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  *  $wgMaxArticleSize*
  38  *
  39  *  * only within ParserOptions
  40  * </pre>
  41  *
  42  * @addtogroup Parser
  43  */
  44 class Parser
  45 {
  46         /**
  47          * Update this version number when the ParserOutput format
  48          * changes in an incompatible way, so the parser cache
  49          * can automatically discard old data.
  50          */
  51         const VERSION = '1.6.3';
  52
  53         # Flags for Parser::setFunctionHook
  54         # Also available as global constants from Defines.php
  55         const SFH_NO_HASH = 1;
  56         const SFH_OBJECT_ARGS = 2;
  57
  58         # Constants needed for external link processing
  59         # Everything except bracket, space, or control characters
  60         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  61         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  62                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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         /**#@+
  75          * @private
  76          */
  77         # Persistent:
  78         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  79                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  80                 $mExtLinkBracketedRegex;
  81
  82         # Cleared with clearState():
  83         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  84         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  85         var $mInterwikiLinkHolders, $mLinkHolders;
  86         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
  87         var $mTplExpandCache,// empty-frame expansion cache
  88             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
  89                                 // in this path. Used for loop detection.
  90         var $mTplRedirCache, $mTplDomCache, $mHeadings;
  91
  92         # Temporary
  93         # These are variables reset at least once per parse regardless of $clearState
  94         var $mOptions,      // ParserOptions object
  95                 $mTitle,        // Title context, used for self-link rendering and similar things
  96                 $mOutputType,   // Output type, one of the OT_xxx constants
  97                 $ot,            // Shortcut alias, see setOutputType()
  98                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
  99                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 100                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 101
 102         /**#@-*/
 103
 104         /**
 105          * Constructor
 106          *
 107          * @public
 108          */
 109         function __construct( $conf = array() ) {
 110                 $this->mTagHooks = array();
 111                 $this->mTransparentTagHooks = array();
 112                 $this->mFunctionHooks = array();
 113                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 114                 $this->mStripList = array( 'nowiki', 'gallery' );
 115                 $this->mMarkerSuffix = "-QINU\x7f";
 116                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 117                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 118                 $this->mFirstCall = true;
 119         }
 120
 121         /**
 122          * Do various kinds of initialisation on the first call of the parser
 123          */
 124         function firstCallInit() {
 125                 if ( !$this->mFirstCall ) {
 126                         return;
 127                 }
 128
 129                 wfProfileIn( __METHOD__ );
 130                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 131
 132                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 133
 134                 # Syntax for arguments (see self::setFunctionHook):
 135                 #  "name for lookup in localized magic words array",
 136                 #  function callback,
 137                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 138                 #    instead of {{#int:...}})
 139                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 140                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 164                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 166
 167                 if ( $wgAllowDisplayTitle ) {
 168                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 169                 }
 170                 if ( $wgAllowSlowParserFunctions ) {
 171                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 172                 }
 173
 174                 $this->initialiseVariables();
 175                 $this->mFirstCall = false;
 176                 wfProfileOut( __METHOD__ );
 177         }
 178
 179         /**
 180          * Clear Parser state
 181          *
 182          * @private
 183          */
 184         function clearState() {
 185                 wfProfileIn( __METHOD__ );
 186                 if ( $this->mFirstCall ) {
 187                         $this->firstCallInit();
 188                 }
 189                 $this->mOutput = new ParserOutput;
 190                 $this->mAutonumber = 0;
 191                 $this->mLastSection = '';
 192                 $this->mDTopen = false;
 193                 $this->mIncludeCount = array();
 194                 $this->mStripState = new StripState;
 195                 $this->mArgStack = false;
 196                 $this->mInPre = false;
 197                 $this->mInterwikiLinkHolders = array(
 198                         'texts' => array(),
 199                         'titles' => array()
 200                 );
 201                 $this->mLinkHolders = array(
 202                         'namespaces' => array(),
 203                         'dbkeys' => array(),
 204                         'queries' => array(),
 205                         'texts' => array(),
 206                         'titles' => array()
 207                 );
 208                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 209
 210                 /**
 211                  * Prefix for temporary replacement strings for the multipass parser.
 212                  * \x07 should never appear in input as it's disallowed in XML.
 213                  * Using it at the front also gives us a little extra robustness
 214                  * since it shouldn't match when butted up against identifier-like
 215                  * string constructs.
 216                  *
 217                  * Must not consist of all title characters, or else it will change
 218                  * the behaviour of <nowiki> in a link.
 219                  */
 220                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 221                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplatePath = array();
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236
 237                 wfRunHooks( 'ParserClearState', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         function setOutputType( $ot ) {
 242                 $this->mOutputType = $ot;
 243                 // Shortcut alias
 244                 $this->ot = array(
 245                         'html' => $ot == OT_HTML,
 246                         'wiki' => $ot == OT_WIKI,
 247                         'msg' => $ot == OT_MSG,
 248                         'pre' => $ot == OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Accessor for mUniqPrefix.
 254          *
 255          * @public
 256          */
 257         function uniqPrefix() {
 258                 if( !isset( $this->mUniqPrefix ) ) {
 259                         // @fixme this is probably *horribly wrong*
 260                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 261                         // if this is called for a parser cache hit, the parser may not
 262                         // have ever been initialized in the first place.
 263                         // Not really sure what the heck is supposed to be going on here.
 264                         return '';
 265                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 266                 }
 267                 return $this->mUniqPrefix;
 268         }
 269
 270         /**
 271          * Convert wikitext to HTML
 272          * Do not call this function recursively.
 273          *
 274          * @param string $text Text we want to parse
 275          * @param Title &$title A title object
 276          * @param array $options
 277          * @param boolean $linestart
 278          * @param boolean $clearState
 279          * @param int $revid number to pass in {{REVISIONID}}
 280          * @return ParserOutput a ParserOutput
 281          */
 282         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 283                 /**
 284                  * First pass--just handle <nowiki> sections, pass the rest off
 285                  * to internalParse() which does all the real work.
 286                  */
 287
 288                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 289                 $fname = 'Parser::parse-' . wfGetCaller();
 290                 wfProfileIn( __METHOD__ );
 291                 wfProfileIn( $fname );
 292
 293                 if ( $clearState ) {
 294                         $this->clearState();
 295                 }
 296
 297                 $this->mOptions = $options;
 298                 $this->mTitle =& $title;
 299                 $oldRevisionId = $this->mRevisionId;
 300                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 301                 if( $revid !== null ) {
 302                         $this->mRevisionId = $revid;
 303                         $this->mRevisionTimestamp = null;
 304                 }
 305                 $this->setOutputType( OT_HTML );
 306                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 307                 # No more strip!
 308                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 309                 $text = $this->internalParse( $text );
 310                 $text = $this->mStripState->unstripGeneral( $text );
 311
 312                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 313                 $fixtags = array(
 314                         # french spaces, last one Guillemet-left
 315                         # only if there is something before the space
 316                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 317                         # french spaces, Guillemet-right
 318                         '/(\\302\\253) /' => '\\1&nbsp;',
 319                 );
 320                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 321
 322                 # only once and last
 323                 $text = $this->doBlockLevels( $text, $linestart );
 324
 325                 $this->replaceLinkHolders( $text );
 326
 327                 # the position of the parserConvert() call should not be changed. it
 328                 # assumes that the links are all replaced and the only thing left
 329                 # is the <nowiki> mark.
 330                 # Side-effects: this calls $this->mOutput->setTitleText()
 331                 $text = $wgContLang->parserConvert( $text, $this );
 332
 333                 $text = $this->mStripState->unstripNoWiki( $text );
 334
 335                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 336
 337 //!JF Move to its own function
 338
 339                 $uniq_prefix = $this->mUniqPrefix;
 340                 $matches = array();
 341                 $elements = array_keys( $this->mTransparentTagHooks );
 342                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 343
 344                 foreach( $matches as $marker => $data ) {
 345                         list( $element, $content, $params, $tag ) = $data;
 346                         $tagName = strtolower( $element );
 347                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 348                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 349                                         array( $content, $params, $this ) );
 350                         } else {
 351                                 $output = $tag;
 352                         }
 353                         $this->mStripState->general->setPair( $marker, $output );
 354                 }
 355                 $text = $this->mStripState->unstripGeneral( $text );
 356
 357                 $text = Sanitizer::normalizeCharReferences( $text );
 358
 359                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 360                         $text = Parser::tidy($text);
 361                 } else {
 362                         # attempt to sanitize at least some nesting problems
 363                         # (bug #2702 and quite a few others)
 364                         $tidyregs = array(
 365                                 # ''Something [http://www.cool.com cool''] -->
 366                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 367                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 368                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 369                                 # fix up an anchor inside another anchor, only
 370                                 # at least for a single single nested link (bug 3695)
 371                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 372                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 373                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 374                                 # contains a div, so fix it up here; replace
 375                                 # div with escaped text
 376                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 377                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 378                                 # remove empty italic or bold tag pairs, some
 379                                 # introduced by rules above
 380                                 '/<([bi])><\/\\1>/' => '',
 381                         );
 382
 383                         $text = preg_replace(
 384                                 array_keys( $tidyregs ),
 385                                 array_values( $tidyregs ),
 386                                 $text );
 387                 }
 388
 389                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 390
 391                 # Information on include size limits, for the benefit of users who try to skirt them
 392                 if ( $this->mOptions->getEnableLimitReport() ) {
 393                         $max = $this->mOptions->getMaxIncludeSize();
 394                         $limitReport =
 395                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 396                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 397                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 398                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 399                         $text .= "\n<!-- \n$limitReport-->\n";
 400                 }
 401                 $this->mOutput->setText( $text );
 402                 $this->mRevisionId = $oldRevisionId;
 403                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 404                 wfProfileOut( $fname );
 405                 wfProfileOut( __METHOD__ );
 406
 407                 return $this->mOutput;
 408         }
 409
 410         /**
 411          * Recursive parser entry point that can be called from an extension tag
 412          * hook.
 413          */
 414         function recursiveTagParse( $text ) {
 415                 wfProfileIn( __METHOD__ );
 416                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 417                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 418                 $text = $this->internalParse( $text );
 419                 wfProfileOut( __METHOD__ );
 420                 return $text;
 421         }
 422
 423         /**
 424          * Expand templates and variables in the text, producing valid, static wikitext.
 425          * Also removes comments.
 426          */
 427         function preprocess( $text, $title, $options, $revid = null ) {
 428                 wfProfileIn( __METHOD__ );
 429                 $this->clearState();
 430                 $this->setOutputType( OT_PREPROCESS );
 431                 $this->mOptions = $options;
 432                 $this->mTitle = $title;
 433                 if( $revid !== null ) {
 434                         $this->mRevisionId = $revid;
 435                 }
 436                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 437                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 438                 $text = $this->replaceVariables( $text );
 439                 if ( $this->mOptions->getRemoveComments() ) {
 440                         $text = Sanitizer::removeHTMLcomments( $text );
 441                 }
 442                 $text = $this->mStripState->unstripBoth( $text );
 443                 wfProfileOut( __METHOD__ );
 444                 return $text;
 445         }
 446
 447         /**
 448          * Get a random string
 449          *
 450          * @private
 451          * @static
 452          */
 453         function getRandomString() {
 454                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 455         }
 456
 457         function &getTitle() { return $this->mTitle; }
 458         function getOptions() { return $this->mOptions; }
 459
 460         function getFunctionLang() {
 461                 global $wgLang, $wgContLang;
 462                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 463         }
 464
 465         /**
 466          * Replaces all occurrences of HTML-style comments and the given tags
 467          * in the text with a random marker and returns teh next text. The output
 468          * parameter $matches will be an associative array filled with data in
 469          * the form:
 470          *   'UNIQ-xxxxx' => array(
 471          *     'element',
 472          *     'tag content',
 473          *     array( 'param' => 'x' ),
 474          *     '<element param="x">tag content</element>' ) )
 475          *
 476          * @param $elements list of element names. Comments are always extracted.
 477          * @param $text Source text string.
 478          * @param $uniq_prefix
 479          *
 480          * @public
 481          * @static
 482          */
 483         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 484                 static $n = 1;
 485                 $stripped = '';
 486                 $matches = array();
 487
 488                 $taglist = implode( '|', $elements );
 489                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 490
 491                 while ( '' != $text ) {
 492                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 493                         $stripped .= $p[0];
 494                         if( count( $p ) < 5 ) {
 495                                 break;
 496                         }
 497                         if( count( $p ) > 5 ) {
 498                                 // comment
 499                                 $element    = $p[4];
 500                                 $attributes = '';
 501                                 $close      = '';
 502                                 $inside     = $p[5];
 503                         } else {
 504                                 // tag
 505                                 $element    = $p[1];
 506                                 $attributes = $p[2];
 507                                 $close      = $p[3];
 508                                 $inside     = $p[4];
 509                         }
 510
 511                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 512                         $stripped .= $marker;
 513
 514                         if ( $close === '/>' ) {
 515                                 // Empty element tag, <tag />
 516                                 $content = null;
 517                                 $text = $inside;
 518                                 $tail = null;
 519                         } else {
 520                                 if( $element == '!--' ) {
 521                                         $end = '/(-->)/';
 522                                 } else {
 523                                         $end = "/(<\\/$element\\s*>)/i";
 524                                 }
 525                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 526                                 $content = $q[0];
 527                                 if( count( $q ) < 3 ) {
 528                                         # No end tag -- let it run out to the end of the text.
 529                                         $tail = '';
 530                                         $text = '';
 531                                 } else {
 532                                         $tail = $q[1];
 533                                         $text = $q[2];
 534                                 }
 535                         }
 536
 537                         $matches[$marker] = array( $element,
 538                                 $content,
 539                                 Sanitizer::decodeTagAttributes( $attributes ),
 540                                 "<$element$attributes$close$content$tail" );
 541                 }
 542                 return $stripped;
 543         }
 544
 545         /**
 546          * Get a list of strippable XML-like elements
 547          */
 548         function getStripList() {
 549                 global $wgRawHtml;
 550                 $elements = $this->mStripList;
 551                 if( $wgRawHtml ) {
 552                         $elements[] = 'html';
 553                 }
 554                 if( $this->mOptions->getUseTeX() ) {
 555                         $elements[] = 'math';
 556                 }
 557                 return $elements;
 558         }
 559
 560         /**
 561          * @deprecated use replaceVariables
 562          */
 563         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 564                 return $text;
 565         }
 566
 567         /**
 568          * Restores pre, math, and other extensions removed by strip()
 569          *
 570          * always call unstripNoWiki() after this one
 571          * @private
 572          * @deprecated use $this->mStripState->unstrip()
 573          */
 574         function unstrip( $text, $state ) {
 575                 return $state->unstripGeneral( $text );
 576         }
 577
 578         /**
 579          * Always call this after unstrip() to preserve the order
 580          *
 581          * @private
 582          * @deprecated use $this->mStripState->unstrip()
 583          */
 584         function unstripNoWiki( $text, $state ) {
 585                 return $state->unstripNoWiki( $text );
 586         }
 587
 588         /**
 589          * @deprecated use $this->mStripState->unstripBoth()
 590          */
 591         function unstripForHTML( $text ) {
 592                 return $this->mStripState->unstripBoth( $text );
 593         }
 594
 595         /**
 596          * Add an item to the strip state
 597          * Returns the unique tag which must be inserted into the stripped text
 598          * The tag will be replaced with the original text in unstrip()
 599          *
 600          * @private
 601          */
 602         function insertStripItem( $text ) {
 603                 static $n = 0;
 604                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 605                 ++$n;
 606                 $this->mStripState->general->setPair( $rnd, $text );
 607                 return $rnd;
 608         }
 609
 610         /**
 611          * Interface with html tidy, used if $wgUseTidy = true.
 612          * If tidy isn't able to correct the markup, the original will be
 613          * returned in all its glory with a warning comment appended.
 614          *
 615          * Either the external tidy program or the in-process tidy extension
 616          * will be used depending on availability. Override the default
 617          * $wgTidyInternal setting to disable the internal if it's not working.
 618          *
 619          * @param string $text Hideous HTML input
 620          * @return string Corrected HTML output
 621          * @public
 622          * @static
 623          */
 624         function tidy( $text ) {
 625                 global $wgTidyInternal;
 626                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 627 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 628 '<head><title>test</title></head><body>'.$text.'</body></html>';
 629                 if( $wgTidyInternal ) {
 630                         $correctedtext = Parser::internalTidy( $wrappedtext );
 631                 } else {
 632                         $correctedtext = Parser::externalTidy( $wrappedtext );
 633                 }
 634                 if( is_null( $correctedtext ) ) {
 635                         wfDebug( "Tidy error detected!\n" );
 636                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 637                 }
 638                 return $correctedtext;
 639         }
 640
 641         /**
 642          * Spawn an external HTML tidy process and get corrected markup back from it.
 643          *
 644          * @private
 645          * @static
 646          */
 647         function externalTidy( $text ) {
 648                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 649                 $fname = 'Parser::externalTidy';
 650                 wfProfileIn( $fname );
 651
 652                 $cleansource = '';
 653                 $opts = ' -utf8';
 654
 655                 $descriptorspec = array(
 656                         0 => array('pipe', 'r'),
 657                         1 => array('pipe', 'w'),
 658                         2 => array('file', wfGetNull(), 'a')
 659                 );
 660                 $pipes = array();
 661                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 662                 if (is_resource($process)) {
 663                         // Theoretically, this style of communication could cause a deadlock
 664                         // here. If the stdout buffer fills up, then writes to stdin could
 665                         // block. This doesn't appear to happen with tidy, because tidy only
 666                         // writes to stdout after it's finished reading from stdin. Search
 667                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 668                         fwrite($pipes[0], $text);
 669                         fclose($pipes[0]);
 670                         while (!feof($pipes[1])) {
 671                                 $cleansource .= fgets($pipes[1], 1024);
 672                         }
 673                         fclose($pipes[1]);
 674                         proc_close($process);
 675                 }
 676
 677                 wfProfileOut( $fname );
 678
 679                 if( $cleansource == '' && $text != '') {
 680                         // Some kind of error happened, so we couldn't get the corrected text.
 681                         // Just give up; we'll use the source text and append a warning.
 682                         return null;
 683                 } else {
 684                         return $cleansource;
 685                 }
 686         }
 687
 688         /**
 689          * Use the HTML tidy PECL extension to use the tidy library in-process,
 690          * saving the overhead of spawning a new process.
 691          *
 692          * 'pear install tidy' should be able to compile the extension module.
 693          *
 694          * @private
 695          * @static
 696          */
 697         function internalTidy( $text ) {
 698                 global $wgTidyConf, $IP, $wgDebugTidy;
 699                 $fname = 'Parser::internalTidy';
 700                 wfProfileIn( $fname );
 701
 702                 $tidy = new tidy;
 703                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 704                 $tidy->cleanRepair();
 705                 if( $tidy->getStatus() == 2 ) {
 706                         // 2 is magic number for fatal error
 707                         // http://www.php.net/manual/en/function.tidy-get-status.php
 708                         $cleansource = null;
 709                 } else {
 710                         $cleansource = tidy_get_output( $tidy );
 711                 }
 712                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 713                         $cleansource .= "<!--\nTidy reports:\n" .
 714                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 715                                 "\n-->";
 716                 }
 717
 718                 wfProfileOut( $fname );
 719                 return $cleansource;
 720         }
 721
 722         /**
 723          * parse the wiki syntax used to render tables
 724          *
 725          * @private
 726          */
 727         function doTableStuff ( $text ) {
 728                 $fname = 'Parser::doTableStuff';
 729                 wfProfileIn( $fname );
 730
 731                 $lines = explode ( "\n" , $text );
 732                 $td_history = array (); // Is currently a td tag open?
 733                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 734                 $tr_history = array (); // Is currently a tr tag open?
 735                 $tr_attributes = array (); // history of tr attributes
 736                 $has_opened_tr = array(); // Did this table open a <tr> element?
 737                 $indent_level = 0; // indent level of the table
 738                 foreach ( $lines as $key => $line )
 739                 {
 740                         $line = trim ( $line );
 741
 742                         if( $line == '' ) { // empty line, go to next line
 743                                 continue;
 744                         }
 745                         $first_character = $line{0};
 746                         $matches = array();
 747
 748                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 749                                 // First check if we are starting a new table
 750                                 $indent_level = strlen( $matches[1] );
 751
 752                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 753                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 754
 755                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 756                                 array_push ( $td_history , false );
 757                                 array_push ( $last_tag_history , '' );
 758                                 array_push ( $tr_history , false );
 759                                 array_push ( $tr_attributes , '' );
 760                                 array_push ( $has_opened_tr , false );
 761                         } else if ( count ( $td_history ) == 0 ) {
 762                                 // Don't do any of the following
 763                                 continue;
 764                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 765                                 // We are ending a table
 766                                 $line = '</table>' . substr ( $line , 2 );
 767                                 $last_tag = array_pop ( $last_tag_history );
 768
 769                                 if ( !array_pop ( $has_opened_tr ) ) {
 770                                         $line = "<tr><td></td></tr>{$line}";
 771                                 }
 772
 773                                 if ( array_pop ( $tr_history ) ) {
 774                                         $line = "</tr>{$line}";
 775                                 }
 776
 777                                 if ( array_pop ( $td_history ) ) {
 778                                         $line = "</{$last_tag}>{$line}";
 779                                 }
 780                                 array_pop ( $tr_attributes );
 781                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 782                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 783                                 // Now we have a table row
 784                                 $line = preg_replace( '#^\|-+#', '', $line );
 785
 786                                 // Whats after the tag is now only attributes
 787                                 $attributes = $this->mStripState->unstripBoth( $line );
 788                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 789                                 array_pop ( $tr_attributes );
 790                                 array_push ( $tr_attributes , $attributes );
 791
 792                                 $line = '';
 793                                 $last_tag = array_pop ( $last_tag_history );
 794                                 array_pop ( $has_opened_tr );
 795                                 array_push ( $has_opened_tr , true );
 796
 797                                 if ( array_pop ( $tr_history ) ) {
 798                                         $line = '</tr>';
 799                                 }
 800
 801                                 if ( array_pop ( $td_history ) ) {
 802                                         $line = "</{$last_tag}>{$line}";
 803                                 }
 804
 805                                 $lines[$key] = $line;
 806                                 array_push ( $tr_history , false );
 807                                 array_push ( $td_history , false );
 808                                 array_push ( $last_tag_history , '' );
 809                         }
 810                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 811                                 // This might be cell elements, td, th or captions
 812                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 813                                         $first_character = '+';
 814                                         $line = substr ( $line , 1 );
 815                                 }
 816
 817                                 $line = substr ( $line , 1 );
 818
 819                                 if ( $first_character == '!' ) {
 820                                         $line = str_replace ( '!!' , '||' , $line );
 821                                 }
 822
 823                                 // Split up multiple cells on the same line.
 824                                 // FIXME : This can result in improper nesting of tags processed
 825                                 // by earlier parser steps, but should avoid splitting up eg
 826                                 // attribute values containing literal "||".
