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 three 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  *
  73  * Globals used:
  74  *    objects:   $wgLang, $wgContLang
  75  *
  76  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  77  *
  78  * settings:
  79  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  80  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  81  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  82  *  $wgMaxArticleSize*
  83  *
  84  *  * only within ParserOptions
  85  * </pre>
  86  *
  87  * @package MediaWiki
  88  */
  89 class Parser
  90 {
  91         /**#@+
  92          * @private
  93          */
  94         # Persistent:
  95         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  96
  97         # Cleared with clearState():
  98         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
  99         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 100         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 101         var $mIncludeSizes;
 102         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $ot,            // Shortcut alias, see setOutputType()
 113                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 114
 115         /**#@-*/
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function Parser() {
 123                 $this->mTagHooks = array();
 124                 $this->mFunctionHooks = array();
 125                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 126                 $this->mFirstCall = true;
 127         }
 128
 129         /**
 130          * Do various kinds of initialisation on the first call of the parser
 131          */
 132         function firstCallInit() {
 133                 if ( !$this->mFirstCall ) {
 134                         return;
 135                 }
 136
 137                 wfProfileIn( __METHOD__ );
 138                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 139
 140                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 141
 142                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 164
 165                 if ( $wgAllowDisplayTitle ) {
 166                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 167                 }
 168                 if ( $wgAllowSlowParserFunctions ) {
 169                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 170                 }
 171
 172                 $this->initialiseVariables();
 173
 174                 $this->mFirstCall = false;
 175                 wfProfileOut( __METHOD__ );
 176         }
 177
 178         /**
 179          * Clear Parser state
 180          *
 181          * @private
 182          */
 183         function clearState() {
 184                 wfProfileIn( __METHOD__ );
 185                 if ( $this->mFirstCall ) {
 186                         $this->firstCallInit();
 187                 }
 188                 $this->mOutput = new ParserOutput;
 189                 $this->mAutonumber = 0;
 190                 $this->mLastSection = '';
 191                 $this->mDTopen = false;
 192                 $this->mIncludeCount = array();
 193                 $this->mStripState = array();
 194                 $this->mArgStack = array();
 195                 $this->mInPre = false;
 196                 $this->mInterwikiLinkHolders = array(
 197                         'texts' => array(),
 198                         'titles' => array()
 199                 );
 200                 $this->mLinkHolders = array(
 201                         'namespaces' => array(),
 202                         'dbkeys' => array(),
 203                         'queries' => array(),
 204                         'texts' => array(),
 205                         'titles' => array()
 206                 );
 207                 $this->mRevisionId = null;
 208
 209                 /**
 210                  * Prefix for temporary replacement strings for the multipass parser.
 211                  * \x07 should never appear in input as it's disallowed in XML.
 212                  * Using it at the front also gives us a little extra robustness
 213                  * since it shouldn't match when butted up against identifier-like
 214                  * string constructs.
 215                  */
 216                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTemplates = array();
 220                 $this->mTemplatePath = array();
 221
 222                 $this->mShowToc = true;
 223                 $this->mForceTocPosition = false;
 224                 $this->mIncludeSizes = array(
 225                         'pre-expand' => 0,
 226                         'post-expand' => 0,
 227                         'arg' => 0
 228                 );
 229
 230                 wfRunHooks( 'ParserClearState', array( &$this ) );
 231                 wfProfileOut( __METHOD__ );
 232         }
 233
 234         function setOutputType( $ot ) {
 235                 $this->mOutputType = $ot;
 236                 // Shortcut alias
 237                 $this->ot = array(
 238                         'html' => $ot == OT_HTML,
 239                         'wiki' => $ot == OT_WIKI,
 240                         'msg' => $ot == OT_MSG,
 241                         'pre' => $ot == OT_PREPROCESS,
 242                 );
 243         }
 244
 245         /**
 246          * Accessor for mUniqPrefix.
 247          *
 248          * @public
 249          */
 250         function uniqPrefix() {
 251                 return $this->mUniqPrefix;
 252         }
 253
 254         /**
 255          * Convert wikitext to HTML
 256          * Do not call this function recursively.
 257          *
 258          * @private
 259          * @param string $text Text we want to parse
 260          * @param Title &$title A title object
 261          * @param array $options
 262          * @param boolean $linestart
 263          * @param boolean $clearState
 264          * @param int $revid number to pass in {{REVISIONID}}
 265          * @return ParserOutput a ParserOutput
 266          */
 267         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 268                 /**
 269                  * First pass--just handle <nowiki> sections, pass the rest off
 270                  * to internalParse() which does all the real work.
 271                  */
 272
 273                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 274                 $fname = 'Parser::parse';
 275                 wfProfileIn( $fname );
 276
 277                 if ( $clearState ) {
 278                         $this->clearState();
 279                 }
 280
 281                 $this->mOptions = $options;
 282                 $this->mTitle =& $title;
 283                 $oldRevisionId = $this->mRevisionId;
 284                 if( $revid !== null ) {
 285                         $this->mRevisionId = $revid;
 286                 }
 287                 $this->setOutputType( OT_HTML );
 288
 289                 //$text = $this->strip( $text, $this->mStripState );
 290                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 291                 $x =& $this->mStripState;
 292
 293                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 294                 $text = $this->strip( $text, $x );
 295                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 296
 297                 $text = $this->internalParse( $text );
 298
 299                 $text = $this->unstrip( $text, $this->mStripState );
 300
 301                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 302                 $fixtags = array(
 303                         # french spaces, last one Guillemet-left
 304                         # only if there is something before the space
 305                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 306                         # french spaces, Guillemet-right
 307                         '/(\\302\\253) /' => '\\1&nbsp;',
 308                 );
 309                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 310
 311                 # only once and last
 312                 $text = $this->doBlockLevels( $text, $linestart );
 313
 314                 $this->replaceLinkHolders( $text );
 315
 316                 # the position of the parserConvert() call should not be changed. it
 317                 # assumes that the links are all replaced and the only thing left
 318                 # is the <nowiki> mark.
 319                 # Side-effects: this calls $this->mOutput->setTitleText()
 320                 $text = $wgContLang->parserConvert( $text, $this );
 321
 322                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 323
 324                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 325
 326                 $text = Sanitizer::normalizeCharReferences( $text );
 327
 328                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 329                         $text = Parser::tidy($text);
 330                 } else {
 331                         # attempt to sanitize at least some nesting problems
 332                         # (bug #2702 and quite a few others)
 333                         $tidyregs = array(
 334                                 # ''Something [http://www.cool.com cool''] -->
 335                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 336                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 337                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 338                                 # fix up an anchor inside another anchor, only
 339                                 # at least for a single single nested link (bug 3695)
 340                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 341                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 342                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 343                                 # contains a div, so fix it up here; replace
 344                                 # div with escaped text
 345                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 346                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 347                                 # remove empty italic or bold tag pairs, some
 348                                 # introduced by rules above
 349                                 '/<([bi])><\/\\1>/' => '',
 350                         );
 351
 352                         $text = preg_replace(
 353                                 array_keys( $tidyregs ),
 354                                 array_values( $tidyregs ),
 355                                 $text );
 356                 }
 357
 358                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 359
 360                 # Information on include size limits, for the benefit of users who try to skirt them
 361                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 362                         $max = $this->mOptions->getMaxIncludeSize();
 363                         $text .= "<!-- \n" .
 364                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 365                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 366                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 367                                 "Maximum: $max bytes\n" .
 368                                 "-->\n";
 369                 }
 370                 $this->mOutput->setText( $text );
 371                 $this->mRevisionId = $oldRevisionId;
 372                 wfProfileOut( $fname );
 373
 374                 return $this->mOutput;
 375         }
 376
 377         /**
 378          * Recursive parser entry point that can be called from an extension tag
 379          * hook.
 380          */
 381         function recursiveTagParse( $text ) {
 382                 wfProfileIn( __METHOD__ );
 383                 $x =& $this->mStripState;
 384                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 385                 $text = $this->strip( $text, $x );
 386                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 387                 $text = $this->internalParse( $text );
 388                 wfProfileOut( __METHOD__ );
 389                 return $text;
 390         }
 391
 392         /**
 393          * Expand templates and variables in the text, producing valid, static wikitext.
 394          * Also removes comments.
 395          */
 396         function preprocess( $text, $title, $options ) {
 397                 wfProfileIn( __METHOD__ );
 398                 $this->clearState();
 399                 $this->setOutputType( OT_PREPROCESS );
 400                 $this->mOptions = $options;
 401                 $this->mTitle = $title;
 402                 $x =& $this->mStripState;
 403                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 404                 $text = $this->strip( $text, $x );
 405                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 406                 if ( $this->mOptions->getRemoveComments() ) {
 407                         $text = Sanitizer::removeHTMLcomments( $text );
 408                 }
 409                 $text = $this->replaceVariables( $text );
 410                 $text = $this->unstrip( $text, $x );
 411                 $text = $this->unstripNowiki( $text, $x );
 412                 wfProfileOut( __METHOD__ );
 413                 return $text;
 414         }
 415
 416         /**
 417          * Get a random string
 418          *
 419          * @private
 420          * @static
 421          */
 422         function getRandomString() {
 423                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 424         }
 425
 426         function &getTitle() { return $this->mTitle; }
 427         function getOptions() { return $this->mOptions; }
 428
 429         function getFunctionLang() {
 430                 global $wgLang, $wgContLang;
 431                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 432         }
 433
 434         /**
 435          * Replaces all occurrences of HTML-style comments and the given tags
 436          * in the text with a random marker and returns teh next text. The output
 437          * parameter $matches will be an associative array filled with data in
 438          * the form:
 439          *   'UNIQ-xxxxx' => array(
 440          *     'element',
 441          *     'tag content',
 442          *     array( 'param' => 'x' ),
 443          *     '<element param="x">tag content</element>' ) )
 444          *
 445          * @param $elements list of element names. Comments are always extracted.
 446          * @param $text Source text string.
 447          * @param $uniq_prefix
 448          *
 449          * @private
 450          * @static
 451          */
 452         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 453                 static $n = 1;
 454                 $stripped = '';
 455                 $matches = array();
 456
 457                 $taglist = implode( '|', $elements );
 458                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 459
 460                 while ( '' != $text ) {
 461                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 462                         $stripped .= $p[0];
 463                         if( count( $p ) < 5 ) {
 464                                 break;
 465                         }
 466                         if( count( $p ) > 5 ) {
 467                                 // comment
 468                                 $element    = $p[4];
 469                                 $attributes = '';
 470                                 $close      = '';
 471                                 $inside     = $p[5];
 472                         } else {
 473                                 // tag
 474                                 $element    = $p[1];
 475                                 $attributes = $p[2];
 476                                 $close      = $p[3];
 477                                 $inside     = $p[4];
 478                         }
 479
 480                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 481                         $stripped .= $marker;
 482
 483                         if ( $close === '/>' ) {
 484                                 // Empty element tag, <tag />
 485                                 $content = null;
 486                                 $text = $inside;
 487                                 $tail = null;
 488                         } else {
 489                                 if( $element == '!--' ) {
 490                                         $end = '/(-->)/';
 491                                 } else {
 492                                         $end = "/(<\\/$element\\s*>)/i";
 493                                 }
 494                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 495                                 $content = $q[0];
 496                                 if( count( $q ) < 3 ) {
 497                                         # No end tag -- let it run out to the end of the text.
 498                                         $tail = '';
 499                                         $text = '';
 500                                 } else {
 501                                         $tail = $q[1];
 502                                         $text = $q[2];
 503                                 }
 504                         }
 505
 506                         $matches[$marker] = array( $element,
 507                                 $content,
 508                                 Sanitizer::decodeTagAttributes( $attributes ),
 509                                 "<$element$attributes$close$content$tail" );
 510                 }
 511                 return $stripped;
 512         }
 513
 514         /**
 515          * Strips and renders nowiki, pre, math, hiero
 516          * If $render is set, performs necessary rendering operations on plugins
 517          * Returns the text, and fills an array with data needed in unstrip()
 518          * If the $state is already a valid strip state, it adds to the state
 519          *
 520          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 521          *  will be stripped in addition to other tags. This is important
 522          *  for section editing, where these comments cause confusion when
 523          *  counting the sections in the wikisource
 524          *
 525          * @param array dontstrip contains tags which should not be stripped;
 526          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 527          *
 528          * @private
 529          */
 530         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 531                 wfProfileIn( __METHOD__ );
 532                 $render = ($this->mOutputType == OT_HTML);
 533
 534                 $uniq_prefix = $this->mUniqPrefix;
 535                 $commentState = array();
 536
 537                 $elements = array_merge(
 538                         array( 'nowiki', 'gallery' ),
 539                         array_keys( $this->mTagHooks ) );
 540                 global $wgRawHtml;
 541                 if( $wgRawHtml ) {
 542                         $elements[] = 'html';
 543                 }
 544                 if( $this->mOptions->getUseTeX() ) {
 545                         $elements[] = 'math';
 546                 }
 547
 548                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 549                 foreach ( $elements AS $k => $v ) {
 550                         if ( !in_array ( $v , $dontstrip ) ) continue;
 551                         unset ( $elements[$k] );
 552                 }
 553
 554                 $matches = array();
 555                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 556
 557                 foreach( $matches as $marker => $data ) {
 558                         list( $element, $content, $params, $tag ) = $data;
 559                         if( $render ) {
 560                                 $tagName = strtolower( $element );
 561                                 wfProfileIn( __METHOD__."-render-$tagName" );
 562                                 switch( $tagName ) {
 563                                 case '!--':
 564                                         // Comment
 565                                         if( substr( $tag, -3 ) == '-->' ) {
 566                                                 $output = $tag;
 567                                         } else {
 568                                                 // Unclosed comment in input.
 569                                                 // Close it so later stripping can remove it
 570                                                 $output = "$tag-->";
 571                                         }
 572                                         break;
 573                                 case 'html':
 574                                         if( $wgRawHtml ) {
 575                                                 $output = $content;
 576                                                 break;
 577                                         }
 578                                         // Shouldn't happen otherwise. :)
 579                                 case 'nowiki':
 580                                         $output = wfEscapeHTMLTagsOnly( $content );
 581                                         break;
 582                                 case 'math':
 583                                         $output = MathRenderer::renderMath( $content );
 584                                         break;
 585                                 case 'gallery':
 586                                         $output = $this->renderImageGallery( $content, $params );
 587                                         break;
 588                                 default:
 589                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 590                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 591                                                         array( $content, $params, $this ) );
 592                                         } else {
 593                                                 throw new MWException( "Invalid call hook $element" );
 594                                         }
 595                                 }
 596                                 wfProfileOut( __METHOD__."-render-$tagName" );
 597                         } else {
 598                                 // Just stripping tags; keep the source
 599                                 $output = $tag;
 600                         }
 601
 602                         // Unstrip the output, because unstrip() is no longer recursive so
 603                         // it won't do it itself
 604                         $output = $this->unstrip( $output, $state );
 605
 606                         if( !$stripcomments && $element == '!--' ) {
 607                                 $commentState[$marker] = $output;
 608                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 609                                 $state['nowiki'][$marker] = $output;
 610                         } else {
 611                                 $state['general'][$marker] = $output;
 612                         }
 613                 }
 614
 615                 # Unstrip comments unless explicitly told otherwise.
 616                 # (The comments are always stripped prior to this point, so as to
 617                 # not invoke any extension tags / parser hooks contained within
 618                 # a comment.)
 619                 if ( !$stripcomments ) {
 620                         // Put them all back and forget them
 621                         $text = strtr( $text, $commentState );
 622                 }
 623
 624                 wfProfileOut( __METHOD__ );
 625                 return $text;
 626         }
 627
 628         /**
 629          * Restores pre, math, and other extensions removed by strip()
 630          *
 631          * always call unstripNoWiki() after this one
 632          * @private
 633          */
 634         function unstrip( $text, &$state ) {
 635                 if ( !isset( $state['general'] ) ) {
 636                         return $text;
 637                 }
 638
 639                 wfProfileIn( __METHOD__ );
 640                 # TODO: good candidate for FSS
 641                 $text = strtr( $text, $state['general'] );
 642                 wfProfileOut( __METHOD__ );
 643                 return $text;
 644         }
 645
 646         /**
 647          * Always call this after unstrip() to preserve the order
 648          *
 649          * @private
 650          */
 651         function unstripNoWiki( $text, &$state ) {
 652                 if ( !isset( $state['nowiki'] ) ) {
 653                         return $text;
 654                 }
 655
 656                 wfProfileIn( __METHOD__ );
 657                 # TODO: good candidate for FSS
 658                 $text = strtr( $text, $state['nowiki'] );
 659                 wfProfileOut( __METHOD__ );
 660
 661                 return $text;
 662         }
 663
 664         /**
 665          * Add an item to the strip state
 666          * Returns the unique tag which must be inserted into the stripped text
 667          * The tag will be replaced with the original text in unstrip()
 668          *
 669          * @private
 670          */
 671         function insertStripItem( $text, &$state ) {
 672                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 673                 if ( !$state ) {
 674                         $state = array();
 675                 }
 676                 $state['general'][$rnd] = $text;
 677                 return $rnd;
 678         }
 679
 680         /**
 681          * Interface with html tidy, used if $wgUseTidy = true.
 682          * If tidy isn't able to correct the markup, the original will be
 683          * returned in all its glory with a warning comment appended.
 684          *
 685          * Either the external tidy program or the in-process tidy extension
 686          * will be used depending on availability. Override the default
 687          * $wgTidyInternal setting to disable the internal if it's not working.
 688          *
 689          * @param string $text Hideous HTML input
 690          * @return string Corrected HTML output
 691          * @public
 692          * @static
 693          */
 694         function tidy( $text ) {
 695                 global $wgTidyInternal;
 696                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 697 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 698 '<head><title>test</title></head><body>'.$text.'</body></html>';
 699                 if( $wgTidyInternal ) {
 700                         $correctedtext = Parser::internalTidy( $wrappedtext );
 701                 } else {
 702                         $correctedtext = Parser::externalTidy( $wrappedtext );
 703                 }
 704                 if( is_null( $correctedtext ) ) {
 705                         wfDebug( "Tidy error detected!\n" );
 706                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 707                 }
 708                 return $correctedtext;
 709         }
 710
 711         /**
 712          * Spawn an external HTML tidy process and get corrected markup back from it.
 713          *
 714          * @private
 715          * @static
 716          */
 717         function externalTidy( $text ) {
 718                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 719                 $fname = 'Parser::externalTidy';
 720                 wfProfileIn( $fname );
 721
 722                 $cleansource = '';
 723                 $opts = ' -utf8';
 724
 725                 $descriptorspec = array(
 726                         0 => array('pipe', 'r'),
 727                         1 => array('pipe', 'w'),
 728                         2 => array('file', '/dev/null', 'a')
 729                 );
 730                 $pipes = array();
 731                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 732                 if (is_resource($process)) {
 733                         // Theoretically, this style of communication could cause a deadlock
 734                         // here. If the stdout buffer fills up, then writes to stdin could
 735                         // block. This doesn't appear to happen with tidy, because tidy only
 736                         // writes to stdout after it's finished reading from stdin. Search
 737                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 738                         fwrite($pipes[0], $text);
 739                         fclose($pipes[0]);
 740                         while (!feof($pipes[1])) {
 741                                 $cleansource .= fgets($pipes[1], 1024);
 742                         }
 743                         fclose($pipes[1]);
 744                         proc_close($process);
 745                 }
 746
 747                 wfProfileOut( $fname );
 748
 749                 if( $cleansource == '' && $text != '') {
 750                         // Some kind of error happened, so we couldn't get the corrected text.
