includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /**
  10  * Update this version number when the ParserOutput format
  11  * changes in an incompatible way, so the parser cache
  12  * can automatically discard old data.
  13  */
  14 define( 'MW_PARSER_VERSION', '1.6.1' );
  15
  16 define( 'RLH_FOR_UPDATE', 1 );
  17
  18 # Allowed values for $mOutputType
  19 define( 'OT_HTML', 1 );
  20 define( 'OT_WIKI', 2 );
  21 define( 'OT_MSG' , 3 );
  22 define( 'OT_PREPROCESS', 4 );
  23
  24 # Flags for setFunctionHook
  25 define( 'SFH_NO_HASH', 1 );
  26
  27 # string parameter for extractTags which will cause it
  28 # to strip HTML comments in addition to regular
  29 # <XML>-style tags. This should not be anything we
  30 # may want to use in wikisyntax
  31 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  32
  33 # Constants needed for external link processing
  34 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  35 # Everything except bracket, space, or control characters
  36 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  37 # Including space, but excluding newlines
  38 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  39 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  40 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  41 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  42         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  43 define( 'EXT_IMAGE_REGEX',
  44         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  45         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  46         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  47 );
  48
  49 // State constants for the definition list colon extraction
  50 define( 'MW_COLON_STATE_TEXT', 0 );
  51 define( 'MW_COLON_STATE_TAG', 1 );
  52 define( 'MW_COLON_STATE_TAGSTART', 2 );
  53 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  54 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  55 define( 'MW_COLON_STATE_COMMENT', 5 );
  56 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  57 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  58
  59 /**
  60  * PHP Parser
  61  *
  62  * Processes wiki markup
  63  *
  64  * <pre>
  65  * There are four main entry points into the Parser class:
  66  * parse()
  67  *   produces HTML output
  68  * preSaveTransform().
  69  *   produces altered wiki markup.
  70  * transformMsg()
  71  *   performs brace substitution on MediaWiki messages
  72  * preprocess()
  73  *   removes HTML comments and expands templates
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgContLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  84  *  $wgMaxArticleSize*
  85  *
  86  *  * only within ParserOptions
  87  * </pre>
  88  *
  89  * @package MediaWiki
  90  */
  91 class Parser
  92 {
  93         /**#@+
  94          * @private
  95          */
  96         # Persistent:
  97         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 103         var $mIncludeSizes;
 104         var $mTemplates,        // cache of already loaded templates, avoids
 105                                 // multiple SQL queries for the same string
 106             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 107                                 // in this path. Used for loop detection.
 108
 109         # Temporary
 110         # These are variables reset at least once per parse regardless of $clearState
 111         var $mOptions,      // ParserOptions object
 112                 $mTitle,        // Title context, used for self-link rendering and similar things
 113                 $mOutputType,   // Output type, one of the OT_xxx constants
 114                 $ot,            // Shortcut alias, see setOutputType()
 115                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 116                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 117                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 118
 119         /**#@-*/
 120
 121         /**
 122          * Constructor
 123          *
 124          * @public
 125          */
 126         function Parser() {
 127                 $this->mTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mFirstCall = true;
 131         }
 132
 133         /**
 134          * Do various kinds of initialisation on the first call of the parser
 135          */
 136         function firstCallInit() {
 137                 if ( !$this->mFirstCall ) {
 138                         return;
 139                 }
 140
 141                 wfProfileIn( __METHOD__ );
 142                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 143
 144                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 145
 146                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 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                 $this->mIncludeSizes = array(
 230                         'pre-expand' => 0,
 231                         'post-expand' => 0,
 232                         'arg' => 0
 233                 );
 234
 235                 wfRunHooks( 'ParserClearState', array( &$this ) );
 236                 wfProfileOut( __METHOD__ );
 237         }
 238
 239         function setOutputType( $ot ) {
 240                 $this->mOutputType = $ot;
 241                 // Shortcut alias
 242                 $this->ot = array(
 243                         'html' => $ot == OT_HTML,
 244                         'wiki' => $ot == OT_WIKI,
 245                         'msg' => $ot == OT_MSG,
 246                         'pre' => $ot == OT_PREPROCESS,
 247                 );
 248         }
 249
 250         /**
 251          * Accessor for mUniqPrefix.
 252          *
 253          * @public
 254          */
 255         function uniqPrefix() {
 256                 return $this->mUniqPrefix;
 257         }
 258
 259         /**
 260          * Convert wikitext to HTML
 261          * Do not call this function recursively.
 262          *
 263          * @private
 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         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                 wfProfileIn( __METHOD__ );
 533                 $render = ($this->mOutputType == OT_HTML);
 534
 535                 $uniq_prefix = $this->mUniqPrefix;
 536                 $commentState = new ReplacementArray;
 537
 538                 $elements = array_merge(
 539                         array( 'nowiki', 'gallery' ),
 540                         array_keys( $this->mTagHooks ) );
 541                 global $wgRawHtml;
 542                 if( $wgRawHtml ) {
 543                         $elements[] = 'html';
 544                 }
 545                 if( $this->mOptions->getUseTeX() ) {
 546                         $elements[] = 'math';
 547                 }
 548
 549                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 550                 foreach ( $elements AS $k => $v ) {
 551                         if ( !in_array ( $v , $dontstrip ) ) continue;
 552                         unset ( $elements[$k] );
 553                 }
 554
 555                 $matches = array();
 556                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 557
 558                 foreach( $matches as $marker => $data ) {
 559                         list( $element, $content, $params, $tag ) = $data;
 560                         if( $render ) {
 561                                 $tagName = strtolower( $element );
 562                                 wfProfileIn( __METHOD__."-render-$tagName" );
 563                                 switch( $tagName ) {
 564                                 case '!--':
 565                                         // Comment
 566                                         if( substr( $tag, -3 ) == '-->' ) {
 567                                                 $output = $tag;
 568                                         } else {
 569                                                 // Unclosed comment in input.
 570                                                 // Close it so later stripping can remove it
 571                                                 $output = "$tag-->";
 572                                         }
 573                                         break;
 574                                 case 'html':
 575                                         if( $wgRawHtml ) {
 576                                                 $output = $content;
 577                                                 break;
 578                                         }
 579                                         // Shouldn't happen otherwise. :)
 580                                 case 'nowiki':
 581                                         $output = Xml::escapeTagsOnly( $content );
 582                                         break;
 583                                 case 'math':
 584                                         $output = MathRenderer::renderMath( $content );
 585                                         break;
 586                                 case 'gallery':
 587                                         $output = $this->renderImageGallery( $content, $params );
 588                                         break;
 589                                 default:
 590                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 591                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 592                                                         array( $content, $params, $this ) );
 593                                         } else {
 594                                                 throw new MWException( "Invalid call hook $element" );
 595                                         }
 596                                 }
 597                                 wfProfileOut( __METHOD__."-render-$tagName" );
 598                         } else {
 599                                 // Just stripping tags; keep the source
 600                                 $output = $tag;
 601                         }
 602
 603                         // Unstrip the output, because unstrip() is no longer recursive so
 604                         // it won't do it itself
 605                         $output = $state->unstripBoth( $output );
 606
 607                         if( !$stripcomments && $element == '!--' ) {
 608                                 $commentState->setPair( $marker, $output );
 609                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 610                                 $state->nowiki->setPair( $marker, $output );
 611                         } else {
 612                                 $state->general->setPair( $marker, $output );
 613                         }
 614                 }
 615
 616                 # Unstrip comments unless explicitly told otherwise.
 617                 # (The comments are always stripped prior to this point, so as to
 618                 # not invoke any extension tags / parser hooks contained within
 619                 # a comment.)
 620                 if ( !$stripcomments ) {
 621                         // Put them all back and forget them
 622                         $text = $commentState->replace( $text );
 623                 }
 624
 625                 wfProfileOut( __METHOD__ );
 626                 return $text;
 627         }
 628
 629         /**
 630          * Restores pre, math, and other extensions removed by strip()
 631          *
 632          * always call unstripNoWiki() after this one
 633          * @private
 634          * @deprecated use $this->mStripState->unstrip()
 635          */
 636         function unstrip( $text, $state ) {
 637                 return $state->unstripGeneral( $text );
 638         }
 639
 640         /**
 641          * Always call this after unstrip() to preserve the order
 642          *
 643          * @private
 644          * @deprecated use $this->mStripState->unstrip()
 645          */
 646         function unstripNoWiki( $text, $state ) {
 647                 return $state->unstripNoWiki( $text );
 648         }
 649
 650         /**
 651          * @deprecated use $this->mStripState->unstripBoth()
 652          */
 653         function unstripForHTML( $text ) {
 654                 return $this->mStripState->unstripBoth( $text );
 655         }
 656
 657         /**
 658          * Add an item to the strip state
 659          * Returns the unique tag which must be inserted into the stripped text
 660          * The tag will be replaced with the original text in unstrip()
 661          *
 662          * @private
 663          */
 664         function insertStripItem( $text, &$state ) {
 665                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 666                 $state->general->setPair( $rnd, $text );
 667                 return $rnd;
 668         }
 669
 670         /**
 671          * Interface with html tidy, used if $wgUseTidy = true.
 672          * If tidy isn't able to correct the markup, the original will be
 673          * returned in all its glory with a warning comment appended.
 674          *
 675          * Either the external tidy program or the in-process tidy extension
 676          * will be used depending on availability. Override the default
 677          * $wgTidyInternal setting to disable the internal if it's not working.
 678          *
 679          * @param string $text Hideous HTML input
 680          * @return string Corrected HTML output
 681          * @public
 682          * @static
 683          */
 684         function tidy( $text ) {
 685                 global $wgTidyInternal;
 686                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 687 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 688 '<head><title>test</title></head><body>'.$text.'</body></html>';
 689                 if( $wgTidyInternal ) {
 690                         $correctedtext = Parser::internalTidy( $wrappedtext );
 691                 } else {
 692                         $correctedtext = Parser::externalTidy( $wrappedtext );
 693                 }
 694                 if( is_null( $correctedtext ) ) {
 695                         wfDebug( "Tidy error detected!\n" );
 696                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 697                 }
 698                 return $correctedtext;
 699         }
 700
 701         /**
 702          * Spawn an external HTML tidy process and get corrected markup back from it.
 703          *
 704          * @private
 705          * @static
 706          */
 707         function externalTidy( $text ) {
 708                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 709                 $fname = 'Parser::externalTidy';
 710                 wfProfileIn( $fname );
 711
 712                 $cleansource = '';
 713                 $opts = ' -utf8';
 714
 715                 $descriptorspec = array(
 716                         0 => array('pipe', 'r'),
 717                         1 => array('pipe', 'w'),
 718                         2 => array('file', '/dev/null', 'a')
 719                 );
 720                 $pipes = array();
 721                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 722                 if (is_resource($process)) {
 723                         // Theoretically, this style of communication could cause a deadlock
 724                         // here. If the stdout buffer fills up, then writes to stdin could
 725                         // block. This doesn't appear to happen with tidy, because tidy only
 726                         // writes to stdout after it's finished reading from stdin. Search
 727                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 728                         fwrite($pipes[0], $text);
 729                         fclose($pipes[0]);
 730                         while (!feof($pipes[1])) {
 731                                 $cleansource .= fgets($pipes[1], 1024);
 732                         }
 733                         fclose($pipes[1]);
 734                         proc_close($process);
 735                 }
 736
 737                 wfProfileOut( $fname );
 738
 739                 if( $cleansource == '' && $text != '') {
 740                         // Some kind of error happened, so we couldn't get the corrected text.
 741                         // Just give up; we'll use the source text and append a warning.
 742                         return null;
 743                 } else {
 744                         return $cleansource;
 745                 }
 746         }
 747
 748         /**
 749          * Use the HTML tidy PECL extension to use the tidy library in-process,
 750          * saving the overhead of spawning a new process. Currently written to
 751          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 752          *
 753          * 'pear install tidy' should be able to compile the extension module.
 754          *
 755          * @private
 756          * @static
 757          */
 758         function internalTidy( $text ) {
 759                 global $wgTidyConf;
 760                 $fname = 'Parser::internalTidy';
 761                 wfProfileIn( $fname );
 762
 763                 tidy_load_config( $wgTidyConf );
 764                 tidy_set_encoding( 'utf8' );
 765                 tidy_parse_string( $text );
 766                 tidy_clean_repair();
 767                 if( tidy_get_status() == 2 ) {
 768                         // 2 is magic number for fatal error
 769                         // http://www.php.net/manual/en/function.tidy-get-status.php
 770                         $cleansource = null;
 771                 } else {
 772                         $cleansource = tidy_get_output();
 773                 }
 774                 wfProfileOut( $fname );
 775                 return $cleansource;
 776         }
 777
 778         /**
 779          * parse the wiki syntax used to render tables
 780          *
 781          * @private
 782          */
 783         function doTableStuff ( $t ) {
 784                 $fname = 'Parser::doTableStuff';
 785                 wfProfileIn( $fname );
 786
 787                 $t = explode ( "\n" , $t ) ;
 788                 $td = array () ; # Is currently a td tag open?
 789                 $ltd = array () ; # Was it TD or TH?
 790                 $tr = array () ; # Is currently a tr tag open?
 791                 $ltr = array () ; # tr attributes
 792                 $has_opened_tr = array(); # Did this table open a <tr> element?
 793                 $indent_level = 0; # indent level of the table
 794                 foreach ( $t AS $k => $x )
 795                 {
 796                         $x = trim ( $x ) ;
 797                         $fc = substr ( $x , 0 , 1 ) ;
 798                         $matches = array();
 799                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 800                                 $indent_level = strlen( $matches[1] );
 801
 802                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 803
 804                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 805                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 806                                 array_push ( $td , false ) ;
 807                                 array_push ( $ltd , '' ) ;
 808                                 array_push ( $tr , false ) ;
 809                                 array_push ( $ltr , '' ) ;
 810                                 array_push ( $has_opened_tr, false );
 811                         }
 812                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 813                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 814                                 $z = "</table>" . substr ( $x , 2);
 815                                 $l = array_pop ( $ltd ) ;
 816                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 817                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 818                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 819                                 array_pop ( $ltr ) ;
 820                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 821                         }
 822                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 823                                 $x = substr ( $x , 1 ) ;
 824                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 825                                 $z = '' ;
 826                                 $l = array_pop ( $ltd ) ;
 827                                 array_pop ( $has_opened_tr );
 828                                 array_push ( $has_opened_tr , true ) ;
 829                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 830                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 831                                 array_pop ( $ltr ) ;
 832                                 $t[$k] = $z ;
 833                                 array_push ( $tr , false ) ;
 834                                 array_push ( $td , false ) ;
 835                                 array_push ( $ltd , '' ) ;
 836                                 $attributes = $this->mStripState->unstripBoth( $x );
 837                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 838                         }
 839                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 840                                 # $x is a table row
 841                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 842                                         $fc = '+' ;
 843                                         $x = substr ( $x , 1 ) ;
 844                                 }
 845                                 $after = substr ( $x , 1 ) ;
 846                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 847
 848                                 // Split up multiple cells on the same line.
 849                                 // FIXME: This can result in improper nesting of tags processed
 850                                 // by earlier parser steps, but should avoid splitting up eg
 851                                 // attribute values containing literal "||".
 852                                 $after = StringUtils::explodeMarkup( '||', $after );
 853
 854                                 $t[$k] = '' ;
 855
 856                                 # Loop through each table cell
 857                                 foreach ( $after AS $theline )
 858                                 {
 859                                         $z = '' ;
 860                                         if ( $fc != '+' )
 861                                         {
 862                                                 $tra = array_pop ( $ltr ) ;
 863                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 864                                                 array_push ( $tr , true ) ;
 865                                                 array_push ( $ltr , '' ) ;
 866                                                 array_pop ( $has_opened_tr );
 867                                                 array_push ( $has_opened_tr , true ) ;
 868                                         }
 869
 870                                         $l = array_pop ( $ltd ) ;
 871                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 872                                         if ( $fc == '|' ) {
 873                                             $l = 'td' ;
 874                                         } else if ( $fc == '!' ) {
 875                                             $l = 'th' ;
 876                                         } else if ( $fc == '+' ) {
 877                                             $l = 'caption' ;
 878                                         } else {
 879                                             $l = '' ;
 880                                         }
 881                                         array_push ( $ltd , $l ) ;
 882
 883                                         # Cell parameters
 884                                         $y = explode ( '|' , $theline , 2 ) ;
 885                                         # Note that a '|' inside an invalid link should not
 886                                         # be mistaken as delimiting cell parameters
 887                                         if ( strpos( $y[0], '[[' ) !== false ) {
 888                                                 $y = array ($theline);
 889                                         }
 890                                         if ( count ( $y ) == 1 )
 891                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 892                                         else {
 893                                                 $attributes = $this->mStripState->unstripBoth( $y[0] );
 894                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 895                                         }
 896                                         $t[$k] .= $y ;
 897                                         array_push ( $td , true ) ;
 898                                 }
 899                         }
 900                 }
 901
 902                 # Closing open td, tr && table
 903                 while ( count ( $td ) > 0 )
 904                 {
 905                         $l = array_pop ( $ltd ) ;
 906                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 907                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 908                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 909                         $t[] = '</table>' ;
 910                 }
 911
 912                 $t = implode ( "\n" , $t ) ;
 913                 # special case: don't return empty table
 914                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 915                         $t = '';
 916                 wfProfileOut( $fname );
 917                 return $t ;
 918         }
 919
 920         /**
 921          * Helper function for parse() that transforms wiki markup into
 922          * HTML. Only called for $mOutputType == OT_HTML.
