includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are four main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * transformMsg()
  24  *   performs brace substitution on MediaWiki messages
  25  * preprocess()
  26  *   removes HTML comments and expands templates
  27  *
  28  * Globals used:
  29  *    objects:   $wgLang, $wgContLang
  30  *
  31  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  32  *
  33  * settings:
  34  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  35  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  36  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  37  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.4';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  63
  64         // State constants for the definition list colon extraction
  65         const COLON_STATE_TEXT = 0;
  66         const COLON_STATE_TAG = 1;
  67         const COLON_STATE_TAGSTART = 2;
  68         const COLON_STATE_CLOSETAG = 3;
  69         const COLON_STATE_TAGSLASH = 4;
  70         const COLON_STATE_COMMENT = 5;
  71         const COLON_STATE_COMMENTDASH = 6;
  72         const COLON_STATE_COMMENTDASHDASH = 7;
  73
  74         // Flags for preprocessToDom
  75         const PTD_FOR_INCLUSION = 1;
  76
  77         /**#@+
  78          * @private
  79          */
  80         # Persistent:
  81         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  82                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  83                 $mExtLinkBracketedRegex;
  84
  85         # Cleared with clearState():
  86         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  87         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  88         var $mInterwikiLinkHolders, $mLinkHolders;
  89         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  90         var $mTplExpandCache,// empty-frame expansion cache
  91             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  92                                 // in this path. Used for loop detection.
  93         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  94
  95         # Temporary
  96         # These are variables reset at least once per parse regardless of $clearState
  97         var $mOptions,      // ParserOptions object
  98                 $mTitle,        // Title context, used for self-link rendering and similar things
  99                 $mOutputType,   // Output type, one of the OT_xxx constants
 100                 $ot,            // Shortcut alias, see setOutputType()
 101                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 102                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 103                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 104
 105         /**#@-*/
 106
 107         /**
 108          * Constructor
 109          *
 110          * @public
 111          */
 112         function __construct( $conf = array() ) {
 113                 $this->mTagHooks = array();
 114                 $this->mTransparentTagHooks = array();
 115                 $this->mFunctionHooks = array();
 116                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 117                 $this->mStripList = array( 'nowiki', 'gallery' );
 118                 $this->mMarkerSuffix = "-QINU\x7f";
 119                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 120                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 121                 $this->mFirstCall = true;
 122         }
 123
 124         /**
 125          * Do various kinds of initialisation on the first call of the parser
 126          */
 127         function firstCallInit() {
 128                 if ( !$this->mFirstCall ) {
 129                         return;
 130                 }
 131
 132                 wfProfileIn( __METHOD__ );
 133                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 134
 135                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 136
 137                 # Syntax for arguments (see self::setFunctionHook):
 138                 #  "name for lookup in localized magic words array",
 139                 #  function callback,
 140                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 141                 #    instead of {{#int:...}})
 142                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 167                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 169
 170                 if ( $wgAllowDisplayTitle ) {
 171                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 172                 }
 173                 if ( $wgAllowSlowParserFunctions ) {
 174                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 175                 }
 176
 177                 $this->initialiseVariables();
 178                 $this->mFirstCall = false;
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = false;
 199                 $this->mInPre = false;
 200                 $this->mInterwikiLinkHolders = array(
 201                         'texts' => array(),
 202                         'titles' => array()
 203                 );
 204                 $this->mLinkHolders = array(
 205                         'namespaces' => array(),
 206                         'dbkeys' => array(),
 207                         'queries' => array(),
 208                         'texts' => array(),
 209                         'titles' => array()
 210                 );
 211                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 212
 213                 /**
 214                  * Prefix for temporary replacement strings for the multipass parser.
 215                  * \x07 should never appear in input as it's disallowed in XML.
 216                  * Using it at the front also gives us a little extra robustness
 217                  * since it shouldn't match when butted up against identifier-like
 218                  * string constructs.
 219                  *
 220                  * Must not consist of all title characters, or else it will change
 221                  * the behaviour of <nowiki> in a link.
 222                  */
 223                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 224                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 225
 226                 # Clear these on every parse, bug 4549
 227                 $this->mTemplatePath = array();
 228                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 229
 230                 $this->mShowToc = true;
 231                 $this->mForceTocPosition = false;
 232                 $this->mIncludeSizes = array(
 233                         'post-expand' => 0,
 234                         'arg' => 0,
 235                 );
 236                 $this->mPPNodeCount = 0;
 237                 $this->mDefaultSort = false;
 238                 $this->mHeadings = array();
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == OT_HTML,
 249                         'wiki' => $ot == OT_WIKI,
 250                         'msg' => $ot == OT_MSG,
 251                         'pre' => $ot == OT_PREPROCESS,
 252                 );
 253         }
 254
 255         /**
 256          * Accessor for mUniqPrefix.
 257          *
 258          * @public
 259          */
 260         function uniqPrefix() {
 261                 if( !isset( $this->mUniqPrefix ) ) {
 262                         // @fixme this is probably *horribly wrong*
 263                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 264                         // if this is called for a parser cache hit, the parser may not
 265                         // have ever been initialized in the first place.
 266                         // Not really sure what the heck is supposed to be going on here.
 267                         return '';
 268                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 269                 }
 270                 return $this->mUniqPrefix;
 271         }
 272
 273         /**
 274          * Convert wikitext to HTML
 275          * Do not call this function recursively.
 276          *
 277          * @param string $text Text we want to parse
 278          * @param Title &$title A title object
 279          * @param array $options
 280          * @param boolean $linestart
 281          * @param boolean $clearState
 282          * @param int $revid number to pass in {{REVISIONID}}
 283          * @return ParserOutput a ParserOutput
 284          */
 285         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 286                 /**
 287                  * First pass--just handle <nowiki> sections, pass the rest off
 288                  * to internalParse() which does all the real work.
 289                  */
 290
 291                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 292                 $fname = 'Parser::parse-' . wfGetCaller();
 293                 wfProfileIn( __METHOD__ );
 294                 wfProfileIn( $fname );
 295
 296                 if ( $clearState ) {
 297                         $this->clearState();
 298                 }
 299
 300                 $this->mOptions = $options;
 301                 $this->mTitle =& $title;
 302                 $oldRevisionId = $this->mRevisionId;
 303                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 304                 if( $revid !== null ) {
 305                         $this->mRevisionId = $revid;
 306                         $this->mRevisionTimestamp = null;
 307                 }
 308                 $this->setOutputType( OT_HTML );
 309                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 310                 # No more strip!
 311                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 312                 $text = $this->internalParse( $text );
 313                 $text = $this->mStripState->unstripGeneral( $text );
 314
 315                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 316                 $fixtags = array(
 317                         # french spaces, last one Guillemet-left
 318                         # only if there is something before the space
 319                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 320                         # french spaces, Guillemet-right
 321                         '/(\\302\\253) /' => '\\1&nbsp;',
 322                 );
 323                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 324
 325                 # only once and last
 326                 $text = $this->doBlockLevels( $text, $linestart );
 327
 328                 $this->replaceLinkHolders( $text );
 329
 330                 # the position of the parserConvert() call should not be changed. it
 331                 # assumes that the links are all replaced and the only thing left
 332                 # is the <nowiki> mark.
 333                 # Side-effects: this calls $this->mOutput->setTitleText()
 334                 $text = $wgContLang->parserConvert( $text, $this );
 335
 336                 $text = $this->mStripState->unstripNoWiki( $text );
 337
 338                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 339
 340 //!JF Move to its own function
 341
 342                 $uniq_prefix = $this->mUniqPrefix;
 343                 $matches = array();
 344                 $elements = array_keys( $this->mTransparentTagHooks );
 345                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 346
 347                 foreach( $matches as $marker => $data ) {
 348                         list( $element, $content, $params, $tag ) = $data;
 349                         $tagName = strtolower( $element );
 350                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 351                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 352                                         array( $content, $params, $this ) );
 353                         } else {
 354                                 $output = $tag;
 355                         }
 356                         $this->mStripState->general->setPair( $marker, $output );
 357                 }
 358                 $text = $this->mStripState->unstripGeneral( $text );
 359
 360                 $text = Sanitizer::normalizeCharReferences( $text );
 361
 362                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 363                         $text = Parser::tidy($text);
 364                 } else {
 365                         # attempt to sanitize at least some nesting problems
 366                         # (bug #2702 and quite a few others)
 367                         $tidyregs = array(
 368                                 # ''Something [http://www.cool.com cool''] -->
 369                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 370                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 371                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 372                                 # fix up an anchor inside another anchor, only
 373                                 # at least for a single single nested link (bug 3695)
 374                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 375                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 376                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 377                                 # contains a div, so fix it up here; replace
 378                                 # div with escaped text
 379                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 380                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 381                                 # remove empty italic or bold tag pairs, some
 382                                 # introduced by rules above
 383                                 '/<([bi])><\/\\1>/' => '',
 384                         );
 385
 386                         $text = preg_replace(
 387                                 array_keys( $tidyregs ),
 388                                 array_values( $tidyregs ),
 389                                 $text );
 390                 }
 391
 392                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 393
 394                 # Information on include size limits, for the benefit of users who try to skirt them
 395                 if ( $this->mOptions->getEnableLimitReport() ) {
 396                         $max = $this->mOptions->getMaxIncludeSize();
 397                         $limitReport =
 398                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 399                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 400                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 401                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 402                         $text .= "\n<!-- \n$limitReport-->\n";
 403                 }
 404                 $this->mOutput->setText( $text );
 405                 $this->mRevisionId = $oldRevisionId;
 406                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 407                 wfProfileOut( $fname );
 408                 wfProfileOut( __METHOD__ );
 409
 410                 return $this->mOutput;
 411         }
 412
 413         /**
 414          * Recursive parser entry point that can be called from an extension tag
 415          * hook.
 416          */
 417         function recursiveTagParse( $text ) {
 418                 wfProfileIn( __METHOD__ );
 419                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 420                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 421                 $text = $this->internalParse( $text );
 422                 wfProfileOut( __METHOD__ );
 423                 return $text;
 424         }
 425
 426         /**
 427          * Expand templates and variables in the text, producing valid, static wikitext.
 428          * Also removes comments.
 429          */
 430         function preprocess( $text, $title, $options, $revid = null ) {
 431                 wfProfileIn( __METHOD__ );
 432                 $this->clearState();
 433                 $this->setOutputType( OT_PREPROCESS );
 434                 $this->mOptions = $options;
 435                 $this->mTitle = $title;
 436                 if( $revid !== null ) {
 437                         $this->mRevisionId = $revid;
 438                 }
 439                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 440                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 $text = $this->replaceVariables( $text );
 442                 if ( $this->mOptions->getRemoveComments() ) {
 443                         $text = Sanitizer::removeHTMLcomments( $text );
 444                 }
 445                 $text = $this->mStripState->unstripBoth( $text );
 446                 wfProfileOut( __METHOD__ );
 447                 return $text;
 448         }
 449
 450         /**
 451          * Get a random string
 452          *
 453          * @private
 454          * @static
 455          */
 456         function getRandomString() {
 457                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 458         }
 459
 460         function &getTitle() { return $this->mTitle; }
 461         function getOptions() { return $this->mOptions; }
 462
 463         function getFunctionLang() {
 464                 global $wgLang, $wgContLang;
 465                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 466         }
 467
 468         /**
 469          * Replaces all occurrences of HTML-style comments and the given tags
 470          * in the text with a random marker and returns teh next text. The output
 471          * parameter $matches will be an associative array filled with data in
 472          * the form:
 473          *   'UNIQ-xxxxx' => array(
 474          *     'element',
 475          *     'tag content',
 476          *     array( 'param' => 'x' ),
 477          *     '<element param="x">tag content</element>' ) )
 478          *
 479          * @param $elements list of element names. Comments are always extracted.
 480          * @param $text Source text string.
 481          * @param $uniq_prefix
 482          *
 483          * @public
 484          * @static
 485          */
 486         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 487                 static $n = 1;
 488                 $stripped = '';
 489                 $matches = array();
 490
 491                 $taglist = implode( '|', $elements );
 492                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 493
 494                 while ( '' != $text ) {
 495                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 496                         $stripped .= $p[0];
 497                         if( count( $p ) < 5 ) {
 498                                 break;
 499                         }
 500                         if( count( $p ) > 5 ) {
 501                                 // comment
 502                                 $element    = $p[4];
 503                                 $attributes = '';
 504                                 $close      = '';
 505                                 $inside     = $p[5];
 506                         } else {
 507                                 // tag
 508                                 $element    = $p[1];
 509                                 $attributes = $p[2];
 510                                 $close      = $p[3];
 511                                 $inside     = $p[4];
 512                         }
 513
 514                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 515                         $stripped .= $marker;
 516
 517                         if ( $close === '/>' ) {
 518                                 // Empty element tag, <tag />
 519                                 $content = null;
 520                                 $text = $inside;
 521                                 $tail = null;
 522                         } else {
 523                                 if( $element == '!--' ) {
 524                                         $end = '/(-->)/';
 525                                 } else {
 526                                         $end = "/(<\\/$element\\s*>)/i";
 527                                 }
 528                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 529                                 $content = $q[0];
 530                                 if( count( $q ) < 3 ) {
 531                                         # No end tag -- let it run out to the end of the text.
 532                                         $tail = '';
 533                                         $text = '';
 534                                 } else {
 535                                         $tail = $q[1];
 536                                         $text = $q[2];
 537                                 }
 538                         }
 539
 540                         $matches[$marker] = array( $element,
 541                                 $content,
 542                                 Sanitizer::decodeTagAttributes( $attributes ),
 543                                 "<$element$attributes$close$content$tail" );
 544                 }
 545                 return $stripped;
 546         }
 547
 548         /**
 549          * Get a list of strippable XML-like elements
 550          */
 551         function getStripList() {
 552                 global $wgRawHtml;
 553                 $elements = $this->mStripList;
 554                 if( $wgRawHtml ) {
 555                         $elements[] = 'html';
 556                 }
 557                 if( $this->mOptions->getUseTeX() ) {
 558                         $elements[] = 'math';
 559                 }
 560                 return $elements;
 561         }
 562
 563         /**
 564          * @deprecated use replaceVariables
 565          */
 566         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 567                 return $text;
 568         }
 569
 570         /**
 571          * Restores pre, math, and other extensions removed by strip()
 572          *
 573          * always call unstripNoWiki() after this one
 574          * @private
 575          * @deprecated use $this->mStripState->unstrip()
 576          */
 577         function unstrip( $text, $state ) {
 578                 return $state->unstripGeneral( $text );
 579         }
 580
 581         /**
 582          * Always call this after unstrip() to preserve the order
 583          *
 584          * @private
 585          * @deprecated use $this->mStripState->unstrip()
 586          */
 587         function unstripNoWiki( $text, $state ) {
 588                 return $state->unstripNoWiki( $text );
 589         }
 590
 591         /**
 592          * @deprecated use $this->mStripState->unstripBoth()
 593          */
 594         function unstripForHTML( $text ) {
 595                 return $this->mStripState->unstripBoth( $text );
 596         }
 597
 598         /**
 599          * Add an item to the strip state
 600          * Returns the unique tag which must be inserted into the stripped text
 601          * The tag will be replaced with the original text in unstrip()
 602          *
 603          * @private
 604          */
 605         function insertStripItem( $text ) {
 606                 static $n = 0;
 607                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 608                 ++$n;
 609                 $this->mStripState->general->setPair( $rnd, $text );
 610                 return $rnd;
 611         }
 612
 613         /**
 614          * Interface with html tidy, used if $wgUseTidy = true.
 615          * If tidy isn't able to correct the markup, the original will be
 616          * returned in all its glory with a warning comment appended.
 617          *
 618          * Either the external tidy program or the in-process tidy extension
 619          * will be used depending on availability. Override the default
 620          * $wgTidyInternal setting to disable the internal if it's not working.
 621          *
 622          * @param string $text Hideous HTML input
 623          * @return string Corrected HTML output
 624          * @public
 625          * @static
 626          */
 627         function tidy( $text ) {
 628                 global $wgTidyInternal;
 629                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 630 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 631 '<head><title>test</title></head><body>'.$text.'</body></html>';
 632                 if( $wgTidyInternal ) {
 633                         $correctedtext = Parser::internalTidy( $wrappedtext );
 634                 } else {
 635                         $correctedtext = Parser::externalTidy( $wrappedtext );
 636                 }
 637                 if( is_null( $correctedtext ) ) {
 638                         wfDebug( "Tidy error detected!\n" );
 639                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 640                 }
 641                 return $correctedtext;
 642         }
 643
 644         /**
 645          * Spawn an external HTML tidy process and get corrected markup back from it.
 646          *
 647          * @private
 648          * @static
 649          */
 650         function externalTidy( $text ) {
 651                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 652                 $fname = 'Parser::externalTidy';
 653                 wfProfileIn( $fname );
 654
 655                 $cleansource = '';
 656                 $opts = ' -utf8';
 657
 658                 $descriptorspec = array(
 659                         0 => array('pipe', 'r'),
 660                         1 => array('pipe', 'w'),
 661                         2 => array('file', wfGetNull(), 'a')
 662                 );
 663                 $pipes = array();
 664                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 665                 if (is_resource($process)) {
 666                         // Theoretically, this style of communication could cause a deadlock
 667                         // here. If the stdout buffer fills up, then writes to stdin could
 668                         // block. This doesn't appear to happen with tidy, because tidy only
 669                         // writes to stdout after it's finished reading from stdin. Search
 670                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 671                         fwrite($pipes[0], $text);
 672                         fclose($pipes[0]);
 673                         while (!feof($pipes[1])) {
 674                                 $cleansource .= fgets($pipes[1], 1024);
 675                         }
 676                         fclose($pipes[1]);
 677                         proc_close($process);
 678                 }
 679
 680                 wfProfileOut( $fname );
 681
 682                 if( $cleansource == '' && $text != '') {
 683                         // Some kind of error happened, so we couldn't get the corrected text.
 684                         // Just give up; we'll use the source text and append a warning.
 685                         return null;
 686                 } else {
 687                         return $cleansource;
 688                 }
 689         }
 690
 691         /**
 692          * Use the HTML tidy PECL extension to use the tidy library in-process,
 693          * saving the overhead of spawning a new process.
 694          *
 695          * 'pear install tidy' should be able to compile the extension module.
 696          *
 697          * @private
 698          * @static
 699          */
 700         function internalTidy( $text ) {
 701                 global $wgTidyConf, $IP, $wgDebugTidy;
 702                 $fname = 'Parser::internalTidy';
 703                 wfProfileIn( $fname );
 704
 705                 $tidy = new tidy;
 706                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 707                 $tidy->cleanRepair();
 708                 if( $tidy->getStatus() == 2 ) {
 709                         // 2 is magic number for fatal error
 710                         // http://www.php.net/manual/en/function.tidy-get-status.php
 711                         $cleansource = null;
 712                 } else {
 713                         $cleansource = tidy_get_output( $tidy );
 714                 }
 715                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 716                         $cleansource .= "<!--\nTidy reports:\n" .
 717                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 718                                 "\n-->";
 719                 }
 720
 721                 wfProfileOut( $fname );
 722                 return $cleansource;
 723         }
 724
 725         /**
 726          * parse the wiki syntax used to render tables
 727          *
 728          * @private
 729          */
 730         function doTableStuff ( $text ) {
 731                 $fname = 'Parser::doTableStuff';
 732                 wfProfileIn( $fname );
 733
 734                 $lines = explode ( "\n" , $text );
 735                 $td_history = array (); // Is currently a td tag open?
 736                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 737                 $tr_history = array (); // Is currently a tr tag open?
 738                 $tr_attributes = array (); // history of tr attributes
 739                 $has_opened_tr = array(); // Did this table open a <tr> element?
 740                 $indent_level = 0; // indent level of the table
 741                 foreach ( $lines as $key => $line )
 742                 {
 743                         $line = trim ( $line );
 744
 745                         if( $line == '' ) { // empty line, go to next line
 746                                 continue;
 747                         }
 748                         $first_character = $line{0};
 749                         $matches = array();
 750
 751                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 752                                 // First check if we are starting a new table
 753                                 $indent_level = strlen( $matches[1] );
 754
 755                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 756                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 757
 758                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 759                                 array_push ( $td_history , false );
 760                                 array_push ( $last_tag_history , '' );
 761                                 array_push ( $tr_history , false );
 762                                 array_push ( $tr_attributes , '' );
 763                                 array_push ( $has_opened_tr , false );
 764                         } else if ( count ( $td_history ) == 0 ) {
 765                                 // Don't do any of the following
 766                                 continue;
 767                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 768                                 // We are ending a table
 769                                 $line = '</table>' . substr ( $line , 2 );
 770                                 $last_tag = array_pop ( $last_tag_history );
 771
 772                                 if ( !array_pop ( $has_opened_tr ) ) {
 773                                         $line = "<tr><td></td></tr>{$line}";
 774                                 }
 775
 776                                 if ( array_pop ( $tr_history ) ) {
 777                                         $line = "</tr>{$line}";
 778                                 }
 779
 780                                 if ( array_pop ( $td_history ) ) {
 781                                         $line = "</{$last_tag}>{$line}";
 782                                 }
 783                                 array_pop ( $tr_attributes );
 784                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 785                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 786                                 // Now we have a table row
 787                                 $line = preg_replace( '#^\|-+#', '', $line );
 788
 789                                 // Whats after the tag is now only attributes
 790                                 $attributes = $this->mStripState->unstripBoth( $line );
 791                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 792                                 array_pop ( $tr_attributes );
 793                                 array_push ( $tr_attributes , $attributes );
 794
 795                                 $line = '';
 796                                 $last_tag = array_pop ( $last_tag_history );
 797                                 array_pop ( $has_opened_tr );
 798                                 array_push ( $has_opened_tr , true );
 799
 800                                 if ( array_pop ( $tr_history ) ) {
 801                                         $line = '</tr>';
 802                                 }
 803
 804                                 if ( array_pop ( $td_history ) ) {
 805                                         $line = "</{$last_tag}>{$line}";
 806                                 }
 807
 808                                 $lines[$key] = $line;
 809                                 array_push ( $tr_history , false );
 810                                 array_push ( $td_history , false );
 811                                 array_push ( $last_tag_history , '' );
 812                         }
 813                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 814                                 // This might be cell elements, td, th or captions
 815                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 816                                         $first_character = '+';
 817                                         $line = substr ( $line , 1 );
 818                                 }
 819
 820                                 $line = substr ( $line , 1 );
 821
 822                                 if ( $first_character == '!' ) {
 823                                         $line = str_replace ( '!!' , '||' , $line );
 824                                 }
 825
 826                                 // Split up multiple cells on the same line.
