includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @addtogroup Parser
   6  */
   7
   8 /**
   9  * Update this version number when the ParserOutput format
  10  * changes in an incompatible way, so the parser cache
  11  * can automatically discard old data.
  12  */
  13 define( 'MW_PARSER_VERSION', '1.6.1' );
  14
  15 define( 'RLH_FOR_UPDATE', 1 );
  16
  17 # Allowed values for $mOutputType
  18 define( 'OT_HTML', 1 );
  19 define( 'OT_WIKI', 2 );
  20 define( 'OT_MSG' , 3 );
  21 define( 'OT_PREPROCESS', 4 );
  22
  23 # Flags for setFunctionHook
  24 define( 'SFH_NO_HASH', 1 );
  25
  26 # string parameter for extractTags which will cause it
  27 # to strip HTML comments in addition to regular
  28 # <XML>-style tags. This should not be anything we
  29 # may want to use in wikisyntax
  30 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  31
  32 # Constants needed for external link processing
  33 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  34 # Everything except bracket, space, or control characters
  35 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  36 # Including space, but excluding newlines
  37 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  38 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  39 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  40 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  41         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  42 define( 'EXT_IMAGE_REGEX',
  43         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  44         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  45         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  46 );
  47
  48 // State constants for the definition list colon extraction
  49 define( 'MW_COLON_STATE_TEXT', 0 );
  50 define( 'MW_COLON_STATE_TAG', 1 );
  51 define( 'MW_COLON_STATE_TAGSTART', 2 );
  52 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  53 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  54 define( 'MW_COLON_STATE_COMMENT', 5 );
  55 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  56 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  57
  58 /**
  59  * PHP Parser
  60  *
  61  * Processes wiki markup
  62  *
  63  * <pre>
  64  * There are four main entry points into the Parser class:
  65  * parse()
  66  *   produces HTML output
  67  * preSaveTransform().
  68  *   produces altered wiki markup.
  69  * transformMsg()
  70  *   performs brace substitution on MediaWiki messages
  71  * preprocess()
  72  *   removes HTML comments and expands templates
  73  *
  74  * Globals used:
  75  *    objects:   $wgLang, $wgContLang
  76  *
  77  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  78  *
  79  * settings:
  80  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  81  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  82  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  83  *  $wgMaxArticleSize*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  */
  89 class Parser
  90 {
  91         const VERSION = MW_PARSER_VERSION;
  92         /**#@+
  93          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102         var $mIncludeSizes, $mDefaultSort;
 103         var $mTemplates,        // cache of already loaded templates, avoids
 104                                 // multiple SQL queries for the same string
 105             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 106                                 // in this path. Used for loop detection.
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function Parser() {
 126                 $this->mTagHooks = array();
 127                 $this->mFunctionHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mFirstCall = true;
 130         }
 131
 132         /**
 133          * Do various kinds of initialisation on the first call of the parser
 134          */
 135         function firstCallInit() {
 136                 if ( !$this->mFirstCall ) {
 137                         return;
 138                 }
 139
 140                 wfProfileIn( __METHOD__ );
 141                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 142
 143                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 144
 145                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 169                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 176                 }
 177
 178                 $this->initialiseVariables();
 179                 $this->mFirstCall = false;
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = array();
 200                 $this->mInPre = false;
 201                 $this->mInterwikiLinkHolders = array(
 202                         'texts' => array(),
 203                         'titles' => array()
 204                 );
 205                 $this->mLinkHolders = array(
 206                         'namespaces' => array(),
 207                         'dbkeys' => array(),
 208                         'queries' => array(),
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 213
 214                 /**
 215                  * Prefix for temporary replacement strings for the multipass parser.
 216                  * \x07 should never appear in input as it's disallowed in XML.
 217                  * Using it at the front also gives us a little extra robustness
 218                  * since it shouldn't match when butted up against identifier-like
 219                  * string constructs.
 220                  */
 221                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplates = array();
 225                 $this->mTemplatePath = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'pre-expand' => 0,
 231                         'post-expand' => 0,
 232                         'arg' => 0
 233                 );
 234                 $this->mDefaultSort = false;
 235
 236                 wfRunHooks( 'ParserClearState', array( &$this ) );
 237                 wfProfileOut( __METHOD__ );
 238         }
 239
 240         function setOutputType( $ot ) {
 241                 $this->mOutputType = $ot;
 242                 // Shortcut alias
 243                 $this->ot = array(
 244                         'html' => $ot == OT_HTML,
 245                         'wiki' => $ot == OT_WIKI,
 246                         'msg' => $ot == OT_MSG,
 247                         'pre' => $ot == OT_PREPROCESS,
 248                 );
 249         }
 250
 251         /**
 252          * Accessor for mUniqPrefix.
 253          *
 254          * @public
 255          */
 256         function uniqPrefix() {
 257                 return $this->mUniqPrefix;
 258         }
 259
 260         /**
 261          * Convert wikitext to HTML
 262          * Do not call this function recursively.
 263          *
 264          * @param string $text Text we want to parse
 265          * @param Title &$title A title object
 266          * @param array $options
 267          * @param boolean $linestart
 268          * @param boolean $clearState
 269          * @param int $revid number to pass in {{REVISIONID}}
 270          * @return ParserOutput a ParserOutput
 271          */
 272         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 273                 /**
 274                  * First pass--just handle <nowiki> sections, pass the rest off
 275                  * to internalParse() which does all the real work.
 276                  */
 277
 278                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 279                 $fname = 'Parser::parse-' . wfGetCaller();
 280                 wfProfileIn( __METHOD__ );
 281                 wfProfileIn( $fname );
 282
 283                 if ( $clearState ) {
 284                         $this->clearState();
 285                 }
 286
 287                 $this->mOptions = $options;
 288                 $this->mTitle =& $title;
 289                 $oldRevisionId = $this->mRevisionId;
 290                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 291                 if( $revid !== null ) {
 292                         $this->mRevisionId = $revid;
 293                         $this->mRevisionTimestamp = null;
 294                 }
 295                 $this->setOutputType( OT_HTML );
 296                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 297                 $text = $this->strip( $text, $this->mStripState );
 298                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 299                 $text = $this->internalParse( $text );
 300                 $text = $this->mStripState->unstripGeneral( $text );
 301
 302                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 303                 $fixtags = array(
 304                         # french spaces, last one Guillemet-left
 305                         # only if there is something before the space
 306                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 307                         # french spaces, Guillemet-right
 308                         '/(\\302\\253) /' => '\\1&nbsp;',
 309                 );
 310                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 311
 312                 # only once and last
 313                 $text = $this->doBlockLevels( $text, $linestart );
 314
 315                 $this->replaceLinkHolders( $text );
 316
 317                 # the position of the parserConvert() call should not be changed. it
 318                 # assumes that the links are all replaced and the only thing left
 319                 # is the <nowiki> mark.
 320                 # Side-effects: this calls $this->mOutput->setTitleText()
 321                 $text = $wgContLang->parserConvert( $text, $this );
 322
 323                 $text = $this->mStripState->unstripNoWiki( $text );
 324
 325                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 326
 327                 $text = Sanitizer::normalizeCharReferences( $text );
 328
 329                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 330                         $text = Parser::tidy($text);
 331                 } else {
 332                         # attempt to sanitize at least some nesting problems
 333                         # (bug #2702 and quite a few others)
 334                         $tidyregs = array(
 335                                 # ''Something [http://www.cool.com cool''] -->
 336                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 337                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 338                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 339                                 # fix up an anchor inside another anchor, only
 340                                 # at least for a single single nested link (bug 3695)
 341                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 342                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 343                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 344                                 # contains a div, so fix it up here; replace
 345                                 # div with escaped text
 346                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 347                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 348                                 # remove empty italic or bold tag pairs, some
 349                                 # introduced by rules above
 350                                 '/<([bi])><\/\\1>/' => '',
 351                         );
 352
 353                         $text = preg_replace(
 354                                 array_keys( $tidyregs ),
 355                                 array_values( $tidyregs ),
 356                                 $text );
 357                 }
 358
 359                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 360
 361                 # Information on include size limits, for the benefit of users who try to skirt them
 362                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 363                         $max = $this->mOptions->getMaxIncludeSize();
 364                         $text .= "<!-- \n" .
 365                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 366                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 367                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 368                                 "Maximum: $max bytes\n" .
 369                                 "-->\n";
 370                 }
 371                 $this->mOutput->setText( $text );
 372                 $this->mRevisionId = $oldRevisionId;
 373                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 374                 wfProfileOut( $fname );
 375                 wfProfileOut( __METHOD__ );
 376
 377                 return $this->mOutput;
 378         }
 379
 380         /**
 381          * Recursive parser entry point that can be called from an extension tag
 382          * hook.
 383          */
 384         function recursiveTagParse( $text ) {
 385                 wfProfileIn( __METHOD__ );
 386                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 387                 $text = $this->strip( $text, $this->mStripState );
 388                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 389                 $text = $this->internalParse( $text );
 390                 wfProfileOut( __METHOD__ );
 391                 return $text;
 392         }
 393
 394         /**
 395          * Expand templates and variables in the text, producing valid, static wikitext.
 396          * Also removes comments.
 397          */
 398         function preprocess( $text, $title, $options ) {
 399                 wfProfileIn( __METHOD__ );
 400                 $this->clearState();
 401                 $this->setOutputType( OT_PREPROCESS );
 402                 $this->mOptions = $options;
 403                 $this->mTitle = $title;
 404                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 405                 $text = $this->strip( $text, $this->mStripState );
 406                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 407                 if ( $this->mOptions->getRemoveComments() ) {
 408                         $text = Sanitizer::removeHTMLcomments( $text );
 409                 }
 410                 $text = $this->replaceVariables( $text );
 411                 $text = $this->mStripState->unstripBoth( $text );
 412                 wfProfileOut( __METHOD__ );
 413                 return $text;
 414         }
 415
 416         /**
 417          * Get a random string
 418          *
 419          * @private
 420          * @static
 421          */
 422         function getRandomString() {
 423                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 424         }
 425
 426         function &getTitle() { return $this->mTitle; }
 427         function getOptions() { return $this->mOptions; }
 428
 429         function getFunctionLang() {
 430                 global $wgLang, $wgContLang;
 431                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 432         }
 433
 434         /**
 435          * Replaces all occurrences of HTML-style comments and the given tags
 436          * in the text with a random marker and returns teh next text. The output
 437          * parameter $matches will be an associative array filled with data in
 438          * the form:
 439          *   'UNIQ-xxxxx' => array(
 440          *     'element',
 441          *     'tag content',
 442          *     array( 'param' => 'x' ),
 443          *     '<element param="x">tag content</element>' ) )
 444          *
 445          * @param $elements list of element names. Comments are always extracted.
 446          * @param $text Source text string.
 447          * @param $uniq_prefix
 448          *
 449          * @private
 450          * @static
 451          */
 452         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 453                 static $n = 1;
 454                 $stripped = '';
 455                 $matches = array();
 456
 457                 $taglist = implode( '|', $elements );
 458                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 459
 460                 while ( '' != $text ) {
 461                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 462                         $stripped .= $p[0];
 463                         if( count( $p ) < 5 ) {
 464                                 break;
 465                         }
 466                         if( count( $p ) > 5 ) {
 467                                 // comment
 468                                 $element    = $p[4];
 469                                 $attributes = '';
 470                                 $close      = '';
 471                                 $inside     = $p[5];
 472                         } else {
 473                                 // tag
 474                                 $element    = $p[1];
 475                                 $attributes = $p[2];
 476                                 $close      = $p[3];
 477                                 $inside     = $p[4];
 478                         }
 479
 480                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 481                         $stripped .= $marker;
 482
 483                         if ( $close === '/>' ) {
 484                                 // Empty element tag, <tag />
 485                                 $content = null;
 486                                 $text = $inside;
 487                                 $tail = null;
 488                         } else {
 489                                 if( $element == '!--' ) {
 490                                         $end = '/(-->)/';
 491                                 } else {
 492                                         $end = "/(<\\/$element\\s*>)/i";
 493                                 }
 494                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 495                                 $content = $q[0];
 496                                 if( count( $q ) < 3 ) {
 497                                         # No end tag -- let it run out to the end of the text.
 498                                         $tail = '';
 499                                         $text = '';
 500                                 } else {
 501                                         $tail = $q[1];
 502                                         $text = $q[2];
 503                                 }
 504                         }
 505
 506                         $matches[$marker] = array( $element,
 507                                 $content,
 508                                 Sanitizer::decodeTagAttributes( $attributes ),
 509                                 "<$element$attributes$close$content$tail" );
 510                 }
 511                 return $stripped;
 512         }
 513
 514         /**
 515          * Strips and renders nowiki, pre, math, hiero
 516          * If $render is set, performs necessary rendering operations on plugins
 517          * Returns the text, and fills an array with data needed in unstrip()
 518          *
 519          * @param StripState $state
 520          *
 521          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 522          *  will be stripped in addition to other tags. This is important
 523          *  for section editing, where these comments cause confusion when
 524          *  counting the sections in the wikisource
 525          *
 526          * @param array dontstrip contains tags which should not be stripped;
 527          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 528          *
 529          * @private
 530          */
 531         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 532                 global $wgContLang;
 533                 wfProfileIn( __METHOD__ );
 534                 $render = ($this->mOutputType == OT_HTML);
 535
 536                 $uniq_prefix = $this->mUniqPrefix;
 537                 $commentState = new ReplacementArray;
 538                 $nowikiItems = array();
 539                 $generalItems = array();
 540
 541                 $elements = array_merge(
 542                         array( 'nowiki', 'gallery' ),
 543                         array_keys( $this->mTagHooks ) );
 544                 global $wgRawHtml;
 545                 if( $wgRawHtml ) {
 546                         $elements[] = 'html';
 547                 }
 548                 if( $this->mOptions->getUseTeX() ) {
 549                         $elements[] = 'math';
 550                 }
 551
 552                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 553                 foreach ( $elements AS $k => $v ) {
 554                         if ( !in_array ( $v , $dontstrip ) ) continue;
 555                         unset ( $elements[$k] );
 556                 }
 557
 558                 $matches = array();
 559                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 560
 561                 foreach( $matches as $marker => $data ) {
 562                         list( $element, $content, $params, $tag ) = $data;
 563                         if( $render ) {
 564                                 $tagName = strtolower( $element );
 565                                 wfProfileIn( __METHOD__."-render-$tagName" );
 566                                 switch( $tagName ) {
 567                                 case '!--':
 568                                         // Comment
 569                                         if( substr( $tag, -3 ) == '-->' ) {
 570                                                 $output = $tag;
 571                                         } else {
 572                                                 // Unclosed comment in input.
 573                                                 // Close it so later stripping can remove it
 574                                                 $output = "$tag-->";
 575                                         }
 576                                         break;
 577                                 case 'html':
 578                                         if( $wgRawHtml ) {
 579                                                 $output = $content;
 580                                                 break;
 581                                         }
 582                                         // Shouldn't happen otherwise. :)
 583                                 case 'nowiki':
 584                                         $output = Xml::escapeTagsOnly( $content );
 585                                         break;
 586                                 case 'math':
 587                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 588                                         break;
 589                                 case 'gallery':
 590                                         $output = $this->renderImageGallery( $content, $params );
 591                                         break;
 592                                 default:
 593                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 594                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 595                                                         array( $content, $params, $this ) );
 596                                         } else {
 597                                                 throw new MWException( "Invalid call hook $element" );
 598                                         }
 599                                 }
 600                                 wfProfileOut( __METHOD__."-render-$tagName" );
 601                         } else {
 602                                 // Just stripping tags; keep the source
 603                                 $output = $tag;
 604                         }
 605
 606                         // Unstrip the output, to support recursive strip() calls
 607                         $output = $state->unstripBoth( $output );
 608
 609                         if( !$stripcomments && $element == '!--' ) {
 610                                 $commentState->setPair( $marker, $output );
 611                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 612                                 $nowikiItems[$marker] = $output;
 613                         } else {
 614                                 $generalItems[$marker] = $output;
 615                         }
 616                 }
 617                 # Add the new items to the state
 618                 # We do this after the loop instead of during it to avoid slowing
 619                 # down the recursive unstrip
 620                 $state->nowiki->mergeArray( $nowikiItems );
 621                 $state->general->mergeArray( $generalItems );
 622
 623                 # Unstrip comments unless explicitly told otherwise.
 624                 # (The comments are always stripped prior to this point, so as to
 625                 # not invoke any extension tags / parser hooks contained within
 626                 # a comment.)
 627                 if ( !$stripcomments ) {
 628                         // Put them all back and forget them
 629                         $text = $commentState->replace( $text );
 630                 }
 631
 632                 wfProfileOut( __METHOD__ );
 633                 return $text;
 634         }
 635
 636         /**
 637          * Restores pre, math, and other extensions removed by strip()
 638          *
 639          * always call unstripNoWiki() after this one
 640          * @private
 641          * @deprecated use $this->mStripState->unstrip()
 642          */
 643         function unstrip( $text, $state ) {
 644                 return $state->unstripGeneral( $text );
 645         }
 646
 647         /**
 648          * Always call this after unstrip() to preserve the order
 649          *
 650          * @private
 651          * @deprecated use $this->mStripState->unstrip()
 652          */
 653         function unstripNoWiki( $text, $state ) {
 654                 return $state->unstripNoWiki( $text );
 655         }
 656
 657         /**
 658          * @deprecated use $this->mStripState->unstripBoth()
 659          */
 660         function unstripForHTML( $text ) {
 661                 return $this->mStripState->unstripBoth( $text );
 662         }
 663
 664         /**
 665          * Add an item to the strip state
 666          * Returns the unique tag which must be inserted into the stripped text
 667          * The tag will be replaced with the original text in unstrip()
 668          *
 669          * @private
 670          */
 671         function insertStripItem( $text, &$state ) {
 672                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 673                 $state->general->setPair( $rnd, $text );
 674                 return $rnd;
 675         }
 676
 677         /**
 678          * Interface with html tidy, used if $wgUseTidy = true.
 679          * If tidy isn't able to correct the markup, the original will be
 680          * returned in all its glory with a warning comment appended.
 681          *
 682          * Either the external tidy program or the in-process tidy extension
 683          * will be used depending on availability. Override the default
 684          * $wgTidyInternal setting to disable the internal if it's not working.
 685          *
 686          * @param string $text Hideous HTML input
 687          * @return string Corrected HTML output
 688          * @public
 689          * @static
 690          */
 691         function tidy( $text ) {
 692                 global $wgTidyInternal;
 693                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 694 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 695 '<head><title>test</title></head><body>'.$text.'</body></html>';
 696                 if( $wgTidyInternal ) {
 697                         $correctedtext = Parser::internalTidy( $wrappedtext );
 698                 } else {
 699                         $correctedtext = Parser::externalTidy( $wrappedtext );
 700                 }
 701                 if( is_null( $correctedtext ) ) {
 702                         wfDebug( "Tidy error detected!\n" );
 703                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 704                 }
 705                 return $correctedtext;
 706         }
 707
 708         /**
 709          * Spawn an external HTML tidy process and get corrected markup back from it.
