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