 827                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 828
 829                                 $lines[$key] = '';
 830
 831                                 // Loop through each table cell
 832                                 foreach ( $cells as $cell )
 833                                 {
 834                                         $previous = '';
 835                                         if ( $first_character != '+' )
 836                                         {
 837                                                 $tr_after = array_pop ( $tr_attributes );
 838                                                 if ( !array_pop ( $tr_history ) ) {
 839                                                         $previous = "<tr{$tr_after}>\n";
 840                                                 }
 841                                                 array_push ( $tr_history , true );
 842                                                 array_push ( $tr_attributes , '' );
 843                                                 array_pop ( $has_opened_tr );
 844                                                 array_push ( $has_opened_tr , true );
 845                                         }
 846
 847                                         $last_tag = array_pop ( $last_tag_history );
 848
 849                                         if ( array_pop ( $td_history ) ) {
 850                                                 $previous = "</{$last_tag}>{$previous}";
 851                                         }
 852
 853                                         if ( $first_character == '|' ) {
 854                                                 $last_tag = 'td';
 855                                         } else if ( $first_character == '!' ) {
 856                                                 $last_tag = 'th';
 857                                         } else if ( $first_character == '+' ) {
 858                                                 $last_tag = 'caption';
 859                                         } else {
 860                                                 $last_tag = '';
 861                                         }
 862
 863                                         array_push ( $last_tag_history , $last_tag );
 864
 865                                         // A cell could contain both parameters and data
 866                                         $cell_data = explode ( '|' , $cell , 2 );
 867
 868                                         // Bug 553: Note that a '|' inside an invalid link should not
 869                                         // be mistaken as delimiting cell parameters
 870                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 871                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 872                                         } else if ( count ( $cell_data ) == 1 )
 873                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 874                                         else {
 875                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 876                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 877                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 878                                         }
 879
 880                                         $lines[$key] .= $cell;
 881                                         array_push ( $td_history , true );
 882                                 }
 883                         }
 884                 }
 885
 886                 // Closing open td, tr && table
 887                 while ( count ( $td_history ) > 0 )
 888                 {
 889                         if ( array_pop ( $td_history ) ) {
 890                                 $lines[] = '</td>' ;
 891                         }
 892                         if ( array_pop ( $tr_history ) ) {
 893                                 $lines[] = '</tr>' ;
 894                         }
 895                         if ( !array_pop ( $has_opened_tr ) ) {
 896                                 $lines[] = "<tr><td></td></tr>" ;
 897                         }
 898
 899                         $lines[] = '</table>' ;
 900                 }
 901
 902                 $output = implode ( "\n" , $lines ) ;
 903
 904                 // special case: don't return empty table
 905                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 906                         $output = '';
 907                 }
 908
 909                 wfProfileOut( $fname );
 910
 911                 return $output;
 912         }
 913
 914         /**
 915          * Helper function for parse() that transforms wiki markup into
 916          * HTML. Only called for $mOutputType == OT_HTML.
 917          *
 918          * @private
 919          */
 920         function internalParse( $text ) {
 921                 $isMain = true;
 922                 $fname = 'Parser::internalParse';
 923                 wfProfileIn( $fname );
 924
 925                 # Hook to suspend the parser in this state
 926                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 927                         wfProfileOut( $fname );
 928                         return $text ;
 929                 }
 930
 931                 # Remove <noinclude> tags and <includeonly> sections
 932                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 933                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 934                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 935
 936                 $text = $this->replaceVariables( $text );
 937                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 938                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 939
 940                 // Tables need to come after variable replacement for things to work
 941                 // properly; putting them before other transformations should keep
 942                 // exciting things like link expansions from showing up in surprising
 943                 // places.
 944                 $text = $this->doTableStuff( $text );
 945
 946                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 947
 948                 $text = $this->stripToc( $text );
 949                 $this->stripNoGallery( $text );
 950                 $text = $this->doHeadings( $text );
 951                 if($this->mOptions->getUseDynamicDates()) {
 952                         $df =& DateFormatter::getInstance();
 953                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 954                 }
 955                 $text = $this->doAllQuotes( $text );
 956                 $text = $this->replaceInternalLinks( $text );
 957                 $text = $this->replaceExternalLinks( $text );
 958
 959                 # replaceInternalLinks may sometimes leave behind
 960                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 961                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 962
 963                 $text = $this->doMagicLinks( $text );
 964                 $text = $this->formatHeadings( $text, $isMain );
 965
 966                 wfProfileOut( $fname );
 967                 return $text;
 968         }
 969
 970         /**
 971          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 972          * magic external links.
 973          *
 974          * @private
 975          */
 976         function doMagicLinks( $text ) {
 977                 wfProfileIn( __METHOD__ );
 978                 $text = preg_replace_callback(
 979                         '!(?:                           # Start cases
 980                             <a.*?</a> |                 # Skip link text
 981                             <.*?> |                     # Skip stuff inside HTML elements
 982                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 983                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 984                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 985                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 986                                      [0-9Xx]                 # check digit
 987                                    \b)
 988                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 989                 wfProfileOut( __METHOD__ );
 990                 return $text;
 991         }
 992
 993         function magicLinkCallback( $m ) {
 994                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 995                         # Skip HTML element
 996                         return $m[0];
 997                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 998                         $isbn = $m[2];
 999                         $num = strtr( $isbn, array(
1000                                 '-' => '',
1001                                 ' ' => '',
1002                                 'x' => 'X',
1003                         ));
1004                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1005                         $text = '<a href="' .
1006                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1007                                 "\" class=\"internal\">ISBN $isbn</a>";
1008                 } else {
1009                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1010                                 $keyword = 'RFC';
1011                                 $urlmsg = 'rfcurl';
1012                                 $id = $m[1];
1013                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1014                                 $keyword = 'PMID';
1015                                 $urlmsg = 'pubmedurl';
1016                                 $id = $m[1];
1017                         } else {
1018                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1019                                         substr($m[0], 0, 20 ) . '"' );
1020                         }
1021
1022                         $url = wfMsg( $urlmsg, $id);
1023                         $sk = $this->mOptions->getSkin();
1024                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1025                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1026                 }
1027                 return $text;
1028         }
1029
1030         /**
1031          * Parse headers and return html
1032          *
1033          * @private
1034          */
1035         function doHeadings( $text ) {
1036                 $fname = 'Parser::doHeadings';
1037                 wfProfileIn( $fname );
1038                 for ( $i = 6; $i >= 1; --$i ) {
1039                         $h = str_repeat( '=', $i );
1040                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1041                           "<h$i>\\1</h$i>", $text );
1042                 }
1043                 wfProfileOut( $fname );
1044                 return $text;
1045         }
1046
1047         /**
1048          * Replace single quotes with HTML markup
1049          * @private
1050          * @return string the altered text
1051          */
1052         function doAllQuotes( $text ) {
1053                 $fname = 'Parser::doAllQuotes';
1054                 wfProfileIn( $fname );
1055                 $outtext = '';
1056                 $lines = explode( "\n", $text );
1057                 foreach ( $lines as $line ) {
1058                         $outtext .= $this->doQuotes ( $line ) . "\n";
1059                 }
1060                 $outtext = substr($outtext, 0,-1);
1061                 wfProfileOut( $fname );
1062                 return $outtext;
1063         }
1064
1065         /**
1066          * Helper function for doAllQuotes()
1067          */
1068         public function doQuotes( $text ) {
1069                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1070                 if ( count( $arr ) == 1 )
1071                         return $text;
1072                 else
1073                 {
1074                         # First, do some preliminary work. This may shift some apostrophes from
1075                         # being mark-up to being text. It also counts the number of occurrences
1076                         # of bold and italics mark-ups.
1077                         $i = 0;
1078                         $numbold = 0;
1079                         $numitalics = 0;
1080                         foreach ( $arr as $r )
1081                         {
1082                                 if ( ( $i % 2 ) == 1 )
1083                                 {
1084                                         # If there are ever four apostrophes, assume the first is supposed to
1085                                         # be text, and the remaining three constitute mark-up for bold text.
1086                                         if ( strlen( $arr[$i] ) == 4 )
1087                                         {
1088                                                 $arr[$i-1] .= "'";
1089                                                 $arr[$i] = "'''";
1090                                         }
1091                                         # If there are more than 5 apostrophes in a row, assume they're all
1092                                         # text except for the last 5.
1093                                         else if ( strlen( $arr[$i] ) > 5 )
1094                                         {
1095                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1096                                                 $arr[$i] = "'''''";
1097                                         }
1098                                         # Count the number of occurrences of bold and italics mark-ups.
1099                                         # We are not counting sequences of five apostrophes.
1100                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1101                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1102                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1103                                 }
1104                                 $i++;
1105                         }
1106
1107                         # If there is an odd number of both bold and italics, it is likely
1108                         # that one of the bold ones was meant to be an apostrophe followed
1109                         # by italics. Which one we cannot know for certain, but it is more
1110                         # likely to be one that has a single-letter word before it.
1111                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1112                         {
1113                                 $i = 0;
1114                                 $firstsingleletterword = -1;
1115                                 $firstmultiletterword = -1;
1116                                 $firstspace = -1;
1117                                 foreach ( $arr as $r )
1118                                 {
1119                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1120                                         {
1121                                                 $x1 = substr ($arr[$i-1], -1);
1122                                                 $x2 = substr ($arr[$i-1], -2, 1);
1123                                                 if ($x1 == ' ') {
1124                                                         if ($firstspace == -1) $firstspace = $i;
1125                                                 } else if ($x2 == ' ') {
1126                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1127                                                 } else {
1128                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1129                                                 }
1130                                         }
1131                                         $i++;
1132                                 }
1133
1134                                 # If there is a single-letter word, use it!
1135                                 if ($firstsingleletterword > -1)
1136                                 {
1137                                         $arr [ $firstsingleletterword ] = "''";
1138                                         $arr [ $firstsingleletterword-1 ] .= "'";
1139                                 }
1140                                 # If not, but there's a multi-letter word, use that one.
1141                                 else if ($firstmultiletterword > -1)
1142                                 {
1143                                         $arr [ $firstmultiletterword ] = "''";
1144                                         $arr [ $firstmultiletterword-1 ] .= "'";
1145                                 }
1146                                 # ... otherwise use the first one that has neither.
1147                                 # (notice that it is possible for all three to be -1 if, for example,
1148                                 # there is only one pentuple-apostrophe in the line)
1149                                 else if ($firstspace > -1)
1150                                 {
1151                                         $arr [ $firstspace ] = "''";
1152                                         $arr [ $firstspace-1 ] .= "'";
1153                                 }
1154                         }
1155
1156                         # Now let's actually convert our apostrophic mush to HTML!
1157                         $output = '';
1158                         $buffer = '';
1159                         $state = '';
1160                         $i = 0;
1161                         foreach ($arr as $r)
1162                         {
1163                                 if (($i % 2) == 0)
1164                                 {
1165                                         if ($state == 'both')
1166                                                 $buffer .= $r;
1167                                         else
1168                                                 $output .= $r;
1169                                 }
1170                                 else
1171                                 {
1172                                         if (strlen ($r) == 2)
1173                                         {
1174                                                 if ($state == 'i')
1175                                                 { $output .= '</i>'; $state = ''; }
1176                                                 else if ($state == 'bi')
1177                                                 { $output .= '</i>'; $state = 'b'; }
1178                                                 else if ($state == 'ib')
1179                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1180                                                 else if ($state == 'both')
1181                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1182                                                 else # $state can be 'b' or ''
1183                                                 { $output .= '<i>'; $state .= 'i'; }
1184                                         }
1185                                         else if (strlen ($r) == 3)
1186                                         {
1187                                                 if ($state == 'b')
1188                                                 { $output .= '</b>'; $state = ''; }
1189                                                 else if ($state == 'bi')
1190                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1191                                                 else if ($state == 'ib')
1192                                                 { $output .= '</b>'; $state = 'i'; }
1193                                                 else if ($state == 'both')
1194                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1195                                                 else # $state can be 'i' or ''
1196                                                 { $output .= '<b>'; $state .= 'b'; }
1197                                         }
1198                                         else if (strlen ($r) == 5)
1199                                         {
1200                                                 if ($state == 'b')
1201                                                 { $output .= '</b><i>'; $state = 'i'; }
1202                                                 else if ($state == 'i')
1203                                                 { $output .= '</i><b>'; $state = 'b'; }
1204                                                 else if ($state == 'bi')
1205                                                 { $output .= '</i></b>'; $state = ''; }
1206                                                 else if ($state == 'ib')
1207                                                 { $output .= '</b></i>'; $state = ''; }
1208                                                 else if ($state == 'both')
1209                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1210                                                 else # ($state == '')
1211                                                 { $buffer = ''; $state = 'both'; }
1212                                         }
1213                                 }
1214                                 $i++;
1215                         }
1216                         # Now close all remaining tags.  Notice that the order is important.
1217                         if ($state == 'b' || $state == 'ib')
1218                                 $output .= '</b>';
1219                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1220                                 $output .= '</i>';
1221                         if ($state == 'bi')
1222                                 $output .= '</b>';
1223                         # There might be lonely ''''', so make sure we have a buffer
1224                         if ($state == 'both' && $buffer)
1225                                 $output .= '<b><i>'.$buffer.'</i></b>';
1226                         return $output;
1227                 }
1228         }
1229
1230         /**
1231          * Replace external links
1232          *
1233          * Note: this is all very hackish and the order of execution matters a lot.
1234          * Make sure to run maintenance/parserTests.php if you change this code.
1235          *
1236          * @private
1237          */
1238         function replaceExternalLinks( $text ) {
1239                 global $wgContLang;
1240                 $fname = 'Parser::replaceExternalLinks';
1241                 wfProfileIn( $fname );
1242
1243                 $sk = $this->mOptions->getSkin();
1244
1245                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1246
1247                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1248
1249                 $i = 0;
1250                 while ( $i<count( $bits ) ) {
1251                         $url = $bits[$i++];
1252                         $protocol = $bits[$i++];
1253                         $text = $bits[$i++];
1254                         $trail = $bits[$i++];
1255
1256                         # The characters '<' and '>' (which were escaped by
1257                         # removeHTMLtags()) should not be included in
1258                         # URLs, per RFC 2396.
1259                         $m2 = array();
1260                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1261                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1262                                 $url = substr($url, 0, $m2[0][1]);
1263                         }
1264
1265                         # If the link text is an image URL, replace it with an <img> tag
1266                         # This happened by accident in the original parser, but some people used it extensively
1267                         $img = $this->maybeMakeExternalImage( $text );
1268                         if ( $img !== false ) {
1269                                 $text = $img;
1270                         }
1271
1272                         $dtrail = '';
1273
1274                         # Set linktype for CSS - if URL==text, link is essentially free
1275                         $linktype = ($text == $url) ? 'free' : 'text';
1276
1277                         # No link text, e.g. [http://domain.tld/some.link]
1278                         if ( $text == '' ) {
1279                                 # Autonumber if allowed. See bug #5918
1280                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1281                                         $text = '[' . ++$this->mAutonumber . ']';
1282                                         $linktype = 'autonumber';
1283                                 } else {
1284                                         # Otherwise just use the URL
1285                                         $text = htmlspecialchars( $url );
1286                                         $linktype = 'free';
1287                                 }
1288                         } else {
1289                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1290                                 # Check for trail
1291                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1292                         }
1293
1294                         $text = $wgContLang->markNoConversion($text);
1295
1296                         $url = Sanitizer::cleanUrl( $url );
1297
1298                         # Process the trail (i.e. everything after this link up until start of the next link),
1299                         # replacing any non-bracketed links
1300                         $trail = $this->replaceFreeExternalLinks( $trail );
1301
1302                         # Use the encoded URL
1303                         # This means that users can paste URLs directly into the text
1304                         # Funny characters like &ouml; aren't valid in URLs anyway
1305                         # This was changed in August 2004
1306                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1307
1308                         # Register link in the output object.
1309                         # Replace unnecessary URL escape codes with the referenced character
1310                         # This prevents spammers from hiding links from the filters
1311                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1312                         $this->mOutput->addExternalLink( $pasteurized );
1313                 }
1314
1315                 wfProfileOut( $fname );
1316                 return $s;
1317         }
1318
1319         /**
1320          * Replace anything that looks like a URL with a link
1321          * @private
1322          */
1323         function replaceFreeExternalLinks( $text ) {
1324                 global $wgContLang;
1325                 $fname = 'Parser::replaceFreeExternalLinks';
1326                 wfProfileIn( $fname );
1327
1328                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1329                 $s = array_shift( $bits );
1330                 $i = 0;
1331
1332                 $sk = $this->mOptions->getSkin();
1333
1334                 while ( $i < count( $bits ) ){
1335                         $protocol = $bits[$i++];
1336                         $remainder = $bits[$i++];
1337
1338                         $m = array();
1339                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1340                                 # Found some characters after the protocol that look promising
1341                                 $url = $protocol . $m[1];
1342                                 $trail = $m[2];
1343
1344                                 # special case: handle urls as url args:
1345                                 # http://www.example.com/foo?=http://www.example.com/bar
1346                                 if(strlen($trail) == 0 &&
1347                                         isset($bits[$i]) &&
1348                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1349                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1350                                 {
1351                                         # add protocol, arg
1352                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1353                                         $i += 2;
1354                                         $trail = $m[2];
1355                                 }
1356
1357                                 # The characters '<' and '>' (which were escaped by
1358                                 # removeHTMLtags()) should not be included in
1359                                 # URLs, per RFC 2396.
1360                                 $m2 = array();
1361                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1362                                         $trail = substr($url, $m2[0][1]) . $trail;
1363                                         $url = substr($url, 0, $m2[0][1]);
1364                                 }
1365
1366                                 # Move trailing punctuation to $trail
1367                                 $sep = ',;\.:!?';
1368                                 # If there is no left bracket, then consider right brackets fair game too
1369                                 if ( strpos( $url, '(' ) === false ) {
1370                                         $sep .= ')';
1371                                 }
1372
1373                                 $numSepChars = strspn( strrev( $url ), $sep );
1374                                 if ( $numSepChars ) {
1375                                         $trail = substr( $url, -$numSepChars ) . $trail;
1376                                         $url = substr( $url, 0, -$numSepChars );
1377                                 }
1378
1379                                 $url = Sanitizer::cleanUrl( $url );
1380
1381                                 # Is this an external image?
1382                                 $text = $this->maybeMakeExternalImage( $url );
1383                                 if ( $text === false ) {
1384                                         # Not an image, make a link
1385                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1386                                         # Register it in the output object...
1387                                         # Replace unnecessary URL escape codes with their equivalent characters
1388                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1389                                         $this->mOutput->addExternalLink( $pasteurized );
1390                                 }
1391                                 $s .= $text . $trail;
1392                         } else {
1393                                 $s .= $protocol . $remainder;
1394                         }
1395                 }
1396                 wfProfileOut( $fname );
1397                 return $s;
1398         }
1399
1400         /**
1401          * Replace unusual URL escape codes with their equivalent characters
1402          * @param string
1403          * @return string
1404          * @static
1405          * @todo  This can merge genuinely required bits in the path or query string,
1406          *        breaking legit URLs. A proper fix would treat the various parts of
1407          *        the URL differently; as a workaround, just use the output for
1408          *        statistical records, not for actual linking/output.
1409          */
1410         static function replaceUnusualEscapes( $url ) {
1411                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1412                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1413         }
1414
1415         /**
1416          * Callback function used in replaceUnusualEscapes().
1417          * Replaces unusual URL escape codes with their equivalent character
1418          * @static
1419          * @private
1420          */
1421         private static function replaceUnusualEscapesCallback( $matches ) {
1422                 $char = urldecode( $matches[0] );
1423                 $ord = ord( $char );
1424                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1425                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1426                         // No, shouldn't be escaped
1427                         return $char;
1428                 } else {
1429                         // Yes, leave it escaped
1430                         return $matches[0];
1431                 }
1432         }
1433
1434         /**
1435          * make an image if it's allowed, either through the global
1436          * option or through the exception
1437          * @private
1438          */
1439         function maybeMakeExternalImage( $url ) {
1440                 $sk = $this->mOptions->getSkin();
1441                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1442                 $imagesexception = !empty($imagesfrom);
1443                 $text = false;
1444                 if ( $this->mOptions->getAllowExternalImages()
1445                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1446                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1447                                 # Image found
1448                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1449                         }
1450                 }
1451                 return $text;
1452         }
1453
1454         /**
1455          * Process [[ ]] wikilinks
1456          *
1457          * @private
1458          */
1459         function replaceInternalLinks( $s ) {
1460                 global $wgContLang;
1461                 static $fname = 'Parser::replaceInternalLinks' ;
1462
1463                 wfProfileIn( $fname );
1464
1465                 wfProfileIn( $fname.'-setup' );
1466                 static $tc = FALSE;
1467                 # the % is needed to support urlencoded titles as well
1468                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1469
1470                 $sk = $this->mOptions->getSkin();
1471
1472                 #split the entire text string on occurences of [[
1473                 $a = explode( '[[', ' ' . $s );
1474                 #get the first element (all text up to first [[), and remove the space we added
1475                 $s = array_shift( $a );
1476                 $s = substr( $s, 1 );
1477
1478                 # Match a link having the form [[namespace:link|alternate]]trail
1479                 static $e1 = FALSE;
1480                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1481                 # Match cases where there is no "]]", which might still be images
1482                 static $e1_img = FALSE;
1483                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1484
1485                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1486                 $e2 = null;
1487                 if ( $useLinkPrefixExtension ) {
1488                         # Match the end of a line for a word that's not followed by whitespace,
1489                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1490                         $e2 = wfMsgForContent( 'linkprefix' );
1491                 }
1492
1493                 if( is_null( $this->mTitle ) ) {
1494                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1495                 }
1496                 $nottalk = !$this->mTitle->isTalkPage();
1497
1498                 if ( $useLinkPrefixExtension ) {
1499                         $m = array();
1500                         if ( preg_match( $e2, $s, $m ) ) {
1501                                 $first_prefix = $m[2];
1502                         } else {
1503                                 $first_prefix = false;
1504                         }
1505                 } else {
1506                         $prefix = '';
1507                 }
1508
1509                 if($wgContLang->hasVariants()) {
1510                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1511                 } else {
1512                         $selflink = array($this->mTitle->getPrefixedText());
1513                 }
1514                 $useSubpages = $this->areSubpagesAllowed();
1515                 wfProfileOut( $fname.'-setup' );
1516
1517                 # Loop for each link
1518                 for ($k = 0; isset( $a[$k] ); $k++) {
1519                         $line = $a[$k];
1520                         if ( $useLinkPrefixExtension ) {
1521                                 wfProfileIn( $fname.'-prefixhandling' );
1522                                 if ( preg_match( $e2, $s, $m ) ) {
1523                                         $prefix = $m[2];
1524                                         $s = $m[1];
1525                                 } else {
1526                                         $prefix='';
1527                                 }
1528                                 # first link
1529                                 if($first_prefix) {
1530                                         $prefix = $first_prefix;
1531                                         $first_prefix = false;
1532                                 }
1533                                 wfProfileOut( $fname.'-prefixhandling' );
1534                         }
1535
1536                         $might_be_img = false;
1537
1538                         wfProfileIn( "$fname-e1" );
1539                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1540                                 $text = $m[2];
1541                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1542                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1543                                 # the real problem is with the $e1 regex
1544                                 # See bug 1300.
