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 five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @addtogroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         /**#@+
  87          * @private
  88          */
  89         # Persistent:
  90         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  91                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix,
  92                 $mExtLinkBracketedRegex, $mPreprocessor, $mDefaultStripList, $mVarCache, $mConf;
  93
  94
  95         # Cleared with clearState():
  96         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  97         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  98         var $mInterwikiLinkHolders, $mLinkHolders;
  99         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 100         var $mTplExpandCache; // empty-frame expansion cache
 101         var $mTplRedirCache, $mTplDomCache, $mHeadings;
 102
 103         # Temporary
 104         # These are variables reset at least once per parse regardless of $clearState
 105         var $mOptions,      // ParserOptions object
 106                 $mTitle,        // Title context, used for self-link rendering and similar things
 107                 $mOutputType,   // Output type, one of the OT_xxx constants
 108                 $ot,            // Shortcut alias, see setOutputType()
 109                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 110                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 111                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 112
 113         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @public
 119          */
 120         function __construct( $conf = array() ) {
 121                 $this->mConf = $conf;
 122                 $this->mTagHooks = array();
 123                 $this->mTransparentTagHooks = array();
 124                 $this->mFunctionHooks = array();
 125                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 126                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 127                 $this->mMarkerSuffix = "-QINU\x7f";
 128                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 129                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 130                 $this->mVarCache = array();
 131                 if ( isset( $conf['preprocessorClass'] ) ) {
 132                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 133                 } else {
 134                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 135                 }
 136                 $this->mFirstCall = true;
 137         }
 138
 139         /**
 140          * Do various kinds of initialisation on the first call of the parser
 141          */
 142         function firstCallInit() {
 143                 if ( !$this->mFirstCall ) {
 144                         return;
 145                 }
 146                 $this->mFirstCall = false;
 147
 148                 wfProfileIn( __METHOD__ );
 149                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 150
 151                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 152
 153                 # Syntax for arguments (see self::setFunctionHook):
 154                 #  "name for lookup in localized magic words array",
 155                 #  function callback,
 156                 #  optional SFH_NO_HASH to omit the hash from calls (e.g. {{int:...}
 157                 #    instead of {{#int:...}})
 158                 $this->setFunctionHook( 'int',              array( 'CoreParserFunctions', 'intFunction'      ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'ns',               array( 'CoreParserFunctions', 'ns'               ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'urlencode',        array( 'CoreParserFunctions', 'urlencode'        ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'lcfirst',          array( 'CoreParserFunctions', 'lcfirst'          ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'ucfirst',          array( 'CoreParserFunctions', 'ucfirst'          ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'lc',               array( 'CoreParserFunctions', 'lc'               ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'uc',               array( 'CoreParserFunctions', 'uc'               ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'localurl',         array( 'CoreParserFunctions', 'localurl'         ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'localurle',        array( 'CoreParserFunctions', 'localurle'        ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'fullurl',          array( 'CoreParserFunctions', 'fullurl'          ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'fullurle',         array( 'CoreParserFunctions', 'fullurle'         ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'formatnum',        array( 'CoreParserFunctions', 'formatnum'        ), SFH_NO_HASH );
 170                 $this->setFunctionHook( 'grammar',          array( 'CoreParserFunctions', 'grammar'          ), SFH_NO_HASH );
 171                 $this->setFunctionHook( 'plural',           array( 'CoreParserFunctions', 'plural'           ), SFH_NO_HASH );
 172                 $this->setFunctionHook( 'numberofpages',    array( 'CoreParserFunctions', 'numberofpages'    ), SFH_NO_HASH );
 173                 $this->setFunctionHook( 'numberofusers',    array( 'CoreParserFunctions', 'numberofusers'    ), SFH_NO_HASH );
 174                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 175                 $this->setFunctionHook( 'numberoffiles',    array( 'CoreParserFunctions', 'numberoffiles'    ), SFH_NO_HASH );
 176                 $this->setFunctionHook( 'numberofadmins',   array( 'CoreParserFunctions', 'numberofadmins'   ), SFH_NO_HASH );
 177                 $this->setFunctionHook( 'numberofedits',    array( 'CoreParserFunctions', 'numberofedits'    ), SFH_NO_HASH );
 178                 $this->setFunctionHook( 'language',         array( 'CoreParserFunctions', 'language'         ), SFH_NO_HASH );
 179                 $this->setFunctionHook( 'padleft',          array( 'CoreParserFunctions', 'padleft'          ), SFH_NO_HASH );
 180                 $this->setFunctionHook( 'padright',         array( 'CoreParserFunctions', 'padright'         ), SFH_NO_HASH );
 181                 $this->setFunctionHook( 'anchorencode',     array( 'CoreParserFunctions', 'anchorencode'     ), SFH_NO_HASH );
 182                 $this->setFunctionHook( 'special',          array( 'CoreParserFunctions', 'special'          ) );
 183                 $this->setFunctionHook( 'defaultsort',      array( 'CoreParserFunctions', 'defaultsort'      ), SFH_NO_HASH );
 184                 $this->setFunctionHook( 'filepath',         array( 'CoreParserFunctions', 'filepath'         ), SFH_NO_HASH );
 185                 $this->setFunctionHook( 'tag',              array( 'CoreParserFunctions', 'tagObj'           ), SFH_OBJECT_ARGS );
 186
 187                 if ( $wgAllowDisplayTitle ) {
 188                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 189                 }
 190                 if ( $wgAllowSlowParserFunctions ) {
 191                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 192                 }
 193
 194                 $this->initialiseVariables();
 195
 196                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 197                 wfProfileOut( __METHOD__ );
 198         }
 199
 200         /**
 201          * Clear Parser state
 202          *
 203          * @private
 204          */
 205         function clearState() {
 206                 wfProfileIn( __METHOD__ );
 207                 if ( $this->mFirstCall ) {
 208                         $this->firstCallInit();
 209                 }
 210                 $this->mOutput = new ParserOutput;
 211                 $this->mAutonumber = 0;
 212                 $this->mLastSection = '';
 213                 $this->mDTopen = false;
 214                 $this->mIncludeCount = array();
 215                 $this->mStripState = new StripState;
 216                 $this->mArgStack = false;
 217                 $this->mInPre = false;
 218                 $this->mInterwikiLinkHolders = array(
 219                         'texts' => array(),
 220                         'titles' => array()
 221                 );
 222                 $this->mLinkHolders = array(
 223                         'namespaces' => array(),
 224                         'dbkeys' => array(),
 225                         'queries' => array(),
 226                         'texts' => array(),
 227                         'titles' => array()
 228                 );
 229                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 230
 231                 /**
 232                  * Prefix for temporary replacement strings for the multipass parser.
 233                  * \x07 should never appear in input as it's disallowed in XML.
 234                  * Using it at the front also gives us a little extra robustness
 235                  * since it shouldn't match when butted up against identifier-like
 236                  * string constructs.
 237                  *
 238                  * Must not consist of all title characters, or else it will change
 239                  * the behaviour of <nowiki> in a link.
 240                  */
 241                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 242                 # Changed to \x7f to allow XML double-parsing -- TS
 243                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 244
 245                 # Clear these on every parse, bug 4549
 246                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 247
 248                 $this->mShowToc = true;
 249                 $this->mForceTocPosition = false;
 250                 $this->mIncludeSizes = array(
 251                         'post-expand' => 0,
 252                         'arg' => 0,
 253                 );
 254                 $this->mPPNodeCount = 0;
 255                 $this->mDefaultSort = false;
 256                 $this->mHeadings = array();
 257
 258                 # Fix cloning
 259                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 260                         $this->mPreprocessor = null;
 261                 }
 262
 263                 wfRunHooks( 'ParserClearState', array( &$this ) );
 264                 wfProfileOut( __METHOD__ );
 265         }
 266
 267         function setOutputType( $ot ) {
 268                 $this->mOutputType = $ot;
 269                 // Shortcut alias
 270                 $this->ot = array(
 271                         'html' => $ot == self::OT_HTML,
 272                         'wiki' => $ot == self::OT_WIKI,
 273                         'pre' => $ot == self::OT_PREPROCESS,
 274                 );
 275         }
 276
 277         /**
 278          * Set the context title
 279          */
 280         function setTitle( $t ) {
 281                 if ( !$t || $t instanceof FakeTitle ) {
 282                         $t = Title::newFromText( 'NO TITLE' );
 283                 }
 284                 if ( strval( $t->getFragment() ) !== '' ) {
 285                         # Strip the fragment to avoid various odd effects
 286                         $this->mTitle = clone $t;
 287                         $this->mTitle->setFragment( '' );
 288                 } else {
 289                         $this->mTitle = $t;
 290                 }
 291         }
 292
 293         /**
 294          * Accessor for mUniqPrefix.
 295          *
 296          * @public
 297          */
 298         function uniqPrefix() {
 299                 if( !isset( $this->mUniqPrefix ) ) {
 300                         // @fixme this is probably *horribly wrong*
 301                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 302                         // if this is called for a parser cache hit, the parser may not
 303                         // have ever been initialized in the first place.
 304                         // Not really sure what the heck is supposed to be going on here.
 305                         return '';
 306                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 307                 }
 308                 return $this->mUniqPrefix;
 309         }
 310
 311         /**
 312          * Convert wikitext to HTML
 313          * Do not call this function recursively.
 314          *
 315          * @param string $text Text we want to parse
 316          * @param Title &$title A title object
 317          * @param array $options
 318          * @param boolean $linestart
 319          * @param boolean $clearState
 320          * @param int $revid number to pass in {{REVISIONID}}
 321          * @return ParserOutput a ParserOutput
 322          */
 323         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 324                 /**
 325                  * First pass--just handle <nowiki> sections, pass the rest off
 326                  * to internalParse() which does all the real work.
 327                  */
 328
 329                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 330                 $fname = 'Parser::parse-' . wfGetCaller();
 331                 wfProfileIn( __METHOD__ );
 332                 wfProfileIn( $fname );
 333
 334                 if ( $clearState ) {
 335                         $this->clearState();
 336                 }
 337
 338                 $this->mOptions = $options;
 339                 $this->setTitle( $title );
 340                 $oldRevisionId = $this->mRevisionId;
 341                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 342                 if( $revid !== null ) {
 343                         $this->mRevisionId = $revid;
 344                         $this->mRevisionTimestamp = null;
 345                 }
 346                 $this->setOutputType( self::OT_HTML );
 347                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 348                 # No more strip!
 349                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 350                 $text = $this->internalParse( $text );
 351                 $text = $this->mStripState->unstripGeneral( $text );
 352
 353                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 354                 $fixtags = array(
 355                         # french spaces, last one Guillemet-left
 356                         # only if there is something before the space
 357                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 358                         # french spaces, Guillemet-right
 359                         '/(\\302\\253) /' => '\\1&nbsp;',
 360                 );
 361                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 362
 363                 # only once and last
 364                 $text = $this->doBlockLevels( $text, $linestart );
 365
 366                 $this->replaceLinkHolders( $text );
 367
 368                 # the position of the parserConvert() call should not be changed. it
 369                 # assumes that the links are all replaced and the only thing left
 370                 # is the <nowiki> mark.
 371                 # Side-effects: this calls $this->mOutput->setTitleText()
 372                 $text = $wgContLang->parserConvert( $text, $this );
 373
 374                 $text = $this->mStripState->unstripNoWiki( $text );
 375
 376                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 377
 378 //!JF Move to its own function
 379
 380                 $uniq_prefix = $this->mUniqPrefix;
 381                 $matches = array();
 382                 $elements = array_keys( $this->mTransparentTagHooks );
 383                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 384
 385                 foreach( $matches as $marker => $data ) {
 386                         list( $element, $content, $params, $tag ) = $data;
 387                         $tagName = strtolower( $element );
 388                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 389                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 390                                         array( $content, $params, $this ) );
 391                         } else {
 392                                 $output = $tag;
 393                         }
 394                         $this->mStripState->general->setPair( $marker, $output );
 395                 }
 396                 $text = $this->mStripState->unstripGeneral( $text );
 397
 398                 $text = Sanitizer::normalizeCharReferences( $text );
 399
 400                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 401                         $text = Parser::tidy($text);
 402                 } else {
 403                         # attempt to sanitize at least some nesting problems
 404                         # (bug #2702 and quite a few others)
 405                         $tidyregs = array(
 406                                 # ''Something [http://www.cool.com cool''] -->
 407                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 408                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 409                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 410                                 # fix up an anchor inside another anchor, only
 411                                 # at least for a single single nested link (bug 3695)
 412                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 413                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 414                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 415                                 # contains a div, so fix it up here; replace
 416                                 # div with escaped text
 417                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 418                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 419                                 # remove empty italic or bold tag pairs, some
 420                                 # introduced by rules above
 421                                 '/<([bi])><\/\\1>/' => '',
 422                         );
 423
 424                         $text = preg_replace(
 425                                 array_keys( $tidyregs ),
 426                                 array_values( $tidyregs ),
 427                                 $text );
 428                 }
 429
 430                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 431
 432                 # Information on include size limits, for the benefit of users who try to skirt them
 433                 if ( $this->mOptions->getEnableLimitReport() ) {
 434                         $max = $this->mOptions->getMaxIncludeSize();
 435                         $limitReport =
 436                                 "NewPP limit report\n" .
 437                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 438                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 439                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 440                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 441                         $text .= "\n<!-- \n$limitReport-->\n";
 442                 }
 443                 $this->mOutput->setText( $text );
 444                 $this->mRevisionId = $oldRevisionId;
 445                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 446                 wfProfileOut( $fname );
 447                 wfProfileOut( __METHOD__ );
 448
 449                 return $this->mOutput;
 450         }
 451
 452         /**
 453          * Recursive parser entry point that can be called from an extension tag
 454          * hook.
 455          */
 456         function recursiveTagParse( $text ) {
 457                 wfProfileIn( __METHOD__ );
 458                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 459                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 460                 $text = $this->internalParse( $text );
 461                 wfProfileOut( __METHOD__ );
 462                 return $text;
 463         }
 464
 465         /**
 466          * Expand templates and variables in the text, producing valid, static wikitext.
 467          * Also removes comments.
 468          */
 469         function preprocess( $text, $title, $options, $revid = null ) {
 470                 wfProfileIn( __METHOD__ );
 471                 $this->clearState();
 472                 $this->setOutputType( self::OT_PREPROCESS );
 473                 $this->mOptions = $options;
 474                 $this->setTitle( $title );
 475                 if( $revid !== null ) {
 476                         $this->mRevisionId = $revid;
 477                 }
 478                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 479                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 480                 $text = $this->replaceVariables( $text );
 481                 $text = $this->mStripState->unstripBoth( $text );
 482                 wfProfileOut( __METHOD__ );
 483                 return $text;
 484         }
 485
 486         /**
 487          * Get a random string
 488          *
 489          * @private
 490          * @static
 491          */
 492         function getRandomString() {
 493                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 494         }
 495
 496         function &getTitle() { return $this->mTitle; }
 497         function getOptions() { return $this->mOptions; }
 498
 499         function getFunctionLang() {
 500                 global $wgLang, $wgContLang;
 501                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 502         }
 503
 504         /**
 505          * Get a preprocessor object
 506          */
 507         function getPreprocessor() {
 508                 if ( !isset( $this->mPreprocessor ) ) {
 509                         $class = $this->mPreprocessorClass;
 510                         $this->mPreprocessor = new $class( $this );
 511                 }
 512                 return $this->mPreprocessor;
 513         }
 514
 515         /**
 516          * Replaces all occurrences of HTML-style comments and the given tags
 517          * in the text with a random marker and returns teh next text. The output
 518          * parameter $matches will be an associative array filled with data in
 519          * the form:
 520          *   'UNIQ-xxxxx' => array(
 521          *     'element',
 522          *     'tag content',
 523          *     array( 'param' => 'x' ),
 524          *     '<element param="x">tag content</element>' ) )
 525          *
 526          * @param $elements list of element names. Comments are always extracted.
 527          * @param $text Source text string.
 528          * @param $uniq_prefix
 529          *
 530          * @public
 531          * @static
 532          */
 533         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 534                 static $n = 1;
 535                 $stripped = '';
 536                 $matches = array();
 537
 538                 $taglist = implode( '|', $elements );
 539                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 540
 541                 while ( '' != $text ) {
 542                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 543                         $stripped .= $p[0];
 544                         if( count( $p ) < 5 ) {
 545                                 break;
 546                         }
 547                         if( count( $p ) > 5 ) {
 548                                 // comment
 549                                 $element    = $p[4];
 550                                 $attributes = '';
 551                                 $close      = '';
 552                                 $inside     = $p[5];
 553                         } else {
 554                                 // tag
 555                                 $element    = $p[1];
 556                                 $attributes = $p[2];
 557                                 $close      = $p[3];
 558                                 $inside     = $p[4];
 559                         }
 560
 561                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 562                         $stripped .= $marker;
 563
 564                         if ( $close === '/>' ) {
 565                                 // Empty element tag, <tag />
 566                                 $content = null;
 567                                 $text = $inside;
 568                                 $tail = null;
 569                         } else {
 570                                 if( $element == '!--' ) {
 571                                         $end = '/(-->)/';
 572                                 } else {
 573                                         $end = "/(<\\/$element\\s*>)/i";
 574                                 }
 575                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 576                                 $content = $q[0];
 577                                 if( count( $q ) < 3 ) {
 578                                         # No end tag -- let it run out to the end of the text.
 579                                         $tail = '';
 580                                         $text = '';
 581                                 } else {
 582                                         $tail = $q[1];
 583                                         $text = $q[2];
 584                                 }
 585                         }
 586
 587                         $matches[$marker] = array( $element,
 588                                 $content,
 589                                 Sanitizer::decodeTagAttributes( $attributes ),
 590                                 "<$element$attributes$close$content$tail" );
 591                 }
 592                 return $stripped;
 593         }
 594
 595         /**
 596          * Get a list of strippable XML-like elements
 597          */
 598         function getStripList() {
 599                 global $wgRawHtml;
 600                 $elements = $this->mStripList;
 601                 if( $wgRawHtml ) {
 602                         $elements[] = 'html';
 603                 }
 604                 if( $this->mOptions->getUseTeX() ) {
 605                         $elements[] = 'math';
 606                 }
 607                 return $elements;
 608         }
 609
 610         /**
 611          * @deprecated use replaceVariables
 612          */
 613         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 614                 return $text;
 615         }
 616
 617         /**
 618          * Restores pre, math, and other extensions removed by strip()
 619          *
 620          * always call unstripNoWiki() after this one
 621          * @private
 622          * @deprecated use $this->mStripState->unstrip()
 623          */
 624         function unstrip( $text, $state ) {
 625                 return $state->unstripGeneral( $text );
 626         }
 627
 628         /**
 629          * Always call this after unstrip() to preserve the order
 630          *
 631          * @private
 632          * @deprecated use $this->mStripState->unstrip()
 633          */
 634         function unstripNoWiki( $text, $state ) {
 635                 return $state->unstripNoWiki( $text );
 636         }
 637
 638         /**
 639          * @deprecated use $this->mStripState->unstripBoth()
 640          */
 641         function unstripForHTML( $text ) {
 642                 return $this->mStripState->unstripBoth( $text );
 643         }
 644
 645         /**
 646          * Add an item to the strip state
 647          * Returns the unique tag which must be inserted into the stripped text
 648          * The tag will be replaced with the original text in unstrip()
 649          *
 650          * @private
 651          */
 652         function insertStripItem( $text ) {
 653                 static $n = 0;
 654                 $rnd = "{$this->mUniqPrefix}-item-$n-{$this->mMarkerSuffix}";
 655                 ++$n;
 656                 $this->mStripState->general->setPair( $rnd, $text );
 657                 return $rnd;
 658         }
 659
 660         /**
 661          * Interface with html tidy, used if $wgUseTidy = true.