 923          *
 924          * @private
 925          */
 926         function internalParse( $text ) {
 927                 $args = array();
 928                 $isMain = true;
 929                 $fname = 'Parser::internalParse';
 930                 wfProfileIn( $fname );
 931
 932                 # Hook to suspend the parser in this state
 933                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 934                         wfProfileOut( $fname );
 935                         return $text ;
 936                 }
 937
 938                 # Remove <noinclude> tags and <includeonly> sections
 939                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 940                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 941                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 942
 943                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 944
 945                 $text = $this->replaceVariables( $text, $args );
 946
 947                 // Tables need to come after variable replacement for things to work
 948                 // properly; putting them before other transformations should keep
 949                 // exciting things like link expansions from showing up in surprising
 950                 // places.
 951                 $text = $this->doTableStuff( $text );
 952
 953                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 954
 955                 $text = $this->stripToc( $text );
 956                 $this->stripNoGallery( $text );
 957                 $text = $this->doHeadings( $text );
 958                 if($this->mOptions->getUseDynamicDates()) {
 959                         $df =& DateFormatter::getInstance();
 960                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 961                 }
 962                 $text = $this->doAllQuotes( $text );
 963                 $text = $this->replaceInternalLinks( $text );
 964                 $text = $this->replaceExternalLinks( $text );
 965
 966                 # replaceInternalLinks may sometimes leave behind
 967                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 968                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 969
 970                 $text = $this->doMagicLinks( $text );
 971                 $text = $this->formatHeadings( $text, $isMain );
 972
 973                 wfProfileOut( $fname );
 974                 return $text;
 975         }
 976
 977         /**
 978          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 979          * magic external links.
 980          *
 981          * @private
 982          */
 983         function &doMagicLinks( &$text ) {
 984                 wfProfileIn( __METHOD__ );
 985                 $text = preg_replace_callback(
 986                         '!(?:                           # Start cases
 987                             <a.*?</a> |                 # Skip link text
 988                             <.*?> |                     # Skip stuff inside HTML elements
 989                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 990                             ISBN\s+(\b[0-9Xx\ \-]+)     # ISBN, capture number as m[2]
 991                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 992                 wfProfileOut( __METHOD__ );
 993                 return $text;
 994         }
 995
 996         function magicLinkCallback( $m ) {
 997                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 998                         # Skip HTML element
 999                         return $m[0];
1000                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1001                         $isbn = $m[2];
1002                         $num = strtr( $isbn, array(
1003                                 '-' => '',
1004                                 ' ' => '',
1005                                 'x' => 'X',
1006                         ));
1007                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1008                         $text = '<a href="' .
1009                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1010                                 "\" class=\"internal\">ISBN $isbn</a>";
1011                 } else {
1012                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1013                                 $keyword = 'RFC';
1014                                 $urlmsg = 'rfcurl';
1015                                 $id = $m[1];
1016                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1017                                 $keyword = 'PMID';
1018                                 $urlmsg = 'pubmedurl';
1019                                 $id = $m[1];
1020                         } else {
1021                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1022                                         substr($m[0], 0, 20 ) . '"' );
1023                         }
1024
1025                         $url = wfMsg( $urlmsg, $id);
1026                         $sk =& $this->mOptions->getSkin();
1027                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1028                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1029                 }
1030                 return $text;
1031         }
1032
1033         /**
1034          * Parse headers and return html
1035          *
1036          * @private
1037          */
1038         function doHeadings( $text ) {
1039                 $fname = 'Parser::doHeadings';
1040                 wfProfileIn( $fname );
1041                 for ( $i = 6; $i >= 1; --$i ) {
1042                         $h = str_repeat( '=', $i );
1043                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1044                           "<h{$i}>\\1</h{$i}>\\2", $text );
1045                 }
1046                 wfProfileOut( $fname );
1047                 return $text;
1048         }
1049
1050         /**
1051          * Replace single quotes with HTML markup
1052          * @private
1053          * @return string the altered text
1054          */
1055         function doAllQuotes( $text ) {
1056                 $fname = 'Parser::doAllQuotes';
1057                 wfProfileIn( $fname );
1058                 $outtext = '';
1059                 $lines = explode( "\n", $text );
1060                 foreach ( $lines as $line ) {
1061                         $outtext .= $this->doQuotes ( $line ) . "\n";
1062                 }
1063                 $outtext = substr($outtext, 0,-1);
1064                 wfProfileOut( $fname );
1065                 return $outtext;
1066         }
1067
1068         /**
1069          * Helper function for doAllQuotes()
1070          * @private
1071          */
1072         function doQuotes( $text ) {
1073                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1074                 if ( count( $arr ) == 1 )
1075                         return $text;
1076                 else
1077                 {
1078                         # First, do some preliminary work. This may shift some apostrophes from
1079                         # being mark-up to being text. It also counts the number of occurrences
1080                         # of bold and italics mark-ups.
1081                         $i = 0;
1082                         $numbold = 0;
1083                         $numitalics = 0;
1084                         foreach ( $arr as $r )
1085                         {
1086                                 if ( ( $i % 2 ) == 1 )
1087                                 {
1088                                         # If there are ever four apostrophes, assume the first is supposed to
1089                                         # be text, and the remaining three constitute mark-up for bold text.
1090                                         if ( strlen( $arr[$i] ) == 4 )
1091                                         {
1092                                                 $arr[$i-1] .= "'";
1093                                                 $arr[$i] = "'''";
1094                                         }
1095                                         # If there are more than 5 apostrophes in a row, assume they're all
1096                                         # text except for the last 5.
1097                                         else if ( strlen( $arr[$i] ) > 5 )
1098                                         {
1099                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1100                                                 $arr[$i] = "'''''";
1101                                         }
1102                                         # Count the number of occurrences of bold and italics mark-ups.
1103                                         # We are not counting sequences of five apostrophes.
1104                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1105                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1106                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1107                                 }
1108                                 $i++;
1109                         }
1110
1111                         # If there is an odd number of both bold and italics, it is likely
1112                         # that one of the bold ones was meant to be an apostrophe followed
1113                         # by italics. Which one we cannot know for certain, but it is more
1114                         # likely to be one that has a single-letter word before it.
1115                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1116                         {
1117                                 $i = 0;
1118                                 $firstsingleletterword = -1;
1119                                 $firstmultiletterword = -1;
1120                                 $firstspace = -1;
1121                                 foreach ( $arr as $r )
1122                                 {
1123                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1124                                         {
1125                                                 $x1 = substr ($arr[$i-1], -1);
1126                                                 $x2 = substr ($arr[$i-1], -2, 1);
1127                                                 if ($x1 == ' ') {
1128                                                         if ($firstspace == -1) $firstspace = $i;
1129                                                 } else if ($x2 == ' ') {
1130                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1131                                                 } else {
1132                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1133                                                 }
1134                                         }
1135                                         $i++;
1136                                 }
1137
1138                                 # If there is a single-letter word, use it!
1139                                 if ($firstsingleletterword > -1)
1140                                 {
1141                                         $arr [ $firstsingleletterword ] = "''";
1142                                         $arr [ $firstsingleletterword-1 ] .= "'";
1143                                 }
1144                                 # If not, but there's a multi-letter word, use that one.
1145                                 else if ($firstmultiletterword > -1)
1146                                 {
1147                                         $arr [ $firstmultiletterword ] = "''";
1148                                         $arr [ $firstmultiletterword-1 ] .= "'";
1149                                 }
1150                                 # ... otherwise use the first one that has neither.
1151                                 # (notice that it is possible for all three to be -1 if, for example,
1152                                 # there is only one pentuple-apostrophe in the line)
1153                                 else if ($firstspace > -1)
1154                                 {
1155                                         $arr [ $firstspace ] = "''";
1156                                         $arr [ $firstspace-1 ] .= "'";
1157                                 }
1158                         }
1159
1160                         # Now let's actually convert our apostrophic mush to HTML!
1161                         $output = '';
1162                         $buffer = '';
1163                         $state = '';
1164                         $i = 0;
1165                         foreach ($arr as $r)
1166                         {
1167                                 if (($i % 2) == 0)
1168                                 {
1169                                         if ($state == 'both')
1170                                                 $buffer .= $r;
1171                                         else
1172                                                 $output .= $r;
1173                                 }
1174                                 else
1175                                 {
1176                                         if (strlen ($r) == 2)
1177                                         {
1178                                                 if ($state == 'i')
1179                                                 { $output .= '</i>'; $state = ''; }
1180                                                 else if ($state == 'bi')
1181                                                 { $output .= '</i>'; $state = 'b'; }
1182                                                 else if ($state == 'ib')
1183                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1184                                                 else if ($state == 'both')
1185                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1186                                                 else # $state can be 'b' or ''
1187                                                 { $output .= '<i>'; $state .= 'i'; }
1188                                         }
1189                                         else if (strlen ($r) == 3)
1190                                         {
1191                                                 if ($state == 'b')
1192                                                 { $output .= '</b>'; $state = ''; }
1193                                                 else if ($state == 'bi')
1194                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1195                                                 else if ($state == 'ib')
1196                                                 { $output .= '</b>'; $state = 'i'; }
1197                                                 else if ($state == 'both')
1198                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1199                                                 else # $state can be 'i' or ''
1200                                                 { $output .= '<b>'; $state .= 'b'; }
1201                                         }
1202                                         else if (strlen ($r) == 5)
1203                                         {
1204                                                 if ($state == 'b')
1205                                                 { $output .= '</b><i>'; $state = 'i'; }
1206                                                 else if ($state == 'i')
1207                                                 { $output .= '</i><b>'; $state = 'b'; }
1208                                                 else if ($state == 'bi')
1209                                                 { $output .= '</i></b>'; $state = ''; }
1210                                                 else if ($state == 'ib')
1211                                                 { $output .= '</b></i>'; $state = ''; }
1212                                                 else if ($state == 'both')
1213                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1214                                                 else # ($state == '')
1215                                                 { $buffer = ''; $state = 'both'; }
1216                                         }
1217                                 }
1218                                 $i++;
1219                         }
1220                         # Now close all remaining tags.  Notice that the order is important.
1221                         if ($state == 'b' || $state == 'ib')
1222                                 $output .= '</b>';
1223                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1224                                 $output .= '</i>';
1225                         if ($state == 'bi')
1226                                 $output .= '</b>';
1227                         if ($state == 'both')
1228                                 $output .= '<b><i>'.$buffer.'</i></b>';
1229                         return $output;
1230                 }
1231         }
1232
1233         /**
1234          * Replace external links
1235          *
1236          * Note: this is all very hackish and the order of execution matters a lot.
1237          * Make sure to run maintenance/parserTests.php if you change this code.
1238          *
1239          * @private
1240          */
1241         function replaceExternalLinks( $text ) {
1242                 global $wgContLang;
1243                 $fname = 'Parser::replaceExternalLinks';
1244                 wfProfileIn( $fname );
1245
1246                 $sk =& $this->mOptions->getSkin();
1247
1248                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1249
1250                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1251
1252                 $i = 0;
1253                 while ( $i<count( $bits ) ) {
1254                         $url = $bits[$i++];
1255                         $protocol = $bits[$i++];
1256                         $text = $bits[$i++];
1257                         $trail = $bits[$i++];
1258
1259                         # The characters '<' and '>' (which were escaped by
1260                         # removeHTMLtags()) should not be included in
1261                         # URLs, per RFC 2396.
1262                         $m2 = array();
1263                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1264                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1265                                 $url = substr($url, 0, $m2[0][1]);
1266                         }
1267
1268                         # If the link text is an image URL, replace it with an <img> tag
1269                         # This happened by accident in the original parser, but some people used it extensively
1270                         $img = $this->maybeMakeExternalImage( $text );
1271                         if ( $img !== false ) {
1272                                 $text = $img;
1273                         }
1274
1275                         $dtrail = '';
1276
1277                         # Set linktype for CSS - if URL==text, link is essentially free
1278                         $linktype = ($text == $url) ? 'free' : 'text';
1279
1280                         # No link text, e.g. [http://domain.tld/some.link]
1281                         if ( $text == '' ) {
1282                                 # Autonumber if allowed. See bug #5918
1283                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1284                                         $text = '[' . ++$this->mAutonumber . ']';
1285                                         $linktype = 'autonumber';
1286                                 } else {
1287                                         # Otherwise just use the URL
1288                                         $text = htmlspecialchars( $url );
1289                                         $linktype = 'free';
1290                                 }
1291                         } else {
1292                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1293                                 # Check for trail
1294                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1295                         }
1296
1297                         $text = $wgContLang->markNoConversion($text);
1298
1299                         $url = Sanitizer::cleanUrl( $url );
1300
1301                         # Process the trail (i.e. everything after this link up until start of the next link),
1302                         # replacing any non-bracketed links
1303                         $trail = $this->replaceFreeExternalLinks( $trail );
1304
1305                         # Use the encoded URL
1306                         # This means that users can paste URLs directly into the text
1307                         # Funny characters like &ouml; aren't valid in URLs anyway
1308                         # This was changed in August 2004
1309                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1310
1311                         # Register link in the output object.
1312                         # Replace unnecessary URL escape codes with the referenced character
1313                         # This prevents spammers from hiding links from the filters
1314                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1315                         $this->mOutput->addExternalLink( $pasteurized );
1316                 }
1317
1318                 wfProfileOut( $fname );
1319                 return $s;
1320         }
1321
1322         /**
1323          * Replace anything that looks like a URL with a link
1324          * @private
1325          */
1326         function replaceFreeExternalLinks( $text ) {
1327                 global $wgContLang;
1328                 $fname = 'Parser::replaceFreeExternalLinks';
1329                 wfProfileIn( $fname );
1330
1331                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1332                 $s = array_shift( $bits );
1333                 $i = 0;
1334
1335                 $sk =& $this->mOptions->getSkin();
1336
1337                 while ( $i < count( $bits ) ){
1338                         $protocol = $bits[$i++];
1339                         $remainder = $bits[$i++];
1340
1341                         $m = array();
1342                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1343                                 # Found some characters after the protocol that look promising
1344                                 $url = $protocol . $m[1];
1345                                 $trail = $m[2];
1346
1347                                 # special case: handle urls as url args:
1348                                 # http://www.example.com/foo?=http://www.example.com/bar
1349                                 if(strlen($trail) == 0 &&
1350                                         isset($bits[$i]) &&
1351                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1352                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1353                                 {
1354                                         # add protocol, arg
1355                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1356                                         $i += 2;
1357                                         $trail = $m[2];
1358                                 }
1359
1360                                 # The characters '<' and '>' (which were escaped by
1361                                 # removeHTMLtags()) should not be included in
1362                                 # URLs, per RFC 2396.
1363                                 $m2 = array();
1364                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1365                                         $trail = substr($url, $m2[0][1]) . $trail;
1366                                         $url = substr($url, 0, $m2[0][1]);
1367                                 }
1368
1369                                 # Move trailing punctuation to $trail
1370                                 $sep = ',;\.:!?';
1371                                 # If there is no left bracket, then consider right brackets fair game too
1372                                 if ( strpos( $url, '(' ) === false ) {
1373                                         $sep .= ')';
1374                                 }
1375
1376                                 $numSepChars = strspn( strrev( $url ), $sep );
1377                                 if ( $numSepChars ) {
1378                                         $trail = substr( $url, -$numSepChars ) . $trail;
1379                                         $url = substr( $url, 0, -$numSepChars );
1380                                 }
1381
1382                                 $url = Sanitizer::cleanUrl( $url );
1383
1384                                 # Is this an external image?
1385                                 $text = $this->maybeMakeExternalImage( $url );
1386                                 if ( $text === false ) {
1387                                         # Not an image, make a link
1388                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1389                                         # Register it in the output object...