1545                                 #
1546                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1547                                 # and no image is in sight. See bug 2095.
1548                                 #
1549                                 if( $text !== '' &&
1550                                         substr( $m[3], 0, 1 ) === ']' &&
1551                                         strpos($text, '[') !== false
1552                                 )
1553                                 {
1554                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1555                                         $m[3] = substr( $m[3], 1 );
1556                                 }
1557                                 # fix up urlencoded title texts
1558                                 if( strpos( $m[1], '%' ) !== false ) {
1559                                         # Should anchors '#' also be rejected?
1560                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1561                                 }
1562                                 $trail = $m[3];
1563                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1564                                 $might_be_img = true;
1565                                 $text = $m[2];
1566                                 if ( strpos( $m[1], '%' ) !== false ) {
1567                                         $m[1] = urldecode($m[1]);
1568                                 }
1569                                 $trail = "";
1570                         } else { # Invalid form; output directly
1571                                 $s .= $prefix . '[[' . $line ;
1572                                 wfProfileOut( "$fname-e1" );
1573                                 continue;
1574                         }
1575                         wfProfileOut( "$fname-e1" );
1576                         wfProfileIn( "$fname-misc" );
1577
1578                         # Don't allow internal links to pages containing
1579                         # PROTO: where PROTO is a valid URL protocol; these
1580                         # should be external links.
1581                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1582                                 $s .= $prefix . '[[' . $line ;
1583                                 continue;
1584                         }
1585
1586                         # Make subpage if necessary
1587                         if( $useSubpages ) {
1588                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1589                         } else {
1590                                 $link = $m[1];
1591                         }
1592
1593                         $noforce = (substr($m[1], 0, 1) != ':');
1594                         if (!$noforce) {
1595                                 # Strip off leading ':'
1596                                 $link = substr($link, 1);
1597                         }
1598
1599                         wfProfileOut( "$fname-misc" );
1600                         wfProfileIn( "$fname-title" );
1601                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1602                         if( !$nt ) {
1603                                 $s .= $prefix . '[[' . $line;
1604                                 wfProfileOut( "$fname-title" );
1605                                 continue;
1606                         }
1607
1608                         $ns = $nt->getNamespace();
1609                         $iw = $nt->getInterWiki();
1610                         wfProfileOut( "$fname-title" );
1611
1612                         if ($might_be_img) { # if this is actually an invalid link
1613                                 wfProfileIn( "$fname-might_be_img" );
1614                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1615                                         $found = false;
1616                                         while (isset ($a[$k+1]) ) {
1617                                                 #look at the next 'line' to see if we can close it there
1618                                                 $spliced = array_splice( $a, $k + 1, 1 );
1619                                                 $next_line = array_shift( $spliced );
1620                                                 $m = explode( ']]', $next_line, 3 );
1621                                                 if ( count( $m ) == 3 ) {
1622                                                         # the first ]] closes the inner link, the second the image
1623                                                         $found = true;
1624                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1625                                                         $trail = $m[2];
1626                                                         break;
1627                                                 } elseif ( count( $m ) == 2 ) {
1628                                                         #if there's exactly one ]] that's fine, we'll keep looking
1629                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1630                                                 } else {
1631                                                         #if $next_line is invalid too, we need look no further
1632                                                         $text .= '[[' . $next_line;
1633                                                         break;
1634                                                 }
1635                                         }
1636                                         if ( !$found ) {
1637                                                 # we couldn't find the end of this imageLink, so output it raw
1638                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1639                                                 $text = $this->replaceInternalLinks($text);
1640                                                 $s .= "{$prefix}[[$link|$text";
1641                                                 # note: no $trail, because without an end, there *is* no trail
1642                                                 wfProfileOut( "$fname-might_be_img" );
1643                                                 continue;
1644                                         }
1645                                 } else { #it's not an image, so output it raw
1646                                         $s .= "{$prefix}[[$link|$text";
1647                                         # note: no $trail, because without an end, there *is* no trail
1648                                         wfProfileOut( "$fname-might_be_img" );
1649                                         continue;
1650                                 }
1651                                 wfProfileOut( "$fname-might_be_img" );
1652                         }
1653
1654                         $wasblank = ( '' == $text );
1655                         if( $wasblank ) $text = $link;
1656
1657                         # Link not escaped by : , create the various objects
1658                         if( $noforce ) {
1659
1660                                 # Interwikis
1661                                 wfProfileIn( "$fname-interwiki" );
1662                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1663                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1664                                         $s = rtrim($s . $prefix);
1665                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1666                                         wfProfileOut( "$fname-interwiki" );
1667                                         continue;
1668                                 }
1669                                 wfProfileOut( "$fname-interwiki" );
1670
1671                                 if ( $ns == NS_IMAGE ) {
1672                                         wfProfileIn( "$fname-image" );
1673                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1674                                                 # recursively parse links inside the image caption
1675                                                 # actually, this will parse them in any other parameters, too,
1676                                                 # but it might be hard to fix that, and it doesn't matter ATM
1677                                                 $text = $this->replaceExternalLinks($text);
1678                                                 $text = $this->replaceInternalLinks($text);
1679
1680                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1681                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1682                                                 $this->mOutput->addImage( $nt->getDBkey() );
1683
1684                                                 wfProfileOut( "$fname-image" );
1685                                                 continue;
1686                                         } else {
1687                                                 # We still need to record the image's presence on the page
1688                                                 $this->mOutput->addImage( $nt->getDBkey() );
1689                                         }
1690                                         wfProfileOut( "$fname-image" );
1691
1692                                 }
1693
1694                                 if ( $ns == NS_CATEGORY ) {
1695                                         wfProfileIn( "$fname-category" );
1696                                         $s = rtrim($s . "\n"); # bug 87
1697
1698                                         if ( $wasblank ) {
1699                                                 $sortkey = $this->getDefaultSort();
1700                                         } else {
1701                                                 $sortkey = $text;
1702                                         }
1703                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1704                                         $sortkey = str_replace( "\n", '', $sortkey );
1705                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1706                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1707
1708                                         /**
1709                                          * Strip the whitespace Category links produce, see bug 87
1710                                          * @todo We might want to use trim($tmp, "\n") here.
1711                                          */
1712                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1713
1714                                         wfProfileOut( "$fname-category" );
1715                                         continue;
1716                                 }
1717                         }
1718
1719                         # Self-link checking
1720                         if( $nt->getFragment() === '' ) {
1721                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1722                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1723                                         continue;
1724                                 }
1725                         }
1726
1727                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1728                         if( $ns == NS_MEDIA ) {
1729                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1730                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1731                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1732                                 $this->mOutput->addImage( $nt->getDBkey() );
1733                                 continue;
1734                         } elseif( $ns == NS_SPECIAL ) {
1735                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1736                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1737                                 } else {
1738                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1739                                 }
1740                                 continue;
1741                         } elseif( $ns == NS_IMAGE ) {
1742                                 $img = wfFindFile( $nt );
1743                                 if( $img ) {
1744                                         // Force a blue link if the file exists; may be a remote
1745                                         // upload on the shared repository, and we want to see its
1746                                         // auto-generated page.
1747                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1748                                         $this->mOutput->addLink( $nt );
1749                                         continue;
1750                                 }
1751                         }
1752                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1753                 }
1754                 wfProfileOut( $fname );
1755                 return $s;
1756         }
1757
1758         /**
1759          * Make a link placeholder. The text returned can be later resolved to a real link with
1760          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1761          * parsing of interwiki links, and secondly to allow all existence checks and
1762          * article length checks (for stub links) to be bundled into a single query.
1763          *
1764          */
1765         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1766                 wfProfileIn( __METHOD__ );
1767                 if ( ! is_object($nt) ) {
1768                         # Fail gracefully
1769                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1770                 } else {
1771                         # Separate the link trail from the rest of the link
1772                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1773
1774                         if ( $nt->isExternal() ) {
1775                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1776                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1777                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1778                         } else {
1779                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1780                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1781                                 $this->mLinkHolders['queries'][] = $query;
1782                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1783                                 $this->mLinkHolders['titles'][] = $nt;
1784
1785                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1786                         }
1787                 }
1788                 wfProfileOut( __METHOD__ );
1789                 return $retVal;
1790         }
1791
1792         /**
1793          * Render a forced-blue link inline; protect against double expansion of
1794          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1795          * Since this little disaster has to split off the trail text to avoid
1796          * breaking URLs in the following text without breaking trails on the
1797          * wiki links, it's been made into a horrible function.
1798          *
1799          * @param Title $nt
1800          * @param string $text
1801          * @param string $query
1802          * @param string $trail
1803          * @param string $prefix
1804          * @return string HTML-wikitext mix oh yuck
1805          */
1806         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1807                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1808                 $sk = $this->mOptions->getSkin();
1809                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1810                 return $this->armorLinks( $link ) . $trail;
1811         }
1812
1813         /**
1814          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1815          * going to go through further parsing steps before inline URL expansion.
1816          *
1817          * In particular this is important when using action=render, which causes
1818          * full URLs to be included.
1819          *
1820          * Oh man I hate our multi-layer parser!
1821          *
1822          * @param string more-or-less HTML
1823          * @return string less-or-more HTML with NOPARSE bits
1824          */
1825         function armorLinks( $text ) {
1826                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1827                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1828         }
1829
1830         /**
1831          * Return true if subpage links should be expanded on this page.
1832          * @return bool
1833          */
1834         function areSubpagesAllowed() {
1835                 # Some namespaces don't allow subpages
1836                 global $wgNamespacesWithSubpages;
1837                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1838         }
1839
1840         /**
1841          * Handle link to subpage if necessary
1842          * @param string $target the source of the link
1843          * @param string &$text the link text, modified as necessary
1844          * @return string the full name of the link
1845          * @private
1846          */
1847         function maybeDoSubpageLink($target, &$text) {
1848                 # Valid link forms:
1849                 # Foobar -- normal
1850                 # :Foobar -- override special treatment of prefix (images, language links)
1851                 # /Foobar -- convert to CurrentPage/Foobar
1852                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1853                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1854                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1855
1856                 $fname = 'Parser::maybeDoSubpageLink';
1857                 wfProfileIn( $fname );
1858                 $ret = $target; # default return value is no change
1859
1860                 # Some namespaces don't allow subpages,
1861                 # so only perform processing if subpages are allowed
1862                 if( $this->areSubpagesAllowed() ) {
1863                         $hash = strpos( $target, '#' );
1864                         if( $hash !== false ) {
1865                                 $suffix = substr( $target, $hash );
1866                                 $target = substr( $target, 0, $hash );
1867                         } else {
1868                                 $suffix = '';
1869                         }
1870                         # bug 7425
1871                         $target = trim( $target );
1872                         # Look at the first character
1873                         if( $target != '' && $target{0} == '/' ) {
1874                                 # / at end means we don't want the slash to be shown
1875                                 $m = array();
1876                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1877                                 if( $trailingSlashes ) {
1878                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1879                                 } else {
1880                                         $noslash = substr( $target, 1 );
1881                                 }
1882
1883                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1884                                 if( '' === $text ) {
1885                                         $text = $target . $suffix;
1886                                 } # this might be changed for ugliness reasons
1887                         } else {
1888                                 # check for .. subpage backlinks
1889                                 $dotdotcount = 0;
1890                                 $nodotdot = $target;
1891                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1892                                         ++$dotdotcount;
1893                                         $nodotdot = substr( $nodotdot, 3 );
1894                                 }
1895                                 if($dotdotcount > 0) {
1896                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1897                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1898                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1899                                                 # / at the end means don't show full path
1900                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1901                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1902                                                         if( '' === $text ) {
1903                                                                 $text = $nodotdot . $suffix;
1904                                                         }
1905                                                 }
1906                                                 $nodotdot = trim( $nodotdot );
1907                                                 if( $nodotdot != '' ) {
1908                                                         $ret .= '/' . $nodotdot;
1909                                                 }
1910                                                 $ret .= $suffix;
1911                                         }
1912                                 }
1913                         }
1914                 }
1915
1916                 wfProfileOut( $fname );
1917                 return $ret;
1918         }
1919
1920         /**#@+
1921          * Used by doBlockLevels()
1922          * @private
1923          */
1924         /* private */ function closeParagraph() {
1925                 $result = '';
1926                 if ( '' != $this->mLastSection ) {
1927                         $result = '</' . $this->mLastSection  . ">\n";
1928                 }
1929                 $this->mInPre = false;
1930                 $this->mLastSection = '';
1931                 return $result;
1932         }
1933         # getCommon() returns the length of the longest common substring
1934         # of both arguments, starting at the beginning of both.
1935         #
1936         /* private */ function getCommon( $st1, $st2 ) {
1937                 $fl = strlen( $st1 );
1938                 $shorter = strlen( $st2 );
1939                 if ( $fl < $shorter ) { $shorter = $fl; }
1940
1941                 for ( $i = 0; $i < $shorter; ++$i ) {
1942                         if ( $st1{$i} != $st2{$i} ) { break; }
1943                 }
1944                 return $i;
1945         }
1946         # These next three functions open, continue, and close the list
1947         # element appropriate to the prefix character passed into them.
1948         #
1949         /* private */ function openList( $char ) {
1950                 $result = $this->closeParagraph();
1951
1952                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1953                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1954                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1955                 else if ( ';' == $char ) {
1956                         $result .= '<dl><dt>';
1957                         $this->mDTopen = true;
1958                 }
1959                 else { $result = '<!-- ERR 1 -->'; }
1960
1961                 return $result;
1962         }
1963
1964         /* private */ function nextItem( $char ) {
1965                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1966                 else if ( ':' == $char || ';' == $char ) {
1967                         $close = '</dd>';
1968                         if ( $this->mDTopen ) { $close = '</dt>'; }
1969                         if ( ';' == $char ) {
1970                                 $this->mDTopen = true;
1971                                 return $close . '<dt>';
1972                         } else {
1973                                 $this->mDTopen = false;
1974                                 return $close . '<dd>';
1975                         }
1976                 }
1977                 return '<!-- ERR 2 -->';
1978         }
1979
1980         /* private */ function closeList( $char ) {
1981                 if ( '*' == $char ) { $text = '</li></ul>'; }
1982                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1983                 else if ( ':' == $char ) {
1984                         if ( $this->mDTopen ) {
1985                                 $this->mDTopen = false;
1986                                 $text = '</dt></dl>';
1987                         } else {
1988                                 $text = '</dd></dl>';
1989                         }
1990                 }
1991                 else {  return '<!-- ERR 3 -->'; }
1992                 return $text."\n";
1993         }
1994         /**#@-*/
1995
1996         /**
1997          * Make lists from lines starting with ':', '*', '#', etc.
1998          *
1999          * @private
2000          * @return string the lists rendered as HTML
2001          */
2002         function doBlockLevels( $text, $linestart ) {
2003                 $fname = 'Parser::doBlockLevels';
2004                 wfProfileIn( $fname );
2005
2006                 # Parsing through the text line by line.  The main thing
2007                 # happening here is handling of block-level elements p, pre,
2008                 # and making lists from lines starting with * # : etc.
2009                 #
2010                 $textLines = explode( "\n", $text );
2011
2012                 $lastPrefix = $output = '';
2013                 $this->mDTopen = $inBlockElem = false;
2014                 $prefixLength = 0;
2015                 $paragraphStack = false;
2016
2017                 if ( !$linestart ) {
2018                         $output .= array_shift( $textLines );
2019                 }
2020                 foreach ( $textLines as $oLine ) {
2021                         $lastPrefixLength = strlen( $lastPrefix );
2022                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2023                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2024                         if ( !$this->mInPre ) {
2025                                 # Multiple prefixes may abut each other for nested lists.
2026                                 $prefixLength = strspn( $oLine, '*#:;' );
2027                                 $pref = substr( $oLine, 0, $prefixLength );
2028
2029                                 # eh?
2030                                 $pref2 = str_replace( ';', ':', $pref );
2031                                 $t = substr( $oLine, $prefixLength );
2032                                 $this->mInPre = !empty($preOpenMatch);
2033                         } else {
2034                                 # Don't interpret any other prefixes in preformatted text
2035                                 $prefixLength = 0;
2036                                 $pref = $pref2 = '';
2037                                 $t = $oLine;
2038                         }
2039
2040                         # List generation
2041                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2042                                 # Same as the last item, so no need to deal with nesting or opening stuff
2043                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2044                                 $paragraphStack = false;
2045
2046                                 if ( substr( $pref, -1 ) == ';') {
2047                                         # The one nasty exception: definition lists work like this:
2048                                         # ; title : definition text
2049                                         # So we check for : in the remainder text to split up the
2050                                         # title and definition, without b0rking links.
2051                                         $term = $t2 = '';
2052                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2053                                                 $t = $t2;
2054                                                 $output .= $term . $this->nextItem( ':' );
2055                                         }
2056                                 }
2057                         } elseif( $prefixLength || $lastPrefixLength ) {
2058                                 # Either open or close a level...
2059                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2060                                 $paragraphStack = false;
2061
2062                                 while( $commonPrefixLength < $lastPrefixLength ) {
2063                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2064                                         --$lastPrefixLength;
2065                                 }
2066                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2067                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2068                                 }
2069                                 while ( $prefixLength > $commonPrefixLength ) {
2070                                         $char = substr( $pref, $commonPrefixLength, 1 );
2071                                         $output .= $this->openList( $char );
2072
2073                                         if ( ';' == $char ) {
2074                                                 # FIXME: This is dupe of code above
2075                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2076                                                         $t = $t2;
2077                                                         $output .= $term . $this->nextItem( ':' );
2078                                                 }
2079                                         }
2080                                         ++$commonPrefixLength;
2081                                 }
2082                                 $lastPrefix = $pref2;
2083                         }
2084                         if( 0 == $prefixLength ) {
2085                                 wfProfileIn( "$fname-paragraph" );
2086                                 # No prefix (not in list)--go to paragraph mode
2087                                 // XXX: use a stack for nestable elements like span, table and div
2088                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2089                                 $closematch = preg_match(
2090                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2091                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2092                                 if ( $openmatch or $closematch ) {
2093                                         $paragraphStack = false;
2094                                         # TODO bug 5718: paragraph closed
2095                                         $output .= $this->closeParagraph();
2096                                         if ( $preOpenMatch and !$preCloseMatch ) {
2097                                                 $this->mInPre = true;
2098                                         }
2099                                         if ( $closematch ) {
2100                                                 $inBlockElem = false;
2101                                         } else {
2102                                                 $inBlockElem = true;
2103                                         }
2104                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2105                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2106                                                 // pre
2107                                                 if ($this->mLastSection != 'pre') {
2108                                                         $paragraphStack = false;
2109                                                         $output .= $this->closeParagraph().'<pre>';
2110                                                         $this->mLastSection = 'pre';
2111                                                 }
2112                                                 $t = substr( $t, 1 );
2113                                         } else {
2114                                                 // paragraph
2115                                                 if ( '' == trim($t) ) {
2116                                                         if ( $paragraphStack ) {
2117                                                                 $output .= $paragraphStack.'<br />';
2118                                                                 $paragraphStack = false;
2119                                                                 $this->mLastSection = 'p';
2120                                                         } else {
2121                                                                 if ($this->mLastSection != 'p' ) {
2122                                                                         $output .= $this->closeParagraph();
2123                                                                         $this->mLastSection = '';
2124                                                                         $paragraphStack = '<p>';
2125                                                                 } else {
2126                                                                         $paragraphStack = '</p><p>';
2127                                                                 }
2128                                                         }
2129                                                 } else {
2130                                                         if ( $paragraphStack ) {
2131                                                                 $output .= $paragraphStack;
2132                                                                 $paragraphStack = false;
2133                                                                 $this->mLastSection = 'p';
2134                                                         } else if ($this->mLastSection != 'p') {
2135                                                                 $output .= $this->closeParagraph().'<p>';
2136                                                                 $this->mLastSection = 'p';
2137                                                         }
2138                                                 }
2139                                         }
2140                                 }
2141                                 wfProfileOut( "$fname-paragraph" );
2142                         }
2143                         // somewhere above we forget to get out of pre block (bug 785)
2144                         if($preCloseMatch && $this->mInPre) {
2145                                 $this->mInPre = false;
2146                         }
2147                         if ($paragraphStack === false) {
2148                                 $output .= $t."\n";
2149                         }
2150                 }
2151                 while ( $prefixLength ) {
2152                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2153                         --$prefixLength;
2154                 }
2155                 if ( '' != $this->mLastSection ) {
2156                         $output .= '</' . $this->mLastSection . '>';
2157                         $this->mLastSection = '';
2158                 }
2159
2160                 wfProfileOut( $fname );
2161                 return $output;
2162         }
2163
2164         /**
2165          * Split up a string on ':', ignoring any occurences inside tags
2166          * to prevent illegal overlapping.
2167          * @param string $str the string to split
2168          * @param string &$before set to everything before the ':'
2169          * @param string &$after set to everything after the ':'
2170          * return string the position of the ':', or false if none found
2171          */
2172         function findColonNoLinks($str, &$before, &$after) {
2173                 $fname = 'Parser::findColonNoLinks';
2174                 wfProfileIn( $fname );
2175
2176                 $pos = strpos( $str, ':' );
2177                 if( $pos === false ) {
2178                         // Nothing to find!
2179                         wfProfileOut( $fname );
2180                         return false;
2181                 }
2182
2183                 $lt = strpos( $str, '<' );
2184                 if( $lt === false || $lt > $pos ) {
2185                         // Easy; no tag nesting to worry about
2186                         $before = substr( $str, 0, $pos );
2187                         $after = substr( $str, $pos+1 );
2188                         wfProfileOut( $fname );
2189                         return $pos;
2190                 }
2191
2192                 // Ugly state machine to walk through avoiding tags.
2193                 $state = self::COLON_STATE_TEXT;
2194                 $stack = 0;
2195                 $len = strlen( $str );
2196                 for( $i = 0; $i < $len; $i++ ) {
2197                         $c = $str{$i};
2198
2199                         switch( $state ) {
2200                         // (Using the number is a performance hack for common cases)
2201                         case 0: // self::COLON_STATE_TEXT:
2202                                 switch( $c ) {
2203                                 case "<":
2204                                         // Could be either a <start> tag or an </end> tag
2205                                         $state = self::COLON_STATE_TAGSTART;
2206                                         break;
2207                                 case ":":
2208                                         if( $stack == 0 ) {
2209                                                 // We found it!