 827                                 // FIXME : This can result in improper nesting of tags processed
 828                                 // by earlier parser steps, but should avoid splitting up eg
 829                                 // attribute values containing literal "||".
 830                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 831
 832                                 $lines[$key] = '';
 833
 834                                 // Loop through each table cell
 835                                 foreach ( $cells as $cell )
 836                                 {
 837                                         $previous = '';
 838                                         if ( $first_character != '+' )
 839                                         {
 840                                                 $tr_after = array_pop ( $tr_attributes );
 841                                                 if ( !array_pop ( $tr_history ) ) {
 842                                                         $previous = "<tr{$tr_after}>\n";
 843                                                 }
 844                                                 array_push ( $tr_history , true );
 845                                                 array_push ( $tr_attributes , '' );
 846                                                 array_pop ( $has_opened_tr );
 847                                                 array_push ( $has_opened_tr , true );
 848                                         }
 849
 850                                         $last_tag = array_pop ( $last_tag_history );
 851
 852                                         if ( array_pop ( $td_history ) ) {
 853                                                 $previous = "</{$last_tag}>{$previous}";
 854                                         }
 855
 856                                         if ( $first_character == '|' ) {
 857                                                 $last_tag = 'td';
 858                                         } else if ( $first_character == '!' ) {
 859                                                 $last_tag = 'th';
 860                                         } else if ( $first_character == '+' ) {
 861                                                 $last_tag = 'caption';
 862                                         } else {
 863                                                 $last_tag = '';
 864                                         }
 865
 866                                         array_push ( $last_tag_history , $last_tag );
 867
 868                                         // A cell could contain both parameters and data
 869                                         $cell_data = explode ( '|' , $cell , 2 );
 870
 871                                         // Bug 553: Note that a '|' inside an invalid link should not
 872                                         // be mistaken as delimiting cell parameters
 873                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 874                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 875                                         } else if ( count ( $cell_data ) == 1 )
 876                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 877                                         else {
 878                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 879                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 880                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 881                                         }
 882
 883                                         $lines[$key] .= $cell;
 884                                         array_push ( $td_history , true );
 885                                 }
 886                         }
 887                 }
 888
 889                 // Closing open td, tr && table
 890                 while ( count ( $td_history ) > 0 )
 891                 {
 892                         if ( array_pop ( $td_history ) ) {
 893                                 $lines[] = '</td>' ;
 894                         }
 895                         if ( array_pop ( $tr_history ) ) {
 896                                 $lines[] = '</tr>' ;
 897                         }
 898                         if ( !array_pop ( $has_opened_tr ) ) {
 899                                 $lines[] = "<tr><td></td></tr>" ;
 900                         }
 901
 902                         $lines[] = '</table>' ;
 903                 }
 904
 905                 $output = implode ( "\n" , $lines ) ;
 906
 907                 // special case: don't return empty table
 908                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 909                         $output = '';
 910                 }
 911
 912                 wfProfileOut( $fname );
 913
 914                 return $output;
 915         }
 916
 917         /**
 918          * Helper function for parse() that transforms wiki markup into
 919          * HTML. Only called for $mOutputType == OT_HTML.
 920          *
 921          * @private
 922          */
 923         function internalParse( $text ) {
 924                 $isMain = true;
 925                 $fname = 'Parser::internalParse';
 926                 wfProfileIn( $fname );
 927
 928                 # Hook to suspend the parser in this state
 929                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 930                         wfProfileOut( $fname );
 931                         return $text ;
 932                 }
 933
 934                 $text = $this->replaceVariables( $text );
 935                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 936                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 937
 938                 // Tables need to come after variable replacement for things to work
 939                 // properly; putting them before other transformations should keep
 940                 // exciting things like link expansions from showing up in surprising
 941                 // places.
 942                 $text = $this->doTableStuff( $text );
 943
 944                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 945
 946                 $text = $this->stripToc( $text );
 947                 $this->stripNoGallery( $text );
 948                 $text = $this->doHeadings( $text );
 949                 if($this->mOptions->getUseDynamicDates()) {
 950                         $df =& DateFormatter::getInstance();
 951                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 952                 }
 953                 $text = $this->doAllQuotes( $text );
 954                 $text = $this->replaceInternalLinks( $text );
 955                 $text = $this->replaceExternalLinks( $text );
 956
 957                 # replaceInternalLinks may sometimes leave behind
 958                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 959                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 960
 961                 $text = $this->doMagicLinks( $text );
 962                 $text = $this->formatHeadings( $text, $isMain );
 963
 964                 wfProfileOut( $fname );
 965                 return $text;
 966         }
 967
 968         /**
 969          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 970          * magic external links.
 971          *
 972          * @private
 973          */
 974         function doMagicLinks( $text ) {
 975                 wfProfileIn( __METHOD__ );
 976                 $text = preg_replace_callback(
 977                         '!(?:                           # Start cases
 978                             <a.*?</a> |                 # Skip link text
 979                             <.*?> |                     # Skip stuff inside HTML elements
 980                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 981                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 982                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 983                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 984                                      [0-9Xx]                 # check digit
 985                                    \b)
 986                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 987                 wfProfileOut( __METHOD__ );
 988                 return $text;
 989         }
 990
 991         function magicLinkCallback( $m ) {
 992                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 993                         # Skip HTML element
 994                         return $m[0];
 995                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 996                         $isbn = $m[2];
 997                         $num = strtr( $isbn, array(
 998                                 '-' => '',
 999                                 ' ' => '',
1000                                 'x' => 'X',
1001                         ));
1002                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1003                         $text = '<a href="' .
1004                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1005                                 "\" class=\"internal\">ISBN $isbn</a>";
1006                 } else {
1007                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1008                                 $keyword = 'RFC';
1009                                 $urlmsg = 'rfcurl';
1010                                 $id = $m[1];
1011                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1012                                 $keyword = 'PMID';
1013                                 $urlmsg = 'pubmedurl';
1014                                 $id = $m[1];
1015                         } else {
1016                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1017                                         substr($m[0], 0, 20 ) . '"' );
1018                         }
1019
1020                         $url = wfMsg( $urlmsg, $id);
1021                         $sk = $this->mOptions->getSkin();
1022                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1023                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1024                 }
1025                 return $text;
1026         }
1027
1028         /**
1029          * Parse headers and return html
1030          *
1031          * @private
1032          */
1033         function doHeadings( $text ) {
1034                 $fname = 'Parser::doHeadings';
1035                 wfProfileIn( $fname );
1036                 for ( $i = 6; $i >= 1; --$i ) {
1037                         $h = str_repeat( '=', $i );
1038                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1039                           "<h$i>\\1</h$i>", $text );
1040                 }
1041                 wfProfileOut( $fname );
1042                 return $text;
1043         }
1044
1045         /**
1046          * Replace single quotes with HTML markup
1047          * @private
1048          * @return string the altered text
1049          */
1050         function doAllQuotes( $text ) {
1051                 $fname = 'Parser::doAllQuotes';
1052                 wfProfileIn( $fname );
1053                 $outtext = '';
1054                 $lines = explode( "\n", $text );
1055                 foreach ( $lines as $line ) {
1056                         $outtext .= $this->doQuotes ( $line ) . "\n";
1057                 }
1058                 $outtext = substr($outtext, 0,-1);
1059                 wfProfileOut( $fname );
1060                 return $outtext;
1061         }
1062
1063         /**
1064          * Helper function for doAllQuotes()
1065          */
1066         public function doQuotes( $text ) {
1067                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1068                 if ( count( $arr ) == 1 )
1069                         return $text;
1070                 else
1071                 {
1072                         # First, do some preliminary work. This may shift some apostrophes from
1073                         # being mark-up to being text. It also counts the number of occurrences
1074                         # of bold and italics mark-ups.
1075                         $i = 0;
1076                         $numbold = 0;
1077                         $numitalics = 0;
1078                         foreach ( $arr as $r )
1079                         {
1080                                 if ( ( $i % 2 ) == 1 )
1081                                 {
1082                                         # If there are ever four apostrophes, assume the first is supposed to
1083                                         # be text, and the remaining three constitute mark-up for bold text.
1084                                         if ( strlen( $arr[$i] ) == 4 )
1085                                         {
1086                                                 $arr[$i-1] .= "'";
1087                                                 $arr[$i] = "'''";
1088                                         }
1089                                         # If there are more than 5 apostrophes in a row, assume they're all
1090                                         # text except for the last 5.
1091                                         else if ( strlen( $arr[$i] ) > 5 )
1092                                         {
1093                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1094                                                 $arr[$i] = "'''''";
1095                                         }
1096                                         # Count the number of occurrences of bold and italics mark-ups.
1097                                         # We are not counting sequences of five apostrophes.
1098                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1099                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1100                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1101                                 }
1102                                 $i++;
1103                         }
1104
1105                         # If there is an odd number of both bold and italics, it is likely
1106                         # that one of the bold ones was meant to be an apostrophe followed
1107                         # by italics. Which one we cannot know for certain, but it is more
1108                         # likely to be one that has a single-letter word before it.
1109                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1110                         {
1111                                 $i = 0;
1112                                 $firstsingleletterword = -1;
1113                                 $firstmultiletterword = -1;
1114                                 $firstspace = -1;
1115                                 foreach ( $arr as $r )
1116                                 {
1117                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1118                                         {
1119                                                 $x1 = substr ($arr[$i-1], -1);
1120                                                 $x2 = substr ($arr[$i-1], -2, 1);
1121                                                 if ($x1 == ' ') {
1122                                                         if ($firstspace == -1) $firstspace = $i;
1123                                                 } else if ($x2 == ' ') {
1124                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1125                                                 } else {
1126                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1127                                                 }
1128                                         }
1129                                         $i++;
1130                                 }
1131
1132                                 # If there is a single-letter word, use it!
1133                                 if ($firstsingleletterword > -1)
1134                                 {
1135                                         $arr [ $firstsingleletterword ] = "''";
1136                                         $arr [ $firstsingleletterword-1 ] .= "'";
1137                                 }
1138                                 # If not, but there's a multi-letter word, use that one.
1139                                 else if ($firstmultiletterword > -1)
1140                                 {
1141                                         $arr [ $firstmultiletterword ] = "''";
1142                                         $arr [ $firstmultiletterword-1 ] .= "'";
1143                                 }
1144                                 # ... otherwise use the first one that has neither.
1145                                 # (notice that it is possible for all three to be -1 if, for example,
1146                                 # there is only one pentuple-apostrophe in the line)
1147                                 else if ($firstspace > -1)
1148                                 {
1149                                         $arr [ $firstspace ] = "''";
1150                                         $arr [ $firstspace-1 ] .= "'";
1151                                 }
1152                         }
1153
1154                         # Now let's actually convert our apostrophic mush to HTML!
1155                         $output = '';
1156                         $buffer = '';
1157                         $state = '';
1158                         $i = 0;
1159                         foreach ($arr as $r)
1160                         {
1161                                 if (($i % 2) == 0)
1162                                 {
1163                                         if ($state == 'both')
1164                                                 $buffer .= $r;
1165                                         else
1166                                                 $output .= $r;
1167                                 }
1168                                 else
1169                                 {
1170                                         if (strlen ($r) == 2)
1171                                         {
1172                                                 if ($state == 'i')
1173                                                 { $output .= '</i>'; $state = ''; }
1174                                                 else if ($state == 'bi')
1175                                                 { $output .= '</i>'; $state = 'b'; }
1176                                                 else if ($state == 'ib')
1177                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1178                                                 else if ($state == 'both')
1179                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1180                                                 else # $state can be 'b' or ''
1181                                                 { $output .= '<i>'; $state .= 'i'; }
1182                                         }
1183                                         else if (strlen ($r) == 3)
1184                                         {
1185                                                 if ($state == 'b')
1186                                                 { $output .= '</b>'; $state = ''; }
1187                                                 else if ($state == 'bi')
1188                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1189                                                 else if ($state == 'ib')
1190                                                 { $output .= '</b>'; $state = 'i'; }
1191                                                 else if ($state == 'both')
1192                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1193                                                 else # $state can be 'i' or ''
1194                                                 { $output .= '<b>'; $state .= 'b'; }
1195                                         }
1196                                         else if (strlen ($r) == 5)
1197                                         {
1198                                                 if ($state == 'b')
1199                                                 { $output .= '</b><i>'; $state = 'i'; }
1200                                                 else if ($state == 'i')
1201                                                 { $output .= '</i><b>'; $state = 'b'; }
1202                                                 else if ($state == 'bi')
1203                                                 { $output .= '</i></b>'; $state = ''; }
1204                                                 else if ($state == 'ib')
1205                                                 { $output .= '</b></i>'; $state = ''; }
1206                                                 else if ($state == 'both')
1207                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1208                                                 else # ($state == '')
1209                                                 { $buffer = ''; $state = 'both'; }
1210                                         }
1211                                 }
1212                                 $i++;
1213                         }
1214                         # Now close all remaining tags.  Notice that the order is important.
1215                         if ($state == 'b' || $state == 'ib')
1216                                 $output .= '</b>';
1217                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1218                                 $output .= '</i>';
1219                         if ($state == 'bi')
1220                                 $output .= '</b>';
1221                         # There might be lonely ''''', so make sure we have a buffer
1222                         if ($state == 'both' && $buffer)
1223                                 $output .= '<b><i>'.$buffer.'</i></b>';
1224                         return $output;
1225                 }
1226         }
1227
1228         /**
1229          * Replace external links
1230          *
1231          * Note: this is all very hackish and the order of execution matters a lot.
1232          * Make sure to run maintenance/parserTests.php if you change this code.
1233          *
1234          * @private
1235          */
1236         function replaceExternalLinks( $text ) {
1237                 global $wgContLang;
1238                 $fname = 'Parser::replaceExternalLinks';
1239                 wfProfileIn( $fname );
1240
1241                 $sk = $this->mOptions->getSkin();
1242
1243                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1244
1245                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1246
1247                 $i = 0;
1248                 while ( $i<count( $bits ) ) {
1249                         $url = $bits[$i++];
1250                         $protocol = $bits[$i++];
1251                         $text = $bits[$i++];
1252                         $trail = $bits[$i++];
1253
1254                         # The characters '<' and '>' (which were escaped by
1255                         # removeHTMLtags()) should not be included in
1256                         # URLs, per RFC 2396.
1257                         $m2 = array();
1258                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1259                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1260                                 $url = substr($url, 0, $m2[0][1]);
1261                         }
1262
1263                         # If the link text is an image URL, replace it with an <img> tag
1264                         # This happened by accident in the original parser, but some people used it extensively
1265                         $img = $this->maybeMakeExternalImage( $text );
1266                         if ( $img !== false ) {
1267                                 $text = $img;
1268                         }
1269
1270                         $dtrail = '';
1271
1272                         # Set linktype for CSS - if URL==text, link is essentially free
1273                         $linktype = ($text == $url) ? 'free' : 'text';
1274
1275                         # No link text, e.g. [http://domain.tld/some.link]
1276                         if ( $text == '' ) {
1277                                 # Autonumber if allowed. See bug #5918
1278                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1279                                         $text = '[' . ++$this->mAutonumber . ']';
1280                                         $linktype = 'autonumber';
1281                                 } else {
1282                                         # Otherwise just use the URL
1283                                         $text = htmlspecialchars( $url );
1284                                         $linktype = 'free';
1285                                 }
1286                         } else {
1287                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1288                                 # Check for trail
1289                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1290                         }
1291
1292                         $text = $wgContLang->markNoConversion($text);
1293
1294                         $url = Sanitizer::cleanUrl( $url );
1295
1296                         # Process the trail (i.e. everything after this link up until start of the next link),
1297                         # replacing any non-bracketed links
1298                         $trail = $this->replaceFreeExternalLinks( $trail );
1299
1300                         # Use the encoded URL
1301                         # This means that users can paste URLs directly into the text
1302                         # Funny characters like &ouml; aren't valid in URLs anyway
1303                         # This was changed in August 2004
1304                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1305
1306                         # Register link in the output object.
1307                         # Replace unnecessary URL escape codes with the referenced character
1308                         # This prevents spammers from hiding links from the filters
1309                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1310                         $this->mOutput->addExternalLink( $pasteurized );
1311                 }
1312
1313                 wfProfileOut( $fname );
1314                 return $s;
1315         }
1316
1317         /**
1318          * Replace anything that looks like a URL with a link
1319          * @private
1320          */
1321         function replaceFreeExternalLinks( $text ) {
1322                 global $wgContLang;
1323                 $fname = 'Parser::replaceFreeExternalLinks';
1324                 wfProfileIn( $fname );
1325
1326                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1327                 $s = array_shift( $bits );
1328                 $i = 0;
1329
1330                 $sk = $this->mOptions->getSkin();
1331
1332                 while ( $i < count( $bits ) ){
1333                         $protocol = $bits[$i++];
1334                         $remainder = $bits[$i++];
1335
1336                         $m = array();
1337                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1338                                 # Found some characters after the protocol that look promising
1339                                 $url = $protocol . $m[1];
1340                                 $trail = $m[2];
1341
1342                                 # special case: handle urls as url args:
1343                                 # http://www.example.com/foo?=http://www.example.com/bar
1344                                 if(strlen($trail) == 0 &&
1345                                         isset($bits[$i]) &&
1346                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1347                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1348                                 {
1349                                         # add protocol, arg
1350                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1351                                         $i += 2;
1352                                         $trail = $m[2];
1353                                 }
1354
1355                                 # The characters '<' and '>' (which were escaped by
1356                                 # removeHTMLtags()) should not be included in
1357                                 # URLs, per RFC 2396.
1358                                 $m2 = array();
1359                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1360                                         $trail = substr($url, $m2[0][1]) . $trail;
1361                                         $url = substr($url, 0, $m2[0][1]);
1362                                 }
1363
1364                                 # Move trailing punctuation to $trail
1365                                 $sep = ',;\.:!?';
1366                                 # If there is no left bracket, then consider right brackets fair game too
1367                                 if ( strpos( $url, '(' ) === false ) {
1368                                         $sep .= ')';
1369                                 }
1370
1371                                 $numSepChars = strspn( strrev( $url ), $sep );
1372                                 if ( $numSepChars ) {
1373                                         $trail = substr( $url, -$numSepChars ) . $trail;
1374                                         $url = substr( $url, 0, -$numSepChars );
1375                                 }
1376
1377                                 $url = Sanitizer::cleanUrl( $url );
1378
1379                                 # Is this an external image?
1380                                 $text = $this->maybeMakeExternalImage( $url );
1381                                 if ( $text === false ) {
1382                                         # Not an image, make a link
1383                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1384                                         # Register it in the output object...
1385                                         # Replace unnecessary URL escape codes with their equivalent characters
1386                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1387                                         $this->mOutput->addExternalLink( $pasteurized );
1388                                 }
1389                                 $s .= $text . $trail;
1390                         } else {
1391                                 $s .= $protocol . $remainder;
1392                         }
1393                 }
1394                 wfProfileOut( $fname );
1395                 return $s;
1396         }
1397
1398         /**
1399          * Replace unusual URL escape codes with their equivalent characters
1400          * @param string
1401          * @return string
1402          * @static
1403          * @todo  This can merge genuinely required bits in the path or query string,
1404          *        breaking legit URLs. A proper fix would treat the various parts of
1405          *        the URL differently; as a workaround, just use the output for
1406          *        statistical records, not for actual linking/output.
1407          */
1408         static function replaceUnusualEscapes( $url ) {
1409                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1410                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1411         }
1412
1413         /**
1414          * Callback function used in replaceUnusualEscapes().
1415          * Replaces unusual URL escape codes with their equivalent character
1416          * @static
1417          * @private
1418          */
1419         private static function replaceUnusualEscapesCallback( $matches ) {
1420                 $char = urldecode( $matches[0] );
1421                 $ord = ord( $char );
1422                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1423                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1424                         // No, shouldn't be escaped
1425                         return $char;
1426                 } else {
1427                         // Yes, leave it escaped
1428                         return $matches[0];
1429                 }
1430         }
1431
1432         /**
1433          * make an image if it's allowed, either through the global
1434          * option or through the exception
1435          * @private
1436          */
1437         function maybeMakeExternalImage( $url ) {
1438                 $sk = $this->mOptions->getSkin();
1439                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1440                 $imagesexception = !empty($imagesfrom);
1441                 $text = false;
1442                 if ( $this->mOptions->getAllowExternalImages()
1443                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1444                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1445                                 # Image found
1446                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1447                         }
1448                 }
1449                 return $text;
1450         }
1451
1452         /**
1453          * Process [[ ]] wikilinks
1454          *
1455          * @private
1456          */
1457         function replaceInternalLinks( $s ) {
1458                 global $wgContLang;
1459                 static $fname = 'Parser::replaceInternalLinks' ;
1460
1461                 wfProfileIn( $fname );
1462
1463                 wfProfileIn( $fname.'-setup' );
1464                 static $tc = FALSE;
1465                 # the % is needed to support urlencoded titles as well
1466                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1467
1468                 $sk = $this->mOptions->getSkin();
1469
1470                 #split the entire text string on occurences of [[
1471                 $a = explode( '[[', ' ' . $s );
1472                 #get the first element (all text up to first [[), and remove the space we added
1473                 $s = array_shift( $a );
1474                 $s = substr( $s, 1 );
1475
1476                 # Match a link having the form [[namespace:link|alternate]]trail
1477                 static $e1 = FALSE;
1478                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1479                 # Match cases where there is no "]]", which might still be images
1480                 static $e1_img = FALSE;
1481                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1482
1483                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1484                 $e2 = null;
1485                 if ( $useLinkPrefixExtension ) {
1486                         # Match the end of a line for a word that's not followed by whitespace,
1487                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1488                         $e2 = wfMsgForContent( 'linkprefix' );
1489                 }
1490
1491                 if( is_null( $this->mTitle ) ) {
1492                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1493                 }
1494                 $nottalk = !$this->mTitle->isTalkPage();
1495
1496                 if ( $useLinkPrefixExtension ) {
1497                         $m = array();
1498                         if ( preg_match( $e2, $s, $m ) ) {
1499                                 $first_prefix = $m[2];
1500                         } else {
1501                                 $first_prefix = false;
1502                         }
1503                 } else {
1504                         $prefix = '';
1505                 }
1506
1507                 if($wgContLang->hasVariants()) {
1508                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1509                 } else {
1510                         $selflink = array($this->mTitle->getPrefixedText());
1511                 }
1512                 $useSubpages = $this->areSubpagesAllowed();
1513                 wfProfileOut( $fname.'-setup' );
1514
1515                 # Loop for each link
1516                 for ($k = 0; isset( $a[$k] ); $k++) {
1517                         $line = $a[$k];
1518                         if ( $useLinkPrefixExtension ) {
1519                                 wfProfileIn( $fname.'-prefixhandling' );
1520                                 if ( preg_match( $e2, $s, $m ) ) {
1521                                         $prefix = $m[2];
1522                                         $s = $m[1];
1523                                 } else {
1524                                         $prefix='';
1525                                 }
1526                                 # first link
1527                                 if($first_prefix) {
1528                                         $prefix = $first_prefix;
1529                                         $first_prefix = false;
1530                                 }
1531                                 wfProfileOut( $fname.'-prefixhandling' );
1532                         }
1533
1534                         $might_be_img = false;
1535
1536                         wfProfileIn( "$fname-e1" );
1537                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1538                                 $text = $m[2];
1539                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1540                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1541                                 # the real problem is with the $e1 regex
1542                                 # See bug 1300.