 662          * If tidy isn't able to correct the markup, the original will be
 663          * returned in all its glory with a warning comment appended.
 664          *
 665          * Either the external tidy program or the in-process tidy extension
 666          * will be used depending on availability. Override the default
 667          * $wgTidyInternal setting to disable the internal if it's not working.
 668          *
 669          * @param string $text Hideous HTML input
 670          * @return string Corrected HTML output
 671          * @public
 672          * @static
 673          */
 674         function tidy( $text ) {
 675                 global $wgTidyInternal;
 676                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 677 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 678 '<head><title>test</title></head><body>'.$text.'</body></html>';
 679                 if( $wgTidyInternal ) {
 680                         $correctedtext = Parser::internalTidy( $wrappedtext );
 681                 } else {
 682                         $correctedtext = Parser::externalTidy( $wrappedtext );
 683                 }
 684                 if( is_null( $correctedtext ) ) {
 685                         wfDebug( "Tidy error detected!\n" );
 686                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 687                 }
 688                 return $correctedtext;
 689         }
 690
 691         /**
 692          * Spawn an external HTML tidy process and get corrected markup back from it.
 693          *
 694          * @private
 695          * @static
 696          */
 697         function externalTidy( $text ) {
 698                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 699                 $fname = 'Parser::externalTidy';
 700                 wfProfileIn( $fname );
 701
 702                 $cleansource = '';
 703                 $opts = ' -utf8';
 704
 705                 $descriptorspec = array(
 706                         0 => array('pipe', 'r'),
 707                         1 => array('pipe', 'w'),
 708                         2 => array('file', wfGetNull(), 'a')
 709                 );
 710                 $pipes = array();
 711                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 712                 if (is_resource($process)) {
 713                         // Theoretically, this style of communication could cause a deadlock
 714                         // here. If the stdout buffer fills up, then writes to stdin could
 715                         // block. This doesn't appear to happen with tidy, because tidy only
 716                         // writes to stdout after it's finished reading from stdin. Search
 717                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 718                         fwrite($pipes[0], $text);
 719                         fclose($pipes[0]);
 720                         while (!feof($pipes[1])) {
 721                                 $cleansource .= fgets($pipes[1], 1024);
 722                         }
 723                         fclose($pipes[1]);
 724                         proc_close($process);
 725                 }
 726
 727                 wfProfileOut( $fname );
 728
 729                 if( $cleansource == '' && $text != '') {
 730                         // Some kind of error happened, so we couldn't get the corrected text.
 731                         // Just give up; we'll use the source text and append a warning.
 732                         return null;
 733                 } else {
 734                         return $cleansource;
 735                 }
 736         }
 737
 738         /**
 739          * Use the HTML tidy PECL extension to use the tidy library in-process,
 740          * saving the overhead of spawning a new process.
 741          *
 742          * 'pear install tidy' should be able to compile the extension module.
 743          *
 744          * @private
 745          * @static
 746          */
 747         function internalTidy( $text ) {
 748                 global $wgTidyConf, $IP, $wgDebugTidy;
 749                 $fname = 'Parser::internalTidy';
 750                 wfProfileIn( $fname );
 751
 752                 $tidy = new tidy;
 753                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 754                 $tidy->cleanRepair();
 755                 if( $tidy->getStatus() == 2 ) {
 756                         // 2 is magic number for fatal error
 757                         // http://www.php.net/manual/en/function.tidy-get-status.php
 758                         $cleansource = null;
 759                 } else {
 760                         $cleansource = tidy_get_output( $tidy );
 761                 }
 762                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 763                         $cleansource .= "<!--\nTidy reports:\n" .
 764                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 765                                 "\n-->";
 766                 }
 767
 768                 wfProfileOut( $fname );
 769                 return $cleansource;
 770         }
 771
 772         /**
 773          * parse the wiki syntax used to render tables
 774          *
 775          * @private
 776          */
 777         function doTableStuff ( $text ) {
 778                 $fname = 'Parser::doTableStuff';
 779                 wfProfileIn( $fname );
 780
 781                 $lines = explode ( "\n" , $text );
 782                 $td_history = array (); // Is currently a td tag open?
 783                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 784                 $tr_history = array (); // Is currently a tr tag open?
 785                 $tr_attributes = array (); // history of tr attributes
 786                 $has_opened_tr = array(); // Did this table open a <tr> element?
 787                 $indent_level = 0; // indent level of the table
 788                 foreach ( $lines as $key => $line )
 789                 {
 790                         $line = trim ( $line );
 791
 792                         if( $line == '' ) { // empty line, go to next line
 793                                 continue;
 794                         }
 795                         $first_character = $line{0};
 796                         $matches = array();
 797
 798                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 799                                 // First check if we are starting a new table
 800                                 $indent_level = strlen( $matches[1] );
 801
 802                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 803                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 804
 805                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 806                                 array_push ( $td_history , false );
 807                                 array_push ( $last_tag_history , '' );
 808                                 array_push ( $tr_history , false );
 809                                 array_push ( $tr_attributes , '' );
 810                                 array_push ( $has_opened_tr , false );
 811                         } else if ( count ( $td_history ) == 0 ) {
 812                                 // Don't do any of the following
 813                                 continue;
 814                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 815                                 // We are ending a table
 816                                 $line = '</table>' . substr ( $line , 2 );
 817                                 $last_tag = array_pop ( $last_tag_history );
 818
 819                                 if ( !array_pop ( $has_opened_tr ) ) {
 820                                         $line = "<tr><td></td></tr>{$line}";
 821                                 }
 822
 823                                 if ( array_pop ( $tr_history ) ) {
 824                                         $line = "</tr>{$line}";
 825                                 }
 826
 827                                 if ( array_pop ( $td_history ) ) {
 828                                         $line = "</{$last_tag}>{$line}";
 829                                 }
 830                                 array_pop ( $tr_attributes );
 831                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 832                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 833                                 // Now we have a table row
 834                                 $line = preg_replace( '#^\|-+#', '', $line );
 835
 836                                 // Whats after the tag is now only attributes
 837                                 $attributes = $this->mStripState->unstripBoth( $line );
 838                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 839                                 array_pop ( $tr_attributes );
 840                                 array_push ( $tr_attributes , $attributes );
 841
 842                                 $line = '';
 843                                 $last_tag = array_pop ( $last_tag_history );
 844                                 array_pop ( $has_opened_tr );
 845                                 array_push ( $has_opened_tr , true );
 846
 847                                 if ( array_pop ( $tr_history ) ) {
 848                                         $line = '</tr>';
 849                                 }
 850
 851                                 if ( array_pop ( $td_history ) ) {
 852                                         $line = "</{$last_tag}>{$line}";
 853                                 }
 854
 855                                 $lines[$key] = $line;
 856                                 array_push ( $tr_history , false );
 857                                 array_push ( $td_history , false );
 858                                 array_push ( $last_tag_history , '' );
 859                         }
 860                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 861                                 // This might be cell elements, td, th or captions
 862                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 863                                         $first_character = '+';
 864                                         $line = substr ( $line , 1 );
 865                                 }
 866
 867                                 $line = substr ( $line , 1 );
 868
 869                                 if ( $first_character == '!' ) {
 870                                         $line = str_replace ( '!!' , '||' , $line );
 871                                 }
 872
 873                                 // Split up multiple cells on the same line.
 874                                 // FIXME : This can result in improper nesting of tags processed
 875                                 // by earlier parser steps, but should avoid splitting up eg
 876                                 // attribute values containing literal "||".
 877                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 878
 879                                 $lines[$key] = '';
 880
 881                                 // Loop through each table cell
 882                                 foreach ( $cells as $cell )
 883                                 {
 884                                         $previous = '';
 885                                         if ( $first_character != '+' )
 886                                         {
 887                                                 $tr_after = array_pop ( $tr_attributes );
 888                                                 if ( !array_pop ( $tr_history ) ) {
 889                                                         $previous = "<tr{$tr_after}>\n";
 890                                                 }
 891                                                 array_push ( $tr_history , true );
 892                                                 array_push ( $tr_attributes , '' );
 893                                                 array_pop ( $has_opened_tr );
 894                                                 array_push ( $has_opened_tr , true );
 895                                         }
 896
 897                                         $last_tag = array_pop ( $last_tag_history );
 898
 899                                         if ( array_pop ( $td_history ) ) {
 900                                                 $previous = "</{$last_tag}>{$previous}";
 901                                         }
 902
 903                                         if ( $first_character == '|' ) {
 904                                                 $last_tag = 'td';
 905                                         } else if ( $first_character == '!' ) {
 906                                                 $last_tag = 'th';
 907                                         } else if ( $first_character == '+' ) {
 908                                                 $last_tag = 'caption';
 909                                         } else {
 910                                                 $last_tag = '';
 911                                         }
 912
 913                                         array_push ( $last_tag_history , $last_tag );
 914
 915                                         // A cell could contain both parameters and data
 916                                         $cell_data = explode ( '|' , $cell , 2 );
 917
 918                                         // Bug 553: Note that a '|' inside an invalid link should not
 919                                         // be mistaken as delimiting cell parameters
 920                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 921                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 922                                         } else if ( count ( $cell_data ) == 1 )
 923                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 924                                         else {
 925                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 926                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 927                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 928                                         }
 929
 930                                         $lines[$key] .= $cell;
 931                                         array_push ( $td_history , true );
 932                                 }
 933                         }
 934                 }
 935
 936                 // Closing open td, tr && table
 937                 while ( count ( $td_history ) > 0 )
 938                 {
 939                         if ( array_pop ( $td_history ) ) {
 940                                 $lines[] = '</td>' ;
 941                         }
 942                         if ( array_pop ( $tr_history ) ) {
 943                                 $lines[] = '</tr>' ;
 944                         }
 945                         if ( !array_pop ( $has_opened_tr ) ) {
 946                                 $lines[] = "<tr><td></td></tr>" ;
 947                         }
 948
 949                         $lines[] = '</table>' ;
 950                 }
 951
 952                 $output = implode ( "\n" , $lines ) ;
 953
 954                 // special case: don't return empty table
 955                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 956                         $output = '';
 957                 }
 958
 959                 wfProfileOut( $fname );
 960
 961                 return $output;
 962         }
 963
 964         /**
 965          * Helper function for parse() that transforms wiki markup into
 966          * HTML. Only called for $mOutputType == self::OT_HTML.
 967          *
 968          * @private
 969          */
 970         function internalParse( $text ) {
 971                 $isMain = true;
 972                 $fname = 'Parser::internalParse';
 973                 wfProfileIn( $fname );
 974
 975                 # Hook to suspend the parser in this state
 976                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 977                         wfProfileOut( $fname );
 978                         return $text ;
 979                 }
 980
 981                 $text = $this->replaceVariables( $text );
 982                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 983                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 984
 985                 // Tables need to come after variable replacement for things to work
 986                 // properly; putting them before other transformations should keep
 987                 // exciting things like link expansions from showing up in surprising
 988                 // places.
 989                 $text = $this->doTableStuff( $text );
 990
 991                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 992
 993                 $text = $this->stripToc( $text );
 994                 $this->stripNoGallery( $text );
 995                 $text = $this->doHeadings( $text );
 996                 if($this->mOptions->getUseDynamicDates()) {
 997                         $df =& DateFormatter::getInstance();
 998                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 999                 }
1000                 $text = $this->doAllQuotes( $text );
1001                 $text = $this->replaceInternalLinks( $text );
1002                 $text = $this->replaceExternalLinks( $text );
1003
1004                 # replaceInternalLinks may sometimes leave behind
1005                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1006                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1007
1008                 $text = $this->doMagicLinks( $text );
1009                 $text = $this->formatHeadings( $text, $isMain );
1010
1011                 wfProfileOut( $fname );
1012                 return $text;
1013         }
1014
1015         /**
1016          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1017          * magic external links.
1018          *
1019          * @private
1020          */
1021         function doMagicLinks( $text ) {
1022                 wfProfileIn( __METHOD__ );
1023                 $text = preg_replace_callback(
1024                         '!(?:                           # Start cases
1025                             <a.*?</a> |                 # Skip link text
1026                             <.*?> |                     # Skip stuff inside HTML elements
1027                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1028                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1029                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1030                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1031                                      [0-9Xx]                 # check digit
1032                                    \b)
1033                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1034                 wfProfileOut( __METHOD__ );
1035                 return $text;
1036         }
1037
1038         function magicLinkCallback( $m ) {
1039                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1040                         # Skip HTML element
1041                         return $m[0];
1042                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1043                         $isbn = $m[2];
1044                         $num = strtr( $isbn, array(
1045                                 '-' => '',
1046                                 ' ' => '',
1047                                 'x' => 'X',
1048                         ));
1049                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1050                         $text = '<a href="' .
1051                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1052                                 "\" class=\"internal\">ISBN $isbn</a>";
1053                 } else {
1054                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1055                                 $keyword = 'RFC';
1056                                 $urlmsg = 'rfcurl';
1057                                 $id = $m[1];
1058                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1059                                 $keyword = 'PMID';
1060                                 $urlmsg = 'pubmedurl';
1061                                 $id = $m[1];
1062                         } else {
1063                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1064                                         substr($m[0], 0, 20 ) . '"' );
1065                         }
1066
1067                         $url = wfMsg( $urlmsg, $id);
1068                         $sk = $this->mOptions->getSkin();
1069                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1070                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1071                 }
1072                 return $text;
1073         }
1074
1075         /**
1076          * Parse headers and return html
1077          *
1078          * @private
1079          */
1080         function doHeadings( $text ) {
1081                 $fname = 'Parser::doHeadings';
1082                 wfProfileIn( $fname );
1083                 for ( $i = 6; $i >= 1; --$i ) {
1084                         $h = str_repeat( '=', $i );
1085                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1086                           "<h$i>\\1</h$i>", $text );
1087                 }
1088                 wfProfileOut( $fname );
1089                 return $text;
1090         }
1091
1092         /**
1093          * Replace single quotes with HTML markup
1094          * @private
1095          * @return string the altered text
1096          */
1097         function doAllQuotes( $text ) {
1098                 $fname = 'Parser::doAllQuotes';
1099                 wfProfileIn( $fname );
1100                 $outtext = '';
1101                 $lines = explode( "\n", $text );
1102                 foreach ( $lines as $line ) {
1103                         $outtext .= $this->doQuotes ( $line ) . "\n";
1104                 }
1105                 $outtext = substr($outtext, 0,-1);
1106                 wfProfileOut( $fname );
1107                 return $outtext;
1108         }
1109
1110         /**
1111          * Helper function for doAllQuotes()
1112          */
1113         public function doQuotes( $text ) {
1114                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1115                 if ( count( $arr ) == 1 )
1116                         return $text;
1117                 else
1118                 {
1119                         # First, do some preliminary work. This may shift some apostrophes from
1120                         # being mark-up to being text. It also counts the number of occurrences
1121                         # of bold and italics mark-ups.
1122                         $i = 0;
1123                         $numbold = 0;
1124                         $numitalics = 0;
1125                         foreach ( $arr as $r )
1126                         {
1127                                 if ( ( $i % 2 ) == 1 )
1128                                 {
1129                                         # If there are ever four apostrophes, assume the first is supposed to
1130                                         # be text, and the remaining three constitute mark-up for bold text.
1131                                         if ( strlen( $arr[$i] ) == 4 )
1132                                         {
1133                                                 $arr[$i-1] .= "'";
1134                                                 $arr[$i] = "'''";
1135                                         }
1136                                         # If there are more than 5 apostrophes in a row, assume they're all
1137                                         # text except for the last 5.
1138                                         else if ( strlen( $arr[$i] ) > 5 )
1139                                         {
1140                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1141                                                 $arr[$i] = "'''''";
1142                                         }
1143                                         # Count the number of occurrences of bold and italics mark-ups.
1144                                         # We are not counting sequences of five apostrophes.
1145                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1146                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1147                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1148                                 }
1149                                 $i++;
1150                         }
1151
1152                         # If there is an odd number of both bold and italics, it is likely
1153                         # that one of the bold ones was meant to be an apostrophe followed
1154                         # by italics. Which one we cannot know for certain, but it is more
1155                         # likely to be one that has a single-letter word before it.
1156                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1157                         {
1158                                 $i = 0;
1159                                 $firstsingleletterword = -1;
1160                                 $firstmultiletterword = -1;
1161                                 $firstspace = -1;
1162                                 foreach ( $arr as $r )
1163                                 {
1164                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1165                                         {
1166                                                 $x1 = substr ($arr[$i-1], -1);
1167                                                 $x2 = substr ($arr[$i-1], -2, 1);
1168                                                 if ($x1 == ' ') {
1169                                                         if ($firstspace == -1) $firstspace = $i;
1170                                                 } else if ($x2 == ' ') {
1171                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1172                                                 } else {
1173                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1174                                                 }
1175                                         }
1176                                         $i++;
1177                                 }
1178
1179                                 # If there is a single-letter word, use it!
1180                                 if ($firstsingleletterword > -1)
1181                                 {
1182                                         $arr [ $firstsingleletterword ] = "''";
1183                                         $arr [ $firstsingleletterword-1 ] .= "'";
1184                                 }
1185                                 # If not, but there's a multi-letter word, use that one.
1186                                 else if ($firstmultiletterword > -1)
1187                                 {
1188                                         $arr [ $firstmultiletterword ] = "''";
1189                                         $arr [ $firstmultiletterword-1 ] .= "'";
1190                                 }
1191                                 # ... otherwise use the first one that has neither.
1192                                 # (notice that it is possible for all three to be -1 if, for example,
1193                                 # there is only one pentuple-apostrophe in the line)
1194                                 else if ($firstspace > -1)
1195                                 {
1196                                         $arr [ $firstspace ] = "''";
1197                                         $arr [ $firstspace-1 ] .= "'";
1198                                 }
1199                         }
1200
1201                         # Now let's actually convert our apostrophic mush to HTML!
1202                         $output = '';
1203                         $buffer = '';
1204                         $state = '';
1205                         $i = 0;
1206                         foreach ($arr as $r)
1207                         {
1208                                 if (($i % 2) == 0)
1209                                 {
1210                                         if ($state == 'both')
1211                                                 $buffer .= $r;
1212                                         else
1213                                                 $output .= $r;
1214                                 }
1215                                 else
1216                                 {
1217                                         if (strlen ($r) == 2)
1218                                         {
1219                                                 if ($state == 'i')
1220                                                 { $output .= '</i>'; $state = ''; }
1221                                                 else if ($state == 'bi')
1222                                                 { $output .= '</i>'; $state = 'b'; }
1223                                                 else if ($state == 'ib')
1224                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1225                                                 else if ($state == 'both')
1226                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1227                                                 else # $state can be 'b' or ''
1228                                                 { $output .= '<i>'; $state .= 'i'; }
1229                                         }
1230                                         else if (strlen ($r) == 3)
1231                                         {
1232                                                 if ($state == 'b')
1233                                                 { $output .= '</b>'; $state = ''; }
1234                                                 else if ($state == 'bi')
1235                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1236                                                 else if ($state == 'ib')
1237                                                 { $output .= '</b>'; $state = 'i'; }
1238                                                 else if ($state == 'both')
1239                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1240                                                 else # $state can be 'i' or ''
1241                                                 { $output .= '<b>'; $state .= 'b'; }
1242                                         }
1243                                         else if (strlen ($r) == 5)
1244                                         {
1245                                                 if ($state == 'b')
1246                                                 { $output .= '</b><i>'; $state = 'i'; }
1247                                                 else if ($state == 'i')
1248                                                 { $output .= '</i><b>'; $state = 'b'; }
1249                                                 else if ($state == 'bi')
1250                                                 { $output .= '</i></b>'; $state = ''; }
1251                                                 else if ($state == 'ib')
1252                                                 { $output .= '</b></i>'; $state = ''; }
1253                                                 else if ($state == 'both')
1254                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1255                                                 else # ($state == '')
1256                                                 { $buffer = ''; $state = 'both'; }
1257                                         }
1258                                 }
1259                                 $i++;
1260                         }
1261                         # Now close all remaining tags.  Notice that the order is important.