2210                                                 $before = substr( $str, 0, $i );
2211                                                 $after = substr( $str, $i + 1 );
2212                                                 wfProfileOut( $fname );
2213                                                 return $i;
2214                                         }
2215                                         // Embedded in a tag; don't break it.
2216                                         break;
2217                                 default:
2218                                         // Skip ahead looking for something interesting
2219                                         $colon = strpos( $str, ':', $i );
2220                                         if( $colon === false ) {
2221                                                 // Nothing else interesting
2222                                                 wfProfileOut( $fname );
2223                                                 return false;
2224                                         }
2225                                         $lt = strpos( $str, '<', $i );
2226                                         if( $stack === 0 ) {
2227                                                 if( $lt === false || $colon < $lt ) {
2228                                                         // We found it!
2229                                                         $before = substr( $str, 0, $colon );
2230                                                         $after = substr( $str, $colon + 1 );
2231                                                         wfProfileOut( $fname );
2232                                                         return $i;
2233                                                 }
2234                                         }
2235                                         if( $lt === false ) {
2236                                                 // Nothing else interesting to find; abort!
2237                                                 // We're nested, but there's no close tags left. Abort!
2238                                                 break 2;
2239                                         }
2240                                         // Skip ahead to next tag start
2241                                         $i = $lt;
2242                                         $state = self::COLON_STATE_TAGSTART;
2243                                 }
2244                                 break;
2245                         case 1: // self::COLON_STATE_TAG:
2246                                 // In a <tag>
2247                                 switch( $c ) {
2248                                 case ">":
2249                                         $stack++;
2250                                         $state = self::COLON_STATE_TEXT;
2251                                         break;
2252                                 case "/":
2253                                         // Slash may be followed by >?
2254                                         $state = self::COLON_STATE_TAGSLASH;
2255                                         break;
2256                                 default:
2257                                         // ignore
2258                                 }
2259                                 break;
2260                         case 2: // self::COLON_STATE_TAGSTART:
2261                                 switch( $c ) {
2262                                 case "/":
2263                                         $state = self::COLON_STATE_CLOSETAG;
2264                                         break;
2265                                 case "!":
2266                                         $state = self::COLON_STATE_COMMENT;
2267                                         break;
2268                                 case ">":
2269                                         // Illegal early close? This shouldn't happen D:
2270                                         $state = self::COLON_STATE_TEXT;
2271                                         break;
2272                                 default:
2273                                         $state = self::COLON_STATE_TAG;
2274                                 }
2275                                 break;
2276                         case 3: // self::COLON_STATE_CLOSETAG:
2277                                 // In a </tag>
2278                                 if( $c == ">" ) {
2279                                         $stack--;
2280                                         if( $stack < 0 ) {
2281                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2282                                                 wfProfileOut( $fname );
2283                                                 return false;
2284                                         }
2285                                         $state = self::COLON_STATE_TEXT;
2286                                 }
2287                                 break;
2288                         case self::COLON_STATE_TAGSLASH:
2289                                 if( $c == ">" ) {
2290                                         // Yes, a self-closed tag <blah/>
2291                                         $state = self::COLON_STATE_TEXT;
2292                                 } else {
2293                                         // Probably we're jumping the gun, and this is an attribute
2294                                         $state = self::COLON_STATE_TAG;
2295                                 }
2296                                 break;
2297                         case 5: // self::COLON_STATE_COMMENT:
2298                                 if( $c == "-" ) {
2299                                         $state = self::COLON_STATE_COMMENTDASH;
2300                                 }
2301                                 break;
2302                         case self::COLON_STATE_COMMENTDASH:
2303                                 if( $c == "-" ) {
2304                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2305                                 } else {
2306                                         $state = self::COLON_STATE_COMMENT;
2307                                 }
2308                                 break;
2309                         case self::COLON_STATE_COMMENTDASHDASH:
2310                                 if( $c == ">" ) {
2311                                         $state = self::COLON_STATE_TEXT;
2312                                 } else {
2313                                         $state = self::COLON_STATE_COMMENT;
2314                                 }
2315                                 break;
2316                         default:
2317                                 throw new MWException( "State machine error in $fname" );
2318                         }
2319                 }
2320                 if( $stack > 0 ) {
2321                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2322                         return false;
2323                 }
2324                 wfProfileOut( $fname );
2325                 return false;
2326         }
2327
2328         /**
2329          * Return value of a magic variable (like PAGENAME)
2330          *
2331          * @private
2332          */
2333         function getVariableValue( $index ) {
2334                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2335
2336                 /**
2337                  * Some of these require message or data lookups and can be
2338                  * expensive to check many times.
2339                  */
2340                 static $varCache = array();
2341                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2342                         if ( isset( $varCache[$index] ) ) {
2343                                 return $varCache[$index];
2344                         }
2345                 }
2346
2347                 $ts = time();
2348                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2349
2350                 # Use the time zone
2351                 global $wgLocaltimezone;
2352                 if ( isset( $wgLocaltimezone ) ) {
2353                         $oldtz = getenv( 'TZ' );
2354                         putenv( 'TZ='.$wgLocaltimezone );
2355                 }
2356
2357                 wfSuppressWarnings(); // E_STRICT system time bitching
2358                 $localTimestamp = date( 'YmdHis', $ts );
2359                 $localMonth = date( 'm', $ts );
2360                 $localMonthName = date( 'n', $ts );
2361                 $localDay = date( 'j', $ts );
2362                 $localDay2 = date( 'd', $ts );
2363                 $localDayOfWeek = date( 'w', $ts );
2364                 $localWeek = date( 'W', $ts );
2365                 $localYear = date( 'Y', $ts );
2366                 $localHour = date( 'H', $ts );
2367                 if ( isset( $wgLocaltimezone ) ) {
2368                         putenv( 'TZ='.$oldtz );
2369                 }
2370                 wfRestoreWarnings();
2371
2372                 switch ( $index ) {
2373                         case 'currentmonth':
2374                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2375                         case 'currentmonthname':
2376                                 return $varCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2377                         case 'currentmonthnamegen':
2378                                 return $varCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2379                         case 'currentmonthabbrev':
2380                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2381                         case 'currentday':
2382                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2383                         case 'currentday2':
2384                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2385                         case 'localmonth':
2386                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2387                         case 'localmonthname':
2388                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2389                         case 'localmonthnamegen':
2390                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2391                         case 'localmonthabbrev':
2392                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2393                         case 'localday':
2394                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2395                         case 'localday2':
2396                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2397                         case 'pagename':
2398                                 return wfEscapeWikiText( $this->mTitle->getText() );
2399                         case 'pagenamee':
2400                                 return $this->mTitle->getPartialURL();
2401                         case 'fullpagename':
2402                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2403                         case 'fullpagenamee':
2404                                 return $this->mTitle->getPrefixedURL();
2405                         case 'subpagename':
2406                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2407                         case 'subpagenamee':
2408                                 return $this->mTitle->getSubpageUrlForm();
2409                         case 'basepagename':
2410                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2411                         case 'basepagenamee':
2412                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2413                         case 'talkpagename':
2414                                 if( $this->mTitle->canTalk() ) {
2415                                         $talkPage = $this->mTitle->getTalkPage();
2416                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2417                                 } else {
2418                                         return '';
2419                                 }
2420                         case 'talkpagenamee':
2421                                 if( $this->mTitle->canTalk() ) {
2422                                         $talkPage = $this->mTitle->getTalkPage();
2423                                         return $talkPage->getPrefixedUrl();
2424                                 } else {
2425                                         return '';
2426                                 }
2427                         case 'subjectpagename':
2428                                 $subjPage = $this->mTitle->getSubjectPage();
2429                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2430                         case 'subjectpagenamee':
2431                                 $subjPage = $this->mTitle->getSubjectPage();
2432                                 return $subjPage->getPrefixedUrl();
2433                         case 'revisionid':
2434                                 // Let the edit saving system know we should parse the page
2435                                 // *after* a revision ID has been assigned.
2436                                 $this->mOutput->setFlag( 'vary-revision' );
2437                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2438                                 return $this->mRevisionId;
2439                         case 'revisionday':
2440                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2441                         case 'revisionday2':
2442                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2443                         case 'revisionmonth':
2444                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2445                         case 'revisionyear':
2446                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2447                         case 'revisiontimestamp':
2448                                 return $this->getRevisionTimestamp();
2449                         case 'namespace':
2450                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2451                         case 'namespacee':
2452                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2453                         case 'talkspace':
2454                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2455                         case 'talkspacee':
2456                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2457                         case 'subjectspace':
2458                                 return $this->mTitle->getSubjectNsText();
2459                         case 'subjectspacee':
2460                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2461                         case 'currentdayname':
2462                                 return $varCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2463                         case 'currentyear':
2464                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2465                         case 'currenttime':
2466                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2467                         case 'currenthour':
2468                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2469                         case 'currentweek':
2470                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2471                                 // int to remove the padding
2472                                 return $varCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2473                         case 'currentdow':
2474                                 return $varCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2475                         case 'localdayname':
2476                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2477                         case 'localyear':
2478                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2479                         case 'localtime':
2480                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2481                         case 'localhour':
2482                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2483                         case 'localweek':
2484                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2485                                 // int to remove the padding
2486                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2487                         case 'localdow':
2488                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2489                         case 'numberofarticles':
2490                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2491                         case 'numberoffiles':
2492                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2493                         case 'numberofusers':
2494                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2495                         case 'numberofpages':
2496                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2497                         case 'numberofadmins':
2498                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2499                         case 'numberofedits':
2500                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2501                         case 'currenttimestamp':
2502                                 return $varCache[$index] = wfTimestampNow();
2503                         case 'localtimestamp':
2504                                 return $varCache[$index] = $localTimestamp;
2505                         case 'currentversion':
2506                                 return $varCache[$index] = SpecialVersion::getVersion();
2507                         case 'sitename':
2508                                 return $wgSitename;
2509                         case 'server':
2510                                 return $wgServer;
2511                         case 'servername':
2512                                 return $wgServerName;
2513                         case 'scriptpath':
2514                                 return $wgScriptPath;
2515                         case 'directionmark':
2516                                 return $wgContLang->getDirMark();
2517                         case 'contentlanguage':
2518                                 global $wgContLanguageCode;
2519                                 return $wgContLanguageCode;
2520                         default:
2521                                 $ret = null;
2522                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2523                                         return $ret;
2524                                 else
2525                                         return null;
2526                 }
2527         }
2528
2529         /**
2530          * initialise the magic variables (like CURRENTMONTHNAME)
2531          *
2532          * @private
2533          */
2534         function initialiseVariables() {
2535                 $fname = 'Parser::initialiseVariables';
2536                 wfProfileIn( $fname );
2537                 $variableIDs = MagicWord::getVariableIDs();
2538
2539                 $this->mVariables = new MagicWordArray( $variableIDs );
2540                 wfProfileOut( $fname );
2541         }
2542
2543         /**
2544          * Parse any parentheses in format ((title|part|part)} and return the document tree
2545          * This is the ghost of replace_variables().
2546          *
2547          * @param string $text The text to parse
2548          * @private
2549          */
2550         function preprocessToDom ( $text ) {
2551                 wfProfileIn( __METHOD__ );
2552                 wfProfileIn( __METHOD__.'-makexml' );
2553
2554                 static $msgRules, $normalRules;
2555                 if ( !$msgRules ) {
2556                         $msgRules = array(
2557                                 '{' => array(
2558                                         'end' => '}',
2559                                         'names' => array(
2560                                                 2 => 'template',
2561                                         ),
2562                                         'min' => 2,
2563                                         'max' => 2,
2564                                 ),
2565                                 '[' => array(
2566                                         'end' => ']',
2567                                         'names' => array( 2 => null ),
2568                                         'min' => 2,
2569                                         'max' => 2,
2570                                 )
2571                         );
2572                         $normalRules = array(
2573                                 '{' => array(
2574                                         'end' => '}',
2575                                         'names' => array(
2576                                                 2 => 'template',
2577                                                 3 => 'tplarg',
2578                                         ),
2579                                         'min' => 2,
2580                                         'max' => 3,
2581                                 ),
2582                                 '[' => array(
2583                                         'end' => ']',
2584                                         'names' => array( 2 => null ),
2585                                         'min' => 2,
2586                                         'max' => 2,
2587                                 )
2588                         );
2589                 }
2590                 if ( $this->ot['msg'] ) {
2591                         $rules = $msgRules;
2592                 } else {
2593                         $rules = $normalRules;
2594                 }
2595
2596                 $extElements = implode( '|', $this->getStripList() );
2597                 // Use "A" modifier (anchored) instead of "^", because ^ doesn't work with an offset
2598                 $extElementsRegex = "/($extElements)(?:\s|\/>|>)|(!--)/iA";
2599
2600                 $stack = array();      # Stack of unclosed parentheses
2601                 $stackIndex = -1;      # Stack read pointer
2602
2603                 $searchBase = implode( '', array_keys( $rules ) ) . '<';
2604
2605                 $i = -1; # Input pointer, starts out pointing to a pseudo-newline before the start
2606                 $topAccum = '<root>';      # Top level text accumulator
2607                 $accum =& $topAccum;            # Current text accumulator
2608                 $findEquals = false;            # True to find equals signs in arguments
2609                 $findHeading = false;           # True to look at LF characters for possible headings
2610                 $findPipe = false;              # True to take notice of pipe characters
2611                 $headingIndex = 1;
2612                 $noMoreGT = false;         # True if there are no more greater-than (>) signs right of $i
2613
2614                 while ( $i < strlen( $text ) ) {
2615                         if ( $i == -1 ) {
2616                                 $found = 'line-start';
2617                                 $curChar = '';
2618                         } else {
2619                                 # Find next opening brace, closing brace or pipe
2620                                 $search = $searchBase;
2621                                 if ( $stackIndex == -1 ) {
2622                                         $currentClosing = '';
2623                                         // Look for headings only at the top stack level
2624                                         // Among other things, this resolves the ambiguity between =
2625                                         // for headings and = for template arguments
2626                                         $search .= "\n";
2627                                 } else {
2628                                         $currentClosing = $stack[$stackIndex]['close'];
2629                                         $search .= $currentClosing;
2630                                 }
2631                                 if ( $findPipe ) {
2632                                         $search .= '|';
2633                                 }
2634                                 if ( $findEquals ) {
2635                                         $search .= '=';
2636                                 }
2637                                 $rule = null;
2638                                 # Output literal section, advance input counter
2639                                 $literalLength = strcspn( $text, $search, $i );
2640                                 if ( $literalLength > 0 ) {
2641                                         $accum .= htmlspecialchars( substr( $text, $i, $literalLength ) );
2642                                         $i += $literalLength;
2643                                 }
2644                                 if ( $i >= strlen( $text ) ) {
2645                                         if ( $currentClosing == "\n" ) {
2646                                                 // Do a past-the-end run to finish off the heading
2647                                                 $curChar = '';
2648                                                 $found = 'line-end';
2649                                         } else {
2650                                                 # All done
2651                                                 break;
2652                                         }
2653                                 } else {
2654                                         $curChar = $text[$i];
2655                                         if ( $curChar == '|' ) {
2656                                                 $found = 'pipe';
2657                                         } elseif ( $curChar == '=' ) {
2658                                                 $found = 'equals';
2659                                         } elseif ( $curChar == '<' ) {
2660                                                 $found = 'angle';
2661                                         } elseif ( $curChar == "\n" ) {
2662                                                 if ( $stackIndex == -1 ) {
2663                                                         $found = 'line-start';
2664                                                 } else {
2665                                                         $found = 'line-end';
2666                                                 }
2667                                         } elseif ( $curChar == $currentClosing ) {
2668                                                 $found = 'close';
2669                                         } elseif ( isset( $rules[$curChar] ) ) {
2670                                                 $found = 'open';
2671                                                 $rule = $rules[$curChar];
2672                                         } else {
2673                                                 # Some versions of PHP have a strcspn which stops on null characters
2674                                                 # Ignore and continue
2675                                                 ++$i;
2676                                                 continue;
2677                                         }
2678                                 }
2679                         }
2680
2681                         if ( $found == 'angle' ) {
2682                                 $matches = false;
2683                                 // Determine element name
2684                                 if ( !preg_match( $extElementsRegex, $text, $matches, 0, $i + 1 ) ) {
2685                                         // Element name missing or not listed
2686                                         $accum .= '&lt;';
2687                                         ++$i;
2688                                         continue;
2689                                 }
2690                                 // Handle comments
2691                                 if ( isset( $matches[2] ) && $matches[2] == '!--' ) {
2692                                         // HTML comment, scan to end
2693                                         $endpos = strpos( $text, '-->', $i + 4 );
2694                                         if ( $endpos === false ) {
2695                                                 // Unclosed comment in input, runs to end
2696                                                 $inner = substr( $text, $i );
2697                                                 if ( $this->ot['html'] ) {
2698                                                         // Close it so later stripping can remove it
2699                                                         $inner .= '-->';
2700                                                 }
2701                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2702                                                 $i = strlen( $text );
2703                                         } else {
2704                                                 $inner = substr( $text, $i, $endpos - $i + 3 );
2705                                                 $accum .= '<comment>' . htmlspecialchars( $inner ) . '</comment>';
2706                                                 $i = $endpos + 3;
2707                                         }
2708                                         continue;
2709                                 }
2710                                 $name = $matches[1];
2711                                 $attrStart = $i + strlen( $name ) + 1;
2712
2713                                 // Find end of tag
2714                                 $tagEndPos = $noMoreGT ? false : strpos( $text, '>', $attrStart );
2715                                 if ( $tagEndPos === false ) {
2716                                         // Infinite backtrack
2717                                         // Disable tag search to prevent worst-case O(N^2) performance
2718                                         $noMoreGT = true;
2719                                         $accum .= '&lt;';
2720                                         ++$i;
2721                                         continue;
2722                                 }
2723                                 if ( $text[$tagEndPos-1] == '/' ) {
2724                                         $attrEnd = $tagEndPos - 1;
2725                                         $inner = null;
2726                                         $i = $tagEndPos + 1;
2727                                         $close = '';
2728                                 } else {
2729                                         $attrEnd = $tagEndPos;
2730                                         // Find closing tag
2731                                         if ( preg_match( "/<\/$name\s*>/i", $text, $matches, PREG_OFFSET_CAPTURE, $tagEndPos + 1 ) ) {
2732                                                 $inner = substr( $text, $tagEndPos + 1, $matches[0][1] - $tagEndPos - 1 );
2733                                                 $i = $matches[0][1] + strlen( $matches[0][0] );
2734                                                 $close = '<close>' . htmlspecialchars( $matches[0][0] ) . '</close>';
2735                                         } else {
2736                                                 // No end tag -- let it run out to the end of the text.
2737                                                 $inner = substr( $text, $tagEndPos + 1 );
2738                                                 $i = strlen( $text );
2739                                                 $close = '';
2740                                         }
2741                                 }
2742                                 $accum .= '<ext>';
2743                                 if ( $attrEnd <= $attrStart ) {
2744                                         $attr = '';
2745                                 } else {
2746                                         $attr = substr( $text, $attrStart, $attrEnd - $attrStart );
2747                                 }
2748                                 $accum .= '<name>' . htmlspecialchars( $name ) . '</name>' .
2749                                         // Note that the attr element contains the whitespace between name and attribute,
2750                                         // this is necessary for precise reconstruction during pre-save transform.
2751                                         '<attr>' . htmlspecialchars( $attr ) . '</attr>';
2752                                 if ( $inner !== null ) {
2753                                         $accum .= '<inner>' . htmlspecialchars( $inner ) . '</inner>';
2754                                 }
2755                                 $accum .= $close . '</ext>';
2756                         }
2757
2758                         elseif ( $found == 'line-start' ) {
2759                                 // Is this the start of a heading?
2760                                 // Line break belongs before the heading element in any case
2761                                 $accum .= $curChar;
2762                                 $i++;
2763
2764                                 $count = strspn( $text, '=', $i, 6 );
2765                                 if ( $count > 0 ) {
2766                                         $piece = array(
2767                                                 'open' => "\n",
2768                                                 'close' => "\n",
2769                                                 'parts' => array( str_repeat( '=', $count ) ),
2770                                                 'count' => $count );
2771                                         $stack[++$stackIndex] = $piece;
2772                                         $i += $count;
2773                                         $accum =& $stack[$stackIndex]['parts'][0];
2774                                         $findPipe = false;
2775                                 }
2776                         }
2777
2778                         elseif ( $found == 'line-end' ) {
2779                                 $piece = $stack[$stackIndex];
2780                                 // A heading must be open, otherwise \n wouldn't have been in the search list
2781                                 assert( $piece['open'] == "\n" );
2782                                 assert( $stackIndex == 0 );
2783                                 // Search back through the accumulator to see if it has a proper close
2784                                 // No efficient way to do this in PHP AFAICT: strrev, PCRE search with $ anchor
2785                                 // and rtrim are all O(N) in total size. Optimal would be O(N) in trailing
2786                                 // whitespace size only.
2787                                 $m = false;
2788                                 $count = $piece['count'];
2789                                 if ( preg_match( "/(={{$count}})\s*$/", $accum, $m, 0, $count ) ) {
2790                                         // Found match, output <h>
2791                                         $count = min( strlen( $m[1] ), $count );
2792                                         $element = "<h level=\"$count\" i=\"$headingIndex\">$accum</h>";
2793                                         $headingIndex++;
2794                                 } else {
2795                                         // No match, no <h>, just pass down the inner text
2796                                         $element = $accum;
2797                                 }
2798                                 // Unwind the stack
2799                                 // Headings can only occur on the top level, so this is a bit simpler than the
2800                                 // generic stack unwind operation in the close case
2801                                 unset( $stack[$stackIndex--] );
2802                                 $accum =& $topAccum;
2803                                 $findEquals = false;
2804                                 $findPipe = false;
2805
2806                                 // Append the result to the enclosing accumulator
2807                                 $accum .= $element;
2808                                 // Note that we do NOT increment the input pointer.
2809                                 // This is because the closing linebreak could be the opening linebreak of
2810                                 // another heading. Infinite loops are avoided because the next iteration MUST
2811                                 // hit the heading open case above, which unconditionally increments the
2812                                 // input pointer.