1390                                         # Replace unnecessary URL escape codes with their equivalent characters
1391                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1392                                         $this->mOutput->addExternalLink( $pasteurized );
1393                                 }
1394                                 $s .= $text . $trail;
1395                         } else {
1396                                 $s .= $protocol . $remainder;
1397                         }
1398                 }
1399                 wfProfileOut( $fname );
1400                 return $s;
1401         }
1402
1403         /**
1404          * Replace unusual URL escape codes with their equivalent characters
1405          * @param string
1406          * @return string
1407          * @static
1408          * @fixme This can merge genuinely required bits in the path or query string,
1409          *        breaking legit URLs. A proper fix would treat the various parts of
1410          *        the URL differently; as a workaround, just use the output for
1411          *        statistical records, not for actual linking/output.
1412          */
1413         static function replaceUnusualEscapes( $url ) {
1414                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1415                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1416         }
1417
1418         /**
1419          * Callback function used in replaceUnusualEscapes().
1420          * Replaces unusual URL escape codes with their equivalent character
1421          * @static
1422          * @private
1423          */
1424         private static function replaceUnusualEscapesCallback( $matches ) {
1425                 $char = urldecode( $matches[0] );
1426                 $ord = ord( $char );
1427                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1428                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1429                         // No, shouldn't be escaped
1430                         return $char;
1431                 } else {
1432                         // Yes, leave it escaped
1433                         return $matches[0];
1434                 }
1435         }
1436
1437         /**
1438          * make an image if it's allowed, either through the global
1439          * option or through the exception
1440          * @private
1441          */
1442         function maybeMakeExternalImage( $url ) {
1443                 $sk =& $this->mOptions->getSkin();
1444                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1445                 $imagesexception = !empty($imagesfrom);
1446                 $text = false;
1447                 if ( $this->mOptions->getAllowExternalImages()
1448                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1449                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1450                                 # Image found
1451                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1452                         }
1453                 }
1454                 return $text;
1455         }
1456
1457         /**
1458          * Process [[ ]] wikilinks
1459          *
1460          * @private
1461          */
1462         function replaceInternalLinks( $s ) {
1463                 global $wgContLang;
1464                 static $fname = 'Parser::replaceInternalLinks' ;
1465
1466                 wfProfileIn( $fname );
1467
1468                 wfProfileIn( $fname.'-setup' );
1469                 static $tc = FALSE;
1470                 # the % is needed to support urlencoded titles as well
1471                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1472
1473                 $sk =& $this->mOptions->getSkin();
1474
1475                 #split the entire text string on occurences of [[
1476                 $a = explode( '[[', ' ' . $s );
1477                 #get the first element (all text up to first [[), and remove the space we added
1478                 $s = array_shift( $a );
1479                 $s = substr( $s, 1 );
1480
1481                 # Match a link having the form [[namespace:link|alternate]]trail
1482                 static $e1 = FALSE;
1483                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1484                 # Match cases where there is no "]]", which might still be images
1485                 static $e1_img = FALSE;
1486                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1487                 # Match the end of a line for a word that's not followed by whitespace,
1488                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1489                 $e2 = wfMsgForContent( 'linkprefix' );
1490
1491                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1492
1493                 if( is_null( $this->mTitle ) ) {
1494                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1495                 }
1496                 $nottalk = !$this->mTitle->isTalkPage();
1497
1498                 if ( $useLinkPrefixExtension ) {
1499                         $m = array();
1500                         if ( preg_match( $e2, $s, $m ) ) {
1501                                 $first_prefix = $m[2];
1502                         } else {
1503                                 $first_prefix = false;
1504                         }
1505                 } else {
1506                         $prefix = '';
1507                 }
1508
1509                 $selflink = $this->mTitle->getPrefixedText();
1510                 $useSubpages = $this->areSubpagesAllowed();
1511                 wfProfileOut( $fname.'-setup' );
1512
1513                 # Loop for each link
1514                 for ($k = 0; isset( $a[$k] ); $k++) {
1515                         $line = $a[$k];
1516                         if ( $useLinkPrefixExtension ) {
1517                                 wfProfileIn( $fname.'-prefixhandling' );
1518                                 if ( preg_match( $e2, $s, $m ) ) {
1519                                         $prefix = $m[2];
1520                                         $s = $m[1];
1521                                 } else {
1522                                         $prefix='';
1523                                 }
1524                                 # first link
1525                                 if($first_prefix) {
1526                                         $prefix = $first_prefix;
1527                                         $first_prefix = false;
1528                                 }
1529                                 wfProfileOut( $fname.'-prefixhandling' );
1530                         }
1531
1532                         $might_be_img = false;
1533
1534                         wfProfileIn( "$fname-e1" );
1535                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1536                                 $text = $m[2];
1537                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1538                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1539                                 # the real problem is with the $e1 regex
1540                                 # See bug 1300.
1541                                 #
1542                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1543                                 # and no image is in sight. See bug 2095.
1544                                 #
1545                                 if( $text !== '' &&
1546                                         substr( $m[3], 0, 1 ) === ']' &&
1547                                         strpos($text, '[') !== false
1548                                 )
1549                                 {
1550                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1551                                         $m[3] = substr( $m[3], 1 );
1552                                 }
1553                                 # fix up urlencoded title texts
1554                                 if( strpos( $m[1], '%' ) !== false ) {
1555                                         # Should anchors '#' also be rejected?
1556                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1557                                 }
1558                                 $trail = $m[3];
1559                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1560                                 $might_be_img = true;
1561                                 $text = $m[2];
1562                                 if ( strpos( $m[1], '%' ) !== false ) {
1563                                        $m[1] = urldecode($m[1]);
1564                                 }
1565                                 $trail = "";
1566                         } else { # Invalid form; output directly
1567                                 $s .= $prefix . '[[' . $line ;
1568                                 wfProfileOut( "$fname-e1" );
1569                                 continue;
1570                         }
1571                         wfProfileOut( "$fname-e1" );
1572                         wfProfileIn( "$fname-misc" );
1573
1574                         # Don't allow internal links to pages containing
1575                         # PROTO: where PROTO is a valid URL protocol; these
1576                         # should be external links.
1577                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1578                                 $s .= $prefix . '[[' . $line ;
1579                                 continue;
1580                         }
1581
1582                         # Make subpage if necessary
1583                         if( $useSubpages ) {
1584                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1585                         } else {
1586                                 $link = $m[1];
1587                         }
1588
1589                         $noforce = (substr($m[1], 0, 1) != ':');
1590                         if (!$noforce) {
1591                                 # Strip off leading ':'
1592                                 $link = substr($link, 1);
1593                         }
1594
1595                         wfProfileOut( "$fname-misc" );
1596                         wfProfileIn( "$fname-title" );
1597                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1598                         if( !$nt ) {
1599                                 $s .= $prefix . '[[' . $line;
1600                                 wfProfileOut( "$fname-title" );
1601                                 continue;
1602                         }
1603
1604                         $ns = $nt->getNamespace();
1605                         $iw = $nt->getInterWiki();
1606                         wfProfileOut( "$fname-title" );
1607
1608                         if ($might_be_img) { # if this is actually an invalid link
1609                                 wfProfileIn( "$fname-might_be_img" );
1610                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1611                                         $found = false;
1612                                         while (isset ($a[$k+1]) ) {
1613                                                 #look at the next 'line' to see if we can close it there
1614                                                 $spliced = array_splice( $a, $k + 1, 1 );
1615                                                 $next_line = array_shift( $spliced );
1616                                                 $m = explode( ']]', $next_line, 3 );
1617                                                 if ( count( $m ) == 3 ) {
1618                                                         # the first ]] closes the inner link, the second the image
1619                                                         $found = true;
1620                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1621                                                         $trail = $m[2];
1622                                                         break;
1623                                                 } elseif ( count( $m ) == 2 ) {
1624                                                         #if there's exactly one ]] that's fine, we'll keep looking
1625                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1626                                                 } else {
1627                                                         #if $next_line is invalid too, we need look no further
1628                                                         $text .= '[[' . $next_line;
1629                                                         break;
1630                                                 }
1631                                         }
1632                                         if ( !$found ) {
1633                                                 # we couldn't find the end of this imageLink, so output it raw
1634                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1635                                                 $text = $this->replaceInternalLinks($text);
1636                                                 $s .= "{$prefix}[[$link|$text";
1637                                                 # note: no $trail, because without an end, there *is* no trail
1638                                                 wfProfileOut( "$fname-might_be_img" );
1639                                                 continue;
1640                                         }
1641                                 } else { #it's not an image, so output it raw
1642                                         $s .= "{$prefix}[[$link|$text";
1643                                         # note: no $trail, because without an end, there *is* no trail
1644                                         wfProfileOut( "$fname-might_be_img" );
1645                                         continue;
1646                                 }
1647                                 wfProfileOut( "$fname-might_be_img" );
1648                         }
1649
1650                         $wasblank = ( '' == $text );
1651                         if( $wasblank ) $text = $link;
1652
1653                         # Link not escaped by : , create the various objects
1654                         if( $noforce ) {
1655
1656                                 # Interwikis
1657                                 wfProfileIn( "$fname-interwiki" );
1658                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1659                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1660                                         $s = rtrim($s . "\n");
1661                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1662                                         wfProfileOut( "$fname-interwiki" );
1663                                         continue;
1664                                 }
1665                                 wfProfileOut( "$fname-interwiki" );
1666
1667                                 if ( $ns == NS_IMAGE ) {
1668                                         wfProfileIn( "$fname-image" );
1669                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1670                                                 # recursively parse links inside the image caption
1671                                                 # actually, this will parse them in any other parameters, too,
1672                                                 # but it might be hard to fix that, and it doesn't matter ATM
1673                                                 $text = $this->replaceExternalLinks($text);
1674                                                 $text = $this->replaceInternalLinks($text);
1675
1676                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1677                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1678                                                 $this->mOutput->addImage( $nt->getDBkey() );
1679
1680                                                 wfProfileOut( "$fname-image" );
1681                                                 continue;
1682                                         } else {
1683                                                 # We still need to record the image's presence on the page
1684                                                 $this->mOutput->addImage( $nt->getDBkey() );
1685                                         }
1686                                         wfProfileOut( "$fname-image" );
1687
1688                                 }
1689
1690                                 if ( $ns == NS_CATEGORY ) {
1691                                         wfProfileIn( "$fname-category" );
1692                                         $s = rtrim($s . "\n"); # bug 87
1693
1694                                         if ( $wasblank ) {
1695                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1696                                                         $sortkey = $this->mTitle->getText();
1697                                                 } else {
1698                                                         $sortkey = $this->mTitle->getPrefixedText();
1699                                                 }
1700                                         } else {
1701                                                 $sortkey = $text;
1702                                         }
1703                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1704                                         $sortkey = str_replace( "\n", '', $sortkey );
1705                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1706                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1707
1708                                         /**
1709                                          * Strip the whitespace Category links produce, see bug 87
1710                                          * @todo We might want to use trim($tmp, "\n") here.
1711                                          */
1712                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1713
1714                                         wfProfileOut( "$fname-category" );
1715                                         continue;
1716                                 }
1717                         }
1718
1719                         if( ( $nt->getPrefixedText() === $selflink ) &&
1720                             ( $nt->getFragment() === '' ) ) {
1721                                 # Self-links are handled specially; generally de-link and change to bold.
1722                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1723                                 continue;
1724                         }
1725
1726                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1727                         if( $ns == NS_MEDIA ) {
1728                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1729                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1730                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1731                                 $this->mOutput->addImage( $nt->getDBkey() );
1732                                 continue;
1733                         } elseif( $ns == NS_SPECIAL ) {
1734                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1735                                 continue;
1736                         } elseif( $ns == NS_IMAGE ) {
1737                                 $img = new Image( $nt );
1738                                 if( $img->exists() ) {
1739                                         // Force a blue link if the file exists; may be a remote
1740                                         // upload on the shared repository, and we want to see its
1741                                         // auto-generated page.
1742                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1743                                         $this->mOutput->addLink( $nt );
1744                                         continue;
1745                                 }
1746                         }
1747                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1748                 }
1749                 wfProfileOut( $fname );
1750                 return $s;
1751         }
1752
1753         /**
1754          * Make a link placeholder. The text returned can be later resolved to a real link with
1755          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1756          * parsing of interwiki links, and secondly to allow all existence checks and
1757          * article length checks (for stub links) to be bundled into a single query.
1758          *
1759          */
1760         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1761                 wfProfileIn( __METHOD__ );
1762                 if ( ! is_object($nt) ) {
1763                         # Fail gracefully
1764                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1765                 } else {
1766                         # Separate the link trail from the rest of the link
1767                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1768
1769                         if ( $nt->isExternal() ) {
1770                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1771                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1772                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1773                         } else {
1774                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1775                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1776                                 $this->mLinkHolders['queries'][] = $query;
1777                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1778                                 $this->mLinkHolders['titles'][] = $nt;
1779
1780                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1781                         }
1782                 }
1783                 wfProfileOut( __METHOD__ );
1784                 return $retVal;
1785         }
1786
1787         /**
1788          * Render a forced-blue link inline; protect against double expansion of
1789          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1790          * Since this little disaster has to split off the trail text to avoid
1791          * breaking URLs in the following text without breaking trails on the
1792          * wiki links, it's been made into a horrible function.
1793          *
1794          * @param Title $nt
1795          * @param string $text
1796          * @param string $query
1797          * @param string $trail
1798          * @param string $prefix
1799          * @return string HTML-wikitext mix oh yuck
1800          */
1801         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1802                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1803                 $sk =& $this->mOptions->getSkin();
1804                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1805                 return $this->armorLinks( $link ) . $trail;
1806         }
1807
1808         /**
1809          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1810          * going to go through further parsing steps before inline URL expansion.
1811          *
1812          * In particular this is important when using action=render, which causes
1813          * full URLs to be included.
1814          *
1815          * Oh man I hate our multi-layer parser!
1816          *
1817          * @param string more-or-less HTML
1818          * @return string less-or-more HTML with NOPARSE bits
1819          */
1820         function armorLinks( $text ) {
1821                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1822                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1823         }
1824
1825         /**
1826          * Return true if subpage links should be expanded on this page.
1827          * @return bool
1828          */
1829         function areSubpagesAllowed() {
1830                 # Some namespaces don't allow subpages
1831                 global $wgNamespacesWithSubpages;
1832                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1833         }
1834
1835         /**
1836          * Handle link to subpage if necessary
1837          * @param string $target the source of the link
1838          * @param string &$text the link text, modified as necessary
1839          * @return string the full name of the link
1840          * @private
1841          */
1842         function maybeDoSubpageLink($target, &$text) {
1843                 # Valid link forms:
1844                 # Foobar -- normal
1845                 # :Foobar -- override special treatment of prefix (images, language links)
1846                 # /Foobar -- convert to CurrentPage/Foobar
1847                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1848                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1849                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1850
1851                 $fname = 'Parser::maybeDoSubpageLink';
1852                 wfProfileIn( $fname );
1853                 $ret = $target; # default return value is no change
1854
1855                 # bug 7425
1856                 $target = trim( $target );
1857
1858                 # Some namespaces don't allow subpages,
1859                 # so only perform processing if subpages are allowed
1860                 if( $this->areSubpagesAllowed() ) {
1861                         # Look at the first character
1862                         if( $target != '' && $target{0} == '/' ) {
1863                                 # / at end means we don't want the slash to be shown
1864                                 if( substr( $target, -1, 1 ) == '/' ) {
1865                                         $target = substr( $target, 1, -1 );
1866                                         $noslash = $target;
1867                                 } else {
1868                                         $noslash = substr( $target, 1 );
1869                                 }
1870
1871                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1872                                 if( '' === $text ) {
1873                                         $text = $target;
1874                                 } # this might be changed for ugliness reasons
1875                         } else {
1876                                 # check for .. subpage backlinks
1877                                 $dotdotcount = 0;
1878                                 $nodotdot = $target;
1879                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1880                                         ++$dotdotcount;
1881                                         $nodotdot = substr( $nodotdot, 3 );
1882                                 }
1883                                 if($dotdotcount > 0) {
1884                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1885                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1886                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1887                                                 # / at the end means don't show full path
1888                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1889                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1890                                                         if( '' === $text ) {
1891                                                                 $text = $nodotdot;
1892                                                         }
1893                                                 }
1894                                                 $nodotdot = trim( $nodotdot );
1895                                                 if( $nodotdot != '' ) {
1896                                                         $ret .= '/' . $nodotdot;
1897                                                 }
1898                                         }
1899                                 }
1900                         }
1901                 }
1902
1903                 wfProfileOut( $fname );
1904                 return $ret;
1905         }
1906
1907         /**#@+
1908          * Used by doBlockLevels()
1909          * @private
1910          */
1911         /* private */ function closeParagraph() {
1912                 $result = '';
1913                 if ( '' != $this->mLastSection ) {
1914                         $result = '</' . $this->mLastSection  . ">\n";
1915                 }
1916                 $this->mInPre = false;
1917                 $this->mLastSection = '';
1918                 return $result;
1919         }
1920         # getCommon() returns the length of the longest common substring
1921         # of both arguments, starting at the beginning of both.