1262                         if ($state == 'b' || $state == 'ib')
1263                                 $output .= '</b>';
1264                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1265                                 $output .= '</i>';
1266                         if ($state == 'bi')
1267                                 $output .= '</b>';
1268                         # There might be lonely ''''', so make sure we have a buffer
1269                         if ($state == 'both' && $buffer)
1270                                 $output .= '<b><i>'.$buffer.'</i></b>';
1271                         return $output;
1272                 }
1273         }
1274
1275         /**
1276          * Replace external links
1277          *
1278          * Note: this is all very hackish and the order of execution matters a lot.
1279          * Make sure to run maintenance/parserTests.php if you change this code.
1280          *
1281          * @private
1282          */
1283         function replaceExternalLinks( $text ) {
1284                 global $wgContLang;
1285                 $fname = 'Parser::replaceExternalLinks';
1286                 wfProfileIn( $fname );
1287
1288                 $sk = $this->mOptions->getSkin();
1289
1290                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1291
1292                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1293
1294                 $i = 0;
1295                 while ( $i<count( $bits ) ) {
1296                         $url = $bits[$i++];
1297                         $protocol = $bits[$i++];
1298                         $text = $bits[$i++];
1299                         $trail = $bits[$i++];
1300
1301                         # The characters '<' and '>' (which were escaped by
1302                         # removeHTMLtags()) should not be included in
1303                         # URLs, per RFC 2396.
1304                         $m2 = array();
1305                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1306                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1307                                 $url = substr($url, 0, $m2[0][1]);
1308                         }
1309
1310                         # If the link text is an image URL, replace it with an <img> tag
1311                         # This happened by accident in the original parser, but some people used it extensively
1312                         $img = $this->maybeMakeExternalImage( $text );
1313                         if ( $img !== false ) {
1314                                 $text = $img;
1315                         }
1316
1317                         $dtrail = '';
1318
1319                         # Set linktype for CSS - if URL==text, link is essentially free
1320                         $linktype = ($text == $url) ? 'free' : 'text';
1321
1322                         # No link text, e.g. [http://domain.tld/some.link]
1323                         if ( $text == '' ) {
1324                                 # Autonumber if allowed. See bug #5918
1325                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1326                                         $text = '[' . ++$this->mAutonumber . ']';
1327                                         $linktype = 'autonumber';
1328                                 } else {
1329                                         # Otherwise just use the URL
1330                                         $text = htmlspecialchars( $url );
1331                                         $linktype = 'free';
1332                                 }
1333                         } else {
1334                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1335                                 # Check for trail
1336                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1337                         }
1338
1339                         $text = $wgContLang->markNoConversion($text);
1340
1341                         $url = Sanitizer::cleanUrl( $url );
1342
1343                         # Process the trail (i.e. everything after this link up until start of the next link),
1344                         # replacing any non-bracketed links
1345                         $trail = $this->replaceFreeExternalLinks( $trail );
1346
1347                         # Use the encoded URL
1348                         # This means that users can paste URLs directly into the text
1349                         # Funny characters like &ouml; aren't valid in URLs anyway
1350                         # This was changed in August 2004
1351                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1352
1353                         # Register link in the output object.
1354                         # Replace unnecessary URL escape codes with the referenced character
1355                         # This prevents spammers from hiding links from the filters
1356                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1357                         $this->mOutput->addExternalLink( $pasteurized );
1358                 }
1359
1360                 wfProfileOut( $fname );
1361                 return $s;
1362         }
1363
1364         /**
1365          * Replace anything that looks like a URL with a link
1366          * @private
1367          */
1368         function replaceFreeExternalLinks( $text ) {
1369                 global $wgContLang;
1370                 $fname = 'Parser::replaceFreeExternalLinks';
1371                 wfProfileIn( $fname );
1372
1373                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1374                 $s = array_shift( $bits );
1375                 $i = 0;
1376
1377                 $sk = $this->mOptions->getSkin();
1378
1379                 while ( $i < count( $bits ) ){
1380                         $protocol = $bits[$i++];
1381                         $remainder = $bits[$i++];
1382
1383                         $m = array();
1384                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1385                                 # Found some characters after the protocol that look promising
1386                                 $url = $protocol . $m[1];
1387                                 $trail = $m[2];
1388
1389                                 # special case: handle urls as url args:
1390                                 # http://www.example.com/foo?=http://www.example.com/bar
1391                                 if(strlen($trail) == 0 &&
1392                                         isset($bits[$i]) &&
1393                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1394                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1395                                 {
1396                                         # add protocol, arg
1397                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1398                                         $i += 2;
1399                                         $trail = $m[2];
1400                                 }
1401
1402                                 # The characters '<' and '>' (which were escaped by
1403                                 # removeHTMLtags()) should not be included in
1404                                 # URLs, per RFC 2396.
1405                                 $m2 = array();
1406                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1407                                         $trail = substr($url, $m2[0][1]) . $trail;
1408                                         $url = substr($url, 0, $m2[0][1]);
1409                                 }
1410
1411                                 # Move trailing punctuation to $trail
1412                                 $sep = ',;\.:!?';
1413                                 # If there is no left bracket, then consider right brackets fair game too
1414                                 if ( strpos( $url, '(' ) === false ) {
1415                                         $sep .= ')';
1416                                 }
1417
1418                                 $numSepChars = strspn( strrev( $url ), $sep );
1419                                 if ( $numSepChars ) {
1420                                         $trail = substr( $url, -$numSepChars ) . $trail;
1421                                         $url = substr( $url, 0, -$numSepChars );
1422                                 }
1423
1424                                 $url = Sanitizer::cleanUrl( $url );
1425
1426                                 # Is this an external image?
1427                                 $text = $this->maybeMakeExternalImage( $url );
1428                                 if ( $text === false ) {
1429                                         # Not an image, make a link
1430                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1431                                         # Register it in the output object...
1432                                         # Replace unnecessary URL escape codes with their equivalent characters
1433                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1434                                         $this->mOutput->addExternalLink( $pasteurized );
1435                                 }
1436                                 $s .= $text . $trail;
1437                         } else {
1438                                 $s .= $protocol . $remainder;
1439                         }
1440                 }
1441                 wfProfileOut( $fname );
1442                 return $s;
1443         }
1444
1445         /**
1446          * Replace unusual URL escape codes with their equivalent characters
1447          * @param string
1448          * @return string
1449          * @static
1450          * @todo  This can merge genuinely required bits in the path or query string,
1451          *        breaking legit URLs. A proper fix would treat the various parts of
1452          *        the URL differently; as a workaround, just use the output for
1453          *        statistical records, not for actual linking/output.
1454          */
1455         static function replaceUnusualEscapes( $url ) {
1456                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1457                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1458         }
1459
1460         /**
1461          * Callback function used in replaceUnusualEscapes().
1462          * Replaces unusual URL escape codes with their equivalent character
1463          * @static
1464          * @private
1465          */
1466         private static function replaceUnusualEscapesCallback( $matches ) {
1467                 $char = urldecode( $matches[0] );
1468                 $ord = ord( $char );
1469                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1470                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1471                         // No, shouldn't be escaped
1472                         return $char;
1473                 } else {
1474                         // Yes, leave it escaped
1475                         return $matches[0];
1476                 }
1477         }
1478
1479         /**
1480          * make an image if it's allowed, either through the global
1481          * option or through the exception
1482          * @private
1483          */
1484         function maybeMakeExternalImage( $url ) {
1485                 $sk = $this->mOptions->getSkin();
1486                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1487                 $imagesexception = !empty($imagesfrom);
1488                 $text = false;
1489                 if ( $this->mOptions->getAllowExternalImages()
1490                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1491                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1492                                 # Image found
1493                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1494                         }
1495                 }
1496                 return $text;
1497         }
1498
1499         /**
1500          * Process [[ ]] wikilinks
1501          *
1502          * @private
1503          */
1504         function replaceInternalLinks( $s ) {
1505                 global $wgContLang;
1506                 static $fname = 'Parser::replaceInternalLinks' ;
1507
1508                 wfProfileIn( $fname );
1509
1510                 wfProfileIn( $fname.'-setup' );
1511                 static $tc = FALSE;
1512                 # the % is needed to support urlencoded titles as well
1513                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1514
1515                 $sk = $this->mOptions->getSkin();
1516
1517                 #split the entire text string on occurences of [[
1518                 $a = explode( '[[', ' ' . $s );
1519                 #get the first element (all text up to first [[), and remove the space we added
1520                 $s = array_shift( $a );
1521                 $s = substr( $s, 1 );
1522
1523                 # Match a link having the form [[namespace:link|alternate]]trail
1524                 static $e1 = FALSE;
1525                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1526                 # Match cases where there is no "]]", which might still be images
1527                 static $e1_img = FALSE;
1528                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1529
1530                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1531                 $e2 = null;
1532                 if ( $useLinkPrefixExtension ) {
1533                         # Match the end of a line for a word that's not followed by whitespace,
1534                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1535                         $e2 = wfMsgForContent( 'linkprefix' );
1536                 }
1537
1538                 if( is_null( $this->mTitle ) ) {
1539                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1540                 }
1541                 $nottalk = !$this->mTitle->isTalkPage();
1542
1543                 if ( $useLinkPrefixExtension ) {
1544                         $m = array();
1545                         if ( preg_match( $e2, $s, $m ) ) {
1546                                 $first_prefix = $m[2];
1547                         } else {
1548                                 $first_prefix = false;
1549                         }
1550                 } else {
1551                         $prefix = '';
1552                 }
1553
1554                 if($wgContLang->hasVariants()) {
1555                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1556                 } else {
1557                         $selflink = array($this->mTitle->getPrefixedText());
1558                 }
1559                 $useSubpages = $this->areSubpagesAllowed();
1560                 wfProfileOut( $fname.'-setup' );
1561
1562                 # Loop for each link
1563                 for ($k = 0; isset( $a[$k] ); $k++) {
1564                         $line = $a[$k];
1565                         if ( $useLinkPrefixExtension ) {
1566                                 wfProfileIn( $fname.'-prefixhandling' );
1567                                 if ( preg_match( $e2, $s, $m ) ) {
1568                                         $prefix = $m[2];
1569                                         $s = $m[1];
1570                                 } else {
1571                                         $prefix='';
1572                                 }
1573                                 # first link
1574                                 if($first_prefix) {
1575                                         $prefix = $first_prefix;
1576                                         $first_prefix = false;
1577                                 }
1578                                 wfProfileOut( $fname.'-prefixhandling' );
1579                         }
1580
1581                         $might_be_img = false;
1582
1583                         wfProfileIn( "$fname-e1" );
1584                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1585                                 $text = $m[2];
1586                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1587                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1588                                 # the real problem is with the $e1 regex
1589                                 # See bug 1300.
1590                                 #
1591                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1592                                 # and no image is in sight. See bug 2095.
1593                                 #
1594                                 if( $text !== '' &&
1595                                         substr( $m[3], 0, 1 ) === ']' &&
1596                                         strpos($text, '[') !== false
1597                                 )
1598                                 {
1599                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1600                                         $m[3] = substr( $m[3], 1 );
1601                                 }
1602                                 # fix up urlencoded title texts
1603                                 if( strpos( $m[1], '%' ) !== false ) {
1604                                         # Should anchors '#' also be rejected?
1605                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1606                                 }
1607                                 $trail = $m[3];
1608                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1609                                 $might_be_img = true;
1610                                 $text = $m[2];
1611                                 if ( strpos( $m[1], '%' ) !== false ) {
1612                                         $m[1] = urldecode($m[1]);
1613                                 }
1614                                 $trail = "";
1615                         } else { # Invalid form; output directly
1616                                 $s .= $prefix . '[[' . $line ;
1617                                 wfProfileOut( "$fname-e1" );
1618                                 continue;
1619                         }
1620                         wfProfileOut( "$fname-e1" );
1621                         wfProfileIn( "$fname-misc" );
1622
1623                         # Don't allow internal links to pages containing
1624                         # PROTO: where PROTO is a valid URL protocol; these
1625                         # should be external links.
1626                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1627                                 $s .= $prefix . '[[' . $line ;
1628                                 continue;
1629                         }
1630
1631                         # Make subpage if necessary
1632                         if( $useSubpages ) {
1633                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1634                         } else {
1635                                 $link = $m[1];
1636                         }
1637
1638                         $noforce = (substr($m[1], 0, 1) != ':');
1639                         if (!$noforce) {
1640                                 # Strip off leading ':'
1641                                 $link = substr($link, 1);
1642                         }
1643
1644                         wfProfileOut( "$fname-misc" );
1645                         wfProfileIn( "$fname-title" );
1646                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1647                         if( !$nt ) {
1648                                 $s .= $prefix . '[[' . $line;
1649                                 wfProfileOut( "$fname-title" );
1650                                 continue;
1651                         }
1652
1653                         $ns = $nt->getNamespace();
1654                         $iw = $nt->getInterWiki();
1655                         wfProfileOut( "$fname-title" );
1656
1657                         if ($might_be_img) { # if this is actually an invalid link
1658                                 wfProfileIn( "$fname-might_be_img" );
1659                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1660                                         $found = false;
1661                                         while (isset ($a[$k+1]) ) {
1662                                                 #look at the next 'line' to see if we can close it there
1663                                                 $spliced = array_splice( $a, $k + 1, 1 );
1664                                                 $next_line = array_shift( $spliced );
1665                                                 $m = explode( ']]', $next_line, 3 );
1666                                                 if ( count( $m ) == 3 ) {
1667                                                         # the first ]] closes the inner link, the second the image
1668                                                         $found = true;
1669                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1670                                                         $trail = $m[2];
1671                                                         break;
1672                                                 } elseif ( count( $m ) == 2 ) {
1673                                                         #if there's exactly one ]] that's fine, we'll keep looking
1674                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1675                                                 } else {
1676                                                         #if $next_line is invalid too, we need look no further
1677                                                         $text .= '[[' . $next_line;
1678                                                         break;
1679                                                 }
1680                                         }
1681                                         if ( !$found ) {
1682                                                 # we couldn't find the end of this imageLink, so output it raw
1683                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1684                                                 $text = $this->replaceInternalLinks($text);
1685                                                 $s .= "{$prefix}[[$link|$text";
1686                                                 # note: no $trail, because without an end, there *is* no trail
1687                                                 wfProfileOut( "$fname-might_be_img" );
1688                                                 continue;
1689                                         }
1690                                 } else { #it's not an image, so output it raw
1691                                         $s .= "{$prefix}[[$link|$text";
1692                                         # note: no $trail, because without an end, there *is* no trail
1693                                         wfProfileOut( "$fname-might_be_img" );
1694                                         continue;
1695                                 }
1696                                 wfProfileOut( "$fname-might_be_img" );
1697                         }
1698
1699                         $wasblank = ( '' == $text );
1700                         if( $wasblank ) $text = $link;
1701
1702                         # Link not escaped by : , create the various objects
1703                         if( $noforce ) {
1704
1705                                 # Interwikis
1706                                 wfProfileIn( "$fname-interwiki" );
1707                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1708                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1709                                         $s = rtrim($s . $prefix);
1710                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1711                                         wfProfileOut( "$fname-interwiki" );
1712                                         continue;
1713                                 }
1714                                 wfProfileOut( "$fname-interwiki" );
1715
1716                                 if ( $ns == NS_IMAGE ) {
1717                                         wfProfileIn( "$fname-image" );
1718                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1719                                                 # recursively parse links inside the image caption
1720                                                 # actually, this will parse them in any other parameters, too,
1721                                                 # but it might be hard to fix that, and it doesn't matter ATM
1722                                                 $text = $this->replaceExternalLinks($text);
1723                                                 $text = $this->replaceInternalLinks($text);
1724
1725                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1726                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1727                                                 $this->mOutput->addImage( $nt->getDBkey() );
1728
1729                                                 wfProfileOut( "$fname-image" );
1730                                                 continue;
1731                                         } else {
1732                                                 # We still need to record the image's presence on the page
1733                                                 $this->mOutput->addImage( $nt->getDBkey() );
1734                                         }
1735                                         wfProfileOut( "$fname-image" );
1736
1737                                 }
1738
1739                                 if ( $ns == NS_CATEGORY ) {
1740                                         wfProfileIn( "$fname-category" );
1741                                         $s = rtrim($s . "\n"); # bug 87
1742
1743                                         if ( $wasblank ) {
1744                                                 $sortkey = $this->getDefaultSort();
1745                                         } else {
1746                                                 $sortkey = $text;
1747                                         }
1748                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1749                                         $sortkey = str_replace( "\n", '', $sortkey );
1750                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1751                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1752
1753                                         /**
1754                                          * Strip the whitespace Category links produce, see bug 87
1755                                          * @todo We might want to use trim($tmp, "\n") here.
1756                                          */
1757                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1758
1759                                         wfProfileOut( "$fname-category" );
1760                                         continue;
1761                                 }
1762                         }
1763
1764                         # Self-link checking
1765                         if( $nt->getFragment() === '' ) {
1766                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1767                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1768                                         continue;
1769                                 }
1770                         }
1771
1772                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1773                         if( $ns == NS_MEDIA ) {
1774                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1775                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1776                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1777                                 $this->mOutput->addImage( $nt->getDBkey() );
1778                                 continue;
1779                         } elseif( $ns == NS_SPECIAL ) {
1780                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1781                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1782                                 } else {
1783                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1784                                 }
1785                                 continue;
1786                         } elseif( $ns == NS_IMAGE ) {
1787                                 $img = wfFindFile( $nt );
1788                                 if( $img ) {
1789                                         // Force a blue link if the file exists; may be a remote
1790                                         // upload on the shared repository, and we want to see its
1791                                         // auto-generated page.
1792                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1793                                         $this->mOutput->addLink( $nt );
1794                                         continue;
1795                                 }
1796                         }
1797                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1798                 }
1799                 wfProfileOut( $fname );
1800                 return $s;
1801         }
1802
1803         /**
1804          * Make a link placeholder. The text returned can be later resolved to a real link with
1805          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1806          * parsing of interwiki links, and secondly to allow all existence checks and
1807          * article length checks (for stub links) to be bundled into a single query.
1808          *
1809          */
1810         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1811                 wfProfileIn( __METHOD__ );
1812                 if ( ! is_object($nt) ) {
1813                         # Fail gracefully
1814                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1815                 } else {
1816                         # Separate the link trail from the rest of the link
1817                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1818
1819                         if ( $nt->isExternal() ) {
1820                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1821                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1822                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1823                         } else {
1824                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1825                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1826                                 $this->mLinkHolders['queries'][] = $query;
1827                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1828                                 $this->mLinkHolders['titles'][] = $nt;
1829
1830                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1831                         }
1832                 }
1833                 wfProfileOut( __METHOD__ );
1834                 return $retVal;
1835         }
1836
1837         /**
1838          * Render a forced-blue link inline; protect against double expansion of
1839          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1840          * Since this little disaster has to split off the trail text to avoid
1841          * breaking URLs in the following text without breaking trails on the
1842          * wiki links, it's been made into a horrible function.
1843          *
1844          * @param Title $nt
1845          * @param string $text
1846          * @param string $query
1847          * @param string $trail
1848          * @param string $prefix
1849          * @return string HTML-wikitext mix oh yuck
1850          */
1851         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1852                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1853                 $sk = $this->mOptions->getSkin();
1854                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1855                 return $this->armorLinks( $link ) . $trail;
1856         }
1857
1858         /**
1859          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1860          * going to go through further parsing steps before inline URL expansion.