2813                         }
2814
2815                         elseif ( $found == 'open' ) {
2816                                 # count opening brace characters
2817                                 $count = strspn( $text, $curChar, $i );
2818
2819                                 # we need to add to stack only if opening brace count is enough for one of the rules
2820                                 if ( $count >= $rule['min'] ) {
2821                                         # Add it to the stack
2822                                         $piece = array(
2823                                                 'open' => $curChar,
2824                                                 'close' => $rule['end'],
2825                                                 'count' => $count,
2826                                                 'parts' => array( '' ),
2827                                                 'eqpos' => array(),
2828                                                 'lineStart' => ($i > 0 && $text[$i-1] == "\n"),
2829                                         );
2830
2831                                         $stackIndex ++;
2832                                         $stack[$stackIndex] = $piece;
2833                                         $accum =& $stack[$stackIndex]['parts'][0];
2834                                         $findEquals = false;
2835                                         $findPipe = true;
2836                                 } else {
2837                                         # Add literal brace(s)
2838                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2839                                 }
2840                                 $i += $count;
2841                         }
2842
2843                         elseif ( $found == 'close' ) {
2844                                 $piece = $stack[$stackIndex];
2845                                 # lets check if there are enough characters for closing brace
2846                                 $maxCount = $piece['count'];
2847                                 $count = strspn( $text, $curChar, $i, $maxCount );
2848
2849                                 # check for maximum matching characters (if there are 5 closing
2850                                 # characters, we will probably need only 3 - depending on the rules)
2851                                 $matchingCount = 0;
2852                                 $rule = $rules[$piece['open']];
2853                                 if ( $count > $rule['max'] ) {
2854                                         # The specified maximum exists in the callback array, unless the caller
2855                                         # has made an error
2856                                         $matchingCount = $rule['max'];
2857                                 } else {
2858                                         # Count is less than the maximum
2859                                         # Skip any gaps in the callback array to find the true largest match
2860                                         # Need to use array_key_exists not isset because the callback can be null
2861                                         $matchingCount = $count;
2862                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $rule['names'] ) ) {
2863                                                 --$matchingCount;
2864                                         }
2865                                 }
2866
2867                                 if ($matchingCount <= 0) {
2868                                         # No matching element found in callback array
2869                                         # Output a literal closing brace and continue
2870                                         $accum .= htmlspecialchars( str_repeat( $curChar, $count ) );
2871                                         $i += $count;
2872                                         continue;
2873                                 }
2874                                 $name = $rule['names'][$matchingCount];
2875                                 if ( $name === null ) {
2876                                         // No element, just literal text
2877                                         $element = str_repeat( $piece['open'], $matchingCount ) .
2878                                                 implode( '|', $piece['parts'] ) .
2879                                                 str_repeat( $rule['end'], $matchingCount );
2880                                 } else {
2881                                         # Create XML element
2882                                         # Note: $parts is already XML, does not need to be encoded further
2883                                         $parts = $piece['parts'];
2884                                         $title = $parts[0];
2885                                         unset( $parts[0] );
2886
2887                                         # The invocation is at the start of the line if lineStart is set in
2888                                         # the stack, and all opening brackets are used up.
2889                                         if ( $maxCount == $matchingCount && !empty( $piece['lineStart'] ) ) {
2890                                                 $attr = ' lineStart="1"';
2891                                         } else {
2892                                                 $attr = '';
2893                                         }
2894
2895                                         $element = "<$name$attr>";
2896                                         $element .= "<title>$title</title>";
2897                                         $argIndex = 1;
2898                                         foreach ( $parts as $partIndex => $part ) {
2899                                                 if ( isset( $piece['eqpos'][$partIndex] ) ) {
2900                                                         $eqpos = $piece['eqpos'][$partIndex];
2901                                                         list( $ws1, $argName, $ws2 ) = self::splitWhitespace( substr( $part, 0, $eqpos ) );
2902                                                         list( $ws3, $argValue, $ws4 ) = self::splitWhitespace( substr( $part, $eqpos + 1 ) );
2903                                                         $element .= "<part>$ws1<name>$argName</name>$ws2=$ws3<value>$argValue</value>$ws4</part>";
2904                                                 } else {
2905                                                         list( $ws1, $value, $ws2 ) = self::splitWhitespace( $part );
2906                                                         $element .= "<part>$ws1<name index=\"$argIndex\" /><value>$value</value>$ws2</part>";
2907                                                         $argIndex++;
2908                                                 }
2909                                         }
2910                                         $element .= "</$name>";
2911                                 }
2912
2913                                 # Advance input pointer
2914                                 $i += $matchingCount;
2915
2916                                 # Unwind the stack
2917                                 unset( $stack[$stackIndex--] );
2918                                 if ( $stackIndex == -1 ) {
2919                                         $accum =& $topAccum;
2920                                         $findEquals = false;
2921                                         $findPipe = false;
2922                                 } else {
2923                                         $partCount = count( $stack[$stackIndex]['parts'] );
2924                                         $accum =& $stack[$stackIndex]['parts'][$partCount - 1];
2925                                         $findPipe = $stack[$stackIndex]['open'] != "\n";
2926                                         $findEquals = $findPipe && $partCount > 1
2927                                                 && !isset( $stack[$stackIndex]['eqpos'][$partCount - 1] );
2928                                 }
2929
2930                                 # Re-add the old stack element if it still has unmatched opening characters remaining
2931                                 if ($matchingCount < $piece['count']) {
2932                                         $piece['parts'] = array( '' );
2933                                         $piece['count'] -= $matchingCount;
2934                                         $piece['eqpos'] = array();
2935                                         # do we still qualify for any callback with remaining count?
2936                                         $names = $rules[$piece['open']]['names'];
2937                                         $skippedBraces = 0;
2938                                         $enclosingAccum =& $accum;
2939                                         while ( $piece['count'] ) {
2940                                                 if ( array_key_exists( $piece['count'], $names ) ) {
2941                                                         $stackIndex++;
2942                                                         $stack[$stackIndex] = $piece;
2943                                                         $accum =& $stack[$stackIndex]['parts'][0];
2944                                                         $findEquals = true;
2945                                                         $findPipe = true;
2946                                                         break;
2947                                                 }
2948                                                 --$piece['count'];
2949                                                 $skippedBraces ++;
2950                                         }
2951                                         $enclosingAccum .= str_repeat( $piece['open'], $skippedBraces );
2952                                 }
2953
2954                                 # Add XML element to the enclosing accumulator
2955                                 $accum .= $element;
2956                         }
2957
2958                         elseif ( $found == 'pipe' ) {
2959                                 $stack[$stackIndex]['parts'][] = '';
2960                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2961                                 $accum =& $stack[$stackIndex]['parts'][$partsCount - 1];
2962                                 $findEquals = true;
2963                                 ++$i;
2964                         }
2965
2966                         elseif ( $found == 'equals' ) {
2967                                 $findEquals = false;
2968                                 $partsCount = count( $stack[$stackIndex]['parts'] );
2969                                 $stack[$stackIndex]['eqpos'][$partsCount - 1] = strlen( $accum );
2970                                 $accum .= '=';
2971                                 ++$i;
2972                         }
2973                 }
2974
2975                 # Output any remaining unclosed brackets
2976                 foreach ( $stack as $piece ) {
2977                         if ( $piece['open'] == "\n" ) {
2978                                 $topAccum .= $piece['parts'][0];
2979                         } else {
2980                                 $topAccum .= str_repeat( $piece['open'], $piece['count'] ) . implode( '|', $piece['parts'] );
2981                         }
2982                 }
2983                 $topAccum .= '</root>';
2984
2985                 wfProfileOut( __METHOD__.'-makexml' );
2986                 wfProfileIn( __METHOD__.'-loadXML' );
2987                 $dom = new DOMDocument;
2988                 wfSuppressWarnings();
2989                 $result = $dom->loadXML( $topAccum );
2990                 wfRestoreWarnings();
2991                 if ( !$result ) {
2992                         // Try running the XML through UtfNormal to get rid of invalid characters
2993                         $topAccum = UtfNormal::cleanUp( $topAccum );
2994                         $result = $dom->loadXML( $topAccum );
2995                         if ( !$result ) {
2996                                 throw new MWException( __METHOD__.' generated invalid XML' );
2997                         }
2998                 }
2999                 wfProfileOut( __METHOD__.'-loadXML' );
3000                 wfProfileOut( __METHOD__ );
3001                 return $dom;
3002         }
3003
3004         /*
3005          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3006          */
3007         public static function splitWhitespace( $s ) {
3008                 $ltrimmed = ltrim( $s );
3009                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3010                 $trimmed = rtrim( $ltrimmed );
3011                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3012                 if ( $diff > 0 ) {
3013                         $w2 = substr( $ltrimmed, -$diff );
3014                 } else {
3015                         $w2 = '';
3016                 }
3017                 return array( $w1, $trimmed, $w2 );
3018         }
3019
3020         /**
3021          * Replace magic variables, templates, and template arguments
3022          * with the appropriate text. Templates are substituted recursively,
3023          * taking care to avoid infinite loops.
3024          *
3025          * Note that the substitution depends on value of $mOutputType:
3026          *  OT_WIKI: only {{subst:}} templates
3027          *  OT_MSG: only magic variables
3028          *  OT_HTML: all templates and magic variables
3029          *
3030          * @param string $tex The text to transform
3031          * @param PPFrame $frame Object describing the arguments passed to the template
3032          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
3033          * @private
3034          */
3035         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3036                 # Prevent too big inclusions
3037                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3038                         return $text;
3039                 }
3040
3041                 $fname = __METHOD__;
3042                 wfProfileIn( $fname );
3043
3044                 if ( $frame === false ) {
3045                         $frame = new PPFrame( $this );
3046                 } elseif ( !( $frame instanceof PPFrame ) ) {
3047                         throw new MWException( __METHOD__ . ' called using the old argument format' );
3048                 }
3049
3050                 # Remove comments
3051                 # This could theoretically be merged into preprocessToDom()
3052                 if ( $this->ot['html'] || ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) ) {
3053                         $text = Sanitizer::removeHTMLcomments( $text );
3054                 }
3055
3056                 $dom = $this->preprocessToDom( $text );
3057                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3058                 $text = $frame->expand( $dom, $flags );
3059
3060                 wfProfileOut( $fname );
3061                 return $text;
3062         }
3063
3064         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3065         static function createAssocArgs( $args ) {
3066                 $assocArgs = array();
3067                 $index = 1;
3068                 foreach( $args as $arg ) {
3069                         $eqpos = strpos( $arg, '=' );
3070                         if ( $eqpos === false ) {
3071                                 $assocArgs[$index++] = $arg;
3072                         } else {
3073                                 $name = trim( substr( $arg, 0, $eqpos ) );
3074                                 $value = trim( substr( $arg, $eqpos+1 ) );
3075                                 if ( $value === false ) {
3076                                         $value = '';
3077                                 }
3078                                 if ( $name !== false ) {
3079                                         $assocArgs[$name] = $value;
3080                                 }
3081                         }
3082                 }
3083
3084                 return $assocArgs;
3085         }
3086
3087         /**
3088          * Return the text of a template, after recursively
3089          * replacing any variables or templates within the template.
3090          *
3091          * @param array $piece The parts of the template
3092          *  $piece['text']: matched text
3093          *  $piece['title']: the title, i.e. the part before the |
3094          *  $piece['parts']: the parameter array
3095          * @param PPFrame The current frame, contains template arguments
3096          * @return string the text of the template
3097          * @private
3098          */
3099         function braceSubstitution( $piece, $frame ) {
3100                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
3101                 $fname = __METHOD__;
3102                 wfProfileIn( $fname );
3103                 wfProfileIn( __METHOD__.'-setup' );
3104
3105                 # Flags
3106                 $found = false;             # $text has been filled
3107                 $nowiki = false;            # wiki markup in $text should be escaped
3108                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
3109                 $noargs = false;            # Don't replace triple-brace arguments in $text
3110                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3111                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3112                 $isDOM = false;             # $text is a DOM node needing expansion
3113
3114                 # Title object, where $text came from
3115                 $title = NULL;
3116
3117                 # $part1 is the bit before the first |, and must contain only title characters.
3118                 # Various prefixes will be stripped from it later.
3119                 $titleWithSpaces = $frame->expand( $piece['title'] );
3120                 $part1 = trim( $titleWithSpaces );
3121                 $titleText = false;
3122
3123                 # Original title text preserved for various purposes
3124                 $originalTitle = $part1;
3125
3126                 # $args is a list of argument nodes, starting from index 0, not including $part1
3127                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
3128                 wfProfileOut( __METHOD__.'-setup' );
3129
3130                 # SUBST
3131                 wfProfileIn( __METHOD__.'-modifiers' );
3132                 if ( !$found ) {
3133                         $mwSubst =& MagicWord::get( 'subst' );
3134                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
3135                                 # One of two possibilities is true:
3136                                 # 1) Found SUBST but not in the PST phase
3137                                 # 2) Didn't find SUBST and in the PST phase
3138                                 # In either case, return without further processing
3139                                 $text = '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3140                                 $found = true;
3141                                 $noparse = true;
3142                                 $noargs = true;
3143                         }
3144                 }
3145
3146                 # Variables
3147                 if ( !$found && $args->length == 0 ) {
3148                         $id = $this->mVariables->matchStartToEnd( $part1 );
3149                         if ( $id !== false ) {
3150                                 $text = $this->getVariableValue( $id );
3151                                 $this->mOutput->mContainsOldMagic = true;
3152                                 $found = true;
3153                                 $noparse = true;
3154                                 $noargs = true;
3155                         }
3156                 }
3157
3158                 # MSG, MSGNW and RAW
3159                 if ( !$found ) {
3160                         # Check for MSGNW:
3161                         $mwMsgnw =& MagicWord::get( 'msgnw' );
3162                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3163                                 $nowiki = true;
3164                         } else {
3165                                 # Remove obsolete MSG:
3166                                 $mwMsg =& MagicWord::get( 'msg' );
3167                                 $mwMsg->matchStartAndRemove( $part1 );
3168                         }
3169
3170                         # Check for RAW:
3171                         $mwRaw =& MagicWord::get( 'raw' );
3172                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3173                                 $forceRawInterwiki = true;
3174                         }
3175                 }
3176                 wfProfileOut( __METHOD__.'-modifiers' );
3177
3178                 # Save path level before recursing into functions & templates.
3179                 $lastPathLevel = $this->mTemplatePath;
3180
3181                 # Parser functions
3182                 if ( !$found ) {
3183                         wfProfileIn( __METHOD__ . '-pfunc' );
3184
3185                         $colonPos = strpos( $part1, ':' );
3186                         if ( $colonPos !== false ) {
3187                                 # Case sensitive functions
3188                                 $function = substr( $part1, 0, $colonPos );
3189                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3190                                         $function = $this->mFunctionSynonyms[1][$function];
3191                                 } else {
3192                                         # Case insensitive functions
3193                                         $function = strtolower( $function );
3194                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3195                                                 $function = $this->mFunctionSynonyms[0][$function];
3196                                         } else {
3197                                                 $function = false;
3198                                         }
3199                                 }
3200                                 if ( $function ) {
3201                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3202                                         $initialArgs = array( &$this );
3203                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3204                                         if ( $flags & SFH_OBJECT_ARGS ) {
3205                                                 # Add a frame parameter, and pass the arguments as an array
3206                                                 $allArgs = $initialArgs;
3207                                                 $allArgs[] = $frame;
3208                                                 foreach ( $args as $arg ) {
3209                                                         $funcArgs[] = $arg;
3210                                                 }
3211                                                 $allArgs[] = $funcArgs;
3212                                         } else {
3213                                                 # Convert arguments to plain text
3214                                                 foreach ( $args as $arg ) {
3215                                                         $funcArgs[] = trim( $frame->expand( $arg ) );
3216                                                 }
3217                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3218                                         }
3219
3220                                         $result = call_user_func_array( $callback, $allArgs );
3221                                         $found = true;
3222
3223                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
3224                                         $noargs = true;
3225                                         $noparse = true;
3226
3227                                         if ( is_array( $result ) ) {
3228                                                 if ( isset( $result[0] ) ) {
3229                                                         $text = $result[0];
3230                                                         unset( $result[0] );
3231                                                 }
3232
3233                                                 // Extract flags into the local scope
3234                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3235                                                 extract( $result );
3236                                         } else {
3237                                                 $text = $result;
3238                                         }
3239                                 }
3240                         }
3241                         wfProfileOut( __METHOD__ . '-pfunc' );
3242                 }
3243
3244                 # Finish mangling title and then check for loops.
3245                 # Set $title to a Title object and $titleText to the PDBK
3246                 if ( !$found ) {
3247                         $ns = NS_TEMPLATE;
3248                         # Split the title into page and subpage
3249                         $subpage = '';
3250                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3251                         if ($subpage !== '') {
3252                                 $ns = $this->mTitle->getNamespace();
3253                         }
3254                         $title = Title::newFromText( $part1, $ns );
3255                         if ( $title ) {
3256                                 $titleText = $title->getPrefixedText();
3257                                 # Check for language variants if the template is not found
3258                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3259                                         $wgContLang->findVariantLink($part1, $title);
3260                                 }
3261                                 # Do infinite loop check
3262                                 if ( isset( $this->mTemplatePath[$titleText] ) ) {
3263                                         $noparse = true;
3264                                         $noargs = true;
3265                                         $found = true;
3266                                         $text = "[[$part1]]" . $this->insertStripItem( '<!-- WARNING: template loop detected -->' );
3267                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3268                                 }
3269                         }
3270                 }
3271
3272                 # Load from database
3273                 if ( !$found && $title ) {
3274                         wfProfileIn( __METHOD__ . '-loadtpl' );
3275                         if ( !$title->isExternal() ) {
3276                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3277                                         $text = SpecialPage::capturePath( $title );
3278                                         if ( is_string( $text ) ) {
3279                                                 $found = true;
3280                                                 $noparse = true;
3281                                                 $noargs = true;
3282                                                 $isHTML = true;
3283                                                 $this->disableCache();
3284                                         }
3285                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3286                                         $found = false; //access denied
3287                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3288                                 } else {
3289                                         list( $text, $title ) = $this->getTemplateDom( $title );
3290                                         if ( $text !== false ) {
3291                                                 $found = true;
3292                                                 $isDOM = true;
3293                                         }
3294                                 }
3295
3296                                 # If the title is valid but undisplayable, make a link to it
3297                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3298                                         $text = "[[:$titleText]]";
3299                                         $found = true;
3300                                 }
3301                         } elseif ( $title->isTrans() ) {
3302                                 // Interwiki transclusion
3303                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3304                                         $text = $this->interwikiTransclude( $title, 'render' );
3305                                         $isHTML = true;
3306                                         $noparse = true;
3307                                 } else {
3308                                         $text = $this->interwikiTransclude( $title, 'raw' );
3309                                 }
3310                                 $found = true;
3311                         }
3312                         wfProfileOut( __METHOD__ . '-loadtpl' );
3313                 }
3314
3315                 # Recursive parsing, escaping and link table handling
3316                 # Only for HTML output
3317                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3318                         if ( $isDOM ) {
3319                                 $text = $frame->expand( $text );
3320                         }
3321                         $text = wfEscapeWikiText( $text );
3322                 } elseif ( !$this->ot['msg'] && $found ) {
3323                         if ( $noargs ) {
3324                                 $newFrame = $frame->newChild();
3325                         } else {
3326                                 # Clean up argument array
3327                                 $newFrame = $frame->newChild( $args, $title );
3328                                 # Add a new element to the templace recursion path
3329                                 $this->mTemplatePath[$titleText] = 1;
3330                         }
3331
3332                         if ( !$noparse ) {
3333                                 if ( $isDOM ) {
3334                                         if ( $titleText !== false && count( $newFrame->args ) == 0 ) {
3335                                                 # Expansion is eligible for the empty-frame cache
3336                                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3337                                                         $text = $this->mTplExpandCache[$titleText];
3338                                                 } else {
3339                                                         $text = $newFrame->expand( $text );
3340                                                         $this->mTplExpandCache[$titleText] = $text;
3341                                                 }
3342                                         } else {
3343                                                 $text = $newFrame->expand( $text );
3344                                         }
3345                                 } else {
3346                                         $text = $this->replaceVariables( $text, $newFrame );
3347                                 }
3348
3349                                 # strip woz 'ere 2004-07
3350
3351                                 # Bug 529: if the template begins with a table or block-level
3352                                 # element, it should be treated as beginning a new line.
3353                                 # This behaviour is somewhat controversial.
3354                                 if (!$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3355                                         $text = "\n" . $text;
3356                                 }
3357                         } elseif ( !$noargs ) {
3358                                 # $noparse and !$noargs
3359                                 # Just replace the arguments, not any double-brace items
3360                                 # This is used for rendered interwiki transclusion
3361                                 if ( $isDOM ) {
3362                                         $text = $newFrame->expand( $text, PPFrame::NO_TEMPLATES );
3363                                 } else {
3364                                         $text = $this->replaceVariables( $text, $newFrame, true );
3365                                 }
3366                         } elseif ( $isDOM ) {
3367                                 $text = $frame->expand( $text );
3368                         }
3369                 } elseif ( $isDOM ) {
3370                         $text = $frame->expand( $text, PPFrame::NO_TEMPLATES | PPFrame::NO_ARGS );
3371                 }
3372
3373                 # Prune lower levels off the recursion check path
3374                 $this->mTemplatePath = $lastPathLevel;
3375
3376                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3377                         # Error, oversize inclusion
3378                         $text = "[[$originalTitle]]" .
3379                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3380                         $noparse = true;
3381                         $noargs = true;
3382                 }
3383
3384                 if ( !$found ) {
3385                         wfProfileOut( $fname );
3386                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3387                 } else {
3388                         wfProfileIn( __METHOD__ . '-placeholders' );
3389                         if ( $isHTML ) {
3390                                 # Replace raw HTML by a placeholder
3391                                 # Add a blank line preceding, to prevent it from mucking up
3392                                 # immediately preceding headings
3393                                 $text = "\n\n" . $this->insertStripItem( $text );
3394                         }
3395                         wfProfileOut( __METHOD__ . '-placeholders' );
3396                 }
3397
3398                 # Prune lower levels off the recursion check path
3399                 $this->mTemplatePath = $lastPathLevel;
3400
3401                 if ( !$found ) {
3402                         wfProfileOut( $fname );
3403                         return '{{' . $frame->implode( '|', $titleWithSpaces, $args ) . '}}';
3404                 } else {
3405                         wfProfileOut( $fname );
3406                         return $text;
3407                 }
3408         }
3409
3410         /**
3411          * Get the semi-parsed DOM representation of a template with a given title,
3412          * and its redirect destination title. Cached.
3413          */
3414         function getTemplateDom( $title ) {
3415                 $cacheTitle = $title;
3416                 $titleText = $title->getPrefixedDBkey();
3417
3418                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3419                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3420                         $title = Title::makeTitle( $ns, $dbk );
3421                         $titleText = $title->getPrefixedDBkey();
3422                 }
3423                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3424                         return array( $this->mTplDomCache[$titleText], $title );
3425                 }
3426
3427                 // Cache miss, go to the database
3428                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3429
3430                 if ( $text === false ) {
3431                         $this->mTplDomCache[$titleText] = false;
3432                         return array( false, $title );
3433                 }
3434
3435                 # If there are any <onlyinclude> tags, only include them
3436                 if ( !$this->ot['msg'] ) {
3437                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3438                                 $replacer = new OnlyIncludeReplacer;
3439                                 StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3440                                         array( &$replacer, 'replace' ), $text );
3441                                 $text = $replacer->output;
3442                         }
3443                         # Remove <noinclude> sections and <includeonly> tags
3444                         $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3445                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3446                 }
3447
3448                 # Remove comments
3449                 # This could theoretically be merged into preprocessToDom()
3450                 if ( $this->ot['html'] || ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) ) {
3451                         $text = Sanitizer::removeHTMLcomments( $text );
3452                 }
3453
3454                 $dom = $this->preprocessToDom( $text );
3455
3456                 $this->mTplDomCache[ $titleText ] = $dom;
3457
3458                 if (! $title->equals($cacheTitle)) {
3459                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3460                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3461                 }
3462
3463                 return array( $dom, $title );
3464         }
3465
3466         /**
3467          * Fetch the unparsed text of a template and register a reference to it.