 751                         // Just give up; we'll use the source text and append a warning.
 752                         return null;
 753                 } else {
 754                         return $cleansource;
 755                 }
 756         }
 757
 758         /**
 759          * Use the HTML tidy PECL extension to use the tidy library in-process,
 760          * saving the overhead of spawning a new process. Currently written to
 761          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 762          *
 763          * 'pear install tidy' should be able to compile the extension module.
 764          *
 765          * @private
 766          * @static
 767          */
 768         function internalTidy( $text ) {
 769                 global $wgTidyConf;
 770                 $fname = 'Parser::internalTidy';
 771                 wfProfileIn( $fname );
 772
 773                 tidy_load_config( $wgTidyConf );
 774                 tidy_set_encoding( 'utf8' );
 775                 tidy_parse_string( $text );
 776                 tidy_clean_repair();
 777                 if( tidy_get_status() == 2 ) {
 778                         // 2 is magic number for fatal error
 779                         // http://www.php.net/manual/en/function.tidy-get-status.php
 780                         $cleansource = null;
 781                 } else {
 782                         $cleansource = tidy_get_output();
 783                 }
 784                 wfProfileOut( $fname );
 785                 return $cleansource;
 786         }
 787
 788         /**
 789          * parse the wiki syntax used to render tables
 790          *
 791          * @private
 792          */
 793         function doTableStuff ( $t ) {
 794                 $fname = 'Parser::doTableStuff';
 795                 wfProfileIn( $fname );
 796
 797                 $t = explode ( "\n" , $t ) ;
 798                 $td = array () ; # Is currently a td tag open?
 799                 $ltd = array () ; # Was it TD or TH?
 800                 $tr = array () ; # Is currently a tr tag open?
 801                 $ltr = array () ; # tr attributes
 802                 $has_opened_tr = array(); # Did this table open a <tr> element?
 803                 $indent_level = 0; # indent level of the table
 804                 foreach ( $t AS $k => $x )
 805                 {
 806                         $x = trim ( $x ) ;
 807                         $fc = substr ( $x , 0 , 1 ) ;
 808                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 809                                 $indent_level = strlen( $matches[1] );
 810
 811                                 $attributes = $this->unstripForHTML( $matches[2] );
 812
 813                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 814                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 815                                 array_push ( $td , false ) ;
 816                                 array_push ( $ltd , '' ) ;
 817                                 array_push ( $tr , false ) ;
 818                                 array_push ( $ltr , '' ) ;
 819                                 array_push ( $has_opened_tr, false );
 820                         }
 821                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 822                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 823                                 $z = "</table>" . substr ( $x , 2);
 824                                 $l = array_pop ( $ltd ) ;
 825                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 826                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 827                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 828                                 array_pop ( $ltr ) ;
 829                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 830                         }
 831                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 832                                 $x = substr ( $x , 1 ) ;
 833                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 834                                 $z = '' ;
 835                                 $l = array_pop ( $ltd ) ;
 836                                 array_pop ( $has_opened_tr );
 837                                 array_push ( $has_opened_tr , true ) ;
 838                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 839                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 840                                 array_pop ( $ltr ) ;
 841                                 $t[$k] = $z ;
 842                                 array_push ( $tr , false ) ;
 843                                 array_push ( $td , false ) ;
 844                                 array_push ( $ltd , '' ) ;
 845                                 $attributes = $this->unstripForHTML( $x );
 846                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 847                         }
 848                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 849                                 # $x is a table row
 850                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 851                                         $fc = '+' ;
 852                                         $x = substr ( $x , 1 ) ;
 853                                 }
 854                                 $after = substr ( $x , 1 ) ;
 855                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 856
 857                                 // Split up multiple cells on the same line.
 858                                 // FIXME: This can result in improper nesting of tags processed
 859                                 // by earlier parser steps, but should avoid splitting up eg
 860                                 // attribute values containing literal "||".
 861                                 $after = wfExplodeMarkup( '||', $after );
 862
 863                                 $t[$k] = '' ;
 864
 865                                 # Loop through each table cell
 866                                 foreach ( $after AS $theline )
 867                                 {
 868                                         $z = '' ;
 869                                         if ( $fc != '+' )
 870                                         {
 871                                                 $tra = array_pop ( $ltr ) ;
 872                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 873                                                 array_push ( $tr , true ) ;
 874                                                 array_push ( $ltr , '' ) ;
 875                                                 array_pop ( $has_opened_tr );
 876                                                 array_push ( $has_opened_tr , true ) ;
 877                                         }
 878
 879                                         $l = array_pop ( $ltd ) ;
 880                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 881                                         if ( $fc == '|' ) $l = 'td' ;
 882                                         else if ( $fc == '!' ) $l = 'th' ;
 883                                         else if ( $fc == '+' ) $l = 'caption' ;
 884                                         else $l = '' ;
 885                                         array_push ( $ltd , $l ) ;
 886
 887                                         # Cell parameters
 888                                         $y = explode ( '|' , $theline , 2 ) ;
 889                                         # Note that a '|' inside an invalid link should not
 890                                         # be mistaken as delimiting cell parameters
 891                                         if ( strpos( $y[0], '[[' ) !== false ) {
 892                                                 $y = array ($theline);
 893                                         }
 894                                         if ( count ( $y ) == 1 )
 895                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 896                                         else {
 897                                                 $attributes = $this->unstripForHTML( $y[0] );
 898                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 899                                         }
 900                                         $t[$k] .= $y ;
 901                                         array_push ( $td , true ) ;
 902                                 }
 903                         }
 904                 }
 905
 906                 # Closing open td, tr && table
 907                 while ( count ( $td ) > 0 )
 908                 {
 909                         $l = array_pop ( $ltd ) ;
 910                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 911                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 912                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 913                         $t[] = '</table>' ;
 914                 }
 915
 916                 $t = implode ( "\n" , $t ) ;
 917                 # special case: don't return empty table
 918                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 919                         $t = '';
 920                 wfProfileOut( $fname );
 921                 return $t ;
 922         }
 923
 924         /**
 925          * Helper function for parse() that transforms wiki markup into
 926          * HTML. Only called for $mOutputType == OT_HTML.
 927          *
 928          * @private
 929          */
 930         function internalParse( $text ) {
 931                 $args = array();
 932                 $isMain = true;
 933                 $fname = 'Parser::internalParse';
 934                 wfProfileIn( $fname );
 935
 936                 # Hook to suspend the parser in this state
 937                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 938                         wfProfileOut( $fname );
 939                         return $text ;
 940                 }
 941
 942                 # Remove <noinclude> tags and <includeonly> sections
 943                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 944                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 945                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 946
 947                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 948
 949                 $text = $this->replaceVariables( $text, $args );
 950
 951                 // Tables need to come after variable replacement for things to work
 952                 // properly; putting them before other transformations should keep
 953                 // exciting things like link expansions from showing up in surprising
 954                 // places.
 955                 $text = $this->doTableStuff( $text );
 956
 957                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 958
 959                 $text = $this->stripToc( $text );
 960                 $this->stripNoGallery( $text );
 961                 $text = $this->doHeadings( $text );
 962                 if($this->mOptions->getUseDynamicDates()) {
 963                         $df =& DateFormatter::getInstance();
 964                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 965                 }
 966                 $text = $this->doAllQuotes( $text );
 967                 $text = $this->replaceInternalLinks( $text );
 968                 $text = $this->replaceExternalLinks( $text );
 969
 970                 # replaceInternalLinks may sometimes leave behind
 971                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 972                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 973
 974                 $text = $this->doMagicLinks( $text );
 975                 $text = $this->formatHeadings( $text, $isMain );
 976
 977                 wfProfileOut( $fname );
 978                 return $text;
 979         }
 980
 981         /**
 982          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 983          * magic external links.
 984          *
 985          * @private
 986          */
 987         function &doMagicLinks( &$text ) {
 988                 wfProfileIn( __METHOD__ );
 989                 $text = preg_replace_callback(
 990                         '!(?:                           # Start cases
 991                             <a.*?</a> |                 # Skip link text
 992                             <.*?> |                     # Skip stuff inside HTML elements
 993                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 994                             ISBN\s+([0-9Xx-]+)          # ISBN, capture number as m[2]
 995                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 996                 wfProfileOut( __METHOD__ );
 997                 return $text;
 998         }
 999
1000         function magicLinkCallback( $m ) {
1001                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1002                         # Skip HTML element
1003                         return $m[0];
1004                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1005                         $isbn = $m[2];
1006                         $num = strtr( $isbn, array(
1007                                 '-' => '',
1008                                 ' ' => '',
1009                                 'x' => 'X',
1010                         ));
1011                         $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
1012                         $text = '<a href="' .
1013                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1014                                 "\" class=\"internal\">ISBN $isbn</a>";
1015                 } else {
1016                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1017                                 $keyword = 'RFC';
1018                                 $urlmsg = 'rfcurl';
1019                                 $id = $m[1];
1020                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1021                                 $keyword = 'PMID';
1022                                 $urlmsg = 'pubmedurl';
1023                                 $id = $m[1];
1024                         } else {
1025                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1026                                         substr($m[0], 0, 20 ) . '"' );
1027                         }
1028
1029                         $url = wfMsg( $urlmsg, $id);
1030                         $sk =& $this->mOptions->getSkin();
1031                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1032                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1033                 }
1034                 return $text;
1035         }
1036
1037         /**
1038          * Parse headers and return html
1039          *
1040          * @private
1041          */
1042         function doHeadings( $text ) {
1043                 $fname = 'Parser::doHeadings';
1044                 wfProfileIn( $fname );
1045                 for ( $i = 6; $i >= 1; --$i ) {
1046                         $h = str_repeat( '=', $i );
1047                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1048                           "<h{$i}>\\1</h{$i}>\\2", $text );
1049                 }
1050                 wfProfileOut( $fname );
1051                 return $text;
1052         }
1053
1054         /**
1055          * Replace single quotes with HTML markup
1056          * @private
1057          * @return string the altered text
1058          */
1059         function doAllQuotes( $text ) {
1060                 $fname = 'Parser::doAllQuotes';
1061                 wfProfileIn( $fname );
1062                 $outtext = '';
1063                 $lines = explode( "\n", $text );
1064                 foreach ( $lines as $line ) {
1065                         $outtext .= $this->doQuotes ( $line ) . "\n";
1066                 }
1067                 $outtext = substr($outtext, 0,-1);
1068                 wfProfileOut( $fname );
1069                 return $outtext;
1070         }
1071
1072         /**
1073          * Helper function for doAllQuotes()
1074          * @private
1075          */
1076         function doQuotes( $text ) {
1077                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1078                 if ( count( $arr ) == 1 )
1079                         return $text;
1080                 else
1081                 {
1082                         # First, do some preliminary work. This may shift some apostrophes from
1083                         # being mark-up to being text. It also counts the number of occurrences
1084                         # of bold and italics mark-ups.
1085                         $i = 0;
1086                         $numbold = 0;
1087                         $numitalics = 0;
1088                         foreach ( $arr as $r )
1089                         {
1090                                 if ( ( $i % 2 ) == 1 )
1091                                 {
1092                                         # If there are ever four apostrophes, assume the first is supposed to
1093                                         # be text, and the remaining three constitute mark-up for bold text.
1094                                         if ( strlen( $arr[$i] ) == 4 )
1095                                         {
1096                                                 $arr[$i-1] .= "'";
1097                                                 $arr[$i] = "'''";
1098                                         }
1099                                         # If there are more than 5 apostrophes in a row, assume they're all
1100                                         # text except for the last 5.
1101                                         else if ( strlen( $arr[$i] ) > 5 )
1102                                         {
1103                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1104                                                 $arr[$i] = "'''''";
1105                                         }
1106                                         # Count the number of occurrences of bold and italics mark-ups.
1107                                         # We are not counting sequences of five apostrophes.
1108                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1109                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1110                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1111                                 }
1112                                 $i++;
1113                         }
1114
1115                         # If there is an odd number of both bold and italics, it is likely
1116                         # that one of the bold ones was meant to be an apostrophe followed
1117                         # by italics. Which one we cannot know for certain, but it is more
1118                         # likely to be one that has a single-letter word before it.
1119                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1120                         {
1121                                 $i = 0;
1122                                 $firstsingleletterword = -1;
1123                                 $firstmultiletterword = -1;
1124                                 $firstspace = -1;
1125                                 foreach ( $arr as $r )
1126                                 {
1127                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1128                                         {
1129                                                 $x1 = substr ($arr[$i-1], -1);
1130                                                 $x2 = substr ($arr[$i-1], -2, 1);
1131                                                 if ($x1 == ' ') {
1132                                                         if ($firstspace == -1) $firstspace = $i;
1133                                                 } else if ($x2 == ' ') {
1134                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1135                                                 } else {
1136                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1137                                                 }
1138                                         }
1139                                         $i++;
1140                                 }
1141
1142                                 # If there is a single-letter word, use it!
1143                                 if ($firstsingleletterword > -1)
1144                                 {
1145                                         $arr [ $firstsingleletterword ] = "''";
1146                                         $arr [ $firstsingleletterword-1 ] .= "'";
1147                                 }
1148                                 # If not, but there's a multi-letter word, use that one.
1149                                 else if ($firstmultiletterword > -1)
1150                                 {
1151                                         $arr [ $firstmultiletterword ] = "''";
1152                                         $arr [ $firstmultiletterword-1 ] .= "'";
1153                                 }
1154                                 # ... otherwise use the first one that has neither.
1155                                 # (notice that it is possible for all three to be -1 if, for example,
1156                                 # there is only one pentuple-apostrophe in the line)
1157                                 else if ($firstspace > -1)
1158                                 {
1159                                         $arr [ $firstspace ] = "''";
1160                                         $arr [ $firstspace-1 ] .= "'";
1161                                 }
1162                         }
1163
1164                         # Now let's actually convert our apostrophic mush to HTML!
1165                         $output = '';
1166                         $buffer = '';
1167                         $state = '';
1168                         $i = 0;
1169                         foreach ($arr as $r)
1170                         {
1171                                 if (($i % 2) == 0)
1172                                 {
1173                                         if ($state == 'both')
1174                                                 $buffer .= $r;
1175                                         else
1176                                                 $output .= $r;
1177                                 }
1178                                 else
1179                                 {
1180                                         if (strlen ($r) == 2)
1181                                         {
1182                                                 if ($state == 'i')
1183                                                 { $output .= '</i>'; $state = ''; }
1184                                                 else if ($state == 'bi')
1185                                                 { $output .= '</i>'; $state = 'b'; }
1186                                                 else if ($state == 'ib')
1187                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1188                                                 else if ($state == 'both')
1189                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1190                                                 else # $state can be 'b' or ''
1191                                                 { $output .= '<i>'; $state .= 'i'; }
1192                                         }
1193                                         else if (strlen ($r) == 3)
1194                                         {
1195                                                 if ($state == 'b')
1196                                                 { $output .= '</b>'; $state = ''; }
1197                                                 else if ($state == 'bi')
1198                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1199                                                 else if ($state == 'ib')
1200                                                 { $output .= '</b>'; $state = 'i'; }
1201                                                 else if ($state == 'both')
1202                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1203                                                 else # $state can be 'i' or ''
1204                                                 { $output .= '<b>'; $state .= 'b'; }
1205                                         }
1206                                         else if (strlen ($r) == 5)
1207                                         {
1208                                                 if ($state == 'b')
1209                                                 { $output .= '</b><i>'; $state = 'i'; }
1210                                                 else if ($state == 'i')
1211                                                 { $output .= '</i><b>'; $state = 'b'; }
1212                                                 else if ($state == 'bi')
1213                                                 { $output .= '</i></b>'; $state = ''; }
1214                                                 else if ($state == 'ib')
1215                                                 { $output .= '</b></i>'; $state = ''; }
1216                                                 else if ($state == 'both')
1217                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1218                                                 else # ($state == '')
1219                                                 { $buffer = ''; $state = 'both'; }
1220                                         }
1221                                 }
1222                                 $i++;
1223                         }
1224                         # Now close all remaining tags.  Notice that the order is important.
1225                         if ($state == 'b' || $state == 'ib')
1226                                 $output .= '</b>';
1227                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1228                                 $output .= '</i>';
1229                         if ($state == 'bi')
1230                                 $output .= '</b>';
1231                         if ($state == 'both')
1232                                 $output .= '<b><i>'.$buffer.'</i></b>';
1233                         return $output;
1234                 }
1235         }
1236
1237         /**
1238          * Replace external links
1239          *
1240          * Note: this is all very hackish and the order of execution matters a lot.
1241          * Make sure to run maintenance/parserTests.php if you change this code.
1242          *
1243          * @private
1244          */
1245         function replaceExternalLinks( $text ) {
1246                 global $wgContLang;
1247                 $fname = 'Parser::replaceExternalLinks';
1248                 wfProfileIn( $fname );
1249
1250                 $sk =& $this->mOptions->getSkin();
1251
1252                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1253
1254                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1255
1256                 $i = 0;
1257                 while ( $i<count( $bits ) ) {
1258                         $url = $bits[$i++];
1259                         $protocol = $bits[$i++];
1260                         $text = $bits[$i++];
1261                         $trail = $bits[$i++];
1262
1263                         # The characters '<' and '>' (which were escaped by
1264                         # removeHTMLtags()) should not be included in
1265                         # URLs, per RFC 2396.
1266                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1267                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1268                                 $url = substr($url, 0, $m2[0][1]);
1269                         }
1270
1271                         # If the link text is an image URL, replace it with an <img> tag
1272                         # This happened by accident in the original parser, but some people used it extensively
1273                         $img = $this->maybeMakeExternalImage( $text );
1274                         if ( $img !== false ) {
1275                                 $text = $img;
1276                         }
1277
1278                         $dtrail = '';
1279
1280                         # Set linktype for CSS - if URL==text, link is essentially free
1281                         $linktype = ($text == $url) ? 'free' : 'text';
1282
1283                         # No link text, e.g. [http://domain.tld/some.link]
1284                         if ( $text == '' ) {
1285                                 # Autonumber if allowed. See bug #5918
1286                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1287                                         $text = '[' . ++$this->mAutonumber . ']';
1288                                         $linktype = 'autonumber';
1289                                 } else {
1290                                         # Otherwise just use the URL
1291                                         $text = htmlspecialchars( $url );
1292                                         $linktype = 'free';
1293                                 }
1294                         } else {
1295                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1296                                 # Check for trail
1297                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1298                         }
1299
1300                         $text = $wgContLang->markNoConversion($text);
1301
1302                         $url = Sanitizer::cleanUrl( $url );
1303
1304                         # Process the trail (i.e. everything after this link up until start of the next link),
1305                         # replacing any non-bracketed links
1306                         $trail = $this->replaceFreeExternalLinks( $trail );
1307
1308                         # Use the encoded URL
1309                         # This means that users can paste URLs directly into the text
1310                         # Funny characters like &ouml; aren't valid in URLs anyway
1311                         # This was changed in August 2004
1312                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1313
1314                         # Register link in the output object.