1922         #
1923         /* private */ function getCommon( $st1, $st2 ) {
1924                 $fl = strlen( $st1 );
1925                 $shorter = strlen( $st2 );
1926                 if ( $fl < $shorter ) { $shorter = $fl; }
1927
1928                 for ( $i = 0; $i < $shorter; ++$i ) {
1929                         if ( $st1{$i} != $st2{$i} ) { break; }
1930                 }
1931                 return $i;
1932         }
1933         # These next three functions open, continue, and close the list
1934         # element appropriate to the prefix character passed into them.
1935         #
1936         /* private */ function openList( $char ) {
1937                 $result = $this->closeParagraph();
1938
1939                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1940                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1941                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1942                 else if ( ';' == $char ) {
1943                         $result .= '<dl><dt>';
1944                         $this->mDTopen = true;
1945                 }
1946                 else { $result = '<!-- ERR 1 -->'; }
1947
1948                 return $result;
1949         }
1950
1951         /* private */ function nextItem( $char ) {
1952                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1953                 else if ( ':' == $char || ';' == $char ) {
1954                         $close = '</dd>';
1955                         if ( $this->mDTopen ) { $close = '</dt>'; }
1956                         if ( ';' == $char ) {
1957                                 $this->mDTopen = true;
1958                                 return $close . '<dt>';
1959                         } else {
1960                                 $this->mDTopen = false;
1961                                 return $close . '<dd>';
1962                         }
1963                 }
1964                 return '<!-- ERR 2 -->';
1965         }
1966
1967         /* private */ function closeList( $char ) {
1968                 if ( '*' == $char ) { $text = '</li></ul>'; }
1969                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1970                 else if ( ':' == $char ) {
1971                         if ( $this->mDTopen ) {
1972                                 $this->mDTopen = false;
1973                                 $text = '</dt></dl>';
1974                         } else {
1975                                 $text = '</dd></dl>';
1976                         }
1977                 }
1978                 else {  return '<!-- ERR 3 -->'; }
1979                 return $text."\n";
1980         }
1981         /**#@-*/
1982
1983         /**
1984          * Make lists from lines starting with ':', '*', '#', etc.
1985          *
1986          * @private
1987          * @return string the lists rendered as HTML
1988          */
1989         function doBlockLevels( $text, $linestart ) {
1990                 $fname = 'Parser::doBlockLevels';
1991                 wfProfileIn( $fname );
1992
1993                 # Parsing through the text line by line.  The main thing
1994                 # happening here is handling of block-level elements p, pre,
1995                 # and making lists from lines starting with * # : etc.
1996                 #
1997                 $textLines = explode( "\n", $text );
1998
1999                 $lastPrefix = $output = '';
2000                 $this->mDTopen = $inBlockElem = false;
2001                 $prefixLength = 0;
2002                 $paragraphStack = false;
2003
2004                 if ( !$linestart ) {
2005                         $output .= array_shift( $textLines );
2006                 }
2007                 foreach ( $textLines as $oLine ) {
2008                         $lastPrefixLength = strlen( $lastPrefix );
2009                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2010                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2011                         if ( !$this->mInPre ) {
2012                                 # Multiple prefixes may abut each other for nested lists.
2013                                 $prefixLength = strspn( $oLine, '*#:;' );
2014                                 $pref = substr( $oLine, 0, $prefixLength );
2015
2016                                 # eh?
2017                                 $pref2 = str_replace( ';', ':', $pref );
2018                                 $t = substr( $oLine, $prefixLength );
2019                                 $this->mInPre = !empty($preOpenMatch);
2020                         } else {
2021                                 # Don't interpret any other prefixes in preformatted text
2022                                 $prefixLength = 0;
2023                                 $pref = $pref2 = '';
2024                                 $t = $oLine;
2025                         }
2026
2027                         # List generation
2028                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2029                                 # Same as the last item, so no need to deal with nesting or opening stuff
2030                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2031                                 $paragraphStack = false;
2032
2033                                 if ( substr( $pref, -1 ) == ';') {
2034                                         # The one nasty exception: definition lists work like this:
2035                                         # ; title : definition text
2036                                         # So we check for : in the remainder text to split up the
2037                                         # title and definition, without b0rking links.
2038                                         $term = $t2 = '';
2039                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2040                                                 $t = $t2;
2041                                                 $output .= $term . $this->nextItem( ':' );
2042                                         }
2043                                 }
2044                         } elseif( $prefixLength || $lastPrefixLength ) {
2045                                 # Either open or close a level...
2046                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2047                                 $paragraphStack = false;
2048
2049                                 while( $commonPrefixLength < $lastPrefixLength ) {
2050                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2051                                         --$lastPrefixLength;
2052                                 }
2053                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2054                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2055                                 }
2056                                 while ( $prefixLength > $commonPrefixLength ) {
2057                                         $char = substr( $pref, $commonPrefixLength, 1 );
2058                                         $output .= $this->openList( $char );
2059
2060                                         if ( ';' == $char ) {
2061                                                 # FIXME: This is dupe of code above
2062                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2063                                                         $t = $t2;
2064                                                         $output .= $term . $this->nextItem( ':' );
2065                                                 }
2066                                         }
2067                                         ++$commonPrefixLength;
2068                                 }
2069                                 $lastPrefix = $pref2;
2070                         }
2071                         if( 0 == $prefixLength ) {
2072                                 wfProfileIn( "$fname-paragraph" );
2073                                 # No prefix (not in list)--go to paragraph mode
2074                                 // XXX: use a stack for nestable elements like span, table and div
2075                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
2076                                 $closematch = preg_match(
2077                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2078                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
2079                                 if ( $openmatch or $closematch ) {
2080                                         $paragraphStack = false;
2081                                         # TODO bug 5718: paragraph closed
2082                                         $output .= $this->closeParagraph();
2083                                         if ( $preOpenMatch and !$preCloseMatch ) {
2084                                                 $this->mInPre = true;
2085                                         }
2086                                         if ( $closematch ) {
2087                                                 $inBlockElem = false;
2088                                         } else {
2089                                                 $inBlockElem = true;
2090                                         }
2091                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2092                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2093                                                 // pre
2094                                                 if ($this->mLastSection != 'pre') {
2095                                                         $paragraphStack = false;
2096                                                         $output .= $this->closeParagraph().'<pre>';
2097                                                         $this->mLastSection = 'pre';
2098                                                 }
2099                                                 $t = substr( $t, 1 );
2100                                         } else {
2101                                                 // paragraph
2102                                                 if ( '' == trim($t) ) {
2103                                                         if ( $paragraphStack ) {
2104                                                                 $output .= $paragraphStack.'<br />';
2105                                                                 $paragraphStack = false;
2106                                                                 $this->mLastSection = 'p';
2107                                                         } else {
2108                                                                 if ($this->mLastSection != 'p' ) {
2109                                                                         $output .= $this->closeParagraph();
2110                                                                         $this->mLastSection = '';
2111                                                                         $paragraphStack = '<p>';
2112                                                                 } else {
2113                                                                         $paragraphStack = '</p><p>';
2114                                                                 }
2115                                                         }
2116                                                 } else {
2117                                                         if ( $paragraphStack ) {
2118                                                                 $output .= $paragraphStack;
2119                                                                 $paragraphStack = false;
2120                                                                 $this->mLastSection = 'p';
2121                                                         } else if ($this->mLastSection != 'p') {
2122                                                                 $output .= $this->closeParagraph().'<p>';
2123                                                                 $this->mLastSection = 'p';
2124                                                         }
2125                                                 }
2126                                         }
2127                                 }
2128                                 wfProfileOut( "$fname-paragraph" );
2129                         }
2130                         // somewhere above we forget to get out of pre block (bug 785)
2131                         if($preCloseMatch && $this->mInPre) {
2132                                 $this->mInPre = false;
2133                         }
2134                         if ($paragraphStack === false) {
2135                                 $output .= $t."\n";
2136                         }
2137                 }
2138                 while ( $prefixLength ) {
2139                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2140                         --$prefixLength;
2141                 }
2142                 if ( '' != $this->mLastSection ) {
2143                         $output .= '</' . $this->mLastSection . '>';
2144                         $this->mLastSection = '';
2145                 }
2146
2147                 wfProfileOut( $fname );
2148                 return $output;
2149         }
2150
2151         /**
2152          * Split up a string on ':', ignoring any occurences inside tags
2153          * to prevent illegal overlapping.
2154          * @param string $str the string to split
2155          * @param string &$before set to everything before the ':'
2156          * @param string &$after set to everything after the ':'
2157          * return string the position of the ':', or false if none found
2158          */
2159         function findColonNoLinks($str, &$before, &$after) {
2160                 $fname = 'Parser::findColonNoLinks';
2161                 wfProfileIn( $fname );
2162
2163                 $pos = strpos( $str, ':' );
2164                 if( $pos === false ) {
2165                         // Nothing to find!
2166                         wfProfileOut( $fname );
2167                         return false;
2168                 }
2169
2170                 $lt = strpos( $str, '<' );
2171                 if( $lt === false || $lt > $pos ) {
2172                         // Easy; no tag nesting to worry about
2173                         $before = substr( $str, 0, $pos );
2174                         $after = substr( $str, $pos+1 );
2175                         wfProfileOut( $fname );
2176                         return $pos;
2177                 }
2178
2179                 // Ugly state machine to walk through avoiding tags.
2180                 $state = MW_COLON_STATE_TEXT;
2181                 $stack = 0;
2182                 $len = strlen( $str );
2183                 for( $i = 0; $i < $len; $i++ ) {
2184                         $c = $str{$i};
2185
2186                         switch( $state ) {
2187                         // (Using the number is a performance hack for common cases)
2188                         case 0: // MW_COLON_STATE_TEXT:
2189                                 switch( $c ) {
2190                                 case "<":
2191                                         // Could be either a <start> tag or an </end> tag
2192                                         $state = MW_COLON_STATE_TAGSTART;
2193                                         break;
2194                                 case ":":
2195                                         if( $stack == 0 ) {
2196                                                 // We found it!
2197                                                 $before = substr( $str, 0, $i );
2198                                                 $after = substr( $str, $i + 1 );
2199                                                 wfProfileOut( $fname );
2200                                                 return $i;
2201                                         }
2202                                         // Embedded in a tag; don't break it.
2203                                         break;
2204                                 default:
2205                                         // Skip ahead looking for something interesting
2206                                         $colon = strpos( $str, ':', $i );
2207                                         if( $colon === false ) {
2208                                                 // Nothing else interesting
2209                                                 wfProfileOut( $fname );
2210                                                 return false;
2211                                         }
2212                                         $lt = strpos( $str, '<', $i );
2213                                         if( $stack === 0 ) {
2214                                                 if( $lt === false || $colon < $lt ) {
2215                                                         // We found it!
2216                                                         $before = substr( $str, 0, $colon );
2217                                                         $after = substr( $str, $colon + 1 );
2218                                                         wfProfileOut( $fname );
2219                                                         return $i;
2220                                                 }
2221                                         }
2222                                         if( $lt === false ) {
2223                                                 // Nothing else interesting to find; abort!
2224                                                 // We're nested, but there's no close tags left. Abort!
2225                                                 break 2;
2226                                         }
2227                                         // Skip ahead to next tag start
2228                                         $i = $lt;
2229                                         $state = MW_COLON_STATE_TAGSTART;
2230                                 }
2231                                 break;
2232                         case 1: // MW_COLON_STATE_TAG:
2233                                 // In a <tag>
2234                                 switch( $c ) {
2235                                 case ">":
2236                                         $stack++;
2237                                         $state = MW_COLON_STATE_TEXT;
2238                                         break;
2239                                 case "/":
2240                                         // Slash may be followed by >?
2241                                         $state = MW_COLON_STATE_TAGSLASH;
2242                                         break;
2243                                 default:
2244                                         // ignore
2245                                 }
2246                                 break;
2247                         case 2: // MW_COLON_STATE_TAGSTART:
2248                                 switch( $c ) {
2249                                 case "/":
2250                                         $state = MW_COLON_STATE_CLOSETAG;
2251                                         break;
2252                                 case "!":
2253                                         $state = MW_COLON_STATE_COMMENT;
2254                                         break;
2255                                 case ">":
2256                                         // Illegal early close? This shouldn't happen D:
2257                                         $state = MW_COLON_STATE_TEXT;
2258                                         break;
2259                                 default:
2260                                         $state = MW_COLON_STATE_TAG;
2261                                 }
2262                                 break;
2263                         case 3: // MW_COLON_STATE_CLOSETAG:
2264                                 // In a </tag>
2265                                 if( $c == ">" ) {
2266                                         $stack--;
2267                                         if( $stack < 0 ) {
2268                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2269                                                 wfProfileOut( $fname );
2270                                                 return false;
2271                                         }
2272                                         $state = MW_COLON_STATE_TEXT;
2273                                 }
2274                                 break;
2275                         case MW_COLON_STATE_TAGSLASH:
2276                                 if( $c == ">" ) {
2277                                         // Yes, a self-closed tag <blah/>
2278                                         $state = MW_COLON_STATE_TEXT;
2279                                 } else {
2280                                         // Probably we're jumping the gun, and this is an attribute
2281                                         $state = MW_COLON_STATE_TAG;
2282                                 }
2283                                 break;
2284                         case 5: // MW_COLON_STATE_COMMENT:
2285                                 if( $c == "-" ) {
2286                                         $state = MW_COLON_STATE_COMMENTDASH;
2287                                 }
2288                                 break;
2289                         case MW_COLON_STATE_COMMENTDASH:
2290                                 if( $c == "-" ) {
2291                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2292                                 } else {
2293                                         $state = MW_COLON_STATE_COMMENT;
2294                                 }
2295                                 break;
2296                         case MW_COLON_STATE_COMMENTDASHDASH:
2297                                 if( $c == ">" ) {
2298                                         $state = MW_COLON_STATE_TEXT;
2299                                 } else {
2300                                         $state = MW_COLON_STATE_COMMENT;
2301                                 }
2302                                 break;
2303                         default:
2304                                 throw new MWException( "State machine error in $fname" );
2305                         }
2306                 }
2307                 if( $stack > 0 ) {
2308                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2309                         return false;
2310                 }
2311                 wfProfileOut( $fname );
2312                 return false;
2313         }
2314
2315         /**
2316          * Return value of a magic variable (like PAGENAME)
2317          *
2318          * @private
2319          */
2320         function getVariableValue( $index ) {
2321                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2322
2323                 /**
2324                  * Some of these require message or data lookups and can be
2325                  * expensive to check many times.