3468          */
3469         function fetchTemplateAndTitle( $title ) {
3470                 $templateCb = $this->mOptions->getTemplateCallback();
3471                 $stuff = call_user_func( $templateCb, $title );
3472                 $text = $stuff['text'];
3473                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3474                 if ( isset( $stuff['deps'] ) ) {
3475                         foreach ( $stuff['deps'] as $dep ) {
3476                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3477                         }
3478                 }
3479                 return array($text,$finalTitle);
3480         }
3481
3482         function fetchTemplate( $title ) {
3483                 $rv = $this->fetchTemplateAndTitle($title);
3484                 return $rv[0];
3485         }
3486
3487         /**
3488          * Static function to get a template
3489          * Can be overridden via ParserOptions::setTemplateCallback().
3490          */
3491         static function statelessFetchTemplate( $title ) {
3492                 $text = $skip = false;
3493                 $finalTitle = $title;
3494                 $deps = array();
3495
3496                 // Loop to fetch the article, with up to 1 redirect
3497                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3498                         # Give extensions a chance to select the revision instead
3499                         $id = false; // Assume current
3500                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3501
3502                         if( $skip ) {
3503                                 $text = false;
3504                                 $deps[] = array(
3505                                         'title' => $title,
3506                                         'page_id' => $title->getArticleID(),
3507                                         'rev_id' => null );
3508                                 break;
3509                         }
3510                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3511                         $rev_id = $rev ? $rev->getId() : 0;
3512
3513                         $deps[] = array(
3514                                 'title' => $title,
3515                                 'page_id' => $title->getArticleID(),
3516                                 'rev_id' => $rev_id );
3517
3518                         if( $rev ) {
3519                                 $text = $rev->getText();
3520                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3521                                 global $wgLang;
3522                                 $message = $wgLang->lcfirst( $title->getText() );
3523                                 $text = wfMsgForContentNoTrans( $message );
3524                                 if( wfEmptyMsg( $message, $text ) ) {
3525                                         $text = false;
3526                                         break;
3527                                 }
3528                         } else {
3529                                 break;
3530                         }
3531                         if ( $text === false ) {
3532                                 break;
3533                         }
3534                         // Redirect?
3535                         $finalTitle = $title;
3536                         $title = Title::newFromRedirect( $text );
3537                 }
3538                 return array(
3539                         'text' => $text,
3540                         'finalTitle' => $finalTitle,
3541                         'deps' => $deps );
3542         }
3543
3544         /**
3545          * Transclude an interwiki link.
3546          */
3547         function interwikiTransclude( $title, $action ) {
3548                 global $wgEnableScaryTranscluding;
3549
3550                 if (!$wgEnableScaryTranscluding)
3551                         return wfMsg('scarytranscludedisabled');
3552
3553                 $url = $title->getFullUrl( "action=$action" );
3554
3555                 if (strlen($url) > 255)
3556                         return wfMsg('scarytranscludetoolong');
3557                 return $this->fetchScaryTemplateMaybeFromCache($url);
3558         }
3559
3560         function fetchScaryTemplateMaybeFromCache($url) {
3561                 global $wgTranscludeCacheExpiry;
3562                 $dbr = wfGetDB(DB_SLAVE);
3563                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3564                                 array('tc_url' => $url));
3565                 if ($obj) {
3566                         $time = $obj->tc_time;
3567                         $text = $obj->tc_contents;
3568                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3569                                 return $text;
3570                         }
3571                 }
3572
3573                 $text = Http::get($url);
3574                 if (!$text)
3575                         return wfMsg('scarytranscludefailed', $url);
3576
3577                 $dbw = wfGetDB(DB_MASTER);
3578                 $dbw->replace('transcache', array('tc_url'), array(
3579                         'tc_url' => $url,
3580                         'tc_time' => time(),
3581                         'tc_contents' => $text));
3582                 return $text;
3583         }
3584
3585
3586         /**
3587          * Triple brace replacement -- used for template arguments
3588          * @private
3589          */
3590         function argSubstitution( $piece, $frame ) {
3591                 wfProfileIn( __METHOD__ );
3592
3593                 $text = false;
3594                 $error = false;
3595                 $parts = $piece['parts'];
3596                 $argWithSpaces = $frame->expand( $piece['title'] );
3597                 $arg = trim( $argWithSpaces );
3598
3599                 if ( isset( $frame->args[$arg] ) ) {
3600                         $text = $frame->parent->expand( $frame->args[$arg] );
3601                 } else if ( ( $this->ot['html'] || $this->ot['pre'] ) && $parts->length > 0 ) {
3602                         $text = $frame->expand( $parts->item( 0 ) );
3603                 }
3604                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3605                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3606                 }
3607
3608                 if ( $text === false ) {
3609                         $text = '{{{' . $frame->implode( '|', $argWithSpaces, $parts ) . '}}}';
3610                 }
3611                 if ( $error !== false ) {
3612                         $text .= $error;
3613                 }
3614
3615                 wfProfileOut( __METHOD__ );
3616                 return $text;
3617         }
3618
3619         /**
3620          * Return the text to be used for a given extension tag.
3621          * This is the ghost of strip().
3622          *
3623          * @param array $params Associative array of parameters:
3624          *     name       DOMNode for the tag name
3625          *     attrText   DOMNode for unparsed text where tag attributes are thought to be
3626          *     inner      Contents of extension element
3627          *     noClose    Original text did not have a close tag
3628          * @param PPFrame $frame
3629          */
3630         function extensionSubstitution( $params, $frame ) {
3631                 global $wgRawHtml, $wgContLang;
3632                 static $n = 1;
3633
3634                 $name = $frame->expand( $params['name'] );
3635                 $attrText = is_null( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3636                 $content = is_null( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3637
3638                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3639
3640                 if ( $this->ot['html'] ) {
3641                         $name = strtolower( $name );
3642
3643                         $params = Sanitizer::decodeTagAttributes( $attrText );
3644                         switch ( $name ) {
3645                                 case 'html':
3646                                         if( $wgRawHtml ) {
3647                                                 $output = $content;
3648                                                 break;
3649                                         } else {
3650                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3651                                         }
3652                                 case 'nowiki':
3653                                         $output = Xml::escapeTagsOnly( $content );
3654                                         break;
3655                                 case 'math':
3656                                         $output = $wgContLang->armourMath(
3657                                                 MathRenderer::renderMath( $content, $params ) );
3658                                         break;
3659                                 case 'gallery':
3660                                         $output = $this->renderImageGallery( $content, $params );
3661                                         break;
3662                                 default:
3663                                         if( isset( $this->mTagHooks[$name] ) ) {
3664                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3665                                                         array( $content, $params, $this ) );
3666                                         } else {
3667                                                 throw new MWException( "Invalid call hook $name" );
3668                                         }
3669                         }
3670                 } else {
3671                         if ( $content === null ) {
3672                                 $output = "<$name$attrText/>";
3673                         } else {
3674                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3675                                 $output = "<$name$attrText>$content$close";
3676                         }
3677                 }
3678
3679                 if ( $name == 'html' || $name == 'nowiki' ) {
3680                         $this->mStripState->nowiki->setPair( $marker, $output );
3681                 } else {
3682                         $this->mStripState->general->setPair( $marker, $output );
3683                 }
3684                 return $marker;
3685         }
3686
3687         /**
3688          * Increment an include size counter
3689          *
3690          * @param string $type The type of expansion
3691          * @param integer $size The size of the text
3692          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3693          */
3694         function incrementIncludeSize( $type, $size ) {
3695                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3696                         return false;
3697                 } else {
3698                         $this->mIncludeSizes[$type] += $size;
3699                         return true;
3700                 }
3701         }
3702
3703         /**
3704          * Detect __NOGALLERY__ magic word and set a placeholder
3705          */
3706         function stripNoGallery( &$text ) {
3707                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3708                 # do not add TOC
3709                 $mw = MagicWord::get( 'nogallery' );
3710                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3711         }
3712
3713         /**
3714          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3715          * placeholder that will then be replaced by the real TOC in
3716          * ->formatHeadings, this works because at this points real
3717          * comments will have already been discarded by the sanitizer.
3718          *
3719          * Any additional __TOC__ magic words left over will be discarded
3720          * as there can only be one TOC on the page.
3721          */
3722         function stripToc( $text ) {
3723                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3724                 # do not add TOC
3725                 $mw = MagicWord::get( 'notoc' );
3726                 if( $mw->matchAndRemove( $text ) ) {
3727                         $this->mShowToc = false;
3728                 }
3729
3730                 $mw = MagicWord::get( 'toc' );
3731                 if( $mw->match( $text ) ) {
3732                         $this->mShowToc = true;
3733                         $this->mForceTocPosition = true;
3734
3735                         // Set a placeholder. At the end we'll fill it in with the TOC.
3736                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3737
3738                         // Only keep the first one.
3739                         $text = $mw->replace( '', $text );
3740                 }
3741                 return $text;
3742         }
3743
3744         /**
3745          * This function accomplishes several tasks:
3746          * 1) Auto-number headings if that option is enabled
3747          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3748          * 3) Add a Table of contents on the top for users who have enabled the option
3749          * 4) Auto-anchor headings
3750          *
3751          * It loops through all headlines, collects the necessary data, then splits up the
3752          * string and re-inserts the newly formatted headlines.
3753          *
3754          * @param string $text
3755          * @param boolean $isMain
3756          * @private
3757          */
3758         function formatHeadings( $text, $isMain=true ) {
3759                 global $wgMaxTocLevel, $wgContLang;
3760
3761                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3762                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3763                         $showEditLink = 0;
3764                 } else {
3765                         $showEditLink = $this->mOptions->getEditSection();
3766                 }
3767
3768                 # Inhibit editsection links if requested in the page
3769                 $esw =& MagicWord::get( 'noeditsection' );
3770                 if( $esw->matchAndRemove( $text ) ) {
3771                         $showEditLink = 0;
3772                 }
3773
3774                 # Get all headlines for numbering them and adding funky stuff like [edit]
3775                 # links - this is for later, but we need the number of headlines right now
3776                 $matches = array();
3777                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3778
3779                 # if there are fewer than 4 headlines in the article, do not show TOC
3780                 # unless it's been explicitly enabled.
3781                 $enoughToc = $this->mShowToc &&
3782                         (($numMatches >= 4) || $this->mForceTocPosition);
3783
3784                 # Allow user to stipulate that a page should have a "new section"
3785                 # link added via __NEWSECTIONLINK__
3786                 $mw =& MagicWord::get( 'newsectionlink' );
3787                 if( $mw->matchAndRemove( $text ) )
3788                         $this->mOutput->setNewSection( true );
3789
3790                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3791                 # override above conditions and always show TOC above first header
3792                 $mw =& MagicWord::get( 'forcetoc' );
3793                 if ($mw->matchAndRemove( $text ) ) {
3794                         $this->mShowToc = true;
3795                         $enoughToc = true;
3796                 }
3797
3798                 # We need this to perform operations on the HTML
3799                 $sk = $this->mOptions->getSkin();
3800
3801                 # headline counter
3802                 $headlineCount = 0;
3803                 $numVisible = 0;
3804
3805                 # Ugh .. the TOC should have neat indentation levels which can be
3806                 # passed to the skin functions. These are determined here
3807                 $toc = '';
3808                 $full = '';
3809                 $head = array();
3810                 $sublevelCount = array();
3811                 $levelCount = array();
3812                 $toclevel = 0;
3813                 $level = 0;
3814                 $prevlevel = 0;
3815                 $toclevel = 0;
3816                 $prevtoclevel = 0;
3817                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3818                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3819                 $tocraw = array();
3820
3821                 foreach( $matches[3] as $headline ) {
3822                         $isTemplate = false;
3823                         $titleText = false;
3824                         $sectionIndex = false;
3825                         $numbering = '';
3826                         $markerMatches = array();
3827                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3828                                 $serial = $markerMatches[1];
3829                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3830                                 $isTemplate = ($titleText != $baseTitleText);
3831                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3832                         }
3833
3834                         if( $toclevel ) {
3835                                 $prevlevel = $level;
3836                                 $prevtoclevel = $toclevel;
3837                         }
3838                         $level = $matches[1][$headlineCount];
3839
3840                         if( $doNumberHeadings || $enoughToc ) {
3841
3842                                 if ( $level > $prevlevel ) {
3843                                         # Increase TOC level
3844                                         $toclevel++;
3845                                         $sublevelCount[$toclevel] = 0;
3846                                         if( $toclevel<$wgMaxTocLevel ) {
3847                                                 $prevtoclevel = $toclevel;
3848                                                 $toc .= $sk->tocIndent();
3849                                                 $numVisible++;
3850                                         }
3851                                 }
3852                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3853                                         # Decrease TOC level, find level to jump to
3854
3855                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3856                                                 # Can only go down to level 1
3857                                                 $toclevel = 1;
3858                                         } else {
3859                                                 for ($i = $toclevel; $i > 0; $i--) {
3860                                                         if ( $levelCount[$i] == $level ) {
3861                                                                 # Found last matching level
3862                                                                 $toclevel = $i;
3863                                                                 break;
3864                                                         }
3865                                                         elseif ( $levelCount[$i] < $level ) {
3866                                                                 # Found first matching level below current level
3867                                                                 $toclevel = $i + 1;
3868                                                                 break;
3869                                                         }
3870                                                 }
3871                                         }
3872                                         if( $toclevel<$wgMaxTocLevel ) {
3873                                                 if($prevtoclevel < $wgMaxTocLevel) {
3874                                                         # Unindent only if the previous toc level was shown :p
3875                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3876                                                 } else {
3877                                                         $toc .= $sk->tocLineEnd();
3878                                                 }
3879                                         }
3880                                 }
3881                                 else {
3882                                         # No change in level, end TOC line
3883                                         if( $toclevel<$wgMaxTocLevel ) {
3884                                                 $toc .= $sk->tocLineEnd();
3885                                         }
3886                                 }
3887
3888                                 $levelCount[$toclevel] = $level;
3889
3890                                 # count number of headlines for each level
3891                                 @$sublevelCount[$toclevel]++;
3892                                 $dot = 0;
3893                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3894                                         if( !empty( $sublevelCount[$i] ) ) {
3895                                                 if( $dot ) {
3896                                                         $numbering .= '.';
3897                                                 }
3898                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3899                                                 $dot = 1;
3900                                         }
3901                                 }
3902                         }
3903
3904                         # The safe header is a version of the header text safe to use for links
3905                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3906                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3907
3908                         # Remove link placeholders by the link text.
3909                         #     <!--LINK number-->
3910                         # turns into
3911                         #     link text with suffix
3912                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3913                                                             "\$this->mLinkHolders['texts'][\$1]",
3914                                                             $safeHeadline );
3915                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3916                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3917                                                             $safeHeadline );
3918
3919                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3920                         $tocline = preg_replace(
3921                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3922                                 array( '',                          '<$1>'),
3923                                 $safeHeadline
3924                         );
3925                         $tocline = trim( $tocline );
3926
3927                         # For the anchor, strip out HTML-y stuff period
3928                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3929                         $safeHeadline = trim( $safeHeadline );
3930
3931                         # Save headline for section edit hint before it's escaped
3932                         $headlineHint = $safeHeadline;
3933                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3934                         $refers[$headlineCount] = $safeHeadline;
3935
3936                         # count how many in assoc. array so we can track dupes in anchors
3937                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3938                         $refcount[$headlineCount] = $refers[$safeHeadline];
3939
3940                         # Don't number the heading if it is the only one (looks silly)
3941                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3942                                 # the two are different if the line contains a link
3943                                 $headline=$numbering . ' ' . $headline;
3944                         }
3945
3946                         # Create the anchor for linking from the TOC to the section
3947                         $anchor = $safeHeadline;
3948                         if($refcount[$headlineCount] > 1 ) {
3949                                 $anchor .= '_' . $refcount[$headlineCount];
3950                         }
3951                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3952                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3953                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3954                         }
3955                         # give headline the correct <h#> tag
3956                         if( $showEditLink && $sectionIndex !== false ) {
3957                                 if( $isTemplate )
3958                                         $editlink = $sk->editSectionLinkForOther($titleText, $sectionIndex);
3959                                 else
3960                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3961                         } else {
3962                                 $editlink = '';
3963                         }
3964                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3965
3966                         $headlineCount++;
3967                 }
3968
3969                 $this->mOutput->setSections( $tocraw );
3970
3971                 # Never ever show TOC if no headers
3972                 if( $numVisible < 1 ) {
3973                         $enoughToc = false;
3974                 }
3975
3976                 if( $enoughToc ) {
3977                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3978                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3979                         }
3980                         $toc = $sk->tocList( $toc );
3981                 }
3982
3983                 # split up and insert constructed headlines
3984
3985                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3986                 $i = 0;
3987
3988                 foreach( $blocks as $block ) {
3989                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3990                                 # This is the [edit] link that appears for the top block of text when
3991                                 # section editing is enabled
3992
3993                                 # Disabled because it broke block formatting
3994                                 # For example, a bullet point in the top line
3995                                 # $full .= $sk->editSectionLink(0);
3996                         }
3997                         $full .= $block;
3998                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3999                                 # Top anchor now in skin
4000                                 $full = $full.$toc;
4001                         }
4002
4003                         if( !empty( $head[$i] ) ) {
4004                                 $full .= $head[$i];
4005                         }
4006                         $i++;
4007                 }
4008                 if( $this->mForceTocPosition ) {
4009                         return str_replace( '<!--MWTOC-->', $toc, $full );
4010                 } else {
4011                         return $full;
4012                 }
4013         }
4014
4015         /**
4016          * Transform wiki markup when saving a page by doing \r\n -> \n
4017          * conversion, substitting signatures, {{subst:}} templates, etc.
4018          *
4019          * @param string $text the text to transform
4020          * @param Title &$title the Title object for the current article
4021          * @param User &$user the User object describing the current user
4022          * @param ParserOptions $options parsing options
4023          * @param bool $clearState whether to clear the parser state first
4024          * @return string the altered wiki markup
4025          * @public
4026          */
4027         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
4028                 $this->mOptions = $options;
4029                 $this->mTitle =& $title;
4030                 $this->setOutputType( OT_WIKI );
4031
4032                 if ( $clearState ) {
4033                         $this->clearState();
4034                 }
4035
4036                 $pairs = array(
4037                         "\r\n" => "\n",
4038                 );
4039                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4040                 $text = $this->pstPass2( $text, $user );
4041                 $text = $this->mStripState->unstripBoth( $text );
4042                 return $text;
4043         }
4044
4045         /**
4046          * Pre-save transform helper function
4047          * @private
4048          */
4049         function pstPass2( $text, $user ) {
4050                 global $wgContLang, $wgLocaltimezone;
4051
4052                 /* Note: This is the timestamp saved as hardcoded wikitext to
4053                  * the database, we use $wgContLang here in order to give
4054                  * everyone the same signature and use the default one rather
4055                  * than the one selected in each user's preferences.
4056                  */
4057                 if ( isset( $wgLocaltimezone ) ) {
4058                         $oldtz = getenv( 'TZ' );
4059                         putenv( 'TZ='.$wgLocaltimezone );
4060                 }
4061                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
4062                   ' (' . date( 'T' ) . ')';
4063                 if ( isset( $wgLocaltimezone ) ) {
4064                         putenv( 'TZ='.$oldtz );
4065                 }
4066
4067                 # Variable replacement
4068                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4069                 $text = $this->replaceVariables( $text );
4070
4071                 # Strip out <nowiki> etc. added via replaceVariables
4072                 #$text = $this->strip( $text, $this->mStripState, false, array( 'gallery' ) );
4073
4074                 # Signatures
4075                 $sigText = $this->getUserSig( $user );
4076                 $text = strtr( $text, array(
4077                         '~~~~~' => $d,
4078                         '~~~~' => "$sigText $d",
4079                         '~~~' => $sigText
4080                 ) );
4081
4082                 # Context links: [[|name]] and [[name (context)|]]
4083                 #
4084                 global $wgLegalTitleChars;
4085                 $tc = "[$wgLegalTitleChars]";
4086                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
4087
4088                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4089                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4090                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4091
4092                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4093                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4094                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4095
4096                 $t = $this->mTitle->getText();
4097                 $m = array();
4098                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4099                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4100                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
4101                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4102                 } else {
4103                         # if there's no context, don't bother duplicating the title
4104                         $text = preg_replace( $p2, '[[\\1]]', $text );
4105                 }
4106
4107                 # Trim trailing whitespace
4108                 $text = rtrim( $text );
4109
4110                 return $text;
4111         }
4112
4113         /**
4114          * Fetch the user's signature text, if any, and normalize to
4115          * validated, ready-to-insert wikitext.
4116          *
4117          * @param User $user
4118          * @return string
4119          * @private
4120          */
4121         function getUserSig( &$user ) {
4122                 global $wgMaxSigChars;
4123
4124                 $username = $user->getName();
4125                 $nickname = $user->getOption( 'nickname' );
4126                 $nickname = $nickname === '' ? $username : $nickname;
4127
4128                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4129                         $nickname = $username;
4130                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4131                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
4132                         # Sig. might contain markup; validate this
4133                         if( $this->validateSig( $nickname ) !== false ) {
4134                                 # Validated; clean up (if needed) and return it
4135                                 return $this->cleanSig( $nickname, true );
4136                         } else {
4137                                 # Failed to validate; fall back to the default
4138                                 $nickname = $username;
4139                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
4140                         }
4141                 }
4142
4143                 // Make sure nickname doesnt get a sig in a sig
4144                 $nickname = $this->cleanSigInSig( $nickname );
4145
4146                 # If we're still here, make it a link to the user page
4147                 $userText = wfEscapeWikiText( $username );
4148                 $nickText = wfEscapeWikiText( $nickname );
4149                 if ( $user->isAnon() )  {
4150                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4151                 } else {
4152                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4153                 }
4154         }
4155
4156         /**
4157          * Check that the user's signature contains no bad XML
4158          *
4159          * @param string $text
4160          * @return mixed An expanded string, or false if invalid.