1543                                 #
1544                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1545                                 # and no image is in sight. See bug 2095.
1546                                 #
1547                                 if( $text !== '' &&
1548                                         substr( $m[3], 0, 1 ) === ']' &&
1549                                         strpos($text, '[') !== false
1550                                 )
1551                                 {
1552                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1553                                         $m[3] = substr( $m[3], 1 );
1554                                 }
1555                                 # fix up urlencoded title texts
1556                                 if( strpos( $m[1], '%' ) !== false ) {
1557                                         # Should anchors '#' also be rejected?
1558                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1559                                 }
1560                                 $trail = $m[3];
1561                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1562                                 $might_be_img = true;
1563                                 $text = $m[2];
1564                                 if ( strpos( $m[1], '%' ) !== false ) {
1565                                         $m[1] = urldecode($m[1]);
1566                                 }
1567                                 $trail = "";
1568                         } else { # Invalid form; output directly
1569                                 $s .= $prefix . '[[' . $line ;
1570                                 wfProfileOut( "$fname-e1" );
1571                                 continue;
1572                         }
1573                         wfProfileOut( "$fname-e1" );
1574                         wfProfileIn( "$fname-misc" );
1575
1576                         # Don't allow internal links to pages containing
1577                         # PROTO: where PROTO is a valid URL protocol; these
1578                         # should be external links.
1579                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1580                                 $s .= $prefix . '[[' . $line ;
1581                                 continue;
1582                         }
1583
1584                         # Make subpage if necessary
1585                         if( $useSubpages ) {
1586                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1587                         } else {
1588                                 $link = $m[1];
1589                         }
1590
1591                         $noforce = (substr($m[1], 0, 1) != ':');
1592                         if (!$noforce) {
1593                                 # Strip off leading ':'
1594                                 $link = substr($link, 1);
1595                         }
1596
1597                         wfProfileOut( "$fname-misc" );
1598                         wfProfileIn( "$fname-title" );
1599                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1600                         if( !$nt ) {
1601                                 $s .= $prefix . '[[' . $line;
1602                                 wfProfileOut( "$fname-title" );
1603                                 continue;
1604                         }
1605
1606                         $ns = $nt->getNamespace();
1607                         $iw = $nt->getInterWiki();
1608                         wfProfileOut( "$fname-title" );
1609
1610                         if ($might_be_img) { # if this is actually an invalid link
1611                                 wfProfileIn( "$fname-might_be_img" );
1612                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1613                                         $found = false;
1614                                         while (isset ($a[$k+1]) ) {
1615                                                 #look at the next 'line' to see if we can close it there
1616                                                 $spliced = array_splice( $a, $k + 1, 1 );
1617                                                 $next_line = array_shift( $spliced );
1618                                                 $m = explode( ']]', $next_line, 3 );
1619                                                 if ( count( $m ) == 3 ) {
1620                                                         # the first ]] closes the inner link, the second the image
1621                                                         $found = true;
1622                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1623                                                         $trail = $m[2];
1624                                                         break;
1625                                                 } elseif ( count( $m ) == 2 ) {
1626                                                         #if there's exactly one ]] that's fine, we'll keep looking
1627                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1628                                                 } else {
1629                                                         #if $next_line is invalid too, we need look no further
1630                                                         $text .= '[[' . $next_line;
1631                                                         break;
1632                                                 }
1633                                         }
1634                                         if ( !$found ) {
1635                                                 # we couldn't find the end of this imageLink, so output it raw
1636                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1637                                                 $text = $this->replaceInternalLinks($text);
1638                                                 $s .= "{$prefix}[[$link|$text";
1639                                                 # note: no $trail, because without an end, there *is* no trail
1640                                                 wfProfileOut( "$fname-might_be_img" );
1641                                                 continue;
1642                                         }
1643                                 } else { #it's not an image, so output it raw
1644                                         $s .= "{$prefix}[[$link|$text";
1645                                         # note: no $trail, because without an end, there *is* no trail
1646                                         wfProfileOut( "$fname-might_be_img" );
1647                                         continue;
1648                                 }
1649                                 wfProfileOut( "$fname-might_be_img" );
1650                         }
1651
1652                         $wasblank = ( '' == $text );
1653                         if( $wasblank ) $text = $link;
1654
1655                         # Link not escaped by : , create the various objects
1656                         if( $noforce ) {
1657
1658                                 # Interwikis
1659                                 wfProfileIn( "$fname-interwiki" );
1660                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1661                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1662                                         $s = rtrim($s . $prefix);
1663                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1664                                         wfProfileOut( "$fname-interwiki" );
1665                                         continue;
1666                                 }
1667                                 wfProfileOut( "$fname-interwiki" );
1668
1669                                 if ( $ns == NS_IMAGE ) {
1670                                         wfProfileIn( "$fname-image" );
1671                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1672                                                 # recursively parse links inside the image caption
1673                                                 # actually, this will parse them in any other parameters, too,
1674                                                 # but it might be hard to fix that, and it doesn't matter ATM
1675                                                 $text = $this->replaceExternalLinks($text);
1676                                                 $text = $this->replaceInternalLinks($text);
1677
1678                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1679                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1680                                                 $this->mOutput->addImage( $nt->getDBkey() );
1681
1682                                                 wfProfileOut( "$fname-image" );
1683                                                 continue;
1684                                         } else {
1685                                                 # We still need to record the image's presence on the page
1686                                                 $this->mOutput->addImage( $nt->getDBkey() );
1687                                         }
1688                                         wfProfileOut( "$fname-image" );
1689
1690                                 }
1691
1692                                 if ( $ns == NS_CATEGORY ) {
1693                                         wfProfileIn( "$fname-category" );
1694                                         $s = rtrim($s . "\n"); # bug 87
1695
1696                                         if ( $wasblank ) {
1697                                                 $sortkey = $this->getDefaultSort();
1698                                         } else {
1699                                                 $sortkey = $text;
1700                                         }
1701                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1702                                         $sortkey = str_replace( "\n", '', $sortkey );
1703                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1704                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1705
1706                                         /**
1707                                          * Strip the whitespace Category links produce, see bug 87
1708                                          * @todo We might want to use trim($tmp, "\n") here.
1709                                          */
1710                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1711
1712                                         wfProfileOut( "$fname-category" );
1713                                         continue;
1714                                 }
1715                         }
1716
1717                         # Self-link checking
1718                         if( $nt->getFragment() === '' ) {
1719                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1720                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1721                                         continue;
1722                                 }
1723                         }
1724
1725                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1726                         if( $ns == NS_MEDIA ) {
1727                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1728                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1729                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1730                                 $this->mOutput->addImage( $nt->getDBkey() );
1731                                 continue;
1732                         } elseif( $ns == NS_SPECIAL ) {
1733                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1734                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1735                                 } else {
1736                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1737                                 }
1738                                 continue;
1739                         } elseif( $ns == NS_IMAGE ) {
1740                                 $img = wfFindFile( $nt );
1741                                 if( $img ) {
1742                                         // Force a blue link if the file exists; may be a remote
1743                                         // upload on the shared repository, and we want to see its
1744                                         // auto-generated page.
1745                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1746                                         $this->mOutput->addLink( $nt );
1747                                         continue;
1748                                 }
1749                         }
1750                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1751                 }
1752                 wfProfileOut( $fname );
1753                 return $s;
1754         }
1755
1756         /**
1757          * Make a link placeholder. The text returned can be later resolved to a real link with
1758          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1759          * parsing of interwiki links, and secondly to allow all existence checks and
1760          * article length checks (for stub links) to be bundled into a single query.
1761          *
1762          */
1763         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1764                 wfProfileIn( __METHOD__ );
1765                 if ( ! is_object($nt) ) {
1766                         # Fail gracefully
1767                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1768                 } else {
1769                         # Separate the link trail from the rest of the link
1770                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1771
1772                         if ( $nt->isExternal() ) {
1773                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1774                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1775                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1776                         } else {
1777                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1778                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1779                                 $this->mLinkHolders['queries'][] = $query;
1780                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1781                                 $this->mLinkHolders['titles'][] = $nt;
1782
1783                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1784                         }
1785                 }
1786                 wfProfileOut( __METHOD__ );
1787                 return $retVal;
1788         }
1789
1790         /**
1791          * Render a forced-blue link inline; protect against double expansion of
1792          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1793          * Since this little disaster has to split off the trail text to avoid
1794          * breaking URLs in the following text without breaking trails on the
1795          * wiki links, it's been made into a horrible function.
1796          *
1797          * @param Title $nt
1798          * @param string $text
1799          * @param string $query
1800          * @param string $trail
1801          * @param string $prefix
1802          * @return string HTML-wikitext mix oh yuck
1803          */
1804         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1805                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1806                 $sk = $this->mOptions->getSkin();
1807                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1808                 return $this->armorLinks( $link ) . $trail;
1809         }
1810
1811         /**
1812          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1813          * going to go through further parsing steps before inline URL expansion.
1814          *
1815          * In particular this is important when using action=render, which causes
1816          * full URLs to be included.
1817          *
1818          * Oh man I hate our multi-layer parser!
1819          *
1820          * @param string more-or-less HTML
1821          * @return string less-or-more HTML with NOPARSE bits
1822          */
1823         function armorLinks( $text ) {
1824                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1825                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1826         }
1827
1828         /**
1829          * Return true if subpage links should be expanded on this page.
1830          * @return bool
1831          */
1832         function areSubpagesAllowed() {
1833                 # Some namespaces don't allow subpages
1834                 global $wgNamespacesWithSubpages;
1835                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1836         }
1837
1838         /**
1839          * Handle link to subpage if necessary
1840          * @param string $target the source of the link
1841          * @param string &$text the link text, modified as necessary
1842          * @return string the full name of the link
1843          * @private
1844          */
1845         function maybeDoSubpageLink($target, &$text) {
1846                 # Valid link forms:
1847                 # Foobar -- normal
1848                 # :Foobar -- override special treatment of prefix (images, language links)
1849                 # /Foobar -- convert to CurrentPage/Foobar
1850                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1851                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1852                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1853
1854                 $fname = 'Parser::maybeDoSubpageLink';
1855                 wfProfileIn( $fname );
1856                 $ret = $target; # default return value is no change
1857
1858                 # Some namespaces don't allow subpages,
1859                 # so only perform processing if subpages are allowed
1860                 if( $this->areSubpagesAllowed() ) {
1861                         $hash = strpos( $target, '#' );
1862                         if( $hash !== false ) {
1863                                 $suffix = substr( $target, $hash );
1864                                 $target = substr( $target, 0, $hash );
1865                         } else {
1866                                 $suffix = '';
1867                         }
1868                         # bug 7425
1869                         $target = trim( $target );
1870                         # Look at the first character
1871                         if( $target != '' && $target{0} == '/' ) {
1872                                 # / at end means we don't want the slash to be shown
1873                                 $m = array();
1874                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1875                                 if( $trailingSlashes ) {
1876                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1877                                 } else {
1878                                         $noslash = substr( $target, 1 );
1879                                 }
1880
1881                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1882                                 if( '' === $text ) {
1883                                         $text = $target . $suffix;
1884                                 } # this might be changed for ugliness reasons
1885                         } else {
1886                                 # check for .. subpage backlinks
1887                                 $dotdotcount = 0;
1888                                 $nodotdot = $target;
1889                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1890                                         ++$dotdotcount;
1891                                         $nodotdot = substr( $nodotdot, 3 );
1892                                 }
1893                                 if($dotdotcount > 0) {
1894                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1895                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1896                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1897                                                 # / at the end means don't show full path
1898                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1899                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1900                                                         if( '' === $text ) {
1901                                                                 $text = $nodotdot . $suffix;
1902                                                         }
1903                                                 }
1904                                                 $nodotdot = trim( $nodotdot );
1905                                                 if( $nodotdot != '' ) {
1906                                                         $ret .= '/' . $nodotdot;
1907                                                 }
1908                                                 $ret .= $suffix;
1909                                         }
1910                                 }
1911                         }
1912                 }
1913
1914                 wfProfileOut( $fname );
1915                 return $ret;
1916         }
1917
1918         /**#@+
1919          * Used by doBlockLevels()
1920          * @private
1921          */
1922         /* private */ function closeParagraph() {
1923                 $result = '';
1924                 if ( '' != $this->mLastSection ) {
1925                         $result = '</' . $this->mLastSection  . ">\n";
1926                 }
1927                 $this->mInPre = false;
1928                 $this->mLastSection = '';
1929                 return $result;
1930         }
1931         # getCommon() returns the length of the longest common substring
1932         # of both arguments, starting at the beginning of both.
1933         #
1934         /* private */ function getCommon( $st1, $st2 ) {
1935                 $fl = strlen( $st1 );
1936                 $shorter = strlen( $st2 );
1937                 if ( $fl < $shorter ) { $shorter = $fl; }
1938
1939                 for ( $i = 0; $i < $shorter; ++$i ) {
1940                         if ( $st1{$i} != $st2{$i} ) { break; }
1941                 }
1942                 return $i;
1943         }
1944         # These next three functions open, continue, and close the list
1945         # element appropriate to the prefix character passed into them.
1946         #
1947         /* private */ function openList( $char ) {
1948                 $result = $this->closeParagraph();
1949
1950                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1951                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1952                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1953                 else if ( ';' == $char ) {
1954                         $result .= '<dl><dt>';
1955                         $this->mDTopen = true;
1956                 }
1957                 else { $result = '<!-- ERR 1 -->'; }
1958
1959                 return $result;
1960         }
1961
1962         /* private */ function nextItem( $char ) {
1963                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1964                 else if ( ':' == $char || ';' == $char ) {
1965                         $close = '</dd>';
1966                         if ( $this->mDTopen ) { $close = '</dt>'; }
1967                         if ( ';' == $char ) {
1968                                 $this->mDTopen = true;
1969                                 return $close . '<dt>';
1970                         } else {
1971                                 $this->mDTopen = false;
1972                                 return $close . '<dd>';
1973                         }
1974                 }
1975                 return '<!-- ERR 2 -->';
1976         }
1977
1978         /* private */ function closeList( $char ) {
1979                 if ( '*' == $char ) { $text = '</li></ul>'; }
1980                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1981                 else if ( ':' == $char ) {
1982                         if ( $this->mDTopen ) {
1983                                 $this->mDTopen = false;
1984                                 $text = '</dt></dl>';
1985                         } else {
1986                                 $text = '</dd></dl>';
1987                         }
1988                 }
1989                 else {  return '<!-- ERR 3 -->'; }
1990                 return $text."\n";
1991         }
1992         /**#@-*/
1993
1994         /**
1995          * Make lists from lines starting with ':', '*', '#', etc.
1996          *
1997          * @private
1998          * @return string the lists rendered as HTML
1999          */
2000         function doBlockLevels( $text, $linestart ) {
2001                 $fname = 'Parser::doBlockLevels';
2002                 wfProfileIn( $fname );
2003
2004                 # Parsing through the text line by line.  The main thing
2005                 # happening here is handling of block-level elements p, pre,
2006                 # and making lists from lines starting with * # : etc.
2007                 #
2008                 $textLines = explode( "\n", $text );
2009
2010                 $lastPrefix = $output = '';
2011                 $this->mDTopen = $inBlockElem = false;
2012                 $prefixLength = 0;
2013                 $paragraphStack = false;
2014
2015                 if ( !$linestart ) {
2016                         $output .= array_shift( $textLines );
2017                 }
2018                 foreach ( $textLines as $oLine ) {
2019                         $lastPrefixLength = strlen( $lastPrefix );
2020                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2021                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2022                         if ( !$this->mInPre ) {
2023                                 # Multiple prefixes may abut each other for nested lists.
2024                                 $prefixLength = strspn( $oLine, '*#:;' );
2025                                 $pref = substr( $oLine, 0, $prefixLength );
2026
2027                                 # eh?
2028                                 $pref2 = str_replace( ';', ':', $pref );
2029                                 $t = substr( $oLine, $prefixLength );
2030                                 $this->mInPre = !empty($preOpenMatch);
2031                         } else {
2032                                 # Don't interpret any other prefixes in preformatted text
2033                                 $prefixLength = 0;
2034                                 $pref = $pref2 = '';
2035                                 $t = $oLine;
2036                         }
2037
2038                         # List generation
2039                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2040                                 # Same as the last item, so no need to deal with nesting or opening stuff
2041                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2042                                 $paragraphStack = false;
2043
2044                                 if ( substr( $pref, -1 ) == ';') {
2045                                         # The one nasty exception: definition lists work like this:
2046                                         # ; title : definition text
2047                                         # So we check for : in the remainder text to split up the
2048                                         # title and definition, without b0rking links.
2049                                         $term = $t2 = '';
2050                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2051                                                 $t = $t2;
2052                                                 $output .= $term . $this->nextItem( ':' );
2053                                         }
2054                                 }
2055                         } elseif( $prefixLength || $lastPrefixLength ) {
2056                                 # Either open or close a level...
2057                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2058                                 $paragraphStack = false;
2059
2060                                 while( $commonPrefixLength < $lastPrefixLength ) {
2061                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2062                                         --$lastPrefixLength;
2063                                 }
2064                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2065                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2066                                 }
2067                                 while ( $prefixLength > $commonPrefixLength ) {
2068                                         $char = substr( $pref, $commonPrefixLength, 1 );
2069                                         $output .= $this->openList( $char );
2070
2071                                         if ( ';' == $char ) {
2072                                                 # FIXME: This is dupe of code above
2073                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2074                                                         $t = $t2;
2075                                                         $output .= $term . $this->nextItem( ':' );
2076                                                 }
2077                                         }
2078                                         ++$commonPrefixLength;
2079                                 }
2080                                 $lastPrefix = $pref2;
2081                         }
2082                         if( 0 == $prefixLength ) {
2083                                 wfProfileIn( "$fname-paragraph" );
2084                                 # No prefix (not in list)--go to paragraph mode
2085                                 // XXX: use a stack for nestable elements like span, table and div
2086                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2087                                 $closematch = preg_match(
2088                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2089                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2090                                 if ( $openmatch or $closematch ) {
2091                                         $paragraphStack = false;
2092                                         # TODO bug 5718: paragraph closed
2093                                         $output .= $this->closeParagraph();
2094                                         if ( $preOpenMatch and !$preCloseMatch ) {
2095                                                 $this->mInPre = true;
2096                                         }
2097                                         if ( $closematch ) {
2098                                                 $inBlockElem = false;
2099                                         } else {
2100                                                 $inBlockElem = true;
2101                                         }
2102                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2103                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2104                                                 // pre
2105                                                 if ($this->mLastSection != 'pre') {
2106                                                         $paragraphStack = false;
2107                                                         $output .= $this->closeParagraph().'<pre>';
2108                                                         $this->mLastSection = 'pre';
2109                                                 }
2110                                                 $t = substr( $t, 1 );
2111                                         } else {
2112                                                 // paragraph
2113                                                 if ( '' == trim($t) ) {
2114                                                         if ( $paragraphStack ) {
2115                                                                 $output .= $paragraphStack.'<br />';
2116                                                                 $paragraphStack = false;
2117                                                                 $this->mLastSection = 'p';
2118                                                         } else {
2119                                                                 if ($this->mLastSection != 'p' ) {
2120                                                                         $output .= $this->closeParagraph();
2121                                                                         $this->mLastSection = '';
2122                                                                         $paragraphStack = '<p>';
2123                                                                 } else {
2124                                                                         $paragraphStack = '</p><p>';
2125                                                                 }
2126                                                         }
2127                                                 } else {
2128                                                         if ( $paragraphStack ) {
2129                                                                 $output .= $paragraphStack;
2130                                                                 $paragraphStack = false;
2131                                                                 $this->mLastSection = 'p';
2132                                                         } else if ($this->mLastSection != 'p') {
2133                                                                 $output .= $this->closeParagraph().'<p>';
2134                                                                 $this->mLastSection = 'p';
2135                                                         }
2136                                                 }
2137                                         }
2138                                 }
2139                                 wfProfileOut( "$fname-paragraph" );
2140                         }
2141                         // somewhere above we forget to get out of pre block (bug 785)
2142                         if($preCloseMatch && $this->mInPre) {
2143                                 $this->mInPre = false;
2144                         }
2145                         if ($paragraphStack === false) {
2146                                 $output .= $t."\n";
2147                         }
2148                 }
2149                 while ( $prefixLength ) {
2150                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2151                         --$prefixLength;
2152                 }
2153                 if ( '' != $this->mLastSection ) {
2154                         $output .= '</' . $this->mLastSection . '>';
2155                         $this->mLastSection = '';
2156                 }
2157
2158                 wfProfileOut( $fname );
2159                 return $output;
2160         }
2161
2162         /**
2163          * Split up a string on ':', ignoring any occurences inside tags
2164          * to prevent illegal overlapping.
2165          * @param string $str the string to split
2166          * @param string &$before set to everything before the ':'
2167          * @param string &$after set to everything after the ':'
2168          * return string the position of the ':', or false if none found
2169          */
2170         function findColonNoLinks($str, &$before, &$after) {
2171                 $fname = 'Parser::findColonNoLinks';
2172                 wfProfileIn( $fname );
2173
2174                 $pos = strpos( $str, ':' );
2175                 if( $pos === false ) {
2176                         // Nothing to find!
2177                         wfProfileOut( $fname );
2178                         return false;
2179                 }
2180
2181                 $lt = strpos( $str, '<' );
2182                 if( $lt === false || $lt > $pos ) {
2183                         // Easy; no tag nesting to worry about
2184                         $before = substr( $str, 0, $pos );
2185                         $after = substr( $str, $pos+1 );
2186                         wfProfileOut( $fname );
2187                         return $pos;
2188                 }
2189
2190                 // Ugly state machine to walk through avoiding tags.
2191                 $state = self::COLON_STATE_TEXT;
2192                 $stack = 0;
2193                 $len = strlen( $str );
2194                 for( $i = 0; $i < $len; $i++ ) {
2195                         $c = $str{$i};
2196
2197                         switch( $state ) {
2198                         // (Using the number is a performance hack for common cases)
2199                         case 0: // self::COLON_STATE_TEXT:
2200                                 switch( $c ) {
2201                                 case "<":
2202                                         // Could be either a <start> tag or an </end> tag
2203                                         $state = self::COLON_STATE_TAGSTART;
2204                                         break;
2205                                 case ":":
2206                                         if( $stack == 0 ) {
2207                                                 // We found it!