1861          *
1862          * In particular this is important when using action=render, which causes
1863          * full URLs to be included.
1864          *
1865          * Oh man I hate our multi-layer parser!
1866          *
1867          * @param string more-or-less HTML
1868          * @return string less-or-more HTML with NOPARSE bits
1869          */
1870         function armorLinks( $text ) {
1871                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1872                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1873         }
1874
1875         /**
1876          * Return true if subpage links should be expanded on this page.
1877          * @return bool
1878          */
1879         function areSubpagesAllowed() {
1880                 # Some namespaces don't allow subpages
1881                 global $wgNamespacesWithSubpages;
1882                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1883         }
1884
1885         /**
1886          * Handle link to subpage if necessary
1887          * @param string $target the source of the link
1888          * @param string &$text the link text, modified as necessary
1889          * @return string the full name of the link
1890          * @private
1891          */
1892         function maybeDoSubpageLink($target, &$text) {
1893                 # Valid link forms:
1894                 # Foobar -- normal
1895                 # :Foobar -- override special treatment of prefix (images, language links)
1896                 # /Foobar -- convert to CurrentPage/Foobar
1897                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1898                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1899                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1900
1901                 $fname = 'Parser::maybeDoSubpageLink';
1902                 wfProfileIn( $fname );
1903                 $ret = $target; # default return value is no change
1904
1905                 # Some namespaces don't allow subpages,
1906                 # so only perform processing if subpages are allowed
1907                 if( $this->areSubpagesAllowed() ) {
1908                         $hash = strpos( $target, '#' );
1909                         if( $hash !== false ) {
1910                                 $suffix = substr( $target, $hash );
1911                                 $target = substr( $target, 0, $hash );
1912                         } else {
1913                                 $suffix = '';
1914                         }
1915                         # bug 7425
1916                         $target = trim( $target );
1917                         # Look at the first character
1918                         if( $target != '' && $target{0} == '/' ) {
1919                                 # / at end means we don't want the slash to be shown
1920                                 $m = array();
1921                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1922                                 if( $trailingSlashes ) {
1923                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1924                                 } else {
1925                                         $noslash = substr( $target, 1 );
1926                                 }
1927
1928                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1929                                 if( '' === $text ) {
1930                                         $text = $target . $suffix;
1931                                 } # this might be changed for ugliness reasons
1932                         } else {
1933                                 # check for .. subpage backlinks
1934                                 $dotdotcount = 0;
1935                                 $nodotdot = $target;
1936                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1937                                         ++$dotdotcount;
1938                                         $nodotdot = substr( $nodotdot, 3 );
1939                                 }
1940                                 if($dotdotcount > 0) {
1941                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1942                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1943                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1944                                                 # / at the end means don't show full path
1945                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1946                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1947                                                         if( '' === $text ) {
1948                                                                 $text = $nodotdot . $suffix;
1949                                                         }
1950                                                 }
1951                                                 $nodotdot = trim( $nodotdot );
1952                                                 if( $nodotdot != '' ) {
1953                                                         $ret .= '/' . $nodotdot;
1954                                                 }
1955                                                 $ret .= $suffix;
1956                                         }
1957                                 }
1958                         }
1959                 }
1960
1961                 wfProfileOut( $fname );
1962                 return $ret;
1963         }
1964
1965         /**#@+
1966          * Used by doBlockLevels()
1967          * @private
1968          */
1969         /* private */ function closeParagraph() {
1970                 $result = '';
1971                 if ( '' != $this->mLastSection ) {
1972                         $result = '</' . $this->mLastSection  . ">\n";
1973                 }
1974                 $this->mInPre = false;
1975                 $this->mLastSection = '';
1976                 return $result;
1977         }
1978         # getCommon() returns the length of the longest common substring
1979         # of both arguments, starting at the beginning of both.
1980         #
1981         /* private */ function getCommon( $st1, $st2 ) {
1982                 $fl = strlen( $st1 );
1983                 $shorter = strlen( $st2 );
1984                 if ( $fl < $shorter ) { $shorter = $fl; }
1985
1986                 for ( $i = 0; $i < $shorter; ++$i ) {
1987                         if ( $st1{$i} != $st2{$i} ) { break; }
1988                 }
1989                 return $i;
1990         }
1991         # These next three functions open, continue, and close the list
1992         # element appropriate to the prefix character passed into them.
1993         #
1994         /* private */ function openList( $char ) {
1995                 $result = $this->closeParagraph();
1996
1997                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1998                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1999                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2000                 else if ( ';' == $char ) {
2001                         $result .= '<dl><dt>';
2002                         $this->mDTopen = true;
2003                 }
2004                 else { $result = '<!-- ERR 1 -->'; }
2005
2006                 return $result;
2007         }
2008
2009         /* private */ function nextItem( $char ) {
2010                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2011                 else if ( ':' == $char || ';' == $char ) {
2012                         $close = '</dd>';
2013                         if ( $this->mDTopen ) { $close = '</dt>'; }
2014                         if ( ';' == $char ) {
2015                                 $this->mDTopen = true;
2016                                 return $close . '<dt>';
2017                         } else {
2018                                 $this->mDTopen = false;
2019                                 return $close . '<dd>';
2020                         }
2021                 }
2022                 return '<!-- ERR 2 -->';
2023         }
2024
2025         /* private */ function closeList( $char ) {
2026                 if ( '*' == $char ) { $text = '</li></ul>'; }
2027                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2028                 else if ( ':' == $char ) {
2029                         if ( $this->mDTopen ) {
2030                                 $this->mDTopen = false;
2031                                 $text = '</dt></dl>';
2032                         } else {
2033                                 $text = '</dd></dl>';
2034                         }
2035                 }
2036                 else {  return '<!-- ERR 3 -->'; }
2037                 return $text."\n";
2038         }
2039         /**#@-*/
2040
2041         /**
2042          * Make lists from lines starting with ':', '*', '#', etc.
2043          *
2044          * @private
2045          * @return string the lists rendered as HTML
2046          */
2047         function doBlockLevels( $text, $linestart ) {
2048                 $fname = 'Parser::doBlockLevels';
2049                 wfProfileIn( $fname );
2050
2051                 # Parsing through the text line by line.  The main thing
2052                 # happening here is handling of block-level elements p, pre,
2053                 # and making lists from lines starting with * # : etc.
2054                 #
2055                 $textLines = explode( "\n", $text );
2056
2057                 $lastPrefix = $output = '';
2058                 $this->mDTopen = $inBlockElem = false;
2059                 $prefixLength = 0;
2060                 $paragraphStack = false;
2061
2062                 if ( !$linestart ) {
2063                         $output .= array_shift( $textLines );
2064                 }
2065                 foreach ( $textLines as $oLine ) {
2066                         $lastPrefixLength = strlen( $lastPrefix );
2067                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2068                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2069                         if ( !$this->mInPre ) {
2070                                 # Multiple prefixes may abut each other for nested lists.
2071                                 $prefixLength = strspn( $oLine, '*#:;' );
2072                                 $pref = substr( $oLine, 0, $prefixLength );
2073
2074                                 # eh?
2075                                 $pref2 = str_replace( ';', ':', $pref );
2076                                 $t = substr( $oLine, $prefixLength );
2077                                 $this->mInPre = !empty($preOpenMatch);
2078                         } else {
2079                                 # Don't interpret any other prefixes in preformatted text
2080                                 $prefixLength = 0;
2081                                 $pref = $pref2 = '';
2082                                 $t = $oLine;
2083                         }
2084
2085                         # List generation
2086                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2087                                 # Same as the last item, so no need to deal with nesting or opening stuff
2088                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2089                                 $paragraphStack = false;
2090
2091                                 if ( substr( $pref, -1 ) == ';') {
2092                                         # The one nasty exception: definition lists work like this:
2093                                         # ; title : definition text
2094                                         # So we check for : in the remainder text to split up the
2095                                         # title and definition, without b0rking links.
2096                                         $term = $t2 = '';
2097                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2098                                                 $t = $t2;
2099                                                 $output .= $term . $this->nextItem( ':' );
2100                                         }
2101                                 }
2102                         } elseif( $prefixLength || $lastPrefixLength ) {
2103                                 # Either open or close a level...
2104                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2105                                 $paragraphStack = false;
2106
2107                                 while( $commonPrefixLength < $lastPrefixLength ) {
2108                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2109                                         --$lastPrefixLength;
2110                                 }
2111                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2112                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2113                                 }
2114                                 while ( $prefixLength > $commonPrefixLength ) {
2115                                         $char = substr( $pref, $commonPrefixLength, 1 );
2116                                         $output .= $this->openList( $char );
2117
2118                                         if ( ';' == $char ) {
2119                                                 # FIXME: This is dupe of code above
2120                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2121                                                         $t = $t2;
2122                                                         $output .= $term . $this->nextItem( ':' );
2123                                                 }
2124                                         }
2125                                         ++$commonPrefixLength;
2126                                 }
2127                                 $lastPrefix = $pref2;
2128                         }
2129                         if( 0 == $prefixLength ) {
2130                                 wfProfileIn( "$fname-paragraph" );
2131                                 # No prefix (not in list)--go to paragraph mode
2132                                 // XXX: use a stack for nestable elements like span, table and div
2133                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2134                                 $closematch = preg_match(
2135                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2136                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2137                                 if ( $openmatch or $closematch ) {
2138                                         $paragraphStack = false;
2139                                         # TODO bug 5718: paragraph closed
2140                                         $output .= $this->closeParagraph();
2141                                         if ( $preOpenMatch and !$preCloseMatch ) {
2142                                                 $this->mInPre = true;
2143                                         }
2144                                         if ( $closematch ) {
2145                                                 $inBlockElem = false;
2146                                         } else {
2147                                                 $inBlockElem = true;
2148                                         }
2149                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2150                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2151                                                 // pre
2152                                                 if ($this->mLastSection != 'pre') {
2153                                                         $paragraphStack = false;
2154                                                         $output .= $this->closeParagraph().'<pre>';
2155                                                         $this->mLastSection = 'pre';
2156                                                 }
2157                                                 $t = substr( $t, 1 );
2158                                         } else {
2159                                                 // paragraph
2160                                                 if ( '' == trim($t) ) {
2161                                                         if ( $paragraphStack ) {
2162                                                                 $output .= $paragraphStack.'<br />';
2163                                                                 $paragraphStack = false;
2164                                                                 $this->mLastSection = 'p';
2165                                                         } else {
2166                                                                 if ($this->mLastSection != 'p' ) {
2167                                                                         $output .= $this->closeParagraph();
2168                                                                         $this->mLastSection = '';
2169                                                                         $paragraphStack = '<p>';
2170                                                                 } else {
2171                                                                         $paragraphStack = '</p><p>';
2172                                                                 }
2173                                                         }
2174                                                 } else {
2175                                                         if ( $paragraphStack ) {
2176                                                                 $output .= $paragraphStack;
2177                                                                 $paragraphStack = false;
2178                                                                 $this->mLastSection = 'p';
2179                                                         } else if ($this->mLastSection != 'p') {
2180                                                                 $output .= $this->closeParagraph().'<p>';
2181                                                                 $this->mLastSection = 'p';
2182                                                         }
2183                                                 }
2184                                         }
2185                                 }
2186                                 wfProfileOut( "$fname-paragraph" );
2187                         }
2188                         // somewhere above we forget to get out of pre block (bug 785)
2189                         if($preCloseMatch && $this->mInPre) {
2190                                 $this->mInPre = false;
2191                         }
2192                         if ($paragraphStack === false) {
2193                                 $output .= $t."\n";
2194                         }
2195                 }
2196                 while ( $prefixLength ) {
2197                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2198                         --$prefixLength;
2199                 }
2200                 if ( '' != $this->mLastSection ) {
2201                         $output .= '</' . $this->mLastSection . '>';
2202                         $this->mLastSection = '';
2203                 }
2204
2205                 wfProfileOut( $fname );
2206                 return $output;
2207         }
2208
2209         /**
2210          * Split up a string on ':', ignoring any occurences inside tags
2211          * to prevent illegal overlapping.
2212          * @param string $str the string to split
2213          * @param string &$before set to everything before the ':'
2214          * @param string &$after set to everything after the ':'
2215          * return string the position of the ':', or false if none found
2216          */
2217         function findColonNoLinks($str, &$before, &$after) {
2218                 $fname = 'Parser::findColonNoLinks';
2219                 wfProfileIn( $fname );
2220
2221                 $pos = strpos( $str, ':' );
2222                 if( $pos === false ) {
2223                         // Nothing to find!
2224                         wfProfileOut( $fname );
2225                         return false;
2226                 }
2227
2228                 $lt = strpos( $str, '<' );
2229                 if( $lt === false || $lt > $pos ) {
2230                         // Easy; no tag nesting to worry about
2231                         $before = substr( $str, 0, $pos );
2232                         $after = substr( $str, $pos+1 );
2233                         wfProfileOut( $fname );
2234                         return $pos;
2235                 }
2236
2237                 // Ugly state machine to walk through avoiding tags.
2238                 $state = self::COLON_STATE_TEXT;
2239                 $stack = 0;
2240                 $len = strlen( $str );
2241                 for( $i = 0; $i < $len; $i++ ) {
2242                         $c = $str{$i};
2243
2244                         switch( $state ) {
2245                         // (Using the number is a performance hack for common cases)
2246                         case 0: // self::COLON_STATE_TEXT:
2247                                 switch( $c ) {
2248                                 case "<":
2249                                         // Could be either a <start> tag or an </end> tag
2250                                         $state = self::COLON_STATE_TAGSTART;
2251                                         break;
2252                                 case ":":
2253                                         if( $stack == 0 ) {
2254                                                 // We found it!
2255                                                 $before = substr( $str, 0, $i );
2256                                                 $after = substr( $str, $i + 1 );
2257                                                 wfProfileOut( $fname );
2258                                                 return $i;
2259                                         }
2260                                         // Embedded in a tag; don't break it.
2261                                         break;
2262                                 default:
2263                                         // Skip ahead looking for something interesting
2264                                         $colon = strpos( $str, ':', $i );
2265                                         if( $colon === false ) {
2266                                                 // Nothing else interesting
2267                                                 wfProfileOut( $fname );
2268                                                 return false;
2269                                         }
2270                                         $lt = strpos( $str, '<', $i );
2271                                         if( $stack === 0 ) {
2272                                                 if( $lt === false || $colon < $lt ) {
2273                                                         // We found it!
2274                                                         $before = substr( $str, 0, $colon );
2275                                                         $after = substr( $str, $colon + 1 );
2276                                                         wfProfileOut( $fname );
2277                                                         return $i;
2278                                                 }
2279                                         }
2280                                         if( $lt === false ) {
2281                                                 // Nothing else interesting to find; abort!
2282                                                 // We're nested, but there's no close tags left. Abort!
2283                                                 break 2;
2284                                         }
2285                                         // Skip ahead to next tag start
2286                                         $i = $lt;
2287                                         $state = self::COLON_STATE_TAGSTART;
2288                                 }
2289                                 break;
2290                         case 1: // self::COLON_STATE_TAG:
2291                                 // In a <tag>
2292                                 switch( $c ) {
2293                                 case ">":
2294                                         $stack++;
2295                                         $state = self::COLON_STATE_TEXT;
2296                                         break;
2297                                 case "/":
2298                                         // Slash may be followed by >?
2299                                         $state = self::COLON_STATE_TAGSLASH;
2300                                         break;
2301                                 default:
2302                                         // ignore
2303                                 }
2304                                 break;
2305                         case 2: // self::COLON_STATE_TAGSTART:
2306                                 switch( $c ) {
2307                                 case "/":
2308                                         $state = self::COLON_STATE_CLOSETAG;
2309                                         break;
2310                                 case "!":
2311                                         $state = self::COLON_STATE_COMMENT;
2312                                         break;
2313                                 case ">":
2314                                         // Illegal early close? This shouldn't happen D:
2315                                         $state = self::COLON_STATE_TEXT;
2316                                         break;
2317                                 default:
2318                                         $state = self::COLON_STATE_TAG;
2319                                 }
2320                                 break;
2321                         case 3: // self::COLON_STATE_CLOSETAG:
2322                                 // In a </tag>
2323                                 if( $c == ">" ) {
2324                                         $stack--;
2325                                         if( $stack < 0 ) {
2326                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2327                                                 wfProfileOut( $fname );
2328                                                 return false;
2329                                         }
2330                                         $state = self::COLON_STATE_TEXT;
2331                                 }
2332                                 break;
2333                         case self::COLON_STATE_TAGSLASH:
2334                                 if( $c == ">" ) {
2335                                         // Yes, a self-closed tag <blah/>
2336                                         $state = self::COLON_STATE_TEXT;
2337                                 } else {
2338                                         // Probably we're jumping the gun, and this is an attribute
2339                                         $state = self::COLON_STATE_TAG;
2340                                 }
2341                                 break;
2342                         case 5: // self::COLON_STATE_COMMENT:
2343                                 if( $c == "-" ) {
2344                                         $state = self::COLON_STATE_COMMENTDASH;
2345                                 }
2346                                 break;
2347                         case self::COLON_STATE_COMMENTDASH:
2348                                 if( $c == "-" ) {
2349                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2350                                 } else {
2351                                         $state = self::COLON_STATE_COMMENT;
2352                                 }
2353                                 break;
2354                         case self::COLON_STATE_COMMENTDASHDASH:
2355                                 if( $c == ">" ) {
2356                                         $state = self::COLON_STATE_TEXT;
2357                                 } else {
2358                                         $state = self::COLON_STATE_COMMENT;
2359                                 }
2360                                 break;
2361                         default:
2362                                 throw new MWException( "State machine error in $fname" );
2363                         }
2364                 }
2365                 if( $stack > 0 ) {
2366                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2367                         return false;
2368                 }
2369                 wfProfileOut( $fname );
2370                 return false;
2371         }
2372
2373         /**
2374          * Return value of a magic variable (like PAGENAME)
2375          *
2376          * @private
2377          */
2378         function getVariableValue( $index ) {
2379                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2380
2381                 /**
2382                  * Some of these require message or data lookups and can be
2383                  * expensive to check many times.
2384                  */
2385                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2386                         if ( isset( $this->mVarCache[$index] ) ) {
2387                                 return $this->mVarCache[$index];
2388                         }
2389                 }
2390
2391                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2392                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2393
2394                 # Use the time zone
2395                 global $wgLocaltimezone;
2396                 if ( isset( $wgLocaltimezone ) ) {
2397                         $oldtz = getenv( 'TZ' );
2398                         putenv( 'TZ='.$wgLocaltimezone );
2399                 }
2400
2401                 wfSuppressWarnings(); // E_STRICT system time bitching
2402                 $localTimestamp = date( 'YmdHis', $ts );
2403                 $localMonth = date( 'm', $ts );
2404                 $localMonthName = date( 'n', $ts );
2405                 $localDay = date( 'j', $ts );
2406                 $localDay2 = date( 'd', $ts );
2407                 $localDayOfWeek = date( 'w', $ts );
2408                 $localWeek = date( 'W', $ts );
2409                 $localYear = date( 'Y', $ts );
2410                 $localHour = date( 'H', $ts );
2411                 if ( isset( $wgLocaltimezone ) ) {
2412                         putenv( 'TZ='.$oldtz );
2413                 }
2414                 wfRestoreWarnings();
2415
2416                 switch ( $index ) {
2417                         case 'currentmonth':
2418                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2419                         case 'currentmonthname':
2420                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2421                         case 'currentmonthnamegen':
2422                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2423                         case 'currentmonthabbrev':
2424                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2425                         case 'currentday':
2426                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2427                         case 'currentday2':
2428                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2429                         case 'localmonth':
2430                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2431                         case 'localmonthname':
2432                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2433                         case 'localmonthnamegen':
2434                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2435                         case 'localmonthabbrev':
2436                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2437                         case 'localday':
2438                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2439                         case 'localday2':
2440                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2441                         case 'pagename':
2442                                 return wfEscapeWikiText( $this->mTitle->getText() );
2443                         case 'pagenamee':
2444                                 return $this->mTitle->getPartialURL();
2445                         case 'fullpagename':
2446                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2447                         case 'fullpagenamee':
2448                                 return $this->mTitle->getPrefixedURL();
2449                         case 'subpagename':
2450                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2451                         case 'subpagenamee':
2452                                 return $this->mTitle->getSubpageUrlForm();
2453                         case 'basepagename':
2454                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2455                         case 'basepagenamee':
2456                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2457                         case 'talkpagename':
2458                                 if( $this->mTitle->canTalk() ) {
2459                                         $talkPage = $this->mTitle->getTalkPage();
2460                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2461                                 } else {
2462                                         return '';
2463                                 }
2464                         case 'talkpagenamee':
2465                                 if( $this->mTitle->canTalk() ) {
2466                                         $talkPage = $this->mTitle->getTalkPage();
2467                                         return $talkPage->getPrefixedUrl();
2468                                 } else {
2469                                         return '';
2470                                 }
2471                         case 'subjectpagename':
2472                                 $subjPage = $this->mTitle->getSubjectPage();
2473                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2474                         case 'subjectpagenamee':
2475                                 $subjPage = $this->mTitle->getSubjectPage();
2476                                 return $subjPage->getPrefixedUrl();
2477                         case 'revisionid':
2478                                 // Let the edit saving system know we should parse the page
2479                                 // *after* a revision ID has been assigned.