 710          *
 711          * @private
 712          * @static
 713          */
 714         function externalTidy( $text ) {
 715                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 716                 $fname = 'Parser::externalTidy';
 717                 wfProfileIn( $fname );
 718
 719                 $cleansource = '';
 720                 $opts = ' -utf8';
 721
 722                 $descriptorspec = array(
 723                         0 => array('pipe', 'r'),
 724                         1 => array('pipe', 'w'),
 725                         2 => array('file', '/dev/null', 'a')  // FIXME: this line in UNIX-specific, it generates a warning on Windows, because /dev/null is not a valid Windows file.
 726                 );
 727                 $pipes = array();
 728                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 729                 if (is_resource($process)) {
 730                         // Theoretically, this style of communication could cause a deadlock
 731                         // here. If the stdout buffer fills up, then writes to stdin could
 732                         // block. This doesn't appear to happen with tidy, because tidy only
 733                         // writes to stdout after it's finished reading from stdin. Search
 734                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 735                         fwrite($pipes[0], $text);
 736                         fclose($pipes[0]);
 737                         while (!feof($pipes[1])) {
 738                                 $cleansource .= fgets($pipes[1], 1024);
 739                         }
 740                         fclose($pipes[1]);
 741                         proc_close($process);
 742                 }
 743
 744                 wfProfileOut( $fname );
 745
 746                 if( $cleansource == '' && $text != '') {
 747                         // Some kind of error happened, so we couldn't get the corrected text.
 748                         // Just give up; we'll use the source text and append a warning.
 749                         return null;
 750                 } else {
 751                         return $cleansource;
 752                 }
 753         }
 754
 755         /**
 756          * Use the HTML tidy PECL extension to use the tidy library in-process,
 757          * saving the overhead of spawning a new process. Currently written to
 758          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 759          *
 760          * 'pear install tidy' should be able to compile the extension module.
 761          *
 762          * @private
 763          * @static
 764          */
 765         function internalTidy( $text ) {
 766                 global $wgTidyConf;
 767                 $fname = 'Parser::internalTidy';
 768                 wfProfileIn( $fname );
 769
 770                 tidy_load_config( $wgTidyConf );
 771                 tidy_set_encoding( 'utf8' );
 772                 tidy_parse_string( $text );
 773                 tidy_clean_repair();
 774                 if( tidy_get_status() == 2 ) {
 775                         // 2 is magic number for fatal error
 776                         // http://www.php.net/manual/en/function.tidy-get-status.php
 777                         $cleansource = null;
 778                 } else {
 779                         $cleansource = tidy_get_output();
 780                 }
 781                 wfProfileOut( $fname );
 782                 return $cleansource;
 783         }
 784
 785         /**
 786          * parse the wiki syntax used to render tables
 787          *
 788          * @private
 789          */
 790         function doTableStuff ( $text ) {
 791                 $fname = 'Parser::doTableStuff';
 792                 wfProfileIn( $fname );
 793
 794                 $lines = explode ( "\n" , $text );
 795                 $td_history = array (); // Is currently a td tag open?
 796                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 797                 $tr_history = array (); // Is currently a tr tag open?
 798                 $tr_attributes = array (); // history of tr attributes
 799                 $has_opened_tr = array(); // Did this table open a <tr> element?
 800                 $indent_level = 0; // indent level of the table
 801                 foreach ( $lines as $key => $line )
 802                 {
 803                         $line = trim ( $line );
 804
 805                         if( $line == '' ) { // empty line, go to next line
 806                                 continue;
 807                         }
 808                         $first_character = $line{0};
 809                         $matches = array();
 810
 811                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 812                                 // First check if we are starting a new table
 813                                 $indent_level = strlen( $matches[1] );
 814
 815                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 816                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 817
 818                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 819                                 array_push ( $td_history , false );
 820                                 array_push ( $last_tag_history , '' );
 821                                 array_push ( $tr_history , false );
 822                                 array_push ( $tr_attributes , '' );
 823                                 array_push ( $has_opened_tr , false );
 824                         } else if ( count ( $td_history ) == 0 ) {
 825                                 // Don't do any of the following
 826                                 continue;
 827                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 828                                 // We are ending a table
 829                                 $line = '</table>' . substr ( $line , 2 );
 830                                 $last_tag = array_pop ( $last_tag_history );
 831
 832                                 if ( !array_pop ( $has_opened_tr ) ) {
 833                                         $line = "<tr><td></td></tr>{$line}";
 834                                 }
 835
 836                                 if ( array_pop ( $tr_history ) ) {
 837                                         $line = "</tr>{$line}";
 838                                 }
 839
 840                                 if ( array_pop ( $td_history ) ) {
 841                                         $line = "</{$last_tag}>{$line}";
 842                                 }
 843                                 array_pop ( $tr_attributes );
 844                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 845                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 846                                 // Now we have a table row
 847                                 $line = preg_replace( '#^\|-+#', '', $line );
 848
 849                                 // Whats after the tag is now only attributes
 850                                 $attributes = $this->mStripState->unstripBoth( $line );
 851                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 852                                 array_pop ( $tr_attributes );
 853                                 array_push ( $tr_attributes , $attributes );
 854
 855                                 $line = '';
 856                                 $last_tag = array_pop ( $last_tag_history );
 857                                 array_pop ( $has_opened_tr );
 858                                 array_push ( $has_opened_tr , true );
 859
 860                                 if ( array_pop ( $tr_history ) ) {
 861                                         $line = '</tr>';
 862                                 }
 863
 864                                 if ( array_pop ( $td_history ) ) {
 865                                         $line = "</{$last_tag}>{$line}";
 866                                 }
 867
 868                                 $lines[$key] = $line;
 869                                 array_push ( $tr_history , false );
 870                                 array_push ( $td_history , false );
 871                                 array_push ( $last_tag_history , '' );
 872                         }
 873                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 874                                 // This might be cell elements, td, th or captions
 875                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 876                                         $first_character = '+';
 877                                         $line = substr ( $line , 1 );
 878                                 }
 879
 880                                 $line = substr ( $line , 1 );
 881
 882                                 if ( $first_character == '!' ) {
 883                                         $line = str_replace ( '!!' , '||' , $line );
 884                                 }
 885
 886                                 // Split up multiple cells on the same line.
 887                                 // FIXME : This can result in improper nesting of tags processed
 888                                 // by earlier parser steps, but should avoid splitting up eg
 889                                 // attribute values containing literal "||".
 890                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 891
 892                                 $lines[$key] = '';
 893
 894                                 // Loop through each table cell
 895                                 foreach ( $cells as $cell )
 896                                 {
 897                                         $previous = '';
 898                                         if ( $first_character != '+' )
 899                                         {
 900                                                 $tr_after = array_pop ( $tr_attributes );
 901                                                 if ( !array_pop ( $tr_history ) ) {
 902                                                         $previous = "<tr{$tr_after}>\n";
 903                                                 }
 904                                                 array_push ( $tr_history , true );
 905                                                 array_push ( $tr_attributes , '' );
 906                                                 array_pop ( $has_opened_tr );
 907                                                 array_push ( $has_opened_tr , true );
 908                                         }
 909
 910                                         $last_tag = array_pop ( $last_tag_history );
 911
 912                                         if ( array_pop ( $td_history ) ) {
 913                                                 $previous = "</{$last_tag}>{$previous}";
 914                                         }
 915
 916                                         if ( $first_character == '|' ) {
 917                                                 $last_tag = 'td';
 918                                         } else if ( $first_character == '!' ) {
 919                                                 $last_tag = 'th';
 920                                         } else if ( $first_character == '+' ) {
 921                                                 $last_tag = 'caption';
 922                                         } else {
 923                                                 $last_tag = '';
 924                                         }
 925
 926                                         array_push ( $last_tag_history , $last_tag );
 927
 928                                         // A cell could contain both parameters and data
 929                                         $cell_data = explode ( '|' , $cell , 2 );
 930
 931                                         // Bug 553: Note that a '|' inside an invalid link should not
 932                                         // be mistaken as delimiting cell parameters
 933                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 934                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 935                                         } else if ( count ( $cell_data ) == 1 )
 936                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 937                                         else {
 938                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 939                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 940                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 941                                         }
 942
 943                                         $lines[$key] .= $cell;
 944                                         array_push ( $td_history , true );
 945                                 }
 946                         }
 947                 }
 948
 949                 // Closing open td, tr && table
 950                 while ( count ( $td_history ) > 0 )
 951                 {
 952                         if ( array_pop ( $td_history ) ) {
 953                                 $lines[] = '</td>' ;
 954                         }
 955                         if ( array_pop ( $tr_history ) ) {
 956                                 $lines[] = '</tr>' ;
 957                         }
 958                         if ( !array_pop ( $has_opened_tr ) ) {
 959                                 $lines[] = "<tr><td></td></tr>" ;
 960                         }
 961
 962                         $lines[] = '</table>' ;
 963                 }
 964
 965                 $output = implode ( "\n" , $lines ) ;
 966
 967                 // special case: don't return empty table
 968                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 969                         $output = '';
 970                 }
 971
 972                 wfProfileOut( $fname );
 973
 974                 return $output;
 975         }
 976
 977         /**
 978          * Helper function for parse() that transforms wiki markup into
 979          * HTML. Only called for $mOutputType == OT_HTML.
 980          *
 981          * @private
 982          */
 983         function internalParse( $text ) {
 984                 $args = array();
 985                 $isMain = true;
 986                 $fname = 'Parser::internalParse';
 987                 wfProfileIn( $fname );
 988
 989                 # Hook to suspend the parser in this state
 990                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 991                         wfProfileOut( $fname );
 992                         return $text ;
 993                 }
 994
 995                 # Remove <noinclude> tags and <includeonly> sections
 996                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 997                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 998                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 999
1000                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1001
1002                 $text = $this->replaceVariables( $text, $args );
1003                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text ) );
1004
1005                 // Tables need to come after variable replacement for things to work
1006                 // properly; putting them before other transformations should keep
1007                 // exciting things like link expansions from showing up in surprising
1008                 // places.
1009                 $text = $this->doTableStuff( $text );
1010
1011                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1012
1013                 $text = $this->stripToc( $text );
1014                 $this->stripNoGallery( $text );
1015                 $text = $this->doHeadings( $text );
1016                 if($this->mOptions->getUseDynamicDates()) {
1017                         $df =& DateFormatter::getInstance();
1018                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1019                 }
1020                 $text = $this->doAllQuotes( $text );
1021                 $text = $this->replaceInternalLinks( $text );
1022                 $text = $this->replaceExternalLinks( $text );
1023
1024                 # replaceInternalLinks may sometimes leave behind
1025                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1026                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1027
1028                 $text = $this->doMagicLinks( $text );
1029                 $text = $this->formatHeadings( $text, $isMain );
1030
1031                 wfProfileOut( $fname );
1032                 return $text;
1033         }
1034
1035         /**
1036          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1037          * magic external links.
1038          *
1039          * @private
1040          */
1041         function &doMagicLinks( &$text ) {
1042                 wfProfileIn( __METHOD__ );
1043                 $text = preg_replace_callback(
1044                         '!(?:                           # Start cases
1045                             <a.*?</a> |                 # Skip link text
1046                             <.*?> |                     # Skip stuff inside HTML elements
1047                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1048                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1049                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1050                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1051                                      [0-9Xx]                 # check digit
1052                                    \b)
1053                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1054                 wfProfileOut( __METHOD__ );
1055                 return $text;
1056         }
1057
1058         function magicLinkCallback( $m ) {
1059                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1060                         # Skip HTML element
1061                         return $m[0];
1062                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1063                         $isbn = $m[2];
1064                         $num = strtr( $isbn, array(
1065                                 '-' => '',
1066                                 ' ' => '',
1067                                 'x' => 'X',
1068                         ));
1069                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1070                         $text = '<a href="' .
1071                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1072                                 "\" class=\"internal\">ISBN $isbn</a>";
1073                 } else {
1074                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1075                                 $keyword = 'RFC';
1076                                 $urlmsg = 'rfcurl';
1077                                 $id = $m[1];
1078                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1079                                 $keyword = 'PMID';
1080                                 $urlmsg = 'pubmedurl';
1081                                 $id = $m[1];
1082                         } else {
1083                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1084                                         substr($m[0], 0, 20 ) . '"' );
1085                         }
1086
1087                         $url = wfMsg( $urlmsg, $id);
1088                         $sk = $this->mOptions->getSkin();
1089                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1090                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1091                 }
1092                 return $text;
1093         }
1094
1095         /**
1096          * Parse headers and return html
1097          *
1098          * @private
1099          */
1100         function doHeadings( $text ) {
1101                 $fname = 'Parser::doHeadings';
1102                 wfProfileIn( $fname );
1103                 for ( $i = 6; $i >= 1; --$i ) {
1104                         $h = str_repeat( '=', $i );
1105                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1106                           "<h{$i}>\\1</h{$i}>\\2", $text );
1107                 }
1108                 wfProfileOut( $fname );
1109                 return $text;
1110         }
1111
1112         /**
1113          * Replace single quotes with HTML markup
1114          * @private
1115          * @return string the altered text
1116          */
1117         function doAllQuotes( $text ) {
1118                 $fname = 'Parser::doAllQuotes';
1119                 wfProfileIn( $fname );
1120                 $outtext = '';
1121                 $lines = explode( "\n", $text );
1122                 foreach ( $lines as $line ) {
1123                         $outtext .= $this->doQuotes ( $line ) . "\n";
1124                 }
1125                 $outtext = substr($outtext, 0,-1);
1126                 wfProfileOut( $fname );
1127                 return $outtext;
1128         }
1129
1130         /**
1131          * Helper function for doAllQuotes()
1132          * @private
1133          */
1134         function doQuotes( $text ) {
1135                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1136                 if ( count( $arr ) == 1 )
1137                         return $text;
1138                 else
1139                 {
1140                         # First, do some preliminary work. This may shift some apostrophes from
1141                         # being mark-up to being text. It also counts the number of occurrences
1142                         # of bold and italics mark-ups.
1143                         $i = 0;
1144                         $numbold = 0;
1145                         $numitalics = 0;
1146                         foreach ( $arr as $r )
1147                         {
1148                                 if ( ( $i % 2 ) == 1 )
1149                                 {
1150                                         # If there are ever four apostrophes, assume the first is supposed to
1151                                         # be text, and the remaining three constitute mark-up for bold text.
1152                                         if ( strlen( $arr[$i] ) == 4 )
1153                                         {
1154                                                 $arr[$i-1] .= "'";
1155                                                 $arr[$i] = "'''";
1156                                         }
1157                                         # If there are more than 5 apostrophes in a row, assume they're all
1158                                         # text except for the last 5.
1159                                         else if ( strlen( $arr[$i] ) > 5 )
1160                                         {
1161                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1162                                                 $arr[$i] = "'''''";
1163                                         }
1164                                         # Count the number of occurrences of bold and italics mark-ups.
1165                                         # We are not counting sequences of five apostrophes.
1166                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1167                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1168                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1169                                 }
1170                                 $i++;
1171                         }
1172
1173                         # If there is an odd number of both bold and italics, it is likely
1174                         # that one of the bold ones was meant to be an apostrophe followed
1175                         # by italics. Which one we cannot know for certain, but it is more
1176                         # likely to be one that has a single-letter word before it.
1177                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1178                         {
1179                                 $i = 0;
1180                                 $firstsingleletterword = -1;
1181                                 $firstmultiletterword = -1;
1182                                 $firstspace = -1;
1183                                 foreach ( $arr as $r )
1184                                 {
1185                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1186                                         {
1187                                                 $x1 = substr ($arr[$i-1], -1);
1188                                                 $x2 = substr ($arr[$i-1], -2, 1);
1189                                                 if ($x1 == ' ') {
1190                                                         if ($firstspace == -1) $firstspace = $i;
1191                                                 } else if ($x2 == ' ') {
1192                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1193                                                 } else {
1194                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1195                                                 }
1196                                         }
1197                                         $i++;
1198                                 }
1199
1200                                 # If there is a single-letter word, use it!
1201                                 if ($firstsingleletterword > -1)
1202                                 {
1203                                         $arr [ $firstsingleletterword ] = "''";
1204                                         $arr [ $firstsingleletterword-1 ] .= "'";
1205                                 }
1206                                 # If not, but there's a multi-letter word, use that one.
1207                                 else if ($firstmultiletterword > -1)
1208                                 {
1209                                         $arr [ $firstmultiletterword ] = "''";
1210                                         $arr [ $firstmultiletterword-1 ] .= "'";
1211                                 }
1212                                 # ... otherwise use the first one that has neither.
1213                                 # (notice that it is possible for all three to be -1 if, for example,
1214                                 # there is only one pentuple-apostrophe in the line)
1215                                 else if ($firstspace > -1)
1216                                 {
1217                                         $arr [ $firstspace ] = "''";
1218                                         $arr [ $firstspace-1 ] .= "'";
1219                                 }
1220                         }
1221
1222                         # Now let's actually convert our apostrophic mush to HTML!
1223                         $output = '';
1224                         $buffer = '';
1225                         $state = '';
1226                         $i = 0;
1227                         foreach ($arr as $r)
1228                         {
1229                                 if (($i % 2) == 0)
1230                                 {
1231                                         if ($state == 'both')
1232                                                 $buffer .= $r;
1233                                         else
1234                                                 $output .= $r;
1235                                 }
1236                                 else
1237                                 {
1238                                         if (strlen ($r) == 2)
1239                                         {
1240                                                 if ($state == 'i')
1241                                                 { $output .= '</i>'; $state = ''; }
1242                                                 else if ($state == 'bi')
1243                                                 { $output .= '</i>'; $state = 'b'; }
1244                                                 else if ($state == 'ib')
1245                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1246                                                 else if ($state == 'both')
1247                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1248                                                 else # $state can be 'b' or ''
1249                                                 { $output .= '<i>'; $state .= 'i'; }
1250                                         }
1251                                         else if (strlen ($r) == 3)
1252                                         {
1253                                                 if ($state == 'b')
1254                                                 { $output .= '</b>'; $state = ''; }
1255                                                 else if ($state == 'bi')
1256                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1257                                                 else if ($state == 'ib')
1258                                                 { $output .= '</b>'; $state = 'i'; }
1259                                                 else if ($state == 'both')
1260                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1261                                                 else # $state can be 'i' or ''
1262                                                 { $output .= '<b>'; $state .= 'b'; }
1263                                         }
1264                                         else if (strlen ($r) == 5)
1265                                         {
1266                                                 if ($state == 'b')
1267                                                 { $output .= '</b><i>'; $state = 'i'; }
1268                                                 else if ($state == 'i')
1269                                                 { $output .= '</i><b>'; $state = 'b'; }
1270                                                 else if ($state == 'bi')
1271                                                 { $output .= '</i></b>'; $state = ''; }
1272                                                 else if ($state == 'ib')
1273                                                 { $output .= '</b></i>'; $state = ''; }
1274                                                 else if ($state == 'both')
1275                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1276                                                 else # ($state == '')
1277                                                 { $buffer = ''; $state = 'both'; }
1278                                         }
1279                                 }
1280                                 $i++;
1281                         }
1282                         # Now close all remaining tags.  Notice that the order is important.