1315                         # Replace unnecessary URL escape codes with the referenced character
1316                         # This prevents spammers from hiding links from the filters
1317                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1318                         $this->mOutput->addExternalLink( $pasteurized );
1319                 }
1320
1321                 wfProfileOut( $fname );
1322                 return $s;
1323         }
1324
1325         /**
1326          * Replace anything that looks like a URL with a link
1327          * @private
1328          */
1329         function replaceFreeExternalLinks( $text ) {
1330                 global $wgContLang;
1331                 $fname = 'Parser::replaceFreeExternalLinks';
1332                 wfProfileIn( $fname );
1333
1334                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1335                 $s = array_shift( $bits );
1336                 $i = 0;
1337
1338                 $sk =& $this->mOptions->getSkin();
1339
1340                 while ( $i < count( $bits ) ){
1341                         $protocol = $bits[$i++];
1342                         $remainder = $bits[$i++];
1343
1344                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1345                                 # Found some characters after the protocol that look promising
1346                                 $url = $protocol . $m[1];
1347                                 $trail = $m[2];
1348
1349                                 # special case: handle urls as url args:
1350                                 # http://www.example.com/foo?=http://www.example.com/bar
1351                                 if(strlen($trail) == 0 &&
1352                                         isset($bits[$i]) &&
1353                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1354                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1355                                 {
1356                                         # add protocol, arg
1357                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1358                                         $i += 2;
1359                                         $trail = $m[2];
1360                                 }
1361
1362                                 # The characters '<' and '>' (which were escaped by
1363                                 # removeHTMLtags()) should not be included in
1364                                 # URLs, per RFC 2396.
1365                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1366                                         $trail = substr($url, $m2[0][1]) . $trail;
1367                                         $url = substr($url, 0, $m2[0][1]);
1368                                 }
1369
1370                                 # Move trailing punctuation to $trail
1371                                 $sep = ',;\.:!?';
1372                                 # If there is no left bracket, then consider right brackets fair game too
1373                                 if ( strpos( $url, '(' ) === false ) {
1374                                         $sep .= ')';
1375                                 }
1376
1377                                 $numSepChars = strspn( strrev( $url ), $sep );
1378                                 if ( $numSepChars ) {
1379                                         $trail = substr( $url, -$numSepChars ) . $trail;
1380                                         $url = substr( $url, 0, -$numSepChars );
1381                                 }
1382
1383                                 $url = Sanitizer::cleanUrl( $url );
1384
1385                                 # Is this an external image?
1386                                 $text = $this->maybeMakeExternalImage( $url );
1387                                 if ( $text === false ) {
1388                                         # Not an image, make a link
1389                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1390                                         # Register it in the output object...
1391                                         # Replace unnecessary URL escape codes with their equivalent characters
1392                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1393                                         $this->mOutput->addExternalLink( $pasteurized );
1394                                 }
1395                                 $s .= $text . $trail;
1396                         } else {
1397                                 $s .= $protocol . $remainder;
1398                         }
1399                 }
1400                 wfProfileOut( $fname );
1401                 return $s;
1402         }
1403
1404         /**
1405          * Replace unusual URL escape codes with their equivalent characters
1406          * @param string
1407          * @return string
1408          * @static
1409          * @fixme This can merge genuinely required bits in the path or query string,
1410          *        breaking legit URLs. A proper fix would treat the various parts of
1411          *        the URL differently; as a workaround, just use the output for
1412          *        statistical records, not for actual linking/output.
1413          */
1414         static function replaceUnusualEscapes( $url ) {
1415                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1416                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1417         }
1418
1419         /**
1420          * Callback function used in replaceUnusualEscapes().
1421          * Replaces unusual URL escape codes with their equivalent character
1422          * @static
1423          * @private
1424          */
1425         private static function replaceUnusualEscapesCallback( $matches ) {
1426                 $char = urldecode( $matches[0] );
1427                 $ord = ord( $char );
1428                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1429                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1430                         // No, shouldn't be escaped
1431                         return $char;
1432                 } else {
1433                         // Yes, leave it escaped
1434                         return $matches[0];
1435                 }
1436         }
1437
1438         /**
1439          * make an image if it's allowed, either through the global
1440          * option or through the exception
1441          * @private
1442          */
1443         function maybeMakeExternalImage( $url ) {
1444                 $sk =& $this->mOptions->getSkin();
1445                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1446                 $imagesexception = !empty($imagesfrom);
1447                 $text = false;
1448                 if ( $this->mOptions->getAllowExternalImages()
1449                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1450                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1451                                 # Image found
1452                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1453                         }
1454                 }
1455                 return $text;
1456         }
1457
1458         /**
1459          * Process [[ ]] wikilinks
1460          *
1461          * @private
1462          */
1463         function replaceInternalLinks( $s ) {
1464                 global $wgContLang;
1465                 static $fname = 'Parser::replaceInternalLinks' ;
1466
1467                 wfProfileIn( $fname );
1468
1469                 wfProfileIn( $fname.'-setup' );
1470                 static $tc = FALSE;
1471                 # the % is needed to support urlencoded titles as well
1472                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1473
1474                 $sk =& $this->mOptions->getSkin();
1475
1476                 #split the entire text string on occurences of [[
1477                 $a = explode( '[[', ' ' . $s );
1478                 #get the first element (all text up to first [[), and remove the space we added
1479                 $s = array_shift( $a );
1480                 $s = substr( $s, 1 );
1481
1482                 # Match a link having the form [[namespace:link|alternate]]trail
1483                 static $e1 = FALSE;
1484                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1485                 # Match cases where there is no "]]", which might still be images
1486                 static $e1_img = FALSE;
1487                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1488                 # Match the end of a line for a word that's not followed by whitespace,
1489                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1490                 $e2 = wfMsgForContent( 'linkprefix' );
1491
1492                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1493
1494                 if( is_null( $this->mTitle ) ) {
1495                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1496                 }
1497                 $nottalk = !$this->mTitle->isTalkPage();
1498
1499                 if ( $useLinkPrefixExtension ) {
1500                         if ( preg_match( $e2, $s, $m ) ) {
1501                                 $first_prefix = $m[2];
1502                         } else {
1503                                 $first_prefix = false;
1504                         }
1505                 } else {
1506                         $prefix = '';
1507                 }
1508
1509                 $selflink = $this->mTitle->getPrefixedText();
1510                 $useSubpages = $this->areSubpagesAllowed();
1511                 wfProfileOut( $fname.'-setup' );
1512
1513                 # Loop for each link
1514                 for ($k = 0; isset( $a[$k] ); $k++) {
1515                         $line = $a[$k];
1516                         if ( $useLinkPrefixExtension ) {
1517                                 wfProfileIn( $fname.'-prefixhandling' );
1518                                 if ( preg_match( $e2, $s, $m ) ) {
1519                                         $prefix = $m[2];
1520                                         $s = $m[1];
1521                                 } else {
1522                                         $prefix='';
1523                                 }
1524                                 # first link
1525                                 if($first_prefix) {
1526                                         $prefix = $first_prefix;
1527                                         $first_prefix = false;
1528                                 }
1529                                 wfProfileOut( $fname.'-prefixhandling' );
1530                         }
1531
1532                         $might_be_img = false;
1533
1534                         wfProfileIn( "$fname-e1" );
1535                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1536                                 $text = $m[2];
1537                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1538                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1539                                 # the real problem is with the $e1 regex
1540                                 # See bug 1300.
1541                                 #
1542                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1543                                 # and no image is in sight. See bug 2095.
1544                                 #
1545                                 if( $text !== '' &&
1546                                         substr( $m[3], 0, 1 ) === ']' &&
1547                                         strpos($text, '[') !== false
1548                                 )
1549                                 {
1550                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1551                                         $m[3] = substr( $m[3], 1 );
1552                                 }
1553                                 # fix up urlencoded title texts
1554                                 if( strpos( $m[1], '%' ) !== false ) {
1555                                         # Should anchors '#' also be rejected?
1556                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1557                                 }
1558                                 $trail = $m[3];
1559                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1560                                 $might_be_img = true;
1561                                 $text = $m[2];
1562                                 if ( strpos( $m[1], '%' ) !== false ) {
1563                                        $m[1] = urldecode($m[1]);
1564                                 }
1565                                 $trail = "";
1566                         } else { # Invalid form; output directly
1567                                 $s .= $prefix . '[[' . $line ;
1568                                 wfProfileOut( "$fname-e1" );
1569                                 continue;
1570                         }
1571                         wfProfileOut( "$fname-e1" );
1572                         wfProfileIn( "$fname-misc" );
1573
1574                         # Don't allow internal links to pages containing
1575                         # PROTO: where PROTO is a valid URL protocol; these
1576                         # should be external links.
1577                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1578                                 $s .= $prefix . '[[' . $line ;
1579                                 continue;
1580                         }
1581
1582                         # Make subpage if necessary
1583                         if( $useSubpages ) {
1584                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1585                         } else {
1586                                 $link = $m[1];
1587                         }
1588
1589                         $noforce = (substr($m[1], 0, 1) != ':');
1590                         if (!$noforce) {
1591                                 # Strip off leading ':'
1592                                 $link = substr($link, 1);
1593                         }
1594
1595                         wfProfileOut( "$fname-misc" );
1596                         wfProfileIn( "$fname-title" );
1597                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1598                         if( !$nt ) {
1599                                 $s .= $prefix . '[[' . $line;
1600                                 wfProfileOut( "$fname-title" );
1601                                 continue;
1602                         }
1603
1604                         $ns = $nt->getNamespace();
1605                         $iw = $nt->getInterWiki();
1606                         wfProfileOut( "$fname-title" );
1607
1608                         if ($might_be_img) { # if this is actually an invalid link
1609                                 wfProfileIn( "$fname-might_be_img" );
1610                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1611                                         $found = false;
1612                                         while (isset ($a[$k+1]) ) {
1613                                                 #look at the next 'line' to see if we can close it there
1614                                                 $spliced = array_splice( $a, $k + 1, 1 );
1615                                                 $next_line = array_shift( $spliced );
1616                                                 $m = explode( ']]', $next_line, 3 );
1617                                                 if ( count( $m ) == 3 ) {
1618                                                         # the first ]] closes the inner link, the second the image
1619                                                         $found = true;
1620                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1621                                                         $trail = $m[2];
1622                                                         break;
1623                                                 } elseif ( count( $m ) == 2 ) {
1624                                                         #if there's exactly one ]] that's fine, we'll keep looking
1625                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1626                                                 } else {
1627                                                         #if $next_line is invalid too, we need look no further
1628                                                         $text .= '[[' . $next_line;
1629                                                         break;
1630                                                 }
1631                                         }
1632                                         if ( !$found ) {
1633                                                 # we couldn't find the end of this imageLink, so output it raw
1634                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1635                                                 $text = $this->replaceInternalLinks($text);
1636                                                 $s .= "{$prefix}[[$link|$text";
1637                                                 # note: no $trail, because without an end, there *is* no trail
1638                                                 wfProfileOut( "$fname-might_be_img" );
1639                                                 continue;
1640                                         }
1641                                 } else { #it's not an image, so output it raw
1642                                         $s .= "{$prefix}[[$link|$text";
1643                                         # note: no $trail, because without an end, there *is* no trail
1644                                         wfProfileOut( "$fname-might_be_img" );
1645                                         continue;
1646                                 }
1647                                 wfProfileOut( "$fname-might_be_img" );
1648                         }
1649
1650                         $wasblank = ( '' == $text );
1651                         if( $wasblank ) $text = $link;
1652
1653                         # Link not escaped by : , create the various objects
1654                         if( $noforce ) {
1655
1656                                 # Interwikis
1657                                 wfProfileIn( "$fname-interwiki" );
1658                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1659                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1660                                         $s = rtrim($s . "\n");
1661                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1662                                         wfProfileOut( "$fname-interwiki" );
1663                                         continue;
1664                                 }
1665                                 wfProfileOut( "$fname-interwiki" );
1666
1667                                 if ( $ns == NS_IMAGE ) {
1668                                         wfProfileIn( "$fname-image" );
1669                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1670                                                 # recursively parse links inside the image caption
1671                                                 # actually, this will parse them in any other parameters, too,
1672                                                 # but it might be hard to fix that, and it doesn't matter ATM
1673                                                 $text = $this->replaceExternalLinks($text);
1674                                                 $text = $this->replaceInternalLinks($text);
1675
1676                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1677                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1678                                                 $this->mOutput->addImage( $nt->getDBkey() );
1679
1680                                                 wfProfileOut( "$fname-image" );
1681                                                 continue;
1682                                         } else {
1683                                                 # We still need to record the image's presence on the page
1684                                                 $this->mOutput->addImage( $nt->getDBkey() );
1685                                         }
1686                                         wfProfileOut( "$fname-image" );
1687
1688                                 }
1689
1690                                 if ( $ns == NS_CATEGORY ) {
1691                                         wfProfileIn( "$fname-category" );
1692                                         $s = rtrim($s . "\n"); # bug 87
1693
1694                                         if ( $wasblank ) {
1695                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1696                                                         $sortkey = $this->mTitle->getText();
1697                                                 } else {
1698                                                         $sortkey = $this->mTitle->getPrefixedText();
1699                                                 }
1700                                         } else {
1701                                                 $sortkey = $text;
1702                                         }
1703                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1704                                         $sortkey = str_replace( "\n", '', $sortkey );
1705                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1706                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1707
1708                                         /**
1709                                          * Strip the whitespace Category links produce, see bug 87
1710                                          * @todo We might want to use trim($tmp, "\n") here.
1711                                          */
1712                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1713
1714                                         wfProfileOut( "$fname-category" );
1715                                         continue;
1716                                 }
1717                         }
1718
1719                         if( ( $nt->getPrefixedText() === $selflink ) &&
1720                             ( $nt->getFragment() === '' ) ) {
1721                                 # Self-links are handled specially; generally de-link and change to bold.
1722                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1723                                 continue;
1724                         }
1725
1726                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1727                         if( $ns == NS_MEDIA ) {
1728                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1729                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1730                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1731                                 $this->mOutput->addImage( $nt->getDBkey() );
1732                                 continue;
1733                         } elseif( $ns == NS_SPECIAL ) {
1734                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1735                                 continue;
1736                         } elseif( $ns == NS_IMAGE ) {
1737                                 $img = new Image( $nt );
1738                                 if( $img->exists() ) {
1739                                         // Force a blue link if the file exists; may be a remote
1740                                         // upload on the shared repository, and we want to see its
1741                                         // auto-generated page.
1742                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1743                                         $this->mOutput->addLink( $nt );
1744                                         continue;
1745                                 }
1746                         }
1747                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1748                 }
1749                 wfProfileOut( $fname );
1750                 return $s;
1751         }
1752
1753         /**
1754          * Make a link placeholder. The text returned can be later resolved to a real link with
1755          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1756          * parsing of interwiki links, and secondly to allow all existence checks and
1757          * article length checks (for stub links) to be bundled into a single query.
1758          *
1759          */
1760         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1761                 wfProfileIn( __METHOD__ );
1762                 if ( ! is_object($nt) ) {
1763                         # Fail gracefully
1764                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1765                 } else {
1766                         # Separate the link trail from the rest of the link
1767                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1768
1769                         if ( $nt->isExternal() ) {
1770                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1771                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1772                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1773                         } else {
1774                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1775                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1776                                 $this->mLinkHolders['queries'][] = $query;
1777                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1778                                 $this->mLinkHolders['titles'][] = $nt;
1779
1780                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1781                         }
1782                 }
1783                 wfProfileOut( __METHOD__ );
1784                 return $retVal;
1785         }
1786
1787         /**
1788          * Render a forced-blue link inline; protect against double expansion of
1789          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1790          * Since this little disaster has to split off the trail text to avoid
1791          * breaking URLs in the following text without breaking trails on the
1792          * wiki links, it's been made into a horrible function.
1793          *
1794          * @param Title $nt
1795          * @param string $text
1796          * @param string $query
1797          * @param string $trail
1798          * @param string $prefix
1799          * @return string HTML-wikitext mix oh yuck
1800          */
1801         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1802                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1803                 $sk =& $this->mOptions->getSkin();
1804                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1805                 return $this->armorLinks( $link ) . $trail;
1806         }
1807
1808         /**
1809          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1810          * going to go through further parsing steps before inline URL expansion.
1811          *
1812          * In particular this is important when using action=render, which causes
1813          * full URLs to be included.
1814          *
1815          * Oh man I hate our multi-layer parser!
1816          *
1817          * @param string more-or-less HTML
1818          * @return string less-or-more HTML with NOPARSE bits
1819          */
1820         function armorLinks( $text ) {
1821                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1822                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1823         }
1824
1825         /**
1826          * Return true if subpage links should be expanded on this page.
1827          * @return bool
1828          */
1829         function areSubpagesAllowed() {
1830                 # Some namespaces don't allow subpages
1831                 global $wgNamespacesWithSubpages;
1832                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1833         }
1834
1835         /**
1836          * Handle link to subpage if necessary
1837          * @param string $target the source of the link
1838          * @param string &$text the link text, modified as necessary
1839          * @return string the full name of the link
1840          * @private
1841          */
1842         function maybeDoSubpageLink($target, &$text) {
1843                 # Valid link forms:
1844                 # Foobar -- normal
1845                 # :Foobar -- override special treatment of prefix (images, language links)
1846                 # /Foobar -- convert to CurrentPage/Foobar
1847                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1848                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1849                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1850
1851                 $fname = 'Parser::maybeDoSubpageLink';
1852                 wfProfileIn( $fname );
1853                 $ret = $target; # default return value is no change
1854
1855                 # Some namespaces don't allow subpages,
1856                 # so only perform processing if subpages are allowed
1857                 if( $this->areSubpagesAllowed() ) {
1858                         # Look at the first character
1859                         if( $target != '' && $target{0} == '/' ) {
1860                                 # / at end means we don't want the slash to be shown
1861                                 if( substr( $target, -1, 1 ) == '/' ) {
1862                                         $target = substr( $target, 1, -1 );
1863                                         $noslash = $target;
1864                                 } else {
1865                                         $noslash = substr( $target, 1 );
1866                                 }
1867
1868                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1869                                 if( '' === $text ) {
1870                                         $text = $target;
1871                                 } # this might be changed for ugliness reasons
1872                         } else {
1873                                 # check for .. subpage backlinks
1874                                 $dotdotcount = 0;
1875                                 $nodotdot = $target;
1876                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1877                                         ++$dotdotcount;
1878                                         $nodotdot = substr( $nodotdot, 3 );
1879                                 }
1880                                 if($dotdotcount > 0) {
1881                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1882                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1883                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1884                                                 # / at the end means don't show full path
1885                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1886                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1887                                                         if( '' === $text ) {
1888                                                                 $text = $nodotdot;
1889                                                         }
1890                                                 }
1891                                                 $nodotdot = trim( $nodotdot );
1892                                                 if( $nodotdot != '' ) {
1893                                                         $ret .= '/' . $nodotdot;
1894                                                 }
1895                                         }
1896                                 }
1897                         }
1898                 }
1899
1900                 wfProfileOut( $fname );
1901                 return $ret;
1902         }
1903
1904         /**#@+
1905          * Used by doBlockLevels()
1906          * @private
1907          */
1908         /* private */ function closeParagraph() {
1909                 $result = '';
1910                 if ( '' != $this->mLastSection ) {
1911                         $result = '</' . $this->mLastSection  . ">\n";
1912                 }
1913                 $this->mInPre = false;
1914                 $this->mLastSection = '';
1915                 return $result;
1916         }
1917         # getCommon() returns the length of the longest common substring
1918         # of both arguments, starting at the beginning of both.