2480                                 $this->mOutput->setFlag( 'vary-revision' );
2481                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2482                                 return $this->mRevisionId;
2483                         case 'revisionday':
2484                                 // Let the edit saving system know we should parse the page
2485                                 // *after* a revision ID has been assigned. This is for null edits.
2486                                 $this->mOutput->setFlag( 'vary-revision' );
2487                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2488                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2489                         case 'revisionday2':
2490                                 // Let the edit saving system know we should parse the page
2491                                 // *after* a revision ID has been assigned. This is for null edits.
2492                                 $this->mOutput->setFlag( 'vary-revision' );
2493                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2494                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2495                         case 'revisionmonth':
2496                                 // Let the edit saving system know we should parse the page
2497                                 // *after* a revision ID has been assigned. This is for null edits.
2498                                 $this->mOutput->setFlag( 'vary-revision' );
2499                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2500                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2501                         case 'revisionyear':
2502                                 // Let the edit saving system know we should parse the page
2503                                 // *after* a revision ID has been assigned. This is for null edits.
2504                                 $this->mOutput->setFlag( 'vary-revision' );
2505                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2506                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2507                         case 'revisiontimestamp':
2508                                 // Let the edit saving system know we should parse the page
2509                                 // *after* a revision ID has been assigned. This is for null edits.
2510                                 $this->mOutput->setFlag( 'vary-revision' );
2511                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2512                                 return $this->getRevisionTimestamp();
2513                         case 'namespace':
2514                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2515                         case 'namespacee':
2516                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2517                         case 'talkspace':
2518                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2519                         case 'talkspacee':
2520                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2521                         case 'subjectspace':
2522                                 return $this->mTitle->getSubjectNsText();
2523                         case 'subjectspacee':
2524                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2525                         case 'currentdayname':
2526                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2527                         case 'currentyear':
2528                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2529                         case 'currenttime':
2530                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2531                         case 'currenthour':
2532                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2533                         case 'currentweek':
2534                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2535                                 // int to remove the padding
2536                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2537                         case 'currentdow':
2538                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2539                         case 'localdayname':
2540                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2541                         case 'localyear':
2542                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2543                         case 'localtime':
2544                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2545                         case 'localhour':
2546                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2547                         case 'localweek':
2548                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2549                                 // int to remove the padding
2550                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2551                         case 'localdow':
2552                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2553                         case 'numberofarticles':
2554                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2555                         case 'numberoffiles':
2556                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2557                         case 'numberofusers':
2558                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2559                         case 'numberofpages':
2560                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2561                         case 'numberofadmins':
2562                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2563                         case 'numberofedits':
2564                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2565                         case 'currenttimestamp':
2566                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2567                         case 'localtimestamp':
2568                                 return $this->mVarCache[$index] = $localTimestamp;
2569                         case 'currentversion':
2570                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2571                         case 'sitename':
2572                                 return $wgSitename;
2573                         case 'server':
2574                                 return $wgServer;
2575                         case 'servername':
2576                                 return $wgServerName;
2577                         case 'scriptpath':
2578                                 return $wgScriptPath;
2579                         case 'directionmark':
2580                                 return $wgContLang->getDirMark();
2581                         case 'contentlanguage':
2582                                 global $wgContLanguageCode;
2583                                 return $wgContLanguageCode;
2584                         default:
2585                                 $ret = null;
2586                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2587                                         return $ret;
2588                                 else
2589                                         return null;
2590                 }
2591         }
2592
2593         /**
2594          * initialise the magic variables (like CURRENTMONTHNAME)
2595          *
2596          * @private
2597          */
2598         function initialiseVariables() {
2599                 $fname = 'Parser::initialiseVariables';
2600                 wfProfileIn( $fname );
2601                 $variableIDs = MagicWord::getVariableIDs();
2602
2603                 $this->mVariables = new MagicWordArray( $variableIDs );
2604                 wfProfileOut( $fname );
2605         }
2606
2607         /**
2608          * Preprocess some wikitext and return the document tree.
2609          * This is the ghost of replace_variables().
2610          *
2611          * @param string $text The text to parse
2612          * @param integer flags Bitwise combination of:
2613          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2614          *                                     included. Default is to assume a direct page view.
2615          *
2616          * The generated DOM tree must depend only on the input text and the flags.
2617          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2618          *
2619          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2620          * change in the DOM tree for a given text, must be passed through the section identifier
2621          * in the section edit link and thus back to extractSections().
2622          *
2623          * The output of this function is currently only cached in process memory, but a persistent
2624          * cache may be implemented at a later date which takes further advantage of these strict
2625          * dependency requirements.
2626          *
2627          * @private
2628          */
2629         function preprocessToDom ( $text, $flags = 0 ) {
2630                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2631                 return $dom;
2632         }
2633
2634         /*
2635          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2636          */
2637         public static function splitWhitespace( $s ) {
2638                 $ltrimmed = ltrim( $s );
2639                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2640                 $trimmed = rtrim( $ltrimmed );
2641                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2642                 if ( $diff > 0 ) {
2643                         $w2 = substr( $ltrimmed, -$diff );
2644                 } else {
2645                         $w2 = '';
2646                 }
2647                 return array( $w1, $trimmed, $w2 );
2648         }
2649
2650         /**
2651          * Replace magic variables, templates, and template arguments
2652          * with the appropriate text. Templates are substituted recursively,
2653          * taking care to avoid infinite loops.
2654          *
2655          * Note that the substitution depends on value of $mOutputType:
2656          *  self::OT_WIKI: only {{subst:}} templates
2657          *  self::OT_PREPROCESS: templates but not extension tags
2658          *  self::OT_HTML: all templates and extension tags
2659          *
2660          * @param string $tex The text to transform
2661          * @param PPFrame $frame Object describing the arguments passed to the template
2662          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2663          * @private
2664          */
2665         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2666                 # Prevent too big inclusions
2667                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2668                         return $text;
2669                 }
2670
2671                 $fname = __METHOD__;
2672                 wfProfileIn( $fname );
2673
2674                 if ( $frame === false ) {
2675                         $frame = $this->getPreprocessor()->newFrame();
2676                 } elseif ( !( $frame instanceof PPFrame ) ) {
2677                         throw new MWException( __METHOD__ . ' called using the old argument format' );
2678                 }
2679
2680                 $dom = $this->preprocessToDom( $text );
2681                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2682                 $text = $frame->expand( $dom, $flags );
2683
2684                 wfProfileOut( $fname );
2685                 return $text;
2686         }
2687
2688         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2689         static function createAssocArgs( $args ) {
2690                 $assocArgs = array();
2691                 $index = 1;
2692                 foreach( $args as $arg ) {
2693                         $eqpos = strpos( $arg, '=' );
2694                         if ( $eqpos === false ) {
2695                                 $assocArgs[$index++] = $arg;
2696                         } else {
2697                                 $name = trim( substr( $arg, 0, $eqpos ) );
2698                                 $value = trim( substr( $arg, $eqpos+1 ) );
2699                                 if ( $value === false ) {
2700                                         $value = '';
2701                                 }
2702                                 if ( $name !== false ) {
2703                                         $assocArgs[$name] = $value;
2704                                 }
2705                         }
2706                 }
2707
2708                 return $assocArgs;
2709         }
2710
2711         /**
2712          * Return the text of a template, after recursively
2713          * replacing any variables or templates within the template.
2714          *
2715          * @param array $piece The parts of the template
2716          *  $piece['title']: the title, i.e. the part before the |
2717          *  $piece['parts']: the parameter array
2718          *  $piece['lineStart']: whether the brace was at the start of a line
2719          * @param PPFrame The current frame, contains template arguments
2720          * @return string the text of the template
2721          * @private
2722          */
2723         function braceSubstitution( $piece, $frame ) {
2724                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2725                 $fname = __METHOD__;
2726                 wfProfileIn( $fname );
2727                 wfProfileIn( __METHOD__.'-setup' );
2728
2729                 # Flags
2730                 $found = false;             # $text has been filled
2731                 $nowiki = false;            # wiki markup in $text should be escaped
2732                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2733                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2734                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2735                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2736
2737                 # Title object, where $text came from
2738                 $title = NULL;
2739
2740                 # $part1 is the bit before the first |, and must contain only title characters.
2741                 # Various prefixes will be stripped from it later.
2742                 $titleWithSpaces = $frame->expand( $piece['title'] );
2743                 $part1 = trim( $titleWithSpaces );
2744                 $titleText = false;
2745
2746                 # Original title text preserved for various purposes
2747                 $originalTitle = $part1;
2748
2749                 # $args is a list of argument nodes, starting from index 0, not including $part1
2750                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2751                 wfProfileOut( __METHOD__.'-setup' );
2752
2753                 # SUBST
2754                 wfProfileIn( __METHOD__.'-modifiers' );
2755                 if ( !$found ) {
2756                         $mwSubst =& MagicWord::get( 'subst' );
2757                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2758                                 # One of two possibilities is true:
2759                                 # 1) Found SUBST but not in the PST phase
2760                                 # 2) Didn't find SUBST and in the PST phase
2761                                 # In either case, return without further processing
2762                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2763                                 $isLocalObj = true;
2764                                 $found = true;
2765                         }
2766                 }
2767
2768                 # Variables
2769                 if ( !$found && $args->getLength() == 0 ) {
2770                         $id = $this->mVariables->matchStartToEnd( $part1 );
2771                         if ( $id !== false ) {
2772                                 $text = $this->getVariableValue( $id );
2773                                 if (MagicWord::getCacheTTL($id)>-1)
2774                                         $this->mOutput->mContainsOldMagic = true;
2775                                 $found = true;
2776                         }
2777                 }
2778
2779                 # MSG, MSGNW and RAW
2780                 if ( !$found ) {
2781                         # Check for MSGNW:
2782                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2783                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2784                                 $nowiki = true;
2785                         } else {
2786                                 # Remove obsolete MSG:
2787                                 $mwMsg =& MagicWord::get( 'msg' );
2788                                 $mwMsg->matchStartAndRemove( $part1 );
2789                         }
2790
2791                         # Check for RAW:
2792                         $mwRaw =& MagicWord::get( 'raw' );
2793                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2794                                 $forceRawInterwiki = true;
2795                         }
2796                 }
2797                 wfProfileOut( __METHOD__.'-modifiers' );
2798
2799                 # Parser functions
2800                 if ( !$found ) {
2801                         wfProfileIn( __METHOD__ . '-pfunc' );
2802
2803                         $colonPos = strpos( $part1, ':' );
2804                         if ( $colonPos !== false ) {
2805                                 # Case sensitive functions
2806                                 $function = substr( $part1, 0, $colonPos );
2807                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2808                                         $function = $this->mFunctionSynonyms[1][$function];
2809                                 } else {
2810                                         # Case insensitive functions
2811                                         $function = strtolower( $function );
2812                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2813                                                 $function = $this->mFunctionSynonyms[0][$function];
2814                                         } else {
2815                                                 $function = false;
2816                                         }
2817                                 }
2818                                 if ( $function ) {
2819                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2820                                         $initialArgs = array( &$this );
2821                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2822                                         if ( $flags & SFH_OBJECT_ARGS ) {
2823                                                 # Add a frame parameter, and pass the arguments as an array
2824                                                 $allArgs = $initialArgs;
2825                                                 $allArgs[] = $frame;
2826                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2827                                                         $funcArgs[] = $args->item( $i );
2828                                                 }
2829                                                 $allArgs[] = $funcArgs;
2830                                         } else {
2831                                                 # Convert arguments to plain text
2832                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2833                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2834                                                 }
2835                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2836                                         }
2837
2838                                         # Workaround for PHP bug 35229 and similar
2839                                         if ( !is_callable( $callback ) ) {
2840                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2841                                         }
2842                                         $result = call_user_func_array( $callback, $allArgs );
2843                                         $found = true;
2844
2845                                         if ( is_array( $result ) ) {
2846                                                 if ( isset( $result[0] ) ) {
2847                                                         $text = $result[0];
2848                                                         unset( $result[0] );
2849                                                 }
2850
2851                                                 // Extract flags into the local scope
2852                                                 // This allows callers to set flags such as nowiki, found, etc.
2853                                                 extract( $result );
2854                                         } else {
2855                                                 $text = $result;
2856                                         }
2857                                 }
2858                         }
2859                         wfProfileOut( __METHOD__ . '-pfunc' );
2860                 }
2861
2862                 # Finish mangling title and then check for loops.
2863                 # Set $title to a Title object and $titleText to the PDBK
2864                 if ( !$found ) {
2865                         $ns = NS_TEMPLATE;
2866                         # Split the title into page and subpage
2867                         $subpage = '';
2868                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2869                         if ($subpage !== '') {
2870                                 $ns = $this->mTitle->getNamespace();
2871                         }
2872                         $title = Title::newFromText( $part1, $ns );
2873                         if ( $title ) {
2874                                 $titleText = $title->getPrefixedText();
2875                                 # Check for language variants if the template is not found
2876                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2877                                         $wgContLang->findVariantLink($part1, $title);
2878                                 }
2879                                 # Do infinite loop check
2880                                 if ( !$frame->loopCheck( $title ) ) {
2881                                         $found = true;
2882                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2883                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2884                                 }
2885                                 # Do recursion depth check
2886                                 $limit = $this->mOptions->getMaxTemplateDepth();
2887                                 if ( $frame->depth >= $limit ) {
2888                                         $found = true;
2889                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2890                                 }
2891                         }
2892                 }
2893
2894                 # Load from database
2895                 if ( !$found && $title ) {
2896                         wfProfileIn( __METHOD__ . '-loadtpl' );
2897                         if ( !$title->isExternal() ) {
2898                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2899                                         $text = SpecialPage::capturePath( $title );
2900                                         if ( is_string( $text ) ) {
2901                                                 $found = true;
2902                                                 $isHTML = true;
2903                                                 $this->disableCache();
2904                                         }
2905                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2906                                         $found = false; //access denied
2907                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2908                                 } else {
2909                                         list( $text, $title ) = $this->getTemplateDom( $title );
2910                                         if ( $text !== false ) {
2911                                                 $found = true;
2912                                                 $isChildObj = true;
2913                                         }
2914                                 }
2915
2916                                 # If the title is valid but undisplayable, make a link to it
2917                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2918                                         $text = "[[:$titleText]]";
2919                                         $found = true;
2920                                 }
2921                         } elseif ( $title->isTrans() ) {
2922                                 // Interwiki transclusion
2923                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2924                                         $text = $this->interwikiTransclude( $title, 'render' );
2925                                         $isHTML = true;
2926                                 } else {
2927                                         $text = $this->interwikiTransclude( $title, 'raw' );
2928                                         // Preprocess it like a template
2929                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2930                                         $isChildObj = true;
2931                                 }
2932                                 $found = true;
2933                         }
2934                         wfProfileOut( __METHOD__ . '-loadtpl' );
2935                 }
2936
2937                 # If we haven't found text to substitute by now, we're done
2938                 # Recover the source wikitext and return it
2939                 if ( !$found ) {
2940                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2941                         wfProfileOut( $fname );
2942                         return array( 'object' => $text );
2943                 }
2944
2945                 # Expand DOM-style return values in a child frame
2946                 if ( $isChildObj ) {
2947                         # Clean up argument array
2948                         $newFrame = $frame->newChild( $args, $title );
2949
2950                         if ( $nowiki ) {
2951                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2952                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2953                                 # Expansion is eligible for the empty-frame cache
2954                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2955                                         $text = $this->mTplExpandCache[$titleText];
2956                                 } else {
2957                                         $text = $newFrame->expand( $text );
2958                                         $this->mTplExpandCache[$titleText] = $text;
2959                                 }
2960                         } else {
2961                                 # Uncached expansion
2962                                 $text = $newFrame->expand( $text );
2963                         }
2964                 }
2965                 if ( $isLocalObj && $nowiki ) {
2966                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2967                         $isLocalObj = false;
2968                 }
2969
2970                 # Replace raw HTML by a placeholder
2971                 # Add a blank line preceding, to prevent it from mucking up
2972                 # immediately preceding headings
2973                 if ( $isHTML ) {
2974                         $text = "\n\n" . $this->insertStripItem( $text );
2975                 }
2976                 # Escape nowiki-style return values
2977                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2978                         $text = wfEscapeWikiText( $text );
2979                 }
2980                 # Bug 529: if the template begins with a table or block-level
2981                 # element, it should be treated as beginning a new line.
2982                 # This behaviour is somewhat controversial.
2983                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2984                         $text = "\n" . $text;
2985                 }
2986
2987                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2988                         # Error, oversize inclusion
2989                         $text = "[[$originalTitle]]" .
2990                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2991                 }
2992
2993                 if ( $isLocalObj ) {
2994                         $ret = array( 'object' => $text );
2995                 } else {
2996                         $ret = array( 'text' => $text );
2997                 }
2998
2999                 wfProfileOut( $fname );
3000                 return $ret;
3001         }
3002
3003         /**
3004          * Get the semi-parsed DOM representation of a template with a given title,
3005          * and its redirect destination title. Cached.
3006          */
3007         function getTemplateDom( $title ) {
3008                 $cacheTitle = $title;
3009                 $titleText = $title->getPrefixedDBkey();
3010
3011                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3012                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3013                         $title = Title::makeTitle( $ns, $dbk );
3014                         $titleText = $title->getPrefixedDBkey();
3015                 }
3016                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3017                         return array( $this->mTplDomCache[$titleText], $title );
3018                 }
3019
3020                 // Cache miss, go to the database
3021                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3022
3023                 if ( $text === false ) {
3024                         $this->mTplDomCache[$titleText] = false;
3025                         return array( false, $title );
3026                 }
3027
3028                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3029                 $this->mTplDomCache[ $titleText ] = $dom;
3030
3031                 if (! $title->equals($cacheTitle)) {
3032                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3033                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3034                 }
3035
3036                 return array( $dom, $title );
3037         }
3038
3039         /**
3040          * Fetch the unparsed text of a template and register a reference to it.