2208                                                 $before = substr( $str, 0, $i );
2209                                                 $after = substr( $str, $i + 1 );
2210                                                 wfProfileOut( $fname );
2211                                                 return $i;
2212                                         }
2213                                         // Embedded in a tag; don't break it.
2214                                         break;
2215                                 default:
2216                                         // Skip ahead looking for something interesting
2217                                         $colon = strpos( $str, ':', $i );
2218                                         if( $colon === false ) {
2219                                                 // Nothing else interesting
2220                                                 wfProfileOut( $fname );
2221                                                 return false;
2222                                         }
2223                                         $lt = strpos( $str, '<', $i );
2224                                         if( $stack === 0 ) {
2225                                                 if( $lt === false || $colon < $lt ) {
2226                                                         // We found it!
2227                                                         $before = substr( $str, 0, $colon );
2228                                                         $after = substr( $str, $colon + 1 );
2229                                                         wfProfileOut( $fname );
2230                                                         return $i;
2231                                                 }
2232                                         }
2233                                         if( $lt === false ) {
2234                                                 // Nothing else interesting to find; abort!
2235                                                 // We're nested, but there's no close tags left. Abort!
2236                                                 break 2;
2237                                         }
2238                                         // Skip ahead to next tag start
2239                                         $i = $lt;
2240                                         $state = self::COLON_STATE_TAGSTART;
2241                                 }
2242                                 break;
2243                         case 1: // self::COLON_STATE_TAG:
2244                                 // In a <tag>
2245                                 switch( $c ) {
2246                                 case ">":
2247                                         $stack++;
2248                                         $state = self::COLON_STATE_TEXT;
2249                                         break;
2250                                 case "/":
2251                                         // Slash may be followed by >?
2252                                         $state = self::COLON_STATE_TAGSLASH;
2253                                         break;
2254                                 default:
2255                                         // ignore
2256                                 }
2257                                 break;
2258                         case 2: // self::COLON_STATE_TAGSTART:
2259                                 switch( $c ) {
2260                                 case "/":
2261                                         $state = self::COLON_STATE_CLOSETAG;
2262                                         break;
2263                                 case "!":
2264                                         $state = self::COLON_STATE_COMMENT;
2265                                         break;
2266                                 case ">":
2267                                         // Illegal early close? This shouldn't happen D:
2268                                         $state = self::COLON_STATE_TEXT;
2269                                         break;
2270                                 default:
2271                                         $state = self::COLON_STATE_TAG;
2272                                 }
2273                                 break;
2274                         case 3: // self::COLON_STATE_CLOSETAG:
2275                                 // In a </tag>
2276                                 if( $c == ">" ) {
2277                                         $stack--;
2278                                         if( $stack < 0 ) {
2279                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2280                                                 wfProfileOut( $fname );
2281                                                 return false;
2282                                         }
2283                                         $state = self::COLON_STATE_TEXT;
2284                                 }
2285                                 break;
2286                         case self::COLON_STATE_TAGSLASH:
2287                                 if( $c == ">" ) {
2288                                         // Yes, a self-closed tag <blah/>
2289                                         $state = self::COLON_STATE_TEXT;
2290                                 } else {
2291                                         // Probably we're jumping the gun, and this is an attribute
2292                                         $state = self::COLON_STATE_TAG;
2293                                 }
2294                                 break;
2295                         case 5: // self::COLON_STATE_COMMENT:
2296                                 if( $c == "-" ) {
2297                                         $state = self::COLON_STATE_COMMENTDASH;
2298                                 }
2299                                 break;
2300                         case self::COLON_STATE_COMMENTDASH:
2301                                 if( $c == "-" ) {
2302                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2303                                 } else {
2304                                         $state = self::COLON_STATE_COMMENT;
2305                                 }
2306                                 break;
2307                         case self::COLON_STATE_COMMENTDASHDASH:
2308                                 if( $c == ">" ) {
2309                                         $state = self::COLON_STATE_TEXT;
2310                                 } else {
2311                                         $state = self::COLON_STATE_COMMENT;
2312                                 }
2313                                 break;
2314                         default:
2315                                 throw new MWException( "State machine error in $fname" );
2316                         }
2317                 }
2318                 if( $stack > 0 ) {
2319                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2320                         return false;
2321                 }
2322                 wfProfileOut( $fname );
2323                 return false;
2324         }
2325
2326         /**
2327          * Return value of a magic variable (like PAGENAME)
2328          *
2329          * @private
2330          */
2331         function getVariableValue( $index ) {
2332                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2333
2334                 /**
2335                  * Some of these require message or data lookups and can be
2336                  * expensive to check many times.
2337                  */
2338                 static $varCache = array();
2339                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2340                         if ( isset( $varCache[$index] ) ) {
2341                                 return $varCache[$index];
2342                         }
2343                 }
2344
2345                 $ts = time();
2346                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2347
2348                 # Use the time zone
2349                 global $wgLocaltimezone;
2350                 if ( isset( $wgLocaltimezone ) ) {
2351                         $oldtz = getenv( 'TZ' );
2352                         putenv( 'TZ='.$wgLocaltimezone );
2353                 }
2354
2355                 wfSuppressWarnings(); // E_STRICT system time bitching
2356                 $localTimestamp = date( 'YmdHis', $ts );
2357                 $localMonth = date( 'm', $ts );
2358                 $localMonthName = date( 'n', $ts );
2359                 $localDay = date( 'j', $ts );
2360                 $localDay2 = date( 'd', $ts );
2361                 $localDayOfWeek = date( 'w', $ts );
2362                 $localWeek = date( 'W', $ts );
2363                 $localYear = date( 'Y', $ts );
2364                 $localHour = date( 'H', $ts );
2365                 if ( isset( $wgLocaltimezone ) ) {
2366                         putenv( 'TZ='.$oldtz );
2367                 }
2368                 wfRestoreWarnings();
2369
2370                 switch ( $index ) {
2371                         case 'currentmonth':
2372                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2373                         case 'currentmonthname':
2374                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2375                         case 'currentmonthnamegen':
2376                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2377                         case 'currentmonthabbrev':
2378                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2379                         case 'currentday':
2380                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2381                         case 'currentday2':
2382                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2383                         case 'localmonth':
2384                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2385                         case 'localmonthname':
2386                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2387                         case 'localmonthnamegen':
2388                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2389                         case 'localmonthabbrev':
2390                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2391                         case 'localday':
2392                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2393                         case 'localday2':
2394                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2395                         case 'pagename':
2396                                 return wfEscapeWikiText( $this->mTitle->getText() );
2397                         case 'pagenamee':
2398                                 return $this->mTitle->getPartialURL();
2399                         case 'fullpagename':
2400                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2401                         case 'fullpagenamee':
2402                                 return $this->mTitle->getPrefixedURL();
2403                         case 'subpagename':
2404                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2405                         case 'subpagenamee':
2406                                 return $this->mTitle->getSubpageUrlForm();
2407                         case 'basepagename':
2408                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2409                         case 'basepagenamee':
2410                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2411                         case 'talkpagename':
2412                                 if( $this->mTitle->canTalk() ) {
2413                                         $talkPage = $this->mTitle->getTalkPage();
2414                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2415                                 } else {
2416                                         return '';
2417                                 }
2418                         case 'talkpagenamee':
2419                                 if( $this->mTitle->canTalk() ) {
2420                                         $talkPage = $this->mTitle->getTalkPage();
2421                                         return $talkPage->getPrefixedUrl();
2422                                 } else {
2423                                         return '';
2424                                 }
2425                         case 'subjectpagename':
2426                                 $subjPage = $this->mTitle->getSubjectPage();
2427                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2428                         case 'subjectpagenamee':
2429                                 $subjPage = $this->mTitle->getSubjectPage();
2430                                 return $subjPage->getPrefixedUrl();
2431                         case 'revisionid':
2432                                 // Let the edit saving system know we should parse the page
2433                                 // *after* a revision ID has been assigned.
2434                                 $this->mOutput->setFlag( 'vary-revision' );
2435                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2436                                 return $this->mRevisionId;
2437                         case 'revisionday':
2438                                 // Let the edit saving system know we should parse the page
2439                                 // *after* a revision ID has been assigned. This is for null edits.
2440                                 $this->mOutput->setFlag( 'vary-revision' );
2441                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2442                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2443                         case 'revisionday2':
2444                                 // Let the edit saving system know we should parse the page
2445                                 // *after* a revision ID has been assigned. This is for null edits.
2446                                 $this->mOutput->setFlag( 'vary-revision' );
2447                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2448                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2449                         case 'revisionmonth':
2450                                 // Let the edit saving system know we should parse the page
2451                                 // *after* a revision ID has been assigned. This is for null edits.
2452                                 $this->mOutput->setFlag( 'vary-revision' );
2453                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2454                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2455                         case 'revisionyear':
2456                                 // Let the edit saving system know we should parse the page
2457                                 // *after* a revision ID has been assigned. This is for null edits.
2458                                 $this->mOutput->setFlag( 'vary-revision' );
2459                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2460                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2461                         case 'revisiontimestamp':
2462                                 // Let the edit saving system know we should parse the page
2463                                 // *after* a revision ID has been assigned. This is for null edits.
2464                                 $this->mOutput->setFlag( 'vary-revision' );
2465                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2466                                 return $this->getRevisionTimestamp();
2467                         case 'namespace':
2468                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2469                         case 'namespacee':
2470                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2471                         case 'talkspace':
2472                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2473                         case 'talkspacee':
2474                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2475                         case 'subjectspace':
2476                                 return $this->mTitle->getSubjectNsText();
2477                         case 'subjectspacee':
2478                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2479                         case 'currentdayname':
2480                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2481                         case 'currentyear':
2482                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2483                         case 'currenttime':
2484                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2485                         case 'currenthour':
2486                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2487                         case 'currentweek':
2488                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2489                                 // int to remove the padding
2490                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2491                         case 'currentdow':
2492                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2493                         case 'localdayname':
2494                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2495                         case 'localyear':
2496                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2497                         case 'localtime':
2498                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2499                         case 'localhour':
2500                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2501                         case 'localweek':
2502                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2503                                 // int to remove the padding
2504                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2505                         case 'localdow':
2506                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2507                         case 'numberofarticles':
2508                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2509                         case 'numberoffiles':
2510                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2511                         case 'numberofusers':
2512                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2513                         case 'numberofpages':
2514                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2515                         case 'numberofadmins':
2516                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2517                         case 'numberofedits':
2518                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2519                         case 'currenttimestamp':
2520                                 return $varCache[$index] = wfTimestampNow();
2521                         case 'localtimestamp':
2522                                 return $varCache[$index] = $localTimestamp;
2523                         case 'currentversion':
2524                                 return $varCache[$index] = SpecialVersion::getVersion();
2525                         case 'sitename':
2526                                 return $wgSitename;
2527                         case 'server':
2528                                 return $wgServer;
2529                         case 'servername':
2530                                 return $wgServerName;
2531                         case 'scriptpath':
2532                                 return $wgScriptPath;
2533                         case 'directionmark':
2534                                 return $wgContLang->getDirMark();
2535                         case 'contentlanguage':
2536                                 global $wgContLanguageCode;
2537                                 return $wgContLanguageCode;
2538                         default:
2539                                 $ret = null;
2540                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2541                                         return $ret;
2542                                 else
2543                                         return null;
2544                 }
2545         }
2546
2547         /**
2548          * initialise the magic variables (like CURRENTMONTHNAME)
2549          *
2550          * @private
2551          */
2552         function initialiseVariables() {
2553                 $fname = 'Parser::initialiseVariables';
2554                 wfProfileIn( $fname );
2555                 $variableIDs = MagicWord::getVariableIDs();
2556
2557                 $this->mVariables = new MagicWordArray( $variableIDs );
2558                 wfProfileOut( $fname );
2559         }
2560
2561         /**
2562          * Preprocess some wikitext and return the document tree.
2563          * This is the ghost of replace_variables().
2564          *
2565          * @param string $text The text to parse
2566          * @param integer flags Bitwise combination of:
2567          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2568          *                                     included. Default is to assume a direct page view.
2569          *
2570          * The generated DOM tree must depend only on the input text, the flags, and $this->ot['msg'].
2571          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2572          *
2573          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2574          * change in the DOM tree for a given text, must be passed through the section identifier
2575          * in the section edit link and thus back to extractSections().
2576          *
2577          * The output of this function is currently only cached in process memory, but a persistent
2578          * cache may be implemented at a later date which takes further advantage of these strict
2579          * dependency requirements.
2580          *
2581          * @private
2582          */
2583         function preprocessToDom ( $text, $flags = 0 ) {
2584                 wfProfileIn( __METHOD__ );
2585                 wfProfileIn( __METHOD__.'-makexml' );
2586
2587                 static $msgRules, $normalRules, $inclusionSupertags, $nonInclusionSupertags;
2588                 if ( !$msgRules ) {
2589                         $msgRules = array(
2590                                 '{' => array(
2591                                         'end' => '}',
2592                                         'names' => array(
2593                                                 2 => 'template',
2594                                         ),
2595                                         'min' => 2,
2596                                         'max' => 2,
2597                                 ),
2598                                 '[' => array(
2599                                         'end' => ']',
2600                                         'names' => array( 2 => null ),
2601                                         'min' => 2,
2602                                         'max' => 2,
2603                                 )
2604                         );
2605                         $normalRules = array(
2606                                 '{' => array(
2607                                         'end' => '}',
2608                                         'names' => array(
2609                                                 2 => 'template',
2610                                                 3 => 'tplarg',
2611                                         ),
2612                                         'min' => 2,
2613                                         'max' => 3,
2614                                 ),
2615                                 '[' => array(
2616                                         'end' => ']',
2617                                         'names' => array( 2 => null ),
2618                                         'min' => 2,
2619                                         'max' => 2,
2620                                 )
2621                         );
2622                 }
2623                 if ( $this->ot['msg'] ) {
2624                         $rules = $msgRules;
2625                 } else {
2626                         $rules = $normalRules;
2627                 }
2628                 $forInclusion = $flags & self::PTD_FOR_INCLUSION;
2629
2630                 $xmlishElements = $this->getStripList();
2631                 $enableOnlyinclude = false;
2632                 if ( $forInclusion ) {
2633                         $ignoredTags = array( 'includeonly', '/includeonly' );
2634                         $ignoredElements = array( 'noinclude' );
2635                         $xmlishElements[] = 'noinclude';
2636                         if ( strpos( $text, '<onlyinclude>' ) !== false && strpos( $text, '</onlyinclude>' ) !== false ) {
2637                                 $enableOnlyinclude = true;
2638                         }
2639                 } else {
2640                         $ignoredTags = array( 'noinclude', '/noinclude', 'onlyinclude', '/onlyinclude' );
2641                         $ignoredElements = array( 'includeonly' );
2642                         $xmlishElements[] = 'includeonly';
2643                 }
2644                 $xmlishRegex = implode( '|', array_merge( $xmlishElements, $ignoredTags ) );
2645
2646                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2647                 $elementsRegex = "~($xmlishRegex)(?:\s|\/>|>)|(!--)~iA";
2648
2649                 $stack = array();      # Stack of unclosed parentheses
2650                 $stackIndex = -1;      # Stack read pointer
2651
2652                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2653                 $revText = strrev( $text ); // For fast reverse searches
2654
2655                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2656                 $topAccum = '<root>';      # Top level text accumulator
2657                 $accum =& $topAccum;            # Current text accumulator
2658                 $findEquals = false;            # True to find equals signs in arguments
2659                 $findHeading = false;           # True to look at LF characters for possible headings
2660                 $findPipe = false;              # True to take notice of pipe characters
2661                 $headingIndex = 1;
2662                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2663                 $findOnlyinclude = $enableOnlyinclude; # True to ignore all input up to the next <onlyinclude>
2664
2665                 if ( $enableOnlyinclude ) {
2666                         $i = 0;
2667                 }
2668
2669                 while ( true ) {
2670                         if ( $findOnlyinclude ) {
2671                                 // Ignore all input up to the next <onlyinclude>
2672                                 $startPos = strpos( $text, '<onlyinclude>', $i );
2673                                 if ( $startPos === false ) {
2674                                         // Ignored section runs to the end
2675                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i ) ) . '</ignore>';
2676                                         break;
2677                                 }
2678                                 $tagEndPos = $startPos + strlen( '<onlyinclude>' ); // past-the-end
2679                                 $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i ) ) . '</ignore>';
2680                                 $i = $tagEndPos;
2681                                 $findOnlyinclude = false;
2682                         }
2683
2684                         if ( $i == -1 ) {
2685                                 $found = 'line-start';
2686                                 $curChar = '';
2687                         } else {
2688                                 # Find next opening brace, closing brace or pipe
2689                                 $search = $searchBase;
2690                                 if ( $stackIndex == -1 ) {
2691                                         $currentClosing = '';
2692                                         // Look for headings only at the top stack level
2693                                         // Among other things, this resolves the ambiguity between =
2694                                         // for headings and = for template arguments
2695                                         $search .= "\n";
2696                                 } else {
2697                                         $currentClosing = $stack[$stackIndex]['close'];
2698                                         $search .= $currentClosing;
2699                                 }
2700                                 if ( $findPipe ) {
2701                                         $search .= '|';
2702                                 }
2703                                 if ( $findEquals ) {
2704                                         $search .= '=';
2705                                 }
2706                                 $rule = null;
2707                                 # Output literal section, advance input counter
2708                                 $literalLength = strcspn( $text, $search, $i );
2709                                 if ( $literalLength > 0 ) {
2710                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2711                                         $i += $literalLength;
2712                                 }
2713                                 if ( $i >= strlen( $text ) ) {
2714                                         if ( $currentClosing == "\n" ) {
2715                                                 // Do a past-the-end run to finish off the heading
2716                                                 $curChar = '';
2717                                                 $found = 'line-end';
2718                                         } else {
2719                                                 # All done
2720                                                 break;
2721                                         }
2722                                 } else {
2723                                         $curChar = $text[$i];
2724                                         if ( $curChar == '|' ) {
2725                                                 $found = 'pipe';
2726                                         } elseif ( $curChar == '=' ) {
2727                                                 $found = 'equals';
2728                                         } elseif ( $curChar == '<' ) {
2729                                                 $found = 'angle';
2730                                         } elseif ( $curChar == "\n" ) {
2731                                                 if ( $stackIndex == -1 ) {
2732                                                         $found = 'line-start';
2733                                                 } else {
2734                                                         $found = 'line-end';
2735                                                 }
2736                                         } elseif ( $curChar == $currentClosing ) {
2737                                                 $found = 'close';
2738                                         } elseif ( isset( $rules[$curChar] ) ) {
2739                                                 $found = 'open';
2740                                                 $rule = $rules[$curChar];
2741                                         } else {
2742                                                 # Some versions of PHP have a strcspn which stops on null characters
2743                                                 # Ignore and continue
2744                                                 ++$i;
2745                                                 continue;
2746                                         }
2747                                 }
2748                         }
2749
2750                         if ( $found == 'angle' ) {
2751                                 $matches = false;
2752                                 // Handle </onlyinclude>
2753                                 if ( $enableOnlyinclude && substr( $text, $i, strlen( '</onlyinclude>' ) ) == '</onlyinclude>' ) {
2754                                         $findOnlyinclude = true;
2755                                         continue;
2756                                 }
2757
2758                                 // Determine element name
2759                                 if ( !preg_match( $elementsRegex, $text, $matches, 0, $i + 1 ) ) {
2760                                         // Element name missing or not listed
2761                                         $accum .= '&lt;';
2762                                         ++$i;
2763                                         continue;
2764                                 }
2765                                 // Handle comments
2766                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2767                                         // To avoid leaving blank lines, when a comment is both preceded
2768                                         // and followed by a newline (ignoring spaces), trim leading and
2769                                         // trailing spaces and one of the newlines.
2770
2771                                         // Find the end
2772                                         $endPos = strpos( $text, '-->', $i + 4 );
2773                                         if ( $endPos === false ) {
2774                                                 // Unclosed comment in input, runs to end
2775                                                 $inner = substr( $text, $i );
2776                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2777                                                 $i = strlen( $text );
2778                                         } else {
2779                                                 // Search backwards for leading whitespace
2780                                                 $wsStart = $i ? ( $i - strspn( $revText, ' ', strlen( $text ) - $i ) ) : 0;
2781                                                 // Search forwards for trailing whitespace
2782                                                 // $wsEnd will be the position of the last space
2783                                                 $wsEnd = $endPos + 2 + strspn( $text, ' ', $endPos + 3 );
2784                                                 // Eat the line if possible
2785                                                 if ( $wsStart > 0 && substr( $text, $wsStart - 1, 1 ) == "\n"
2786                                                         && substr( $text, $wsEnd + 1, 1 ) == "\n" )
2787                                                 {
2788                                                         $startPos = $wsStart;
2789                                                         $endPos = $wsEnd + 1;
2790                                                         // Remove leading whitespace from the end of the accumulator
2791                                                         // Sanity check first though
2792                                                         $wsLength = $i - $wsStart;
2793                                                         if ( $wsLength > 0 && substr( $accum, -$wsLength ) === str_repeat( ' ', $wsLength ) ) {
2794                                                                 $accum = substr( $accum, 0, -$wsLength );
2795                                                         }
2796                                                 } else {
2797                                                         // No line to eat, just take the comment itself
2798                                                         $startPos = $i;
2799                                                         $endPos += 2;
2800                                                 }
2801
2802                                                 $inner = substr( $text, $startPos, $endPos - $startPos + 1 );
2803                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2804                                                 $i = $endPos + 1;
2805                                         }
2806                                         continue;
2807                                 }
2808                                 $name = $matches[1];
2809                                 $attrStart = $i + strlen( $name ) + 1;
2810
2811                                 // Find end of tag
2812                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2813                                 if ( $tagEndPos === false ) {
2814                                         // Infinite backtrack
2815                                         // Disable tag search to prevent worst-case O(N^2) performance
2816                                         $noMoreGT = true;
2817                                         $accum .= '&lt;';
2818                                         ++$i;
2819                                         continue;
2820                                 }
2821
2822                                 // Handle ignored tags
2823                                 if ( in_array( $name, $ignoredTags ) ) {
2824                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $i, $tagEndPos - $i + 1 ) ) . '</ignore>';
2825                                         $i = $tagEndPos + 1;
2826                                         continue;
2827                                 }
2828
2829                                 $tagStartPos = $i;
2830                                 if ( $text[$tagEndPos-1] == '/' ) {
2831                                         $attrEnd = $tagEndPos - 1;
2832                                         $inner = null;
2833                                         $i = $tagEndPos + 1;
2834                                         $close = '';
2835                                 } else {
2836                                         $attrEnd = $tagEndPos;
2837                                         // Find closing tag
2838                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2839                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2840                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2841                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2842                                         } else {
2843                                                 // No end tag -- let it run out to the end of the text.