2326                  */
2327                 static $varCache = array();
2328                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2329                         if ( isset( $varCache[$index] ) ) {
2330                                 return $varCache[$index];
2331                         }
2332                 }
2333
2334                 $ts = time();
2335                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2336
2337                 # Use the time zone
2338                 global $wgLocaltimezone;
2339                 if ( isset( $wgLocaltimezone ) ) {
2340                         $oldtz = getenv( 'TZ' );
2341                         putenv( 'TZ='.$wgLocaltimezone );
2342                 }
2343                 $localTimestamp = date( 'YmdHis', $ts );
2344                 $localMonth = date( 'm', $ts );
2345                 $localMonthName = date( 'n', $ts );
2346                 $localDay = date( 'j', $ts );
2347                 $localDay2 = date( 'd', $ts );
2348                 $localDayOfWeek = date( 'w', $ts );
2349                 $localWeek = date( 'W', $ts );
2350                 $localYear = date( 'Y', $ts );
2351                 $localHour = date( 'H', $ts );
2352                 if ( isset( $wgLocaltimezone ) ) {
2353                         putenv( 'TZ='.$oldtz );
2354                 }
2355
2356                 switch ( $index ) {
2357                         case 'currentmonth':
2358                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2359                         case 'currentmonthname':
2360                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2361                         case 'currentmonthnamegen':
2362                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2363                         case 'currentmonthabbrev':
2364                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2365                         case 'currentday':
2366                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2367                         case 'currentday2':
2368                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2369                         case 'localmonth':
2370                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2371                         case 'localmonthname':
2372                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2373                         case 'localmonthnamegen':
2374                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2375                         case 'localmonthabbrev':
2376                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2377                         case 'localday':
2378                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2379                         case 'localday2':
2380                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2381                         case 'pagename':
2382                                 return $this->mTitle->getText();
2383                         case 'pagenamee':
2384                                 return $this->mTitle->getPartialURL();
2385                         case 'fullpagename':
2386                                 return $this->mTitle->getPrefixedText();
2387                         case 'fullpagenamee':
2388                                 return $this->mTitle->getPrefixedURL();
2389                         case 'subpagename':
2390                                 return $this->mTitle->getSubpageText();
2391                         case 'subpagenamee':
2392                                 return $this->mTitle->getSubpageUrlForm();
2393                         case 'basepagename':
2394                                 return $this->mTitle->getBaseText();
2395                         case 'basepagenamee':
2396                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2397                         case 'talkpagename':
2398                                 if( $this->mTitle->canTalk() ) {
2399                                         $talkPage = $this->mTitle->getTalkPage();
2400                                         return $talkPage->getPrefixedText();
2401                                 } else {
2402                                         return '';
2403                                 }
2404                         case 'talkpagenamee':
2405                                 if( $this->mTitle->canTalk() ) {
2406                                         $talkPage = $this->mTitle->getTalkPage();
2407                                         return $talkPage->getPrefixedUrl();
2408                                 } else {
2409                                         return '';
2410                                 }
2411                         case 'subjectpagename':
2412                                 $subjPage = $this->mTitle->getSubjectPage();
2413                                 return $subjPage->getPrefixedText();
2414                         case 'subjectpagenamee':
2415                                 $subjPage = $this->mTitle->getSubjectPage();
2416                                 return $subjPage->getPrefixedUrl();
2417                         case 'revisionid':
2418                                 return $this->mRevisionId;
2419                         case 'revisionday':
2420                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2421                         case 'revisionday2':
2422                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2423                         case 'revisionmonth':
2424                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2425                         case 'revisionyear':
2426                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2427                         case 'revisiontimestamp':
2428                                 return $this->getRevisionTimestamp();
2429                         case 'namespace':
2430                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2431                         case 'namespacee':
2432                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2433                         case 'talkspace':
2434                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2435                         case 'talkspacee':
2436                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2437                         case 'subjectspace':
2438                                 return $this->mTitle->getSubjectNsText();
2439                         case 'subjectspacee':
2440                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2441                         case 'currentdayname':
2442                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2443                         case 'currentyear':
2444                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2445                         case 'currenttime':
2446                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2447                         case 'currenthour':
2448                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2449                         case 'currentweek':
2450                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2451                                 // int to remove the padding
2452                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2453                         case 'currentdow':
2454                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2455                         case 'localdayname':
2456                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2457                         case 'localyear':
2458                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2459                         case 'localtime':
2460                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2461                         case 'localhour':
2462                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2463                         case 'localweek':
2464                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2465                                 // int to remove the padding
2466                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2467                         case 'localdow':
2468                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2469                         case 'numberofarticles':
2470                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2471                         case 'numberoffiles':
2472                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2473                         case 'numberofusers':
2474                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2475                         case 'numberofpages':
2476                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2477                         case 'numberofadmins':
2478                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2479                         case 'currenttimestamp':
2480                                 return $varCache[$index] = wfTimestampNow();
2481                         case 'localtimestamp':
2482                                 return $varCache[$index] = $localTimestamp;
2483                         case 'currentversion':
2484                                 return $varCache[$index] = SpecialVersion::getVersion();
2485                         case 'sitename':
2486                                 return $wgSitename;
2487                         case 'server':
2488                                 return $wgServer;
2489                         case 'servername':
2490                                 return $wgServerName;
2491                         case 'scriptpath':
2492                                 return $wgScriptPath;
2493                         case 'directionmark':
2494                                 return $wgContLang->getDirMark();
2495                         case 'contentlanguage':
2496                                 global $wgContLanguageCode;
2497                                 return $wgContLanguageCode;
2498                         default:
2499                                 $ret = null;
2500                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2501                                         return $ret;
2502                                 else
2503                                         return null;
2504                 }
2505         }
2506
2507         /**
2508          * initialise the magic variables (like CURRENTMONTHNAME)
2509          *
2510          * @private
2511          */
2512         function initialiseVariables() {
2513                 $fname = 'Parser::initialiseVariables';
2514                 wfProfileIn( $fname );
2515                 $variableIDs = MagicWord::getVariableIDs();
2516
2517                 $this->mVariables = array();
2518                 foreach ( $variableIDs as $id ) {
2519                         $mw =& MagicWord::get( $id );
2520                         $mw->addToArray( $this->mVariables, $id );
2521                 }
2522                 wfProfileOut( $fname );
2523         }
2524
2525         /**
2526          * parse any parentheses in format ((title|part|part))
2527          * and call callbacks to get a replacement text for any found piece
2528          *
2529          * @param string $text The text to parse
2530          * @param array $callbacks rules in form:
2531          *     '{' => array(                            # opening parentheses
2532          *                                      'end' => '}',   # closing parentheses
2533          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2534          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2535          *                                                                )
2536          *                                      )
2537          *                                      'min' => 2,     # Minimum parenthesis count in cb
2538          *                                      'max' => 3,     # Maximum parenthesis count in cb
2539          * @private
2540          */
2541         function replace_callback ($text, $callbacks) {
2542                 wfProfileIn( __METHOD__ );
2543                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2544                 $lastOpeningBrace = -1;                 # last not closed parentheses
2545
2546                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2547
2548                 $i = 0;
2549                 while ( $i < strlen( $text ) ) {
2550                         # Find next opening brace, closing brace or pipe
2551                         if ( $lastOpeningBrace == -1 ) {
2552                                 $currentClosing = '';
2553                                 $search = $validOpeningBraces;
2554                         } else {
2555                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2556                                 $search = $validOpeningBraces . '|' . $currentClosing;
2557                         }
2558                         $rule = null;
2559                         $i += strcspn( $text, $search, $i );
2560                         if ( $i < strlen( $text ) ) {
2561                                 if ( $text[$i] == '|' ) {
2562                                         $found = 'pipe';
2563                                 } elseif ( $text[$i] == $currentClosing ) {
2564                                         $found = 'close';
2565                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2566                                         $found = 'open';
2567                                         $rule = $callbacks[$text[$i]];
2568                                 } else {
2569                                         # Some versions of PHP have a strcspn which stops on null characters
2570                                         # Ignore and continue
2571                                         ++$i;
2572                                         continue;
2573                                 }
2574                         } else {
2575                                 # All done
2576                                 break;
2577                         }
2578
2579                         if ( $found == 'open' ) {
2580                                 # found opening brace, let's add it to parentheses stack
2581                                 $piece = array('brace' => $text[$i],
2582                                                            'braceEnd' => $rule['end'],
2583                                                            'title' => '',
2584                                                            'parts' => null);
2585
2586                                 # count opening brace characters
2587                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2588                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2589                                 $i += $piece['count'];
2590
2591                                 # we need to add to stack only if opening brace count is enough for one of the rules
2592                                 if ( $piece['count'] >= $rule['min'] ) {
2593                                         $lastOpeningBrace ++;
2594                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2595                                 }
2596                         } elseif ( $found == 'close' ) {
2597                                 # lets check if it is enough characters for closing brace
2598                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2599                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2600
2601                                 # check for maximum matching characters (if there are 5 closing
2602                                 # characters, we will probably need only 3 - depending on the rules)
2603                                 $matchingCount = 0;
2604                                 $matchingCallback = null;
2605                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2606                                 if ( $count > $cbType['max'] ) {
2607                                         # The specified maximum exists in the callback array, unless the caller
2608                                         # has made an error
2609                                         $matchingCount = $cbType['max'];
2610                                 } else {
2611                                         # Count is less than the maximum
2612                                         # Skip any gaps in the callback array to find the true largest match
2613                                         # Need to use array_key_exists not isset because the callback can be null
2614                                         $matchingCount = $count;
2615                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2616                                                 --$matchingCount;
2617                                         }
2618                                 }
2619
2620                                 if ($matchingCount <= 0) {
2621                                         $i += $count;
2622                                         continue;
2623                                 }
2624                                 $matchingCallback = $cbType['cb'][$matchingCount];
2625
2626                                 # let's set a title or last part (if '|' was found)
2627                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2628                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2629                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2630                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2631                                 } else {
2632                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2633                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2634                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2635                                 }
2636
2637                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2638                                 $pieceEnd = $i + $matchingCount;
2639
2640                                 if( is_callable( $matchingCallback ) ) {
2641                                         $cbArgs = array (
2642                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2643                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2644                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2645                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2646                                                                          );
2647                                         # finally we can call a user callback and replace piece of text
2648                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2649                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2650                                         $i = $pieceStart + strlen($replaceWith);
2651                                 } else {
2652                                         # null value for callback means that parentheses should be parsed, but not replaced
2653                                         $i += $matchingCount;
2654                                 }
2655
2656                                 # reset last opening parentheses, but keep it in case there are unused characters
2657                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2658                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2659                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2660                                                            'title' => '',
2661                                                            'parts' => null,
2662                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2663                                 $openingBraceStack[$lastOpeningBrace--] = null;
2664
2665                                 if ($matchingCount < $piece['count']) {
2666                                         $piece['count'] -= $matchingCount;
2667                                         $piece['startAt'] -= $matchingCount;
2668                                         $piece['partStart'] = $piece['startAt'];
2669                                         # do we still qualify for any callback with remaining count?
2670                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2671                                         while ( $piece['count'] ) {
2672                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2673                                                         $lastOpeningBrace++;
2674                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2675                                                         break;
2676                                                 }
2677                                                 --$piece['count'];
2678                                         }
2679                                 }
2680                         } elseif ( $found == 'pipe' ) {
2681                                 # lets set a title if it is a first separator, or next part otherwise
2682                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2683                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2684                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2685                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2686                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2687                                 } else {
2688                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2689                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2690                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2691                                 }
2692                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2693                         }
2694                 }
2695
2696                 wfProfileOut( __METHOD__ );
2697                 return $text;
2698         }
2699
2700         /**
2701          * Replace magic variables, templates, and template arguments
2702          * with the appropriate text. Templates are substituted recursively,
2703          * taking care to avoid infinite loops.
2704          *
2705          * Note that the substitution depends on value of $mOutputType:
2706          *  OT_WIKI: only {{subst:}} templates
2707          *  OT_MSG: only magic variables
2708          *  OT_HTML: all templates and magic variables
2709          *
2710          * @param string $tex The text to transform
2711          * @param array $args Key-value pairs representing template parameters to substitute
2712          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2713          * @private
2714          */
2715         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2716                 # Prevent too big inclusions
2717                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2718                         return $text;
2719                 }
2720
2721                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2722                 wfProfileIn( $fname );
2723
2724                 # This function is called recursively. To keep track of arguments we need a stack:
2725                 array_push( $this->mArgStack, $args );
2726
2727                 $braceCallbacks = array();
2728                 if ( !$argsOnly ) {
2729                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2730                 }
2731                 if ( $this->mOutputType != OT_MSG ) {
2732                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2733                 }
2734                 if ( $braceCallbacks ) {
2735                         $callbacks = array(
2736                                 '{' => array(
2737                                         'end' => '}',
2738                                         'cb' => $braceCallbacks,
2739                                         'min' => $argsOnly ? 3 : 2,
2740                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2741                                 ),
2742                                 '[' => array(
2743                                         'end' => ']',
2744                                         'cb' => array(2=>null),
2745                                         'min' => 2,
2746                                         'max' => 2,
2747                                 )
2748                         );
2749                         $text = $this->replace_callback ($text, $callbacks);
2750
2751                         array_pop( $this->mArgStack );
2752                 }
2753                 wfProfileOut( $fname );
2754                 return $text;
2755         }
2756
2757         /**
2758          * Replace magic variables
2759          * @private
2760          */
2761         function variableSubstitution( $matches ) {
2762                 global $wgContLang;
2763                 $fname = 'Parser::variableSubstitution';
2764                 $varname = $wgContLang->lc($matches[1]);
2765                 wfProfileIn( $fname );
2766                 $skip = false;
2767                 if ( $this->mOutputType == OT_WIKI ) {
2768                         # Do only magic variables prefixed by SUBST
2769                         $mwSubst =& MagicWord::get( 'subst' );
2770                         if (!$mwSubst->matchStartAndRemove( $varname ))
2771                                 $skip = true;
2772                         # Note that if we don't substitute the variable below,
2773                         # we don't remove the {{subst:}} magic word, in case
2774                         # it is a template rather than a magic variable.
2775                 }
2776                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2777                         $id = $this->mVariables[$varname];
2778                         # Now check if we did really match, case sensitive or not
2779                         $mw =& MagicWord::get( $id );
2780                         if ($mw->match($matches[1])) {
2781                                 $text = $this->getVariableValue( $id );
2782                                 $this->mOutput->mContainsOldMagic = true;
2783                         } else {
2784                                 $text = $matches[0];
2785                         }
2786                 } else {
2787                         $text = $matches[0];
2788                 }
2789                 wfProfileOut( $fname );
2790                 return $text;
2791         }
2792
2793         /**
2794          * Return the text of a template, after recursively
2795          * replacing any variables or templates within the template.
2796          *
2797          * @param array $piece The parts of the template
2798          *  $piece['text']: matched text
2799          *  $piece['title']: the title, i.e. the part before the |
2800          *  $piece['parts']: the parameter array
2801          * @return string the text of the template
2802          * @private
2803          */
2804         function braceSubstitution( $piece ) {
2805                 global $wgContLang, $wgLang, $wgAllowDisplayTitle;
2806                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2807                 wfProfileIn( $fname );
2808                 wfProfileIn( __METHOD__.'-setup' );
2809
2810                 # Flags
2811                 $found = false;             # $text has been filled
2812                 $nowiki = false;            # wiki markup in $text should be escaped
2813                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2814                 $noargs = false;            # Don't replace triple-brace arguments in $text
2815                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2816                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2817                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2818
2819                 # Title object, where $text came from
2820                 $title = NULL;
2821
2822                 $linestart = '';
2823
2824
2825                 # $part1 is the bit before the first |, and must contain only title characters
2826                 # $args is a list of arguments, starting from index 0, not including $part1
2827
2828                 $titleText = $part1 = $piece['title'];
2829                 # If the third subpattern matched anything, it will start with |
2830
2831                 if (null == $piece['parts']) {
2832                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2833                         if ($replaceWith != $piece['text']) {
2834                                 $text = $replaceWith;
2835                                 $found = true;
2836                                 $noparse = true;
2837                                 $noargs = true;
2838                         }
2839                 }
2840
2841                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2842                 wfProfileOut( __METHOD__.'-setup' );
2843
2844                 # SUBST
2845                 wfProfileIn( __METHOD__.'-modifiers' );
2846                 if ( !$found ) {
2847                         $mwSubst =& MagicWord::get( 'subst' );
2848                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2849                                 # One of two possibilities is true:
2850                                 # 1) Found SUBST but not in the PST phase
2851                                 # 2) Didn't find SUBST and in the PST phase
2852                                 # In either case, return without further processing
2853                                 $text = $piece['text'];
2854                                 $found = true;
2855                                 $noparse = true;
2856                                 $noargs = true;
2857                         }
2858                 }
2859
2860                 # MSG, MSGNW and RAW
2861                 if ( !$found ) {
2862                         # Check for MSGNW:
2863                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2864                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2865                                 $nowiki = true;
2866                         } else {
2867                                 # Remove obsolete MSG:
2868                                 $mwMsg =& MagicWord::get( 'msg' );
2869                                 $mwMsg->matchStartAndRemove( $part1 );
2870                         }
2871
2872                         # Check for RAW:
2873                         $mwRaw =& MagicWord::get( 'raw' );
2874                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2875                                 $forceRawInterwiki = true;
2876                         }
2877                 }
2878                 wfProfileOut( __METHOD__.'-modifiers' );
2879
2880                 # Parser functions
2881                 if ( !$found ) {
2882                         wfProfileIn( __METHOD__ . '-pfunc' );
2883
2884                         $colonPos = strpos( $part1, ':' );
2885                         if ( $colonPos !== false ) {
2886                                 # Case sensitive functions
2887                                 $function = substr( $part1, 0, $colonPos );
2888                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2889                                         $function = $this->mFunctionSynonyms[1][$function];
2890                                 } else {
2891                                         # Case insensitive functions
2892                                         $function = strtolower( $function );
2893                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2894                                                 $function = $this->mFunctionSynonyms[0][$function];
2895                                         } else {
2896                                                 $function = false;
2897                                         }
2898                                 }
2899                                 if ( $function ) {
2900                                         $funcArgs = array_map( 'trim', $args );
2901                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2902                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2903                                         $found = true;
2904
2905                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2906                                         //$noargs = true;
2907                                         //$noparse = true;
2908
2909                                         if ( is_array( $result ) ) {
2910                                                 if ( isset( $result[0] ) ) {
2911                                                         $text = $linestart . $result[0];
2912                                                         unset( $result[0] );
2913                                                 }
2914
2915                                                 // Extract flags into the local scope
2916                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2917                                                 extract( $result );
2918                                         } else {
2919                                                 $text = $linestart . $result;
2920                                         }
2921                                 }
2922                         }
2923                         wfProfileOut( __METHOD__ . '-pfunc' );
2924                 }
2925
2926                 # Template table test
2927
2928                 # Did we encounter this template already? If yes, it is in the cache
2929                 # and we need to check for loops.