1283                         if ($state == 'b' || $state == 'ib')
1284                                 $output .= '</b>';
1285                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1286                                 $output .= '</i>';
1287                         if ($state == 'bi')
1288                                 $output .= '</b>';
1289                         # There might be lonely ''''', so make sure we have a buffer
1290                         if ($state == 'both' && $buffer)
1291                                 $output .= '<b><i>'.$buffer.'</i></b>';
1292                         return $output;
1293                 }
1294         }
1295
1296         /**
1297          * Replace external links
1298          *
1299          * Note: this is all very hackish and the order of execution matters a lot.
1300          * Make sure to run maintenance/parserTests.php if you change this code.
1301          *
1302          * @private
1303          */
1304         function replaceExternalLinks( $text ) {
1305                 global $wgContLang;
1306                 $fname = 'Parser::replaceExternalLinks';
1307                 wfProfileIn( $fname );
1308
1309                 $sk = $this->mOptions->getSkin();
1310
1311                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1312
1313                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1314
1315                 $i = 0;
1316                 while ( $i<count( $bits ) ) {
1317                         $url = $bits[$i++];
1318                         $protocol = $bits[$i++];
1319                         $text = $bits[$i++];
1320                         $trail = $bits[$i++];
1321
1322                         # The characters '<' and '>' (which were escaped by
1323                         # removeHTMLtags()) should not be included in
1324                         # URLs, per RFC 2396.
1325                         $m2 = array();
1326                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1327                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1328                                 $url = substr($url, 0, $m2[0][1]);
1329                         }
1330
1331                         # If the link text is an image URL, replace it with an <img> tag
1332                         # This happened by accident in the original parser, but some people used it extensively
1333                         $img = $this->maybeMakeExternalImage( $text );
1334                         if ( $img !== false ) {
1335                                 $text = $img;
1336                         }
1337
1338                         $dtrail = '';
1339
1340                         # Set linktype for CSS - if URL==text, link is essentially free
1341                         $linktype = ($text == $url) ? 'free' : 'text';
1342
1343                         # No link text, e.g. [http://domain.tld/some.link]
1344                         if ( $text == '' ) {
1345                                 # Autonumber if allowed. See bug #5918
1346                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1347                                         $text = '[' . ++$this->mAutonumber . ']';
1348                                         $linktype = 'autonumber';
1349                                 } else {
1350                                         # Otherwise just use the URL
1351                                         $text = htmlspecialchars( $url );
1352                                         $linktype = 'free';
1353                                 }
1354                         } else {
1355                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1356                                 # Check for trail
1357                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1358                         }
1359
1360                         $text = $wgContLang->markNoConversion($text);
1361
1362                         $url = Sanitizer::cleanUrl( $url );
1363
1364                         # Process the trail (i.e. everything after this link up until start of the next link),
1365                         # replacing any non-bracketed links
1366                         $trail = $this->replaceFreeExternalLinks( $trail );
1367
1368                         # Use the encoded URL
1369                         # This means that users can paste URLs directly into the text
1370                         # Funny characters like &ouml; aren't valid in URLs anyway
1371                         # This was changed in August 2004
1372                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1373
1374                         # Register link in the output object.
1375                         # Replace unnecessary URL escape codes with the referenced character
1376                         # This prevents spammers from hiding links from the filters
1377                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1378                         $this->mOutput->addExternalLink( $pasteurized );
1379                 }
1380
1381                 wfProfileOut( $fname );
1382                 return $s;
1383         }
1384
1385         /**
1386          * Replace anything that looks like a URL with a link
1387          * @private
1388          */
1389         function replaceFreeExternalLinks( $text ) {
1390                 global $wgContLang;
1391                 $fname = 'Parser::replaceFreeExternalLinks';
1392                 wfProfileIn( $fname );
1393
1394                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1395                 $s = array_shift( $bits );
1396                 $i = 0;
1397
1398                 $sk = $this->mOptions->getSkin();
1399
1400                 while ( $i < count( $bits ) ){
1401                         $protocol = $bits[$i++];
1402                         $remainder = $bits[$i++];
1403
1404                         $m = array();
1405                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1406                                 # Found some characters after the protocol that look promising
1407                                 $url = $protocol . $m[1];
1408                                 $trail = $m[2];
1409
1410                                 # special case: handle urls as url args:
1411                                 # http://www.example.com/foo?=http://www.example.com/bar
1412                                 if(strlen($trail) == 0 &&
1413                                         isset($bits[$i]) &&
1414                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1415                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1416                                 {
1417                                         # add protocol, arg
1418                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1419                                         $i += 2;
1420                                         $trail = $m[2];
1421                                 }
1422
1423                                 # The characters '<' and '>' (which were escaped by
1424                                 # removeHTMLtags()) should not be included in
1425                                 # URLs, per RFC 2396.
1426                                 $m2 = array();
1427                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1428                                         $trail = substr($url, $m2[0][1]) . $trail;
1429                                         $url = substr($url, 0, $m2[0][1]);
1430                                 }
1431
1432                                 # Move trailing punctuation to $trail
1433                                 $sep = ',;\.:!?';
1434                                 # If there is no left bracket, then consider right brackets fair game too
1435                                 if ( strpos( $url, '(' ) === false ) {
1436                                         $sep .= ')';
1437                                 }
1438
1439                                 $numSepChars = strspn( strrev( $url ), $sep );
1440                                 if ( $numSepChars ) {
1441                                         $trail = substr( $url, -$numSepChars ) . $trail;
1442                                         $url = substr( $url, 0, -$numSepChars );
1443                                 }
1444
1445                                 $url = Sanitizer::cleanUrl( $url );
1446
1447                                 # Is this an external image?
1448                                 $text = $this->maybeMakeExternalImage( $url );
1449                                 if ( $text === false ) {
1450                                         # Not an image, make a link
1451                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1452                                         # Register it in the output object...
1453                                         # Replace unnecessary URL escape codes with their equivalent characters
1454                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1455                                         $this->mOutput->addExternalLink( $pasteurized );
1456                                 }
1457                                 $s .= $text . $trail;
1458                         } else {
1459                                 $s .= $protocol . $remainder;
1460                         }
1461                 }
1462                 wfProfileOut( $fname );
1463                 return $s;
1464         }
1465
1466         /**
1467          * Replace unusual URL escape codes with their equivalent characters
1468          * @param string
1469          * @return string
1470          * @static
1471          * @fixme This can merge genuinely required bits in the path or query string,
1472          *        breaking legit URLs. A proper fix would treat the various parts of
1473          *        the URL differently; as a workaround, just use the output for
1474          *        statistical records, not for actual linking/output.
1475          */
1476         static function replaceUnusualEscapes( $url ) {
1477                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1478                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1479         }
1480
1481         /**
1482          * Callback function used in replaceUnusualEscapes().
1483          * Replaces unusual URL escape codes with their equivalent character
1484          * @static
1485          * @private
1486          */
1487         private static function replaceUnusualEscapesCallback( $matches ) {
1488                 $char = urldecode( $matches[0] );
1489                 $ord = ord( $char );
1490                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1491                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1492                         // No, shouldn't be escaped
1493                         return $char;
1494                 } else {
1495                         // Yes, leave it escaped
1496                         return $matches[0];
1497                 }
1498         }
1499
1500         /**
1501          * make an image if it's allowed, either through the global
1502          * option or through the exception
1503          * @private
1504          */
1505         function maybeMakeExternalImage( $url ) {
1506                 $sk = $this->mOptions->getSkin();
1507                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1508                 $imagesexception = !empty($imagesfrom);
1509                 $text = false;
1510                 if ( $this->mOptions->getAllowExternalImages()
1511                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1512                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1513                                 # Image found
1514                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1515                         }
1516                 }
1517                 return $text;
1518         }
1519
1520         /**
1521          * Process [[ ]] wikilinks
1522          *
1523          * @private
1524          */
1525         function replaceInternalLinks( $s ) {
1526                 global $wgContLang;
1527                 static $fname = 'Parser::replaceInternalLinks' ;
1528
1529                 wfProfileIn( $fname );
1530
1531                 wfProfileIn( $fname.'-setup' );
1532                 static $tc = FALSE;
1533                 # the % is needed to support urlencoded titles as well
1534                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1535
1536                 $sk = $this->mOptions->getSkin();
1537
1538                 #split the entire text string on occurences of [[
1539                 $a = explode( '[[', ' ' . $s );
1540                 #get the first element (all text up to first [[), and remove the space we added
1541                 $s = array_shift( $a );
1542                 $s = substr( $s, 1 );
1543
1544                 # Match a link having the form [[namespace:link|alternate]]trail
1545                 static $e1 = FALSE;
1546                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1547                 # Match cases where there is no "]]", which might still be images
1548                 static $e1_img = FALSE;
1549                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1550                 # Match the end of a line for a word that's not followed by whitespace,
1551                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1552                 $e2 = wfMsgForContent( 'linkprefix' );
1553
1554                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1555                 if( is_null( $this->mTitle ) ) {
1556                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1557                 }
1558                 $nottalk = !$this->mTitle->isTalkPage();
1559
1560                 if ( $useLinkPrefixExtension ) {
1561                         $m = array();
1562                         if ( preg_match( $e2, $s, $m ) ) {
1563                                 $first_prefix = $m[2];
1564                         } else {
1565                                 $first_prefix = false;
1566                         }
1567                 } else {
1568                         $prefix = '';
1569                 }
1570
1571                 if($wgContLang->hasVariants()) {
1572                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1573                 } else {
1574                         $selflink = array($this->mTitle->getPrefixedText());
1575                 }
1576                 $useSubpages = $this->areSubpagesAllowed();
1577                 wfProfileOut( $fname.'-setup' );
1578
1579                 # Loop for each link
1580                 for ($k = 0; isset( $a[$k] ); $k++) {
1581                         $line = $a[$k];
1582                         if ( $useLinkPrefixExtension ) {
1583                                 wfProfileIn( $fname.'-prefixhandling' );
1584                                 if ( preg_match( $e2, $s, $m ) ) {
1585                                         $prefix = $m[2];
1586                                         $s = $m[1];
1587                                 } else {
1588                                         $prefix='';
1589                                 }
1590                                 # first link
1591                                 if($first_prefix) {
1592                                         $prefix = $first_prefix;
1593                                         $first_prefix = false;
1594                                 }
1595                                 wfProfileOut( $fname.'-prefixhandling' );
1596                         }
1597
1598                         $might_be_img = false;
1599
1600                         wfProfileIn( "$fname-e1" );
1601                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1602                                 $text = $m[2];
1603                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1604                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1605                                 # the real problem is with the $e1 regex
1606                                 # See bug 1300.
1607                                 #
1608                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1609                                 # and no image is in sight. See bug 2095.
1610                                 #
1611                                 if( $text !== '' &&
1612                                         substr( $m[3], 0, 1 ) === ']' &&
1613                                         strpos($text, '[') !== false
1614                                 )
1615                                 {
1616                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1617                                         $m[3] = substr( $m[3], 1 );
1618                                 }
1619                                 # fix up urlencoded title texts
1620                                 if( strpos( $m[1], '%' ) !== false ) {
1621                                         # Should anchors '#' also be rejected?
1622                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1623                                 }
1624                                 $trail = $m[3];
1625                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1626                                 $might_be_img = true;
1627                                 $text = $m[2];
1628                                 if ( strpos( $m[1], '%' ) !== false ) {
1629                                         $m[1] = urldecode($m[1]);
1630                                 }
1631                                 $trail = "";
1632                         } else { # Invalid form; output directly
1633                                 $s .= $prefix . '[[' . $line ;
1634                                 wfProfileOut( "$fname-e1" );
1635                                 continue;
1636                         }
1637                         wfProfileOut( "$fname-e1" );
1638                         wfProfileIn( "$fname-misc" );
1639
1640                         # Don't allow internal links to pages containing
1641                         # PROTO: where PROTO is a valid URL protocol; these
1642                         # should be external links.
1643                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1644                                 $s .= $prefix . '[[' . $line ;
1645                                 continue;
1646                         }
1647
1648                         # Make subpage if necessary
1649                         if( $useSubpages ) {
1650                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1651                         } else {
1652                                 $link = $m[1];
1653                         }
1654
1655                         $noforce = (substr($m[1], 0, 1) != ':');
1656                         if (!$noforce) {
1657                                 # Strip off leading ':'
1658                                 $link = substr($link, 1);
1659                         }
1660
1661                         wfProfileOut( "$fname-misc" );
1662                         wfProfileIn( "$fname-title" );
1663                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1664                         if( !$nt ) {
1665                                 $s .= $prefix . '[[' . $line;
1666                                 wfProfileOut( "$fname-title" );
1667                                 continue;
1668                         }
1669
1670                         $ns = $nt->getNamespace();
1671                         $iw = $nt->getInterWiki();
1672                         wfProfileOut( "$fname-title" );
1673
1674                         if ($might_be_img) { # if this is actually an invalid link
1675                                 wfProfileIn( "$fname-might_be_img" );
1676                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1677                                         $found = false;
1678                                         while (isset ($a[$k+1]) ) {
1679                                                 #look at the next 'line' to see if we can close it there
1680                                                 $spliced = array_splice( $a, $k + 1, 1 );
1681                                                 $next_line = array_shift( $spliced );
1682                                                 $m = explode( ']]', $next_line, 3 );
1683                                                 if ( count( $m ) == 3 ) {
1684                                                         # the first ]] closes the inner link, the second the image
1685                                                         $found = true;
1686                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1687                                                         $trail = $m[2];
1688                                                         break;
1689                                                 } elseif ( count( $m ) == 2 ) {
1690                                                         #if there's exactly one ]] that's fine, we'll keep looking
1691                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1692                                                 } else {
1693                                                         #if $next_line is invalid too, we need look no further
1694                                                         $text .= '[[' . $next_line;
1695                                                         break;
1696                                                 }
1697                                         }
1698                                         if ( !$found ) {
1699                                                 # we couldn't find the end of this imageLink, so output it raw
1700                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1701                                                 $text = $this->replaceInternalLinks($text);
1702                                                 $s .= "{$prefix}[[$link|$text";
1703                                                 # note: no $trail, because without an end, there *is* no trail
1704                                                 wfProfileOut( "$fname-might_be_img" );
1705                                                 continue;
1706                                         }
1707                                 } else { #it's not an image, so output it raw
1708                                         $s .= "{$prefix}[[$link|$text";
1709                                         # note: no $trail, because without an end, there *is* no trail
1710                                         wfProfileOut( "$fname-might_be_img" );
1711                                         continue;
1712                                 }
1713                                 wfProfileOut( "$fname-might_be_img" );
1714                         }
1715
1716                         $wasblank = ( '' == $text );
1717                         if( $wasblank ) $text = $link;
1718
1719                         # Link not escaped by : , create the various objects
1720                         if( $noforce ) {
1721
1722                                 # Interwikis
1723                                 wfProfileIn( "$fname-interwiki" );
1724                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1725                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1726                                         $s = rtrim($s . $prefix);
1727                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1728                                         wfProfileOut( "$fname-interwiki" );
1729                                         continue;
1730                                 }
1731                                 wfProfileOut( "$fname-interwiki" );
1732
1733                                 if ( $ns == NS_IMAGE ) {
1734                                         wfProfileIn( "$fname-image" );
1735                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1736                                                 # recursively parse links inside the image caption
1737                                                 # actually, this will parse them in any other parameters, too,
1738                                                 # but it might be hard to fix that, and it doesn't matter ATM
1739                                                 $text = $this->replaceExternalLinks($text);
1740                                                 $text = $this->replaceInternalLinks($text);
1741
1742                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1743                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1744                                                 $this->mOutput->addImage( $nt->getDBkey() );
1745
1746                                                 wfProfileOut( "$fname-image" );
1747                                                 continue;
1748                                         } else {
1749                                                 # We still need to record the image's presence on the page
1750                                                 $this->mOutput->addImage( $nt->getDBkey() );
1751                                         }
1752                                         wfProfileOut( "$fname-image" );
1753
1754                                 }
1755
1756                                 if ( $ns == NS_CATEGORY ) {
1757                                         wfProfileIn( "$fname-category" );
1758                                         $s = rtrim($s . "\n"); # bug 87
1759
1760                                         if ( $wasblank ) {
1761                                                 $sortkey = $this->getDefaultSort();
1762                                         } else {
1763                                                 $sortkey = $text;
1764                                         }
1765                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1766                                         $sortkey = str_replace( "\n", '', $sortkey );
1767                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1768                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1769
1770                                         /**
1771                                          * Strip the whitespace Category links produce, see bug 87
1772                                          * @todo We might want to use trim($tmp, "\n") here.
1773                                          */
1774                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1775
1776                                         wfProfileOut( "$fname-category" );
1777                                         continue;
1778                                 }
1779                         }
1780
1781                         # Self-link checking
1782                         if( $nt->getFragment() === '' ) {
1783                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1784                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1785                                         continue;
1786                                 }
1787                         }
1788
1789                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1790                         if( $ns == NS_MEDIA ) {
1791                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1792                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1793                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1794                                 $this->mOutput->addImage( $nt->getDBkey() );
1795                                 continue;
1796                         } elseif( $ns == NS_SPECIAL ) {
1797                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1798                                 continue;
1799                         } elseif( $ns == NS_IMAGE ) {
1800                                 $img = new Image( $nt );
1801                                 if( $img->exists() ) {
1802                                         // Force a blue link if the file exists; may be a remote
1803                                         // upload on the shared repository, and we want to see its
1804                                         // auto-generated page.
1805                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1806                                         $this->mOutput->addLink( $nt );
1807                                         continue;
1808                                 }
1809                         }
1810                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1811                 }
1812                 wfProfileOut( $fname );
1813                 return $s;
1814         }
1815
1816         /**
1817          * Make a link placeholder. The text returned can be later resolved to a real link with
1818          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1819          * parsing of interwiki links, and secondly to allow all existence checks and
1820          * article length checks (for stub links) to be bundled into a single query.
1821          *
1822          */
1823         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1824                 wfProfileIn( __METHOD__ );
1825                 if ( ! is_object($nt) ) {
1826                         # Fail gracefully
1827                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1828                 } else {
1829                         # Separate the link trail from the rest of the link
1830                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1831
1832                         if ( $nt->isExternal() ) {
1833                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1834                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1835                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1836                         } else {
1837                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1838                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1839                                 $this->mLinkHolders['queries'][] = $query;
1840                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1841                                 $this->mLinkHolders['titles'][] = $nt;
1842
1843                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1844                         }
1845                 }
1846                 wfProfileOut( __METHOD__ );
1847                 return $retVal;
1848         }
1849
1850         /**
1851          * Render a forced-blue link inline; protect against double expansion of
1852          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1853          * Since this little disaster has to split off the trail text to avoid
1854          * breaking URLs in the following text without breaking trails on the
1855          * wiki links, it's been made into a horrible function.