1919         #
1920         /* private */ function getCommon( $st1, $st2 ) {
1921                 $fl = strlen( $st1 );
1922                 $shorter = strlen( $st2 );
1923                 if ( $fl < $shorter ) { $shorter = $fl; }
1924
1925                 for ( $i = 0; $i < $shorter; ++$i ) {
1926                         if ( $st1{$i} != $st2{$i} ) { break; }
1927                 }
1928                 return $i;
1929         }
1930         # These next three functions open, continue, and close the list
1931         # element appropriate to the prefix character passed into them.
1932         #
1933         /* private */ function openList( $char ) {
1934                 $result = $this->closeParagraph();
1935
1936                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1937                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1938                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1939                 else if ( ';' == $char ) {
1940                         $result .= '<dl><dt>';
1941                         $this->mDTopen = true;
1942                 }
1943                 else { $result = '<!-- ERR 1 -->'; }
1944
1945                 return $result;
1946         }
1947
1948         /* private */ function nextItem( $char ) {
1949                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1950                 else if ( ':' == $char || ';' == $char ) {
1951                         $close = '</dd>';
1952                         if ( $this->mDTopen ) { $close = '</dt>'; }
1953                         if ( ';' == $char ) {
1954                                 $this->mDTopen = true;
1955                                 return $close . '<dt>';
1956                         } else {
1957                                 $this->mDTopen = false;
1958                                 return $close . '<dd>';
1959                         }
1960                 }
1961                 return '<!-- ERR 2 -->';
1962         }
1963
1964         /* private */ function closeList( $char ) {
1965                 if ( '*' == $char ) { $text = '</li></ul>'; }
1966                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1967                 else if ( ':' == $char ) {
1968                         if ( $this->mDTopen ) {
1969                                 $this->mDTopen = false;
1970                                 $text = '</dt></dl>';
1971                         } else {
1972                                 $text = '</dd></dl>';
1973                         }
1974                 }
1975                 else {  return '<!-- ERR 3 -->'; }
1976                 return $text."\n";
1977         }
1978         /**#@-*/
1979
1980         /**
1981          * Make lists from lines starting with ':', '*', '#', etc.
1982          *
1983          * @private
1984          * @return string the lists rendered as HTML
1985          */
1986         function doBlockLevels( $text, $linestart ) {
1987                 $fname = 'Parser::doBlockLevels';
1988                 wfProfileIn( $fname );
1989
1990                 # Parsing through the text line by line.  The main thing
1991                 # happening here is handling of block-level elements p, pre,
1992                 # and making lists from lines starting with * # : etc.
1993                 #
1994                 $textLines = explode( "\n", $text );
1995
1996                 $lastPrefix = $output = '';
1997                 $this->mDTopen = $inBlockElem = false;
1998                 $prefixLength = 0;
1999                 $paragraphStack = false;
2000
2001                 if ( !$linestart ) {
2002                         $output .= array_shift( $textLines );
2003                 }
2004                 foreach ( $textLines as $oLine ) {
2005                         $lastPrefixLength = strlen( $lastPrefix );
2006                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2007                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2008                         if ( !$this->mInPre ) {
2009                                 # Multiple prefixes may abut each other for nested lists.
2010                                 $prefixLength = strspn( $oLine, '*#:;' );
2011                                 $pref = substr( $oLine, 0, $prefixLength );
2012
2013                                 # eh?
2014                                 $pref2 = str_replace( ';', ':', $pref );
2015                                 $t = substr( $oLine, $prefixLength );
2016                                 $this->mInPre = !empty($preOpenMatch);
2017                         } else {
2018                                 # Don't interpret any other prefixes in preformatted text
2019                                 $prefixLength = 0;
2020                                 $pref = $pref2 = '';
2021                                 $t = $oLine;
2022                         }
2023
2024                         # List generation
2025                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2026                                 # Same as the last item, so no need to deal with nesting or opening stuff
2027                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2028                                 $paragraphStack = false;
2029
2030                                 if ( substr( $pref, -1 ) == ';') {
2031                                         # The one nasty exception: definition lists work like this:
2032                                         # ; title : definition text
2033                                         # So we check for : in the remainder text to split up the
2034                                         # title and definition, without b0rking links.
2035                                         $term = $t2 = '';
2036                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2037                                                 $t = $t2;
2038                                                 $output .= $term . $this->nextItem( ':' );
2039                                         }
2040                                 }
2041                         } elseif( $prefixLength || $lastPrefixLength ) {
2042                                 # Either open or close a level...
2043                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2044                                 $paragraphStack = false;
2045
2046                                 while( $commonPrefixLength < $lastPrefixLength ) {
2047                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2048                                         --$lastPrefixLength;
2049                                 }
2050                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2051                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2052                                 }
2053                                 while ( $prefixLength > $commonPrefixLength ) {
2054                                         $char = substr( $pref, $commonPrefixLength, 1 );
2055                                         $output .= $this->openList( $char );
2056
2057                                         if ( ';' == $char ) {
2058                                                 # FIXME: This is dupe of code above
2059                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2060                                                         $t = $t2;
2061                                                         $output .= $term . $this->nextItem( ':' );
2062                                                 }
2063                                         }
2064                                         ++$commonPrefixLength;
2065                                 }
2066                                 $lastPrefix = $pref2;
2067                         }
2068                         if( 0 == $prefixLength ) {
2069                                 wfProfileIn( "$fname-paragraph" );
2070                                 # No prefix (not in list)--go to paragraph mode
2071                                 // XXX: use a stack for nestable elements like span, table and div
2072                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
2073                                 $closematch = preg_match(
2074                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2075                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
2076                                 if ( $openmatch or $closematch ) {
2077                                         $paragraphStack = false;
2078                                         # TODO bug 5718: paragraph closed
2079                                         $output .= $this->closeParagraph();
2080                                         if ( $preOpenMatch and !$preCloseMatch ) {
2081                                                 $this->mInPre = true;
2082                                         }
2083                                         if ( $closematch ) {
2084                                                 $inBlockElem = false;
2085                                         } else {
2086                                                 $inBlockElem = true;
2087                                         }
2088                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2089                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2090                                                 // pre
2091                                                 if ($this->mLastSection != 'pre') {
2092                                                         $paragraphStack = false;
2093                                                         $output .= $this->closeParagraph().'<pre>';
2094                                                         $this->mLastSection = 'pre';
2095                                                 }
2096                                                 $t = substr( $t, 1 );
2097                                         } else {
2098                                                 // paragraph
2099                                                 if ( '' == trim($t) ) {
2100                                                         if ( $paragraphStack ) {
2101                                                                 $output .= $paragraphStack.'<br />';
2102                                                                 $paragraphStack = false;
2103                                                                 $this->mLastSection = 'p';
2104                                                         } else {
2105                                                                 if ($this->mLastSection != 'p' ) {
2106                                                                         $output .= $this->closeParagraph();
2107                                                                         $this->mLastSection = '';
2108                                                                         $paragraphStack = '<p>';
2109                                                                 } else {
2110                                                                         $paragraphStack = '</p><p>';
2111                                                                 }
2112                                                         }
2113                                                 } else {
2114                                                         if ( $paragraphStack ) {
2115                                                                 $output .= $paragraphStack;
2116                                                                 $paragraphStack = false;
2117                                                                 $this->mLastSection = 'p';
2118                                                         } else if ($this->mLastSection != 'p') {
2119                                                                 $output .= $this->closeParagraph().'<p>';
2120                                                                 $this->mLastSection = 'p';
2121                                                         }
2122                                                 }
2123                                         }
2124                                 }
2125                                 wfProfileOut( "$fname-paragraph" );
2126                         }
2127                         // somewhere above we forget to get out of pre block (bug 785)
2128                         if($preCloseMatch && $this->mInPre) {
2129                                 $this->mInPre = false;
2130                         }
2131                         if ($paragraphStack === false) {
2132                                 $output .= $t."\n";
2133                         }
2134                 }
2135                 while ( $prefixLength ) {
2136                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2137                         --$prefixLength;
2138                 }
2139                 if ( '' != $this->mLastSection ) {
2140                         $output .= '</' . $this->mLastSection . '>';
2141                         $this->mLastSection = '';
2142                 }
2143
2144                 wfProfileOut( $fname );
2145                 return $output;
2146         }
2147
2148         /**
2149          * Split up a string on ':', ignoring any occurences inside tags
2150          * to prevent illegal overlapping.
2151          * @param string $str the string to split
2152          * @param string &$before set to everything before the ':'
2153          * @param string &$after set to everything after the ':'
2154          * return string the position of the ':', or false if none found
2155          */
2156         function findColonNoLinks($str, &$before, &$after) {
2157                 $fname = 'Parser::findColonNoLinks';
2158                 wfProfileIn( $fname );
2159
2160                 $pos = strpos( $str, ':' );
2161                 if( $pos === false ) {
2162                         // Nothing to find!
2163                         wfProfileOut( $fname );
2164                         return false;
2165                 }
2166
2167                 $lt = strpos( $str, '<' );
2168                 if( $lt === false || $lt > $pos ) {
2169                         // Easy; no tag nesting to worry about
2170                         $before = substr( $str, 0, $pos );
2171                         $after = substr( $str, $pos+1 );
2172                         wfProfileOut( $fname );
2173                         return $pos;
2174                 }
2175
2176                 // Ugly state machine to walk through avoiding tags.
2177                 $state = MW_COLON_STATE_TEXT;
2178                 $stack = 0;
2179                 $len = strlen( $str );
2180                 for( $i = 0; $i < $len; $i++ ) {
2181                         $c = $str{$i};
2182
2183                         switch( $state ) {
2184                         // (Using the number is a performance hack for common cases)
2185                         case 0: // MW_COLON_STATE_TEXT:
2186                                 switch( $c ) {
2187                                 case "<":
2188                                         // Could be either a <start> tag or an </end> tag
2189                                         $state = MW_COLON_STATE_TAGSTART;
2190                                         break;
2191                                 case ":":
2192                                         if( $stack == 0 ) {
2193                                                 // We found it!
2194                                                 $before = substr( $str, 0, $i );
2195                                                 $after = substr( $str, $i + 1 );
2196                                                 wfProfileOut( $fname );
2197                                                 return $i;
2198                                         }
2199                                         // Embedded in a tag; don't break it.
2200                                         break;
2201                                 default:
2202                                         // Skip ahead looking for something interesting
2203                                         $colon = strpos( $str, ':', $i );
2204                                         if( $colon === false ) {
2205                                                 // Nothing else interesting
2206                                                 wfProfileOut( $fname );
2207                                                 return false;
2208                                         }
2209                                         $lt = strpos( $str, '<', $i );
2210                                         if( $stack === 0 ) {
2211                                                 if( $lt === false || $colon < $lt ) {
2212                                                         // We found it!
2213                                                         $before = substr( $str, 0, $colon );
2214                                                         $after = substr( $str, $colon + 1 );
2215                                                         wfProfileOut( $fname );
2216                                                         return $i;
2217                                                 }
2218                                         }
2219                                         if( $lt === false ) {
2220                                                 // Nothing else interesting to find; abort!
2221                                                 // We're nested, but there's no close tags left. Abort!
2222                                                 break 2;
2223                                         }
2224                                         // Skip ahead to next tag start
2225                                         $i = $lt;
2226                                         $state = MW_COLON_STATE_TAGSTART;
2227                                 }
2228                                 break;
2229                         case 1: // MW_COLON_STATE_TAG:
2230                                 // In a <tag>
2231                                 switch( $c ) {
2232                                 case ">":
2233                                         $stack++;
2234                                         $state = MW_COLON_STATE_TEXT;
2235                                         break;
2236                                 case "/":
2237                                         // Slash may be followed by >?
2238                                         $state = MW_COLON_STATE_TAGSLASH;
2239                                         break;
2240                                 default:
2241                                         // ignore
2242                                 }
2243                                 break;
2244                         case 2: // MW_COLON_STATE_TAGSTART:
2245                                 switch( $c ) {
2246                                 case "/":
2247                                         $state = MW_COLON_STATE_CLOSETAG;
2248                                         break;
2249                                 case "!":
2250                                         $state = MW_COLON_STATE_COMMENT;
2251                                         break;
2252                                 case ">":
2253                                         // Illegal early close? This shouldn't happen D:
2254                                         $state = MW_COLON_STATE_TEXT;
2255                                         break;
2256                                 default:
2257                                         $state = MW_COLON_STATE_TAG;
2258                                 }
2259                                 break;
2260                         case 3: // MW_COLON_STATE_CLOSETAG:
2261                                 // In a </tag>
2262                                 if( $c == ">" ) {
2263                                         $stack--;
2264                                         if( $stack < 0 ) {
2265                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2266                                                 wfProfileOut( $fname );
2267                                                 return false;
2268                                         }
2269                                         $state = MW_COLON_STATE_TEXT;
2270                                 }
2271                                 break;
2272                         case MW_COLON_STATE_TAGSLASH:
2273                                 if( $c == ">" ) {
2274                                         // Yes, a self-closed tag <blah/>
2275                                         $state = MW_COLON_STATE_TEXT;
2276                                 } else {
2277                                         // Probably we're jumping the gun, and this is an attribute
2278                                         $state = MW_COLON_STATE_TAG;
2279                                 }
2280                                 break;
2281                         case 5: // MW_COLON_STATE_COMMENT:
2282                                 if( $c == "-" ) {
2283                                         $state = MW_COLON_STATE_COMMENTDASH;
2284                                 }
2285                                 break;
2286                         case MW_COLON_STATE_COMMENTDASH:
2287                                 if( $c == "-" ) {
2288                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2289                                 } else {
2290                                         $state = MW_COLON_STATE_COMMENT;
2291                                 }
2292                                 break;
2293                         case MW_COLON_STATE_COMMENTDASHDASH:
2294                                 if( $c == ">" ) {
2295                                         $state = MW_COLON_STATE_TEXT;
2296                                 } else {
2297                                         $state = MW_COLON_STATE_COMMENT;
2298                                 }
2299                                 break;
2300                         default:
2301                                 throw new MWException( "State machine error in $fname" );
2302                         }
2303                 }
2304                 if( $stack > 0 ) {
2305                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2306                         return false;
2307                 }
2308                 wfProfileOut( $fname );
2309                 return false;
2310         }
2311
2312         /**
2313          * Return value of a magic variable (like PAGENAME)
2314          *
2315          * @private
2316          */
2317         function getVariableValue( $index ) {
2318                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2319
2320                 /**
2321                  * Some of these require message or data lookups and can be
2322                  * expensive to check many times.