4161          */
4162         function validateSig( $text ) {
4163                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
4164         }
4165
4166         /**
4167          * Clean up signature text
4168          *
4169          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4170          * 2) Substitute all transclusions
4171          *
4172          * @param string $text
4173          * @param $parsing Whether we're cleaning (preferences save) or parsing
4174          * @return string Signature text
4175          */
4176         function cleanSig( $text, $parsing = false ) {
4177                 if ( !$parsing ) {
4178                         global $wgTitle;
4179                         $this->startExternalParse( $wgTitle, new ParserOptions(), OT_MSG );
4180                 }
4181
4182                 # FIXME: regex doesn't respect extension tags or nowiki
4183                 #  => Move this logic to braceSubstitution()
4184                 $substWord = MagicWord::get( 'subst' );
4185                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4186                 $substText = '{{' . $substWord->getSynonym( 0 );
4187
4188                 $text = preg_replace( $substRegex, $substText, $text );
4189                 $text = $this->cleanSigInSig( $text );
4190                 $dom = $this->preprocessToDom( $text );
4191                 $frame = new PPFrame( $this );
4192                 $text = $frame->expand( $dom->documentElement );
4193
4194                 if ( !$parsing ) {
4195                         $text = $this->mStripState->unstripBoth( $text );
4196                 }
4197
4198                 return $text;
4199         }
4200
4201         /**
4202          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4203          * @param string $text
4204          * @return string Signature text with /~{3,5}/ removed
4205          */
4206         function cleanSigInSig( $text ) {
4207                 $text = preg_replace( '/~{3,5}/', '', $text );
4208                 return $text;
4209         }
4210
4211         /**
4212          * Set up some variables which are usually set up in parse()
4213          * so that an external function can call some class members with confidence
4214          * @public
4215          */
4216         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4217                 $this->mTitle =& $title;
4218                 $this->mOptions = $options;
4219                 $this->setOutputType( $outputType );
4220                 if ( $clearState ) {
4221                         $this->clearState();
4222                 }
4223         }
4224
4225         /**
4226          * Transform a MediaWiki message by replacing magic variables.
4227          *
4228          * For some unknown reason, it also expands templates, but only to the
4229          * first recursion level. This is wrong and broken, probably introduced
4230          * accidentally during refactoring, but probably relied upon by thousands
4231          * of users.
4232          *
4233          * @param string $text the text to transform
4234          * @param ParserOptions $options  options
4235          * @return string the text with variables substituted
4236          * @public
4237          */
4238         function transformMsg( $text, $options ) {
4239                 global $wgTitle;
4240                 static $executing = false;
4241
4242                 $fname = "Parser::transformMsg";
4243
4244                 # Guard against infinite recursion
4245                 if ( $executing ) {
4246                         return $text;
4247                 }
4248                 $executing = true;
4249
4250                 wfProfileIn($fname);
4251
4252                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
4253                         $this->mTitle = $wgTitle;
4254                 } else {
4255                         $this->mTitle = Title::newFromText('msg');
4256                 }
4257                 $this->mOptions = $options;
4258                 $this->setOutputType( OT_MSG );
4259                 $this->clearState();
4260                 $text = $this->replaceVariables( $text );
4261                 $text = $this->mStripState->unstripBoth( $text );
4262
4263                 $executing = false;
4264                 wfProfileOut($fname);
4265                 return $text;
4266         }
4267
4268         /**
4269          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4270          * The callback should have the following form:
4271          *    function myParserHook( $text, $params, &$parser ) { ... }
4272          *
4273          * Transform and return $text. Use $parser for any required context, e.g. use
4274          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4275          *
4276          * @public
4277          *
4278          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4279          * @param mixed $callback The callback function (and object) to use for the tag
4280          *
4281          * @return The old value of the mTagHooks array associated with the hook
4282          */
4283         function setHook( $tag, $callback ) {
4284                 $tag = strtolower( $tag );
4285                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4286                 $this->mTagHooks[$tag] = $callback;
4287                 $this->mStripList[] = $tag;
4288
4289                 return $oldVal;
4290         }
4291
4292         function setTransparentTagHook( $tag, $callback ) {
4293                 $tag = strtolower( $tag );
4294                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4295                 $this->mTransparentTagHooks[$tag] = $callback;
4296
4297                 return $oldVal;
4298         }
4299
4300         /**
4301          * Create a function, e.g. {{sum:1|2|3}}
4302          * The callback function should have the form:
4303          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4304          *
4305          * The callback may either return the text result of the function, or an array with the text
4306          * in element 0, and a number of flags in the other elements. The names of the flags are
4307          * specified in the keys. Valid flags are:
4308          *   found                     The text returned is valid, stop processing the template. This
4309          *                             is on by default.
4310          *   nowiki                    Wiki markup in the return value should be escaped
4311          *   noparse                   Unsafe HTML tags should not be stripped, etc.
4312          *   noargs                    Don't replace triple-brace arguments in the return value
4313          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4314          *
4315          * @public
4316          *
4317          * @param string $id The magic word ID
4318          * @param mixed $callback The callback function (and object) to use
4319          * @param integer $flags a combination of the following flags:
4320          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4321          *
4322          * @return The old callback function for this name, if any
4323          */
4324         function setFunctionHook( $id, $callback, $flags = 0 ) {
4325                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4326                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4327
4328                 # Add to function cache
4329                 $mw = MagicWord::get( $id );
4330                 if( !$mw )
4331                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4332
4333                 $synonyms = $mw->getSynonyms();
4334                 $sensitive = intval( $mw->isCaseSensitive() );
4335
4336                 foreach ( $synonyms as $syn ) {
4337                         # Case
4338                         if ( !$sensitive ) {
4339                                 $syn = strtolower( $syn );
4340                         }
4341                         # Add leading hash
4342                         if ( !( $flags & SFH_NO_HASH ) ) {
4343                                 $syn = '#' . $syn;
4344                         }
4345                         # Remove trailing colon
4346                         if ( substr( $syn, -1, 1 ) == ':' ) {
4347                                 $syn = substr( $syn, 0, -1 );
4348                         }
4349                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4350                 }
4351                 return $oldVal;
4352         }
4353
4354         /**
4355          * Get all registered function hook identifiers
4356          *
4357          * @return array
4358          */
4359         function getFunctionHooks() {
4360                 return array_keys( $this->mFunctionHooks );
4361         }
4362
4363         /**
4364          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4365          * Placeholders created in Skin::makeLinkObj()
4366          * Returns an array of link CSS classes, indexed by PDBK.
4367          * $options is a bit field, RLH_FOR_UPDATE to select for update
4368          */
4369         function replaceLinkHolders( &$text, $options = 0 ) {
4370                 global $wgUser;
4371                 global $wgContLang;
4372
4373                 $fname = 'Parser::replaceLinkHolders';
4374                 wfProfileIn( $fname );
4375
4376                 $pdbks = array();
4377                 $colours = array();
4378                 $linkcolour_ids = array();
4379                 $sk = $this->mOptions->getSkin();
4380                 $linkCache =& LinkCache::singleton();
4381
4382                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4383                         wfProfileIn( $fname.'-check' );
4384                         $dbr = wfGetDB( DB_SLAVE );
4385                         $page = $dbr->tableName( 'page' );
4386                         $threshold = $wgUser->getOption('stubthreshold');
4387
4388                         # Sort by namespace
4389                         asort( $this->mLinkHolders['namespaces'] );
4390
4391                         # Generate query
4392                         $query = false;
4393                         $current = null;
4394                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4395                                 # Make title object
4396                                 $title = $this->mLinkHolders['titles'][$key];
4397
4398                                 # Skip invalid entries.
4399                                 # Result will be ugly, but prevents crash.
4400                                 if ( is_null( $title ) ) {
4401                                         continue;
4402                                 }
4403                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4404
4405                                 # Check if it's a static known link, e.g. interwiki
4406                                 if ( $title->isAlwaysKnown() ) {
4407                                         $colours[$pdbk] = '';
4408                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4409                                         $colours[$pdbk] = '';
4410                                         $this->mOutput->addLink( $title, $id );
4411                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4412                                         $colours[$pdbk] = 'new';
4413                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4414                                         $colours[$pdbk] = 'new';
4415                                 } else {
4416                                         # Not in the link cache, add it to the query
4417                                         if ( !isset( $current ) ) {
4418                                                 $current = $ns;
4419                                                 $query =  "SELECT page_id, page_namespace, page_title";
4420                                                 if ( $threshold > 0 ) {
4421                                                         $query .= ', page_len, page_is_redirect';
4422                                                 }
4423                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4424                                         } elseif ( $current != $ns ) {
4425                                                 $current = $ns;
4426                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4427                                         } else {
4428                                                 $query .= ', ';
4429                                         }
4430
4431                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4432                                 }
4433                         }
4434                         if ( $query ) {
4435                                 $query .= '))';
4436                                 if ( $options & RLH_FOR_UPDATE ) {
4437                                         $query .= ' FOR UPDATE';
4438                                 }
4439
4440                                 $res = $dbr->query( $query, $fname );
4441
4442                                 # Fetch data and form into an associative array
4443                                 # non-existent = broken
4444                                 while ( $s = $dbr->fetchObject($res) ) {
4445                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4446                                         $pdbk = $title->getPrefixedDBkey();
4447                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4448                                         $this->mOutput->addLink( $title, $s->page_id );
4449                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4450                                         //add id to the extension todolist
4451                                         $linkcolour_ids[$s->page_id] = $pdbk;
4452                                 }
4453                                 //pass an array of page_ids to an extension
4454                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4455                         }
4456                         wfProfileOut( $fname.'-check' );
4457
4458                         # Do a second query for different language variants of links and categories
4459                         if($wgContLang->hasVariants()){
4460                                 $linkBatch = new LinkBatch();
4461                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4462                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4463                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4464
4465                                 $categories = $this->mOutput->getCategoryLinks();
4466
4467                                 // Add variants of links to link batch
4468                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4469                                         $title = $this->mLinkHolders['titles'][$key];
4470                                         if ( is_null( $title ) )
4471                                                 continue;
4472
4473                                         $pdbk = $title->getPrefixedDBkey();
4474                                         $titleText = $title->getText();
4475
4476                                         // generate all variants of the link title text
4477                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4478
4479                                         // if link was not found (in first query), add all variants to query
4480                                         if ( !isset($colours[$pdbk]) ){
4481                                                 foreach($allTextVariants as $textVariant){
4482                                                         if($textVariant != $titleText){
4483                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4484                                                                 if(is_null($variantTitle)) continue;
4485                                                                 $linkBatch->addObj( $variantTitle );
4486                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4487                                                         }
4488                                                 }
4489                                         }
4490                                 }
4491
4492                                 // process categories, check if a category exists in some variant
4493                                 foreach( $categories as $category ){
4494                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4495                                         foreach($variants as $variant){
4496                                                 if($variant != $category){
4497                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4498                                                         if(is_null($variantTitle)) continue;
4499                                                         $linkBatch->addObj( $variantTitle );
4500                                                         $categoryMap[$variant] = $category;
4501                                                 }
4502                                         }
4503                                 }
4504
4505
4506                                 if(!$linkBatch->isEmpty()){
4507                                         // construct query
4508                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4509
4510                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4511                                         if ( $threshold > 0 ) {
4512                                                 $variantQuery .= ', page_len, page_is_redirect';
4513                                         }
4514
4515                                         $variantQuery .= " FROM $page WHERE $titleClause";
4516                                         if ( $options & RLH_FOR_UPDATE ) {
4517                                                 $variantQuery .= ' FOR UPDATE';
4518                                         }
4519
4520                                         $varRes = $dbr->query( $variantQuery, $fname );
4521
4522                                         // for each found variants, figure out link holders and replace
4523                                         while ( $s = $dbr->fetchObject($varRes) ) {
4524
4525                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4526                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4527                                                 $vardbk = $variantTitle->getDBkey();
4528
4529                                                 $holderKeys = array();
4530                                                 if(isset($variantMap[$varPdbk])){
4531                                                         $holderKeys = $variantMap[$varPdbk];
4532                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4533                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4534                                                 }
4535
4536                                                 // loop over link holders
4537                                                 foreach($holderKeys as $key){
4538                                                         $title = $this->mLinkHolders['titles'][$key];
4539                                                         if ( is_null( $title ) ) continue;
4540
4541                                                         $pdbk = $title->getPrefixedDBkey();
4542
4543                                                         if(!isset($colours[$pdbk])){
4544                                                                 // found link in some of the variants, replace the link holder data
4545                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4546                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4547
4548                                                                 // set pdbk and colour
4549                                                                 $pdbks[$key] = $varPdbk;
4550                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4551                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4552                                                         }
4553                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4554                                                 }
4555
4556                                                 // check if the object is a variant of a category
4557                                                 if(isset($categoryMap[$vardbk])){
4558                                                         $oldkey = $categoryMap[$vardbk];
4559                                                         if($oldkey != $vardbk)
4560                                                                 $varCategories[$oldkey]=$vardbk;
4561                                                 }
4562                                         }
4563
4564                                         // rebuild the categories in original order (if there are replacements)
4565                                         if(count($varCategories)>0){
4566                                                 $newCats = array();
4567                                                 $originalCats = $this->mOutput->getCategories();
4568                                                 foreach($originalCats as $cat => $sortkey){
4569                                                         // make the replacement
4570                                                         if( array_key_exists($cat,$varCategories) )
4571                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4572                                                         else $newCats[$cat] = $sortkey;
4573                                                 }
4574                                                 $this->mOutput->setCategoryLinks($newCats);
4575                                         }
4576                                 }
4577                         }
4578
4579                         # Construct search and replace arrays
4580                         wfProfileIn( $fname.'-construct' );
4581                         $replacePairs = array();
4582                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4583                                 $pdbk = $pdbks[$key];
4584                                 $searchkey = "<!--LINK $key-->";
4585                                 $title = $this->mLinkHolders['titles'][$key];
4586                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4587                                         $linkCache->addBadLinkObj( $title );
4588                                         $colours[$pdbk] = 'new';
4589                                         $this->mOutput->addLink( $title, 0 );
4590                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4591                                                                         $this->mLinkHolders['texts'][$key],
4592                                                                         $this->mLinkHolders['queries'][$key] );
4593                                 } else {
4594                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4595                                                                         $this->mLinkHolders['texts'][$key],
4596                                                                         $this->mLinkHolders['queries'][$key] );
4597                                 }
4598                         }
4599                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4600                         wfProfileOut( $fname.'-construct' );
4601
4602                         # Do the thing
4603                         wfProfileIn( $fname.'-replace' );
4604                         $text = preg_replace_callback(
4605                                 '/(<!--LINK .*?-->)/',
4606                                 $replacer->cb(),
4607                                 $text);
4608
4609                         wfProfileOut( $fname.'-replace' );
4610                 }
4611
4612                 # Now process interwiki link holders
4613                 # This is quite a bit simpler than internal links
4614                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4615                         wfProfileIn( $fname.'-interwiki' );
4616                         # Make interwiki link HTML
4617                         $replacePairs = array();
4618                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4619                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4620                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4621                         }
4622                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4623
4624                         $text = preg_replace_callback(
4625                                 '/<!--IWLINK (.*?)-->/',
4626                                 $replacer->cb(),
4627                                 $text );
4628                         wfProfileOut( $fname.'-interwiki' );
4629                 }
4630
4631                 wfProfileOut( $fname );
4632                 return $colours;
4633         }
4634
4635         /**
4636          * Replace <!--LINK--> link placeholders with plain text of links
4637          * (not HTML-formatted).
4638          * @param string $text
4639          * @return string
4640          */
4641         function replaceLinkHoldersText( $text ) {
4642                 $fname = 'Parser::replaceLinkHoldersText';
4643                 wfProfileIn( $fname );
4644
4645                 $text = preg_replace_callback(
4646                         '/<!--(LINK|IWLINK) (.*?)-->/',
4647                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4648                         $text );
4649
4650                 wfProfileOut( $fname );
4651                 return $text;
4652         }
4653
4654         /**
4655          * @param array $matches
4656          * @return string
4657          * @private
4658          */
4659         function replaceLinkHoldersTextCallback( $matches ) {
4660                 $type = $matches[1];
4661                 $key  = $matches[2];
4662                 if( $type == 'LINK' ) {
4663                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4664                                 return $this->mLinkHolders['texts'][$key];
4665                         }
4666                 } elseif( $type == 'IWLINK' ) {
4667                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4668                                 return $this->mInterwikiLinkHolders['texts'][$key];
4669                         }
4670                 }
4671                 return $matches[0];
4672         }
4673
4674         /**
4675          * Tag hook handler for 'pre'.
4676          */
4677         function renderPreTag( $text, $attribs ) {
4678                 // Backwards-compatibility hack
4679                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4680
4681                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4682                 return wfOpenElement( 'pre', $attribs ) .
4683                         Xml::escapeTagsOnly( $content ) .
4684                         '</pre>';
4685         }
4686
4687         /**
4688          * Renders an image gallery from a text with one line per image.
4689          * text labels may be given by using |-style alternative text. E.g.
4690          *   Image:one.jpg|The number "1"
4691          *   Image:tree.jpg|A tree
4692          * given as text will return the HTML of a gallery with two images,
4693          * labeled 'The number "1"' and
4694          * 'A tree'.
4695          */
4696         function renderImageGallery( $text, $params ) {
4697                 $ig = new ImageGallery();
4698                 $ig->setContextTitle( $this->mTitle );
4699                 $ig->setShowBytes( false );
4700                 $ig->setShowFilename( false );
4701                 $ig->setParser( $this );
4702                 $ig->setHideBadImages();
4703                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4704                 $ig->useSkin( $this->mOptions->getSkin() );
4705                 $ig->mRevisionId = $this->mRevisionId;
4706
4707                 if( isset( $params['caption'] ) ) {
4708                         $caption = $params['caption'];
4709                         $caption = htmlspecialchars( $caption );
4710                         $caption = $this->replaceInternalLinks( $caption );
4711                         $ig->setCaptionHtml( $caption );
4712                 }
4713                 if( isset( $params['perrow'] ) ) {
4714                         $ig->setPerRow( $params['perrow'] );
4715                 }
4716                 if( isset( $params['widths'] ) ) {
4717                         $ig->setWidths( $params['widths'] );
4718                 }
4719                 if( isset( $params['heights'] ) ) {
4720                         $ig->setHeights( $params['heights'] );
4721                 }
4722
4723                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4724
4725                 $lines = explode( "\n", $text );
4726                 foreach ( $lines as $line ) {
4727                         # match lines like these:
4728                         # Image:someimage.jpg|This is some image
4729                         $matches = array();
4730                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4731                         # Skip empty lines
4732                         if ( count( $matches ) == 0 ) {
4733                                 continue;
4734                         }
4735                         $tp = Title::newFromText( $matches[1] );
4736                         $nt =& $tp;
4737                         if( is_null( $nt ) ) {
4738                                 # Bogus title. Ignore these so we don't bomb out later.
4739                                 continue;
4740                         }
4741                         if ( isset( $matches[3] ) ) {
4742                                 $label = $matches[3];
4743                         } else {
4744                                 $label = '';
4745                         }
4746
4747                         $pout = $this->parse( $label,
4748                                 $this->mTitle,
4749                                 $this->mOptions,
4750                                 false, // Strip whitespace...?
4751                                 false  // Don't clear state!
4752                         );
4753                         $html = $pout->getText();
4754
4755                         $ig->add( $nt, $html );
4756
4757                         # Only add real images (bug #5586)
4758                         if ( $nt->getNamespace() == NS_IMAGE ) {
4759                                 $this->mOutput->addImage( $nt->getDBkey() );
4760                         }
4761                 }
4762                 return $ig->toHTML();
4763         }
4764
4765         function getImageParams( $handler ) {
4766                 if ( $handler ) {
4767                         $handlerClass = get_class( $handler );
4768                 } else {
4769                         $handlerClass = '';
4770                 }
4771                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4772                         // Initialise static lists
4773                         static $internalParamNames = array(
4774                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4775                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4776                                         'bottom', 'text-bottom' ),
4777                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4778                                         'upright', 'border' ),
4779                         );
4780                         static $internalParamMap;
4781                         if ( !$internalParamMap ) {
4782                                 $internalParamMap = array();
4783                                 foreach ( $internalParamNames as $type => $names ) {
4784                                         foreach ( $names as $name ) {
4785                                                 $magicName = str_replace( '-', '_', "img_$name" );
4786                                                 $internalParamMap[$magicName] = array( $type, $name );
4787                                         }
4788                                 }
4789                         }
4790
4791                         // Add handler params
4792                         $paramMap = $internalParamMap;
4793                         if ( $handler ) {
4794                                 $handlerParamMap = $handler->getParamMap();
4795                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4796                                         $paramMap[$magic] = array( 'handler', $paramName );
4797                                 }
4798                         }
4799                         $this->mImageParams[$handlerClass] = $paramMap;
4800                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4801                 }
4802                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4803         }
4804
4805         /**
4806          * Parse image options text and use it to make an image
4807          */
4808         function makeImage( $title, $options ) {
4809                 # @TODO: let the MediaHandler specify its transform parameters
4810                 #
4811                 # Check if the options text is of the form "options|alt text"
4812                 # Options are:
4813                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4814                 #  * left               no resizing, just left align. label is used for alt= only
4815                 #  * right              same, but right aligned
4816                 #  * none               same, but not aligned
4817                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4818                 #  * center             center the image
4819                 #  * framed             Keep original image size, no magnify-button.
4820                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4821                 #  * upright            reduce width for upright images, rounded to full __0 px
4822                 #  * border             draw a 1px border around the image
4823                 # vertical-align values (no % or length right now):
4824                 #  * baseline
4825                 #  * sub
4826                 #  * super
4827                 #  * top
4828                 #  * text-top
4829                 #  * middle
4830                 #  * bottom
4831                 #  * text-bottom
4832
4833                 $parts = array_map( 'trim', explode( '|', $options) );
4834                 $sk = $this->mOptions->getSkin();
4835
4836                 # Give extensions a chance to select the file revision for us
4837                 $skip = $time = false;
4838                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4839
4840                 if ( $skip ) {
4841                         return $sk->makeLinkObj( $title );
4842                 }
4843
4844                 # Get parameter map
4845                 $file = wfFindFile( $title, $time );
4846                 $handler = $file ? $file->getHandler() : false;
4847
4848                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4849
4850                 # Process the input parameters
4851                 $caption = '';
4852                 $params = array( 'frame' => array(), 'handler' => array(),
4853                         'horizAlign' => array(), 'vertAlign' => array() );
4854                 foreach( $parts as $part ) {
4855                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4856                         if ( isset( $paramMap[$magicName] ) ) {
4857                                 list( $type, $paramName ) = $paramMap[$magicName];
4858                                 $params[$type][$paramName] = $value;
4859
4860                                 // Special case; width and height come in one variable together
4861                                 if( $type == 'handler' && $paramName == 'width' ) {
4862                                         $m = array();
4863                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4864                                                 $params[$type]['width'] = intval( $m[1] );
4865                                                 $params[$type]['height'] = intval( $m[2] );
4866                                         } else {
4867                                                 $params[$type]['width'] = intval( $value );
4868                                         }
4869                                 }
4870                         } else {
4871                                 $caption = $part;
4872                         }
4873                 }
4874
4875                 # Process alignment parameters
4876                 if ( $params['horizAlign'] ) {
4877                         $params['frame']['align'] = key( $params['horizAlign'] );
4878                 }
4879                 if ( $params['vertAlign'] ) {
4880                         $params['frame']['valign'] = key( $params['vertAlign'] );
4881                 }
4882
4883                 # Validate the handler parameters
4884                 if ( $handler ) {
4885                         foreach ( $params['handler'] as $name => $value ) {
4886                                 if ( !$handler->validateParam( $name, $value ) ) {
4887                                         unset( $params['handler'][$name] );
4888                                 }
4889                         }
4890                 }
4891
4892                 # Strip bad stuff out of the alt text
4893                 $alt = $this->replaceLinkHoldersText( $caption );
4894
4895                 # make sure there are no placeholders in thumbnail attributes
4896                 # that are later expanded to html- so expand them now and
4897                 # remove the tags
4898                 $alt = $this->mStripState->unstripBoth( $alt );
4899                 $alt = Sanitizer::stripAllTags( $alt );
4900
4901                 $params['frame']['alt'] = $alt;
4902                 $params['frame']['caption'] = $caption;
4903
4904                 # Linker does the rest
4905                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4906
4907                 # Give the handler a chance to modify the parser object
4908                 if ( $handler ) {
4909                         $handler->parserTransformHook( $this, $file );
4910                 }
4911
4912                 return $ret;
4913         }
4914
4915         /**
4916          * Set a flag in the output object indicating that the content is dynamic and
4917          * shouldn't be cached.