1856          *
1857          * @param Title $nt
1858          * @param string $text
1859          * @param string $query
1860          * @param string $trail
1861          * @param string $prefix
1862          * @return string HTML-wikitext mix oh yuck
1863          */
1864         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1865                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1866                 $sk = $this->mOptions->getSkin();
1867                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1868                 return $this->armorLinks( $link ) . $trail;
1869         }
1870
1871         /**
1872          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1873          * going to go through further parsing steps before inline URL expansion.
1874          *
1875          * In particular this is important when using action=render, which causes
1876          * full URLs to be included.
1877          *
1878          * Oh man I hate our multi-layer parser!
1879          *
1880          * @param string more-or-less HTML
1881          * @return string less-or-more HTML with NOPARSE bits
1882          */
1883         function armorLinks( $text ) {
1884                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1885                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1886         }
1887
1888         /**
1889          * Return true if subpage links should be expanded on this page.
1890          * @return bool
1891          */
1892         function areSubpagesAllowed() {
1893                 # Some namespaces don't allow subpages
1894                 global $wgNamespacesWithSubpages;
1895                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1896         }
1897
1898         /**
1899          * Handle link to subpage if necessary
1900          * @param string $target the source of the link
1901          * @param string &$text the link text, modified as necessary
1902          * @return string the full name of the link
1903          * @private
1904          */
1905         function maybeDoSubpageLink($target, &$text) {
1906                 # Valid link forms:
1907                 # Foobar -- normal
1908                 # :Foobar -- override special treatment of prefix (images, language links)
1909                 # /Foobar -- convert to CurrentPage/Foobar
1910                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1911                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1912                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1913
1914                 $fname = 'Parser::maybeDoSubpageLink';
1915                 wfProfileIn( $fname );
1916                 $ret = $target; # default return value is no change
1917
1918                 # bug 7425
1919                 $target = trim( $target );
1920
1921                 # Some namespaces don't allow subpages,
1922                 # so only perform processing if subpages are allowed
1923                 if( $this->areSubpagesAllowed() ) {
1924                         # Look at the first character
1925                         if( $target != '' && $target{0} == '/' ) {
1926                                 # / at end means we don't want the slash to be shown
1927                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1928                                 if( $trailingSlashes ) {
1929                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1930                                 } else {
1931                                         $noslash = substr( $target, 1 );
1932                                 }
1933
1934                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1935                                 if( '' === $text ) {
1936                                         $text = $target;
1937                                 } # this might be changed for ugliness reasons
1938                         } else {
1939                                 # check for .. subpage backlinks
1940                                 $dotdotcount = 0;
1941                                 $nodotdot = $target;
1942                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1943                                         ++$dotdotcount;
1944                                         $nodotdot = substr( $nodotdot, 3 );
1945                                 }
1946                                 if($dotdotcount > 0) {
1947                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1948                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1949                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1950                                                 # / at the end means don't show full path
1951                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1952                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1953                                                         if( '' === $text ) {
1954                                                                 $text = $nodotdot;
1955                                                         }
1956                                                 }
1957                                                 $nodotdot = trim( $nodotdot );
1958                                                 if( $nodotdot != '' ) {
1959                                                         $ret .= '/' . $nodotdot;
1960                                                 }
1961                                         }
1962                                 }
1963                         }
1964                 }
1965
1966                 wfProfileOut( $fname );
1967                 return $ret;
1968         }
1969
1970         /**#@+
1971          * Used by doBlockLevels()
1972          * @private
1973          */
1974         /* private */ function closeParagraph() {
1975                 $result = '';
1976                 if ( '' != $this->mLastSection ) {
1977                         $result = '</' . $this->mLastSection  . ">\n";
1978                 }
1979                 $this->mInPre = false;
1980                 $this->mLastSection = '';
1981                 return $result;
1982         }
1983         # getCommon() returns the length of the longest common substring
1984         # of both arguments, starting at the beginning of both.
1985         #
1986         /* private */ function getCommon( $st1, $st2 ) {
1987                 $fl = strlen( $st1 );
1988                 $shorter = strlen( $st2 );
1989                 if ( $fl < $shorter ) { $shorter = $fl; }
1990
1991                 for ( $i = 0; $i < $shorter; ++$i ) {
1992                         if ( $st1{$i} != $st2{$i} ) { break; }
1993                 }
1994                 return $i;
1995         }
1996         # These next three functions open, continue, and close the list
1997         # element appropriate to the prefix character passed into them.
1998         #
1999         /* private */ function openList( $char ) {
2000                 $result = $this->closeParagraph();
2001
2002                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2003                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2004                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2005                 else if ( ';' == $char ) {
2006                         $result .= '<dl><dt>';
2007                         $this->mDTopen = true;
2008                 }
2009                 else { $result = '<!-- ERR 1 -->'; }
2010
2011                 return $result;
2012         }
2013
2014         /* private */ function nextItem( $char ) {
2015                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2016                 else if ( ':' == $char || ';' == $char ) {
2017                         $close = '</dd>';
2018                         if ( $this->mDTopen ) { $close = '</dt>'; }
2019                         if ( ';' == $char ) {
2020                                 $this->mDTopen = true;
2021                                 return $close . '<dt>';
2022                         } else {
2023                                 $this->mDTopen = false;
2024                                 return $close . '<dd>';
2025                         }
2026                 }
2027                 return '<!-- ERR 2 -->';
2028         }
2029
2030         /* private */ function closeList( $char ) {
2031                 if ( '*' == $char ) { $text = '</li></ul>'; }
2032                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2033                 else if ( ':' == $char ) {
2034                         if ( $this->mDTopen ) {
2035                                 $this->mDTopen = false;
2036                                 $text = '</dt></dl>';
2037                         } else {
2038                                 $text = '</dd></dl>';
2039                         }
2040                 }
2041                 else {  return '<!-- ERR 3 -->'; }
2042                 return $text."\n";
2043         }
2044         /**#@-*/
2045
2046         /**
2047          * Make lists from lines starting with ':', '*', '#', etc.
2048          *
2049          * @private
2050          * @return string the lists rendered as HTML
2051          */
2052         function doBlockLevels( $text, $linestart ) {
2053                 $fname = 'Parser::doBlockLevels';
2054                 wfProfileIn( $fname );
2055
2056                 # Parsing through the text line by line.  The main thing
2057                 # happening here is handling of block-level elements p, pre,
2058                 # and making lists from lines starting with * # : etc.
2059                 #
2060                 $textLines = explode( "\n", $text );
2061
2062                 $lastPrefix = $output = '';
2063                 $this->mDTopen = $inBlockElem = false;
2064                 $prefixLength = 0;
2065                 $paragraphStack = false;
2066
2067                 if ( !$linestart ) {
2068                         $output .= array_shift( $textLines );
2069                 }
2070                 foreach ( $textLines as $oLine ) {
2071                         $lastPrefixLength = strlen( $lastPrefix );
2072                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2073                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2074                         if ( !$this->mInPre ) {
2075                                 # Multiple prefixes may abut each other for nested lists.
2076                                 $prefixLength = strspn( $oLine, '*#:;' );
2077                                 $pref = substr( $oLine, 0, $prefixLength );
2078
2079                                 # eh?
2080                                 $pref2 = str_replace( ';', ':', $pref );
2081                                 $t = substr( $oLine, $prefixLength );
2082                                 $this->mInPre = !empty($preOpenMatch);
2083                         } else {
2084                                 # Don't interpret any other prefixes in preformatted text
2085                                 $prefixLength = 0;
2086                                 $pref = $pref2 = '';
2087                                 $t = $oLine;
2088                         }
2089
2090                         # List generation
2091                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2092                                 # Same as the last item, so no need to deal with nesting or opening stuff
2093                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2094                                 $paragraphStack = false;
2095
2096                                 if ( substr( $pref, -1 ) == ';') {
2097                                         # The one nasty exception: definition lists work like this:
2098                                         # ; title : definition text
2099                                         # So we check for : in the remainder text to split up the
2100                                         # title and definition, without b0rking links.
2101                                         $term = $t2 = '';
2102                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2103                                                 $t = $t2;
2104                                                 $output .= $term . $this->nextItem( ':' );
2105                                         }
2106                                 }
2107                         } elseif( $prefixLength || $lastPrefixLength ) {
2108                                 # Either open or close a level...
2109                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2110                                 $paragraphStack = false;
2111
2112                                 while( $commonPrefixLength < $lastPrefixLength ) {
2113                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2114                                         --$lastPrefixLength;
2115                                 }
2116                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2117                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2118                                 }
2119                                 while ( $prefixLength > $commonPrefixLength ) {
2120                                         $char = substr( $pref, $commonPrefixLength, 1 );
2121                                         $output .= $this->openList( $char );
2122
2123                                         if ( ';' == $char ) {
2124                                                 # FIXME: This is dupe of code above
2125                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2126                                                         $t = $t2;
2127                                                         $output .= $term . $this->nextItem( ':' );
2128                                                 }
2129                                         }
2130                                         ++$commonPrefixLength;
2131                                 }
2132                                 $lastPrefix = $pref2;
2133                         }
2134                         if( 0 == $prefixLength ) {
2135                                 wfProfileIn( "$fname-paragraph" );
2136                                 # No prefix (not in list)--go to paragraph mode
2137                                 // XXX: use a stack for nestable elements like span, table and div
2138                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2139                                 $closematch = preg_match(
2140                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2141                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2142                                 if ( $openmatch or $closematch ) {
2143                                         $paragraphStack = false;
2144                                         # TODO bug 5718: paragraph closed
2145                                         $output .= $this->closeParagraph();
2146                                         if ( $preOpenMatch and !$preCloseMatch ) {
2147                                                 $this->mInPre = true;
2148                                         }
2149                                         if ( $closematch ) {
2150                                                 $inBlockElem = false;
2151                                         } else {
2152                                                 $inBlockElem = true;
2153                                         }
2154                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2155                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2156                                                 // pre
2157                                                 if ($this->mLastSection != 'pre') {
2158                                                         $paragraphStack = false;
2159                                                         $output .= $this->closeParagraph().'<pre>';
2160                                                         $this->mLastSection = 'pre';
2161                                                 }
2162                                                 $t = substr( $t, 1 );
2163                                         } else {
2164                                                 // paragraph
2165                                                 if ( '' == trim($t) ) {
2166                                                         if ( $paragraphStack ) {
2167                                                                 $output .= $paragraphStack.'<br />';
2168                                                                 $paragraphStack = false;
2169                                                                 $this->mLastSection = 'p';
2170                                                         } else {
2171                                                                 if ($this->mLastSection != 'p' ) {
2172                                                                         $output .= $this->closeParagraph();
2173                                                                         $this->mLastSection = '';
2174                                                                         $paragraphStack = '<p>';
2175                                                                 } else {
2176                                                                         $paragraphStack = '</p><p>';
2177                                                                 }
2178                                                         }
2179                                                 } else {
2180                                                         if ( $paragraphStack ) {
2181                                                                 $output .= $paragraphStack;
2182                                                                 $paragraphStack = false;
2183                                                                 $this->mLastSection = 'p';
2184                                                         } else if ($this->mLastSection != 'p') {
2185                                                                 $output .= $this->closeParagraph().'<p>';
2186                                                                 $this->mLastSection = 'p';
2187                                                         }
2188                                                 }
2189                                         }
2190                                 }
2191                                 wfProfileOut( "$fname-paragraph" );
2192                         }
2193                         // somewhere above we forget to get out of pre block (bug 785)
2194                         if($preCloseMatch && $this->mInPre) {
2195                                 $this->mInPre = false;
2196                         }
2197                         if ($paragraphStack === false) {
2198                                 $output .= $t."\n";
2199                         }
2200                 }
2201                 while ( $prefixLength ) {
2202                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2203                         --$prefixLength;
2204                 }
2205                 if ( '' != $this->mLastSection ) {
2206                         $output .= '</' . $this->mLastSection . '>';
2207                         $this->mLastSection = '';
2208                 }
2209
2210                 wfProfileOut( $fname );
2211                 return $output;
2212         }
2213
2214         /**
2215          * Split up a string on ':', ignoring any occurences inside tags
2216          * to prevent illegal overlapping.
2217          * @param string $str the string to split
2218          * @param string &$before set to everything before the ':'
2219          * @param string &$after set to everything after the ':'
2220          * return string the position of the ':', or false if none found
2221          */
2222         function findColonNoLinks($str, &$before, &$after) {
2223                 $fname = 'Parser::findColonNoLinks';
2224                 wfProfileIn( $fname );
2225
2226                 $pos = strpos( $str, ':' );
2227                 if( $pos === false ) {
2228                         // Nothing to find!
2229                         wfProfileOut( $fname );
2230                         return false;
2231                 }
2232
2233                 $lt = strpos( $str, '<' );
2234                 if( $lt === false || $lt > $pos ) {
2235                         // Easy; no tag nesting to worry about
2236                         $before = substr( $str, 0, $pos );
2237                         $after = substr( $str, $pos+1 );
2238                         wfProfileOut( $fname );
2239                         return $pos;
2240                 }
2241
2242                 // Ugly state machine to walk through avoiding tags.
2243                 $state = MW_COLON_STATE_TEXT;
2244                 $stack = 0;
2245                 $len = strlen( $str );
2246                 for( $i = 0; $i < $len; $i++ ) {
2247                         $c = $str{$i};
2248
2249                         switch( $state ) {
2250                         // (Using the number is a performance hack for common cases)
2251                         case 0: // MW_COLON_STATE_TEXT:
2252                                 switch( $c ) {
2253                                 case "<":
2254                                         // Could be either a <start> tag or an </end> tag
2255                                         $state = MW_COLON_STATE_TAGSTART;
2256                                         break;
2257                                 case ":":
2258                                         if( $stack == 0 ) {
2259                                                 // We found it!
2260                                                 $before = substr( $str, 0, $i );
2261                                                 $after = substr( $str, $i + 1 );
2262                                                 wfProfileOut( $fname );
2263                                                 return $i;
2264                                         }
2265                                         // Embedded in a tag; don't break it.
2266                                         break;
2267                                 default:
2268                                         // Skip ahead looking for something interesting
2269                                         $colon = strpos( $str, ':', $i );
2270                                         if( $colon === false ) {
2271                                                 // Nothing else interesting
2272                                                 wfProfileOut( $fname );
2273                                                 return false;
2274                                         }
2275                                         $lt = strpos( $str, '<', $i );
2276                                         if( $stack === 0 ) {
2277                                                 if( $lt === false || $colon < $lt ) {
2278                                                         // We found it!
2279                                                         $before = substr( $str, 0, $colon );
2280                                                         $after = substr( $str, $colon + 1 );
2281                                                         wfProfileOut( $fname );
2282                                                         return $i;
2283                                                 }
2284                                         }
2285                                         if( $lt === false ) {
2286                                                 // Nothing else interesting to find; abort!
2287                                                 // We're nested, but there's no close tags left. Abort!
2288                                                 break 2;
2289                                         }
2290                                         // Skip ahead to next tag start
2291                                         $i = $lt;
2292                                         $state = MW_COLON_STATE_TAGSTART;
2293                                 }
2294                                 break;
2295                         case 1: // MW_COLON_STATE_TAG:
2296                                 // In a <tag>
2297                                 switch( $c ) {
2298                                 case ">":
2299                                         $stack++;
2300                                         $state = MW_COLON_STATE_TEXT;
2301                                         break;
2302                                 case "/":
2303                                         // Slash may be followed by >?
2304                                         $state = MW_COLON_STATE_TAGSLASH;
2305                                         break;
2306                                 default:
2307                                         // ignore
2308                                 }
2309                                 break;
2310                         case 2: // MW_COLON_STATE_TAGSTART:
2311                                 switch( $c ) {
2312                                 case "/":
2313                                         $state = MW_COLON_STATE_CLOSETAG;
2314                                         break;
2315                                 case "!":
2316                                         $state = MW_COLON_STATE_COMMENT;
2317                                         break;
2318                                 case ">":
2319                                         // Illegal early close? This shouldn't happen D:
2320                                         $state = MW_COLON_STATE_TEXT;
2321                                         break;
2322                                 default:
2323                                         $state = MW_COLON_STATE_TAG;
2324                                 }
2325                                 break;
2326                         case 3: // MW_COLON_STATE_CLOSETAG:
2327                                 // In a </tag>
2328                                 if( $c == ">" ) {
2329                                         $stack--;
2330                                         if( $stack < 0 ) {
2331                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2332                                                 wfProfileOut( $fname );
2333                                                 return false;
2334                                         }
2335                                         $state = MW_COLON_STATE_TEXT;
2336                                 }
2337                                 break;
2338                         case MW_COLON_STATE_TAGSLASH:
2339                                 if( $c == ">" ) {
2340                                         // Yes, a self-closed tag <blah/>
2341                                         $state = MW_COLON_STATE_TEXT;
2342                                 } else {
2343                                         // Probably we're jumping the gun, and this is an attribute
2344                                         $state = MW_COLON_STATE_TAG;
2345                                 }
2346                                 break;
2347                         case 5: // MW_COLON_STATE_COMMENT:
2348                                 if( $c == "-" ) {
2349                                         $state = MW_COLON_STATE_COMMENTDASH;
2350                                 }
2351                                 break;
2352                         case MW_COLON_STATE_COMMENTDASH:
2353                                 if( $c == "-" ) {
2354                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2355                                 } else {
2356                                         $state = MW_COLON_STATE_COMMENT;
2357                                 }
2358                                 break;
2359                         case MW_COLON_STATE_COMMENTDASHDASH:
2360                                 if( $c == ">" ) {
2361                                         $state = MW_COLON_STATE_TEXT;
2362                                 } else {
2363                                         $state = MW_COLON_STATE_COMMENT;
2364                                 }
2365                                 break;
2366                         default:
2367                                 throw new MWException( "State machine error in $fname" );
2368                         }
2369                 }
2370                 if( $stack > 0 ) {
2371                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2372                         return false;
2373                 }
2374                 wfProfileOut( $fname );
2375                 return false;
2376         }
2377
2378         /**
2379          * Return value of a magic variable (like PAGENAME)
2380          *
2381          * @private
2382          */
2383         function getVariableValue( $index ) {
2384                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2385
2386                 /**
2387                  * Some of these require message or data lookups and can be
2388                  * expensive to check many times.