2844                                                 $inner = substr( $text, $tagEndPos + 1 );
2845                                                 $i = strlen( $text );
2846                                                 $close = '';
2847                                         }
2848                                 }
2849                                 // <includeonly> and <noinclude> just become <ignore> tags
2850                                 if ( in_array( $name, $ignoredElements ) ) {
2851                                         $accum .= '<ignore>' . htmlspecialchars( substr( $text, $tagStartPos, $i - $tagStartPos ) )
2852                                                 . '</ignore>';
2853                                         continue;
2854                                 }
2855
2856                                 $accum .= '<ext>';
2857                                 if ( $attrEnd <= $attrStart ) {
2858                                         $attr = '';
2859                                 } else {
2860                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2861                                 }
2862                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2863                                         // Note that the attr element contains the whitespace between name and attribute,
2864                                         // this is necessary for precise reconstruction during pre-save transform.
2865                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2866                                 if ( $inner !== null ) {
2867                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2868                                 }
2869                                 $accum .= $close . '</ext>';
2870                         }
2871
2872                         elseif ( $found == 'line-start' ) {
2873                                 // Is this the start of a heading?
2874                                 // Line break belongs before the heading element in any case
2875                                 $accum .= $curChar;
2876                                 $i++;
2877
2878                                 $count = strspn( $text, '=', $i, 6 );
2879                                 if ( $count > 0 ) {
2880                                         $piece = array(
2881                                                 'open' => "\n",
2882                                                 'close' => "\n",
2883                                                 'parts' => array( str_repeat( '=', $count ) ),
2884                                                 'count' => $count );
2885                                         $stack[++$stackIndex] = $piece;
2886                                         $i += $count;
2887                                         $accum =& $stack[$stackIndex]['parts'][0];
2888                                         $findPipe = false;
2889                                 }
2890                         }
2891
2892                         elseif ( $found == 'line-end' ) {
2893                                 $piece = $stack[$stackIndex];
2894                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2895                                 assert( $piece['open'] == "\n" );
2896                                 assert( $stackIndex == 0 );
2897                                 // Search back through the input to see if it has a proper close
2898                                 // Do this using the reversed string since the other solutions (end anchor, etc.) are inefficient
2899                                 $m = false;
2900                                 $count = $piece['count'];
2901                                 if ( preg_match( "/\s*(={{$count}})/A", $revText, $m, 0, strlen( $text ) - $i ) ) {
2902                                         // Found match, output <h>
2903                                         $count = min( strlen( $m[1] ), $count );
2904                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2905                                         $headingIndex++;
2906                                 } else {
2907                                         // No match, no <h>, just pass down the inner text
2908                                         $element = $accum;
2909                                 }
2910                                 // Unwind the stack
2911                                 // Headings can only occur on the top level, so this is a bit simpler than the
2912                                 // generic stack unwind operation in the close case
2913                                 unset( $stack[$stackIndex--] );
2914                                 $accum =& $topAccum;
2915                                 $findEquals = false;
2916                                 $findPipe = false;
2917
2918                                 // Append the result to the enclosing accumulator
2919                                 $accum .= $element;
2920                                 // Note that we do NOT increment the input pointer.
2921                                 // This is because the closing linebreak could be the opening linebreak of
2922                                 // another heading. Infinite loops are avoided because the next iteration MUST
2923                                 // hit the heading open case above, which unconditionally increments the
2924                                 // input pointer.
2925                         }
2926
2927                         elseif ( $found == 'open' ) {
2928                                 # count opening brace characters
2929                                 $count = strspn( $text, $curChar, $i );
2930
2931                                 # we need to add to stack only if opening brace count is enough for one of the rules
2932                                 if ( $count >= $rule['min'] ) {
2933                                         # Add it to the stack
2934                                         $piece = array(
2935                                                 'open' => $curChar,
2936                                                 'close' => $rule['end'],
2937                                                 'count' => $count,
2938                                                 'parts' => array( '' ),
2939                                                 'eqpos' => array(),
2940                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2941                                         );
2942
2943                                         $stackIndex ++;
2944                                         $stack[$stackIndex] = $piece;
2945                                         $accum =& $stack[$stackIndex]['parts'][0];
2946                                         $findEquals = false;
2947                                         $findPipe = true;
2948                                 } else {
2949                                         # Add literal brace(s)
2950                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2951                                 }
2952                                 $i += $count;
2953                         }
2954
2955                         elseif ( $found == 'close' ) {
2956                                 $piece = $stack[$stackIndex];
2957                                 # lets check if there are enough characters for closing brace
2958                                 $maxCount = $piece['count'];
2959                                 $count = strspn( $text, $curChar, $i, $maxCount );
2960
2961                                 # check for maximum matching characters (if there are 5 closing
2962                                 # characters, we will probably need only 3 - depending on the rules)
2963                                 $matchingCount = 0;
2964                                 $rule = $rules[$piece['open']];
2965                                 if ( $count > $rule['max'] ) {
2966                                         # The specified maximum exists in the callback array, unless the caller
2967                                         # has made an error
2968                                         $matchingCount = $rule['max'];
2969                                 } else {
2970                                         # Count is less than the maximum
2971                                         # Skip any gaps in the callback array to find the true largest match
2972                                         # Need to use array_key_exists not isset because the callback can be null
2973                                         $matchingCount = $count;
2974                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2975                                                 --$matchingCount;
2976                                         }
2977                                 }
2978
2979                                 if ($matchingCount <= 0) {
2980                                         # No matching element found in callback array
2981                                         # Output a literal closing brace and continue
2982                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2983                                         $i += $count;
2984                                         continue;
2985                                 }
2986                                 $name = $rule['names'][$matchingCount];
2987                                 if ( $name === null ) {
2988                                         // No element, just literal text
2989                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2990                                                 implode( '|', $piece['parts'] ) .
2991                                                 str_repeat( $rule['end'], $matchingCount );
2992                                 } else {
2993                                         # Create XML element
2994                                         # Note: $parts is already XML, does not need to be encoded further
2995                                         $parts = $piece['parts'];
2996                                         $title = $parts[0];
2997                                         unset( $parts[0] );
2998
2999                                         # The invocation is at the start of the line if lineStart is set in
3000                                         # the stack, and all opening brackets are used up.
3001                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
3002                                                 $attr = ' lineStart="1"';
3003                                         } else {
3004                                                 $attr = '';
3005                                         }
3006
3007                                         $element = "<$name$attr>";
3008                                         $element .= "<title>$title</title>";
3009                                         $argIndex = 1;
3010                                         foreach ( $parts as $partIndex => $part ) {
3011                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
3012                                                         $eqpos = $piece['eqpos'][$partIndex];
3013                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
3014                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
3015                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
3016                                                 } else {
3017                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
3018                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
3019                                                         $argIndex++;
3020                                                 }
3021                                         }
3022                                         $element .= "</$name>";
3023                                 }
3024
3025                                 # Advance input pointer
3026                                 $i += $matchingCount;
3027
3028                                 # Unwind the stack
3029                                 unset( $stack[$stackIndex--] );
3030                                 if ( $stackIndex == -1 ) {
3031                                         $accum =& $topAccum;
3032                                         $findEquals = false;
3033                                         $findPipe = false;
3034                                 } else {
3035                                         $partCount = count( $stack[$stackIndex]['parts'] );
3036                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
3037                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
3038                                         $findEquals = $findPipe && $partCount > 1
3039                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
3040                                 }
3041
3042                                 # Re-add the old stack element if it still has unmatched opening characters remaining
3043                                 if ($matchingCount < $piece['count']) {
3044                                         $piece['parts'] = array( '' );
3045                                         $piece['count'] -= $matchingCount;
3046                                         $piece['eqpos'] = array();
3047                                         # do we still qualify for any callback with remaining count?
3048                                         $names = $rules[$piece['open']]['names'];
3049                                         $skippedBraces = 0;
3050                                         $enclosingAccum =& $accum;
3051                                         while ( $piece['count'] ) {
3052                                                 if ( array_key_exists( $piece['count'], $names ) ) {
3053                                                         $stackIndex++;
3054                                                         $stack[$stackIndex] = $piece;
3055                                                         $accum =& $stack[$stackIndex]['parts'][0];
3056                                                         $findEquals = true;
3057                                                         $findPipe = true;
3058                                                         break;
3059                                                 }
3060                                                 --$piece['count'];
3061                                                 $skippedBraces ++;
3062                                         }
3063                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
3064                                 }
3065
3066                                 # Add XML element to the enclosing accumulator
3067                                 $accum .= $element;
3068                         }
3069
3070                         elseif ( $found == 'pipe' ) {
3071                                 $stack[$stackIndex]['parts'][] = '';
3072                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3073                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
3074                                 $findEquals = true;
3075                                 ++$i;
3076                         }
3077
3078                         elseif ( $found == 'equals' ) {
3079                                 $findEquals = false;
3080                                 $partsCount = count( $stack[$stackIndex]['parts'] );
3081                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
3082                                 $accum .= '=';
3083                                 ++$i;
3084                         }
3085                 }
3086
3087                 # Output any remaining unclosed brackets
3088                 foreach ( $stack as $piece ) {
3089                         if ( $piece['open'] == "\n" ) {
3090                                 $topAccum .= $piece['parts'][0];
3091                         } else {
3092                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
3093                         }
3094                 }
3095                 $topAccum .= '</root>';
3096
3097                 wfProfileOut( __METHOD__.'-makexml' );
3098                 wfProfileIn( __METHOD__.'-loadXML' );
3099                 $dom = new DOMDocument;
3100                 wfSuppressWarnings();
3101                 $result = $dom->loadXML( $topAccum );
3102                 wfRestoreWarnings();
3103                 if ( !$result ) {
3104                         // Try running the XML through UtfNormal to get rid of invalid characters
3105                         $topAccum = UtfNormal::cleanUp( $topAccum );
3106                         $result = $dom->loadXML( $topAccum );
3107                         if ( !$result ) {
3108                                 throw new MWException( __METHOD__.' generated invalid XML' );
3109                         }
3110                 }
3111                 wfProfileOut( __METHOD__.'-loadXML' );
3112                 wfProfileOut( __METHOD__ );
3113                 return $dom;
3114         }
3115
3116         /*
3117          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3118          */
3119         public static function splitWhitespace( $s ) {
3120                 $ltrimmed = ltrim( $s );
3121                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3122                 $trimmed = rtrim( $ltrimmed );
3123                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3124                 if ( $diff > 0 ) {
3125                         $w2 = substr( $ltrimmed, -$diff );
3126                 } else {
3127                         $w2 = '';
3128                 }
3129                 return array( $w1, $trimmed, $w2 );
3130         }
3131
3132         /**
3133          * Replace magic variables, templates, and template arguments
3134          * with the appropriate text. Templates are substituted recursively,
3135          * taking care to avoid infinite loops.
3136          *
3137          * Note that the substitution depends on value of $mOutputType:
3138          *  OT_WIKI: only {{subst:}} templates
3139          *  OT_MSG: only magic variables
3140          *  OT_HTML: all templates and magic variables
3141          *
3142          * @param string $tex The text to transform
3143          * @param PPFrame $frame Object describing the arguments passed to the template
3144          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3145          * @private
3146          */
3147         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3148                 # Prevent too big inclusions
3149                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3150                         return $text;
3151                 }
3152
3153                 $fname = __METHOD__;
3154                 wfProfileIn( $fname );
3155
3156                 if ( $frame === false ) {
3157                         $frame = new PPFrame( $this );
3158                 } elseif ( !( $frame instanceof PPFrame ) ) {
3159                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3160                 }
3161
3162                 $dom = $this->preprocessToDom( $text );
3163                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3164                 $text = $frame->expand( $dom, $flags );
3165
3166                 wfProfileOut( $fname );
3167                 return $text;
3168         }
3169
3170         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3171         static function createAssocArgs( $args ) {
3172                 $assocArgs = array();
3173                 $index = 1;
3174                 foreach( $args as $arg ) {
3175                         $eqpos = strpos( $arg, '=' );
3176                         if ( $eqpos === false ) {
3177                                 $assocArgs[$index++] = $arg;
3178                         } else {
3179                                 $name = trim( substr( $arg, 0, $eqpos ) );
3180                                 $value = trim( substr( $arg, $eqpos+1 ) );
3181                                 if ( $value === false ) {
3182                                         $value = '';
3183                                 }
3184                                 if ( $name !== false ) {
3185                                         $assocArgs[$name] = $value;
3186                                 }
3187                         }
3188                 }
3189
3190                 return $assocArgs;
3191         }
3192
3193         /**
3194          * Return the text of a template, after recursively
3195          * replacing any variables or templates within the template.
3196          *
3197          * @param array $piece The parts of the template
3198          *  $piece['text']: matched text
3199          *  $piece['title']: the title, i.e. the part before the |
3200          *  $piece['parts']: the parameter array
3201          * @param PPFrame The current frame, contains template arguments
3202          * @return string the text of the template
3203          * @private
3204          */
3205         function braceSubstitution( $piece, $frame ) {
3206                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3207                 $fname = __METHOD__;
3208                 wfProfileIn( $fname );
3209                 wfProfileIn( __METHOD__.'-setup' );
3210
3211                 # Flags
3212                 $found = false;             # $text has been filled
3213                 $nowiki = false;            # wiki markup in $text should be escaped
3214                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3215                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3216                 $isDOM = false;             # $text is a DOM node needing expansion
3217
3218                 # Title object, where $text came from
3219                 $title = NULL;
3220
3221                 # $part1 is the bit before the first |, and must contain only title characters.
3222                 # Various prefixes will be stripped from it later.
3223                 $titleWithSpaces = $frame->expand( $piece['title'] );
3224                 $part1 = trim( $titleWithSpaces );
3225                 $titleText = false;
3226
3227                 # Original title text preserved for various purposes
3228                 $originalTitle = $part1;
3229
3230                 # $args is a list of argument nodes, starting from index 0, not including $part1
3231                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3232                 wfProfileOut( __METHOD__.'-setup' );
3233
3234                 # SUBST
3235                 wfProfileIn( __METHOD__.'-modifiers' );
3236                 if ( !$found ) {
3237                         $mwSubst =& MagicWord::get( 'subst' );
3238                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3239                                 # One of two possibilities is true:
3240                                 # 1) Found SUBST but not in the PST phase
3241                                 # 2) Didn't find SUBST and in the PST phase
3242                                 # In either case, return without further processing
3243                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3244                                 $found = true;
3245                         }
3246                 }
3247
3248                 # Variables
3249                 if ( !$found && $args->length == 0 ) {
3250                         $id = $this->mVariables->matchStartToEnd( $part1 );
3251                         if ( $id !== false ) {
3252                                 $text = $this->getVariableValue( $id );
3253                                 $this->mOutput->mContainsOldMagic = true;
3254                                 $found = true;
3255                         }
3256                 }
3257
3258                 # MSG, MSGNW and RAW
3259                 if ( !$found ) {
3260                         # Check for MSGNW:
3261                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3262                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3263                                 $nowiki = true;
3264                         } else {
3265                                 # Remove obsolete MSG:
3266                                 $mwMsg =& MagicWord::get( 'msg' );
3267                                 $mwMsg->matchStartAndRemove( $part1 );
3268                         }
3269
3270                         # Check for RAW:
3271                         $mwRaw =& MagicWord::get( 'raw' );
3272                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3273                                 $forceRawInterwiki = true;
3274                         }
3275                 }
3276                 wfProfileOut( __METHOD__.'-modifiers' );
3277
3278                 # Save path level before recursing into functions & templates.
3279                 $lastPathLevel = $this->mTemplatePath;
3280
3281                 # Parser functions
3282                 if ( !$found ) {
3283                         wfProfileIn( __METHOD__ . '-pfunc' );
3284
3285                         $colonPos = strpos( $part1, ':' );
3286                         if ( $colonPos !== false ) {
3287                                 # Case sensitive functions
3288                                 $function = substr( $part1, 0, $colonPos );
3289                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3290                                         $function = $this->mFunctionSynonyms[1][$function];
3291                                 } else {
3292                                         # Case insensitive functions
3293                                         $function = strtolower( $function );
3294                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3295                                                 $function = $this->mFunctionSynonyms[0][$function];
3296                                         } else {
3297                                                 $function = false;
3298                                         }
3299                                 }
3300                                 if ( $function ) {
3301                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3302                                         $initialArgs = array( &$this );
3303                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3304                                         if ( $flags & SFH_OBJECT_ARGS ) {
3305                                                 # Add a frame parameter, and pass the arguments as an array
3306                                                 $allArgs = $initialArgs;
3307                                                 $allArgs[] = $frame;
3308                                                 foreach ( $args as $arg ) {
3309                                                         $funcArgs[] = $arg;
3310                                                 }
3311                                                 $allArgs[] = $funcArgs;
3312                                         } else {
3313                                                 # Convert arguments to plain text
3314                                                 foreach ( $args as $arg ) {
3315                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3316                                                 }
3317                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3318                                         }
3319
3320                                         $result = call_user_func_array( $callback, $allArgs );
3321                                         $found = true;
3322
3323                                         if ( is_array( $result ) ) {
3324                                                 if ( isset( $result[0] ) ) {
3325                                                         $text = $result[0];
3326                                                         unset( $result[0] );
3327                                                 }
3328
3329                                                 // Extract flags into the local scope
3330                                                 // This allows callers to set flags such as nowiki, found, etc.
3331                                                 extract( $result );
3332                                         } else {
3333                                                 $text = $result;
3334                                         }
3335                                 }
3336                         }
3337                         wfProfileOut( __METHOD__ . '-pfunc' );
3338                 }
3339
3340                 # Finish mangling title and then check for loops.
3341                 # Set $title to a Title object and $titleText to the PDBK
3342                 if ( !$found ) {
3343                         $ns = NS_TEMPLATE;
3344                         # Split the title into page and subpage
3345                         $subpage = '';
3346                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3347                         if ($subpage !== '') {
3348                                 $ns = $this->mTitle->getNamespace();
3349                         }
3350                         $title = Title::newFromText( $part1, $ns );
3351                         if ( $title ) {
3352                                 $titleText = $title->getPrefixedText();
3353                                 # Check for language variants if the template is not found
3354                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3355                                         $wgContLang->findVariantLink($part1, $title);
3356                                 }
3357                                 # Do infinite loop check
3358                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3359                                         $found = true;
3360                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3361                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3362                                 }
3363                         }
3364                 }
3365
3366                 # Load from database
3367                 if ( !$found && $title ) {
3368                         wfProfileIn( __METHOD__ . '-loadtpl' );
3369                         if ( !$title->isExternal() ) {
3370                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3371                                         $text = SpecialPage::capturePath( $title );
3372                                         if ( is_string( $text ) ) {
3373                                                 $found = true;
3374                                                 $isHTML = true;
3375                                                 $this->disableCache();
3376                                         }
3377                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3378                                         $found = false; //access denied
3379                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3380                                 } else {
3381                                         list( $text, $title ) = $this->getTemplateDom( $title );
3382                                         if ( $text !== false ) {
3383                                                 $found = true;
3384                                                 $isDOM = true;
3385                                         }
3386                                 }
3387
3388                                 # If the title is valid but undisplayable, make a link to it
3389                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3390                                         $text = "[[:$titleText]]";
3391                                         $found = true;
3392                                 }
3393                         } elseif ( $title->isTrans() ) {
3394                                 // Interwiki transclusion
3395                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3396                                         $text = $this->interwikiTransclude( $title, 'render' );
3397                                         $isHTML = true;
3398                                 } else {
3399                                         $text = $this->interwikiTransclude( $title, 'raw' );
3400                                         // Preprocess it like a template
3401                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3402                                         $isDOM = true;
3403                                 }
3404                                 $found = true;
3405                         }
3406                         wfProfileOut( __METHOD__ . '-loadtpl' );
3407                 }
3408
3409                 # If we haven't found text to substitute by now, we're done
3410                 # Recover the source wikitext and return it
3411                 if ( !$found ) {
3412                         $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3413                         # Prune lower levels off the recursion check path
3414                         $this->mTemplatePath = $lastPathLevel;
3415                         wfProfileOut( $fname );
3416                         return $text;
3417                 }
3418
3419                 # Expand DOM-style return values in a child frame
3420                 if ( $isDOM ) {
3421                         # Clean up argument array
3422                         $newFrame = $frame->newChild( $args, $title );
3423                         # Add a new element to the templace recursion path
3424                         $this->mTemplatePath[$titleText] = 1;
3425
3426                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3427                                 # Expansion is eligible for the empty-frame cache
3428                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3429                                         $text = $this->mTplExpandCache[$titleText];
3430                                 } else {
3431                                         $text = $newFrame->expand( $text );
3432                                         $this->mTplExpandCache[$titleText] = $text;
3433                                 }
3434                         } else {
3435                                 $text = $newFrame->expand( $text );
3436                         }
3437                 }
3438
3439                 # Replace raw HTML by a placeholder
3440                 # Add a blank line preceding, to prevent it from mucking up
3441                 # immediately preceding headings
3442                 if ( $isHTML ) {
3443                         $text = "\n\n" . $this->insertStripItem( $text );
3444                 }
3445                 # Escape nowiki-style return values
3446                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3447                         $text = wfEscapeWikiText( $text );
3448                 }
3449                 # Bug 529: if the template begins with a table or block-level
3450                 # element, it should be treated as beginning a new line.
3451                 # This behaviour is somewhat controversial.
3452                 elseif ( !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3453                         $text = "\n" . $text;
3454                 }
3455
3456                 # Prune lower levels off the recursion check path
3457                 $this->mTemplatePath = $lastPathLevel;
3458
3459                 if ( !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3460                         # Error, oversize inclusion
3461                         $text = "[[$originalTitle]]" .
3462                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3463                 }
3464
3465                 wfProfileOut( $fname );
3466                 return $text;
3467         }
3468
3469         /**
3470          * Get the semi-parsed DOM representation of a template with a given title,
3471          * and its redirect destination title. Cached.
3472          */
3473         function getTemplateDom( $title ) {
3474                 $cacheTitle = $title;
3475                 $titleText = $title->getPrefixedDBkey();
3476
3477                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3478                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3479                         $title = Title::makeTitle( $ns, $dbk );
3480                         $titleText = $title->getPrefixedDBkey();
3481                 }
3482                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3483                         return array( $this->mTplDomCache[$titleText], $title );
3484                 }
3485
3486                 // Cache miss, go to the database
3487                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3488
3489                 if ( $text === false ) {
3490                         $this->mTplDomCache[$titleText] = false;
3491                         return array( false, $title );
3492                 }
3493
3494                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3495                 $this->mTplDomCache[ $titleText ] = $dom;
3496
3497                 if (! $title->equals($cacheTitle)) {
3498                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3499                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3500                 }
3501
3502                 return array( $dom, $title );
3503         }
3504
3505         /**
3506          * Fetch the unparsed text of a template and register a reference to it.