4918          */
4919         function disableCache() {
4920                 wfDebug( "Parser output marked as uncacheable.\n" );
4921                 $this->mOutput->mCacheTime = -1;
4922         }
4923
4924         /**#@+
4925          * Callback from the Sanitizer for expanding items found in HTML attribute
4926          * values, so they can be safely tested and escaped.
4927          * @param string $text
4928          * @param PPFrame $frame
4929          * @return string
4930          * @private
4931          */
4932         function attributeStripCallback( &$text, $frame = false ) {
4933                 $text = $this->replaceVariables( $text, $frame );
4934                 $text = $this->mStripState->unstripBoth( $text );
4935                 return $text;
4936         }
4937
4938         /**#@-*/
4939
4940         /**#@+
4941          * Accessor/mutator
4942          */
4943         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4944         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4945         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4946         /**#@-*/
4947
4948         /**#@+
4949          * Accessor
4950          */
4951         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4952         /**#@-*/
4953
4954
4955         /**
4956          * Break wikitext input into sections, and either pull or replace
4957          * some particular section's text.
4958          *
4959          * External callers should use the getSection and replaceSection methods.
4960          *
4961          * @param $text Page wikitext
4962          * @param $section Numbered section. 0 pulls the text before the first
4963          *                 heading; other numbers will pull the given section
4964          *                 along with its lower-level subsections. If the section is
4965          *                 not found, $mode=get will return $newtext, and
4966          *                 $mode=replace will return $text.
4967          * @param $mode One of "get" or "replace"
4968          * @param $newText Replacement text for section data.
4969          * @return string for "get", the extracted section text.
4970          *                for "replace", the whole page with the section replaced.
4971          */
4972         private function extractSections( $text, $section, $mode, $newText='' ) {
4973                 global $wgTitle;
4974                 $this->clearState();
4975                 $this->mTitle = $wgTitle; // not generally used but removes an ugly failure mode
4976                 $this->mOptions = new ParserOptions;
4977                 $this->setOutputType( OT_WIKI );
4978                 $curIndex = 0;
4979                 $outText = '';
4980                 $frame = new PPFrame( $this );
4981
4982                 // Preprocess the text
4983                 $dom = $this->preprocessToDom( $text );
4984                 $root = $dom->documentElement;
4985
4986                 // <h> nodes indicate section breaks
4987                 // They can only occur at the top level, so we can find them by iterating the root's children
4988                 $node = $root->firstChild;
4989
4990                 // Find the target section
4991                 if ( $section == 0 ) {
4992                         // Section zero doesn't nest, level=big
4993                         $targetLevel = 1000;
4994                 } else {
4995                         while ( $node ) {
4996                                 if ( $node->nodeName == 'h' ) {
4997                                         if ( $curIndex + 1 == $section ) {
4998                                                 break;
4999                                         }
5000                                         $curIndex++;
5001                                 }
5002                                 if ( $mode == 'replace' ) {
5003                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5004                                 }
5005                                 $node = $node->nextSibling;
5006                         }
5007                         if ( $node ) {
5008                                 $targetLevel = $node->getAttribute( 'level' );
5009                         }
5010                 }
5011
5012                 if ( !$node ) {
5013                         // Not found
5014                         if ( $mode == 'get' ) {
5015                                 return $newText;
5016                         } else {
5017                                 return $text;
5018                         }
5019                 }
5020
5021                 // Find the end of the section, including nested sections
5022                 do {
5023                         if ( $node->nodeName == 'h' ) {
5024                                 $curIndex++;
5025                                 $curLevel = $node->getAttribute( 'level' );
5026                                 if ( $curIndex != $section && $curLevel <= $targetLevel ) {
5027                                         break;
5028                                 }
5029                         }
5030                         if ( $mode == 'get' ) {
5031                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5032                         }
5033                         $node = $node->nextSibling;
5034                 } while ( $node );
5035
5036                 // Write out the remainder (in replace mode only)
5037                 if ( $mode == 'replace' ) {
5038                         // Output the replacement text
5039                         // Add two newlines on -- trailing whitespace in $newText is conventionally
5040                         // stripped by the editor, so we need both newlines to restore the paragraph gap
5041                         $outText .= $newText . "\n\n";
5042                         while ( $node ) {
5043                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5044                                 $node = $node->nextSibling;
5045                         }
5046                 }
5047
5048                 if ( is_string( $outText ) ) {
5049                         // Re-insert stripped tags
5050                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
5051                 }
5052
5053                 return $outText;
5054         }
5055
5056         /**
5057          * This function returns the text of a section, specified by a number ($section).
5058          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5059          * the first section before any such heading (section 0).
5060          *
5061          * If a section contains subsections, these are also returned.
5062          *
5063          * @param $text String: text to look in
5064          * @param $section Integer: section number
5065          * @param $deftext: default to return if section is not found
5066          * @return string text of the requested section
5067          */
5068         public function getSection( $text, $section, $deftext='' ) {
5069                 return $this->extractSections( $text, $section, "get", $deftext );
5070         }
5071
5072         public function replaceSection( $oldtext, $section, $text ) {
5073                 return $this->extractSections( $oldtext, $section, "replace", $text );
5074         }
5075
5076         /**
5077          * Get the timestamp associated with the current revision, adjusted for
5078          * the default server-local timestamp
5079          */
5080         function getRevisionTimestamp() {
5081                 if ( is_null( $this->mRevisionTimestamp ) ) {
5082                         wfProfileIn( __METHOD__ );
5083                         global $wgContLang;
5084                         $dbr = wfGetDB( DB_SLAVE );
5085                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
5086                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
5087
5088                         // Normalize timestamp to internal MW format for timezone processing.
5089                         // This has the added side-effect of replacing a null value with
5090                         // the current time, which gives us more sensible behavior for
5091                         // previews.
5092                         $timestamp = wfTimestamp( TS_MW, $timestamp );
5093
5094                         // The cryptic '' timezone parameter tells to use the site-default
5095                         // timezone offset instead of the user settings.
5096                         //
5097                         // Since this value will be saved into the parser cache, served
5098                         // to other users, and potentially even used inside links and such,
5099                         // it needs to be consistent for all visitors.
5100                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5101
5102                         wfProfileOut( __METHOD__ );
5103                 }
5104                 return $this->mRevisionTimestamp;
5105         }
5106
5107         /**
5108          * Mutator for $mDefaultSort
5109          *
5110          * @param $sort New value
5111          */
5112         public function setDefaultSort( $sort ) {
5113                 $this->mDefaultSort = $sort;
5114         }
5115
5116         /**
5117          * Accessor for $mDefaultSort
5118          * Will use the title/prefixed title if none is set
5119          *
5120          * @return string
5121          */
5122         public function getDefaultSort() {
5123                 if( $this->mDefaultSort !== false ) {
5124                         return $this->mDefaultSort;
5125                 } else {
5126                         return $this->mTitle->getNamespace() == NS_CATEGORY
5127                                         ? $this->mTitle->getText()
5128                                         : $this->mTitle->getPrefixedText();
5129                 }
5130         }
5131
5132         /**
5133          * Try to guess the section anchor name based on a wikitext fragment
5134          * presumably extracted from a heading, for example "Header" from
5135          * "== Header ==".
5136          */
5137         public function guessSectionNameFromWikiText( $text ) {
5138                 # Strip out wikitext links(they break the anchor)
5139                 $text = $this->stripSectionName( $text );
5140                 $headline = Sanitizer::decodeCharReferences( $text );
5141                 # strip out HTML
5142                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5143                 $headline = trim( $headline );
5144                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5145                 $replacearray = array(
5146                         '%3A' => ':',
5147                         '%' => '.'
5148                 );
5149                 return str_replace(
5150                         array_keys( $replacearray ),
5151                         array_values( $replacearray ),
5152                         $sectionanchor );
5153         }
5154
5155         /**
5156          * Strips a text string of wikitext for use in a section anchor
5157          *
5158          * Accepts a text string and then removes all wikitext from the
5159          * string and leaves only the resultant text (i.e. the result of
5160          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5161          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5162          * to create valid section anchors by mimicing the output of the
5163          * parser when headings are parsed.
5164          *
5165          * @param $text string Text string to be stripped of wikitext
5166          * for use in a Section anchor
5167          * @return Filtered text string
5168          */
5169         public function stripSectionName( $text ) {
5170                 # Strip internal link markup
5171                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5172                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5173
5174                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5175                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5176                 # on how many empty links there are on the page - need to figure that out.
5177                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5178
5179                 # Parse wikitext quotes (italics & bold)
5180                 $text = $this->doQuotes($text);
5181
5182                 # Strip HTML tags
5183                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5184                 return $text;
5185         }
5186
5187         /**
5188          * strip/replaceVariables/unstrip for preprocessor regression testing
5189          */
5190         function srvus( $text ) {
5191                 $text = $this->replaceVariables( $text );
5192                 $text = $this->mStripState->unstripBoth( $text );
5193                 $text = Sanitizer::removeHTMLtags( $text );
5194                 return $text;
5195         }
5196 }
5197
5198 /**
5199  * @todo document, briefly.
5200  * @addtogroup Parser
5201  */
5202 class OnlyIncludeReplacer {
5203         var $output = '';
5204
5205         function replace( $matches ) {
5206                 if ( substr( $matches[1], -1 ) == "\n" ) {
5207                         $this->output .= substr( $matches[1], 0, -1 );
5208                 } else {
5209                         $this->output .= $matches[1];
5210                 }
5211         }
5212 }
5213
5214 /**
5215  * @todo document, briefly.
5216  * @addtogroup Parser
5217  */
5218 class StripState {
5219         var $general, $nowiki;
5220
5221         function __construct() {
5222                 $this->general = new ReplacementArray;
5223                 $this->nowiki = new ReplacementArray;
5224         }
5225
5226         function unstripGeneral( $text ) {
5227                 wfProfileIn( __METHOD__ );
5228                 do {
5229                         $oldText = $text;
5230                         $text = $this->general->replace( $text );
5231                 } while ( $text != $oldText );
5232                 wfProfileOut( __METHOD__ );
5233                 return $text;
5234         }
5235
5236         function unstripNoWiki( $text ) {
5237                 wfProfileIn( __METHOD__ );
5238                 do {
5239                         $oldText = $text;
5240                         $text = $this->nowiki->replace( $text );
5241                 } while ( $text != $oldText );
5242                 wfProfileOut( __METHOD__ );
5243                 return $text;
5244         }
5245
5246         function unstripBoth( $text ) {
5247                 wfProfileIn( __METHOD__ );
5248                 do {
5249                         $oldText = $text;
5250                         $text = $this->general->replace( $text );
5251                         $text = $this->nowiki->replace( $text );
5252                 } while ( $text != $oldText );
5253                 wfProfileOut( __METHOD__ );
5254                 return $text;
5255         }
5256 }
5257
5258 /**
5259  * An expansion frame, used as a context to expand the result of preprocessToDom()
5260  */
5261 class PPFrame {
5262         var $parser, $title;
5263         var $titleCache;
5264
5265         const NO_ARGS = 1;
5266         const NO_TEMPLATES = 2;
5267         const STRIP_COMMENTS = 4;
5268
5269         const RECOVER_ORIG = 3;
5270
5271         /**
5272          * Construct a new preprocessor frame.
5273          * @param Parser $parser The parent parser
5274          * @param Title $title The context title, or false if there isn't one
5275          */
5276         function __construct( $parser ) {
5277                 $this->parser = $parser;
5278                 $this->title = $parser->mTitle;
5279                 $this->titleCache = array( $this->title ? $this->title->getPrefixedDBkey() : false );
5280         }
5281
5282         /**
5283          * Create a new child frame
5284          * $args is optionally a DOMNodeList containing the template arguments
5285          */
5286         function newChild( $args = false, $title = false ) {
5287                 $assocArgs = array();
5288                 if ( $title === false ) {
5289                         $title = $this->title;
5290                 }
5291                 if ( $args !== false ) {
5292                         $xpath = false;
5293                         foreach ( $args as $arg ) {
5294                                 if ( !$xpath ) {
5295                                         $xpath = new DOMXPath( $arg->ownerDocument );
5296                                 }
5297
5298                                 $nameNodes = $xpath->query( 'name', $arg );
5299                                 if ( $nameNodes->item( 0 )->hasAttributes() ) {
5300                                         // Numbered parameter
5301                                         $name = $nameNodes->item( 0 )->attributes->getNamedItem( 'index' )->textContent;
5302                                 } else {
5303                                         // Named parameter
5304                                         $name = $this->expand( $nameNodes->item( 0 ), PPFrame::STRIP_COMMENTS );
5305                                 }
5306
5307                                 $value = $xpath->query( 'value', $arg );
5308                                 $assocArgs[$name] = $value->item( 0 );
5309                         }
5310                 }
5311                 return new PPTemplateFrame( $this->parser, $this, $assocArgs, $title );
5312         }
5313
5314         /**
5315          * Expand a DOMNode describing a preprocessed document into plain wikitext,
5316          * using the current context
5317          * @param $root the node
5318          */
5319         function expand( $root, $flags = 0 ) {
5320                 if ( is_string( $root ) ) {
5321                         return $root;
5322                 }
5323
5324                 if ( $this->parser->ot['html']
5325                         && ++$this->parser->mPPNodeCount > $this->parser->mOptions->mMaxPPNodeCount )
5326                 {
5327                         return $this->parser->insertStripItem( '<!-- node-count limit exceeded -->' );
5328                 }
5329
5330                 if ( is_array( $root ) ) {
5331                         $s = '';
5332                         foreach ( $root as $node ) {
5333                                 $s .= $this->expand( $node, $flags );
5334                         }
5335                 } elseif ( $root instanceof DOMNodeList ) {
5336                         $s = '';
5337                         foreach ( $root as $node ) {
5338                                 $s .= $this->expand( $node, $flags );
5339                         }
5340                 } elseif ( $root instanceof DOMNode ) {
5341                         if ( $root->nodeType == XML_TEXT_NODE ) {
5342                                 $s = $root->nodeValue;
5343                         } elseif ( $root->nodeName == 'template' ) {
5344                                 # Double-brace expansion
5345                                 $xpath = new DOMXPath( $root->ownerDocument );
5346                                 $titles = $xpath->query( 'title', $root );
5347                                 $title = $titles->item( 0 );
5348                                 $parts = $xpath->query( 'part', $root );
5349                                 if ( $flags & self::NO_TEMPLATES ) {
5350                                         $s = '{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}';
5351                                 } else {
5352                                         $lineStart = $root->getAttribute( 'lineStart' );
5353                                         $params = array(
5354                                                 'title' => $title,
5355                                                 'parts' => $parts,
5356                                                 'lineStart' => $lineStart,
5357                                                 'text' => 'FIXME' );
5358                                         $s = $this->parser->braceSubstitution( $params, $this );
5359                                 }
5360                         } elseif ( $root->nodeName == 'tplarg' ) {
5361                                 # Triple-brace expansion
5362                                 $xpath = new DOMXPath( $root->ownerDocument );
5363                                 $titles = $xpath->query( 'title', $root );
5364                                 $title = $titles->item( 0 );
5365                                 $parts = $xpath->query( 'part', $root );
5366                                 if ( $flags & self::NO_ARGS || $this->parser->ot['msg'] ) {
5367                                         $s = '{{{' . $this->implodeWithFlags( '|', $flags, $title, $parts ) . '}}}';
5368                                 } else {
5369                                         $params = array( 'title' => $title, 'parts' => $parts, 'text' => 'FIXME' );
5370                                         $s = $this->parser->argSubstitution( $params, $this );
5371                                 }
5372                         } elseif ( $root->nodeName == 'comment' ) {
5373                                 # HTML-style comment
5374                                 if ( $flags & self::STRIP_COMMENTS ) {
5375                                         $s = '';
5376                                 } else {
5377                                         $s = $root->textContent;
5378                                 }
5379                         } elseif ( $root->nodeName == 'ext' ) {
5380                                 # Extension tag
5381                                 $xpath = new DOMXPath( $root->ownerDocument );
5382                                 $names = $xpath->query( 'name', $root );
5383                                 $attrs = $xpath->query( 'attr', $root );
5384                                 $inners = $xpath->query( 'inner', $root );
5385                                 $closes = $xpath->query( 'close', $root );
5386                                 $params = array(
5387                                         'name' => $names->item( 0 ),
5388                                         'attr' => $attrs->length > 0 ? $attrs->item( 0 ) : null,
5389                                         'inner' => $inners->length > 0 ? $inners->item( 0 ) : null,
5390                                         'close' => $closes->length > 0 ? $closes->item( 0 ) : null,
5391                                 );
5392                                 $s = $this->parser->extensionSubstitution( $params, $this );
5393                         } elseif ( $root->nodeName == 'h' ) {
5394                                 # Heading
5395                                 $s = $this->expand( $root->childNodes, $flags );
5396
5397                                 if ( $this->parser->ot['html'] ) {
5398                                         # Insert heading index marker
5399                                         $headingIndex = $root->getAttribute( 'i' );
5400                                         $titleText = $this->title->getPrefixedDBkey();
5401                                         $this->parser->mHeadings[] = array( $titleText, $headingIndex );
5402                                         $serial = count( $this->parser->mHeadings ) - 1;
5403                                         $marker = "{$this->parser->mUniqPrefix}-h-$serial-{$this->parser->mMarkerSuffix}";
5404                                         $count = $root->getAttribute( 'level' );
5405                                         $s = substr( $s, 0, $count ) . $marker . substr( $s, $count );
5406                                         $this->parser->mStripState->general->setPair( $marker, '' );
5407                                 }
5408                         } else {
5409                                 # Generic recursive expansion
5410                                 $s = '';
5411                                 for ( $node = $root->firstChild; $node; $node = $node->nextSibling ) {
5412                                         if ( $node->nodeType == XML_TEXT_NODE ) {
5413                                                 $s .= $node->nodeValue;
5414                                         } elseif ( $node->nodeType == XML_ELEMENT_NODE ) {
5415                                                 $s .= $this->expand( $node, $flags );
5416                                         }
5417                                 }
5418                         }
5419                 } else {
5420                         throw new MWException( __METHOD__.': Invalid parameter type' );
5421                 }
5422                 return $s;
5423         }
5424
5425         function implodeWithFlags( $sep, $flags /*, ... */ ) {
5426                 $args = array_slice( func_get_args(), 2 );
5427
5428                 $first = true;
5429                 $s = '';
5430                 foreach ( $args as $root ) {
5431                         if ( !is_array( $root ) && !( $root instanceof DOMNodeList ) ) {
5432                                 $root = array( $root );
5433                         }
5434                         foreach ( $root as $node ) {
5435                                 if ( $first ) {
5436                                         $first = false;
5437                                 } else {
5438                                         $s .= $sep;
5439                                 }
5440                                 $s .= $this->expand( $node, $flags );
5441                         }
5442                 }
5443                 return $s;
5444         }
5445
5446         function implode( $sep /*, ... */ ) {
5447                 $args = func_get_args();
5448                 $args = array_merge( array_slice( $args, 0, 1 ), array( 0 ), array_slice( $args, 1 ) );
5449                 return call_user_func_array( array( $this, 'implodeWithFlags' ), $args );
5450         }
5451
5452         /**
5453          * Split an <arg> or <template> node into a three-element array:
5454          *    DOMNode name, string index and DOMNode value
5455          */
5456         function splitBraceNode( $node ) {
5457                 $xpath = new DOMXPath( $node->ownerDocument );
5458                 $names = $xpath->query( 'name', $node );
5459                 $values = $xpath->query( 'value', $node );
5460                 if ( !$names->length || !$values->length ) {
5461                         throw new MWException( 'Invalid brace node passed to ' . __METHOD__ );
5462                 }
5463                 $name = $names->item( 0 );
5464                 $index = $name->getAttribute( 'index' );
5465                 return array( $name, $index, $values->item( 0 ) );
5466         }
5467
5468         function __toString() {
5469                 return 'frame{}';
5470         }
5471
5472         function getPDBK( $level = false ) {
5473                 if ( $level === false ) {
5474                         return $this->title->getPrefixedDBkey();
5475                 } else {
5476                         return isset( $this->titleCache[$level] ) ? $this->titleCache[$level] : false;
5477                 }
5478         }
5479 }
5480
5481 /**
5482  * Expansion frame with template arguments
5483  */
5484 class PPTemplateFrame extends PPFrame {
5485         var $parser, $args, $parent;
5486         var $titleCache;
5487
5488         function __construct( $parser, $parent = false, $args = array(), $title = false ) {
5489                 $this->parser = $parser;
5490                 $this->parent = $parent;
5491                 $this->args = $args;
5492                 $this->title = $title;
5493                 $this->titleCache = $parent->titleCache;
5494                 $this->titleCache[] = $title ? $title->getPrefixedDBkey() : false;
5495         }
5496
5497         function __toString() {
5498                 $s = 'tplframe{';
5499                 $first = true;
5500                 foreach ( $this->args as $name => $value ) {
5501                         if ( $first ) {
5502                                 $first = false;
5503                         } else {
5504                                 $s .= ', ';
5505                         }
5506                         $s .= "\"$name\":\"" .
5507                                 str_replace( '"', '\\"', $value->ownerDocument->saveXML( $value ) ) . '"';
5508                 }
5509                 $s .= '}';
5510                 return $s;
5511         }
5512 }
5513