2389                  */
2390                 static $varCache = array();
2391                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2392                         if ( isset( $varCache[$index] ) ) {
2393                                 return $varCache[$index];
2394                         }
2395                 }
2396
2397                 $ts = time();
2398                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2399
2400                 # Use the time zone
2401                 global $wgLocaltimezone;
2402                 if ( isset( $wgLocaltimezone ) ) {
2403                         $oldtz = getenv( 'TZ' );
2404                         putenv( 'TZ='.$wgLocaltimezone );
2405                 }
2406                 $localTimestamp = date( 'YmdHis', $ts );
2407                 $localMonth = date( 'm', $ts );
2408                 $localMonthName = date( 'n', $ts );
2409                 $localDay = date( 'j', $ts );
2410                 $localDay2 = date( 'd', $ts );
2411                 $localDayOfWeek = date( 'w', $ts );
2412                 $localWeek = date( 'W', $ts );
2413                 $localYear = date( 'Y', $ts );
2414                 $localHour = date( 'H', $ts );
2415                 if ( isset( $wgLocaltimezone ) ) {
2416                         putenv( 'TZ='.$oldtz );
2417                 }
2418
2419                 switch ( $index ) {
2420                         case 'currentmonth':
2421                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2422                         case 'currentmonthname':
2423                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2424                         case 'currentmonthnamegen':
2425                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2426                         case 'currentmonthabbrev':
2427                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2428                         case 'currentday':
2429                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2430                         case 'currentday2':
2431                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2432                         case 'localmonth':
2433                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2434                         case 'localmonthname':
2435                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2436                         case 'localmonthnamegen':
2437                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2438                         case 'localmonthabbrev':
2439                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2440                         case 'localday':
2441                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2442                         case 'localday2':
2443                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2444                         case 'pagename':
2445                                 return $this->mTitle->getText();
2446                         case 'pagenamee':
2447                                 return $this->mTitle->getPartialURL();
2448                         case 'fullpagename':
2449                                 return $this->mTitle->getPrefixedText();
2450                         case 'fullpagenamee':
2451                                 return $this->mTitle->getPrefixedURL();
2452                         case 'subpagename':
2453                                 return $this->mTitle->getSubpageText();
2454                         case 'subpagenamee':
2455                                 return $this->mTitle->getSubpageUrlForm();
2456                         case 'basepagename':
2457                                 return $this->mTitle->getBaseText();
2458                         case 'basepagenamee':
2459                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2460                         case 'talkpagename':
2461                                 if( $this->mTitle->canTalk() ) {
2462                                         $talkPage = $this->mTitle->getTalkPage();
2463                                         return $talkPage->getPrefixedText();
2464                                 } else {
2465                                         return '';
2466                                 }
2467                         case 'talkpagenamee':
2468                                 if( $this->mTitle->canTalk() ) {
2469                                         $talkPage = $this->mTitle->getTalkPage();
2470                                         return $talkPage->getPrefixedUrl();
2471                                 } else {
2472                                         return '';
2473                                 }
2474                         case 'subjectpagename':
2475                                 $subjPage = $this->mTitle->getSubjectPage();
2476                                 return $subjPage->getPrefixedText();
2477                         case 'subjectpagenamee':
2478                                 $subjPage = $this->mTitle->getSubjectPage();
2479                                 return $subjPage->getPrefixedUrl();
2480                         case 'revisionid':
2481                                 return $this->mRevisionId;
2482                         case 'revisionday':
2483                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2484                         case 'revisionday2':
2485                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2486                         case 'revisionmonth':
2487                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2488                         case 'revisionyear':
2489                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2490                         case 'revisiontimestamp':
2491                                 return $this->getRevisionTimestamp();
2492                         case 'namespace':
2493                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2494                         case 'namespacee':
2495                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2496                         case 'talkspace':
2497                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2498                         case 'talkspacee':
2499                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2500                         case 'subjectspace':
2501                                 return $this->mTitle->getSubjectNsText();
2502                         case 'subjectspacee':
2503                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2504                         case 'currentdayname':
2505                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2506                         case 'currentyear':
2507                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2508                         case 'currenttime':
2509                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2510                         case 'currenthour':
2511                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2512                         case 'currentweek':
2513                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2514                                 // int to remove the padding
2515                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2516                         case 'currentdow':
2517                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2518                         case 'localdayname':
2519                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2520                         case 'localyear':
2521                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2522                         case 'localtime':
2523                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2524                         case 'localhour':
2525                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2526                         case 'localweek':
2527                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2528                                 // int to remove the padding
2529                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2530                         case 'localdow':
2531                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2532                         case 'numberofarticles':
2533                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2534                         case 'numberoffiles':
2535                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2536                         case 'numberofusers':
2537                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2538                         case 'numberofpages':
2539                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2540                         case 'numberofadmins':
2541                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2542                         case 'currenttimestamp':
2543                                 return $varCache[$index] = wfTimestampNow();
2544                         case 'localtimestamp':
2545                                 return $varCache[$index] = $localTimestamp;
2546                         case 'currentversion':
2547                                 return $varCache[$index] = SpecialVersion::getVersion();
2548                         case 'sitename':
2549                                 return $wgSitename;
2550                         case 'server':
2551                                 return $wgServer;
2552                         case 'servername':
2553                                 return $wgServerName;
2554                         case 'scriptpath':
2555                                 return $wgScriptPath;
2556                         case 'directionmark':
2557                                 return $wgContLang->getDirMark();
2558                         case 'contentlanguage':
2559                                 global $wgContLanguageCode;
2560                                 return $wgContLanguageCode;
2561                         default:
2562                                 $ret = null;
2563                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2564                                         return $ret;
2565                                 else
2566                                         return null;
2567                 }
2568         }
2569
2570         /**
2571          * initialise the magic variables (like CURRENTMONTHNAME)
2572          *
2573          * @private
2574          */
2575         function initialiseVariables() {
2576                 $fname = 'Parser::initialiseVariables';
2577                 wfProfileIn( $fname );
2578                 $variableIDs = MagicWord::getVariableIDs();
2579
2580                 $this->mVariables = array();
2581                 foreach ( $variableIDs as $id ) {
2582                         $mw =& MagicWord::get( $id );
2583                         $mw->addToArray( $this->mVariables, $id );
2584                 }
2585                 wfProfileOut( $fname );
2586         }
2587
2588         /**
2589          * parse any parentheses in format ((title|part|part))
2590          * and call callbacks to get a replacement text for any found piece
2591          *
2592          * @param string $text The text to parse
2593          * @param array $callbacks rules in form:
2594          *     '{' => array(                            # opening parentheses
2595          *                                      'end' => '}',   # closing parentheses
2596          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2597          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2598          *                                                                )
2599          *                                      )
2600          *                                      'min' => 2,     # Minimum parenthesis count in cb
2601          *                                      'max' => 3,     # Maximum parenthesis count in cb
2602          * @private
2603          */
2604         function replace_callback ($text, $callbacks) {
2605                 wfProfileIn( __METHOD__ );
2606                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2607                 $lastOpeningBrace = -1;                 # last not closed parentheses
2608
2609                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2610
2611                 $i = 0;
2612                 while ( $i < strlen( $text ) ) {
2613                         # Find next opening brace, closing brace or pipe
2614                         if ( $lastOpeningBrace == -1 ) {
2615                                 $currentClosing = '';
2616                                 $search = $validOpeningBraces;
2617                         } else {
2618                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2619                                 $search = $validOpeningBraces . '|' . $currentClosing;
2620                         }
2621                         $rule = null;
2622                         $i += strcspn( $text, $search, $i );
2623                         if ( $i < strlen( $text ) ) {
2624                                 if ( $text[$i] == '|' ) {
2625                                         $found = 'pipe';
2626                                 } elseif ( $text[$i] == $currentClosing ) {
2627                                         $found = 'close';
2628                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2629                                         $found = 'open';
2630                                         $rule = $callbacks[$text[$i]];
2631                                 } else {
2632                                         # Some versions of PHP have a strcspn which stops on null characters
2633                                         # Ignore and continue
2634                                         ++$i;
2635                                         continue;
2636                                 }
2637                         } else {
2638                                 # All done
2639                                 break;
2640                         }
2641
2642                         if ( $found == 'open' ) {
2643                                 # found opening brace, let's add it to parentheses stack
2644                                 $piece = array('brace' => $text[$i],
2645                                                            'braceEnd' => $rule['end'],
2646                                                            'title' => '',
2647                                                            'parts' => null);
2648
2649                                 # count opening brace characters
2650                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2651                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2652                                 $i += $piece['count'];
2653
2654                                 # we need to add to stack only if opening brace count is enough for one of the rules
2655                                 if ( $piece['count'] >= $rule['min'] ) {
2656                                         $lastOpeningBrace ++;
2657                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2658                                 }
2659                         } elseif ( $found == 'close' ) {
2660                                 # lets check if it is enough characters for closing brace
2661                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2662                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2663
2664                                 # check for maximum matching characters (if there are 5 closing
2665                                 # characters, we will probably need only 3 - depending on the rules)
2666                                 $matchingCount = 0;
2667                                 $matchingCallback = null;
2668                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2669                                 if ( $count > $cbType['max'] ) {
2670                                         # The specified maximum exists in the callback array, unless the caller
2671                                         # has made an error
2672                                         $matchingCount = $cbType['max'];
2673                                 } else {
2674                                         # Count is less than the maximum
2675                                         # Skip any gaps in the callback array to find the true largest match
2676                                         # Need to use array_key_exists not isset because the callback can be null
2677                                         $matchingCount = $count;
2678                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2679                                                 --$matchingCount;
2680                                         }
2681                                 }
2682
2683                                 if ($matchingCount <= 0) {
2684                                         $i += $count;
2685                                         continue;
2686                                 }
2687                                 $matchingCallback = $cbType['cb'][$matchingCount];
2688
2689                                 # let's set a title or last part (if '|' was found)
2690                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2691                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2692                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2693                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2694                                 } else {
2695                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2696                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2697                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2698                                 }
2699
2700                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2701                                 $pieceEnd = $i + $matchingCount;
2702
2703                                 if( is_callable( $matchingCallback ) ) {
2704                                         $cbArgs = array (
2705                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2706                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2707                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2708                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2709                                                                          );
2710                                         # finally we can call a user callback and replace piece of text
2711                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2712                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2713                                         $i = $pieceStart + strlen($replaceWith);
2714                                 } else {
2715                                         # null value for callback means that parentheses should be parsed, but not replaced
2716                                         $i += $matchingCount;
2717                                 }
2718
2719                                 # reset last opening parentheses, but keep it in case there are unused characters
2720                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2721                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2722                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2723                                                            'title' => '',
2724                                                            'parts' => null,
2725                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2726                                 $openingBraceStack[$lastOpeningBrace--] = null;
2727
2728                                 if ($matchingCount < $piece['count']) {
2729                                         $piece['count'] -= $matchingCount;
2730                                         $piece['startAt'] -= $matchingCount;
2731                                         $piece['partStart'] = $piece['startAt'];
2732                                         # do we still qualify for any callback with remaining count?
2733                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2734                                         while ( $piece['count'] ) {
2735                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2736                                                         $lastOpeningBrace++;
2737                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2738                                                         break;
2739                                                 }
2740                                                 --$piece['count'];
2741                                         }
2742                                 }
2743                         } elseif ( $found == 'pipe' ) {
2744                                 # lets set a title if it is a first separator, or next part otherwise
2745                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2746                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2747                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2748                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2749                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2750                                 } else {
2751                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2752                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2753                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2754                                 }
2755                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2756                         }
2757                 }
2758
2759                 wfProfileOut( __METHOD__ );
2760                 return $text;
2761         }
2762
2763         /**
2764          * Replace magic variables, templates, and template arguments
2765          * with the appropriate text. Templates are substituted recursively,
2766          * taking care to avoid infinite loops.
2767          *
2768          * Note that the substitution depends on value of $mOutputType:
2769          *  OT_WIKI: only {{subst:}} templates
2770          *  OT_MSG: only magic variables
2771          *  OT_HTML: all templates and magic variables
2772          *
2773          * @param string $tex The text to transform
2774          * @param array $args Key-value pairs representing template parameters to substitute
2775          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2776          * @private
2777          */
2778         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2779                 # Prevent too big inclusions
2780                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2781                         return $text;
2782                 }
2783
2784                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2785                 wfProfileIn( $fname );
2786
2787                 # This function is called recursively. To keep track of arguments we need a stack:
2788                 array_push( $this->mArgStack, $args );
2789
2790                 $braceCallbacks = array();
2791                 if ( !$argsOnly ) {
2792                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2793                 }
2794                 if ( $this->mOutputType != OT_MSG ) {
2795                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2796                 }
2797                 if ( $braceCallbacks ) {
2798                         $callbacks = array(
2799                                 '{' => array(
2800                                         'end' => '}',
2801                                         'cb' => $braceCallbacks,
2802                                         'min' => $argsOnly ? 3 : 2,
2803                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2804                                 ),
2805                                 '[' => array(
2806                                         'end' => ']',
2807                                         'cb' => array(2=>null),
2808                                         'min' => 2,
2809                                         'max' => 2,
2810                                 )
2811                         );
2812                         $text = $this->replace_callback ($text, $callbacks);
2813
2814                         array_pop( $this->mArgStack );
2815                 }
2816                 wfProfileOut( $fname );
2817                 return $text;
2818         }
2819
2820         /**
2821          * Replace magic variables
2822          * @private
2823          */
2824         function variableSubstitution( $matches ) {
2825                 global $wgContLang;
2826                 $fname = 'Parser::variableSubstitution';
2827                 $varname = $wgContLang->lc($matches[1]);
2828                 wfProfileIn( $fname );
2829                 $skip = false;
2830                 if ( $this->mOutputType == OT_WIKI ) {
2831                         # Do only magic variables prefixed by SUBST
2832                         $mwSubst =& MagicWord::get( 'subst' );
2833                         if (!$mwSubst->matchStartAndRemove( $varname ))
2834                                 $skip = true;
2835                         # Note that if we don't substitute the variable below,
2836                         # we don't remove the {{subst:}} magic word, in case
2837                         # it is a template rather than a magic variable.
2838                 }
2839                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2840                         $id = $this->mVariables[$varname];
2841                         # Now check if we did really match, case sensitive or not
2842                         $mw =& MagicWord::get( $id );
2843                         if ($mw->match($matches[1])) {
2844                                 $text = $this->getVariableValue( $id );
2845                                 $this->mOutput->mContainsOldMagic = true;
2846                         } else {
2847                                 $text = $matches[0];
2848                         }
2849                 } else {
2850                         $text = $matches[0];
2851                 }
2852                 wfProfileOut( $fname );
2853                 return $text;
2854         }
2855
2856
2857         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2858         static function createAssocArgs( $args ) {
2859                 $assocArgs = array();
2860                 $index = 1;
2861                 foreach( $args as $arg ) {
2862                         $eqpos = strpos( $arg, '=' );
2863                         if ( $eqpos === false ) {
2864                                 $assocArgs[$index++] = $arg;
2865                         } else {
2866                                 $name = trim( substr( $arg, 0, $eqpos ) );
2867                                 $value = trim( substr( $arg, $eqpos+1 ) );
2868                                 if ( $value === false ) {
2869                                         $value = '';
2870                                 }
2871                                 if ( $name !== false ) {
2872                                         $assocArgs[$name] = $value;
2873                                 }
2874                         }
2875                 }
2876
2877                 return $assocArgs;
2878         }
2879
2880         /**
2881          * Return the text of a template, after recursively
2882          * replacing any variables or templates within the template.
2883          *
2884          * @param array $piece The parts of the template
2885          *  $piece['text']: matched text
2886          *  $piece['title']: the title, i.e. the part before the |
2887          *  $piece['parts']: the parameter array
2888          * @return string the text of the template
2889          * @private
2890          */
2891         function braceSubstitution( $piece ) {
2892                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2893                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2894                 wfProfileIn( $fname );
2895                 wfProfileIn( __METHOD__.'-setup' );
2896
2897                 # Flags
2898                 $found = false;             # $text has been filled
2899                 $nowiki = false;            # wiki markup in $text should be escaped
2900                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2901                 $noargs = false;            # Don't replace triple-brace arguments in $text
2902                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2903                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2904                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2905                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2906
2907                 # Title object, where $text came from
2908                 $title = NULL;
2909
2910                 $linestart = '';
2911
2912
2913                 # $part1 is the bit before the first |, and must contain only title characters
2914                 # $args is a list of arguments, starting from index 0, not including $part1
2915
2916                 $titleText = $part1 = $piece['title'];
2917                 # If the third subpattern matched anything, it will start with |
2918
2919                 if (null == $piece['parts']) {
2920                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2921                         if ($replaceWith != $piece['text']) {
2922                                 $text = $replaceWith;
2923                                 $found = true;
2924                                 $noparse = true;
2925                                 $noargs = true;
2926                         }
2927                 }
2928
2929                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2930                 wfProfileOut( __METHOD__.'-setup' );
2931
2932                 # SUBST
2933                 wfProfileIn( __METHOD__.'-modifiers' );
2934                 if ( !$found ) {
2935                         $mwSubst =& MagicWord::get( 'subst' );
2936                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2937                                 # One of two possibilities is true:
2938                                 # 1) Found SUBST but not in the PST phase
2939                                 # 2) Didn't find SUBST and in the PST phase
2940                                 # In either case, return without further processing
2941                                 $text = $piece['text'];
2942                                 $found = true;
2943                                 $noparse = true;
2944                                 $noargs = true;
2945                         }
2946                 }
2947
2948                 # MSG, MSGNW and RAW
2949                 if ( !$found ) {
2950                         # Check for MSGNW:
2951                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2952                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2953                                 $nowiki = true;
2954                         } else {
2955                                 # Remove obsolete MSG:
2956                                 $mwMsg =& MagicWord::get( 'msg' );
2957                                 $mwMsg->matchStartAndRemove( $part1 );
2958                         }
2959
2960                         # Check for RAW:
2961                         $mwRaw =& MagicWord::get( 'raw' );
2962                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2963                                 $forceRawInterwiki = true;
2964                         }
2965                 }
2966                 wfProfileOut( __METHOD__.'-modifiers' );
2967
2968                 //save path level before recursing into functions & templates.
2969                 $lastPathLevel = $this->mTemplatePath;
2970
2971                 # Parser functions
2972                 if ( !$found ) {
2973                         wfProfileIn( __METHOD__ . '-pfunc' );
2974
2975                         $colonPos = strpos( $part1, ':' );
2976                         if ( $colonPos !== false ) {
2977                                 # Case sensitive functions
2978                                 $function = substr( $part1, 0, $colonPos );
2979                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2980                                         $function = $this->mFunctionSynonyms[1][$function];
2981                                 } else {
2982                                         # Case insensitive functions
2983                                         $function = strtolower( $function );
2984                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2985                                                 $function = $this->mFunctionSynonyms[0][$function];
2986                                         } else {
2987                                                 $function = false;
2988                                         }
2989                                 }
2990                                 if ( $function ) {
2991                                         $funcArgs = array_map( 'trim', $args );
2992                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2993                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2994                                         $found = true;
2995
2996                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2997                                         //$noargs = true;
2998                                         //$noparse = true;
2999
3000                                         if ( is_array( $result ) ) {
3001                                                 if ( isset( $result[0] ) ) {
3002                                                         $text = $linestart . $result[0];
3003                                                         unset( $result[0] );
3004                                                 }
3005
3006                                                 // Extract flags into the local scope
3007                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3008                                                 extract( $result );
3009                                         } else {
3010                                                 $text = $linestart . $result;
3011                                         }
3012                                 }
3013                         }
3014                         wfProfileOut( __METHOD__ . '-pfunc' );
3015                 }
3016
3017                 # Template table test
3018
3019                 # Did we encounter this template already? If yes, it is in the cache
3020                 # and we need to check for loops.