2323                  */
2324                 static $varCache = array();
2325                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2326                         if ( isset( $varCache[$index] ) )
2327                                 return $varCache[$index];
2328
2329                 $ts = time();
2330                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2331
2332                 # Use the time zone
2333                 global $wgLocaltimezone;
2334                 if ( isset( $wgLocaltimezone ) ) {
2335                         $oldtz = getenv( 'TZ' );
2336                         putenv( 'TZ='.$wgLocaltimezone );
2337                 }
2338                 $localTimestamp = date( 'YmdHis', $ts );
2339                 $localMonth = date( 'm', $ts );
2340                 $localMonthName = date( 'n', $ts );
2341                 $localDay = date( 'j', $ts );
2342                 $localDay2 = date( 'd', $ts );
2343                 $localDayOfWeek = date( 'w', $ts );
2344                 $localWeek = date( 'W', $ts );
2345                 $localYear = date( 'Y', $ts );
2346                 $localHour = date( 'H', $ts );
2347                 if ( isset( $wgLocaltimezone ) ) {
2348                         putenv( 'TZ='.$oldtz );
2349                 }
2350
2351                 switch ( $index ) {
2352                         case 'currentmonth':
2353                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2354                         case 'currentmonthname':
2355                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2356                         case 'currentmonthnamegen':
2357                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2358                         case 'currentmonthabbrev':
2359                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2360                         case 'currentday':
2361                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2362                         case 'currentday2':
2363                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2364                         case 'localmonth':
2365                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2366                         case 'localmonthname':
2367                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2368                         case 'localmonthnamegen':
2369                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2370                         case 'localmonthabbrev':
2371                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2372                         case 'localday':
2373                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2374                         case 'localday2':
2375                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2376                         case 'pagename':
2377                                 return $this->mTitle->getText();
2378                         case 'pagenamee':
2379                                 return $this->mTitle->getPartialURL();
2380                         case 'fullpagename':
2381                                 return $this->mTitle->getPrefixedText();
2382                         case 'fullpagenamee':
2383                                 return $this->mTitle->getPrefixedURL();
2384                         case 'subpagename':
2385                                 return $this->mTitle->getSubpageText();
2386                         case 'subpagenamee':
2387                                 return $this->mTitle->getSubpageUrlForm();
2388                         case 'basepagename':
2389                                 return $this->mTitle->getBaseText();
2390                         case 'basepagenamee':
2391                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2392                         case 'talkpagename':
2393                                 if( $this->mTitle->canTalk() ) {
2394                                         $talkPage = $this->mTitle->getTalkPage();
2395                                         return $talkPage->getPrefixedText();
2396                                 } else {
2397                                         return '';
2398                                 }
2399                         case 'talkpagenamee':
2400                                 if( $this->mTitle->canTalk() ) {
2401                                         $talkPage = $this->mTitle->getTalkPage();
2402                                         return $talkPage->getPrefixedUrl();
2403                                 } else {
2404                                         return '';
2405                                 }
2406                         case 'subjectpagename':
2407                                 $subjPage = $this->mTitle->getSubjectPage();
2408                                 return $subjPage->getPrefixedText();
2409                         case 'subjectpagenamee':
2410                                 $subjPage = $this->mTitle->getSubjectPage();
2411                                 return $subjPage->getPrefixedUrl();
2412                         case 'revisionid':
2413                                 return $this->mRevisionId;
2414                         case 'namespace':
2415                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2416                         case 'namespacee':
2417                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2418                         case 'talkspace':
2419                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2420                         case 'talkspacee':
2421                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2422                         case 'subjectspace':
2423                                 return $this->mTitle->getSubjectNsText();
2424                         case 'subjectspacee':
2425                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2426                         case 'currentdayname':
2427                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2428                         case 'currentyear':
2429                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2430                         case 'currenttime':
2431                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2432                         case 'currenthour':
2433                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2434                         case 'currentweek':
2435                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2436                                 // int to remove the padding
2437                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2438                         case 'currentdow':
2439                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2440                         case 'localdayname':
2441                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2442                         case 'localyear':
2443                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2444                         case 'localtime':
2445                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2446                         case 'localhour':
2447                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2448                         case 'localweek':
2449                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2450                                 // int to remove the padding
2451                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2452                         case 'localdow':
2453                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2454                         case 'numberofarticles':
2455                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2456                         case 'numberoffiles':
2457                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2458                         case 'numberofusers':
2459                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2460                         case 'numberofpages':
2461                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2462                         case 'numberofadmins':
2463                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2464                         case 'currenttimestamp':
2465                                 return $varCache[$index] = wfTimestampNow();
2466                         case 'localtimestamp':
2467                                 return $varCache[$index] = $localTimestamp;
2468                         case 'currentversion':
2469                                 return $varCache[$index] = SpecialVersion::getVersion();
2470                         case 'sitename':
2471                                 return $wgSitename;
2472                         case 'server':
2473                                 return $wgServer;
2474                         case 'servername':
2475                                 return $wgServerName;
2476                         case 'scriptpath':
2477                                 return $wgScriptPath;
2478                         case 'directionmark':
2479                                 return $wgContLang->getDirMark();
2480                         case 'contentlanguage':
2481                                 global $wgContLanguageCode;
2482                                 return $wgContLanguageCode;
2483                         default:
2484                                 $ret = null;
2485                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2486                                         return $ret;
2487                                 else
2488                                         return null;
2489                 }
2490         }
2491
2492         /**
2493          * initialise the magic variables (like CURRENTMONTHNAME)
2494          *
2495          * @private
2496          */
2497         function initialiseVariables() {
2498                 $fname = 'Parser::initialiseVariables';
2499                 wfProfileIn( $fname );
2500                 $variableIDs = MagicWord::getVariableIDs();
2501
2502                 $this->mVariables = array();
2503                 foreach ( $variableIDs as $id ) {
2504                         $mw =& MagicWord::get( $id );
2505                         $mw->addToArray( $this->mVariables, $id );
2506                 }
2507                 wfProfileOut( $fname );
2508         }
2509
2510         /**
2511          * parse any parentheses in format ((title|part|part))
2512          * and call callbacks to get a replacement text for any found piece
2513          *
2514          * @param string $text The text to parse
2515          * @param array $callbacks rules in form:
2516          *     '{' => array(                            # opening parentheses
2517          *                                      'end' => '}',   # closing parentheses
2518          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2519          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2520          *                                                                )
2521          *                                      )
2522          *                                      'min' => 2,     # Minimum parenthesis count in cb
2523          *                                      'max' => 3,     # Maximum parenthesis count in cb
2524          * @private
2525          */
2526         function replace_callback ($text, $callbacks) {
2527                 wfProfileIn( __METHOD__ );
2528                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2529                 $lastOpeningBrace = -1;                 # last not closed parentheses
2530
2531                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2532
2533                 $i = 0;
2534                 while ( $i < strlen( $text ) ) {
2535                         # Find next opening brace, closing brace or pipe
2536                         if ( $lastOpeningBrace == -1 ) {
2537                                 $currentClosing = '';
2538                                 $search = $validOpeningBraces;
2539                         } else {
2540                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2541                                 $search = $validOpeningBraces . '|' . $currentClosing;
2542                         }
2543                         $rule = null;
2544                         $i += strcspn( $text, $search, $i );
2545                         if ( $i < strlen( $text ) ) {
2546                                 if ( $text[$i] == '|' ) {
2547                                         $found = 'pipe';
2548                                 } elseif ( $text[$i] == $currentClosing ) {
2549                                         $found = 'close';
2550                                 } else {
2551                                         $found = 'open';
2552                                         $rule = $callbacks[$text[$i]];
2553                                 }
2554                         } else {
2555                                 # All done
2556                                 break;
2557                         }
2558
2559                         if ( $found == 'open' ) {
2560                                 # found opening brace, let's add it to parentheses stack
2561                                 $piece = array('brace' => $text[$i],
2562                                                            'braceEnd' => $rule['end'],
2563                                                            'title' => '',
2564                                                            'parts' => null);
2565
2566                                 # count opening brace characters
2567                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2568                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2569                                 $i += $piece['count'];
2570
2571                                 # we need to add to stack only if opening brace count is enough for one of the rules
2572                                 if ( $piece['count'] >= $rule['min'] ) {
2573                                         $lastOpeningBrace ++;
2574                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2575                                 }
2576                         } elseif ( $found == 'close' ) {
2577                                 # lets check if it is enough characters for closing brace
2578                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2579                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2580
2581                                 # check for maximum matching characters (if there are 5 closing
2582                                 # characters, we will probably need only 3 - depending on the rules)
2583                                 $matchingCount = 0;
2584                                 $matchingCallback = null;
2585                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2586                                 if ( $count > $cbType['max'] ) {
2587                                         # The specified maximum exists in the callback array, unless the caller
2588                                         # has made an error
2589                                         $matchingCount = $cbType['max'];
2590                                 } else {
2591                                         # Count is less than the maximum
2592                                         # Skip any gaps in the callback array to find the true largest match
2593                                         # Need to use array_key_exists not isset because the callback can be null
2594                                         $matchingCount = $count;
2595                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2596                                                 --$matchingCount;
2597                                         }
2598                                 }
2599
2600                                 if ($matchingCount <= 0) {
2601                                         $i += $count;
2602                                         continue;
2603                                 }
2604                                 $matchingCallback = $cbType['cb'][$matchingCount];
2605
2606                                 # let's set a title or last part (if '|' was found)
2607                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2608                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2609                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2610                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2611                                 } else {
2612                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2613                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2614                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2615                                 }
2616
2617                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2618                                 $pieceEnd = $i + $matchingCount;
2619
2620                                 if( is_callable( $matchingCallback ) ) {
2621                                         $cbArgs = array (
2622                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2623                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2624                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2625                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2626                                                                          );
2627                                         # finally we can call a user callback and replace piece of text
2628                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2629                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2630                                         $i = $pieceStart + strlen($replaceWith);
2631                                 } else {
2632                                         # null value for callback means that parentheses should be parsed, but not replaced
2633                                         $i += $matchingCount;
2634                                 }
2635
2636                                 # reset last opening parentheses, but keep it in case there are unused characters
2637                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2638                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2639                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2640                                                            'title' => '',
2641                                                            'parts' => null,
2642                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2643                                 $openingBraceStack[$lastOpeningBrace--] = null;
2644
2645                                 if ($matchingCount < $piece['count']) {
2646                                         $piece['count'] -= $matchingCount;
2647                                         $piece['startAt'] -= $matchingCount;
2648                                         $piece['partStart'] = $piece['startAt'];
2649                                         # do we still qualify for any callback with remaining count?
2650                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2651                                         while ( $piece['count'] ) {
2652                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2653                                                         $lastOpeningBrace++;
2654                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2655                                                         break;
2656                                                 }
2657                                                 --$piece['count'];
2658                                         }
2659                                 }
2660                         } elseif ( $found == 'pipe' ) {
2661                                 # lets set a title if it is a first separator, or next part otherwise
2662                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2663                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2664                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2665                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2666                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2667                                 } else {
2668                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2669                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2670                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2671                                 }
2672                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2673                         }
2674                 }
2675
2676                 wfProfileOut( __METHOD__ );
2677                 return $text;
2678         }
2679
2680         /**
2681          * Replace magic variables, templates, and template arguments
2682          * with the appropriate text. Templates are substituted recursively,
2683          * taking care to avoid infinite loops.
2684          *
2685          * Note that the substitution depends on value of $mOutputType:
2686          *  OT_WIKI: only {{subst:}} templates
2687          *  OT_MSG: only magic variables
2688          *  OT_HTML: all templates and magic variables
2689          *
2690          * @param string $tex The text to transform
2691          * @param array $args Key-value pairs representing template parameters to substitute
2692          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2693          * @private
2694          */
2695         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2696                 # Prevent too big inclusions
2697                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2698                         return $text;
2699                 }
2700
2701                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2702                 wfProfileIn( $fname );
2703
2704                 # This function is called recursively. To keep track of arguments we need a stack:
2705                 array_push( $this->mArgStack, $args );
2706
2707                 $braceCallbacks = array();
2708                 if ( !$argsOnly ) {
2709                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2710                 }
2711                 if ( $this->mOutputType != OT_MSG ) {
2712                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2713                 }
2714                 if ( $braceCallbacks ) {
2715                         $callbacks = array(
2716                                 '{' => array(
2717                                         'end' => '}',
2718                                         'cb' => $braceCallbacks,
2719                                         'min' => $argsOnly ? 3 : 2,
2720                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2721                                 ),
2722                                 '[' => array(
2723                                         'end' => ']',
2724                                         'cb' => array(2=>null),
2725                                         'min' => 2,
2726                                         'max' => 2,
2727                                 )
2728                         );
2729                         $text = $this->replace_callback ($text, $callbacks);
2730
2731                         array_pop( $this->mArgStack );
2732                 }
2733                 wfProfileOut( $fname );
2734                 return $text;
2735         }
2736
2737         /**
2738          * Replace magic variables
2739          * @private
2740          */
2741         function variableSubstitution( $matches ) {
2742                 $fname = 'Parser::variableSubstitution';
2743                 $varname = $matches[1];
2744                 wfProfileIn( $fname );
2745                 $skip = false;
2746                 if ( $this->mOutputType == OT_WIKI ) {
2747                         # Do only magic variables prefixed by SUBST
2748                         $mwSubst =& MagicWord::get( 'subst' );
2749                         if (!$mwSubst->matchStartAndRemove( $varname ))
2750                                 $skip = true;
2751                         # Note that if we don't substitute the variable below,
2752                         # we don't remove the {{subst:}} magic word, in case
2753                         # it is a template rather than a magic variable.
2754                 }
2755                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2756                         $id = $this->mVariables[$varname];
2757                         $text = $this->getVariableValue( $id );
2758                         $this->mOutput->mContainsOldMagic = true;
2759                 } else {
2760                         $text = $matches[0];
2761                 }
2762                 wfProfileOut( $fname );
2763                 return $text;
2764         }
2765
2766         # Split template arguments
2767         function getTemplateArgs( $argsString ) {
2768                 if ( $argsString === '' ) {
2769                         return array();
2770                 }
2771
2772                 $args = explode( '|', substr( $argsString, 1 ) );
2773
2774                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2775                 # merged with the next arg because the '|' character between belongs
2776                 # to the link syntax and not the template parameter syntax.
2777                 $argc = count($args);
2778
2779                 for ( $i = 0; $i < $argc-1; $i++ ) {
2780                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2781                                 $args[$i] .= '|'.$args[$i+1];
2782                                 array_splice($args, $i+1, 1);
2783                                 $i--;
2784                                 $argc--;
2785                         }
2786                 }
2787
2788                 return $args;
2789         }
2790
2791         /**
2792          * Return the text of a template, after recursively
2793          * replacing any variables or templates within the template.
2794          *
2795          * @param array $piece The parts of the template
2796          *  $piece['text']: matched text
2797          *  $piece['title']: the title, i.e. the part before the |
2798          *  $piece['parts']: the parameter array
2799          * @return string the text of the template
2800          * @private
2801          */
2802         function braceSubstitution( $piece ) {
2803                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2804                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2805                 wfProfileIn( $fname );
2806                 wfProfileIn( __METHOD__.'-setup' );
2807
2808                 # Flags
2809                 $found = false;             # $text has been filled
2810                 $nowiki = false;            # wiki markup in $text should be escaped
2811                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2812                 $noargs = false;            # Don't replace triple-brace arguments in $text
2813                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2814                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2815                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2816
2817                 # Title object, where $text came from
2818                 $title = NULL;
2819
2820                 $linestart = '';
2821
2822
2823                 # $part1 is the bit before the first |, and must contain only title characters
2824                 # $args is a list of arguments, starting from index 0, not including $part1
2825
2826                 $titleText = $part1 = $piece['title'];
2827                 # If the third subpattern matched anything, it will start with |
2828
2829                 if (null == $piece['parts']) {
2830                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2831                         if ($replaceWith != $piece['text']) {
2832                                 $text = $replaceWith;
2833                                 $found = true;
2834                                 $noparse = true;
2835                                 $noargs = true;
2836                         }
2837                 }
2838
2839                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2840                 $argc = count( $args );
2841                 wfProfileOut( __METHOD__.'-setup' );
2842
2843                 # SUBST
2844                 wfProfileIn( __METHOD__.'-modifiers' );
2845                 if ( !$found ) {
2846                         $mwSubst =& MagicWord::get( 'subst' );
2847                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2848                                 # One of two possibilities is true:
2849                                 # 1) Found SUBST but not in the PST phase
2850                                 # 2) Didn't find SUBST and in the PST phase
2851                                 # In either case, return without further processing
2852                                 $text = $piece['text'];
2853                                 $found = true;
2854                                 $noparse = true;
2855                                 $noargs = true;
2856                         }
2857                 }
2858
2859                 # MSG, MSGNW, INT and RAW
2860                 if ( !$found ) {
2861                         # Check for MSGNW:
2862                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2863                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2864                                 $nowiki = true;
2865                         } else {
2866                                 # Remove obsolete MSG:
2867                                 $mwMsg =& MagicWord::get( 'msg' );
2868                                 $mwMsg->matchStartAndRemove( $part1 );
2869                         }
2870
2871                         # Check for RAW:
2872                         $mwRaw =& MagicWord::get( 'raw' );
2873                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2874                                 $forceRawInterwiki = true;
2875                         }
2876
2877                         # Check if it is an internal message
2878                         $mwInt =& MagicWord::get( 'int' );
2879                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2880                                 $text = $linestart . wfMsgReal( $part1, $args, true );
2881                                 $found = true;
2882                         }
2883                 }
2884                 wfProfileOut( __METHOD__.'-modifiers' );
2885
2886                 # Parser functions
2887                 if ( !$found ) {
2888                         wfProfileIn( __METHOD__ . '-pfunc' );
2889
2890                         $colonPos = strpos( $part1, ':' );
2891                         if ( $colonPos !== false ) {
2892                                 # Case sensitive functions
2893                                 $function = substr( $part1, 0, $colonPos );
2894                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2895                                         $function = $this->mFunctionSynonyms[1][$function];
2896                                 } else {
2897                                         # Case insensitive functions
2898                                         $function = strtolower( $function );
2899                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2900                                                 $function = $this->mFunctionSynonyms[0][$function];
2901                                         } else {
2902                                                 $function = false;
2903                                         }
2904                                 }
2905                                 if ( $function ) {
2906                                         $funcArgs = array_map( 'trim', $args );
2907                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2908                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2909                                         $found = true;
2910
2911                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2912                                         //$noargs = true;
2913                                         //$noparse = true;
2914
2915                                         if ( is_array( $result ) ) {
2916                                                 if ( isset( $result[0] ) ) {
2917                                                         $text = $linestart . $result[0];
2918                                                         unset( $result[0] );
2919                                                 }
2920
2921                                                 // Extract flags into the local scope
2922                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2923                                                 extract( $result );
2924                                         } else {
2925                                                 $text = $linestart . $result;
2926                                         }
2927                                 }
2928                         }
2929                         wfProfileOut( __METHOD__ . '-pfunc' );
2930                 }
2931
2932                 # Template table test
2933
2934                 # Did we encounter this template already? If yes, it is in the cache
2935                 # and we need to check for loops.
2936                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2937                         $found = true;
2938
2939                         # Infinite loop test
2940                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2941                                 $noparse = true;
2942                                 $noargs = true;
2943                                 $found = true;
2944                                 $text = $linestart .
2945                                         "[[$part1]]<!-- WARNING: template loop detected -->";
2946                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
2947                         } else {
2948                                 # set $text to cached message.
2949                                 $text = $linestart . $this->mTemplates[$piece['title']];
2950                         }
2951                 }
2952
2953                 # Load from database
2954                 $lastPathLevel = $this->mTemplatePath;
2955                 if ( !$found ) {
2956                         wfProfileIn( __METHOD__ . '-loadtpl' );
2957                         $ns = NS_TEMPLATE;
2958                         # declaring $subpage directly in the function call
2959                         # does not work correctly with references and breaks
2960                         # {{/subpage}}-style inclusions
2961                         $subpage = '';
2962                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2963                         if ($subpage !== '') {
2964                                 $ns = $this->mTitle->getNamespace();
2965                         }
2966                         $title = Title::newFromText( $part1, $ns );
2967
2968
2969                         if ( !is_null( $title ) ) {
2970                                 $titleText = $title->getPrefixedText();
2971                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2972                                 # Check for language variants if the template is not found
2973                                 if($checkVariantLink && $title->getArticleID() == 0){
2974                                         $wgContLang->findVariantLink($part1, $title);
2975                                 }
2976
2977                                 if ( !$title->isExternal() ) {
2978                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2979                                                 $text = SpecialPage::capturePath( $title );
2980                                                 if ( is_string( $text ) ) {
2981                                                         $found = true;
2982                                                         $noparse = true;
2983                                                         $noargs = true;
2984                                                         $isHTML = true;
2985                                                         $this->disableCache();
2986                                                 }
2987                                         } else {
2988                                                 $articleContent = $this->fetchTemplate( $title );
2989                                                 if ( $articleContent !== false ) {
2990                                                         $found = true;
2991                                                         $text = $articleContent;
2992                                                         $replaceHeadings = true;
2993                                                 }
2994                                         }
2995
2996                                         # If the title is valid but undisplayable, make a link to it
2997                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2998                                                 $text = "[[$titleText]]";
2999                                                 $found = true;
3000                                         }
3001                                 } elseif ( $title->isTrans() ) {
3002                                         // Interwiki transclusion
3003                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3004                                                 $text = $this->interwikiTransclude( $title, 'render' );
3005                                                 $isHTML = true;
3006                                                 $noparse = true;
3007                                         } else {
3008                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3009                                                 $replaceHeadings = true;
3010                                         }
3011                                         $found = true;
3012                                 }
3013
3014                                 # Template cache array insertion
3015                                 # Use the original $piece['title'] not the mangled $part1, so that
3016                                 # modifiers such as RAW: produce separate cache entries
3017                                 if( $found ) {
3018                                         if( $isHTML ) {
3019                                                 // A special page; don't store it in the template cache.
3020                                         } else {
3021                                                 $this->mTemplates[$piece['title']] = $text;
3022                                         }
3023                                         $text = $linestart . $text;
3024                                 }
3025                         }
3026                         wfProfileOut( __METHOD__ . '-loadtpl' );
3027                 }
3028
3029                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3030                         # Error, oversize inclusion
3031                         $text = $linestart .
3032                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3033                         $noparse = true;
3034                         $noargs = true;
3035                 }
3036
3037                 # Recursive parsing, escaping and link table handling
3038                 # Only for HTML output
3039                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3040                         $text = wfEscapeWikiText( $text );
3041                 } elseif ( !$this->ot['msg'] && $found ) {
3042                         if ( $noargs ) {
3043                                 $assocArgs = array();
3044                         } else {
3045                                 # Clean up argument array
3046                                 $assocArgs = array();
3047                                 $index = 1;
3048                                 foreach( $args as $arg ) {
3049                                         $eqpos = strpos( $arg, '=' );
3050                                         if ( $eqpos === false ) {
3051                                                 $assocArgs[$index++] = $arg;
3052                                         } else {
3053                                                 $name = trim( substr( $arg, 0, $eqpos ) );
3054                                                 $value = trim( substr( $arg, $eqpos+1 ) );
3055                                                 if ( $value === false ) {
3056                                                         $value = '';
3057                                                 }
3058                                                 if ( $name !== false ) {
3059                                                         $assocArgs[$name] = $value;
3060                                                 }
3061                                         }
3062                                 }
3063
3064                                 # Add a new element to the templace recursion path
3065                                 $this->mTemplatePath[$part1] = 1;
3066                         }
3067
3068                         if ( !$noparse ) {
3069                                 # If there are any <onlyinclude> tags, only include them
3070                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3071                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
3072                                         $text = '';
3073                                         foreach ($m[1] as $piece)
3074                                                 $text .= $piece;
3075                                 }
3076                                 # Remove <noinclude> sections and <includeonly> tags
3077                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
3078                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3079
3080                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3081                                         # Strip <nowiki>, <pre>, etc.
3082                                         $text = $this->strip( $text, $this->mStripState );
3083                                         if ( $this->ot['html'] ) {
3084                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3085                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3086                                                 $text = Sanitizer::removeHTMLcomments( $text );
3087                                         }
3088                                 }
3089                                 $text = $this->replaceVariables( $text, $assocArgs );
3090
3091                                 # If the template begins with a table or block-level
3092                                 # element, it should be treated as beginning a new line.