3041          */
3042         function fetchTemplateAndTitle( $title ) {
3043                 $templateCb = $this->mOptions->getTemplateCallback();
3044                 $stuff = call_user_func( $templateCb, $title );
3045                 $text = $stuff['text'];
3046                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3047                 if ( isset( $stuff['deps'] ) ) {
3048                         foreach ( $stuff['deps'] as $dep ) {
3049                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3050                         }
3051                 }
3052                 return array($text,$finalTitle);
3053         }
3054
3055         function fetchTemplate( $title ) {
3056                 $rv = $this->fetchTemplateAndTitle($title);
3057                 return $rv[0];
3058         }
3059
3060         /**
3061          * Static function to get a template
3062          * Can be overridden via ParserOptions::setTemplateCallback().
3063          */
3064         static function statelessFetchTemplate( $title ) {
3065                 $text = $skip = false;
3066                 $finalTitle = $title;
3067                 $deps = array();
3068
3069                 // Loop to fetch the article, with up to 1 redirect
3070                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3071                         # Give extensions a chance to select the revision instead
3072                         $id = false; // Assume current
3073                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3074
3075                         if( $skip ) {
3076                                 $text = false;
3077                                 $deps[] = array(
3078                                         'title' => $title,
3079                                         'page_id' => $title->getArticleID(),
3080                                         'rev_id' => null );
3081                                 break;
3082                         }
3083                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3084                         $rev_id = $rev ? $rev->getId() : 0;
3085
3086                         $deps[] = array(
3087                                 'title' => $title,
3088                                 'page_id' => $title->getArticleID(),
3089                                 'rev_id' => $rev_id );
3090
3091                         if( $rev ) {
3092                                 $text = $rev->getText();
3093                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3094                                 global $wgLang;
3095                                 $message = $wgLang->lcfirst( $title->getText() );
3096                                 $text = wfMsgForContentNoTrans( $message );
3097                                 if( wfEmptyMsg( $message, $text ) ) {
3098                                         $text = false;
3099                                         break;
3100                                 }
3101                         } else {
3102                                 break;
3103                         }
3104                         if ( $text === false ) {
3105                                 break;
3106                         }
3107                         // Redirect?
3108                         $finalTitle = $title;
3109                         $title = Title::newFromRedirect( $text );
3110                 }
3111                 return array(
3112                         'text' => $text,
3113                         'finalTitle' => $finalTitle,
3114                         'deps' => $deps );
3115         }
3116
3117         /**
3118          * Transclude an interwiki link.
3119          */
3120         function interwikiTransclude( $title, $action ) {
3121                 global $wgEnableScaryTranscluding;
3122
3123                 if (!$wgEnableScaryTranscluding)
3124                         return wfMsg('scarytranscludedisabled');
3125
3126                 $url = $title->getFullUrl( "action=$action" );
3127
3128                 if (strlen($url) > 255)
3129                         return wfMsg('scarytranscludetoolong');
3130                 return $this->fetchScaryTemplateMaybeFromCache($url);
3131         }
3132
3133         function fetchScaryTemplateMaybeFromCache($url) {
3134                 global $wgTranscludeCacheExpiry;
3135                 $dbr = wfGetDB(DB_SLAVE);
3136                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3137                                 array('tc_url' => $url));
3138                 if ($obj) {
3139                         $time = $obj->tc_time;
3140                         $text = $obj->tc_contents;
3141                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3142                                 return $text;
3143                         }
3144                 }
3145
3146                 $text = Http::get($url);
3147                 if (!$text)
3148                         return wfMsg('scarytranscludefailed', $url);
3149
3150                 $dbw = wfGetDB(DB_MASTER);
3151                 $dbw->replace('transcache', array('tc_url'), array(
3152                         'tc_url' => $url,
3153                         'tc_time' => time(),
3154                         'tc_contents' => $text));
3155                 return $text;
3156         }
3157
3158
3159         /**
3160          * Triple brace replacement -- used for template arguments
3161          * @private
3162          */
3163         function argSubstitution( $piece, $frame ) {
3164                 wfProfileIn( __METHOD__ );
3165
3166                 $error = false;
3167                 $parts = $piece['parts'];
3168                 $nameWithSpaces = $frame->expand( $piece['title'] );
3169                 $argName = trim( $nameWithSpaces );
3170                 $object = false;
3171                 $text = $frame->getArgument( $argName );
3172                 if (  $text === false && $parts->getLength() > 0
3173                   && (
3174                     $this->ot['html']
3175                     || $this->ot['pre']
3176                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3177                   )
3178                 ) {
3179                         # No match in frame, use the supplied default
3180                         $object = $parts->item( 0 )->getChildren();
3181                 }
3182                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3183                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3184                 }
3185
3186                 if ( $text === false && $object === false ) {
3187                         # No match anywhere
3188                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3189                 }
3190                 if ( $error !== false ) {
3191                         $text .= $error;
3192                 }
3193                 if ( $object !== false ) {
3194                         $ret = array( 'object' => $object );
3195                 } else {
3196                         $ret = array( 'text' => $text );
3197                 }
3198
3199                 wfProfileOut( __METHOD__ );
3200                 return $ret;
3201         }
3202
3203         /**
3204          * Return the text to be used for a given extension tag.
3205          * This is the ghost of strip().
3206          *
3207          * @param array $params Associative array of parameters:
3208          *     name       PPNode for the tag name
3209          *     attr       PPNode for unparsed text where tag attributes are thought to be
3210          *     attributes Optional associative array of parsed attributes
3211          *     inner      Contents of extension element
3212          *     noClose    Original text did not have a close tag
3213          * @param PPFrame $frame
3214          */
3215         function extensionSubstitution( $params, $frame ) {
3216                 global $wgRawHtml, $wgContLang;
3217                 static $n = 1;
3218
3219                 $name = $frame->expand( $params['name'] );
3220                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3221                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3222
3223                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
3224
3225                 if ( $this->ot['html'] ) {
3226                         $name = strtolower( $name );
3227
3228                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3229                         if ( isset( $params['attributes'] ) ) {
3230                                 $attributes = $attributes + $params['attributes'];
3231                         }
3232                         switch ( $name ) {
3233                                 case 'html':
3234                                         if( $wgRawHtml ) {
3235                                                 $output = $content;
3236                                                 break;
3237                                         } else {
3238                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3239                                         }
3240                                 case 'nowiki':
3241                                         $output = Xml::escapeTagsOnly( $content );
3242                                         break;
3243                                 case 'math':
3244                                         $output = $wgContLang->armourMath(
3245                                                 MathRenderer::renderMath( $content, $attributes ) );
3246                                         break;
3247                                 case 'gallery':
3248                                         $output = $this->renderImageGallery( $content, $attributes );
3249                                         break;
3250                                 default:
3251                                         if( isset( $this->mTagHooks[$name] ) ) {
3252                                                 # Workaround for PHP bug 35229 and similar
3253                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3254                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3255                                                 }
3256                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3257                                                         array( $content, $attributes, $this ) );
3258                                         } else {
3259                                                 throw new MWException( "Invalid call hook $name" );
3260                                         }
3261                         }
3262                 } else {
3263                         if ( is_null( $attrText ) ) {
3264                                 $attrText = '';
3265                         }
3266                         if ( isset( $params['attributes'] ) ) {
3267                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3268                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3269                                                 htmlspecialchars( $attrValue ) . '"';
3270                                 }
3271                         }
3272                         if ( $content === null ) {
3273                                 $output = "<$name$attrText/>";
3274                         } else {
3275                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3276                                 $output = "<$name$attrText>$content$close";
3277                         }
3278                 }
3279
3280                 if ( $name == 'html' || $name == 'nowiki' ) {
3281                         $this->mStripState->nowiki->setPair( $marker, $output );
3282                 } else {
3283                         $this->mStripState->general->setPair( $marker, $output );
3284                 }
3285                 return $marker;
3286         }
3287
3288         /**
3289          * Increment an include size counter
3290          *
3291          * @param string $type The type of expansion
3292          * @param integer $size The size of the text
3293          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3294          */
3295         function incrementIncludeSize( $type, $size ) {
3296                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3297                         return false;
3298                 } else {
3299                         $this->mIncludeSizes[$type] += $size;
3300                         return true;
3301                 }
3302         }
3303
3304         /**
3305          * Detect __NOGALLERY__ magic word and set a placeholder
3306          */
3307         function stripNoGallery( &$text ) {
3308                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3309                 # do not add TOC
3310                 $mw = MagicWord::get( 'nogallery' );
3311                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3312         }
3313
3314         /**
3315          * Find the first __TOC__ magic word and set a <!--MWTOC-->
3316          * placeholder that will then be replaced by the real TOC in
3317          * ->formatHeadings, this works because at this points real
3318          * comments will have already been discarded by the sanitizer.
3319          *
3320          * Any additional __TOC__ magic words left over will be discarded
3321          * as there can only be one TOC on the page.
3322          */
3323         function stripToc( $text ) {
3324                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3325                 # do not add TOC
3326                 $mw = MagicWord::get( 'notoc' );
3327                 if( $mw->matchAndRemove( $text ) ) {
3328                         $this->mShowToc = false;
3329                 }
3330
3331                 $mw = MagicWord::get( 'toc' );
3332                 if( $mw->match( $text ) ) {
3333                         $this->mShowToc = true;
3334                         $this->mForceTocPosition = true;
3335
3336                         // Set a placeholder. At the end we'll fill it in with the TOC.
3337                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3338
3339                         // Only keep the first one.
3340                         $text = $mw->replace( '', $text );
3341                 }
3342                 return $text;
3343         }
3344
3345         /**
3346          * This function accomplishes several tasks:
3347          * 1) Auto-number headings if that option is enabled
3348          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3349          * 3) Add a Table of contents on the top for users who have enabled the option
3350          * 4) Auto-anchor headings
3351          *
3352          * It loops through all headlines, collects the necessary data, then splits up the
3353          * string and re-inserts the newly formatted headlines.
3354          *
3355          * @param string $text
3356          * @param boolean $isMain
3357          * @private
3358          */
3359         function formatHeadings( $text, $isMain=true ) {
3360                 global $wgMaxTocLevel, $wgContLang;
3361
3362                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3363                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3364                         $showEditLink = 0;
3365                 } else {
3366                         $showEditLink = $this->mOptions->getEditSection();
3367                 }
3368
3369                 # Inhibit editsection links if requested in the page
3370                 $esw =& MagicWord::get( 'noeditsection' );
3371                 if( $esw->matchAndRemove( $text ) ) {
3372                         $showEditLink = 0;
3373                 }
3374
3375                 # Get all headlines for numbering them and adding funky stuff like [edit]
3376                 # links - this is for later, but we need the number of headlines right now
3377                 $matches = array();
3378                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3379
3380                 # if there are fewer than 4 headlines in the article, do not show TOC
3381                 # unless it's been explicitly enabled.
3382                 $enoughToc = $this->mShowToc &&
3383                         (($numMatches >= 4) || $this->mForceTocPosition);
3384
3385                 # Allow user to stipulate that a page should have a "new section"
3386                 # link added via __NEWSECTIONLINK__
3387                 $mw =& MagicWord::get( 'newsectionlink' );
3388                 if( $mw->matchAndRemove( $text ) )
3389                         $this->mOutput->setNewSection( true );
3390
3391                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3392                 # override above conditions and always show TOC above first header
3393                 $mw =& MagicWord::get( 'forcetoc' );
3394                 if ($mw->matchAndRemove( $text ) ) {
3395                         $this->mShowToc = true;
3396                         $enoughToc = true;
3397                 }
3398
3399                 # We need this to perform operations on the HTML
3400                 $sk = $this->mOptions->getSkin();
3401
3402                 # headline counter
3403                 $headlineCount = 0;
3404                 $numVisible = 0;
3405
3406                 # Ugh .. the TOC should have neat indentation levels which can be
3407                 # passed to the skin functions. These are determined here
3408                 $toc = '';
3409                 $full = '';
3410                 $head = array();
3411                 $sublevelCount = array();
3412                 $levelCount = array();
3413                 $toclevel = 0;
3414                 $level = 0;
3415                 $prevlevel = 0;
3416                 $toclevel = 0;
3417                 $prevtoclevel = 0;
3418                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3419                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3420                 $tocraw = array();
3421
3422                 foreach( $matches[3] as $headline ) {
3423                         $isTemplate = false;
3424                         $titleText = false;
3425                         $sectionIndex = false;
3426                         $numbering = '';
3427                         $markerMatches = array();
3428                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3429                                 $serial = $markerMatches[1];
3430                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3431                                 $isTemplate = ($titleText != $baseTitleText);
3432                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3433                         }
3434
3435                         if( $toclevel ) {
3436                                 $prevlevel = $level;
3437                                 $prevtoclevel = $toclevel;
3438                         }
3439                         $level = $matches[1][$headlineCount];
3440
3441                         if( $doNumberHeadings || $enoughToc ) {
3442
3443                                 if ( $level > $prevlevel ) {
3444                                         # Increase TOC level
3445                                         $toclevel++;
3446                                         $sublevelCount[$toclevel] = 0;
3447                                         if( $toclevel<$wgMaxTocLevel ) {
3448                                                 $prevtoclevel = $toclevel;
3449                                                 $toc .= $sk->tocIndent();
3450                                                 $numVisible++;
3451                                         }
3452                                 }
3453                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3454                                         # Decrease TOC level, find level to jump to
3455
3456                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3457                                                 # Can only go down to level 1
3458                                                 $toclevel = 1;
3459                                         } else {
3460                                                 for ($i = $toclevel; $i > 0; $i--) {
3461                                                         if ( $levelCount[$i] == $level ) {
3462                                                                 # Found last matching level
3463                                                                 $toclevel = $i;
3464                                                                 break;
3465                                                         }
3466                                                         elseif ( $levelCount[$i] < $level ) {
3467                                                                 # Found first matching level below current level
3468                                                                 $toclevel = $i + 1;
3469                                                                 break;
3470                                                         }
3471                                                 }
3472                                         }
3473                                         if( $toclevel<$wgMaxTocLevel ) {
3474                                                 if($prevtoclevel < $wgMaxTocLevel) {
3475                                                         # Unindent only if the previous toc level was shown :p
3476                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3477                                                 } else {
3478                                                         $toc .= $sk->tocLineEnd();
3479                                                 }
3480                                         }
3481                                 }
3482                                 else {
3483                                         # No change in level, end TOC line
3484                                         if( $toclevel<$wgMaxTocLevel ) {
3485                                                 $toc .= $sk->tocLineEnd();
3486                                         }
3487                                 }
3488
3489                                 $levelCount[$toclevel] = $level;
3490
3491                                 # count number of headlines for each level
3492                                 @$sublevelCount[$toclevel]++;
3493                                 $dot = 0;
3494                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3495                                         if( !empty( $sublevelCount[$i] ) ) {
3496                                                 if( $dot ) {
3497                                                         $numbering .= '.';
3498                                                 }
3499                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3500                                                 $dot = 1;
3501                                         }
3502                                 }
3503                         }
3504
3505                         # The safe header is a version of the header text safe to use for links
3506                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3507                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3508
3509                         # Remove link placeholders by the link text.
3510                         #     <!--LINK number-->
3511                         # turns into
3512                         #     link text with suffix
3513                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3514                                                             "\$this->mLinkHolders['texts'][\$1]",
3515                                                             $safeHeadline );
3516                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3517                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3518                                                             $safeHeadline );
3519
3520                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3521                         $tocline = preg_replace(
3522                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3523                                 array( '',                          '<$1>'),
3524                                 $safeHeadline
3525                         );
3526                         $tocline = trim( $tocline );
3527
3528                         # For the anchor, strip out HTML-y stuff period
3529                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3530                         $safeHeadline = trim( $safeHeadline );
3531
3532                         # Save headline for section edit hint before it's escaped
3533                         $headlineHint = $safeHeadline;
3534                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3535                         $refers[$headlineCount] = $safeHeadline;
3536
3537                         # count how many in assoc. array so we can track dupes in anchors
3538                         isset( $refers[$safeHeadline] ) ? $refers[$safeHeadline]++ : $refers[$safeHeadline] = 1;
3539                         $refcount[$headlineCount] = $refers[$safeHeadline];
3540
3541                         # Don't number the heading if it is the only one (looks silly)
3542                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3543                                 # the two are different if the line contains a link
3544                                 $headline=$numbering . ' ' . $headline;
3545                         }
3546
3547                         # Create the anchor for linking from the TOC to the section
3548                         $anchor = $safeHeadline;
3549                         if($refcount[$headlineCount] > 1 ) {
3550                                 $anchor .= '_' . $refcount[$headlineCount];
3551                         }
3552                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3553                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3554                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3555                         }
3556                         # give headline the correct <h#> tag
3557                         if( $showEditLink && $sectionIndex !== false ) {
3558                                 if( $isTemplate ) {
3559                                         # Put a T flag in the section identifier, to indicate to extractSections()
3560                                         # that sections inside <includeonly> should be counted.
3561                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3562                                 } else {
3563                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3564                                 }
3565                         } else {
3566                                 $editlink = '';
3567                         }
3568                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3569
3570                         $headlineCount++;
3571                 }
3572
3573                 $this->mOutput->setSections( $tocraw );
3574
3575                 # Never ever show TOC if no headers
3576                 if( $numVisible < 1 ) {
3577                         $enoughToc = false;
3578                 }
3579
3580                 if( $enoughToc ) {
3581                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3582                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3583                         }
3584                         $toc = $sk->tocList( $toc );
3585                 }
3586
3587                 # split up and insert constructed headlines
3588
3589                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3590                 $i = 0;
3591
3592                 foreach( $blocks as $block ) {
3593                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3594                                 # This is the [edit] link that appears for the top block of text when
3595                                 # section editing is enabled
3596
3597                                 # Disabled because it broke block formatting
3598                                 # For example, a bullet point in the top line
3599                                 # $full .= $sk->editSectionLink(0);
3600                         }
3601                         $full .= $block;
3602                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3603                                 # Top anchor now in skin
3604                                 $full = $full.$toc;
3605                         }
3606
3607                         if( !empty( $head[$i] ) ) {
3608                                 $full .= $head[$i];
3609                         }
3610                         $i++;
3611                 }
3612                 if( $this->mForceTocPosition ) {
3613                         return str_replace( '<!--MWTOC-->', $toc, $full );
3614                 } else {
3615                         return $full;
3616                 }
3617         }
3618
3619         /**
3620          * Transform wiki markup when saving a page by doing \r\n -> \n
3621          * conversion, substitting signatures, {{subst:}} templates, etc.
3622          *
3623          * @param string $text the text to transform
3624          * @param Title &$title the Title object for the current article
3625          * @param User &$user the User object describing the current user
3626          * @param ParserOptions $options parsing options
3627          * @param bool $clearState whether to clear the parser state first
3628          * @return string the altered wiki markup
3629          * @public
3630          */
3631         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3632                 $this->mOptions = $options;
3633                 $this->setTitle( $title );
3634                 $this->setOutputType( self::OT_WIKI );
3635
3636                 if ( $clearState ) {
3637                         $this->clearState();
3638                 }
3639
3640                 $pairs = array(
3641                         "\r\n" => "\n",
3642                 );
3643                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3644                 $text = $this->pstPass2( $text, $user );
3645                 $text = $this->mStripState->unstripBoth( $text );
3646                 return $text;
3647         }
3648
3649         /**
3650          * Pre-save transform helper function
3651          * @private
3652          */
3653         function pstPass2( $text, $user ) {
3654                 global $wgContLang, $wgLocaltimezone;
3655
3656                 /* Note: This is the timestamp saved as hardcoded wikitext to
3657                  * the database, we use $wgContLang here in order to give
3658                  * everyone the same signature and use the default one rather
3659                  * than the one selected in each user's preferences.
3660                  */
3661                 if ( isset( $wgLocaltimezone ) ) {
3662                         $oldtz = getenv( 'TZ' );
3663                         putenv( 'TZ='.$wgLocaltimezone );
3664                 }
3665                 $d = $wgContLang->timeanddate( $this->mOptions->getTimestamp(), false, false) .