3021                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3022                         $found = true;
3023
3024                         # Infinite loop test
3025                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3026                                 $noparse = true;
3027                                 $noargs = true;
3028                                 $found = true;
3029                                 $text = $linestart .
3030                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3031                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3032                         } else {
3033                                 # set $text to cached message.
3034                                 $text = $linestart . $this->mTemplates[$piece['title']];
3035                                 #treat title for cached page the same as others
3036                                 $ns = NS_TEMPLATE;
3037                                 $subpage = '';
3038                                 $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3039                                 if ($subpage !== '') {
3040                                   $ns = $this->mTitle->getNamespace();
3041                                 }
3042                                 $title = Title::newFromText( $part1, $ns );
3043                                 //used by include size checking
3044                                 $titleText = $title->getPrefixedText();
3045                                 //used by edit section links
3046                                 $replaceHeadings = true;
3047
3048                         }
3049                 }
3050
3051                 # Load from database
3052                 if ( !$found ) {
3053                         wfProfileIn( __METHOD__ . '-loadtpl' );
3054                         $ns = NS_TEMPLATE;
3055                         # declaring $subpage directly in the function call
3056                         # does not work correctly with references and breaks
3057                         # {{/subpage}}-style inclusions
3058                         $subpage = '';
3059                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3060                         if ($subpage !== '') {
3061                                 $ns = $this->mTitle->getNamespace();
3062                         }
3063                         $title = Title::newFromText( $part1, $ns );
3064
3065
3066                         if ( !is_null( $title ) ) {
3067                                 $titleText = $title->getPrefixedText();
3068                                 # Check for language variants if the template is not found
3069                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3070                                         $wgContLang->findVariantLink($part1, $title);
3071                                 }
3072
3073                                 if ( !$title->isExternal() ) {
3074                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3075                                                 $text = SpecialPage::capturePath( $title );
3076                                                 if ( is_string( $text ) ) {
3077                                                         $found = true;
3078                                                         $noparse = true;
3079                                                         $noargs = true;
3080                                                         $isHTML = true;
3081                                                         $this->disableCache();
3082                                                 }
3083                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3084                                                 $found = false; //access denied
3085                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3086                                         } else {
3087                                                 $articleContent = $this->fetchTemplate( $title );
3088                                                 if ( $articleContent !== false ) {
3089                                                         $found = true;
3090                                                         $text = $articleContent;
3091                                                         $replaceHeadings = true;
3092                                                 }
3093                                         }
3094
3095                                         # If the title is valid but undisplayable, make a link to it
3096                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3097                                                 $text = "[[:$titleText]]";
3098                                                 $found = true;
3099                                         }
3100                                 } elseif ( $title->isTrans() ) {
3101                                         // Interwiki transclusion
3102                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3103                                                 $text = $this->interwikiTransclude( $title, 'render' );
3104                                                 $isHTML = true;
3105                                                 $noparse = true;
3106                                         } else {
3107                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3108                                                 $replaceHeadings = true;
3109                                         }
3110                                         $found = true;
3111                                 }
3112
3113                                 # Template cache array insertion
3114                                 # Use the original $piece['title'] not the mangled $part1, so that
3115                                 # modifiers such as RAW: produce separate cache entries
3116                                 if( $found ) {
3117                                         if( $isHTML ) {
3118                                                 // A special page; don't store it in the template cache.
3119                                         } else {
3120                                                 $this->mTemplates[$piece['title']] = $text;
3121                                         }
3122                                         $text = $linestart . $text;
3123                                 }
3124                         }
3125                         wfProfileOut( __METHOD__ . '-loadtpl' );
3126                 }
3127
3128                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3129                         # Error, oversize inclusion
3130                         $text = $linestart .
3131                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3132                         $noparse = true;
3133                         $noargs = true;
3134                 }
3135
3136                 # Recursive parsing, escaping and link table handling
3137                 # Only for HTML output
3138                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3139                         $text = wfEscapeWikiText( $text );
3140                 } elseif ( !$this->ot['msg'] && $found ) {
3141                         if ( $noargs ) {
3142                                 $assocArgs = array();
3143                         } else {
3144                                 # Clean up argument array
3145                                 $assocArgs = self::createAssocArgs($args);
3146                                 # Add a new element to the templace recursion path
3147                                 $this->mTemplatePath[$part1] = 1;
3148                         }
3149
3150                         if ( !$noparse ) {
3151                                 # If there are any <onlyinclude> tags, only include them
3152                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3153                                         $replacer = new OnlyIncludeReplacer;
3154                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3155                                                 array( &$replacer, 'replace' ), $text );
3156                                         $text = $replacer->output;
3157                                 }
3158                                 # Remove <noinclude> sections and <includeonly> tags
3159                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3160                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3161
3162                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3163                                         # Strip <nowiki>, <pre>, etc.
3164                                         $text = $this->strip( $text, $this->mStripState );
3165                                         if ( $this->ot['html'] ) {
3166                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3167                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3168                                                 $text = Sanitizer::removeHTMLcomments( $text );
3169                                         }
3170                                 }
3171                                 $text = $this->replaceVariables( $text, $assocArgs );
3172
3173                                 # If the template begins with a table or block-level
3174                                 # element, it should be treated as beginning a new line.
3175                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3176                                         $text = "\n" . $text;
3177                                 }
3178                         } elseif ( !$noargs ) {
3179                                 # $noparse and !$noargs
3180                                 # Just replace the arguments, not any double-brace items
3181                                 # This is used for rendered interwiki transclusion
3182                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3183                         }
3184                 }
3185                 # Prune lower levels off the recursion check path
3186                 $this->mTemplatePath = $lastPathLevel;
3187
3188                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3189                         # Error, oversize inclusion
3190                         $text = $linestart .
3191                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3192                         $noparse = true;
3193                         $noargs = true;
3194                 }
3195
3196                 if ( !$found ) {
3197                         wfProfileOut( $fname );
3198                         return $piece['text'];
3199                 } else {
3200                         wfProfileIn( __METHOD__ . '-placeholders' );
3201                         if ( $isHTML ) {
3202                                 # Replace raw HTML by a placeholder
3203                                 # Add a blank line preceding, to prevent it from mucking up
3204                                 # immediately preceding headings
3205                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3206                         } else {
3207                                 # replace ==section headers==
3208                                 # XXX this needs to go away once we have a better parser.
3209                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3210                                         if( !is_null( $title ) )
3211                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3212                                         else
3213                                                 $encodedname = base64_encode("");
3214                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3215                                                 PREG_SPLIT_DELIM_CAPTURE);
3216                                         $text = '';
3217                                         $nsec = $headingOffset;
3218                                         for( $i = 0; $i < count($m); $i += 2 ) {
3219                                                 $text .= $m[$i];
3220                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3221                                                 $hl = $m[$i + 1];
3222                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3223                                                         $text .= $hl;
3224                                                         continue;
3225                                                 }
3226                                                 $m2 = array();
3227                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3228                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3229                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3230
3231                                                 $nsec++;
3232                                         }
3233                                 }
3234                         }
3235                         wfProfileOut( __METHOD__ . '-placeholders' );
3236                 }
3237
3238                 # Prune lower levels off the recursion check path
3239                 $this->mTemplatePath = $lastPathLevel;
3240
3241                 if ( !$found ) {
3242                         wfProfileOut( $fname );
3243                         return $piece['text'];
3244                 } else {
3245                         wfProfileOut( $fname );
3246                         return $text;
3247                 }
3248         }
3249
3250         /**
3251          * Fetch the unparsed text of a template and register a reference to it.
3252          */
3253         function fetchTemplate( $title ) {
3254                 $text = false;
3255                 // Loop to fetch the article, with up to 1 redirect
3256                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3257                         $rev = Revision::newFromTitle( $title );
3258                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3259                         if ( $rev ) {
3260                                 $text = $rev->getText();
3261                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3262                                 global $wgLang;
3263                                 $message = $wgLang->lcfirst( $title->getText() );
3264                                 $text = wfMsgForContentNoTrans( $message );
3265                                 if( wfEmptyMsg( $message, $text ) ) {
3266                                         $text = false;
3267                                         break;
3268                                 }
3269                         } else {
3270                                 break;
3271                         }
3272                         if ( $text === false ) {
3273                                 break;
3274                         }
3275                         // Redirect?
3276                         $title = Title::newFromRedirect( $text );
3277                 }
3278                 return $text;
3279         }
3280
3281         /**
3282          * Transclude an interwiki link.
3283          */
3284         function interwikiTransclude( $title, $action ) {
3285                 global $wgEnableScaryTranscluding;
3286
3287                 if (!$wgEnableScaryTranscluding)
3288                         return wfMsg('scarytranscludedisabled');
3289
3290                 $url = $title->getFullUrl( "action=$action" );
3291
3292                 if (strlen($url) > 255)
3293                         return wfMsg('scarytranscludetoolong');
3294                 return $this->fetchScaryTemplateMaybeFromCache($url);
3295         }
3296
3297         function fetchScaryTemplateMaybeFromCache($url) {
3298                 global $wgTranscludeCacheExpiry;
3299                 $dbr = wfGetDB(DB_SLAVE);
3300                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3301                                 array('tc_url' => $url));
3302                 if ($obj) {
3303                         $time = $obj->tc_time;
3304                         $text = $obj->tc_contents;
3305                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3306                                 return $text;
3307                         }
3308                 }
3309
3310                 $text = Http::get($url);
3311                 if (!$text)
3312                         return wfMsg('scarytranscludefailed', $url);
3313
3314                 $dbw = wfGetDB(DB_MASTER);
3315                 $dbw->replace('transcache', array('tc_url'), array(
3316                         'tc_url' => $url,
3317                         'tc_time' => time(),
3318                         'tc_contents' => $text));
3319                 return $text;
3320         }
3321
3322
3323         /**
3324          * Triple brace replacement -- used for template arguments
3325          * @private
3326          */
3327         function argSubstitution( $matches ) {
3328                 $arg = trim( $matches['title'] );
3329                 $text = $matches['text'];
3330                 $inputArgs = end( $this->mArgStack );
3331
3332                 if ( array_key_exists( $arg, $inputArgs ) ) {
3333                         $text = $inputArgs[$arg];
3334                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3335                 null != $matches['parts'] && count($matches['parts']) > 0) {
3336                         $text = $matches['parts'][0];
3337                 }
3338                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3339                         $text = $matches['text'] .
3340                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3341                 }
3342
3343                 return $text;
3344         }
3345
3346         /**
3347          * Increment an include size counter
3348          *
3349          * @param string $type The type of expansion
3350          * @param integer $size The size of the text
3351          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3352          */
3353         function incrementIncludeSize( $type, $size ) {
3354                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3355                         return false;
3356                 } else {
3357                         $this->mIncludeSizes[$type] += $size;
3358                         return true;
3359                 }
3360         }
3361
3362         /**
3363          * Detect __NOGALLERY__ magic word and set a placeholder
3364          */
3365         function stripNoGallery( &$text ) {
3366                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3367                 # do not add TOC
3368                 $mw = MagicWord::get( 'nogallery' );
3369                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3370         }
3371
3372         /**
3373          * Detect __TOC__ magic word and set a placeholder
3374          */
3375         function stripToc( $text ) {
3376                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3377                 # do not add TOC
3378                 $mw = MagicWord::get( 'notoc' );
3379                 if( $mw->matchAndRemove( $text ) ) {
3380                         $this->mShowToc = false;
3381                 }
3382
3383                 $mw = MagicWord::get( 'toc' );
3384                 if( $mw->match( $text ) ) {
3385                         $this->mShowToc = true;
3386                         $this->mForceTocPosition = true;
3387
3388                         // Set a placeholder. At the end we'll fill it in with the TOC.
3389                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3390
3391                         // Only keep the first one.
3392                         $text = $mw->replace( '', $text );
3393                 }
3394                 return $text;
3395         }
3396
3397         /**
3398          * This function accomplishes several tasks:
3399          * 1) Auto-number headings if that option is enabled
3400          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3401          * 3) Add a Table of contents on the top for users who have enabled the option
3402          * 4) Auto-anchor headings
3403          *
3404          * It loops through all headlines, collects the necessary data, then splits up the
3405          * string and re-inserts the newly formatted headlines.
3406          *
3407          * @param string $text
3408          * @param boolean $isMain
3409          * @private
3410          */
3411         function formatHeadings( $text, $isMain=true ) {
3412                 global $wgMaxTocLevel, $wgContLang;
3413
3414                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3415                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3416                         $showEditLink = 0;
3417                 } else {
3418                         $showEditLink = $this->mOptions->getEditSection();
3419                 }
3420
3421                 # Inhibit editsection links if requested in the page
3422                 $esw =& MagicWord::get( 'noeditsection' );
3423                 if( $esw->matchAndRemove( $text ) ) {
3424                         $showEditLink = 0;
3425                 }
3426
3427                 # Get all headlines for numbering them and adding funky stuff like [edit]
3428                 # links - this is for later, but we need the number of headlines right now
3429                 $matches = array();
3430                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3431
3432                 # if there are fewer than 4 headlines in the article, do not show TOC
3433                 # unless it's been explicitly enabled.
3434                 $enoughToc = $this->mShowToc &&
3435                         (($numMatches >= 4) || $this->mForceTocPosition);
3436
3437                 # Allow user to stipulate that a page should have a "new section"
3438                 # link added via __NEWSECTIONLINK__
3439                 $mw =& MagicWord::get( 'newsectionlink' );
3440                 if( $mw->matchAndRemove( $text ) )
3441                         $this->mOutput->setNewSection( true );
3442
3443                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3444                 # override above conditions and always show TOC above first header
3445                 $mw =& MagicWord::get( 'forcetoc' );
3446                 if ($mw->matchAndRemove( $text ) ) {
3447                         $this->mShowToc = true;
3448                         $enoughToc = true;
3449                 }
3450
3451                 # Never ever show TOC if no headers
3452                 if( $numMatches < 1 ) {
3453                         $enoughToc = false;
3454                 }
3455
3456                 # We need this to perform operations on the HTML
3457                 $sk = $this->mOptions->getSkin();
3458
3459                 # headline counter
3460                 $headlineCount = 0;
3461                 $sectionCount = 0; # headlineCount excluding template sections
3462
3463                 # Ugh .. the TOC should have neat indentation levels which can be
3464                 # passed to the skin functions. These are determined here
3465                 $toc = '';
3466                 $full = '';
3467                 $head = array();
3468                 $sublevelCount = array();
3469                 $levelCount = array();
3470                 $toclevel = 0;
3471                 $level = 0;
3472                 $prevlevel = 0;
3473                 $toclevel = 0;
3474                 $prevtoclevel = 0;
3475
3476                 foreach( $matches[3] as $headline ) {
3477                         $istemplate = 0;
3478                         $templatetitle = '';
3479                         $templatesection = 0;
3480                         $numbering = '';
3481                         $mat = array();
3482                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3483                                 $istemplate = 1;
3484                                 $templatetitle = base64_decode($mat[1]);
3485                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3486                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3487                         }
3488
3489                         if( $toclevel ) {
3490                                 $prevlevel = $level;
3491                                 $prevtoclevel = $toclevel;
3492                         }
3493                         $level = $matches[1][$headlineCount];
3494
3495                         if( $doNumberHeadings || $enoughToc ) {
3496
3497                                 if ( $level > $prevlevel ) {
3498                                         # Increase TOC level
3499                                         $toclevel++;
3500                                         $sublevelCount[$toclevel] = 0;
3501                                         if( $toclevel<$wgMaxTocLevel ) {
3502                                                 $toc .= $sk->tocIndent();
3503                                         }
3504                                 }
3505                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3506                                         # Decrease TOC level, find level to jump to
3507
3508                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3509                                                 # Can only go down to level 1
3510                                                 $toclevel = 1;
3511                                         } else {
3512                                                 for ($i = $toclevel; $i > 0; $i--) {
3513                                                         if ( $levelCount[$i] == $level ) {
3514                                                                 # Found last matching level
3515                                                                 $toclevel = $i;
3516                                                                 break;
3517                                                         }
3518                                                         elseif ( $levelCount[$i] < $level ) {
3519                                                                 # Found first matching level below current level
3520                                                                 $toclevel = $i + 1;
3521                                                                 break;
3522                                                         }
3523                                                 }
3524                                         }
3525                                         if( $toclevel<$wgMaxTocLevel ) {
3526                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3527                                         }
3528                                 }
3529                                 else {
3530                                         # No change in level, end TOC line
3531                                         if( $toclevel<$wgMaxTocLevel ) {
3532                                                 $toc .= $sk->tocLineEnd();
3533                                         }
3534                                 }
3535
3536                                 $levelCount[$toclevel] = $level;
3537
3538                                 # count number of headlines for each level
3539                                 @$sublevelCount[$toclevel]++;
3540                                 $dot = 0;
3541                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3542                                         if( !empty( $sublevelCount[$i] ) ) {
3543                                                 if( $dot ) {
3544                                                         $numbering .= '.';
3545                                                 }
3546                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3547                                                 $dot = 1;
3548                                         }
3549                                 }
3550                         }
3551
3552                         # The canonized header is a version of the header text safe to use for links
3553                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3554                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3555
3556                         # Remove link placeholders by the link text.