2930                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2931                         $found = true;
2932
2933                         # Infinite loop test
2934                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2935                                 $noparse = true;
2936                                 $noargs = true;
2937                                 $found = true;
2938                                 $text = $linestart .
2939                                         "[[$part1]]<!-- WARNING: template loop detected -->";
2940                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
2941                         } else {
2942                                 # set $text to cached message.
2943                                 $text = $linestart . $this->mTemplates[$piece['title']];
2944                         }
2945                 }
2946
2947                 # Load from database
2948                 $lastPathLevel = $this->mTemplatePath;
2949                 if ( !$found ) {
2950                         wfProfileIn( __METHOD__ . '-loadtpl' );
2951                         $ns = NS_TEMPLATE;
2952                         # declaring $subpage directly in the function call
2953                         # does not work correctly with references and breaks
2954                         # {{/subpage}}-style inclusions
2955                         $subpage = '';
2956                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2957                         if ($subpage !== '') {
2958                                 $ns = $this->mTitle->getNamespace();
2959                         }
2960                         $title = Title::newFromText( $part1, $ns );
2961
2962
2963                         if ( !is_null( $title ) ) {
2964                                 $titleText = $title->getPrefixedText();
2965                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2966                                 # Check for language variants if the template is not found
2967                                 if($checkVariantLink && $title->getArticleID() == 0){
2968                                         $wgContLang->findVariantLink($part1, $title);
2969                                 }
2970
2971                                 if ( !$title->isExternal() ) {
2972                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2973                                                 $text = SpecialPage::capturePath( $title );
2974                                                 if ( is_string( $text ) ) {
2975                                                         $found = true;
2976                                                         $noparse = true;
2977                                                         $noargs = true;
2978                                                         $isHTML = true;
2979                                                         $this->disableCache();
2980                                                 }
2981                                         } else {
2982                                                 $articleContent = $this->fetchTemplate( $title );
2983                                                 if ( $articleContent !== false ) {
2984                                                         $found = true;
2985                                                         $text = $articleContent;
2986                                                         $replaceHeadings = true;
2987                                                 }
2988                                         }
2989
2990                                         # If the title is valid but undisplayable, make a link to it
2991                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2992                                                 $text = "[[:$titleText]]";
2993                                                 $found = true;
2994                                         }
2995                                 } elseif ( $title->isTrans() ) {
2996                                         // Interwiki transclusion
2997                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
2998                                                 $text = $this->interwikiTransclude( $title, 'render' );
2999                                                 $isHTML = true;
3000                                                 $noparse = true;
3001                                         } else {
3002                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3003                                                 $replaceHeadings = true;
3004                                         }
3005                                         $found = true;
3006                                 }
3007
3008                                 # Template cache array insertion
3009                                 # Use the original $piece['title'] not the mangled $part1, so that
3010                                 # modifiers such as RAW: produce separate cache entries
3011                                 if( $found ) {
3012                                         if( $isHTML ) {
3013                                                 // A special page; don't store it in the template cache.
3014                                         } else {
3015                                                 $this->mTemplates[$piece['title']] = $text;
3016                                         }
3017                                         $text = $linestart . $text;
3018                                 }
3019                         }
3020                         wfProfileOut( __METHOD__ . '-loadtpl' );
3021                 }
3022
3023                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3024                         # Error, oversize inclusion
3025                         $text = $linestart .
3026                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3027                         $noparse = true;
3028                         $noargs = true;
3029                 }
3030
3031                 # Recursive parsing, escaping and link table handling
3032                 # Only for HTML output
3033                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3034                         $text = wfEscapeWikiText( $text );
3035                 } elseif ( !$this->ot['msg'] && $found ) {
3036                         if ( $noargs ) {
3037                                 $assocArgs = array();
3038                         } else {
3039                                 # Clean up argument array
3040                                 $assocArgs = array();
3041                                 $index = 1;
3042                                 foreach( $args as $arg ) {
3043                                         $eqpos = strpos( $arg, '=' );
3044                                         if ( $eqpos === false ) {
3045                                                 $assocArgs[$index++] = $arg;
3046                                         } else {
3047                                                 $name = trim( substr( $arg, 0, $eqpos ) );
3048                                                 $value = trim( substr( $arg, $eqpos+1 ) );
3049                                                 if ( $value === false ) {
3050                                                         $value = '';
3051                                                 }
3052                                                 if ( $name !== false ) {
3053                                                         $assocArgs[$name] = $value;
3054                                                 }
3055                                         }
3056                                 }
3057
3058                                 # Add a new element to the templace recursion path
3059                                 $this->mTemplatePath[$part1] = 1;
3060                         }
3061
3062                         if ( !$noparse ) {
3063                                 # If there are any <onlyinclude> tags, only include them
3064                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3065                                         $replacer = new OnlyIncludeReplacer;
3066                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3067                                                 array( &$replacer, 'replace' ), $text );
3068                                         $text = $replacer->output;
3069                                 }
3070                                 # Remove <noinclude> sections and <includeonly> tags
3071                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3072                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3073
3074                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3075                                         # Strip <nowiki>, <pre>, etc.
3076                                         $text = $this->strip( $text, $this->mStripState );
3077                                         if ( $this->ot['html'] ) {
3078                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3079                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3080                                                 $text = Sanitizer::removeHTMLcomments( $text );
3081                                         }
3082                                 }
3083                                 $text = $this->replaceVariables( $text, $assocArgs );
3084
3085                                 # If the template begins with a table or block-level
3086                                 # element, it should be treated as beginning a new line.
3087                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3088                                         $text = "\n" . $text;
3089                                 }
3090                         } elseif ( !$noargs ) {
3091                                 # $noparse and !$noargs
3092                                 # Just replace the arguments, not any double-brace items
3093                                 # This is used for rendered interwiki transclusion
3094                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3095                         }
3096                 }
3097                 # Prune lower levels off the recursion check path
3098                 $this->mTemplatePath = $lastPathLevel;
3099
3100                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3101                         # Error, oversize inclusion
3102                         $text = $linestart .
3103                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3104                         $noparse = true;
3105                         $noargs = true;
3106                 }
3107
3108                 if ( !$found ) {
3109                         wfProfileOut( $fname );
3110                         return $piece['text'];
3111                 } else {
3112                         wfProfileIn( __METHOD__ . '-placeholders' );
3113                         if ( $isHTML ) {
3114                                 # Replace raw HTML by a placeholder
3115                                 # Add a blank line preceding, to prevent it from mucking up
3116                                 # immediately preceding headings
3117                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3118                         } else {
3119                                 # replace ==section headers==
3120                                 # XXX this needs to go away once we have a better parser.
3121                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3122                                         if( !is_null( $title ) )
3123                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3124                                         else
3125                                                 $encodedname = base64_encode("");
3126                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3127                                                 PREG_SPLIT_DELIM_CAPTURE);
3128                                         $text = '';
3129                                         $nsec = 0;
3130                                         for( $i = 0; $i < count($m); $i += 2 ) {
3131                                                 $text .= $m[$i];
3132                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3133                                                 $hl = $m[$i + 1];
3134                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3135                                                         $text .= $hl;
3136                                                         continue;
3137                                                 }
3138                                                 $m2 = array();
3139                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3140                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3141                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3142
3143                                                 $nsec++;
3144                                         }
3145                                 }
3146                         }
3147                         wfProfileOut( __METHOD__ . '-placeholders' );
3148                 }
3149
3150                 # Prune lower levels off the recursion check path
3151                 $this->mTemplatePath = $lastPathLevel;
3152
3153                 if ( !$found ) {
3154                         wfProfileOut( $fname );
3155                         return $piece['text'];
3156                 } else {
3157                         wfProfileOut( $fname );
3158                         return $text;
3159                 }
3160         }
3161
3162         /**
3163          * Fetch the unparsed text of a template and register a reference to it.
3164          */
3165         function fetchTemplate( $title ) {
3166                 $text = false;
3167                 // Loop to fetch the article, with up to 1 redirect
3168                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3169                         $rev = Revision::newFromTitle( $title );
3170                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3171                         if ( !$rev ) {
3172                                 break;
3173                         }
3174                         $text = $rev->getText();
3175                         if ( $text === false ) {
3176                                 break;
3177                         }
3178                         // Redirect?
3179                         $title = Title::newFromRedirect( $text );
3180                 }
3181                 return $text;
3182         }
3183
3184         /**
3185          * Transclude an interwiki link.
3186          */
3187         function interwikiTransclude( $title, $action ) {
3188                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3189
3190                 if (!$wgEnableScaryTranscluding)
3191                         return wfMsg('scarytranscludedisabled');
3192
3193                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3194                 // But we'll handle it generally anyway
3195                 if ( $title->getNamespace() ) {
3196                         // Use the canonical namespace, which should work anywhere
3197                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3198                 } else {
3199                         $articleName = $title->getDBkey();
3200                 }
3201
3202                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3203                 $url .= "?action=$action";
3204                 if (strlen($url) > 255)
3205                         return wfMsg('scarytranscludetoolong');
3206                 return $this->fetchScaryTemplateMaybeFromCache($url);
3207         }
3208
3209         function fetchScaryTemplateMaybeFromCache($url) {
3210                 global $wgTranscludeCacheExpiry;
3211                 $dbr =& wfGetDB(DB_SLAVE);
3212                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3213                                 array('tc_url' => $url));
3214                 if ($obj) {
3215                         $time = $obj->tc_time;
3216                         $text = $obj->tc_contents;
3217                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3218                                 return $text;
3219                         }
3220                 }
3221
3222                 $text = Http::get($url);
3223                 if (!$text)
3224                         return wfMsg('scarytranscludefailed', $url);
3225
3226                 $dbw =& wfGetDB(DB_MASTER);
3227                 $dbw->replace('transcache', array('tc_url'), array(
3228                         'tc_url' => $url,
3229                         'tc_time' => time(),
3230                         'tc_contents' => $text));
3231                 return $text;
3232         }
3233
3234
3235         /**
3236          * Triple brace replacement -- used for template arguments
3237          * @private
3238          */
3239         function argSubstitution( $matches ) {
3240                 $arg = trim( $matches['title'] );
3241                 $text = $matches['text'];
3242                 $inputArgs = end( $this->mArgStack );
3243
3244                 if ( array_key_exists( $arg, $inputArgs ) ) {
3245                         $text = $inputArgs[$arg];
3246                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3247                 null != $matches['parts'] && count($matches['parts']) > 0) {
3248                         $text = $matches['parts'][0];
3249                 }
3250                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3251                         $text = $matches['text'] .
3252                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3253                 }
3254
3255                 return $text;
3256         }
3257
3258         /**
3259          * Increment an include size counter
3260          *
3261          * @param string $type The type of expansion
3262          * @param integer $size The size of the text
3263          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3264          */
3265         function incrementIncludeSize( $type, $size ) {
3266                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3267                         return false;
3268                 } else {
3269                         $this->mIncludeSizes[$type] += $size;
3270                         return true;
3271                 }
3272         }
3273
3274         /**
3275          * Detect __NOGALLERY__ magic word and set a placeholder
3276          */
3277         function stripNoGallery( &$text ) {
3278                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3279                 # do not add TOC
3280                 $mw = MagicWord::get( 'nogallery' );
3281                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3282         }
3283
3284         /**
3285          * Detect __TOC__ magic word and set a placeholder
3286          */
3287         function stripToc( $text ) {
3288                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3289                 # do not add TOC
3290                 $mw = MagicWord::get( 'notoc' );
3291                 if( $mw->matchAndRemove( $text ) ) {
3292                         $this->mShowToc = false;
3293                 }
3294
3295                 $mw = MagicWord::get( 'toc' );
3296                 if( $mw->match( $text ) ) {
3297                         $this->mShowToc = true;
3298                         $this->mForceTocPosition = true;
3299
3300                         // Set a placeholder. At the end we'll fill it in with the TOC.
3301                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3302
3303                         // Only keep the first one.
3304                         $text = $mw->replace( '', $text );
3305                 }
3306                 return $text;
3307         }
3308
3309         /**
3310          * This function accomplishes several tasks:
3311          * 1) Auto-number headings if that option is enabled
3312          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3313          * 3) Add a Table of contents on the top for users who have enabled the option
3314          * 4) Auto-anchor headings
3315          *
3316          * It loops through all headlines, collects the necessary data, then splits up the
3317          * string and re-inserts the newly formatted headlines.
3318          *
3319          * @param string $text
3320          * @param boolean $isMain
3321          * @private
3322          */
3323         function formatHeadings( $text, $isMain=true ) {
3324                 global $wgMaxTocLevel, $wgContLang;
3325
3326                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3327                 if( !$this->mTitle->userCanEdit() ) {
3328                         $showEditLink = 0;
3329                 } else {
3330                         $showEditLink = $this->mOptions->getEditSection();
3331                 }
3332
3333                 # Inhibit editsection links if requested in the page
3334                 $esw =& MagicWord::get( 'noeditsection' );
3335                 if( $esw->matchAndRemove( $text ) ) {
3336                         $showEditLink = 0;
3337                 }
3338
3339                 # Get all headlines for numbering them and adding funky stuff like [edit]
3340                 # links - this is for later, but we need the number of headlines right now
3341                 $matches = array();
3342                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3343
3344                 # if there are fewer than 4 headlines in the article, do not show TOC
3345                 # unless it's been explicitly enabled.
3346                 $enoughToc = $this->mShowToc &&
3347                         (($numMatches >= 4) || $this->mForceTocPosition);
3348
3349                 # Allow user to stipulate that a page should have a "new section"
3350                 # link added via __NEWSECTIONLINK__
3351                 $mw =& MagicWord::get( 'newsectionlink' );
3352                 if( $mw->matchAndRemove( $text ) )
3353                         $this->mOutput->setNewSection( true );
3354
3355                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3356                 # override above conditions and always show TOC above first header
3357                 $mw =& MagicWord::get( 'forcetoc' );
3358                 if ($mw->matchAndRemove( $text ) ) {
3359                         $this->mShowToc = true;
3360                         $enoughToc = true;
3361                 }
3362
3363                 # Never ever show TOC if no headers
3364                 if( $numMatches < 1 ) {
3365                         $enoughToc = false;
3366                 }
3367
3368                 # We need this to perform operations on the HTML
3369                 $sk =& $this->mOptions->getSkin();
3370
3371                 # headline counter
3372                 $headlineCount = 0;
3373                 $sectionCount = 0; # headlineCount excluding template sections
3374
3375                 # Ugh .. the TOC should have neat indentation levels which can be
3376                 # passed to the skin functions. These are determined here
3377                 $toc = '';
3378                 $full = '';
3379                 $head = array();
3380                 $sublevelCount = array();
3381                 $levelCount = array();
3382                 $toclevel = 0;
3383                 $level = 0;
3384                 $prevlevel = 0;
3385                 $toclevel = 0;
3386                 $prevtoclevel = 0;
3387
3388                 foreach( $matches[3] as $headline ) {
3389                         $istemplate = 0;
3390                         $templatetitle = '';
3391                         $templatesection = 0;
3392                         $numbering = '';
3393                         $mat = array();
3394                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3395                                 $istemplate = 1;
3396                                 $templatetitle = base64_decode($mat[1]);
3397                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3398                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3399                         }
3400
3401                         if( $toclevel ) {
3402                                 $prevlevel = $level;
3403                                 $prevtoclevel = $toclevel;
3404                         }
3405                         $level = $matches[1][$headlineCount];
3406
3407                         if( $doNumberHeadings || $enoughToc ) {
3408
3409                                 if ( $level > $prevlevel ) {
3410                                         # Increase TOC level
3411                                         $toclevel++;
3412                                         $sublevelCount[$toclevel] = 0;
3413                                         if( $toclevel<$wgMaxTocLevel ) {
3414                                                 $toc .= $sk->tocIndent();
3415                                         }
3416                                 }
3417                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3418                                         # Decrease TOC level, find level to jump to
3419
3420                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3421                                                 # Can only go down to level 1
3422                                                 $toclevel = 1;
3423                                         } else {
3424                                                 for ($i = $toclevel; $i > 0; $i--) {
3425                                                         if ( $levelCount[$i] == $level ) {
3426                                                                 # Found last matching level
3427                                                                 $toclevel = $i;
3428                                                                 break;
3429                                                         }
3430                                                         elseif ( $levelCount[$i] < $level ) {
3431                                                                 # Found first matching level below current level
3432                                                                 $toclevel = $i + 1;
3433                                                                 break;
3434                                                         }
3435                                                 }
3436                                         }
3437                                         if( $toclevel<$wgMaxTocLevel ) {
3438                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3439                                         }
3440                                 }
3441                                 else {
3442                                         # No change in level, end TOC line
3443                                         if( $toclevel<$wgMaxTocLevel ) {
3444                                                 $toc .= $sk->tocLineEnd();
3445                                         }
3446                                 }
3447
3448                                 $levelCount[$toclevel] = $level;
3449
3450                                 # count number of headlines for each level
3451                                 @$sublevelCount[$toclevel]++;
3452                                 $dot = 0;
3453                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3454                                         if( !empty( $sublevelCount[$i] ) ) {
3455                                                 if( $dot ) {
3456                                                         $numbering .= '.';
3457                                                 }
3458                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3459                                                 $dot = 1;
3460                                         }
3461                                 }
3462                         }
3463
3464                         # The canonized header is a version of the header text safe to use for links
3465                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3466                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3467
3468                         # Remove link placeholders by the link text.