3507          */
3508         function fetchTemplateAndTitle( $title ) {
3509                 $templateCb = $this->mOptions->getTemplateCallback();
3510                 $stuff = call_user_func( $templateCb, $title );
3511                 $text = $stuff['text'];
3512                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3513                 if ( isset( $stuff['deps'] ) ) {
3514                         foreach ( $stuff['deps'] as $dep ) {
3515                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3516                         }
3517                 }
3518                 return array($text,$finalTitle);
3519         }
3520
3521         function fetchTemplate( $title ) {
3522                 $rv = $this->fetchTemplateAndTitle($title);
3523                 return $rv[0];
3524         }
3525
3526         /**
3527          * Static function to get a template
3528          * Can be overridden via ParserOptions::setTemplateCallback().
3529          */
3530         static function statelessFetchTemplate( $title ) {
3531                 $text = $skip = false;
3532                 $finalTitle = $title;
3533                 $deps = array();
3534
3535                 // Loop to fetch the article, with up to 1 redirect
3536                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3537                         # Give extensions a chance to select the revision instead
3538                         $id = false; // Assume current
3539                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3540
3541                         if( $skip ) {
3542                                 $text = false;
3543                                 $deps[] = array(
3544                                         'title' => $title,
3545                                         'page_id' => $title->getArticleID(),
3546                                         'rev_id' => null );
3547                                 break;
3548                         }
3549                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3550                         $rev_id = $rev ? $rev->getId() : 0;
3551
3552                         $deps[] = array(
3553                                 'title' => $title,
3554                                 'page_id' => $title->getArticleID(),
3555                                 'rev_id' => $rev_id );
3556
3557                         if( $rev ) {
3558                                 $text = $rev->getText();
3559                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3560                                 global $wgLang;
3561                                 $message = $wgLang->lcfirst( $title->getText() );
3562                                 $text = wfMsgForContentNoTrans( $message );
3563                                 if( wfEmptyMsg( $message, $text ) ) {
3564                                         $text = false;
3565                                         break;
3566                                 }
3567                         } else {
3568                                 break;
3569                         }
3570                         if ( $text === false ) {
3571                                 break;
3572                         }
3573                         // Redirect?
3574                         $finalTitle = $title;
3575                         $title = Title::newFromRedirect( $text );
3576                 }
3577                 return array(
3578                         'text' => $text,
3579                         'finalTitle' => $finalTitle,
3580                         'deps' => $deps );
3581         }
3582
3583         /**
3584          * Transclude an interwiki link.
3585          */
3586         function interwikiTransclude( $title, $action ) {
3587                 global $wgEnableScaryTranscluding;
3588
3589                 if (!$wgEnableScaryTranscluding)
3590                         return wfMsg('scarytranscludedisabled');
3591
3592                 $url = $title->getFullUrl( "action=$action" );
3593
3594                 if (strlen($url) > 255)
3595                         return wfMsg('scarytranscludetoolong');
3596                 return $this->fetchScaryTemplateMaybeFromCache($url);
3597         }
3598
3599         function fetchScaryTemplateMaybeFromCache($url) {
3600                 global $wgTranscludeCacheExpiry;
3601                 $dbr = wfGetDB(DB_SLAVE);
3602                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3603                                 array('tc_url' => $url));
3604                 if ($obj) {
3605                         $time = $obj->tc_time;
3606                         $text = $obj->tc_contents;
3607                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3608                                 return $text;
3609                         }
3610                 }
3611
3612                 $text = Http::get($url);
3613                 if (!$text)
3614                         return wfMsg('scarytranscludefailed', $url);
3615
3616                 $dbw = wfGetDB(DB_MASTER);
3617                 $dbw->replace('transcache', array('tc_url'), array(
3618                         'tc_url' => $url,
3619                         'tc_time' => time(),
3620                         'tc_contents' => $text));
3621                 return $text;
3622         }
3623
3624
3625         /**
3626          * Triple brace replacement -- used for template arguments
3627          * @private
3628          */
3629         function argSubstitution( $piece, $frame ) {
3630                 wfProfileIn( __METHOD__ );
3631
3632                 $text = false;
3633                 $error = false;
3634                 $parts = $piece['parts'];
3635                 $argWithSpaces = $frame->expand( $piece['title'] );
3636                 $arg = trim( $argWithSpaces );
3637
3638                 if ( isset( $frame->args[$arg] ) ) {
3639                         $text = $frame->parent->expand( $frame->args[$arg] );
3640                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3641                         $text = $frame->expand( $parts->item( 0 ) );
3642                 }
3643                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3644                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3645                 }
3646
3647                 if ( $text === false ) {
3648                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3649                 }
3650                 if ( $error !== false ) {
3651                         $text .= $error;
3652                 }
3653
3654                 wfProfileOut( __METHOD__ );
3655                 return $text;
3656         }
3657
3658         /**
3659          * Return the text to be used for a given extension tag.
3660          * This is the ghost of strip().
3661          *
3662          * @param array $params Associative array of parameters:
3663          *     name       DOMNode for the tag name
3664          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3665          *     inner      Contents of extension element
3666          *     noClose    Original text did not have a close tag
3667          * @param PPFrame $frame
3668          */
3669         function extensionSubstitution( $params, $frame ) {
3670                 global $wgRawHtml, $wgContLang;
3671                 static $n = 1;
3672
3673                 $name = $frame->expand( $params['name'] );
3674                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3675                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3676
3677                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3678
3679                 if ( $this->ot['html'] ) {
3680                         $name = strtolower( $name );
3681
3682                         $params = Sanitizer::decodeTagAttributes( $attrText );
3683                         switch ( $name ) {
3684                                 case 'html':
3685                                         if( $wgRawHtml ) {
3686                                                 $output = $content;
3687                                                 break;
3688                                         } else {
3689                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3690                                         }
3691                                 case 'nowiki':
3692                                         $output = Xml::escapeTagsOnly( $content );
3693                                         break;
3694                                 case 'math':
3695                                         $output = $wgContLang->armourMath(
3696                                                 MathRenderer::renderMath( $content, $params ) );
3697                                         break;
3698                                 case 'gallery':
3699                                         $output = $this->renderImageGallery( $content, $params );
3700                                         break;
3701                                 default:
3702                                         if( isset( $this->mTagHooks[$name] ) ) {
3703                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3704                                                         array( $content, $params, $this ) );
3705                                         } else {
3706                                                 throw new MWException( "Invalid call hook $name" );
3707                                         }
3708                         }
3709                 } else {
3710                         if ( $content === null ) {
3711                                 $output = "<$name$attrText/>";
3712                         } else {
3713                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3714                                 $output = "<$name$attrText>$content$close";
3715                         }
3716                 }
3717
3718                 if ( $name == 'html' || $name == 'nowiki' ) {
3719                         $this->mStripState->nowiki->setPair( $marker, $output );
3720                 } else {
3721                         $this->mStripState->general->setPair( $marker, $output );
3722                 }
3723                 return $marker;
3724         }
3725
3726         /**
3727          * Increment an include size counter
3728          *
3729          * @param string $type The type of expansion
3730          * @param integer $size The size of the text
3731          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3732          */
3733         function incrementIncludeSize( $type, $size ) {
3734                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3735                         return false;
3736                 } else {
3737                         $this->mIncludeSizes[$type] += $size;
3738                         return true;
3739                 }
3740         }
3741
3742         /**
3743          * Detect __NOGALLERY__ magic word and set a placeholder
3744          */
3745         function stripNoGallery( &$text ) {
3746                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3747                 # do not add TOC
3748                 $mw = MagicWord::get( 'nogallery' );
3749                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3750         }
3751
3752         /**
3753          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3754          * placeholder that will then be replaced by the real TOC in
3755          * ->formatHeadings, this works because at this points real
3756          * comments will have already been discarded by the sanitizer.
3757          *
3758          * Any additional __TOC__ magic words left over will be discarded
3759          * as there can only be one TOC on the page.
3760          */
3761         function stripToc( $text ) {
3762                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3763                 # do not add TOC
3764                 $mw = MagicWord::get( 'notoc' );
3765                 if( $mw->matchAndRemove( $text ) ) {
3766                         $this->mShowToc = false;
3767                 }
3768
3769                 $mw = MagicWord::get( 'toc' );
3770                 if( $mw->match( $text ) ) {
3771                         $this->mShowToc = true;
3772                         $this->mForceTocPosition = true;
3773
3774                         // Set a placeholder. At the end we'll fill it in with the TOC.
3775                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3776
3777                         // Only keep the first one.
3778                         $text = $mw->replace( '', $text );
3779                 }
3780                 return $text;
3781         }
3782
3783         /**
3784          * This function accomplishes several tasks:
3785          * 1) Auto-number headings if that option is enabled
3786          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3787          * 3) Add a Table of contents on the top for users who have enabled the option
3788          * 4) Auto-anchor headings
3789          *
3790          * It loops through all headlines, collects the necessary data, then splits up the
3791          * string and re-inserts the newly formatted headlines.
3792          *
3793          * @param string $text
3794          * @param boolean $isMain
3795          * @private
3796          */
3797         function formatHeadings( $text, $isMain=true ) {
3798                 global $wgMaxTocLevel, $wgContLang;
3799
3800                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3801                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3802                         $showEditLink = 0;
3803                 } else {
3804                         $showEditLink = $this->mOptions->getEditSection();
3805                 }
3806
3807                 # Inhibit editsection links if requested in the page
3808                 $esw =& MagicWord::get( 'noeditsection' );
3809                 if( $esw->matchAndRemove( $text ) ) {
3810                         $showEditLink = 0;
3811                 }
3812
3813                 # Get all headlines for numbering them and adding funky stuff like [edit]
3814                 # links - this is for later, but we need the number of headlines right now
3815                 $matches = array();
3816                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3817
3818                 # if there are fewer than 4 headlines in the article, do not show TOC
3819                 # unless it's been explicitly enabled.
3820                 $enoughToc = $this->mShowToc &&
3821                         (($numMatches >= 4) || $this->mForceTocPosition);
3822
3823                 # Allow user to stipulate that a page should have a "new section"
3824                 # link added via __NEWSECTIONLINK__
3825                 $mw =& MagicWord::get( 'newsectionlink' );
3826                 if( $mw->matchAndRemove( $text ) )
3827                         $this->mOutput->setNewSection( true );
3828
3829                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3830                 # override above conditions and always show TOC above first header
3831                 $mw =& MagicWord::get( 'forcetoc' );
3832                 if ($mw->matchAndRemove( $text ) ) {
3833                         $this->mShowToc = true;
3834                         $enoughToc = true;
3835                 }
3836
3837                 # We need this to perform operations on the HTML
3838                 $sk = $this->mOptions->getSkin();
3839
3840                 # headline counter
3841                 $headlineCount = 0;
3842                 $numVisible = 0;
3843
3844                 # Ugh .. the TOC should have neat indentation levels which can be
3845                 # passed to the skin functions. These are determined here
3846                 $toc = '';
3847                 $full = '';
3848                 $head = array();
3849                 $sublevelCount = array();
3850                 $levelCount = array();
3851                 $toclevel = 0;
3852                 $level = 0;
3853                 $prevlevel = 0;
3854                 $toclevel = 0;
3855                 $prevtoclevel = 0;
3856                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3857                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3858                 $tocraw = array();
3859
3860                 foreach( $matches[3] as $headline ) {
3861                         $isTemplate = false;
3862                         $titleText = false;
3863                         $sectionIndex = false;
3864                         $numbering = '';
3865                         $markerMatches = array();
3866                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3867                                 $serial = $markerMatches[1];
3868                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3869                                 $isTemplate = ($titleText != $baseTitleText);
3870                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3871                         }
3872
3873                         if( $toclevel ) {
3874                                 $prevlevel = $level;
3875                                 $prevtoclevel = $toclevel;
3876                         }
3877                         $level = $matches[1][$headlineCount];
3878
3879                         if( $doNumberHeadings || $enoughToc ) {
3880
3881                                 if ( $level > $prevlevel ) {
3882                                         # Increase TOC level
3883                                         $toclevel++;
3884                                         $sublevelCount[$toclevel] = 0;
3885                                         if( $toclevel<$wgMaxTocLevel ) {
3886                                                 $prevtoclevel = $toclevel;
3887                                                 $toc .= $sk->tocIndent();
3888                                                 $numVisible++;
3889                                         }
3890                                 }
3891                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3892                                         # Decrease TOC level, find level to jump to
3893
3894                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3895                                                 # Can only go down to level 1
3896                                                 $toclevel = 1;
3897                                         } else {
3898                                                 for ($i = $toclevel; $i > 0; $i--) {
3899                                                         if ( $levelCount[$i] == $level ) {
3900                                                                 # Found last matching level
3901                                                                 $toclevel = $i;
3902                                                                 break;
3903                                                         }
3904                                                         elseif ( $levelCount[$i] < $level ) {
3905                                                                 # Found first matching level below current level
3906                                                                 $toclevel = $i + 1;
3907                                                                 break;
3908                                                         }
3909                                                 }
3910                                         }
3911                                         if( $toclevel<$wgMaxTocLevel ) {
3912                                                 if($prevtoclevel < $wgMaxTocLevel) {
3913                                                         # Unindent only if the previous toc level was shown :p
3914                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3915                                                 } else {
3916                                                         $toc .= $sk->tocLineEnd();
3917                                                 }
3918                                         }
3919                                 }
3920                                 else {
3921                                         # No change in level, end TOC line
3922                                         if( $toclevel<$wgMaxTocLevel ) {
3923                                                 $toc .= $sk->tocLineEnd();
3924                                         }
3925                                 }
3926
3927                                 $levelCount[$toclevel] = $level;
3928
3929                                 # count number of headlines for each level
3930                                 @$sublevelCount[$toclevel]++;
3931                                 $dot = 0;
3932                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3933                                         if( !empty( $sublevelCount[$i] ) ) {
3934                                                 if( $dot ) {
3935                                                         $numbering .= '.';
3936                                                 }
3937                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3938                                                 $dot = 1;
3939                                         }
3940                                 }
3941                         }
3942
3943                         # The safe header is a version of the header text safe to use for links
3944                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3945                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3946
3947                         # Remove link placeholders by the link text.
3948                         #     <!--LINK number-->
3949                         # turns into
3950                         #     link text with suffix
3951                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3952                                                             "\$this->mLinkHolders['texts'][\$1]",
3953                                                             $safeHeadline );
3954                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3955                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3956                                                             $safeHeadline );
3957
3958                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3959                         $tocline = preg_replace(
3960                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3961                                 array( '',                          '<$1>'),
3962                                 $safeHeadline
3963                         );
3964                         $tocline = trim( $tocline );
3965
3966                         # For the anchor, strip out HTML-y stuff period
3967                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3968                         $safeHeadline = trim( $safeHeadline );
3969
3970                         # Save headline for section edit hint before it's escaped
3971                         $headlineHint = $safeHeadline;
3972                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3973                         $refers[$headlineCount] = $safeHeadline;
3974
3975                         # count how many in assoc. array so we can track dupes in anchors
3976                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3977                         $refcount[$headlineCount] = $refers[$safeHeadline];
3978
3979                         # Don't number the heading if it is the only one (looks silly)
3980                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3981                                 # the two are different if the line contains a link
3982                                 $headline=$numbering . ' ' . $headline;
3983                         }
3984
3985                         # Create the anchor for linking from the TOC to the section
3986                         $anchor = $safeHeadline;
3987                         if($refcount[$headlineCount] > 1 ) {
3988                                 $anchor .= '_' . $refcount[$headlineCount];
3989                         }
3990                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3991                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3992                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3993                         }
3994                         # give headline the correct <h#> tag
3995                         if( $showEditLink && $sectionIndex !== false ) {
3996                                 if( $isTemplate ) {
3997                                         # Put a T flag in the section identifier, to indicate to extractSections()
3998                                         # that sections inside <includeonly> should be counted.
3999                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
4000                                 } else {
4001                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
4002                                 }
4003                         } else {
4004                                 $editlink = '';
4005                         }
4006                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
4007
4008                         $headlineCount++;
4009                 }
4010
4011                 $this->mOutput->setSections( $tocraw );
4012
4013                 # Never ever show TOC if no headers
4014                 if( $numVisible < 1 ) {
4015                         $enoughToc = false;
4016                 }
4017
4018                 if( $enoughToc ) {
4019                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4020                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
4021                         }
4022                         $toc = $sk->tocList( $toc );
4023                 }
4024
4025                 # split up and insert constructed headlines
4026
4027                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4028                 $i = 0;
4029
4030                 foreach( $blocks as $block ) {
4031                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
4032                                 # This is the [edit] link that appears for the top block of text when
4033                                 # section editing is enabled
4034
4035                                 # Disabled because it broke block formatting
4036                                 # For example, a bullet point in the top line
4037                                 # $full .= $sk->editSectionLink(0);
4038                         }
4039                         $full .= $block;
4040                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4041                                 # Top anchor now in skin
4042                                 $full = $full.$toc;
4043                         }
4044
4045                         if( !empty( $head[$i] ) ) {
4046                                 $full .= $head[$i];
4047                         }
4048                         $i++;
4049                 }
4050                 if( $this->mForceTocPosition ) {
4051                         return str_replace( '<!--MWTOC-->', $toc, $full );
4052                 } else {
4053                         return $full;
4054                 }
4055         }
4056
4057         /**
4058          * Transform wiki markup when saving a page by doing \r\n -> \n
4059          * conversion, substitting signatures, {{subst:}} templates, etc.
4060          *
4061          * @param string $text the text to transform
4062          * @param Title &$title the Title object for the current article
4063          * @param User &$user the User object describing the current user
4064          * @param ParserOptions $options parsing options
4065          * @param bool $clearState whether to clear the parser state first
4066          * @return string the altered wiki markup
4067          * @public
4068          */
4069         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4070                 $this->mOptions = $options;
4071                 $this->mTitle =& $title;
4072                 $this->setOutputType( OT_WIKI );
4073
4074                 if ( $clearState ) {
4075                         $this->clearState();
4076                 }
4077
4078                 $pairs = array(
4079                         "\r\n" => "\n",
4080                 );
4081                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4082                 $text = $this->pstPass2( $text, $user );
4083                 $text = $this->mStripState->unstripBoth( $text );
4084                 return $text;
4085         }
4086
4087         /**
4088          * Pre-save transform helper function
4089          * @private
4090          */
4091         function pstPass2( $text, $user ) {
4092                 global $wgContLang, $wgLocaltimezone;
4093
4094                 /* Note: This is the timestamp saved as hardcoded wikitext to
4095                  * the database, we use $wgContLang here in order to give
4096                  * everyone the same signature and use the default one rather
4097                  * than the one selected in each user's preferences.
4098                  */
4099                 if ( isset( $wgLocaltimezone ) ) {
4100                         $oldtz = getenv( 'TZ' );
4101                         putenv( 'TZ='.$wgLocaltimezone );
4102                 }
4103                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4104                   ' (' . date( 'T' ) . ')';
4105                 if ( isset( $wgLocaltimezone ) ) {
4106                         putenv( 'TZ='.$oldtz );
4107                 }
4108
4109                 # Variable replacement
4110                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4111                 $text = $this->replaceVariables( $text );
4112
4113                 # Strip out <nowiki> etc. added via replaceVariables
4114                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4115
4116                 # Signatures
4117                 $sigText = $this->getUserSig( $user );
4118                 $text = strtr( $text, array(
4119                         '~~~~~' => $d,
4120                         '~~~~' => "$sigText $d",
4121                         '~~~' => $sigText
4122                 ) );
4123
4124                 # Context links: [[|name]] and [[name (context)|]]
4125                 #
4126                 global $wgLegalTitleChars;
4127                 $tc = "[$wgLegalTitleChars]";
4128                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4129
4130                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4131                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4132                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4133
4134                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4135                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4136                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4137
4138                 $t = $this->mTitle->getText();
4139                 $m = array();
4140                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4141                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4142                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4143                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4144                 } else {
4145                         # if there's no context, don't bother duplicating the title
4146                         $text = preg_replace( $p2, '[[\\1]]', $text );
4147                 }
4148
4149                 # Trim trailing whitespace
4150                 $text = rtrim( $text );
4151
4152                 return $text;
4153         }
4154
4155         /**
4156          * Fetch the user's signature text, if any, and normalize to
4157          * validated, ready-to-insert wikitext.
4158          *
4159          * @param User $user
4160          * @return string
4161          * @private
4162          */
4163         function getUserSig( &$user ) {
4164                 global $wgMaxSigChars;
4165
4166                 $username = $user->getName();
4167                 $nickname = $user->getOption( 'nickname' );
4168                 $nickname = $nickname === '' ? $username : $nickname;
4169
4170                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4171                         $nickname = $username;
4172                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4173                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4174                         # Sig. might contain markup; validate this
4175                         if( $this->validateSig( $nickname ) !== false ) {
4176                                 # Validated; clean up (if needed) and return it
4177                                 return $this->cleanSig( $nickname, true );
4178                         } else {
4179                                 # Failed to validate; fall back to the default
4180                                 $nickname = $username;
4181                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4182                         }
4183                 }
4184
4185                 // Make sure nickname doesnt get a sig in a sig
4186                 $nickname = $this->cleanSigInSig( $nickname );
4187
4188                 # If we're still here, make it a link to the user page
4189                 $userText = wfEscapeWikiText( $username );
4190                 $nickText = wfEscapeWikiText( $nickname );
4191                 if ( $user->isAnon() )  {
4192                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4193                 } else {
4194                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4195                 }
4196         }
4197
4198         /**
4199          * Check that the user's signature contains no bad XML
4200          *
4201          * @param string $text
4202          * @return mixed An expanded string, or false if invalid.