3557                         #     <!--LINK number-->
3558                         # turns into
3559                         #     link text with suffix
3560                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3561                                                             "\$this->mLinkHolders['texts'][\$1]",
3562                                                             $canonized_headline );
3563                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3564                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3565                                                             $canonized_headline );
3566
3567                         # strip out HTML
3568                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3569                         $tocline = trim( $canonized_headline );
3570                         # Save headline for section edit hint before it's escaped
3571                         $headline_hint = trim( $canonized_headline );
3572                         $canonized_headline = Sanitizer::escapeId( $tocline );
3573                         $refers[$headlineCount] = $canonized_headline;
3574
3575                         # count how many in assoc. array so we can track dupes in anchors
3576                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3577                         $refcount[$headlineCount]=$refers[$canonized_headline];
3578
3579                         # Don't number the heading if it is the only one (looks silly)
3580                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3581                                 # the two are different if the line contains a link
3582                                 $headline=$numbering . ' ' . $headline;
3583                         }
3584
3585                         # Create the anchor for linking from the TOC to the section
3586                         $anchor = $canonized_headline;
3587                         if($refcount[$headlineCount] > 1 ) {
3588                                 $anchor .= '_' . $refcount[$headlineCount];
3589                         }
3590                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3591                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3592                         }
3593                         # give headline the correct <h#> tag
3594                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3595                                 if( $istemplate )
3596                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3597                                 else
3598                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3599                         } else {
3600                                 $editlink = '';
3601                         }
3602                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3603
3604                         $headlineCount++;
3605                         if( !$istemplate )
3606                                 $sectionCount++;
3607                 }
3608
3609                 if( $enoughToc ) {
3610                         if( $toclevel<$wgMaxTocLevel ) {
3611                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3612                         }
3613                         $toc = $sk->tocList( $toc );
3614                 }
3615
3616                 # split up and insert constructed headlines
3617
3618                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3619                 $i = 0;
3620
3621                 foreach( $blocks as $block ) {
3622                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3623                                 # This is the [edit] link that appears for the top block of text when
3624                                 # section editing is enabled
3625
3626                                 # Disabled because it broke block formatting
3627                                 # For example, a bullet point in the top line
3628                                 # $full .= $sk->editSectionLink(0);
3629                         }
3630                         $full .= $block;
3631                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3632                                 # Top anchor now in skin
3633                                 $full = $full.$toc;
3634                         }
3635
3636                         if( !empty( $head[$i] ) ) {
3637                                 $full .= $head[$i];
3638                         }
3639                         $i++;
3640                 }
3641                 if( $this->mForceTocPosition ) {
3642                         return str_replace( '<!--MWTOC-->', $toc, $full );
3643                 } else {
3644                         return $full;
3645                 }
3646         }
3647
3648         /**
3649          * Transform wiki markup when saving a page by doing \r\n -> \n
3650          * conversion, substitting signatures, {{subst:}} templates, etc.
3651          *
3652          * @param string $text the text to transform
3653          * @param Title &$title the Title object for the current article
3654          * @param User &$user the User object describing the current user
3655          * @param ParserOptions $options parsing options
3656          * @param bool $clearState whether to clear the parser state first
3657          * @return string the altered wiki markup
3658          * @public
3659          */
3660         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3661                 $this->mOptions = $options;
3662                 $this->mTitle =& $title;
3663                 $this->setOutputType( OT_WIKI );
3664
3665                 if ( $clearState ) {
3666                         $this->clearState();
3667                 }
3668
3669                 $stripState = new StripState;
3670                 $pairs = array(
3671                         "\r\n" => "\n",
3672                 );
3673                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3674                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3675                 $text = $this->pstPass2( $text, $stripState, $user );
3676                 $text = $stripState->unstripBoth( $text );
3677                 return $text;
3678         }
3679
3680         /**
3681          * Pre-save transform helper function
3682          * @private
3683          */
3684         function pstPass2( $text, &$stripState, $user ) {
3685                 global $wgContLang, $wgLocaltimezone;
3686
3687                 /* Note: This is the timestamp saved as hardcoded wikitext to
3688                  * the database, we use $wgContLang here in order to give
3689                  * everyone the same signature and use the default one rather
3690                  * than the one selected in each user's preferences.
3691                  */
3692                 if ( isset( $wgLocaltimezone ) ) {
3693                         $oldtz = getenv( 'TZ' );
3694                         putenv( 'TZ='.$wgLocaltimezone );
3695                 }
3696                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3697                   ' (' . date( 'T' ) . ')';
3698                 if ( isset( $wgLocaltimezone ) ) {
3699                         putenv( 'TZ='.$oldtz );
3700                 }
3701
3702                 # Variable replacement
3703                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3704                 $text = $this->replaceVariables( $text );
3705
3706                 # Strip out <nowiki> etc. added via replaceVariables
3707                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3708
3709                 # Signatures
3710                 $sigText = $this->getUserSig( $user );
3711                 $text = strtr( $text, array(
3712                         '~~~~~' => $d,
3713                         '~~~~' => "$sigText $d",
3714                         '~~~' => $sigText
3715                 ) );
3716
3717                 # Context links: [[|name]] and [[name (context)|]]
3718                 #
3719                 global $wgLegalTitleChars;
3720                 $tc = "[$wgLegalTitleChars]";
3721                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3722
3723                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3724                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3725                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3726
3727                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3728                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3729                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3730
3731                 $t = $this->mTitle->getText();
3732                 $m = array();
3733                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3734                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3735                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3736                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3737                 } else {
3738                         # if there's no context, don't bother duplicating the title
3739                         $text = preg_replace( $p2, '[[\\1]]', $text );
3740                 }
3741
3742                 # Trim trailing whitespace
3743                 $text = rtrim( $text );
3744
3745                 return $text;
3746         }
3747
3748         /**
3749          * Fetch the user's signature text, if any, and normalize to
3750          * validated, ready-to-insert wikitext.
3751          *
3752          * @param User $user
3753          * @return string
3754          * @private
3755          */
3756         function getUserSig( &$user ) {
3757                 $username = $user->getName();
3758                 $nickname = $user->getOption( 'nickname' );
3759                 $nickname = $nickname === '' ? $username : $nickname;
3760
3761                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3762                         # Sig. might contain markup; validate this
3763                         if( $this->validateSig( $nickname ) !== false ) {
3764                                 # Validated; clean up (if needed) and return it
3765                                 return $this->cleanSig( $nickname, true );
3766                         } else {
3767                                 # Failed to validate; fall back to the default
3768                                 $nickname = $username;
3769                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3770                         }
3771                 }
3772
3773                 // Make sure nickname doesnt get a sig in a sig
3774                 $nickname = $this->cleanSigInSig( $nickname );
3775
3776                 # If we're still here, make it a link to the user page
3777                 $userpage = $user->getUserPage();
3778                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3779         }
3780
3781         /**
3782          * Check that the user's signature contains no bad XML
3783          *
3784          * @param string $text
3785          * @return mixed An expanded string, or false if invalid.
3786          */
3787         function validateSig( $text ) {
3788                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3789         }
3790
3791         /**
3792          * Clean up signature text
3793          *
3794          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3795          * 2) Substitute all transclusions
3796          *
3797          * @param string $text
3798          * @param $parsing Whether we're cleaning (preferences save) or parsing
3799          * @return string Signature text
3800          */
3801         function cleanSig( $text, $parsing = false ) {
3802                 global $wgTitle;
3803                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3804
3805                 $substWord = MagicWord::get( 'subst' );
3806                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3807                 $substText = '{{' . $substWord->getSynonym( 0 );
3808
3809                 $text = preg_replace( $substRegex, $substText, $text );
3810                 $text = $this->cleanSigInSig( $text );
3811                 $text = $this->replaceVariables( $text );
3812
3813                 $this->clearState();
3814                 return $text;
3815         }
3816
3817         /**
3818          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3819          * @param string $text
3820          * @return string Signature text with /~{3,5}/ removed
3821          */
3822         function cleanSigInSig( $text ) {
3823                 $text = preg_replace( '/~{3,5}/', '', $text );
3824                 return $text;
3825         }
3826
3827         /**
3828          * Set up some variables which are usually set up in parse()
3829          * so that an external function can call some class members with confidence
3830          * @public
3831          */
3832         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3833                 $this->mTitle =& $title;
3834                 $this->mOptions = $options;
3835                 $this->setOutputType( $outputType );
3836                 if ( $clearState ) {
3837                         $this->clearState();
3838                 }
3839         }
3840
3841         /**
3842          * Transform a MediaWiki message by replacing magic variables.
3843          *
3844          * @param string $text the text to transform
3845          * @param ParserOptions $options  options
3846          * @return string the text with variables substituted
3847          * @public
3848          */
3849         function transformMsg( $text, $options ) {
3850                 global $wgTitle;
3851                 static $executing = false;
3852
3853                 $fname = "Parser::transformMsg";
3854
3855                 # Guard against infinite recursion
3856                 if ( $executing ) {
3857                         return $text;
3858                 }
3859                 $executing = true;
3860
3861                 wfProfileIn($fname);
3862
3863                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3864                         $this->mTitle = $wgTitle;
3865                 } else {
3866                         $this->mTitle = Title::newFromText('msg');
3867                 }
3868                 $this->mOptions = $options;
3869                 $this->setOutputType( OT_MSG );
3870                 $this->clearState();
3871                 $text = $this->replaceVariables( $text );
3872
3873                 $executing = false;
3874                 wfProfileOut($fname);
3875                 return $text;
3876         }
3877
3878         /**
3879          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3880          * The callback should have the following form:
3881          *    function myParserHook( $text, $params, &$parser ) { ... }
3882          *
3883          * Transform and return $text. Use $parser for any required context, e.g. use
3884          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3885          *
3886          * @public
3887          *
3888          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3889          * @param mixed $callback The callback function (and object) to use for the tag
3890          *
3891          * @return The old value of the mTagHooks array associated with the hook
3892          */
3893         function setHook( $tag, $callback ) {
3894                 $tag = strtolower( $tag );
3895                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3896                 $this->mTagHooks[$tag] = $callback;
3897
3898                 return $oldVal;
3899         }
3900
3901         /**
3902          * Create a function, e.g. {{sum:1|2|3}}
3903          * The callback function should have the form:
3904          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3905          *
3906          * The callback may either return the text result of the function, or an array with the text
3907          * in element 0, and a number of flags in the other elements. The names of the flags are
3908          * specified in the keys. Valid flags are:
3909          *   found                     The text returned is valid, stop processing the template. This
3910          *                             is on by default.
3911          *   nowiki                    Wiki markup in the return value should be escaped
3912          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3913          *   noargs                    Don't replace triple-brace arguments in the return value
3914          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3915          *
3916          * @public
3917          *
3918          * @param string $id The magic word ID
3919          * @param mixed $callback The callback function (and object) to use
3920          * @param integer $flags a combination of the following flags:
3921          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3922          *
3923          * @return The old callback function for this name, if any
3924          */
3925         function setFunctionHook( $id, $callback, $flags = 0 ) {
3926                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3927                 $this->mFunctionHooks[$id] = $callback;
3928
3929                 # Add to function cache
3930                 $mw = MagicWord::get( $id );
3931                 if( !$mw )
3932                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3933
3934                 $synonyms = $mw->getSynonyms();
3935                 $sensitive = intval( $mw->isCaseSensitive() );
3936
3937                 foreach ( $synonyms as $syn ) {
3938                         # Case
3939                         if ( !$sensitive ) {
3940                                 $syn = strtolower( $syn );
3941                         }
3942                         # Add leading hash
3943                         if ( !( $flags & SFH_NO_HASH ) ) {
3944                                 $syn = '#' . $syn;
3945                         }
3946                         # Remove trailing colon
3947                         if ( substr( $syn, -1, 1 ) == ':' ) {
3948                                 $syn = substr( $syn, 0, -1 );
3949                         }
3950                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3951                 }
3952                 return $oldVal;
3953         }
3954
3955         /**
3956          * Get all registered function hook identifiers
3957          *
3958          * @return array
3959          */
3960         function getFunctionHooks() {
3961                 return array_keys( $this->mFunctionHooks );
3962         }
3963
3964         /**
3965          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3966          * Placeholders created in Skin::makeLinkObj()
3967          * Returns an array of links found, indexed by PDBK:
3968          *  0 - broken
3969          *  1 - normal link
3970          *  2 - stub
3971          * $options is a bit field, RLH_FOR_UPDATE to select for update
3972          */
3973         function replaceLinkHolders( &$text, $options = 0 ) {
3974                 global $wgUser;
3975                 global $wgContLang;
3976
3977                 $fname = 'Parser::replaceLinkHolders';
3978                 wfProfileIn( $fname );
3979
3980                 $pdbks = array();
3981                 $colours = array();
3982                 $sk = $this->mOptions->getSkin();
3983                 $linkCache =& LinkCache::singleton();
3984
3985                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3986                         wfProfileIn( $fname.'-check' );
3987                         $dbr = wfGetDB( DB_SLAVE );
3988                         $page = $dbr->tableName( 'page' );
3989                         $threshold = $wgUser->getOption('stubthreshold');
3990
3991                         # Sort by namespace
3992                         asort( $this->mLinkHolders['namespaces'] );
3993
3994                         # Generate query
3995                         $query = false;
3996                         $current = null;
3997                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3998                                 # Make title object
3999                                 $title = $this->mLinkHolders['titles'][$key];
4000
4001                                 # Skip invalid entries.
4002                                 # Result will be ugly, but prevents crash.
4003                                 if ( is_null( $title ) ) {
4004                                         continue;
4005                                 }
4006                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4007
4008                                 # Check if it's a static known link, e.g. interwiki
4009                                 if ( $title->isAlwaysKnown() ) {
4010                                         $colours[$pdbk] = 1;
4011                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4012                                         $colours[$pdbk] = 1;
4013                                         $this->mOutput->addLink( $title, $id );
4014                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4015                                         $colours[$pdbk] = 0;
4016                                 } else {
4017                                         # Not in the link cache, add it to the query
4018                                         if ( !isset( $current ) ) {
4019                                                 $current = $ns;
4020                                                 $query =  "SELECT page_id, page_namespace, page_title";
4021                                                 if ( $threshold > 0 ) {
4022                                                         $query .= ', page_len, page_is_redirect';
4023                                                 }
4024                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4025                                         } elseif ( $current != $ns ) {
4026                                                 $current = $ns;
4027                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4028                                         } else {
4029                                                 $query .= ', ';
4030                                         }
4031
4032                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4033                                 }
4034                         }
4035                         if ( $query ) {
4036                                 $query .= '))';
4037                                 if ( $options & RLH_FOR_UPDATE ) {
4038                                         $query .= ' FOR UPDATE';
4039                                 }
4040
4041                                 $res = $dbr->query( $query, $fname );
4042
4043                                 # Fetch data and form into an associative array
4044                                 # non-existent = broken
4045                                 # 1 = known
4046                                 # 2 = stub
4047                                 while ( $s = $dbr->fetchObject($res) ) {
4048                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4049                                         $pdbk = $title->getPrefixedDBkey();
4050                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4051                                         $this->mOutput->addLink( $title, $s->page_id );
4052
4053                                         if ( $threshold >  0 ) {
4054                                                 $size = $s->page_len;
4055                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4056                                                         $colours[$pdbk] = 1;
4057                                                 } else {
4058                                                         $colours[$pdbk] = 2;
4059                                                 }
4060                                         } else {
4061                                                 $colours[$pdbk] = 1;
4062                                         }
4063                                 }
4064                         }
4065                         wfProfileOut( $fname.'-check' );
4066
4067                         # Do a second query for different language variants of links and categories
4068                         if($wgContLang->hasVariants()){
4069                                 $linkBatch = new LinkBatch();
4070                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4071                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4072                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4073
4074                                 $categories = $this->mOutput->getCategoryLinks();
4075
4076                                 // Add variants of links to link batch
4077                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4078                                         $title = $this->mLinkHolders['titles'][$key];
4079                                         if ( is_null( $title ) )
4080                                                 continue;
4081
4082                                         $pdbk = $title->getPrefixedDBkey();
4083                                         $titleText = $title->getText();
4084
4085                                         // generate all variants of the link title text
4086                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4087
4088                                         // if link was not found (in first query), add all variants to query
4089                                         if ( !isset($colours[$pdbk]) ){
4090                                                 foreach($allTextVariants as $textVariant){
4091                                                         if($textVariant != $titleText){
4092                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4093                                                                 if(is_null($variantTitle)) continue;
4094                                                                 $linkBatch->addObj( $variantTitle );
4095                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4096                                                         }
4097                                                 }
4098                                         }
4099                                 }
4100
4101                                 // process categories, check if a category exists in some variant
4102                                 foreach( $categories as $category){
4103                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4104                                         foreach($variants as $variant){
4105                                                 if($variant != $category){
4106                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4107                                                         if(is_null($variantTitle)) continue;
4108                                                         $linkBatch->addObj( $variantTitle );
4109                                                         $categoryMap[$variant] = $category;
4110                                                 }
4111                                         }
4112                                 }
4113
4114
4115                                 if(!$linkBatch->isEmpty()){
4116                                         // construct query
4117                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4118
4119                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4120                                         if ( $threshold > 0 ) {
4121                                                 $variantQuery .= ', page_len, page_is_redirect';
4122                                         }
4123
4124                                         $variantQuery .= " FROM $page WHERE $titleClause";
4125                                         if ( $options & RLH_FOR_UPDATE ) {
4126                                                 $variantQuery .= ' FOR UPDATE';
4127                                         }
4128
4129                                         $varRes = $dbr->query( $variantQuery, $fname );
4130
4131                                         // for each found variants, figure out link holders and replace
4132                                         while ( $s = $dbr->fetchObject($varRes) ) {
4133
4134                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4135                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4136                                                 $vardbk = $variantTitle->getDBkey();
4137
4138                                                 $holderKeys = array();
4139                                                 if(isset($variantMap[$varPdbk])){
4140                                                         $holderKeys = $variantMap[$varPdbk];
4141                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4142                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4143                                                 }
4144
4145                                                 // loop over link holders
4146                                                 foreach($holderKeys as $key){
4147                                                         $title = $this->mLinkHolders['titles'][$key];
4148                                                         if ( is_null( $title ) ) continue;
4149
4150                                                         $pdbk = $title->getPrefixedDBkey();
4151
4152                                                         if(!isset($colours[$pdbk])){
4153                                                                 // found link in some of the variants, replace the link holder data
4154                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4155                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4156
4157                                                                 // set pdbk and colour
4158                                                                 $pdbks[$key] = $varPdbk;
4159                                                                 if ( $threshold >  0 ) {
4160                                                                         $size = $s->page_len;
4161                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4162                                                                                 $colours[$varPdbk] = 1;
4163                                                                         } else {
4164                                                                                 $colours[$varPdbk] = 2;
4165                                                                         }
4166                                                                 }
4167                                                                 else {
4168                                                                         $colours[$varPdbk] = 1;
4169                                                                 }
4170                                                         }
4171                                                 }
4172
4173                                                 // check if the object is a variant of a category
4174                                                 if(isset($categoryMap[$vardbk])){
4175                                                         $oldkey = $categoryMap[$vardbk];
4176                                                         if($oldkey != $vardbk)
4177                                                                 $varCategories[$oldkey]=$vardbk;
4178                                                 }
4179                                         }
4180
4181                                         // rebuild the categories in original order (if there are replacements)
4182                                         if(count($varCategories)>0){
4183                                                 $newCats = array();
4184                                                 $originalCats = $this->mOutput->getCategories();
4185                                                 foreach($originalCats as $cat => $sortkey){
4186                                                         // make the replacement
4187                                                         if( array_key_exists($cat,$varCategories) )
4188                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4189                                                         else $newCats[$cat] = $sortkey;
4190                                                 }
4191                                                 $this->mOutput->setCategoryLinks($newCats);
4192                                         }
4193                                 }
4194                         }
4195
4196                         # Construct search and replace arrays
4197                         wfProfileIn( $fname.'-construct' );
4198                         $replacePairs = array();
4199                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4200                                 $pdbk = $pdbks[$key];
4201                                 $searchkey = "<!--LINK $key-->";
4202                                 $title = $this->mLinkHolders['titles'][$key];
4203                                 if ( empty( $colours[$pdbk] ) ) {
4204                                         $linkCache->addBadLinkObj( $title );
4205                                         $colours[$pdbk] = 0;
4206                                         $this->mOutput->addLink( $title, 0 );
4207                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4208                                                                         $this->mLinkHolders['texts'][$key],
4209                                                                         $this->mLinkHolders['queries'][$key] );
4210                                 } elseif ( $colours[$pdbk] == 1 ) {
4211                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4212                                                                         $this->mLinkHolders['texts'][$key],
4213                                                                         $this->mLinkHolders['queries'][$key] );
4214                                 } elseif ( $colours[$pdbk] == 2 ) {
4215                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4216                                                                         $this->mLinkHolders['texts'][$key],
4217                                                                         $this->mLinkHolders['queries'][$key] );
4218                                 }
4219                         }
4220                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4221                         wfProfileOut( $fname.'-construct' );
4222
4223                         # Do the thing
4224                         wfProfileIn( $fname.'-replace' );
4225                         $text = preg_replace_callback(
4226                                 '/(<!--LINK .*?-->)/',
4227                                 $replacer->cb(),
4228                                 $text);
4229
4230                         wfProfileOut( $fname.'-replace' );
4231                 }
4232
4233                 # Now process interwiki link holders
4234                 # This is quite a bit simpler than internal links
4235                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4236                         wfProfileIn( $fname.'-interwiki' );
4237                         # Make interwiki link HTML
4238                         $replacePairs = array();
4239                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4240                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4241                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4242                         }
4243                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4244
4245                         $text = preg_replace_callback(
4246                                 '/<!--IWLINK (.*?)-->/',
4247                                 $replacer->cb(),
4248                                 $text );
4249                         wfProfileOut( $fname.'-interwiki' );
4250                 }
4251
4252                 wfProfileOut( $fname );
4253                 return $colours;
4254         }
4255
4256         /**
4257          * Replace <!--LINK--> link placeholders with plain text of links
4258          * (not HTML-formatted).