3469                         #     <!--LINK number-->
3470                         # turns into
3471                         #     link text with suffix
3472                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3473                                                             "\$this->mLinkHolders['texts'][\$1]",
3474                                                             $canonized_headline );
3475                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3476                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3477                                                             $canonized_headline );
3478
3479                         # strip out HTML
3480                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3481                         $tocline = trim( $canonized_headline );
3482                         # Save headline for section edit hint before it's escaped
3483                         $headline_hint = trim( $canonized_headline );
3484                         $canonized_headline = Sanitizer::escapeId( $tocline );
3485                         $refers[$headlineCount] = $canonized_headline;
3486
3487                         # count how many in assoc. array so we can track dupes in anchors
3488                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3489                         $refcount[$headlineCount]=$refers[$canonized_headline];
3490
3491                         # Don't number the heading if it is the only one (looks silly)
3492                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3493                                 # the two are different if the line contains a link
3494                                 $headline=$numbering . ' ' . $headline;
3495                         }
3496
3497                         # Create the anchor for linking from the TOC to the section
3498                         $anchor = $canonized_headline;
3499                         if($refcount[$headlineCount] > 1 ) {
3500                                 $anchor .= '_' . $refcount[$headlineCount];
3501                         }
3502                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3503                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3504                         }
3505                         # give headline the correct <h#> tag
3506                         $head[$headlineCount] = "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount];
3507
3508                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3509                                 if ( empty( $head[$headlineCount] ) ) {
3510                                         $head[$headlineCount] = '';
3511                                 }
3512                                 if( $istemplate )
3513                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3514                                 else
3515                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3516                         }
3517                         // Yes, the headline logically goes before the edit section.  Why isn't it there
3518                         // in source?  Ask the CSS people.  The float gets screwed up if you do that.
3519                         // This might be moved to before the editsection at some point so that it will
3520                         // display a bit more prettily without CSS, so please don't rely on the order.
3521                         $head[$headlineCount] .= ' <span class="mw-headline">'.$headline.'</span></h'.$level.'>';
3522
3523                         $headlineCount++;
3524                         if( !$istemplate )
3525                                 $sectionCount++;
3526                 }
3527
3528                 if( $enoughToc ) {
3529                         if( $toclevel<$wgMaxTocLevel ) {
3530                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3531                         }
3532                         $toc = $sk->tocList( $toc );
3533                 }
3534
3535                 # split up and insert constructed headlines
3536
3537                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3538                 $i = 0;
3539
3540                 foreach( $blocks as $block ) {
3541                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3542                                 # This is the [edit] link that appears for the top block of text when
3543                                 # section editing is enabled
3544
3545                                 # Disabled because it broke block formatting
3546                                 # For example, a bullet point in the top line
3547                                 # $full .= $sk->editSectionLink(0);
3548                         }
3549                         $full .= $block;
3550                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3551                                 # Top anchor now in skin
3552                                 $full = $full.$toc;
3553                         }
3554
3555                         if( !empty( $head[$i] ) ) {
3556                                 $full .= $head[$i];
3557                         }
3558                         $i++;
3559                 }
3560                 if( $this->mForceTocPosition ) {
3561                         return str_replace( '<!--MWTOC-->', $toc, $full );
3562                 } else {
3563                         return $full;
3564                 }
3565         }
3566
3567         /**
3568          * Transform wiki markup when saving a page by doing \r\n -> \n
3569          * conversion, substitting signatures, {{subst:}} templates, etc.
3570          *
3571          * @param string $text the text to transform
3572          * @param Title &$title the Title object for the current article
3573          * @param User &$user the User object describing the current user
3574          * @param ParserOptions $options parsing options
3575          * @param bool $clearState whether to clear the parser state first
3576          * @return string the altered wiki markup
3577          * @public
3578          */
3579         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3580                 $this->mOptions = $options;
3581                 $this->mTitle =& $title;
3582                 $this->setOutputType( OT_WIKI );
3583
3584                 if ( $clearState ) {
3585                         $this->clearState();
3586                 }
3587
3588                 $stripState = new StripState;
3589                 $pairs = array(
3590                         "\r\n" => "\n",
3591                 );
3592                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3593                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3594                 $text = $this->pstPass2( $text, $stripState, $user );
3595                 $text = $stripState->unstripBoth( $text );
3596                 return $text;
3597         }
3598
3599         /**
3600          * Pre-save transform helper function
3601          * @private
3602          */
3603         function pstPass2( $text, &$stripState, $user ) {
3604                 global $wgContLang, $wgLocaltimezone;
3605
3606                 /* Note: This is the timestamp saved as hardcoded wikitext to
3607                  * the database, we use $wgContLang here in order to give
3608                  * everyone the same signature and use the default one rather
3609                  * than the one selected in each user's preferences.
3610                  */
3611                 if ( isset( $wgLocaltimezone ) ) {
3612                         $oldtz = getenv( 'TZ' );
3613                         putenv( 'TZ='.$wgLocaltimezone );
3614                 }
3615                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3616                   ' (' . date( 'T' ) . ')';
3617                 if ( isset( $wgLocaltimezone ) ) {
3618                         putenv( 'TZ='.$oldtz );
3619                 }
3620
3621                 # Variable replacement
3622                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3623                 $text = $this->replaceVariables( $text );
3624
3625                 # Strip out <nowiki> etc. added via replaceVariables
3626                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3627
3628                 # Signatures
3629                 $sigText = $this->getUserSig( $user );
3630                 $text = strtr( $text, array(
3631                         '~~~~~' => $d,
3632                         '~~~~' => "$sigText $d",
3633                         '~~~' => $sigText
3634                 ) );
3635
3636                 # Context links: [[|name]] and [[name (context)|]]
3637                 #
3638                 global $wgLegalTitleChars;
3639                 $tc = "[$wgLegalTitleChars]";
3640                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3641
3642                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3643                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3644                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3645
3646                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3647                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3648                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3649
3650                 $t = $this->mTitle->getText();
3651                 $m = array();
3652                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3653                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3654                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3655                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3656                 } else {
3657                         # if there's no context, don't bother duplicating the title
3658                         $text = preg_replace( $p2, '[[\\1]]', $text );
3659                 }
3660
3661                 # Trim trailing whitespace
3662                 # __END__ tag allows for trailing
3663                 # whitespace to be deliberately included
3664                 $text = rtrim( $text );
3665                 $mw =& MagicWord::get( 'end' );
3666                 $mw->matchAndRemove( $text );
3667
3668                 return $text;
3669         }
3670
3671         /**
3672          * Fetch the user's signature text, if any, and normalize to
3673          * validated, ready-to-insert wikitext.
3674          *
3675          * @param User $user
3676          * @return string
3677          * @private
3678          */
3679         function getUserSig( &$user ) {
3680                 $username = $user->getName();
3681                 $nickname = $user->getOption( 'nickname' );
3682                 $nickname = $nickname === '' ? $username : $nickname;
3683
3684                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3685                         # Sig. might contain markup; validate this
3686                         if( $this->validateSig( $nickname ) !== false ) {
3687                                 # Validated; clean up (if needed) and return it
3688                                 return $this->cleanSig( $nickname, true );
3689                         } else {
3690                                 # Failed to validate; fall back to the default
3691                                 $nickname = $username;
3692                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3693                         }
3694                 }
3695
3696                 // Make sure nickname doesnt get a sig in a sig
3697                 $nickname = $this->cleanSigInSig( $nickname );
3698
3699                 # If we're still here, make it a link to the user page
3700                 $userpage = $user->getUserPage();
3701                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3702         }
3703
3704         /**
3705          * Check that the user's signature contains no bad XML
3706          *
3707          * @param string $text
3708          * @return mixed An expanded string, or false if invalid.
3709          */
3710         function validateSig( $text ) {
3711                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3712         }
3713
3714         /**
3715          * Clean up signature text
3716          *
3717          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3718          * 2) Substitute all transclusions
3719          *
3720          * @param string $text
3721          * @param $parsing Whether we're cleaning (preferences save) or parsing
3722          * @return string Signature text
3723          */
3724         function cleanSig( $text, $parsing = false ) {
3725                 global $wgTitle;
3726                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3727
3728                 $substWord = MagicWord::get( 'subst' );
3729                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3730                 $substText = '{{' . $substWord->getSynonym( 0 );
3731
3732                 $text = preg_replace( $substRegex, $substText, $text );
3733                 $text = $this->cleanSigInSig( $text );
3734                 $text = $this->replaceVariables( $text );
3735
3736                 $this->clearState();
3737                 return $text;
3738         }
3739
3740         /**
3741          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3742          * @param string $text
3743          * @return string Signature text with /~{3,5}/ removed
3744          */
3745         function cleanSigInSig( $text ) {
3746                 $text = preg_replace( '/~{3,5}/', '', $text );
3747                 return $text;
3748         }
3749
3750         /**
3751          * Set up some variables which are usually set up in parse()
3752          * so that an external function can call some class members with confidence
3753          * @public
3754          */
3755         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3756                 $this->mTitle =& $title;
3757                 $this->mOptions = $options;
3758                 $this->setOutputType( $outputType );
3759                 if ( $clearState ) {
3760                         $this->clearState();
3761                 }
3762         }
3763
3764         /**
3765          * Transform a MediaWiki message by replacing magic variables.
3766          *
3767          * @param string $text the text to transform
3768          * @param ParserOptions $options  options
3769          * @return string the text with variables substituted
3770          * @public
3771          */
3772         function transformMsg( $text, $options ) {
3773                 global $wgTitle;
3774                 static $executing = false;
3775
3776                 $fname = "Parser::transformMsg";
3777
3778                 # Guard against infinite recursion
3779                 if ( $executing ) {
3780                         return $text;
3781                 }
3782                 $executing = true;
3783
3784                 wfProfileIn($fname);
3785
3786                 if ( $wgTitle ) {
3787                         $this->mTitle = $wgTitle;
3788                 } else {
3789                         $this->mTitle = Title::newFromText('msg');
3790                 }
3791                 $this->mOptions = $options;
3792                 $this->setOutputType( OT_MSG );
3793                 $this->clearState();
3794                 $text = $this->replaceVariables( $text );
3795
3796                 $executing = false;
3797                 wfProfileOut($fname);
3798                 return $text;
3799         }
3800
3801         /**
3802          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3803          * The callback should have the following form:
3804          *    function myParserHook( $text, $params, &$parser ) { ... }
3805          *
3806          * Transform and return $text. Use $parser for any required context, e.g. use
3807          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3808          *
3809          * @public
3810          *
3811          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3812          * @param mixed $callback The callback function (and object) to use for the tag
3813          *
3814          * @return The old value of the mTagHooks array associated with the hook
3815          */
3816         function setHook( $tag, $callback ) {
3817                 $tag = strtolower( $tag );
3818                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3819                 $this->mTagHooks[$tag] = $callback;
3820
3821                 return $oldVal;
3822         }
3823
3824         /**
3825          * Create a function, e.g. {{sum:1|2|3}}
3826          * The callback function should have the form:
3827          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3828          *
3829          * The callback may either return the text result of the function, or an array with the text
3830          * in element 0, and a number of flags in the other elements. The names of the flags are
3831          * specified in the keys. Valid flags are:
3832          *   found                     The text returned is valid, stop processing the template. This
3833          *                             is on by default.
3834          *   nowiki                    Wiki markup in the return value should be escaped
3835          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3836          *   noargs                    Don't replace triple-brace arguments in the return value
3837          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3838          *
3839          * @public
3840          *
3841          * @param string $id The magic word ID
3842          * @param mixed $callback The callback function (and object) to use
3843          * @param integer $flags a combination of the following flags:
3844          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3845          *
3846          * @return The old callback function for this name, if any
3847          */
3848         function setFunctionHook( $id, $callback, $flags = 0 ) {
3849                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3850                 $this->mFunctionHooks[$id] = $callback;
3851
3852                 # Add to function cache
3853                 $mw = MagicWord::get( $id );
3854                 if( !$mw )
3855                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3856
3857                 $synonyms = $mw->getSynonyms();
3858                 $sensitive = intval( $mw->isCaseSensitive() );
3859
3860                 foreach ( $synonyms as $syn ) {
3861                         # Case
3862                         if ( !$sensitive ) {
3863                                 $syn = strtolower( $syn );
3864                         }
3865                         # Add leading hash
3866                         if ( !( $flags & SFH_NO_HASH ) ) {
3867                                 $syn = '#' . $syn;
3868                         }
3869                         # Remove trailing colon
3870                         if ( substr( $syn, -1, 1 ) == ':' ) {
3871                                 $syn = substr( $syn, 0, -1 );
3872                         }
3873                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3874                 }
3875                 return $oldVal;
3876         }
3877
3878         /**
3879          * Get all registered function hook identifiers
3880          *
3881          * @return array
3882          */
3883         function getFunctionHooks() {
3884                 return array_keys( $this->mFunctionHooks );
3885         }
3886
3887         /**
3888          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3889          * Placeholders created in Skin::makeLinkObj()
3890          * Returns an array of links found, indexed by PDBK:
3891          *  0 - broken
3892          *  1 - normal link
3893          *  2 - stub
3894          * $options is a bit field, RLH_FOR_UPDATE to select for update
3895          */
3896         function replaceLinkHolders( &$text, $options = 0 ) {
3897                 global $wgUser;
3898                 global $wgContLang;
3899
3900                 $fname = 'Parser::replaceLinkHolders';
3901                 wfProfileIn( $fname );
3902
3903                 $pdbks = array();
3904                 $colours = array();
3905                 $sk =& $this->mOptions->getSkin();
3906                 $linkCache =& LinkCache::singleton();
3907
3908                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3909                         wfProfileIn( $fname.'-check' );
3910                         $dbr =& wfGetDB( DB_SLAVE );
3911                         $page = $dbr->tableName( 'page' );
3912                         $threshold = $wgUser->getOption('stubthreshold');
3913
3914                         # Sort by namespace
3915                         asort( $this->mLinkHolders['namespaces'] );
3916
3917                         # Generate query
3918                         $query = false;
3919                         $current = null;
3920                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3921                                 # Make title object
3922                                 $title = $this->mLinkHolders['titles'][$key];
3923
3924                                 # Skip invalid entries.
3925                                 # Result will be ugly, but prevents crash.