3093                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3094                                         $text = "\n" . $text;
3095                                 }
3096                         } elseif ( !$noargs ) {
3097                                 # $noparse and !$noargs
3098                                 # Just replace the arguments, not any double-brace items
3099                                 # This is used for rendered interwiki transclusion
3100                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3101                         }
3102                 }
3103                 # Prune lower levels off the recursion check path
3104                 $this->mTemplatePath = $lastPathLevel;
3105
3106                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3107                         # Error, oversize inclusion
3108                         $text = $linestart .
3109                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3110                         $noparse = true;
3111                         $noargs = true;
3112                 }
3113
3114                 if ( !$found ) {
3115                         wfProfileOut( $fname );
3116                         return $piece['text'];
3117                 } else {
3118                         wfProfileIn( __METHOD__ . '-placeholders' );
3119                         if ( $isHTML ) {
3120                                 # Replace raw HTML by a placeholder
3121                                 # Add a blank line preceding, to prevent it from mucking up
3122                                 # immediately preceding headings
3123                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3124                         } else {
3125                                 # replace ==section headers==
3126                                 # XXX this needs to go away once we have a better parser.
3127                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3128                                         if( !is_null( $title ) )
3129                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3130                                         else
3131                                                 $encodedname = base64_encode("");
3132                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3133                                                 PREG_SPLIT_DELIM_CAPTURE);
3134                                         $text = '';
3135                                         $nsec = 0;
3136                                         for( $i = 0; $i < count($m); $i += 2 ) {
3137                                                 $text .= $m[$i];
3138                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3139                                                 $hl = $m[$i + 1];
3140                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3141                                                         $text .= $hl;
3142                                                         continue;
3143                                                 }
3144                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3145                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3146                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3147
3148                                                 $nsec++;
3149                                         }
3150                                 }
3151                         }
3152                         wfProfileOut( __METHOD__ . '-placeholders' );
3153                 }
3154
3155                 # Prune lower levels off the recursion check path
3156                 $this->mTemplatePath = $lastPathLevel;
3157
3158                 if ( !$found ) {
3159                         wfProfileOut( $fname );
3160                         return $piece['text'];
3161                 } else {
3162                         wfProfileOut( $fname );
3163                         return $text;
3164                 }
3165         }
3166
3167         /**
3168          * Fetch the unparsed text of a template and register a reference to it.
3169          */
3170         function fetchTemplate( $title ) {
3171                 $text = false;
3172                 // Loop to fetch the article, with up to 1 redirect
3173                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3174                         $rev = Revision::newFromTitle( $title );
3175                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3176                         if ( !$rev ) {
3177                                 break;
3178                         }
3179                         $text = $rev->getText();
3180                         if ( $text === false ) {
3181                                 break;
3182                         }
3183                         // Redirect?
3184                         $title = Title::newFromRedirect( $text );
3185                 }
3186                 return $text;
3187         }
3188
3189         /**
3190          * Transclude an interwiki link.
3191          */
3192         function interwikiTransclude( $title, $action ) {
3193                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3194
3195                 if (!$wgEnableScaryTranscluding)
3196                         return wfMsg('scarytranscludedisabled');
3197
3198                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3199                 // But we'll handle it generally anyway
3200                 if ( $title->getNamespace() ) {
3201                         // Use the canonical namespace, which should work anywhere
3202                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3203                 } else {
3204                         $articleName = $title->getDBkey();
3205                 }
3206
3207                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3208                 $url .= "?action=$action";
3209                 if (strlen($url) > 255)
3210                         return wfMsg('scarytranscludetoolong');
3211                 return $this->fetchScaryTemplateMaybeFromCache($url);
3212         }
3213
3214         function fetchScaryTemplateMaybeFromCache($url) {
3215                 global $wgTranscludeCacheExpiry;
3216                 $dbr =& wfGetDB(DB_SLAVE);
3217                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3218                                 array('tc_url' => $url));
3219                 if ($obj) {
3220                         $time = $obj->tc_time;
3221                         $text = $obj->tc_contents;
3222                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3223                                 return $text;
3224                         }
3225                 }
3226
3227                 $text = Http::get($url);
3228                 if (!$text)
3229                         return wfMsg('scarytranscludefailed', $url);
3230
3231                 $dbw =& wfGetDB(DB_MASTER);
3232                 $dbw->replace('transcache', array('tc_url'), array(
3233                         'tc_url' => $url,
3234                         'tc_time' => time(),
3235                         'tc_contents' => $text));
3236                 return $text;
3237         }
3238
3239
3240         /**
3241          * Triple brace replacement -- used for template arguments
3242          * @private
3243          */
3244         function argSubstitution( $matches ) {
3245                 $arg = trim( $matches['title'] );
3246                 $text = $matches['text'];
3247                 $inputArgs = end( $this->mArgStack );
3248
3249                 if ( array_key_exists( $arg, $inputArgs ) ) {
3250                         $text = $inputArgs[$arg];
3251                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3252                 null != $matches['parts'] && count($matches['parts']) > 0) {
3253                         $text = $matches['parts'][0];
3254                 }
3255                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3256                         $text = $matches['text'] .
3257                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3258                 }
3259
3260                 return $text;
3261         }
3262
3263         /**
3264          * Increment an include size counter
3265          *
3266          * @param string $type The type of expansion
3267          * @param integer $size The size of the text
3268          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3269          */
3270         function incrementIncludeSize( $type, $size ) {
3271                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3272                         return false;
3273                 } else {
3274                         $this->mIncludeSizes[$type] += $size;
3275                         return true;
3276                 }
3277         }
3278
3279         /**
3280          * Detect __NOGALLERY__ magic word and set a placeholder
3281          */
3282         function stripNoGallery( &$text ) {
3283                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3284                 # do not add TOC
3285                 $mw = MagicWord::get( 'nogallery' );
3286                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3287         }
3288
3289         /**
3290          * Detect __TOC__ magic word and set a placeholder
3291          */
3292         function stripToc( $text ) {
3293                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3294                 # do not add TOC
3295                 $mw = MagicWord::get( 'notoc' );
3296                 if( $mw->matchAndRemove( $text ) ) {
3297                         $this->mShowToc = false;
3298                 }
3299
3300                 $mw = MagicWord::get( 'toc' );
3301                 if( $mw->match( $text ) ) {
3302                         $this->mShowToc = true;
3303                         $this->mForceTocPosition = true;
3304
3305                         // Set a placeholder. At the end we'll fill it in with the TOC.
3306                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3307
3308                         // Only keep the first one.
3309                         $text = $mw->replace( '', $text );
3310                 }
3311                 return $text;
3312         }
3313
3314         /**
3315          * This function accomplishes several tasks:
3316          * 1) Auto-number headings if that option is enabled
3317          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3318          * 3) Add a Table of contents on the top for users who have enabled the option
3319          * 4) Auto-anchor headings
3320          *
3321          * It loops through all headlines, collects the necessary data, then splits up the
3322          * string and re-inserts the newly formatted headlines.
3323          *
3324          * @param string $text
3325          * @param boolean $isMain
3326          * @private
3327          */
3328         function formatHeadings( $text, $isMain=true ) {
3329                 global $wgMaxTocLevel, $wgContLang;
3330
3331                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3332                 if( !$this->mTitle->userCanEdit() ) {
3333                         $showEditLink = 0;
3334                 } else {
3335                         $showEditLink = $this->mOptions->getEditSection();
3336                 }
3337
3338                 # Inhibit editsection links if requested in the page
3339                 $esw =& MagicWord::get( 'noeditsection' );
3340                 if( $esw->matchAndRemove( $text ) ) {
3341                         $showEditLink = 0;
3342                 }
3343
3344                 # Get all headlines for numbering them and adding funky stuff like [edit]
3345                 # links - this is for later, but we need the number of headlines right now
3346                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3347
3348                 # if there are fewer than 4 headlines in the article, do not show TOC
3349                 # unless it's been explicitly enabled.
3350                 $enoughToc = $this->mShowToc &&
3351                         (($numMatches >= 4) || $this->mForceTocPosition);
3352
3353                 # Allow user to stipulate that a page should have a "new section"
3354                 # link added via __NEWSECTIONLINK__
3355                 $mw =& MagicWord::get( 'newsectionlink' );
3356                 if( $mw->matchAndRemove( $text ) )
3357                         $this->mOutput->setNewSection( true );
3358
3359                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3360                 # override above conditions and always show TOC above first header
3361                 $mw =& MagicWord::get( 'forcetoc' );
3362                 if ($mw->matchAndRemove( $text ) ) {
3363                         $this->mShowToc = true;
3364                         $enoughToc = true;
3365                 }
3366
3367                 # Never ever show TOC if no headers
3368                 if( $numMatches < 1 ) {
3369                         $enoughToc = false;
3370                 }
3371
3372                 # We need this to perform operations on the HTML
3373                 $sk =& $this->mOptions->getSkin();
3374
3375                 # headline counter
3376                 $headlineCount = 0;
3377                 $sectionCount = 0; # headlineCount excluding template sections
3378
3379                 # Ugh .. the TOC should have neat indentation levels which can be
3380                 # passed to the skin functions. These are determined here
3381                 $toc = '';
3382                 $full = '';
3383                 $head = array();
3384                 $sublevelCount = array();
3385                 $levelCount = array();
3386                 $toclevel = 0;
3387                 $level = 0;
3388                 $prevlevel = 0;
3389                 $toclevel = 0;
3390                 $prevtoclevel = 0;
3391
3392                 foreach( $matches[3] as $headline ) {
3393                         $istemplate = 0;
3394                         $templatetitle = '';
3395                         $templatesection = 0;
3396                         $numbering = '';
3397
3398                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3399                                 $istemplate = 1;
3400                                 $templatetitle = base64_decode($mat[1]);
3401                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3402                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3403                         }
3404
3405                         if( $toclevel ) {
3406                                 $prevlevel = $level;
3407                                 $prevtoclevel = $toclevel;
3408                         }
3409                         $level = $matches[1][$headlineCount];
3410
3411                         if( $doNumberHeadings || $enoughToc ) {
3412
3413                                 if ( $level > $prevlevel ) {
3414                                         # Increase TOC level
3415                                         $toclevel++;
3416                                         $sublevelCount[$toclevel] = 0;
3417                                         if( $toclevel<$wgMaxTocLevel ) {
3418                                                 $toc .= $sk->tocIndent();
3419                                         }
3420                                 }
3421                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3422                                         # Decrease TOC level, find level to jump to
3423
3424                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3425                                                 # Can only go down to level 1
3426                                                 $toclevel = 1;
3427                                         } else {
3428                                                 for ($i = $toclevel; $i > 0; $i--) {
3429                                                         if ( $levelCount[$i] == $level ) {
3430                                                                 # Found last matching level
3431                                                                 $toclevel = $i;
3432                                                                 break;
3433                                                         }
3434                                                         elseif ( $levelCount[$i] < $level ) {
3435                                                                 # Found first matching level below current level
3436                                                                 $toclevel = $i + 1;
3437                                                                 break;
3438                                                         }
3439                                                 }
3440                                         }
3441                                         if( $toclevel<$wgMaxTocLevel ) {
3442                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3443                                         }
3444                                 }
3445                                 else {
3446                                         # No change in level, end TOC line
3447                                         if( $toclevel<$wgMaxTocLevel ) {
3448                                                 $toc .= $sk->tocLineEnd();
3449                                         }
3450                                 }
3451
3452                                 $levelCount[$toclevel] = $level;
3453
3454                                 # count number of headlines for each level
3455                                 @$sublevelCount[$toclevel]++;
3456                                 $dot = 0;
3457                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3458                                         if( !empty( $sublevelCount[$i] ) ) {
3459                                                 if( $dot ) {
3460                                                         $numbering .= '.';
3461                                                 }
3462                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3463                                                 $dot = 1;
3464                                         }
3465                                 }
3466                         }
3467
3468                         # The canonized header is a version of the header text safe to use for links
3469                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3470                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3471                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3472
3473                         # Remove link placeholders by the link text.
3474                         #     <!--LINK number-->
3475                         # turns into
3476                         #     link text with suffix
3477                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3478                                                             "\$this->mLinkHolders['texts'][\$1]",
3479                                                             $canonized_headline );
3480                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3481                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3482                                                             $canonized_headline );
3483
3484                         # strip out HTML
3485                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3486                         $tocline = trim( $canonized_headline );
3487                         # Save headline for section edit hint before it's escaped
3488                         $headline_hint = trim( $canonized_headline );
3489                         $canonized_headline = Sanitizer::escapeId( $tocline );
3490                         $refers[$headlineCount] = $canonized_headline;
3491
3492                         # count how many in assoc. array so we can track dupes in anchors
3493                         @$refers[$canonized_headline]++;
3494                         $refcount[$headlineCount]=$refers[$canonized_headline];
3495
3496                         # Don't number the heading if it is the only one (looks silly)
3497                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3498                                 # the two are different if the line contains a link
3499                                 $headline=$numbering . ' ' . $headline;
3500                         }
3501
3502                         # Create the anchor for linking from the TOC to the section
3503                         $anchor = $canonized_headline;
3504                         if($refcount[$headlineCount] > 1 ) {
3505                                 $anchor .= '_' . $refcount[$headlineCount];
3506                         }
3507                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3508                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3509                         }
3510                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3511                                 if ( empty( $head[$headlineCount] ) ) {
3512                                         $head[$headlineCount] = '';
3513                                 }
3514                                 if( $istemplate )
3515                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3516                                 else
3517                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3518                         }
3519
3520                         # give headline the correct <h#> tag
3521                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3522
3523                         $headlineCount++;
3524                         if( !$istemplate )
3525                                 $sectionCount++;
3526                 }
3527
3528                 if( $enoughToc ) {
3529                         if( $toclevel<$wgMaxTocLevel ) {
3530                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3531                         }
3532                         $toc = $sk->tocList( $toc );
3533                 }
3534
3535                 # split up and insert constructed headlines
3536
3537                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3538                 $i = 0;
3539
3540                 foreach( $blocks as $block ) {
3541                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3542                                 # This is the [edit] link that appears for the top block of text when
3543                                 # section editing is enabled
3544
3545                                 # Disabled because it broke block formatting
3546                                 # For example, a bullet point in the top line
3547                                 # $full .= $sk->editSectionLink(0);
3548                         }
3549                         $full .= $block;
3550                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3551                                 # Top anchor now in skin
3552                                 $full = $full.$toc;
3553                         }
3554
3555                         if( !empty( $head[$i] ) ) {
3556                                 $full .= $head[$i];
3557                         }
3558                         $i++;
3559                 }
3560                 if( $this->mForceTocPosition ) {
3561                         return str_replace( '<!--MWTOC-->', $toc, $full );
3562                 } else {
3563                         return $full;
3564                 }
3565         }
3566
3567         /**
3568          * Transform wiki markup when saving a page by doing \r\n -> \n
3569          * conversion, substitting signatures, {{subst:}} templates, etc.
3570          *
3571          * @param string $text the text to transform
3572          * @param Title &$title the Title object for the current article
3573          * @param User &$user the User object describing the current user
3574          * @param ParserOptions $options parsing options
3575          * @param bool $clearState whether to clear the parser state first
3576          * @return string the altered wiki markup
3577          * @public
3578          */
3579         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3580                 $this->mOptions = $options;
3581                 $this->mTitle =& $title;
3582                 $this->setOutputType( OT_WIKI );
3583
3584                 if ( $clearState ) {
3585                         $this->clearState();
3586                 }
3587
3588                 $stripState = false;
3589                 $pairs = array(
3590                         "\r\n" => "\n",
3591                 );
3592                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3593                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3594                 $text = $this->pstPass2( $text, $stripState, $user );
3595                 $text = $this->unstrip( $text, $stripState );
3596                 $text = $this->unstripNoWiki( $text, $stripState );
3597                 return $text;
3598         }
3599
3600         /**
3601          * Pre-save transform helper function
3602          * @private
3603          */
3604         function pstPass2( $text, &$stripState, &$user ) {
3605                 global $wgContLang, $wgLocaltimezone;
3606
3607                 /* Note: This is the timestamp saved as hardcoded wikitext to
3608                  * the database, we use $wgContLang here in order to give
3609                  * everyone the same signature and use the default one rather
3610                  * than the one selected in each user's preferences.
3611                  */
3612                 if ( isset( $wgLocaltimezone ) ) {
3613                         $oldtz = getenv( 'TZ' );
3614                         putenv( 'TZ='.$wgLocaltimezone );
3615                 }
3616                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3617                   ' (' . date( 'T' ) . ')';
3618                 if ( isset( $wgLocaltimezone ) ) {
3619                         putenv( 'TZ='.$oldtz );
3620                 }
3621
3622                 # Variable replacement
3623                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3624                 $text = $this->replaceVariables( $text );
3625
3626                 # Strip out <nowiki> etc. added via replaceVariables
3627                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3628
3629                 # Signatures
3630                 $sigText = $this->getUserSig( $user );
3631                 $text = strtr( $text, array(
3632                         '~~~~~' => $d,
3633                         '~~~~' => "$sigText $d",
3634                         '~~~' => $sigText
3635                 ) );
3636
3637                 # Context links: [[|name]] and [[name (context)|]]
3638                 #
3639                 global $wgLegalTitleChars;
3640                 $tc = "[$wgLegalTitleChars]";
3641
3642                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3643                 $conpat = "/^{$tc}+?( \\({$tc}+\\)|)$/";
3644
3645                 $p1 = "/\[\[(:?$namespacechar+:|:|)({$tc}+?)( \\({$tc}+\\)|)\\|]]/";    # [[ns:page (context)|]]
3646                 $p2 = "/\[\[\\|({$tc}+)]]/";                                            # [[|page]]
3647
3648                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3649
3650                 $t = $this->mTitle->getText();
3651                 if ( preg_match( $conpat, $t, $m ) && '' != $m[1] ) {
3652                         $text = preg_replace( $p2, "[[\\1{$m[1]}|\\1]]", $text );
3653                 } else {
3654                         # if $m[1] is empty, don't bother duplicating the title
3655                         $text = preg_replace( $p2, '[[\\1]]', $text );
3656                 }
3657
3658                 # Trim trailing whitespace
3659                 # __END__ tag allows for trailing
3660                 # whitespace to be deliberately included
3661                 $text = rtrim( $text );
3662                 $mw =& MagicWord::get( 'end' );
3663                 $mw->matchAndRemove( $text );
3664
3665                 return $text;
3666         }
3667
3668         /**
3669          * Fetch the user's signature text, if any, and normalize to
3670          * validated, ready-to-insert wikitext.
3671          *
3672          * @param User $user
3673          * @return string
3674          * @private
3675          */
3676         function getUserSig( &$user ) {
3677                 $username = $user->getName();
3678                 $nickname = $user->getOption( 'nickname' );
3679                 $nickname = $nickname === '' ? $username : $nickname;
3680
3681                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3682                         # Sig. might contain markup; validate this
3683                         if( $this->validateSig( $nickname ) !== false ) {
3684                                 # Validated; clean up (if needed) and return it
3685                                 return $this->cleanSig( $nickname, true );
3686                         } else {
3687                                 # Failed to validate; fall back to the default
3688                                 $nickname = $username;
3689                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3690                         }
3691                 }
3692
3693                 // Make sure nickname doesnt get a sig in a sig
3694                 $nickname = $this->cleanSigInSig( $nickname );
3695
3696                 # If we're still here, make it a link to the user page
3697                 $userpage = $user->getUserPage();
3698                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3699         }
3700
3701         /**
3702          * Check that the user's signature contains no bad XML
3703          *
3704          * @param string $text
3705          * @return mixed An expanded string, or false if invalid.
