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