3666                   ' (' . date( 'T' ) . ')';
3667                 if ( isset( $wgLocaltimezone ) ) {
3668                         putenv( 'TZ='.$oldtz );
3669                 }
3670
3671                 # Variable replacement
3672                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3673                 $text = $this->replaceVariables( $text );
3674
3675                 # Signatures
3676                 $sigText = $this->getUserSig( $user );
3677                 $text = strtr( $text, array(
3678                         '~~~~~' => $d,
3679                         '~~~~' => "$sigText $d",
3680                         '~~~' => $sigText
3681                 ) );
3682
3683                 # Context links: [[|name]] and [[name (context)|]]
3684                 #
3685                 global $wgLegalTitleChars;
3686                 $tc = "[$wgLegalTitleChars]";
3687                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3688
3689                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3690                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3691                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3692
3693                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3694                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3695                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3696
3697                 $t = $this->mTitle->getText();
3698                 $m = array();
3699                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3700                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3701                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3702                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3703                 } else {
3704                         # if there's no context, don't bother duplicating the title
3705                         $text = preg_replace( $p2, '[[\\1]]', $text );
3706                 }
3707
3708                 # Trim trailing whitespace
3709                 $text = rtrim( $text );
3710
3711                 return $text;
3712         }
3713
3714         /**
3715          * Fetch the user's signature text, if any, and normalize to
3716          * validated, ready-to-insert wikitext.
3717          *
3718          * @param User $user
3719          * @return string
3720          * @private
3721          */
3722         function getUserSig( &$user ) {
3723                 global $wgMaxSigChars;
3724
3725                 $username = $user->getName();
3726                 $nickname = $user->getOption( 'nickname' );
3727                 $nickname = $nickname === '' ? $username : $nickname;
3728
3729                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3730                         $nickname = $username;
3731                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3732                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3733                         # Sig. might contain markup; validate this
3734                         if( $this->validateSig( $nickname ) !== false ) {
3735                                 # Validated; clean up (if needed) and return it
3736                                 return $this->cleanSig( $nickname, true );
3737                         } else {
3738                                 # Failed to validate; fall back to the default
3739                                 $nickname = $username;
3740                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3741                         }
3742                 }
3743
3744                 // Make sure nickname doesnt get a sig in a sig
3745                 $nickname = $this->cleanSigInSig( $nickname );
3746
3747                 # If we're still here, make it a link to the user page
3748                 $userText = wfEscapeWikiText( $username );
3749                 $nickText = wfEscapeWikiText( $nickname );
3750                 if ( $user->isAnon() )  {
3751                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3752                 } else {
3753                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3754                 }
3755         }
3756
3757         /**
3758          * Check that the user's signature contains no bad XML
3759          *
3760          * @param string $text
3761          * @return mixed An expanded string, or false if invalid.
3762          */
3763         function validateSig( $text ) {
3764                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3765         }
3766
3767         /**
3768          * Clean up signature text
3769          *
3770          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3771          * 2) Substitute all transclusions
3772          *
3773          * @param string $text
3774          * @param $parsing Whether we're cleaning (preferences save) or parsing
3775          * @return string Signature text
3776          */
3777         function cleanSig( $text, $parsing = false ) {
3778                 if ( !$parsing ) {
3779                         global $wgTitle;
3780                         $this->clearState();
3781                         $this->setTitle( $wgTitle );
3782                         $this->mOptions = new ParserOptions;
3783                         $this->setOutputType = self::OT_PREPROCESS;
3784                 }
3785
3786                 # FIXME: regex doesn't respect extension tags or nowiki
3787                 #  => Move this logic to braceSubstitution()
3788                 $substWord = MagicWord::get( 'subst' );
3789                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3790                 $substText = '{{' . $substWord->getSynonym( 0 );
3791
3792                 $text = preg_replace( $substRegex, $substText, $text );
3793                 $text = $this->cleanSigInSig( $text );
3794                 $dom = $this->preprocessToDom( $text );
3795                 $frame = $this->getPreprocessor()->newFrame();
3796                 $text = $frame->expand( $dom );
3797
3798                 if ( !$parsing ) {
3799                         $text = $this->mStripState->unstripBoth( $text );
3800                 }
3801
3802                 return $text;
3803         }
3804
3805         /**
3806          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3807          * @param string $text
3808          * @return string Signature text with /~{3,5}/ removed
3809          */
3810         function cleanSigInSig( $text ) {
3811                 $text = preg_replace( '/~{3,5}/', '', $text );
3812                 return $text;
3813         }
3814
3815         /**
3816          * Set up some variables which are usually set up in parse()
3817          * so that an external function can call some class members with confidence
3818          * @public
3819          */
3820         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3821                 $this->setTitle( $title );
3822                 $this->mOptions = $options;
3823                 $this->setOutputType( $outputType );
3824                 if ( $clearState ) {
3825                         $this->clearState();
3826                 }
3827         }
3828
3829         /**
3830          * Wrapper for preprocess()
3831          *
3832          * @param string $text the text to preprocess
3833          * @param ParserOptions $options  options
3834          * @return string
3835          * @public
3836          */
3837         function transformMsg( $text, $options ) {
3838                 global $wgTitle;
3839                 static $executing = false;
3840
3841                 $fname = "Parser::transformMsg";
3842
3843                 # Guard against infinite recursion
3844                 if ( $executing ) {
3845                         return $text;
3846                 }
3847                 $executing = true;
3848
3849                 wfProfileIn($fname);
3850                 $text = $this->preprocess( $text, $wgTitle, $options );
3851
3852                 $executing = false;
3853                 wfProfileOut($fname);
3854                 return $text;
3855         }
3856
3857         /**
3858          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3859          * The callback should have the following form:
3860          *    function myParserHook( $text, $params, &$parser ) { ... }
3861          *
3862          * Transform and return $text. Use $parser for any required context, e.g. use
3863          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3864          *
3865          * @public
3866          *
3867          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3868          * @param mixed $callback The callback function (and object) to use for the tag
3869          *
3870          * @return The old value of the mTagHooks array associated with the hook
3871          */
3872         function setHook( $tag, $callback ) {
3873                 $tag = strtolower( $tag );
3874                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3875                 $this->mTagHooks[$tag] = $callback;
3876                 $this->mStripList[] = $tag;
3877
3878                 return $oldVal;
3879         }
3880
3881         function setTransparentTagHook( $tag, $callback ) {
3882                 $tag = strtolower( $tag );
3883                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3884                 $this->mTransparentTagHooks[$tag] = $callback;
3885
3886                 return $oldVal;
3887         }
3888
3889         /**
3890          * Remove all tag hooks
3891          */
3892         function clearTagHooks() {
3893                 $this->mTagHooks = array();
3894                 $this->mStripList = $this->mDefaultStripList;
3895         }
3896
3897         /**
3898          * Create a function, e.g. {{sum:1|2|3}}
3899          * The callback function should have the form:
3900          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3901          *
3902          * The callback may either return the text result of the function, or an array with the text
3903          * in element 0, and a number of flags in the other elements. The names of the flags are
3904          * specified in the keys. Valid flags are:
3905          *   found                     The text returned is valid, stop processing the template. This
3906          *                             is on by default.
3907          *   nowiki                    Wiki markup in the return value should be escaped
3908          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3909          *
3910          * @public
3911          *
3912          * @param string $id The magic word ID
3913          * @param mixed $callback The callback function (and object) to use
3914          * @param integer $flags a combination of the following flags:
3915          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3916          *
3917          * @return The old callback function for this name, if any
3918          */
3919         function setFunctionHook( $id, $callback, $flags = 0 ) {
3920                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3921                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3922
3923                 # Add to function cache
3924                 $mw = MagicWord::get( $id );
3925                 if( !$mw )
3926                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3927
3928                 $synonyms = $mw->getSynonyms();
3929                 $sensitive = intval( $mw->isCaseSensitive() );
3930
3931                 foreach ( $synonyms as $syn ) {
3932                         # Case
3933                         if ( !$sensitive ) {
3934                                 $syn = strtolower( $syn );
3935                         }
3936                         # Add leading hash
3937                         if ( !( $flags & SFH_NO_HASH ) ) {
3938                                 $syn = '#' . $syn;
3939                         }
3940                         # Remove trailing colon
3941                         if ( substr( $syn, -1, 1 ) == ':' ) {
3942                                 $syn = substr( $syn, 0, -1 );
3943                         }
3944                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3945                 }
3946                 return $oldVal;
3947         }
3948
3949         /**
3950          * Get all registered function hook identifiers
3951          *
3952          * @return array
3953          */
3954         function getFunctionHooks() {
3955                 return array_keys( $this->mFunctionHooks );
3956         }
3957
3958         /**
3959          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3960          * Placeholders created in Skin::makeLinkObj()
3961          * Returns an array of link CSS classes, indexed by PDBK.
3962          * $options is a bit field, RLH_FOR_UPDATE to select for update
3963          */
3964         function replaceLinkHolders( &$text, $options = 0 ) {
3965                 global $wgUser;
3966                 global $wgContLang;
3967
3968                 $fname = 'Parser::replaceLinkHolders';
3969                 wfProfileIn( $fname );
3970
3971                 $pdbks = array();
3972                 $colours = array();
3973                 $linkcolour_ids = array();
3974                 $sk = $this->mOptions->getSkin();
3975                 $linkCache =& LinkCache::singleton();
3976
3977                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3978                         wfProfileIn( $fname.'-check' );
3979                         $dbr = wfGetDB( DB_SLAVE );
3980                         $page = $dbr->tableName( 'page' );
3981                         $threshold = $wgUser->getOption('stubthreshold');
3982
3983                         # Sort by namespace
3984                         asort( $this->mLinkHolders['namespaces'] );
3985
3986                         # Generate query
3987                         $query = false;
3988                         $current = null;
3989                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3990                                 # Make title object
3991                                 $title = $this->mLinkHolders['titles'][$key];
3992
3993                                 # Skip invalid entries.
3994                                 # Result will be ugly, but prevents crash.
3995                                 if ( is_null( $title ) ) {
3996                                         continue;
3997                                 }
3998                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3999
4000                                 # Check if it's a static known link, e.g. interwiki
4001                                 if ( $title->isAlwaysKnown() ) {
4002                                         $colours[$pdbk] = '';
4003                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4004                                         $colours[$pdbk] = '';
4005                                         $this->mOutput->addLink( $title, $id );
4006                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4007                                         $colours[$pdbk] = 'new';
4008                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4009                                         $colours[$pdbk] = 'new';
4010                                 } else {
4011                                         # Not in the link cache, add it to the query
4012                                         if ( !isset( $current ) ) {
4013                                                 $current = $ns;
4014                                                 $query =  "SELECT page_id, page_namespace, page_title";
4015                                                 if ( $threshold > 0 ) {
4016                                                         $query .= ', page_len, page_is_redirect';
4017                                                 }
4018                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4019                                         } elseif ( $current != $ns ) {
4020                                                 $current = $ns;
4021                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4022                                         } else {
4023                                                 $query .= ', ';
4024                                         }
4025
4026                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4027                                 }
4028                         }
4029                         if ( $query ) {
4030                                 $query .= '))';
4031                                 if ( $options & RLH_FOR_UPDATE ) {
4032                                         $query .= ' FOR UPDATE';
4033                                 }
4034
4035                                 $res = $dbr->query( $query, $fname );
4036
4037                                 # Fetch data and form into an associative array
4038                                 # non-existent = broken
4039                                 while ( $s = $dbr->fetchObject($res) ) {
4040                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4041                                         $pdbk = $title->getPrefixedDBkey();
4042                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4043                                         $this->mOutput->addLink( $title, $s->page_id );
4044                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4045                                         //add id to the extension todolist
4046                                         $linkcolour_ids[$s->page_id] = $pdbk;
4047                                 }
4048                                 //pass an array of page_ids to an extension
4049                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4050                         }
4051                         wfProfileOut( $fname.'-check' );
4052
4053                         # Do a second query for different language variants of links and categories
4054                         if($wgContLang->hasVariants()){
4055                                 $linkBatch = new LinkBatch();
4056                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4057                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4058                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4059
4060                                 $categories = $this->mOutput->getCategoryLinks();
4061
4062                                 // Add variants of links to link batch
4063                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4064                                         $title = $this->mLinkHolders['titles'][$key];
4065                                         if ( is_null( $title ) )
4066                                                 continue;
4067
4068                                         $pdbk = $title->getPrefixedDBkey();
4069                                         $titleText = $title->getText();
4070
4071                                         // generate all variants of the link title text
4072                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4073
4074                                         // if link was not found (in first query), add all variants to query
4075                                         if ( !isset($colours[$pdbk]) ){
4076                                                 foreach($allTextVariants as $textVariant){
4077                                                         if($textVariant != $titleText){
4078                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4079                                                                 if(is_null($variantTitle)) continue;
4080                                                                 $linkBatch->addObj( $variantTitle );
4081                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4082                                                         }
4083                                                 }
4084                                         }
4085                                 }
4086
4087                                 // process categories, check if a category exists in some variant
4088                                 foreach( $categories as $category ){
4089                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4090                                         foreach($variants as $variant){
4091                                                 if($variant != $category){
4092                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4093                                                         if(is_null($variantTitle)) continue;
4094                                                         $linkBatch->addObj( $variantTitle );
4095                                                         $categoryMap[$variant] = $category;
4096                                                 }
4097                                         }
4098                                 }
4099
4100
4101                                 if(!$linkBatch->isEmpty()){
4102                                         // construct query
4103                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4104
4105                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4106                                         if ( $threshold > 0 ) {
4107                                                 $variantQuery .= ', page_len, page_is_redirect';
4108                                         }
4109
4110                                         $variantQuery .= " FROM $page WHERE $titleClause";
4111                                         if ( $options & RLH_FOR_UPDATE ) {
4112                                                 $variantQuery .= ' FOR UPDATE';
4113                                         }
4114
4115                                         $varRes = $dbr->query( $variantQuery, $fname );
4116
4117                                         // for each found variants, figure out link holders and replace
4118                                         while ( $s = $dbr->fetchObject($varRes) ) {
4119
4120                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4121                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4122                                                 $vardbk = $variantTitle->getDBkey();
4123
4124                                                 $holderKeys = array();
4125                                                 if(isset($variantMap[$varPdbk])){
4126                                                         $holderKeys = $variantMap[$varPdbk];
4127                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4128                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4129                                                 }
4130
4131                                                 // loop over link holders
4132                                                 foreach($holderKeys as $key){
4133                                                         $title = $this->mLinkHolders['titles'][$key];
4134                                                         if ( is_null( $title ) ) continue;
4135
4136                                                         $pdbk = $title->getPrefixedDBkey();
4137
4138                                                         if(!isset($colours[$pdbk])){
4139                                                                 // found link in some of the variants, replace the link holder data
4140                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4141                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4142
4143                                                                 // set pdbk and colour
4144                                                                 $pdbks[$key] = $varPdbk;
4145                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4146                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4147                                                         }
4148                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4149                                                 }
4150
4151                                                 // check if the object is a variant of a category
4152                                                 if(isset($categoryMap[$vardbk])){
4153                                                         $oldkey = $categoryMap[$vardbk];
4154                                                         if($oldkey != $vardbk)
4155                                                                 $varCategories[$oldkey]=$vardbk;
4156                                                 }
4157                                         }
4158
4159                                         // rebuild the categories in original order (if there are replacements)
4160                                         if(count($varCategories)>0){
4161                                                 $newCats = array();
4162                                                 $originalCats = $this->mOutput->getCategories();
4163                                                 foreach($originalCats as $cat => $sortkey){
4164                                                         // make the replacement
4165                                                         if( array_key_exists($cat,$varCategories) )
4166                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4167                                                         else $newCats[$cat] = $sortkey;
4168                                                 }
4169                                                 $this->mOutput->setCategoryLinks($newCats);
4170                                         }
4171                                 }
4172                         }
4173
4174                         # Construct search and replace arrays
4175                         wfProfileIn( $fname.'-construct' );
4176                         $replacePairs = array();
4177                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4178                                 $pdbk = $pdbks[$key];
4179                                 $searchkey = "<!--LINK $key-->";
4180                                 $title = $this->mLinkHolders['titles'][$key];
4181                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4182                                         $linkCache->addBadLinkObj( $title );
4183                                         $colours[$pdbk] = 'new';
4184                                         $this->mOutput->addLink( $title, 0 );
4185                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4186                                                                         $this->mLinkHolders['texts'][$key],
4187                                                                         $this->mLinkHolders['queries'][$key] );
4188                                 } else {
4189                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4190                                                                         $this->mLinkHolders['texts'][$key],
4191                                                                         $this->mLinkHolders['queries'][$key] );
4192                                 }
4193                         }
4194                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4195                         wfProfileOut( $fname.'-construct' );
4196
4197                         # Do the thing
4198                         wfProfileIn( $fname.'-replace' );
4199                         $text = preg_replace_callback(
4200                                 '/(<!--LINK .*?-->)/',
4201                                 $replacer->cb(),
4202                                 $text);
4203
4204                         wfProfileOut( $fname.'-replace' );
4205                 }
4206
4207                 # Now process interwiki link holders
4208                 # This is quite a bit simpler than internal links
4209                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4210                         wfProfileIn( $fname.'-interwiki' );
4211                         # Make interwiki link HTML
4212                         $replacePairs = array();
4213                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4214                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4215                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4216                         }
4217                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4218
4219                         $text = preg_replace_callback(
4220                                 '/<!--IWLINK (.*?)-->/',
4221                                 $replacer->cb(),
4222                                 $text );
4223                         wfProfileOut( $fname.'-interwiki' );
4224                 }
4225
4226                 wfProfileOut( $fname );
4227                 return $colours;
4228         }
4229
4230         /**
4231          * Replace <!--LINK--> link placeholders with plain text of links
4232          * (not HTML-formatted).
4233          * @param string $text
4234          * @return string
4235          */
4236         function replaceLinkHoldersText( $text ) {
4237                 $fname = 'Parser::replaceLinkHoldersText';
4238                 wfProfileIn( $fname );
4239
4240                 $text = preg_replace_callback(
4241                         '/<!--(LINK|IWLINK) (.*?)-->/',
4242                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4243                         $text );
4244
4245                 wfProfileOut( $fname );
4246                 return $text;
4247         }
4248
4249         /**
4250          * @param array $matches
4251          * @return string
4252          * @private
4253          */
4254         function replaceLinkHoldersTextCallback( $matches ) {
4255                 $type = $matches[1];
4256                 $key  = $matches[2];
4257                 if( $type == 'LINK' ) {
4258                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4259                                 return $this->mLinkHolders['texts'][$key];
4260                         }
4261                 } elseif( $type == 'IWLINK' ) {
4262                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4263                                 return $this->mInterwikiLinkHolders['texts'][$key];
4264                         }
4265                 }
4266                 return $matches[0];
4267         }
4268
4269         /**
4270          * Tag hook handler for 'pre'.
4271          */
4272         function renderPreTag( $text, $attribs ) {
4273                 // Backwards-compatibility hack
4274                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4275
4276                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4277                 return wfOpenElement( 'pre', $attribs ) .
4278                         Xml::escapeTagsOnly( $content ) .
4279                         '</pre>';
4280         }
4281
4282         /**
4283          * Renders an image gallery from a text with one line per image.
4284          * text labels may be given by using |-style alternative text. E.g.
4285          *   Image:one.jpg|The number "1"
4286          *   Image:tree.jpg|A tree
4287          * given as text will return the HTML of a gallery with two images,
4288          * labeled 'The number "1"' and
4289          * 'A tree'.