4259          * @param string $text
4260          * @return string
4261          */
4262         function replaceLinkHoldersText( $text ) {
4263                 $fname = 'Parser::replaceLinkHoldersText';
4264                 wfProfileIn( $fname );
4265
4266                 $text = preg_replace_callback(
4267                         '/<!--(LINK|IWLINK) (.*?)-->/',
4268                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4269                         $text );
4270
4271                 wfProfileOut( $fname );
4272                 return $text;
4273         }
4274
4275         /**
4276          * @param array $matches
4277          * @return string
4278          * @private
4279          */
4280         function replaceLinkHoldersTextCallback( $matches ) {
4281                 $type = $matches[1];
4282                 $key  = $matches[2];
4283                 if( $type == 'LINK' ) {
4284                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4285                                 return $this->mLinkHolders['texts'][$key];
4286                         }
4287                 } elseif( $type == 'IWLINK' ) {
4288                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4289                                 return $this->mInterwikiLinkHolders['texts'][$key];
4290                         }
4291                 }
4292                 return $matches[0];
4293         }
4294
4295         /**
4296          * Tag hook handler for 'pre'.
4297          */
4298         function renderPreTag( $text, $attribs ) {
4299                 // Backwards-compatibility hack
4300                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4301
4302                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4303                 return wfOpenElement( 'pre', $attribs ) .
4304                         Xml::escapeTagsOnly( $content ) .
4305                         '</pre>';
4306         }
4307
4308         /**
4309          * Renders an image gallery from a text with one line per image.
4310          * text labels may be given by using |-style alternative text. E.g.
4311          *   Image:one.jpg|The number "1"
4312          *   Image:tree.jpg|A tree
4313          * given as text will return the HTML of a gallery with two images,
4314          * labeled 'The number "1"' and
4315          * 'A tree'.
4316          */
4317         function renderImageGallery( $text, $params ) {
4318                 $ig = new ImageGallery();
4319                 $ig->setContextTitle( $this->mTitle );
4320                 $ig->setShowBytes( false );
4321                 $ig->setShowFilename( false );
4322                 $ig->setParsing();
4323                 $ig->useSkin( $this->mOptions->getSkin() );
4324
4325                 if( isset( $params['caption'] ) ) {
4326                         $caption = $params['caption'];
4327                         $caption = htmlspecialchars( $caption );
4328                         $caption = $this->replaceInternalLinks( $caption );
4329                         $ig->setCaptionHtml( $caption );
4330                 }
4331                 if( isset( $params['perrow'] ) ) {
4332                         $ig->setPerRow( $params['perrow'] );
4333                 }
4334                 if( isset( $params['widths'] ) ) {
4335                         $ig->setWidths( $params['widths'] );
4336                 }
4337                 if( isset( $params['heights'] ) ) {
4338                         $ig->setHeights( $params['heights'] );
4339                 }
4340
4341                 $lines = explode( "\n", $text );
4342                 foreach ( $lines as $line ) {
4343                         # match lines like these:
4344                         # Image:someimage.jpg|This is some image
4345                         $matches = array();
4346                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4347                         # Skip empty lines
4348                         if ( count( $matches ) == 0 ) {
4349                                 continue;
4350                         }
4351                         $tp = Title::newFromText( $matches[1] );
4352                         $nt =& $tp;
4353                         if( is_null( $nt ) ) {
4354                                 # Bogus title. Ignore these so we don't bomb out later.
4355                                 continue;
4356                         }
4357                         if ( isset( $matches[3] ) ) {
4358                                 $label = $matches[3];
4359                         } else {
4360                                 $label = '';
4361                         }
4362
4363                         $pout = $this->parse( $label,
4364                                 $this->mTitle,
4365                                 $this->mOptions,
4366                                 false, // Strip whitespace...?
4367                                 false  // Don't clear state!
4368                         );
4369                         $html = $pout->getText();
4370
4371                         $ig->add( new Image( $nt ), $html );
4372
4373                         # Only add real images (bug #5586)
4374                         if ( $nt->getNamespace() == NS_IMAGE ) {
4375                                 $this->mOutput->addImage( $nt->getDBkey() );
4376                         }
4377                 }
4378                 return $ig->toHTML();
4379         }
4380
4381         /**
4382          * Parse image options text and use it to make an image
4383          */
4384         function makeImage( $nt, $options ) {
4385                 global $wgUseImageResize, $wgDjvuRenderer;
4386
4387                 # Check if the options text is of the form "options|alt text"
4388                 # Options are:
4389                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4390                 #  * left               no resizing, just left align. label is used for alt= only
4391                 #  * right              same, but right aligned
4392                 #  * none               same, but not aligned
4393                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4394                 #  * center             center the image
4395                 #  * framed             Keep original image size, no magnify-button.
4396                 # vertical-align values (no % or length right now):
4397                 #  * baseline
4398                 #  * sub
4399                 #  * super
4400                 #  * top
4401                 #  * text-top
4402                 #  * middle
4403                 #  * bottom
4404                 #  * text-bottom
4405
4406                 $part = array_map( 'trim', explode( '|', $options) );
4407
4408                 $mwAlign = array();
4409                 $alignments = array( 'left', 'right', 'center', 'none', 'baseline', 'sub', 'super', 'top', 'text-top', 'middle', 'bottom', 'text-bottom' );
4410                 foreach ( $alignments as $alignment ) {
4411                         $mwAlign[$alignment] =& MagicWord::get( 'img_'.$alignment );
4412                 }
4413                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4414                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4415                 $mwWidth  =& MagicWord::get( 'img_width' );
4416                 $mwFramed =& MagicWord::get( 'img_framed' );
4417                 $mwPage   =& MagicWord::get( 'img_page' );
4418                 $caption = '';
4419
4420                 $width = $height = $framed = $thumb = false;
4421                 $page = null;
4422                 $manual_thumb = '' ;
4423                 $align = $valign = '';
4424
4425                 foreach( $part as $val ) {
4426                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4427                                 $thumb=true;
4428                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4429                                 # use manually specified thumbnail
4430                                 $thumb=true;
4431                                 $manual_thumb = $match;
4432                         } else {
4433                                 foreach( $alignments as $alignment ) {
4434                                         if ( ! is_null( $mwAlign[$alignment]->matchVariableStartToEnd($val) ) ) {
4435                                                 switch ( $alignment ) {
4436                                                         case 'left': case 'right': case 'center': case 'none':
4437                                                                 $align = $alignment; break;
4438                                                         default:
4439                                                                 $valign = $alignment;
4440                                                 }
4441                                                 continue 2;
4442                                         }
4443                                 }
4444                                 if ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4445                                 && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4446                                         # Select a page in a multipage document
4447                                         $page = $match;
4448                                 } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4449                                         wfDebug( "img_width match: $match\n" );
4450                                         # $match is the image width in pixels
4451                                         $m = array();
4452                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4453                                                 $width = intval( $m[1] );
4454                                                 $height = intval( $m[2] );
4455                                         } else {
4456                                                 $width = intval($match);
4457                                         }
4458                                 } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4459                                         $framed=true;
4460                                 } else {
4461                                         $caption = $val;
4462                                 }
4463                         }
4464                 }
4465                 # Strip bad stuff out of the alt text
4466                 $alt = $this->replaceLinkHoldersText( $caption );
4467
4468                 # make sure there are no placeholders in thumbnail attributes
4469                 # that are later expanded to html- so expand them now and
4470                 # remove the tags
4471                 $alt = $this->mStripState->unstripBoth( $alt );
4472                 $alt = Sanitizer::stripAllTags( $alt );
4473
4474                 # Linker does the rest
4475                 $sk = $this->mOptions->getSkin();
4476                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page, $valign );
4477         }
4478
4479         /**
4480          * Set a flag in the output object indicating that the content is dynamic and
4481          * shouldn't be cached.
4482          */
4483         function disableCache() {
4484                 wfDebug( "Parser output marked as uncacheable.\n" );
4485                 $this->mOutput->mCacheTime = -1;
4486         }
4487
4488         /**#@+
4489          * Callback from the Sanitizer for expanding items found in HTML attribute
4490          * values, so they can be safely tested and escaped.
4491          * @param string $text
4492          * @param array $args
4493          * @return string
4494          * @private
4495          */
4496         function attributeStripCallback( &$text, $args ) {
4497                 $text = $this->replaceVariables( $text, $args );
4498                 $text = $this->mStripState->unstripBoth( $text );
4499                 return $text;
4500         }
4501
4502         /**#@-*/
4503
4504         /**#@+
4505          * Accessor/mutator
4506          */
4507         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4508         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4509         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4510         /**#@-*/
4511
4512         /**#@+
4513          * Accessor
4514          */
4515         function getTags() { return array_keys( $this->mTagHooks ); }
4516         /**#@-*/
4517
4518
4519         /**
4520          * Break wikitext input into sections, and either pull or replace
4521          * some particular section's text.
4522          *
4523          * External callers should use the getSection and replaceSection methods.
4524          *
4525          * @param $text Page wikitext
4526          * @param $section Numbered section. 0 pulls the text before the first
4527          *                 heading; other numbers will pull the given section
4528          *                 along with its lower-level subsections.
4529          * @param $mode One of "get" or "replace"
4530          * @param $newtext Replacement text for section data.
4531          * @return string for "get", the extracted section text.
4532          *                for "replace", the whole page with the section replaced.
4533          */
4534         private function extractSections( $text, $section, $mode, $newtext='' ) {
4535                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4536                 # comments to be stripped as well)
4537                 $stripState = new StripState;
4538
4539                 $oldOutputType = $this->mOutputType;
4540                 $oldOptions = $this->mOptions;
4541                 $this->mOptions = new ParserOptions();
4542                 $this->setOutputType( OT_WIKI );
4543
4544                 $striptext = $this->strip( $text, $stripState, true );
4545
4546                 $this->setOutputType( $oldOutputType );
4547                 $this->mOptions = $oldOptions;
4548
4549                 # now that we can be sure that no pseudo-sections are in the source,
4550                 # split it up by section
4551                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4552                 $comment = "(?:$uniq-!--.*?QINU)";
4553                 $secs = preg_split(
4554                         "/
4555                         (
4556                                 ^
4557                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4558                                 (=+) # Should this be limited to 6?
4559                                 .+?  # Section title...
4560                                 \\2  # Ending = count must match start
4561                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4562                                 $
4563                         |
4564                                 <h([1-6])\b.*?>
4565                                 .*?
4566                                 <\/h\\3\s*>
4567                         )
4568                         /mix",
4569                         $striptext, -1,
4570                         PREG_SPLIT_DELIM_CAPTURE);
4571
4572                 if( $mode == "get" ) {
4573                         if( $section == 0 ) {
4574                                 // "Section 0" returns the content before any other section.
4575                                 $rv = $secs[0];
4576                         } else {
4577                                 //track missing section, will replace if found.
4578                                 $rv = $newtext;
4579                         }
4580                 } elseif( $mode == "replace" ) {
4581                         if( $section == 0 ) {
4582                                 $rv = $newtext . "\n\n";
4583                                 $remainder = true;
4584                         } else {
4585                                 $rv = $secs[0];
4586                                 $remainder = false;
4587                         }
4588                 }
4589                 $count = 0;
4590                 $sectionLevel = 0;
4591                 for( $index = 1; $index < count( $secs ); ) {
4592                         $headerLine = $secs[$index++];
4593                         if( $secs[$index] ) {
4594                                 // A wiki header
4595                                 $headerLevel = strlen( $secs[$index++] );
4596                         } else {
4597                                 // An HTML header
4598                                 $index++;
4599                                 $headerLevel = intval( $secs[$index++] );
4600                         }
4601                         $content = $secs[$index++];
4602
4603                         $count++;
4604                         if( $mode == "get" ) {
4605                                 if( $count == $section ) {
4606                                         $rv = $headerLine . $content;
4607                                         $sectionLevel = $headerLevel;
4608                                 } elseif( $count > $section ) {
4609                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4610                                                 $rv .= $headerLine . $content;
4611                                         } else {
4612                                                 // Broke out to a higher-level section
4613                                                 break;
4614                                         }
4615                                 }
4616                         } elseif( $mode == "replace" ) {
4617                                 if( $count < $section ) {
4618                                         $rv .= $headerLine . $content;
4619                                 } elseif( $count == $section ) {
4620                                         $rv .= $newtext . "\n\n";
4621                                         $sectionLevel = $headerLevel;
4622                                 } elseif( $count > $section ) {
4623                                         if( $headerLevel <= $sectionLevel ) {
4624                                                 // Passed the section's sub-parts.
4625                                                 $remainder = true;
4626                                         }
4627                                         if( $remainder ) {
4628                                                 $rv .= $headerLine . $content;
4629                                         }
4630                                 }
4631                         }
4632                 }
4633                 if (is_string($rv))
4634                         # reinsert stripped tags
4635                         $rv = trim( $stripState->unstripBoth( $rv ) );
4636
4637                 return $rv;
4638         }
4639
4640         /**
4641          * This function returns the text of a section, specified by a number ($section).
4642          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4643          * the first section before any such heading (section 0).
4644          *
4645          * If a section contains subsections, these are also returned.
4646          *
4647          * @param $text String: text to look in
4648          * @param $section Integer: section number
4649          * @param $deftext: default to return if section is not found
4650          * @return string text of the requested section
4651          */
4652         public function getSection( $text, $section, $deftext='' ) {
4653                 return $this->extractSections( $text, $section, "get", $deftext );
4654         }
4655
4656         public function replaceSection( $oldtext, $section, $text ) {
4657                 return $this->extractSections( $oldtext, $section, "replace", $text );
4658         }
4659
4660         /**
4661          * Get the timestamp associated with the current revision, adjusted for
4662          * the default server-local timestamp
4663          */
4664         function getRevisionTimestamp() {
4665                 if ( is_null( $this->mRevisionTimestamp ) ) {
4666                         wfProfileIn( __METHOD__ );
4667                         global $wgContLang;
4668                         $dbr = wfGetDB( DB_SLAVE );
4669                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4670                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4671
4672                         // Normalize timestamp to internal MW format for timezone processing.
4673                         // This has the added side-effect of replacing a null value with
4674                         // the current time, which gives us more sensible behavior for
4675                         // previews.
4676                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4677
4678                         // The cryptic '' timezone parameter tells to use the site-default
4679                         // timezone offset instead of the user settings.
4680                         //
4681                         // Since this value will be saved into the parser cache, served
4682                         // to other users, and potentially even used inside links and such,
4683                         // it needs to be consistent for all visitors.
4684                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4685
4686                         wfProfileOut( __METHOD__ );
4687                 }
4688                 return $this->mRevisionTimestamp;
4689         }
4690
4691         /**
4692          * Mutator for $mDefaultSort
4693          *
4694          * @param $sort New value
4695          */
4696         public function setDefaultSort( $sort ) {
4697                 $this->mDefaultSort = $sort;
4698         }
4699
4700         /**
4701          * Accessor for $mDefaultSort
4702          * Will use the title/prefixed title if none is set
4703          *
4704          * @return string
4705          */
4706         public function getDefaultSort() {
4707                 if( $this->mDefaultSort !== false ) {
4708                         return $this->mDefaultSort;
4709                 } else {
4710                         return $this->mTitle->getNamespace() == NS_CATEGORY
4711                                         ? $this->mTitle->getText()
4712                                         : $this->mTitle->getPrefixedText();
4713                 }
4714         }
4715
4716 }
4717
4718 class OnlyIncludeReplacer {
4719         var $output = '';
4720
4721         function replace( $matches ) {
4722                 if ( substr( $matches[1], -1 ) == "\n" ) {
4723                         $this->output .= substr( $matches[1], 0, -1 );
4724                 } else {
4725                         $this->output .= $matches[1];
4726                 }
4727         }
4728 }
4729
4730 class StripState {
4731         var $general, $nowiki;
4732
4733         function __construct() {
4734                 $this->general = new ReplacementArray;
4735                 $this->nowiki = new ReplacementArray;
4736         }
4737
4738         function unstripGeneral( $text ) {
4739                 wfProfileIn( __METHOD__ );
4740                 $text = $this->general->replace( $text );
4741                 wfProfileOut( __METHOD__ );
4742                 return $text;
4743         }
4744
4745         function unstripNoWiki( $text ) {
4746                 wfProfileIn( __METHOD__ );
4747                 $text = $this->nowiki->replace( $text );
4748                 wfProfileOut( __METHOD__ );
4749                 return $text;
4750         }
4751
4752         function unstripBoth( $text ) {
4753                 wfProfileIn( __METHOD__ );
4754                 $text = $this->general->replace( $text );
4755                 $text = $this->nowiki->replace( $text );
4756                 wfProfileOut( __METHOD__ );
4757                 return $text;
4758         }
4759 }
4760
4761 ?>