4203          */
4204         function validateSig( $text ) {
4205                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4206         }
4207
4208         /**
4209          * Clean up signature text
4210          *
4211          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4212          * 2) Substitute all transclusions
4213          *
4214          * @param string $text
4215          * @param $parsing Whether we're cleaning (preferences save) or parsing
4216          * @return string Signature text
4217          */
4218         function cleanSig( $text, $parsing = false ) {
4219                 if ( !$parsing ) {
4220                         global $wgTitle;
4221                         $this->startExternalParse( $wgTitle, new ParserOptions(), OT_MSG );
4222                 }
4223
4224                 # FIXME: regex doesn't respect extension tags or nowiki
4225                 #  => Move this logic to braceSubstitution()
4226                 $substWord = MagicWord::get( 'subst' );
4227                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4228                 $substText = '{{' . $substWord->getSynonym( 0 );
4229
4230                 $text = preg_replace( $substRegex, $substText, $text );
4231                 $text = $this->cleanSigInSig( $text );
4232                 $dom = $this->preprocessToDom( $text );
4233                 $frame = new PPFrame( $this );
4234                 $text = $frame->expand( $dom->documentElement );
4235
4236                 if ( !$parsing ) {
4237                         $text = $this->mStripState->unstripBoth( $text );
4238                 }
4239
4240                 return $text;
4241         }
4242
4243         /**
4244          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4245          * @param string $text
4246          * @return string Signature text with /~{3,5}/ removed
4247          */
4248         function cleanSigInSig( $text ) {
4249                 $text = preg_replace( '/~{3,5}/', '', $text );
4250                 return $text;
4251         }
4252
4253         /**
4254          * Set up some variables which are usually set up in parse()
4255          * so that an external function can call some class members with confidence
4256          * @public
4257          */
4258         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4259                 $this->mTitle =& $title;
4260                 $this->mOptions = $options;
4261                 $this->setOutputType( $outputType );
4262                 if ( $clearState ) {
4263                         $this->clearState();
4264                 }
4265         }
4266
4267         /**
4268          * Transform a MediaWiki message by replacing magic variables.
4269          *
4270          * For some unknown reason, it also expands templates, but only to the
4271          * first recursion level. This is wrong and broken, probably introduced
4272          * accidentally during refactoring, but probably relied upon by thousands
4273          * of users.
4274          *
4275          * @param string $text the text to transform
4276          * @param ParserOptions $options  options
4277          * @return string the text with variables substituted
4278          * @public
4279          */
4280         function transformMsg( $text, $options ) {
4281                 global $wgTitle;
4282                 static $executing = false;
4283
4284                 $fname = "Parser::transformMsg";
4285
4286                 # Guard against infinite recursion
4287                 if ( $executing ) {
4288                         return $text;
4289                 }
4290                 $executing = true;
4291
4292                 wfProfileIn($fname);
4293
4294                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4295                         $this->mTitle = $wgTitle;
4296                 } else {
4297                         $this->mTitle = Title::newFromText('msg');
4298                 }
4299                 $this->mOptions = $options;
4300                 $this->setOutputType( OT_MSG );
4301                 $this->clearState();
4302                 $text = $this->replaceVariables( $text );
4303                 $text = $this->mStripState->unstripBoth( $text );
4304
4305                 $executing = false;
4306                 wfProfileOut($fname);
4307                 return $text;
4308         }
4309
4310         /**
4311          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4312          * The callback should have the following form:
4313          *    function myParserHook( $text, $params, &$parser ) { ... }
4314          *
4315          * Transform and return $text. Use $parser for any required context, e.g. use
4316          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4317          *
4318          * @public
4319          *
4320          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4321          * @param mixed $callback The callback function (and object) to use for the tag
4322          *
4323          * @return The old value of the mTagHooks array associated with the hook
4324          */
4325         function setHook( $tag, $callback ) {
4326                 $tag = strtolower( $tag );
4327                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4328                 $this->mTagHooks[$tag] = $callback;
4329                 $this->mStripList[] = $tag;
4330
4331                 return $oldVal;
4332         }
4333
4334         function setTransparentTagHook( $tag, $callback ) {
4335                 $tag = strtolower( $tag );
4336                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4337                 $this->mTransparentTagHooks[$tag] = $callback;
4338
4339                 return $oldVal;
4340         }
4341
4342         /**
4343          * Create a function, e.g. {{sum:1|2|3}}
4344          * The callback function should have the form:
4345          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4346          *
4347          * The callback may either return the text result of the function, or an array with the text
4348          * in element 0, and a number of flags in the other elements. The names of the flags are
4349          * specified in the keys. Valid flags are:
4350          *   found                     The text returned is valid, stop processing the template. This
4351          *                             is on by default.
4352          *   nowiki                    Wiki markup in the return value should be escaped
4353          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4354          *   noargs                    Don't replace triple-brace arguments in the return value
4355          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4356          *
4357          * @public
4358          *
4359          * @param string $id The magic word ID
4360          * @param mixed $callback The callback function (and object) to use
4361          * @param integer $flags a combination of the following flags:
4362          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4363          *
4364          * @return The old callback function for this name, if any
4365          */
4366         function setFunctionHook( $id, $callback, $flags = 0 ) {
4367                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4368                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4369
4370                 # Add to function cache
4371                 $mw = MagicWord::get( $id );
4372                 if( !$mw )
4373                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4374
4375                 $synonyms = $mw->getSynonyms();
4376                 $sensitive = intval( $mw->isCaseSensitive() );
4377
4378                 foreach ( $synonyms as $syn ) {
4379                         # Case
4380                         if ( !$sensitive ) {
4381                                 $syn = strtolower( $syn );
4382                         }
4383                         # Add leading hash
4384                         if ( !( $flags & SFH_NO_HASH ) ) {
4385                                 $syn = '#' . $syn;
4386                         }
4387                         # Remove trailing colon
4388                         if ( substr( $syn, -1, 1 ) == ':' ) {
4389                                 $syn = substr( $syn, 0, -1 );
4390                         }
4391                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4392                 }
4393                 return $oldVal;
4394         }
4395
4396         /**
4397          * Get all registered function hook identifiers
4398          *
4399          * @return array
4400          */
4401         function getFunctionHooks() {
4402                 return array_keys( $this->mFunctionHooks );
4403         }
4404
4405         /**
4406          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4407          * Placeholders created in Skin::makeLinkObj()
4408          * Returns an array of link CSS classes, indexed by PDBK.
4409          * $options is a bit field, RLH_FOR_UPDATE to select for update
4410          */
4411         function replaceLinkHolders( &$text, $options = 0 ) {
4412                 global $wgUser;
4413                 global $wgContLang;
4414
4415                 $fname = 'Parser::replaceLinkHolders';
4416                 wfProfileIn( $fname );
4417
4418                 $pdbks = array();
4419                 $colours = array();
4420                 $linkcolour_ids = array();
4421                 $sk = $this->mOptions->getSkin();
4422                 $linkCache =& LinkCache::singleton();
4423
4424                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4425                         wfProfileIn( $fname.'-check' );
4426                         $dbr = wfGetDB( DB_SLAVE );
4427                         $page = $dbr->tableName( 'page' );
4428                         $threshold = $wgUser->getOption('stubthreshold');
4429
4430                         # Sort by namespace
4431                         asort( $this->mLinkHolders['namespaces'] );
4432
4433                         # Generate query
4434                         $query = false;
4435                         $current = null;
4436                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4437                                 # Make title object
4438                                 $title = $this->mLinkHolders['titles'][$key];
4439
4440                                 # Skip invalid entries.
4441                                 # Result will be ugly, but prevents crash.
4442                                 if ( is_null( $title ) ) {
4443                                         continue;
4444                                 }
4445                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4446
4447                                 # Check if it's a static known link, e.g. interwiki
4448                                 if ( $title->isAlwaysKnown() ) {
4449                                         $colours[$pdbk] = '';
4450                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4451                                         $colours[$pdbk] = '';
4452                                         $this->mOutput->addLink( $title, $id );
4453                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4454                                         $colours[$pdbk] = 'new';
4455                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4456                                         $colours[$pdbk] = 'new';
4457                                 } else {
4458                                         # Not in the link cache, add it to the query
4459                                         if ( !isset( $current ) ) {
4460                                                 $current = $ns;
4461                                                 $query =  "SELECT page_id, page_namespace, page_title";
4462                                                 if ( $threshold > 0 ) {
4463                                                         $query .= ', page_len, page_is_redirect';
4464                                                 }
4465                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4466                                         } elseif ( $current != $ns ) {
4467                                                 $current = $ns;
4468                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4469                                         } else {
4470                                                 $query .= ', ';
4471                                         }
4472
4473                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4474                                 }
4475                         }
4476                         if ( $query ) {
4477                                 $query .= '))';
4478                                 if ( $options & RLH_FOR_UPDATE ) {
4479                                         $query .= ' FOR UPDATE';
4480                                 }
4481
4482                                 $res = $dbr->query( $query, $fname );
4483
4484                                 # Fetch data and form into an associative array
4485                                 # non-existent = broken
4486                                 while ( $s = $dbr->fetchObject($res) ) {
4487                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4488                                         $pdbk = $title->getPrefixedDBkey();
4489                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4490                                         $this->mOutput->addLink( $title, $s->page_id );
4491                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4492                                         //add id to the extension todolist
4493                                         $linkcolour_ids[$s->page_id] = $pdbk;
4494                                 }
4495                                 //pass an array of page_ids to an extension
4496                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4497                         }
4498                         wfProfileOut( $fname.'-check' );
4499
4500                         # Do a second query for different language variants of links and categories
4501                         if($wgContLang->hasVariants()){
4502                                 $linkBatch = new LinkBatch();
4503                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4504                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4505                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4506
4507                                 $categories = $this->mOutput->getCategoryLinks();
4508
4509                                 // Add variants of links to link batch
4510                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4511                                         $title = $this->mLinkHolders['titles'][$key];
4512                                         if ( is_null( $title ) )
4513                                                 continue;
4514
4515                                         $pdbk = $title->getPrefixedDBkey();
4516                                         $titleText = $title->getText();
4517
4518                                         // generate all variants of the link title text
4519                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4520
4521                                         // if link was not found (in first query), add all variants to query
4522                                         if ( !isset($colours[$pdbk]) ){
4523                                                 foreach($allTextVariants as $textVariant){
4524                                                         if($textVariant != $titleText){
4525                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4526                                                                 if(is_null($variantTitle)) continue;
4527                                                                 $linkBatch->addObj( $variantTitle );
4528                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4529                                                         }
4530                                                 }
4531                                         }
4532                                 }
4533
4534                                 // process categories, check if a category exists in some variant
4535                                 foreach( $categories as $category ){
4536                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4537                                         foreach($variants as $variant){
4538                                                 if($variant != $category){
4539                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4540                                                         if(is_null($variantTitle)) continue;
4541                                                         $linkBatch->addObj( $variantTitle );
4542                                                         $categoryMap[$variant] = $category;
4543                                                 }
4544                                         }
4545                                 }
4546
4547
4548                                 if(!$linkBatch->isEmpty()){
4549                                         // construct query
4550                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4551
4552                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4553                                         if ( $threshold > 0 ) {
4554                                                 $variantQuery .= ', page_len, page_is_redirect';
4555                                         }
4556
4557                                         $variantQuery .= " FROM $page WHERE $titleClause";
4558                                         if ( $options & RLH_FOR_UPDATE ) {
4559                                                 $variantQuery .= ' FOR UPDATE';
4560                                         }
4561
4562                                         $varRes = $dbr->query( $variantQuery, $fname );
4563
4564                                         // for each found variants, figure out link holders and replace
4565                                         while ( $s = $dbr->fetchObject($varRes) ) {
4566
4567                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4568                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4569                                                 $vardbk = $variantTitle->getDBkey();
4570
4571                                                 $holderKeys = array();
4572                                                 if(isset($variantMap[$varPdbk])){
4573                                                         $holderKeys = $variantMap[$varPdbk];
4574                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4575                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4576                                                 }
4577
4578                                                 // loop over link holders
4579                                                 foreach($holderKeys as $key){
4580                                                         $title = $this->mLinkHolders['titles'][$key];
4581                                                         if ( is_null( $title ) ) continue;
4582
4583                                                         $pdbk = $title->getPrefixedDBkey();
4584
4585                                                         if(!isset($colours[$pdbk])){
4586                                                                 // found link in some of the variants, replace the link holder data
4587                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4588                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4589
4590                                                                 // set pdbk and colour
4591                                                                 $pdbks[$key] = $varPdbk;
4592                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4593                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4594                                                         }
4595                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4596                                                 }
4597
4598                                                 // check if the object is a variant of a category
4599                                                 if(isset($categoryMap[$vardbk])){
4600                                                         $oldkey = $categoryMap[$vardbk];
4601                                                         if($oldkey != $vardbk)
4602                                                                 $varCategories[$oldkey]=$vardbk;
4603                                                 }
4604                                         }
4605
4606                                         // rebuild the categories in original order (if there are replacements)
4607                                         if(count($varCategories)>0){
4608                                                 $newCats = array();
4609                                                 $originalCats = $this->mOutput->getCategories();
4610                                                 foreach($originalCats as $cat => $sortkey){
4611                                                         // make the replacement
4612                                                         if( array_key_exists($cat,$varCategories) )
4613                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4614                                                         else $newCats[$cat] = $sortkey;
4615                                                 }
4616                                                 $this->mOutput->setCategoryLinks($newCats);
4617                                         }
4618                                 }
4619                         }
4620
4621                         # Construct search and replace arrays
4622                         wfProfileIn( $fname.'-construct' );
4623                         $replacePairs = array();
4624                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4625                                 $pdbk = $pdbks[$key];
4626                                 $searchkey = "<!--LINK $key-->";
4627                                 $title = $this->mLinkHolders['titles'][$key];
4628                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4629                                         $linkCache->addBadLinkObj( $title );
4630                                         $colours[$pdbk] = 'new';
4631                                         $this->mOutput->addLink( $title, 0 );
4632                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4633                                                                         $this->mLinkHolders['texts'][$key],
4634                                                                         $this->mLinkHolders['queries'][$key] );
4635                                 } else {
4636                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4637                                                                         $this->mLinkHolders['texts'][$key],
4638                                                                         $this->mLinkHolders['queries'][$key] );
4639                                 }
4640                         }
4641                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4642                         wfProfileOut( $fname.'-construct' );
4643
4644                         # Do the thing
4645                         wfProfileIn( $fname.'-replace' );
4646                         $text = preg_replace_callback(
4647                                 '/(<!--LINK .*?-->)/',
4648                                 $replacer->cb(),
4649                                 $text);
4650
4651                         wfProfileOut( $fname.'-replace' );
4652                 }
4653
4654                 # Now process interwiki link holders
4655                 # This is quite a bit simpler than internal links
4656                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4657                         wfProfileIn( $fname.'-interwiki' );
4658                         # Make interwiki link HTML
4659                         $replacePairs = array();
4660                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4661                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4662                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4663                         }
4664                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4665
4666                         $text = preg_replace_callback(
4667                                 '/<!--IWLINK (.*?)-->/',
4668                                 $replacer->cb(),
4669                                 $text );
4670                         wfProfileOut( $fname.'-interwiki' );
4671                 }
4672
4673                 wfProfileOut( $fname );
4674                 return $colours;
4675         }
4676
4677         /**
4678          * Replace <!--LINK--> link placeholders with plain text of links
4679          * (not HTML-formatted).
4680          * @param string $text
4681          * @return string
4682          */
4683         function replaceLinkHoldersText( $text ) {
4684                 $fname = 'Parser::replaceLinkHoldersText';
4685                 wfProfileIn( $fname );
4686
4687                 $text = preg_replace_callback(
4688                         '/<!--(LINK|IWLINK) (.*?)-->/',
4689                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4690                         $text );
4691
4692                 wfProfileOut( $fname );
4693                 return $text;
4694         }
4695
4696         /**
4697          * @param array $matches
4698          * @return string
4699          * @private
4700          */
4701         function replaceLinkHoldersTextCallback( $matches ) {
4702                 $type = $matches[1];
4703                 $key  = $matches[2];
4704                 if( $type == 'LINK' ) {
4705                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4706                                 return $this->mLinkHolders['texts'][$key];
4707                         }
4708                 } elseif( $type == 'IWLINK' ) {
4709                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4710                                 return $this->mInterwikiLinkHolders['texts'][$key];
4711                         }
4712                 }
4713                 return $matches[0];
4714         }
4715
4716         /**
4717          * Tag hook handler for 'pre'.
4718          */
4719         function renderPreTag( $text, $attribs ) {
4720                 // Backwards-compatibility hack
4721                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4722
4723                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4724                 return wfOpenElement( 'pre', $attribs ) .
4725                         Xml::escapeTagsOnly( $content ) .
4726                         '</pre>';
4727         }
4728
4729         /**
4730          * Renders an image gallery from a text with one line per image.
4731          * text labels may be given by using |-style alternative text. E.g.
4732          *   Image:one.jpg|The number "1"
4733          *   Image:tree.jpg|A tree
4734          * given as text will return the HTML of a gallery with two images,
4735          * labeled 'The number "1"' and
4736          * 'A tree'.
4737          */
4738         function renderImageGallery( $text, $params ) {
4739                 $ig = new ImageGallery();
4740                 $ig->setContextTitle( $this->mTitle );
4741                 $ig->setShowBytes( false );
4742                 $ig->setShowFilename( false );
4743                 $ig->setParser( $this );
4744                 $ig->setHideBadImages();
4745                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4746                 $ig->useSkin( $this->mOptions->getSkin() );
4747                 $ig->mRevisionId = $this->mRevisionId;
4748
4749                 if( isset( $params['caption'] ) ) {
4750                         $caption = $params['caption'];
4751                         $caption = htmlspecialchars( $caption );
4752                         $caption = $this->replaceInternalLinks( $caption );
4753                         $ig->setCaptionHtml( $caption );
4754                 }
4755                 if( isset( $params['perrow'] ) ) {
4756                         $ig->setPerRow( $params['perrow'] );
4757                 }
4758                 if( isset( $params['widths'] ) ) {
4759                         $ig->setWidths( $params['widths'] );
4760                 }
4761                 if( isset( $params['heights'] ) ) {
4762                         $ig->setHeights( $params['heights'] );
4763                 }
4764
4765                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4766
4767                 $lines = explode( "\n", $text );
4768                 foreach ( $lines as $line ) {
4769                         # match lines like these:
4770                         # Image:someimage.jpg|This is some image
4771                         $matches = array();
4772                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4773                         # Skip empty lines
4774                         if ( count( $matches ) == 0 ) {
4775                                 continue;
4776                         }
4777                         $tp = Title::newFromText( $matches[1] );
4778                         $nt =& $tp;
4779                         if( is_null( $nt ) ) {
4780                                 # Bogus title. Ignore these so we don't bomb out later.
4781                                 continue;
4782                         }
4783                         if ( isset( $matches[3] ) ) {
4784                                 $label = $matches[3];
4785                         } else {
4786                                 $label = '';
4787                         }
4788
4789                         $pout = $this->parse( $label,
4790                                 $this->mTitle,
4791                                 $this->mOptions,
4792                                 false, // Strip whitespace...?
4793                                 false  // Don't clear state!
4794                         );
4795                         $html = $pout->getText();
4796
4797                         $ig->add( $nt, $html );
4798
4799                         # Only add real images (bug #5586)
4800                         if ( $nt->getNamespace() == NS_IMAGE ) {
4801                                 $this->mOutput->addImage( $nt->getDBkey() );
4802                         }
4803                 }
4804                 return $ig->toHTML();
4805         }
4806
4807         function getImageParams( $handler ) {
4808                 if ( $handler ) {
4809                         $handlerClass = get_class( $handler );
4810                 } else {
4811                         $handlerClass = '';
4812                 }
4813                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4814                         // Initialise static lists
4815                         static $internalParamNames = array(
4816                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4817                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4818                                         'bottom', 'text-bottom' ),
4819                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4820                                         'upright', 'border' ),
4821                         );
4822                         static $internalParamMap;
4823                         if ( !$internalParamMap ) {
4824                                 $internalParamMap = array();
4825                                 foreach ( $internalParamNames as $type => $names ) {
4826                                         foreach ( $names as $name ) {
4827                                                 $magicName = str_replace( '-', '_', "img_$name" );
4828                                                 $internalParamMap[$magicName] = array( $type, $name );
4829                                         }
4830                                 }
4831                         }
4832
4833                         // Add handler params
4834                         $paramMap = $internalParamMap;
4835                         if ( $handler ) {
4836                                 $handlerParamMap = $handler->getParamMap();
4837                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4838                                         $paramMap[$magic] = array( 'handler', $paramName );
4839                                 }
4840                         }
4841                         $this->mImageParams[$handlerClass] = $paramMap;
4842                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4843                 }
4844                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4845         }
4846
4847         /**
4848          * Parse image options text and use it to make an image
4849          */
4850         function makeImage( $title, $options ) {
4851                 # @TODO: let the MediaHandler specify its transform parameters
4852                 #
4853                 # Check if the options text is of the form "options|alt text"
4854                 # Options are:
4855                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4856                 #  * left               no resizing, just left align. label is used for alt= only
4857                 #  * right              same, but right aligned
4858                 #  * none               same, but not aligned
4859                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4860                 #  * center             center the image
4861                 #  * framed             Keep original image size, no magnify-button.
4862                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4863                 #  * upright            reduce width for upright images, rounded to full __0 px
4864                 #  * border             draw a 1px border around the image
4865                 # vertical-align values (no % or length right now):
4866                 #  * baseline
4867                 #  * sub
4868                 #  * super
4869                 #  * top
4870                 #  * text-top
4871                 #  * middle
4872                 #  * bottom
4873                 #  * text-bottom
4874
4875                 $parts = array_map( 'trim', explode( '|', $options) );
4876                 $sk = $this->mOptions->getSkin();
4877
4878                 # Give extensions a chance to select the file revision for us
4879                 $skip = $time = false;
4880                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4881
4882                 if ( $skip ) {
4883                         return $sk->makeLinkObj( $title );
4884                 }
4885
4886                 # Get parameter map
4887                 $file = wfFindFile( $title, $time );
4888                 $handler = $file ? $file->getHandler() : false;
4889
4890                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4891
4892                 # Process the input parameters
4893                 $caption = '';
4894                 $params = array( 'frame' => array(), 'handler' => array(),
4895                         'horizAlign' => array(), 'vertAlign' => array() );
4896                 foreach( $parts as $part ) {
4897                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4898                         if ( isset( $paramMap[$magicName] ) ) {
4899                                 list( $type, $paramName ) = $paramMap[$magicName];
4900                                 $params[$type][$paramName] = $value;
4901
4902                                 // Special case; width and height come in one variable together
4903                                 if( $type == 'handler' && $paramName == 'width' ) {
4904                                         $m = array();
4905                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4906                                                 $params[$type]['width'] = intval( $m[1] );
4907                                                 $params[$type]['height'] = intval( $m[2] );
4908                                         } else {
4909                                                 $params[$type]['width'] = intval( $value );
4910                                         }
4911                                 }
4912                         } else {
4913                                 $caption = $part;
4914                         }
4915                 }
4916
4917                 # Process alignment parameters
4918                 if ( $params['horizAlign'] ) {
4919                         $params['frame']['align'] = key( $params['horizAlign'] );
4920                 }
4921                 if ( $params['vertAlign'] ) {
4922                         $params['frame']['valign'] = key( $params['vertAlign'] );
4923                 }
4924
4925                 # Validate the handler parameters
4926                 if ( $handler ) {
4927                         foreach ( $params['handler'] as $name => $value ) {
4928                                 if ( !$handler->validateParam( $name, $value ) ) {
4929                                         unset( $params['handler'][$name] );
4930                                 }
4931                         }
4932                 }
4933
4934                 # Strip bad stuff out of the alt text
4935                 $alt = $this->replaceLinkHoldersText( $caption );
4936
4937                 # make sure there are no placeholders in thumbnail attributes
4938                 # that are later expanded to html- so expand them now and
4939                 # remove the tags
4940                 $alt = $this->mStripState->unstripBoth( $alt );
4941                 $alt = Sanitizer::stripAllTags( $alt );
4942
4943                 $params['frame']['alt'] = $alt;
4944                 $params['frame']['caption'] = $caption;
4945
4946                 # Linker does the rest
4947                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4948
4949                 # Give the handler a chance to modify the parser object
4950                 if ( $handler ) {
4951                         $handler->parserTransformHook( $this, $file );
4952                 }
4953
4954                 return $ret;
4955         }
4956
4957         /**
4958          * Set a flag in the output object indicating that the content is dynamic and
4959          * shouldn't be cached.