4290          */
4291         function renderImageGallery( $text, $params ) {
4292                 $ig = new ImageGallery();
4293                 $ig->setContextTitle( $this->mTitle );
4294                 $ig->setShowBytes( false );
4295                 $ig->setShowFilename( false );
4296                 $ig->setParser( $this );
4297                 $ig->setHideBadImages();
4298                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4299                 $ig->useSkin( $this->mOptions->getSkin() );
4300                 $ig->mRevisionId = $this->mRevisionId;
4301
4302                 if( isset( $params['caption'] ) ) {
4303                         $caption = $params['caption'];
4304                         $caption = htmlspecialchars( $caption );
4305                         $caption = $this->replaceInternalLinks( $caption );
4306                         $ig->setCaptionHtml( $caption );
4307                 }
4308                 if( isset( $params['perrow'] ) ) {
4309                         $ig->setPerRow( $params['perrow'] );
4310                 }
4311                 if( isset( $params['widths'] ) ) {
4312                         $ig->setWidths( $params['widths'] );
4313                 }
4314                 if( isset( $params['heights'] ) ) {
4315                         $ig->setHeights( $params['heights'] );
4316                 }
4317
4318                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4319
4320                 $lines = explode( "\n", $text );
4321                 foreach ( $lines as $line ) {
4322                         # match lines like these:
4323                         # Image:someimage.jpg|This is some image
4324                         $matches = array();
4325                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4326                         # Skip empty lines
4327                         if ( count( $matches ) == 0 ) {
4328                                 continue;
4329                         }
4330                         $tp = Title::newFromText( $matches[1] );
4331                         $nt =& $tp;
4332                         if( is_null( $nt ) ) {
4333                                 # Bogus title. Ignore these so we don't bomb out later.
4334                                 continue;
4335                         }
4336                         if ( isset( $matches[3] ) ) {
4337                                 $label = $matches[3];
4338                         } else {
4339                                 $label = '';
4340                         }
4341
4342                         $html = $this->recursiveTagParse( trim( $label ) );
4343
4344                         $ig->add( $nt, $html );
4345
4346                         # Only add real images (bug #5586)
4347                         if ( $nt->getNamespace() == NS_IMAGE ) {
4348                                 $this->mOutput->addImage( $nt->getDBkey() );
4349                         }
4350                 }
4351                 return $ig->toHTML();
4352         }
4353
4354         function getImageParams( $handler ) {
4355                 if ( $handler ) {
4356                         $handlerClass = get_class( $handler );
4357                 } else {
4358                         $handlerClass = '';
4359                 }
4360                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4361                         // Initialise static lists
4362                         static $internalParamNames = array(
4363                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4364                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4365                                         'bottom', 'text-bottom' ),
4366                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4367                                         'upright', 'border' ),
4368                         );
4369                         static $internalParamMap;
4370                         if ( !$internalParamMap ) {
4371                                 $internalParamMap = array();
4372                                 foreach ( $internalParamNames as $type => $names ) {
4373                                         foreach ( $names as $name ) {
4374                                                 $magicName = str_replace( '-', '_', "img_$name" );
4375                                                 $internalParamMap[$magicName] = array( $type, $name );
4376                                         }
4377                                 }
4378                         }
4379
4380                         // Add handler params
4381                         $paramMap = $internalParamMap;
4382                         if ( $handler ) {
4383                                 $handlerParamMap = $handler->getParamMap();
4384                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4385                                         $paramMap[$magic] = array( 'handler', $paramName );
4386                                 }
4387                         }
4388                         $this->mImageParams[$handlerClass] = $paramMap;
4389                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4390                 }
4391                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4392         }
4393
4394         /**
4395          * Parse image options text and use it to make an image
4396          */
4397         function makeImage( $title, $options ) {
4398                 # @TODO: let the MediaHandler specify its transform parameters
4399                 #
4400                 # Check if the options text is of the form "options|alt text"
4401                 # Options are:
4402                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4403                 #  * left               no resizing, just left align. label is used for alt= only
4404                 #  * right              same, but right aligned
4405                 #  * none               same, but not aligned
4406                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4407                 #  * center             center the image
4408                 #  * framed             Keep original image size, no magnify-button.
4409                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4410                 #  * upright            reduce width for upright images, rounded to full __0 px
4411                 #  * border             draw a 1px border around the image
4412                 # vertical-align values (no % or length right now):
4413                 #  * baseline
4414                 #  * sub
4415                 #  * super
4416                 #  * top
4417                 #  * text-top
4418                 #  * middle
4419                 #  * bottom
4420                 #  * text-bottom
4421
4422                 $parts = array_map( 'trim', explode( '|', $options) );
4423                 $sk = $this->mOptions->getSkin();
4424
4425                 # Give extensions a chance to select the file revision for us
4426                 $skip = $time = false;
4427                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4428
4429                 if ( $skip ) {
4430                         return $sk->makeLinkObj( $title );
4431                 }
4432
4433                 # Get parameter map
4434                 $file = wfFindFile( $title, $time );
4435                 $handler = $file ? $file->getHandler() : false;
4436
4437                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4438
4439                 # Process the input parameters
4440                 $caption = '';
4441                 $params = array( 'frame' => array(), 'handler' => array(),
4442                         'horizAlign' => array(), 'vertAlign' => array() );
4443                 foreach( $parts as $part ) {
4444                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4445                         if ( isset( $paramMap[$magicName] ) ) {
4446                                 list( $type, $paramName ) = $paramMap[$magicName];
4447                                 $params[$type][$paramName] = $value;
4448
4449                                 // Special case; width and height come in one variable together
4450                                 if( $type == 'handler' && $paramName == 'width' ) {
4451                                         $m = array();
4452                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4453                                                 $params[$type]['width'] = intval( $m[1] );
4454                                                 $params[$type]['height'] = intval( $m[2] );
4455                                         } else {
4456                                                 $params[$type]['width'] = intval( $value );
4457                                         }
4458                                 }
4459                         } else {
4460                                 $caption = $part;
4461                         }
4462                 }
4463
4464                 # Process alignment parameters
4465                 if ( $params['horizAlign'] ) {
4466                         $params['frame']['align'] = key( $params['horizAlign'] );
4467                 }
4468                 if ( $params['vertAlign'] ) {
4469                         $params['frame']['valign'] = key( $params['vertAlign'] );
4470                 }
4471
4472                 # Validate the handler parameters
4473                 if ( $handler ) {
4474                         foreach ( $params['handler'] as $name => $value ) {
4475                                 if ( !$handler->validateParam( $name, $value ) ) {
4476                                         unset( $params['handler'][$name] );
4477                                 }
4478                         }
4479                 }
4480
4481                 # Strip bad stuff out of the alt text
4482                 $alt = $this->replaceLinkHoldersText( $caption );
4483
4484                 # make sure there are no placeholders in thumbnail attributes
4485                 # that are later expanded to html- so expand them now and
4486                 # remove the tags
4487                 $alt = $this->mStripState->unstripBoth( $alt );
4488                 $alt = Sanitizer::stripAllTags( $alt );
4489
4490                 $params['frame']['alt'] = $alt;
4491                 $params['frame']['caption'] = $caption;
4492
4493                 # Linker does the rest
4494                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'] );
4495
4496                 # Give the handler a chance to modify the parser object
4497                 if ( $handler ) {
4498                         $handler->parserTransformHook( $this, $file );
4499                 }
4500
4501                 return $ret;
4502         }
4503
4504         /**
4505          * Set a flag in the output object indicating that the content is dynamic and
4506          * shouldn't be cached.
4507          */
4508         function disableCache() {
4509                 wfDebug( "Parser output marked as uncacheable.\n" );
4510                 $this->mOutput->mCacheTime = -1;
4511         }
4512
4513         /**#@+
4514          * Callback from the Sanitizer for expanding items found in HTML attribute
4515          * values, so they can be safely tested and escaped.
4516          * @param string $text
4517          * @param PPFrame $frame
4518          * @return string
4519          * @private
4520          */
4521         function attributeStripCallback( &$text, $frame = false ) {
4522                 $text = $this->replaceVariables( $text, $frame );
4523                 $text = $this->mStripState->unstripBoth( $text );
4524                 return $text;
4525         }
4526
4527         /**#@-*/
4528
4529         /**#@+
4530          * Accessor/mutator
4531          */
4532         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4533         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4534         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4535         /**#@-*/
4536
4537         /**#@+
4538          * Accessor
4539          */
4540         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4541         /**#@-*/
4542
4543
4544         /**
4545          * Break wikitext input into sections, and either pull or replace
4546          * some particular section's text.
4547          *
4548          * External callers should use the getSection and replaceSection methods.
4549          *
4550          * @param string $text Page wikitext
4551          * @param string $section A section identifier string of the form:
4552          *   <flag1> - <flag2> - ... - <section number>
4553          *
4554          * Currently the only recognised flag is "T", which means the target section number
4555          * was derived during a template inclusion parse, in other words this is a template
4556          * section edit link. If no flags are given, it was an ordinary section edit link.
4557          * This flag is required to avoid a section numbering mismatch when a section is
4558          * enclosed by <includeonly> (bug 6563).
4559          *
4560          * The section number 0 pulls the text before the first heading; other numbers will
4561          * pull the given section along with its lower-level subsections. If the section is
4562          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4563          *
4564          * @param string $mode One of "get" or "replace"
4565          * @param string $newText Replacement text for section data.
4566          * @return string for "get", the extracted section text.
4567          *                for "replace", the whole page with the section replaced.
4568          */
4569         private function extractSections( $text, $section, $mode, $newText='' ) {
4570                 global $wgTitle;
4571                 $this->clearState();
4572                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4573                 $this->mOptions = new ParserOptions;
4574                 $this->setOutputType( self::OT_WIKI );
4575                 $outText = '';
4576                 $frame = $this->getPreprocessor()->newFrame();
4577
4578                 // Process section extraction flags
4579                 $flags = 0;
4580                 $sectionParts = explode( '-', $section );
4581                 $sectionIndex = array_pop( $sectionParts );
4582                 foreach ( $sectionParts as $part ) {
4583                         if ( $part == 'T' ) {
4584                                 $flags |= self::PTD_FOR_INCLUSION;
4585                         }
4586                 }
4587                 // Preprocess the text
4588                 $root = $this->preprocessToDom( $text, $flags );
4589
4590                 // <h> nodes indicate section breaks
4591                 // They can only occur at the top level, so we can find them by iterating the root's children
4592                 $node = $root->getFirstChild();
4593
4594                 // Find the target section
4595                 if ( $sectionIndex == 0 ) {
4596                         // Section zero doesn't nest, level=big
4597                         $targetLevel = 1000;
4598                 } else {
4599             while ( $node ) {
4600                 if ( $node->getName() == 'h' ) {
4601                     $bits = $node->splitHeading();
4602                                         if ( $bits['i'] == $sectionIndex ) {
4603                                         $targetLevel = $bits['level'];
4604                                                 break;
4605                                         }
4606                                 }
4607                                 if ( $mode == 'replace' ) {
4608                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4609                                 }
4610                                 $node = $node->getNextSibling();
4611                         }
4612                 }
4613
4614                 if ( !$node ) {
4615                         // Not found
4616                         if ( $mode == 'get' ) {
4617                                 return $newText;
4618                         } else {
4619                                 return $text;
4620                         }
4621                 }
4622
4623                 // Find the end of the section, including nested sections
4624                 do {
4625                         if ( $node->getName() == 'h' ) {
4626                                 $bits = $node->splitHeading();
4627                                 $curLevel = $bits['level'];
4628                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4629                                         break;
4630                                 }
4631                         }
4632                         if ( $mode == 'get' ) {
4633                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4634                         }
4635                         $node = $node->getNextSibling();
4636                 } while ( $node );
4637
4638                 // Write out the remainder (in replace mode only)
4639                 if ( $mode == 'replace' ) {
4640                         // Output the replacement text
4641                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4642                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4643                         $outText .= $newText . "\n\n";
4644                         while ( $node ) {
4645                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4646                                 $node = $node->getNextSibling();
4647                         }
4648                 }
4649
4650                 if ( is_string( $outText ) ) {
4651                         // Re-insert stripped tags
4652                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
4653                 }
4654
4655                 return $outText;
4656         }
4657
4658         /**
4659          * This function returns the text of a section, specified by a number ($section).
4660          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4661          * the first section before any such heading (section 0).
4662          *
4663          * If a section contains subsections, these are also returned.
4664          *
4665          * @param string $text text to look in
4666          * @param string $section section identifier
4667          * @param string $deftext default to return if section is not found
4668          * @return string text of the requested section
4669          */
4670         public function getSection( $text, $section, $deftext='' ) {
4671                 return $this->extractSections( $text, $section, "get", $deftext );
4672         }
4673
4674         public function replaceSection( $oldtext, $section, $text ) {
4675                 return $this->extractSections( $oldtext, $section, "replace", $text );
4676         }
4677
4678         /**
4679          * Get the timestamp associated with the current revision, adjusted for
4680          * the default server-local timestamp
4681          */
4682         function getRevisionTimestamp() {
4683                 if ( is_null( $this->mRevisionTimestamp ) ) {
4684                         wfProfileIn( __METHOD__ );
4685                         global $wgContLang;
4686                         $dbr = wfGetDB( DB_SLAVE );
4687                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4688                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4689
4690                         // Normalize timestamp to internal MW format for timezone processing.
4691                         // This has the added side-effect of replacing a null value with
4692                         // the current time, which gives us more sensible behavior for
4693                         // previews.
4694                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4695
4696                         // The cryptic '' timezone parameter tells to use the site-default
4697                         // timezone offset instead of the user settings.
4698                         //
4699                         // Since this value will be saved into the parser cache, served
4700                         // to other users, and potentially even used inside links and such,
4701                         // it needs to be consistent for all visitors.
4702                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4703
4704                         wfProfileOut( __METHOD__ );
4705                 }
4706                 return $this->mRevisionTimestamp;
4707         }
4708
4709         /**
4710          * Mutator for $mDefaultSort
4711          *
4712          * @param $sort New value
4713          */
4714         public function setDefaultSort( $sort ) {
4715                 $this->mDefaultSort = $sort;
4716         }
4717
4718         /**
4719          * Accessor for $mDefaultSort
4720          * Will use the title/prefixed title if none is set
4721          *
4722          * @return string
4723          */
4724         public function getDefaultSort() {
4725                 if( $this->mDefaultSort !== false ) {
4726                         return $this->mDefaultSort;
4727                 } else {
4728                         return $this->mTitle->getNamespace() == NS_CATEGORY
4729                                         ? $this->mTitle->getText()
4730                                         : $this->mTitle->getPrefixedText();
4731                 }
4732         }
4733
4734         /**
4735          * Try to guess the section anchor name based on a wikitext fragment
4736          * presumably extracted from a heading, for example "Header" from
4737          * "== Header ==".
4738          */
4739         public function guessSectionNameFromWikiText( $text ) {
4740                 # Strip out wikitext links(they break the anchor)
4741                 $text = $this->stripSectionName( $text );
4742                 $headline = Sanitizer::decodeCharReferences( $text );
4743                 # strip out HTML
4744                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4745                 $headline = trim( $headline );
4746                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4747                 $replacearray = array(
4748                         '%3A' => ':',
4749                         '%' => '.'
4750                 );
4751                 return str_replace(
4752                         array_keys( $replacearray ),
4753                         array_values( $replacearray ),
4754                         $sectionanchor );
4755         }
4756
4757         /**
4758          * Strips a text string of wikitext for use in a section anchor
4759          *
4760          * Accepts a text string and then removes all wikitext from the
4761          * string and leaves only the resultant text (i.e. the result of
4762          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4763          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4764          * to create valid section anchors by mimicing the output of the
4765          * parser when headings are parsed.
4766          *
4767          * @param $text string Text string to be stripped of wikitext
4768          * for use in a Section anchor
4769          * @return Filtered text string
4770          */
4771         public function stripSectionName( $text ) {
4772                 # Strip internal link markup
4773                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4774                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4775
4776                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4777                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4778                 # on how many empty links there are on the page - need to figure that out.
4779                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4780
4781                 # Parse wikitext quotes (italics & bold)
4782                 $text = $this->doQuotes($text);
4783
4784                 # Strip HTML tags
4785                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4786                 return $text;
4787         }
4788
4789         function srvus( $text ) {
4790                 return $this->testSrvus( $text, $this->mOutputType );
4791         }
4792
4793         /**
4794          * strip/replaceVariables/unstrip for preprocessor regression testing
4795          */
4796         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4797                 $this->clearState();
4798                 if ( ! ( $title instanceof Title ) ) {
4799                         $title = Title::newFromText( $title );
4800                 }
4801                 $this->mTitle = $title;
4802                 $this->mOptions = $options;
4803                 $this->setOutputType( $outputType );
4804                 $text = $this->replaceVariables( $text );
4805                 $text = $this->mStripState->unstripBoth( $text );
4806                 $text = Sanitizer::removeHTMLtags( $text );
4807                 return $text;
4808         }
4809
4810         function testPst( $text, $title, $options ) {
4811                 global $wgUser;
4812                 if ( ! ( $title instanceof Title ) ) {
4813                         $title = Title::newFromText( $title );
4814                 }
4815                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4816         }
4817
4818         function testPreprocess( $text, $title, $options ) {
4819                 if ( ! ( $title instanceof Title ) ) {
4820                         $title = Title::newFromText( $title );
4821                 }
4822                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4823         }
4824
4825         function markerSkipCallback( $s, $callback ) {
4826                 $i = 0;
4827                 $out = '';
4828                 while ( $i < strlen( $s ) ) {
4829                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4830                         if ( $markerStart === false ) {
4831                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4832                                 break;
4833                         } else {
4834                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4835                                 $markerEnd = strpos( $s, $this->mMarkerSuffix, $markerStart );
4836                                 if ( $markerEnd === false ) {
4837                                         $out .= substr( $s, $markerStart );
4838                                         break;
4839                                 } else {
4840                                         $markerEnd += strlen( $this->mMarkerSuffix );
4841                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4842                                         $i = $markerEnd;
4843                                 }
4844                         }
4845                 }
4846                 return $out;
4847         }
4848 }
4849
4850 /**
4851  * @todo document, briefly.
4852  * @addtogroup Parser
4853  */
4854 class StripState {
4855         var $general, $nowiki;
4856
4857         function __construct() {
4858                 $this->general = new ReplacementArray;
4859                 $this->nowiki = new ReplacementArray;
4860         }
4861
4862         function unstripGeneral( $text ) {
4863                 wfProfileIn( __METHOD__ );
4864                 do {
4865                         $oldText = $text;
4866                         $text = $this->general->replace( $text );
4867                 } while ( $text != $oldText );
4868                 wfProfileOut( __METHOD__ );
4869                 return $text;
4870         }
4871
4872         function unstripNoWiki( $text ) {
4873                 wfProfileIn( __METHOD__ );
4874                 do {
4875                         $oldText = $text;
4876                         $text = $this->nowiki->replace( $text );
4877                 } while ( $text != $oldText );
4878                 wfProfileOut( __METHOD__ );
4879                 return $text;
4880         }
4881
4882         function unstripBoth( $text ) {
4883                 wfProfileIn( __METHOD__ );
4884                 do {
4885                         $oldText = $text;
4886                         $text = $this->general->replace( $text );
4887                         $text = $this->nowiki->replace( $text );
4888                 } while ( $text != $oldText );
4889                 wfProfileOut( __METHOD__ );
4890                 return $text;
4891         }
4892 }
4893
4894 /**
4895  * @todo document, briefly.
4896  * @addtogroup Parser
4897  */
4898 class OnlyIncludeReplacer {
4899         var $output = '';
4900
4901         function replace( $matches ) {
4902                 if ( substr( $matches[1], -1 ) == "\n" ) {
4903                         $this->output .= substr( $matches[1], 0, -1 );
4904                 } else {
4905                         $this->output .= $matches[1];
4906                 }
4907         }
4908 }
4909