3706          */
3707         function validateSig( $text ) {
3708                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3709         }
3710
3711         /**
3712          * Clean up signature text
3713          *
3714          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3715          * 2) Substitute all transclusions
3716          *
3717          * @param string $text
3718          * @param $parsing Whether we're cleaning (preferences save) or parsing
3719          * @return string Signature text
3720          */
3721         function cleanSig( $text, $parsing = false ) {
3722                 global $wgTitle;
3723                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3724
3725                 $substWord = MagicWord::get( 'subst' );
3726                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3727                 $substText = '{{' . $substWord->getSynonym( 0 );
3728
3729                 $text = preg_replace( $substRegex, $substText, $text );
3730                 $text = $this->cleanSigInSig( $text );
3731                 $text = $this->replaceVariables( $text );
3732
3733                 $this->clearState();
3734                 return $text;
3735         }
3736
3737         /**
3738          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3739          * @param string $text
3740          * @return string Signature text with /~{3,5}/ removed
3741          */
3742         function cleanSigInSig( $text ) {
3743                 $text = preg_replace( '/~{3,5}/', '', $text );
3744                 return $text;
3745         }
3746
3747         /**
3748          * Set up some variables which are usually set up in parse()
3749          * so that an external function can call some class members with confidence
3750          * @public
3751          */
3752         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3753                 $this->mTitle =& $title;
3754                 $this->mOptions = $options;
3755                 $this->setOutputType( $outputType );
3756                 if ( $clearState ) {
3757                         $this->clearState();
3758                 }
3759         }
3760
3761         /**
3762          * Transform a MediaWiki message by replacing magic variables.
3763          *
3764          * @param string $text the text to transform
3765          * @param ParserOptions $options  options
3766          * @return string the text with variables substituted
3767          * @public
3768          */
3769         function transformMsg( $text, $options ) {
3770                 global $wgTitle;
3771                 static $executing = false;
3772
3773                 $fname = "Parser::transformMsg";
3774
3775                 # Guard against infinite recursion
3776                 if ( $executing ) {
3777                         return $text;
3778                 }
3779                 $executing = true;
3780
3781                 wfProfileIn($fname);
3782
3783                 $this->mTitle = $wgTitle;
3784                 $this->mOptions = $options;
3785                 $this->setOutputType( OT_MSG );
3786                 $this->clearState();
3787                 $text = $this->replaceVariables( $text );
3788
3789                 $executing = false;
3790                 wfProfileOut($fname);
3791                 return $text;
3792         }
3793
3794         /**
3795          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3796          * The callback should have the following form:
3797          *    function myParserHook( $text, $params, &$parser ) { ... }
3798          *
3799          * Transform and return $text. Use $parser for any required context, e.g. use
3800          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3801          *
3802          * @public
3803          *
3804          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3805          * @param mixed $callback The callback function (and object) to use for the tag
3806          *
3807          * @return The old value of the mTagHooks array associated with the hook
3808          */
3809         function setHook( $tag, $callback ) {
3810                 $tag = strtolower( $tag );
3811                 $oldVal = @$this->mTagHooks[$tag];
3812                 $this->mTagHooks[$tag] = $callback;
3813
3814                 return $oldVal;
3815         }
3816
3817         /**
3818          * Create a function, e.g. {{sum:1|2|3}}
3819          * The callback function should have the form:
3820          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3821          *
3822          * The callback may either return the text result of the function, or an array with the text
3823          * in element 0, and a number of flags in the other elements. The names of the flags are
3824          * specified in the keys. Valid flags are:
3825          *   found                     The text returned is valid, stop processing the template. This
3826          *                             is on by default.
3827          *   nowiki                    Wiki markup in the return value should be escaped
3828          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3829          *   noargs                    Don't replace triple-brace arguments in the return value
3830          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3831          *
3832          * @public
3833          *
3834          * @param string $id The magic word ID
3835          * @param mixed $callback The callback function (and object) to use
3836          * @param integer $flags a combination of the following flags:
3837          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3838          *
3839          * @return The old callback function for this name, if any
3840          */
3841         function setFunctionHook( $id, $callback, $flags = 0 ) {
3842                 $oldVal = @$this->mFunctionHooks[$id];
3843                 $this->mFunctionHooks[$id] = $callback;
3844
3845                 # Add to function cache
3846                 $mw = MagicWord::get( $id );
3847                 if( !$mw )
3848                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3849
3850                 $synonyms = $mw->getSynonyms();
3851                 $sensitive = intval( $mw->isCaseSensitive() );
3852
3853                 foreach ( $synonyms as $syn ) {
3854                         # Case
3855                         if ( !$sensitive ) {
3856                                 $syn = strtolower( $syn );
3857                         }
3858                         # Add leading hash
3859                         if ( !( $flags & SFH_NO_HASH ) ) {
3860                                 $syn = '#' . $syn;
3861                         }
3862                         # Remove trailing colon
3863                         if ( substr( $syn, -1, 1 ) == ':' ) {
3864                                 $syn = substr( $syn, 0, -1 );
3865                         }
3866                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3867                 }
3868                 return $oldVal;
3869         }
3870
3871         /**
3872          * Get all registered function hook identifiers
3873          *
3874          * @return array
3875          */
3876         function getFunctionHooks() {
3877                 return array_keys( $this->mFunctionHooks );
3878         }
3879
3880         /**
3881          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3882          * Placeholders created in Skin::makeLinkObj()
3883          * Returns an array of links found, indexed by PDBK:
3884          *  0 - broken
3885          *  1 - normal link
3886          *  2 - stub
3887          * $options is a bit field, RLH_FOR_UPDATE to select for update
3888          */
3889         function replaceLinkHolders( &$text, $options = 0 ) {
3890                 global $wgUser;
3891                 global $wgOutputReplace;
3892                 global $wgContLang, $wgLanguageCode;
3893
3894                 $fname = 'Parser::replaceLinkHolders';
3895                 wfProfileIn( $fname );
3896
3897                 $pdbks = array();
3898                 $colours = array();
3899                 $sk =& $this->mOptions->getSkin();
3900                 $linkCache =& LinkCache::singleton();
3901
3902                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3903                         wfProfileIn( $fname.'-check' );
3904                         $dbr =& wfGetDB( DB_SLAVE );
3905                         $page = $dbr->tableName( 'page' );
3906                         $threshold = $wgUser->getOption('stubthreshold');
3907
3908                         # Sort by namespace
3909                         asort( $this->mLinkHolders['namespaces'] );
3910
3911                         # Generate query
3912                         $query = false;
3913                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3914                                 # Make title object
3915                                 $title = $this->mLinkHolders['titles'][$key];
3916
3917                                 # Skip invalid entries.
3918                                 # Result will be ugly, but prevents crash.
3919                                 if ( is_null( $title ) ) {
3920                                         continue;
3921                                 }
3922                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3923
3924                                 # Check if it's a static known link, e.g. interwiki
3925                                 if ( $title->isAlwaysKnown() ) {
3926                                         $colours[$pdbk] = 1;
3927                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3928                                         $colours[$pdbk] = 1;
3929                                         $this->mOutput->addLink( $title, $id );
3930                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3931                                         $colours[$pdbk] = 0;
3932                                 } else {
3933                                         # Not in the link cache, add it to the query
3934                                         if ( !isset( $current ) ) {
3935                                                 $current = $ns;
3936                                                 $query =  "SELECT page_id, page_namespace, page_title";
3937                                                 if ( $threshold > 0 ) {
3938                                                         $query .= ', page_len, page_is_redirect';
3939                                                 }
3940                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3941                                         } elseif ( $current != $ns ) {
3942                                                 $current = $ns;
3943                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3944                                         } else {
3945                                                 $query .= ', ';
3946                                         }
3947
3948                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3949                                 }
3950                         }
3951                         if ( $query ) {
3952                                 $query .= '))';
3953                                 if ( $options & RLH_FOR_UPDATE ) {
3954                                         $query .= ' FOR UPDATE';
3955                                 }
3956
3957                                 $res = $dbr->query( $query, $fname );
3958
3959                                 # Fetch data and form into an associative array
3960                                 # non-existent = broken
3961                                 # 1 = known
3962                                 # 2 = stub
3963                                 while ( $s = $dbr->fetchObject($res) ) {
3964                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3965                                         $pdbk = $title->getPrefixedDBkey();
3966                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3967                                         $this->mOutput->addLink( $title, $s->page_id );
3968
3969                                         if ( $threshold >  0 ) {
3970                                                 $size = $s->page_len;
3971                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3972                                                         $colours[$pdbk] = 1;
3973                                                 } else {
3974                                                         $colours[$pdbk] = 2;
3975                                                 }
3976                                         } else {
3977                                                 $colours[$pdbk] = 1;
3978                                         }
3979                                 }
3980                         }
3981                         wfProfileOut( $fname.'-check' );
3982
3983                         # Do a second query for different language variants of links (if needed)
3984                         if($wgContLang->hasVariants()){
3985                                 $linkBatch = new LinkBatch();
3986                                 $variantMap = array(); // maps $pdbkey_Variant => $pdbkey_original
3987
3988                                 // Add variants of links to link batch
3989                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3990                                         $title = $this->mLinkHolders['titles'][$key];
3991                                         if ( is_null( $title ) )
3992                                                 continue;
3993
3994                                         $pdbk = $title->getPrefixedDBkey();
3995
3996                                         // generate all variants of the link title text
3997                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
3998
3999                                         // if link was not found (in first query), add all variants to query
4000                                         if ( !isset($colours[$pdbk]) ){
4001                                                 foreach($allTextVariants as $textVariant){
4002                                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
4003                                                         if(is_null($variantTitle)) continue;
4004                                                         $linkBatch->addObj( $variantTitle );
4005                                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4006                                                 }
4007                                         }
4008                                 }
4009
4010
4011                                 if(!$linkBatch->isEmpty()){
4012                                         // construct query
4013                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4014
4015                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4016                                         if ( $threshold > 0 ) {
4017                                                 $variantQuery .= ', page_len, page_is_redirect';
4018                                         }
4019
4020                                         $variantQuery .= " FROM $page WHERE $titleClause";
4021                                         if ( $options & RLH_FOR_UPDATE ) {
4022                                                 $variantQuery .= ' FOR UPDATE';
4023                                         }
4024
4025                                         $varRes = $dbr->query( $variantQuery, $fname );
4026
4027                                         // for each found variants, figure out link holders and replace
4028                                         while ( $s = $dbr->fetchObject($varRes) ) {
4029
4030                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4031                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4032                                                 $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4033                                                 $this->mOutput->addLink( $variantTitle, $s->page_id );
4034
4035                                                 $holderKeys = $variantMap[$varPdbk];
4036
4037                                                 // loop over link holders
4038                                                 foreach($holderKeys as $key){
4039                                                         $title = $this->mLinkHolders['titles'][$key];
4040                                                         if ( is_null( $title ) ) continue;
4041
4042                                                         $pdbk = $title->getPrefixedDBkey();
4043
4044                                                         if(!isset($colours[$pdbk])){
4045                                                                 // found link in some of the variants, replace the link holder data
4046                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4047                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4048
4049                                                                 // set pdbk and colour
4050                                                                 $pdbks[$key] = $varPdbk;
4051                                                                 if ( $threshold >  0 ) {
4052                                                                         $size = $s->page_len;
4053                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4054                                                                                 $colours[$varPdbk] = 1;
4055                                                                         } else {
4056                                                                                 $colours[$varPdbk] = 2;
4057                                                                         }
4058                                                                 }
4059                                                                 else {
4060                                                                         $colours[$varPdbk] = 1;
4061                                                                 }
4062                                                         }
4063                                                 }
4064                                         }
4065                                 }
4066                         }
4067
4068                         # Construct search and replace arrays
4069                         wfProfileIn( $fname.'-construct' );
4070                         $wgOutputReplace = array();
4071                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4072                                 $pdbk = $pdbks[$key];
4073                                 $searchkey = "<!--LINK $key-->";
4074                                 $title = $this->mLinkHolders['titles'][$key];
4075                                 if ( empty( $colours[$pdbk] ) ) {
4076                                         $linkCache->addBadLinkObj( $title );
4077                                         $colours[$pdbk] = 0;
4078                                         $this->mOutput->addLink( $title, 0 );
4079                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
4080                                                                         $this->mLinkHolders['texts'][$key],
4081                                                                         $this->mLinkHolders['queries'][$key] );
4082                                 } elseif ( $colours[$pdbk] == 1 ) {
4083                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
4084                                                                         $this->mLinkHolders['texts'][$key],
4085                                                                         $this->mLinkHolders['queries'][$key] );
4086                                 } elseif ( $colours[$pdbk] == 2 ) {
4087                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
4088                                                                         $this->mLinkHolders['texts'][$key],
4089                                                                         $this->mLinkHolders['queries'][$key] );
4090                                 }
4091                         }
4092                         wfProfileOut( $fname.'-construct' );
4093
4094                         # Do the thing
4095                         wfProfileIn( $fname.'-replace' );
4096
4097                         $text = preg_replace_callback(
4098                                 '/(<!--LINK .*?-->)/',
4099                                 "wfOutputReplaceMatches",
4100                                 $text);
4101
4102                         wfProfileOut( $fname.'-replace' );
4103                 }
4104
4105                 # Now process interwiki link holders
4106                 # This is quite a bit simpler than internal links
4107                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4108                         wfProfileIn( $fname.'-interwiki' );
4109                         # Make interwiki link HTML
4110                         $wgOutputReplace = array();
4111                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4112                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4113                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4114                         }
4115
4116                         $text = preg_replace_callback(
4117                                 '/<!--IWLINK (.*?)-->/',
4118                                 "wfOutputReplaceMatches",
4119                                 $text );
4120                         wfProfileOut( $fname.'-interwiki' );
4121                 }
4122
4123                 wfProfileOut( $fname );
4124                 return $colours;
4125         }
4126
4127         /**
4128          * Replace <!--LINK--> link placeholders with plain text of links
4129          * (not HTML-formatted).
4130          * @param string $text
4131          * @return string
4132          */
4133         function replaceLinkHoldersText( $text ) {
4134                 $fname = 'Parser::replaceLinkHoldersText';
4135                 wfProfileIn( $fname );
4136
4137                 $text = preg_replace_callback(
4138                         '/<!--(LINK|IWLINK) (.*?)-->/',
4139                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4140                         $text );
4141
4142                 wfProfileOut( $fname );
4143                 return $text;
4144         }
4145
4146         /**
4147          * @param array $matches
4148          * @return string
4149          * @private
4150          */
4151         function replaceLinkHoldersTextCallback( $matches ) {
4152                 $type = $matches[1];
4153                 $key  = $matches[2];
4154                 if( $type == 'LINK' ) {
4155                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4156                                 return $this->mLinkHolders['texts'][$key];
4157                         }
4158                 } elseif( $type == 'IWLINK' ) {
4159                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4160                                 return $this->mInterwikiLinkHolders['texts'][$key];
4161                         }
4162                 }
4163                 return $matches[0];
4164         }
4165
4166         /**
4167          * Tag hook handler for 'pre'.
4168          */
4169         function renderPreTag( $text, $attribs, $parser ) {
4170                 // Backwards-compatibility hack
4171                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4172
4173                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4174                 return wfOpenElement( 'pre', $attribs ) .
4175                         wfEscapeHTMLTagsOnly( $content ) .
4176                         '</pre>';
4177         }
4178
4179         /**
4180          * Renders an image gallery from a text with one line per image.
4181          * text labels may be given by using |-style alternative text. E.g.
4182          *   Image:one.jpg|The number "1"
4183          *   Image:tree.jpg|A tree
4184          * given as text will return the HTML of a gallery with two images,
4185          * labeled 'The number "1"' and
4186          * 'A tree'.
4187          */
4188         function renderImageGallery( $text, $params ) {
4189                 $ig = new ImageGallery();
4190                 $ig->setShowBytes( false );
4191                 $ig->setShowFilename( false );
4192                 $ig->setParsing();
4193                 $ig->useSkin( $this->mOptions->getSkin() );
4194
4195                 if( isset( $params['caption'] ) )
4196                         $ig->setCaption( $params['caption'] );
4197
4198                 $lines = explode( "\n", $text );
4199                 foreach ( $lines as $line ) {
4200                         # match lines like these:
4201                         # Image:someimage.jpg|This is some image
4202                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4203                         # Skip empty lines
4204                         if ( count( $matches ) == 0 ) {
4205                                 continue;
4206                         }
4207                         $tp = Title::newFromText( $matches[1] );
4208                         $nt =& $tp;
4209                         if( is_null( $nt ) ) {
4210                                 # Bogus title. Ignore these so we don't bomb out later.
4211                                 continue;
4212                         }
4213                         if ( isset( $matches[3] ) ) {
4214                                 $label = $matches[3];
4215                         } else {
4216                                 $label = '';
4217                         }
4218
4219                         $pout = $this->parse( $label,
4220                                 $this->mTitle,
4221                                 $this->mOptions,
4222                                 false, // Strip whitespace...?
4223                                 false  // Don't clear state!
4224                         );
4225                         $html = $pout->getText();
4226
4227                         $ig->add( new Image( $nt ), $html );
4228
4229                         # Only add real images (bug #5586)
4230                         if ( $nt->getNamespace() == NS_IMAGE ) {
4231                                 $this->mOutput->addImage( $nt->getDBkey() );
4232                         }
4233                 }
4234                 return $ig->toHTML();
4235         }
4236
4237         /**
4238          * Parse image options text and use it to make an image
4239          */
4240         function makeImage( &$nt, $options ) {
4241                 global $wgUseImageResize, $wgDjvuRenderer;
4242
4243                 $align = '';
4244
4245                 # Check if the options text is of the form "options|alt text"
4246                 # Options are:
4247                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4248                 #  * left               no resizing, just left align. label is used for alt= only
4249                 #  * right              same, but right aligned
4250                 #  * none               same, but not aligned
4251                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4252                 #  * center             center the image
4253                 #  * framed             Keep original image size, no magnify-button.