4960          */
4961         function disableCache() {
4962                 wfDebug( "Parser output marked as uncacheable.\n" );
4963                 $this->mOutput->mCacheTime = -1;
4964         }
4965
4966         /**#@+
4967          * Callback from the Sanitizer for expanding items found in HTML attribute
4968          * values, so they can be safely tested and escaped.
4969          * @param string $text
4970          * @param PPFrame $frame
4971          * @return string
4972          * @private
4973          */
4974         function attributeStripCallback( &$text, $frame = false ) {
4975                 $text = $this->replaceVariables( $text, $frame );
4976                 $text = $this->mStripState->unstripBoth( $text );
4977                 return $text;
4978         }
4979
4980         /**#@-*/
4981
4982         /**#@+
4983          * Accessor/mutator
4984          */
4985         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4986         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4987         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4988         /**#@-*/
4989
4990         /**#@+
4991          * Accessor
4992          */
4993         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4994         /**#@-*/
4995
4996
4997         /**
4998          * Break wikitext input into sections, and either pull or replace
4999          * some particular section's text.
5000          *
5001          * External callers should use the getSection and replaceSection methods.
5002          *
5003          * @param string $text Page wikitext
5004          * @param string $section A section identifier string of the form:
5005          *   <flag1> - <flag2> - ... - <section number>
5006          *
5007          * Currently the only recognised flag is "T", which means the target section number
5008          * was derived during a template inclusion parse, in other words this is a template
5009          * section edit link. If no flags are given, it was an ordinary section edit link.
5010          * This flag is required to avoid a section numbering mismatch when a section is
5011          * enclosed by <includeonly> (bug 6563).
5012          *
5013          * The section number 0 pulls the text before the first heading; other numbers will
5014          * pull the given section along with its lower-level subsections. If the section is
5015          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5016          *
5017          * @param string $mode One of "get" or "replace"
5018          * @param string $newText Replacement text for section data.
5019          * @return string for "get", the extracted section text.
5020          *                for "replace", the whole page with the section replaced.
5021          */
5022         private function extractSections( $text, $section, $mode, $newText='' ) {
5023                 global $wgTitle;
5024                 $this->clearState();
5025                 $this->mTitle = $wgTitle; // not generally used but removes an ugly failure mode
5026                 $this->mOptions = new ParserOptions;
5027                 $this->setOutputType( OT_WIKI );
5028                 $curIndex = 0;
5029                 $outText = '';
5030                 $frame = new PPFrame( $this );
5031
5032                 // Process section extraction flags
5033                 $flags = 0;
5034                 $sectionParts = explode( '-', $section );
5035                 $sectionIndex = array_pop( $sectionParts );
5036                 foreach ( $sectionParts as $part ) {
5037                         if ( $part == 'T' ) {
5038                                 $flags |= self::PTD_FOR_INCLUSION;
5039                         }
5040                 }
5041                 // Preprocess the text
5042                 $dom = $this->preprocessToDom( $text, $flags );
5043                 $root = $dom->documentElement;
5044
5045                 // <h> nodes indicate section breaks
5046                 // They can only occur at the top level, so we can find them by iterating the root's children
5047                 $node = $root->firstChild;
5048
5049                 // Find the target section
5050                 if ( $sectionIndex == 0 ) {
5051                         // Section zero doesn't nest, level=big
5052                         $targetLevel = 1000;
5053                 } else {
5054                         while ( $node ) {
5055                                 if ( $node->nodeName == 'h' ) {
5056                                         if ( $curIndex + 1 == $sectionIndex ) {
5057                                                 break;
5058                                         }
5059                                         $curIndex++;
5060                                 }
5061                                 if ( $mode == 'replace' ) {
5062                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5063                                 }
5064                                 $node = $node->nextSibling;
5065                         }
5066                         if ( $node ) {
5067                                 $targetLevel = $node->getAttribute( 'level' );
5068                         }
5069                 }
5070
5071                 if ( !$node ) {
5072                         // Not found
5073                         if ( $mode == 'get' ) {
5074                                 return $newText;
5075                         } else {
5076                                 return $text;
5077                         }
5078                 }
5079
5080                 // Find the end of the section, including nested sections
5081                 do {
5082                         if ( $node->nodeName == 'h' ) {
5083                                 $curIndex++;
5084                                 $curLevel = $node->getAttribute( 'level' );
5085                                 if ( $curIndex != $sectionIndex && $curLevel <= $targetLevel ) {
5086                                         break;
5087                                 }
5088                         }
5089                         if ( $mode == 'get' ) {
5090                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5091                         }
5092                         $node = $node->nextSibling;
5093                 } while ( $node );
5094
5095                 // Write out the remainder (in replace mode only)
5096                 if ( $mode == 'replace' ) {
5097                         // Output the replacement text
5098                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5099                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5100                         $outText .= $newText . "\n\n";
5101                         while ( $node ) {
5102                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5103                                 $node = $node->nextSibling;
5104                         }
5105                 }
5106
5107                 if ( is_string( $outText ) ) {
5108                         // Re-insert stripped tags
5109                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5110                 }
5111
5112                 return $outText;
5113         }
5114
5115         /**
5116          * This function returns the text of a section, specified by a number ($section).
5117          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5118          * the first section before any such heading (section 0).
5119          *
5120          * If a section contains subsections, these are also returned.
5121          *
5122          * @param string $text text to look in
5123          * @param string $section section identifier
5124          * @param string $deftext default to return if section is not found
5125          * @return string text of the requested section
5126          */
5127         public function getSection( $text, $section, $deftext='' ) {
5128                 return $this->extractSections( $text, $section, "get", $deftext );
5129         }
5130
5131         public function replaceSection( $oldtext, $section, $text ) {
5132                 return $this->extractSections( $oldtext, $section, "replace", $text );
5133         }
5134
5135         /**
5136          * Get the timestamp associated with the current revision, adjusted for
5137          * the default server-local timestamp
5138          */
5139         function getRevisionTimestamp() {
5140                 if ( is_null( $this->mRevisionTimestamp ) ) {
5141                         wfProfileIn( __METHOD__ );
5142                         global $wgContLang;
5143                         $dbr = wfGetDB( DB_SLAVE );
5144                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5145                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5146
5147                         // Normalize timestamp to internal MW format for timezone processing.
5148                         // This has the added side-effect of replacing a null value with
5149                         // the current time, which gives us more sensible behavior for
5150                         // previews.
5151                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5152
5153                         // The cryptic '' timezone parameter tells to use the site-default
5154                         // timezone offset instead of the user settings.
5155                         //
5156                         // Since this value will be saved into the parser cache, served
5157                         // to other users, and potentially even used inside links and such,
5158                         // it needs to be consistent for all visitors.
5159                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5160
5161                         wfProfileOut( __METHOD__ );
5162                 }
5163                 return $this->mRevisionTimestamp;
5164         }
5165
5166         /**
5167          * Mutator for $mDefaultSort
5168          *
5169          * @param $sort New value
5170          */
5171         public function setDefaultSort( $sort ) {
5172                 $this->mDefaultSort = $sort;
5173         }
5174
5175         /**
5176          * Accessor for $mDefaultSort
5177          * Will use the title/prefixed title if none is set
5178          *
5179          * @return string
5180          */
5181         public function getDefaultSort() {
5182                 if( $this->mDefaultSort !== false ) {
5183                         return $this->mDefaultSort;
5184                 } else {
5185                         return $this->mTitle->getNamespace() == NS_CATEGORY
5186                                         ? $this->mTitle->getText()
5187                                         : $this->mTitle->getPrefixedText();
5188                 }
5189         }
5190
5191         /**
5192          * Try to guess the section anchor name based on a wikitext fragment
5193          * presumably extracted from a heading, for example "Header" from
5194          * "== Header ==".
5195          */
5196         public function guessSectionNameFromWikiText( $text ) {
5197                 # Strip out wikitext links(they break the anchor)
5198                 $text = $this->stripSectionName( $text );
5199                 $headline = Sanitizer::decodeCharReferences( $text );
5200                 # strip out HTML
5201                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5202                 $headline = trim( $headline );
5203                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5204                 $replacearray = array(
5205                         '%3A' => ':',
5206                         '%' => '.'
5207                 );
5208                 return str_replace(
5209                         array_keys( $replacearray ),
5210                         array_values( $replacearray ),
5211                         $sectionanchor );
5212         }
5213
5214         /**
5215          * Strips a text string of wikitext for use in a section anchor
5216          *
5217          * Accepts a text string and then removes all wikitext from the
5218          * string and leaves only the resultant text (i.e. the result of
5219          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5220          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5221          * to create valid section anchors by mimicing the output of the
5222          * parser when headings are parsed.
5223          *
5224          * @param $text string Text string to be stripped of wikitext
5225          * for use in a Section anchor
5226          * @return Filtered text string
5227          */
5228         public function stripSectionName( $text ) {
5229                 # Strip internal link markup
5230                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5231                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5232
5233                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5234                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5235                 # on how many empty links there are on the page - need to figure that out.
5236                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5237
5238                 # Parse wikitext quotes (italics & bold)
5239                 $text = $this->doQuotes($text);
5240
5241                 # Strip HTML tags
5242                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5243                 return $text;
5244         }
5245
5246         /**
5247          * strip/replaceVariables/unstrip for preprocessor regression testing
5248          */
5249         function srvus( $text ) {
5250                 $text = $this->replaceVariables( $text );
5251                 $text = $this->mStripState->unstripBoth( $text );
5252                 $text = Sanitizer::removeHTMLtags( $text );
5253                 return $text;
5254         }
5255 }
5256
5257 /**
5258  * @todo document, briefly.
5259  * @addtogroup Parser
5260  */
5261 class OnlyIncludeReplacer {
5262         var $output = '';
5263
5264         function replace( $matches ) {
5265                 if ( substr( $matches[1], -1 ) == "\n" ) {
5266                         $this->output .= substr( $matches[1], 0, -1 );
5267                 } else {
5268                         $this->output .= $matches[1];
5269                 }
5270         }
5271 }
5272
5273 /**
5274  * @todo document, briefly.
5275  * @addtogroup Parser
5276  */
5277 class StripState {
5278         var $general, $nowiki;
5279
5280         function __construct() {
5281                 $this->general = new ReplacementArray;
5282                 $this->nowiki = new ReplacementArray;
5283         }
5284
5285         function unstripGeneral( $text ) {
5286                 wfProfileIn( __METHOD__ );
5287                 do {
5288                         $oldText = $text;
5289                         $text = $this->general->replace( $text );
5290                 } while ( $text != $oldText );
5291                 wfProfileOut( __METHOD__ );
5292                 return $text;
5293         }
5294
5295         function unstripNoWiki( $text ) {
5296                 wfProfileIn( __METHOD__ );
5297                 do {
5298                         $oldText = $text;
5299                         $text = $this->nowiki->replace( $text );
5300                 } while ( $text != $oldText );
5301                 wfProfileOut( __METHOD__ );
5302                 return $text;
5303         }
5304
5305         function unstripBoth( $text ) {
5306                 wfProfileIn( __METHOD__ );
5307                 do {
5308                         $oldText = $text;
5309                         $text = $this->general->replace( $text );
5310                         $text = $this->nowiki->replace( $text );
5311                 } while ( $text != $oldText );
5312                 wfProfileOut( __METHOD__ );
5313                 return $text;
5314         }
5315 }
5316
5317 /**
5318  * An expansion frame, used as a context to expand the result of preprocessToDom()
5319  */
5320 class PPFrame {
5321         var $parser, $title;
5322         var $titleCache;
5323
5324         const NO_ARGS = 1;
5325         const NO_TEMPLATES = 2;
5326         const STRIP_COMMENTS = 4;
5327         const NO_IGNORE = 8;
5328
5329         const RECOVER_ORIG = 11;
5330
5331         /**
5332          * Construct a new preprocessor frame.
5333          * @param Parser $parser The parent parser
5334          * @param Title $title The context title, or false if there isn't one
5335          */
5336         function __construct( $parser ) {
5337                 $this->parser = $parser;
5338                 $this->title = $parser->mTitle;
5339                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
5340         }
5341
5342         /**
5343          * Create a new child frame
5344          * $args is optionally a DOMNodeList containing the template arguments
5345          */
5346         function newChild( $args = false, $title = false ) {
5347                 $assocArgs = array();
5348                 if ( $title === false ) {
5349                         $title = $this->title;
5350                 }
5351                 if ( $args !== false ) {
5352                         $xpath = false;
5353                         foreach ( $args as $arg ) {
5354                                 if ( !$xpath ) {
5355                                         $xpath = new DOMXPath( $arg->ownerDocument );
5356                                 }
5357
5358                                 $nameNodes = $xpath->query( 'name', $arg );
5359                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5360                                         // Numbered parameter
5361                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5362                                 } else {
5363                                         // Named parameter
5364                                         $name = $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS );
5365                                 }
5366
5367                                 $value = $xpath->query( 'value', $arg );
5368                                 $assocArgs[$name] = $value->item( 0 );
5369                         }
5370                 }
5371                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5372         }
5373
5374         /**
5375          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5376          * using the current context
5377          * @param $root the node
5378          */
5379         function expand( $root, $flags = 0 ) {
5380                 if ( is_string( $root ) ) {
5381                         return $root;
5382                 }
5383
5384                 if ( $this->parser->ot['html']
5385                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5386                 {
5387                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5388                 }
5389
5390                 if ( is_array( $root ) ) {
5391                         $s = '';
5392                         foreach ( $root as $node ) {
5393                                 $s .= $this->expand( $node, $flags );
5394                         }
5395                 } elseif ( $root instanceof DOMNodeList ) {
5396                         $s = '';
5397                         foreach ( $root as $node ) {
5398                                 $s .= $this->expand( $node, $flags );
5399                         }
5400                 } elseif ( $root instanceof DOMNode ) {
5401                         if ( $root->nodeType == XML_TEXT_NODE ) {
5402                                 $s = $root->nodeValue;
5403                         } elseif ( $root->nodeName == 'template' ) {
5404                                 # Double-brace expansion
5405                                 $xpath = new DOMXPath( $root->ownerDocument );
5406                                 $titles = $xpath->query( 'title', $root );
5407                                 $title = $titles->item( 0 );
5408                                 $parts = $xpath->query( 'part', $root );
5409                                 if ( $flags & self::NO_TEMPLATES ) {
5410                                         $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
5411                                 } else {
5412                                         $lineStart = $root->getAttribute( 'lineStart' );
5413                                         $params = array(
5414                                                 'title' => $title,
5415                                                 'parts' => $parts,
5416                                                 'lineStart' => $lineStart,
5417                                                 'text' => 'FIXME' );
5418                                         $s = $this->parser->braceSubstitution( $params, $this );
5419                                 }
5420                         } elseif ( $root->nodeName == 'tplarg' ) {
5421                                 # Triple-brace expansion
5422                                 $xpath = new DOMXPath( $root->ownerDocument );
5423                                 $titles = $xpath->query( 'title', $root );
5424                                 $title = $titles->item( 0 );
5425                                 $parts = $xpath->query( 'part', $root );
5426                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5427                                         $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
5428                                 } else {
5429                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5430                                         $s = $this->parser->argSubstitution( $params, $this );
5431                                 }
5432                         } elseif ( $root->nodeName == 'comment' ) {
5433                                 # HTML-style comment
5434                                 if ( $this->parser->ot['html']
5435                                         || ( $this->parser->ot['pre'] && $this->parser->mOptions->getRemoveComments() )
5436                                         || ( $flags & self::STRIP_COMMENTS ) )
5437                                 {
5438                                         $s = '';
5439                                 } else {
5440                                         $s = $root->textContent;
5441                                 }
5442                         } elseif ( $root->nodeName == 'ignore' ) {
5443                                 # Output suppression used by <includeonly> etc.
5444                                 # OT_WIKI will only respect <ignore> in substed templates.
5445                                 # The other output types respect it unless NO_IGNORE is set.
5446                                 # extractSections() sets NO_IGNORE and so never respects it.
5447                                 if ( ( !isset( $this->parent ) && $this->parser->ot['wiki'] ) || ( $flags & self::NO_IGNORE ) ) {
5448                                         $s = $root->textContent;
5449                                 } else {
5450                                         $s = '';
5451                                 }
5452                         } elseif ( $root->nodeName == 'ext' ) {
5453                                 # Extension tag
5454                                 $xpath = new DOMXPath( $root->ownerDocument );
5455                                 $names = $xpath->query( 'name', $root );
5456                                 $attrs = $xpath->query( 'attr', $root );
5457                                 $inners = $xpath->query( 'inner', $root );
5458                                 $closes = $xpath->query( 'close', $root );
5459                                 $params = array(
5460                                         'name' => $names->item( 0 ),
5461                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5462                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5463                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5464                                 );
5465                                 $s = $this->parser->extensionSubstitution( $params, $this );
5466                         } elseif ( $root->nodeName == 'h' ) {
5467                                 # Heading
5468                                 $s = $this->expand( $root->childNodes, $flags );
5469
5470                                 if ( $this->parser->ot['html'] ) {
5471                                         # Insert heading index marker
5472                                         $headingIndex = $root->getAttribute( 'i' );
5473                                         $titleText = $this->title->getPrefixedDBkey();
5474                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5475                                         $serial = count( $this->parser->mHeadings ) - 1;
5476                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5477                                         $count = $root->getAttribute( 'level' );
5478                                         $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5479                                         $this->parser->mStripState->general->setPair( $marker, '' );
5480                                 }
5481                         } else {
5482                                 # Generic recursive expansion
5483                                 $s = '';
5484                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5485                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5486                                                 $s .= $node->nodeValue;
5487                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5488                                                 $s .= $this->expand( $node, $flags );
5489                                         }
5490                                 }
5491                         }
5492                 } else {
5493                         throw new MWException( __METHOD__.': Invalid parameter type' );
5494                 }
5495                 return $s;
5496         }
5497
5498         function implodeWithFlags( $sep, $flags /*, ... */ ) {
5499                 $args = array_slice( func_get_args(), 2 );
5500
5501                 $first = true;
5502                 $s = '';
5503                 foreach ( $args as $root ) {
5504                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5505                                 $root = array( $root );
5506                         }
5507                         foreach ( $root as $node ) {
5508                                 if ( $first ) {
5509                                         $first = false;
5510                                 } else {
5511                                         $s .= $sep;
5512                                 }
5513                                 $s .= $this->expand( $node, $flags );
5514                         }
5515                 }
5516                 return $s;
5517         }
5518
5519         function implode( $sep /*, ... */ ) {
5520                 $args = func_get_args();
5521                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
5522                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5523         }
5524
5525         /**
5526          * Split an <arg> or <template> node into a three-element array:
5527          *    DOMNode name, string index and DOMNode value
5528          */
5529         function splitBraceNode( $node ) {
5530                 $xpath = new DOMXPath( $node->ownerDocument );
5531                 $names = $xpath->query( 'name', $node );
5532                 $values = $xpath->query( 'value', $node );
5533                 if ( !$names->length || !$values->length ) {
5534                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
5535                 }
5536                 $name = $names->item( 0 );
5537                 $index = $name->getAttribute( 'index' );
5538                 return array( $name, $index, $values->item( 0 ) );
5539         }
5540
5541         /**
5542          * Split an <ext> node into an associative array containing name, attr, inner and close
5543          * All values in the resulting array are DOMNodes. Inner and close are optional.
5544          */
5545         function splitExtNode( $node ) {
5546                 $xpath = new DOMXPath( $node->ownerDocument );
5547                 $names = $xpath->query( 'name', $node );
5548                 $attrs = $xpath->query( 'attr', $node );
5549                 $inners = $xpath->query( 'inner', $node );
5550                 $closes = $xpath->query( 'close', $node );
5551                 if ( !$names->length || !$attrs->length ) {
5552                         throw new MWException( 'Invalid ext node passed to ' . __METHOD__ );
5553                 }
5554                 $parts = array(
5555                         'name' => $names->item( 0 ),
5556                         'attr' => $attrs->item( 0 ) );
5557                 if ( $inners->length ) {
5558                         $parts['inner'] = $inners->item( 0 );
5559                 }
5560                 if ( $closes->length ) {
5561                         $parts['close'] = $closes->item( 0 );
5562                 }
5563                 return $parts;
5564         }
5565
5566         function __toString() {
5567                 return 'frame{}';
5568         }
5569
5570         function getPDBK( $level = false ) {
5571                 if ( $level === false ) {
5572                         return $this->title->getPrefixedDBkey();
5573                 } else {
5574                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
5575                 }
5576         }
5577 }
5578
5579 /**
5580  * Expansion frame with template arguments
5581  */
5582 class PPTemplateFrame extends PPFrame {
5583         var $parser, $args, $parent;
5584         var $titleCache;
5585
5586         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5587                 $this->parser = $parser;
5588                 $this->parent = $parent;
5589                 $this->args = $args;
5590                 $this->title = $title;
5591                 $this->titleCache = $parent->titleCache;
5592                 $this->titleCache[] = $title ? $title->getPrefixedDBkey() : false;
5593         }
5594
5595         function __toString() {
5596                 $s = 'tplframe{';
5597                 $first = true;
5598                 foreach ( $this->args as $name => $value ) {
5599                         if ( $first ) {
5600                                 $first = false;
5601                         } else {
5602                                 $s .= ', ';
5603                         }
5604                         $s .= "\"$name\":\"" .
5605                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5606                 }
5607                 $s .= '}';
5608                 return $s;
5609         }
5610 }
5611