4254
4255                 $part = explode( '|', $options);
4256
4257                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4258                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4259                 $mwLeft   =& MagicWord::get( 'img_left' );
4260                 $mwRight  =& MagicWord::get( 'img_right' );
4261                 $mwNone   =& MagicWord::get( 'img_none' );
4262                 $mwWidth  =& MagicWord::get( 'img_width' );
4263                 $mwCenter =& MagicWord::get( 'img_center' );
4264                 $mwFramed =& MagicWord::get( 'img_framed' );
4265                 $mwPage   =& MagicWord::get( 'img_page' );
4266                 $caption = '';
4267
4268                 $width = $height = $framed = $thumb = false;
4269                 $page = null;
4270                 $manual_thumb = '' ;
4271
4272                 foreach( $part as $key => $val ) {
4273                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4274                                 $thumb=true;
4275                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4276                                 # use manually specified thumbnail
4277                                 $thumb=true;
4278                                 $manual_thumb = $match;
4279                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4280                                 # remember to set an alignment, don't render immediately
4281                                 $align = 'right';
4282                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4283                                 # remember to set an alignment, don't render immediately
4284                                 $align = 'left';
4285                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4286                                 # remember to set an alignment, don't render immediately
4287                                 $align = 'center';
4288                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4289                                 # remember to set an alignment, don't render immediately
4290                                 $align = 'none';
4291                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4292                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4293                                 # Select a page in a multipage document
4294                                 $page = $match;
4295                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4296                                 wfDebug( "img_width match: $match\n" );
4297                                 # $match is the image width in pixels
4298                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4299                                         $width = intval( $m[1] );
4300                                         $height = intval( $m[2] );
4301                                 } else {
4302                                         $width = intval($match);
4303                                 }
4304                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4305                                 $framed=true;
4306                         } else {
4307                                 $caption = $val;
4308                         }
4309                 }
4310                 # Strip bad stuff out of the alt text
4311                 $alt = $this->replaceLinkHoldersText( $caption );
4312
4313                 # make sure there are no placeholders in thumbnail attributes
4314                 # that are later expanded to html- so expand them now and
4315                 # remove the tags
4316                 $alt = $this->unstrip($alt, $this->mStripState);
4317                 $alt = Sanitizer::stripAllTags( $alt );
4318
4319                 # Linker does the rest
4320                 $sk =& $this->mOptions->getSkin();
4321                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4322         }
4323
4324         /**
4325          * Set a flag in the output object indicating that the content is dynamic and
4326          * shouldn't be cached.
4327          */
4328         function disableCache() {
4329                 wfDebug( "Parser output marked as uncacheable.\n" );
4330                 $this->mOutput->mCacheTime = -1;
4331         }
4332
4333         /**#@+
4334          * Callback from the Sanitizer for expanding items found in HTML attribute
4335          * values, so they can be safely tested and escaped.
4336          * @param string $text
4337          * @param array $args
4338          * @return string
4339          * @private
4340          */
4341         function attributeStripCallback( &$text, $args ) {
4342                 $text = $this->replaceVariables( $text, $args );
4343                 $text = $this->unstripForHTML( $text );
4344                 return $text;
4345         }
4346
4347         function unstripForHTML( $text ) {
4348                 $text = $this->unstrip( $text, $this->mStripState );
4349                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4350                 return $text;
4351         }
4352         /**#@-*/
4353
4354         /**#@+
4355          * Accessor/mutator
4356          */
4357         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4358         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4359         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4360         /**#@-*/
4361
4362         /**#@+
4363          * Accessor
4364          */
4365         function getTags() { return array_keys( $this->mTagHooks ); }
4366         /**#@-*/
4367
4368
4369         /**
4370          * Break wikitext input into sections, and either pull or replace
4371          * some particular section's text.
4372          *
4373          * External callers should use the getSection and replaceSection methods.
4374          *
4375          * @param $text Page wikitext
4376          * @param $section Numbered section. 0 pulls the text before the first
4377          *                 heading; other numbers will pull the given section
4378          *                 along with its lower-level subsections.
4379          * @param $mode One of "get" or "replace"
4380          * @param $newtext Replacement text for section data.
4381          * @return string for "get", the extracted section text.
4382          *                for "replace", the whole page with the section replaced.
4383          */
4384         private function extractSections( $text, $section, $mode, $newtext='' ) {
4385                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4386                 # comments to be stripped as well)
4387                 $striparray = array();
4388
4389                 $oldOutputType = $this->mOutputType;
4390                 $oldOptions = $this->mOptions;
4391                 $this->mOptions = new ParserOptions();
4392                 $this->setOutputType( OT_WIKI );
4393
4394                 $striptext = $this->strip( $text, $striparray, true );
4395
4396                 $this->setOutputType( $oldOutputType );
4397                 $this->mOptions = $oldOptions;
4398
4399                 # now that we can be sure that no pseudo-sections are in the source,
4400                 # split it up by section
4401                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4402                 $comment = "(?:$uniq-!--.*?QINU)";
4403                 $secs = preg_split(
4404                 /*
4405                         "/
4406                         ^(
4407                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4408                         (?:
4409                                 (=+) # Should this be limited to 6?
4410                                 .+?  # Section title...
4411                                 \\2  # Ending = count must match start
4412                         |
4413                                 ^
4414                                 <h([1-6])\b.*?>
4415                                 .*?
4416                                 <\/h\\3\s*>
4417                         )
4418                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4419                         )$
4420                         /mix",
4421                 */
4422                         "/
4423                         (
4424                                 ^
4425                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4426                                 (=+) # Should this be limited to 6?
4427                                 .+?  # Section title...
4428                                 \\2  # Ending = count must match start
4429                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4430                                 $
4431                         |
4432                                 <h([1-6])\b.*?>
4433                                 .*?
4434                                 <\/h\\3\s*>
4435                         )
4436                         /mix",
4437                         $striptext, -1,
4438                         PREG_SPLIT_DELIM_CAPTURE);
4439
4440                 if( $mode == "get" ) {
4441                         if( $section == 0 ) {
4442                                 // "Section 0" returns the content before any other section.
4443                                 $rv = $secs[0];
4444                         } else {
4445                                 $rv = "";
4446                         }
4447                 } elseif( $mode == "replace" ) {
4448                         if( $section == 0 ) {
4449                                 $rv = $newtext . "\n\n";
4450                                 $remainder = true;
4451                         } else {
4452                                 $rv = $secs[0];
4453                                 $remainder = false;
4454                         }
4455                 }
4456                 $count = 0;
4457                 $sectionLevel = 0;
4458                 for( $index = 1; $index < count( $secs ); ) {
4459                         $headerLine = $secs[$index++];
4460                         if( $secs[$index] ) {
4461                                 // A wiki header
4462                                 $headerLevel = strlen( $secs[$index++] );
4463                         } else {
4464                                 // An HTML header
4465                                 $index++;
4466                                 $headerLevel = intval( $secs[$index++] );
4467                         }
4468                         $content = $secs[$index++];
4469
4470                         $count++;
4471                         if( $mode == "get" ) {
4472                                 if( $count == $section ) {
4473                                         $rv = $headerLine . $content;
4474                                         $sectionLevel = $headerLevel;
4475                                 } elseif( $count > $section ) {
4476                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4477                                                 $rv .= $headerLine . $content;
4478                                         } else {
4479                                                 // Broke out to a higher-level section
4480                                                 break;
4481                                         }
4482                                 }
4483                         } elseif( $mode == "replace" ) {
4484                                 if( $count < $section ) {
4485                                         $rv .= $headerLine . $content;
4486                                 } elseif( $count == $section ) {
4487                                         $rv .= $newtext . "\n\n";
4488                                         $sectionLevel = $headerLevel;
4489                                 } elseif( $count > $section ) {
4490                                         if( $headerLevel <= $sectionLevel ) {
4491                                                 // Passed the section's sub-parts.
4492                                                 $remainder = true;
4493                                         }
4494                                         if( $remainder ) {
4495                                                 $rv .= $headerLine . $content;
4496                                         }
4497                                 }
4498                         }
4499                 }
4500                 # reinsert stripped tags
4501                 $rv = $this->unstrip( $rv, $striparray );
4502                 $rv = $this->unstripNoWiki( $rv, $striparray );
4503                 $rv = trim( $rv );
4504                 return $rv;
4505         }
4506
4507         /**
4508          * This function returns the text of a section, specified by a number ($section).
4509          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4510          * the first section before any such heading (section 0).
4511          *
4512          * If a section contains subsections, these are also returned.
4513          *
4514          * @param $text String: text to look in
4515          * @param $section Integer: section number
4516          * @return string text of the requested section
4517          */
4518         function getSection( $text, $section ) {
4519                 return $this->extractSections( $text, $section, "get" );
4520         }
4521
4522         function replaceSection( $oldtext, $section, $text ) {
4523                 return $this->extractSections( $oldtext, $section, "replace", $text );
4524         }
4525
4526 }
4527
4528 /**
4529  * @todo document
4530  * @package MediaWiki
4531  */
4532 class ParserOutput
4533 {
4534         var $mText,             # The output text
4535                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4536                 $mCategories,       # Map of category names to sort keys
4537                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4538                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4539                 $mVersion,          # Compatibility check
4540                 $mTitleText,        # title text of the chosen language variant
4541                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4542                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4543                 $mImages,           # DB keys of the images used, in the array key only
4544                 $mExternalLinks,    # External link URLs, in the key only
4545                 $mHTMLtitle,            # Display HTML title
4546                 $mSubtitle,                     # Additional subtitle
4547                 $mNewSection,           # Show a new section link?
4548                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4549
4550         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4551                 $containsOldMagic = false, $titletext = '' )
4552         {
4553                 $this->mText = $text;
4554                 $this->mLanguageLinks = $languageLinks;
4555                 $this->mCategories = $categoryLinks;
4556                 $this->mContainsOldMagic = $containsOldMagic;
4557                 $this->mCacheTime = '';
4558                 $this->mVersion = MW_PARSER_VERSION;
4559                 $this->mTitleText = $titletext;
4560                 $this->mLinks = array();
4561                 $this->mTemplates = array();
4562                 $this->mImages = array();
4563                 $this->mExternalLinks = array();
4564                 $this->mHTMLtitle = "" ;
4565                 $this->mSubtitle = "" ;
4566                 $this->mNewSection = false;
4567                 $this->mNoGallery = false;
4568         }
4569
4570         function getText()                   { return $this->mText; }
4571         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4572         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4573         function &getCategories()            { return $this->mCategories; }
4574         function getCacheTime()              { return $this->mCacheTime; }
4575         function getTitleText()              { return $this->mTitleText; }
4576         function &getLinks()                 { return $this->mLinks; }
4577         function &getTemplates()             { return $this->mTemplates; }
4578         function &getImages()                { return $this->mImages; }
4579         function &getExternalLinks()         { return $this->mExternalLinks; }
4580         function getNoGallery()              { return $this->mNoGallery; }
4581         function getSubtitle()               { return $this->mSubtitle; }
4582
4583         function containsOldMagic()          { return $this->mContainsOldMagic; }
4584         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4585         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4586         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4587         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4588         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4589         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4590         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4591
4592         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4593         function addImage( $name )           { $this->mImages[$name] = 1; }
4594         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4595         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4596
4597         function setNewSection( $value ) {
4598                 $this->mNewSection = (bool)$value;
4599         }
4600         function getNewSection() {
4601                 return (bool)$this->mNewSection;
4602         }
4603
4604         function addLink( $title, $id = null ) {
4605                 $ns = $title->getNamespace();
4606                 $dbk = $title->getDBkey();
4607                 if ( !isset( $this->mLinks[$ns] ) ) {
4608                         $this->mLinks[$ns] = array();
4609                 }
4610                 if ( is_null( $id ) ) {
4611                         $id = $title->getArticleID();
4612                 }
4613                 $this->mLinks[$ns][$dbk] = $id;
4614         }
4615
4616         function addTemplate( $title, $id ) {
4617                 $ns = $title->getNamespace();
4618                 $dbk = $title->getDBkey();
4619                 if ( !isset( $this->mTemplates[$ns] ) ) {
4620                         $this->mTemplates[$ns] = array();
4621                 }
4622                 $this->mTemplates[$ns][$dbk] = $id;
4623         }
4624
4625         /**
4626          * Return true if this cached output object predates the global or
4627          * per-article cache invalidation timestamps, or if it comes from
4628          * an incompatible older version.
4629          *
4630          * @param string $touched the affected article's last touched timestamp
4631          * @return bool
4632          * @public
4633          */
4634         function expired( $touched ) {
4635                 global $wgCacheEpoch;
4636                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4637                        $this->getCacheTime() < $touched ||
4638                        $this->getCacheTime() <= $wgCacheEpoch ||
4639                        !isset( $this->mVersion ) ||
4640                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4641         }
4642 }
4643
4644 /**
4645  * Set options of the Parser
4646  * @todo document
4647  * @package MediaWiki
4648  */
4649 class ParserOptions
4650 {
4651         # All variables are private
4652         var $mUseTeX;                    # Use texvc to expand <math> tags
4653         var $mUseDynamicDates;           # Use DateFormatter to format dates
4654         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4655         var $mAllowExternalImages;       # Allow external images inline
4656         var $mAllowExternalImagesFrom;   # If not, any exception?
4657         var $mSkin;                      # Reference to the preferred skin
4658         var $mDateFormat;                # Date format index
4659         var $mEditSection;               # Create "edit section" links
4660         var $mNumberHeadings;            # Automatically number headings
4661         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4662         var $mTidy;                      # Ask for tidy cleanup
4663         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4664         var $mMaxIncludeSize;            # Maximum size of template expansions, in bytes
4665         var $mRemoveComments;            # Remove HTML comments. ONLY APPLIES TO PREPROCESS OPERATIONS
4666
4667         var $mUser;                      # Stored user object, just used to initialise the skin
4668
4669         function getUseTeX()                        { return $this->mUseTeX; }
4670         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4671         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4672         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4673         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4674         function getEditSection()                   { return $this->mEditSection; }
4675         function getNumberHeadings()                { return $this->mNumberHeadings; }
4676         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4677         function getTidy()                          { return $this->mTidy; }
4678         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4679         function getMaxIncludeSize()                { return $this->mMaxIncludeSize; }
4680         function getRemoveComments()                { return $this->mRemoveComments; }
4681
4682         function &getSkin() {
4683                 if ( !isset( $this->mSkin ) ) {
4684                         $this->mSkin = $this->mUser->getSkin();
4685                 }
4686                 return $this->mSkin;
4687         }
4688
4689         function getDateFormat() {
4690                 if ( !isset( $this->mDateFormat ) ) {
4691                         $this->mDateFormat = $this->mUser->getDatePreference();
4692                 }
4693                 return $this->mDateFormat;
4694         }
4695
4696         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4697         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4698         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4699         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4700         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4701         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4702         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4703         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4704         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4705         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4706         function setSkin( &$x ) { $this->mSkin =& $x; }
4707         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4708         function setMaxIncludeSize( $x )            { return wfSetVar( $this->mMaxIncludeSize, $x ); }
4709         function setRemoveComments( $x )            { return wfSetVar( $this->mRemoveComments, $x ); }
4710
4711         function ParserOptions( $user = null ) {
4712                 $this->initialiseFromUser( $user );
4713         }
4714
4715         /**
4716          * Get parser options
4717          * @static
4718          */
4719         static function newFromUser( $user ) {
4720                 return new ParserOptions( $user );
4721         }
4722
4723         /** Get user options */
4724         function initialiseFromUser( $userInput ) {
4725                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4726                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
4727                 $fname = 'ParserOptions::initialiseFromUser';
4728                 wfProfileIn( $fname );
4729                 if ( !$userInput ) {
4730                         global $wgUser;
4731                         if ( isset( $wgUser ) ) {
4732                                 $user = $wgUser;
4733                         } else {
4734                                 $user = new User;
4735                                 $user->setLoaded( true );
4736                         }
4737                 } else {
4738                         $user =& $userInput;
4739                 }
4740
4741                 $this->mUser = $user;
4742
4743                 $this->mUseTeX = $wgUseTeX;
4744                 $this->mUseDynamicDates = $wgUseDynamicDates;
4745                 $this->mInterwikiMagic = $wgInterwikiMagic;
4746                 $this->mAllowExternalImages = $wgAllowExternalImages;
4747                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4748                 $this->mSkin = null; # Deferred
4749                 $this->mDateFormat = null; # Deferred
4750                 $this->mEditSection = true;
4751                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4752                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4753                 $this->mTidy = false;
4754                 $this->mInterfaceMessage = false;
4755                 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
4756                 $this->mRemoveComments = true;
4757                 wfProfileOut( $fname );
4758         }
4759 }
4760
4761 /**
4762  * Callback function used by Parser::replaceLinkHolders()
4763  * to substitute link placeholders.
4764  */
4765 function &wfOutputReplaceMatches( $matches ) {
4766         global $wgOutputReplace;
4767         return $wgOutputReplace[$matches[1]];
4768 }
4769
4770 /**
4771  * Return the total number of articles
4772  */
4773 function wfNumberOfArticles() {
4774         global $wgNumberOfArticles;
4775
4776         wfLoadSiteStats();
4777         return $wgNumberOfArticles;
4778 }
4779
4780 /**
4781  * Return the number of files
4782  */
4783 function wfNumberOfFiles() {
4784         $fname = 'wfNumberOfFiles';
4785
4786         wfProfileIn( $fname );
4787         $dbr =& wfGetDB( DB_SLAVE );
4788         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4789         wfProfileOut( $fname );
4790
4791         return $numImages;
4792 }
4793
4794 /**
4795  * Return the number of user accounts
4796  * @return integer
4797  */
4798 function wfNumberOfUsers() {
4799         wfProfileIn( 'wfNumberOfUsers' );
4800         $dbr =& wfGetDB( DB_SLAVE );
4801         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4802         wfProfileOut( 'wfNumberOfUsers' );
4803         return (int)$count;
4804 }
4805
4806 /**
4807  * Return the total number of pages
4808  * @return integer
4809  */
4810 function wfNumberOfPages() {
4811         wfProfileIn( 'wfNumberOfPages' );
4812         $dbr =& wfGetDB( DB_SLAVE );
4813         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4814         wfProfileOut( 'wfNumberOfPages' );
4815         return (int)$count;
4816 }
4817
4818 /**
4819  * Return the total number of admins
4820  *
4821  * @return integer
4822  */
4823 function wfNumberOfAdmins() {
4824         static $admins = -1;
4825         wfProfileIn( 'wfNumberOfAdmins' );
4826         if( $admins == -1 ) {
4827                 $dbr =& wfGetDB( DB_SLAVE );
4828                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4829         }
4830         wfProfileOut( 'wfNumberOfAdmins' );
4831         return (int)$admins;
4832 }
4833
4834 /**
4835  * Count the number of pages in a particular namespace
4836  *
4837  * @param $ns Namespace
4838  * @return integer
4839  */
4840 function wfPagesInNs( $ns ) {
4841         static $pageCount = array();
4842         wfProfileIn( 'wfPagesInNs' );
4843         if( !isset( $pageCount[$ns] ) ) {
4844                 $dbr =& wfGetDB( DB_SLAVE );
4845                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4846         }
4847         wfProfileOut( 'wfPagesInNs' );
4848         return (int)$pageCount[$ns];
4849 }
4850
4851 /**
4852  * Get various statistics from the database
4853  * @private
4854  */
4855 function wfLoadSiteStats() {
4856         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4857         $fname = 'wfLoadSiteStats';
4858
4859         if ( -1 != $wgNumberOfArticles ) return;
4860         $dbr =& wfGetDB( DB_SLAVE );
4861         $s = $dbr->selectRow( 'site_stats',
4862                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4863                 array( 'ss_row_id' => 1 ), $fname
4864         );
4865
4866         if ( $s === false ) {
4867                 return;
4868         } else {
4869                 $wgTotalViews = $s->ss_total_views;
4870                 $wgTotalEdits = $s->ss_total_edits;
4871                 $wgNumberOfArticles = $s->ss_good_articles;
4872         }
4873 }
4874
4875 /**
4876  * Escape html tags
4877  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4878  *
4879  * @param $in String: text that might contain HTML tags.
4880  * @return string Escaped string
4881  */
4882 function wfEscapeHTMLTagsOnly( $in ) {
4883         return str_replace(
4884                 array( '"', '>', '<' ),
4885                 array( '&quot;', '&gt;', '&lt;' ),
4886                 $in );
4887 }
4888
4889 ?>