3926                                 if ( is_null( $title ) ) {
3927                                         continue;
3928                                 }
3929                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3930
3931                                 # Check if it's a static known link, e.g. interwiki
3932                                 if ( $title->isAlwaysKnown() ) {
3933                                         $colours[$pdbk] = 1;
3934                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3935                                         $colours[$pdbk] = 1;
3936                                         $this->mOutput->addLink( $title, $id );
3937                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3938                                         $colours[$pdbk] = 0;
3939                                 } else {
3940                                         # Not in the link cache, add it to the query
3941                                         if ( !isset( $current ) ) {
3942                                                 $current = $ns;
3943                                                 $query =  "SELECT page_id, page_namespace, page_title";
3944                                                 if ( $threshold > 0 ) {
3945                                                         $query .= ', page_len, page_is_redirect';
3946                                                 }
3947                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3948                                         } elseif ( $current != $ns ) {
3949                                                 $current = $ns;
3950                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3951                                         } else {
3952                                                 $query .= ', ';
3953                                         }
3954
3955                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3956                                 }
3957                         }
3958                         if ( $query ) {
3959                                 $query .= '))';
3960                                 if ( $options & RLH_FOR_UPDATE ) {
3961                                         $query .= ' FOR UPDATE';
3962                                 }
3963
3964                                 $res = $dbr->query( $query, $fname );
3965
3966                                 # Fetch data and form into an associative array
3967                                 # non-existent = broken
3968                                 # 1 = known
3969                                 # 2 = stub
3970                                 while ( $s = $dbr->fetchObject($res) ) {
3971                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3972                                         $pdbk = $title->getPrefixedDBkey();
3973                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3974                                         $this->mOutput->addLink( $title, $s->page_id );
3975
3976                                         if ( $threshold >  0 ) {
3977                                                 $size = $s->page_len;
3978                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3979                                                         $colours[$pdbk] = 1;
3980                                                 } else {
3981                                                         $colours[$pdbk] = 2;
3982                                                 }
3983                                         } else {
3984                                                 $colours[$pdbk] = 1;
3985                                         }
3986                                 }
3987                         }
3988                         wfProfileOut( $fname.'-check' );
3989
3990                         # Do a second query for different language variants of links (if needed)
3991                         if($wgContLang->hasVariants()){
3992                                 $linkBatch = new LinkBatch();
3993                                 $variantMap = array(); // maps $pdbkey_Variant => $pdbkey_original
3994
3995                                 // Add variants of links to link batch
3996                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3997                                         $title = $this->mLinkHolders['titles'][$key];
3998                                         if ( is_null( $title ) )
3999                                                 continue;
4000
4001                                         $pdbk = $title->getPrefixedDBkey();
4002
4003                                         // generate all variants of the link title text
4004                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
4005
4006                                         // if link was not found (in first query), add all variants to query
4007                                         if ( !isset($colours[$pdbk]) ){
4008                                                 foreach($allTextVariants as $textVariant){
4009                                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
4010                                                         if(is_null($variantTitle)) continue;
4011                                                         $linkBatch->addObj( $variantTitle );
4012                                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4013                                                 }
4014                                         }
4015                                 }
4016
4017
4018                                 if(!$linkBatch->isEmpty()){
4019                                         // construct query
4020                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4021
4022                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4023                                         if ( $threshold > 0 ) {
4024                                                 $variantQuery .= ', page_len, page_is_redirect';
4025                                         }
4026
4027                                         $variantQuery .= " FROM $page WHERE $titleClause";
4028                                         if ( $options & RLH_FOR_UPDATE ) {
4029                                                 $variantQuery .= ' FOR UPDATE';
4030                                         }
4031
4032                                         $varRes = $dbr->query( $variantQuery, $fname );
4033
4034                                         // for each found variants, figure out link holders and replace
4035                                         while ( $s = $dbr->fetchObject($varRes) ) {
4036
4037                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4038                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4039                                                 $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4040                                                 $this->mOutput->addLink( $variantTitle, $s->page_id );
4041
4042                                                 $holderKeys = $variantMap[$varPdbk];
4043
4044                                                 // loop over link holders
4045                                                 foreach($holderKeys as $key){
4046                                                         $title = $this->mLinkHolders['titles'][$key];
4047                                                         if ( is_null( $title ) ) continue;
4048
4049                                                         $pdbk = $title->getPrefixedDBkey();
4050
4051                                                         if(!isset($colours[$pdbk])){
4052                                                                 // found link in some of the variants, replace the link holder data
4053                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4054                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4055
4056                                                                 // set pdbk and colour
4057                                                                 $pdbks[$key] = $varPdbk;
4058                                                                 if ( $threshold >  0 ) {
4059                                                                         $size = $s->page_len;
4060                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4061                                                                                 $colours[$varPdbk] = 1;
4062                                                                         } else {
4063                                                                                 $colours[$varPdbk] = 2;
4064                                                                         }
4065                                                                 }
4066                                                                 else {
4067                                                                         $colours[$varPdbk] = 1;
4068                                                                 }
4069                                                         }
4070                                                 }
4071                                         }
4072                                 }
4073                         }
4074
4075                         # Construct search and replace arrays
4076                         wfProfileIn( $fname.'-construct' );
4077                         $replacePairs = array();
4078                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4079                                 $pdbk = $pdbks[$key];
4080                                 $searchkey = "<!--LINK $key-->";
4081                                 $title = $this->mLinkHolders['titles'][$key];
4082                                 if ( empty( $colours[$pdbk] ) ) {
4083                                         $linkCache->addBadLinkObj( $title );
4084                                         $colours[$pdbk] = 0;
4085                                         $this->mOutput->addLink( $title, 0 );
4086                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4087                                                                         $this->mLinkHolders['texts'][$key],
4088                                                                         $this->mLinkHolders['queries'][$key] );
4089                                 } elseif ( $colours[$pdbk] == 1 ) {
4090                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4091                                                                         $this->mLinkHolders['texts'][$key],
4092                                                                         $this->mLinkHolders['queries'][$key] );
4093                                 } elseif ( $colours[$pdbk] == 2 ) {
4094                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4095                                                                         $this->mLinkHolders['texts'][$key],
4096                                                                         $this->mLinkHolders['queries'][$key] );
4097                                 }
4098                         }
4099                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4100                         wfProfileOut( $fname.'-construct' );
4101
4102                         # Do the thing
4103                         wfProfileIn( $fname.'-replace' );
4104                         $text = preg_replace_callback(
4105                                 '/(<!--LINK .*?-->)/',
4106                                 $replacer->cb(),
4107                                 $text);
4108
4109                         wfProfileOut( $fname.'-replace' );
4110                 }
4111
4112                 # Now process interwiki link holders
4113                 # This is quite a bit simpler than internal links
4114                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4115                         wfProfileIn( $fname.'-interwiki' );
4116                         # Make interwiki link HTML
4117                         $replacePairs = array();
4118                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4119                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4120                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4121                         }
4122                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4123
4124                         $text = preg_replace_callback(
4125                                 '/<!--IWLINK (.*?)-->/',
4126                                 $replacer->cb(),
4127                                 $text );
4128                         wfProfileOut( $fname.'-interwiki' );
4129                 }
4130
4131                 wfProfileOut( $fname );
4132                 return $colours;
4133         }
4134
4135         /**
4136          * Replace <!--LINK--> link placeholders with plain text of links
4137          * (not HTML-formatted).
4138          * @param string $text
4139          * @return string
4140          */
4141         function replaceLinkHoldersText( $text ) {
4142                 $fname = 'Parser::replaceLinkHoldersText';
4143                 wfProfileIn( $fname );
4144
4145                 $text = preg_replace_callback(
4146                         '/<!--(LINK|IWLINK) (.*?)-->/',
4147                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4148                         $text );
4149
4150                 wfProfileOut( $fname );
4151                 return $text;
4152         }
4153
4154         /**
4155          * @param array $matches
4156          * @return string
4157          * @private
4158          */
4159         function replaceLinkHoldersTextCallback( $matches ) {
4160                 $type = $matches[1];
4161                 $key  = $matches[2];
4162                 if( $type == 'LINK' ) {
4163                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4164                                 return $this->mLinkHolders['texts'][$key];
4165                         }
4166                 } elseif( $type == 'IWLINK' ) {
4167                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4168                                 return $this->mInterwikiLinkHolders['texts'][$key];
4169                         }
4170                 }
4171                 return $matches[0];
4172         }
4173
4174         /**
4175          * Tag hook handler for 'pre'.
4176          */
4177         function renderPreTag( $text, $attribs ) {
4178                 // Backwards-compatibility hack
4179                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4180
4181                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4182                 return wfOpenElement( 'pre', $attribs ) .
4183                         Xml::escapeTagsOnly( $content ) .
4184                         '</pre>';
4185         }
4186
4187         /**
4188          * Renders an image gallery from a text with one line per image.
4189          * text labels may be given by using |-style alternative text. E.g.
4190          *   Image:one.jpg|The number "1"
4191          *   Image:tree.jpg|A tree
4192          * given as text will return the HTML of a gallery with two images,
4193          * labeled 'The number "1"' and
4194          * 'A tree'.
4195          */
4196         function renderImageGallery( $text, $params ) {
4197                 $ig = new ImageGallery();
4198                 $ig->setShowBytes( false );
4199                 $ig->setShowFilename( false );
4200                 $ig->setParsing();
4201                 $ig->useSkin( $this->mOptions->getSkin() );
4202
4203                 if( isset( $params['caption'] ) )
4204                         $ig->setCaption( $params['caption'] );
4205
4206                 $lines = explode( "\n", $text );
4207                 foreach ( $lines as $line ) {
4208                         # match lines like these:
4209                         # Image:someimage.jpg|This is some image
4210                         $matches = array();
4211                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4212                         # Skip empty lines
4213                         if ( count( $matches ) == 0 ) {
4214                                 continue;
4215                         }
4216                         $tp = Title::newFromText( $matches[1] );
4217                         $nt =& $tp;
4218                         if( is_null( $nt ) ) {
4219                                 # Bogus title. Ignore these so we don't bomb out later.
4220                                 continue;
4221                         }
4222                         if ( isset( $matches[3] ) ) {
4223                                 $label = $matches[3];
4224                         } else {
4225                                 $label = '';
4226                         }
4227
4228                         $pout = $this->parse( $label,
4229                                 $this->mTitle,
4230                                 $this->mOptions,
4231                                 false, // Strip whitespace...?
4232                                 false  // Don't clear state!
4233                         );
4234                         $html = $pout->getText();
4235
4236                         $ig->add( new Image( $nt ), $html );
4237
4238                         # Only add real images (bug #5586)
4239                         if ( $nt->getNamespace() == NS_IMAGE ) {
4240                                 $this->mOutput->addImage( $nt->getDBkey() );
4241                         }
4242                 }
4243                 return $ig->toHTML();
4244         }
4245
4246         /**
4247          * Parse image options text and use it to make an image
4248          */
4249         function makeImage( $nt, $options ) {
4250                 global $wgUseImageResize, $wgDjvuRenderer;
4251
4252                 $align = '';
4253
4254                 # Check if the options text is of the form "options|alt text"
4255                 # Options are:
4256                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4257                 #  * left               no resizing, just left align. label is used for alt= only
4258                 #  * right              same, but right aligned
4259                 #  * none               same, but not aligned
4260                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4261                 #  * center             center the image
4262                 #  * framed             Keep original image size, no magnify-button.
4263
4264                 $part = explode( '|', $options);
4265
4266                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4267                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4268                 $mwLeft   =& MagicWord::get( 'img_left' );
4269                 $mwRight  =& MagicWord::get( 'img_right' );
4270                 $mwNone   =& MagicWord::get( 'img_none' );
4271                 $mwWidth  =& MagicWord::get( 'img_width' );
4272                 $mwCenter =& MagicWord::get( 'img_center' );
4273                 $mwFramed =& MagicWord::get( 'img_framed' );
4274                 $mwPage   =& MagicWord::get( 'img_page' );
4275                 $caption = '';
4276
4277                 $width = $height = $framed = $thumb = false;
4278                 $page = null;
4279                 $manual_thumb = '' ;
4280
4281                 foreach( $part as $val ) {
4282                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4283                                 $thumb=true;
4284                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4285                                 # use manually specified thumbnail
4286                                 $thumb=true;
4287                                 $manual_thumb = $match;
4288                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4289                                 # remember to set an alignment, don't render immediately
4290                                 $align = 'right';
4291                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4292                                 # remember to set an alignment, don't render immediately
4293                                 $align = 'left';
4294                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4295                                 # remember to set an alignment, don't render immediately
4296                                 $align = 'center';
4297                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4298                                 # remember to set an alignment, don't render immediately
4299                                 $align = 'none';
4300                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4301                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4302                                 # Select a page in a multipage document
4303                                 $page = $match;
4304                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4305                                 wfDebug( "img_width match: $match\n" );
4306                                 # $match is the image width in pixels
4307                                 $m = array();
4308                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4309                                         $width = intval( $m[1] );
4310                                         $height = intval( $m[2] );
4311                                 } else {
4312                                         $width = intval($match);
4313                                 }
4314                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4315                                 $framed=true;
4316                         } else {
4317                                 $caption = $val;
4318                         }
4319                 }
4320                 # Strip bad stuff out of the alt text
4321                 $alt = $this->replaceLinkHoldersText( $caption );
4322
4323                 # make sure there are no placeholders in thumbnail attributes
4324                 # that are later expanded to html- so expand them now and
4325                 # remove the tags
4326                 $alt = $this->mStripState->unstripBoth( $alt );
4327                 $alt = Sanitizer::stripAllTags( $alt );
4328
4329                 # Linker does the rest
4330                 $sk =& $this->mOptions->getSkin();
4331                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4332         }
4333
4334         /**
4335          * Set a flag in the output object indicating that the content is dynamic and
4336          * shouldn't be cached.
4337          */
4338         function disableCache() {
4339                 wfDebug( "Parser output marked as uncacheable.\n" );
4340                 $this->mOutput->mCacheTime = -1;
4341         }
4342
4343         /**#@+
4344          * Callback from the Sanitizer for expanding items found in HTML attribute
4345          * values, so they can be safely tested and escaped.
4346          * @param string $text
4347          * @param array $args
4348          * @return string
4349          * @private
4350          */
4351         function attributeStripCallback( &$text, $args ) {
4352                 $text = $this->replaceVariables( $text, $args );
4353                 $text = $this->mStripState->unstripBoth( $text );
4354                 return $text;
4355         }
4356
4357         /**#@-*/
4358
4359         /**#@+
4360          * Accessor/mutator
4361          */
4362         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4363         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4364         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4365         /**#@-*/
4366
4367         /**#@+
4368          * Accessor
4369          */
4370         function getTags() { return array_keys( $this->mTagHooks ); }
4371         /**#@-*/
4372
4373
4374         /**
4375          * Break wikitext input into sections, and either pull or replace
4376          * some particular section's text.
4377          *
4378          * External callers should use the getSection and replaceSection methods.
4379          *
4380          * @param $text Page wikitext
4381          * @param $section Numbered section. 0 pulls the text before the first
4382          *                 heading; other numbers will pull the given section
4383          *                 along with its lower-level subsections.
4384          * @param $mode One of "get" or "replace"
4385          * @param $newtext Replacement text for section data.
4386          * @return string for "get", the extracted section text.
4387          *                for "replace", the whole page with the section replaced.
4388          */
4389         private function extractSections( $text, $section, $mode, $newtext='' ) {
4390                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4391                 # comments to be stripped as well)
4392                 $stripState = new StripState;
4393
4394                 $oldOutputType = $this->mOutputType;
4395                 $oldOptions = $this->mOptions;
4396                 $this->mOptions = new ParserOptions();
4397                 $this->setOutputType( OT_WIKI );
4398
4399                 $striptext = $this->strip( $text, $stripState, true );
4400
4401                 $this->setOutputType( $oldOutputType );
4402                 $this->mOptions = $oldOptions;
4403
4404                 # now that we can be sure that no pseudo-sections are in the source,
4405                 # split it up by section
4406                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4407                 $comment = "(?:$uniq-!--.*?QINU)";
4408                 $secs = preg_split(
4409                 /*
4410                         "/
4411                         ^(
4412                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4413                         (?:
4414                                 (=+) # Should this be limited to 6?
4415                                 .+?  # Section title...
4416                                 \\2  # Ending = count must match start
4417                         |
4418                                 ^
4419                                 <h([1-6])\b.*?>
4420                                 .*?
4421                                 <\/h\\3\s*>
4422                         )
4423                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4424                         )$
4425                         /mix",
4426                 */
4427                         "/
4428                         (
4429                                 ^
4430                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4431                                 (=+) # Should this be limited to 6?
4432                                 .+?  # Section title...
4433                                 \\2  # Ending = count must match start
4434                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4435                                 $
4436                         |
4437                                 <h([1-6])\b.*?>
4438                                 .*?
4439                                 <\/h\\3\s*>
4440                         )
4441                         /mix",
4442                         $striptext, -1,
4443                         PREG_SPLIT_DELIM_CAPTURE);
4444
4445                 if( $mode == "get" ) {
4446                         if( $section == 0 ) {
4447                                 // "Section 0" returns the content before any other section.
4448                                 $rv = $secs[0];
4449                         } else {
4450                                 $rv = "";
4451                         }
4452                 } elseif( $mode == "replace" ) {
4453                         if( $section == 0 ) {
4454                                 $rv = $newtext . "\n\n";
4455                                 $remainder = true;
4456                         } else {
4457                                 $rv = $secs[0];
4458                                 $remainder = false;
4459                         }
4460                 }
4461                 $count = 0;
4462                 $sectionLevel = 0;
4463                 for( $index = 1; $index < count( $secs ); ) {
4464                         $headerLine = $secs[$index++];
4465                         if( $secs[$index] ) {
4466                                 // A wiki header
4467                                 $headerLevel = strlen( $secs[$index++] );
4468                         } else {
4469                                 // An HTML header
4470                                 $index++;
4471                                 $headerLevel = intval( $secs[$index++] );
4472                         }
4473                         $content = $secs[$index++];
4474
4475                         $count++;
4476                         if( $mode == "get" ) {
4477                                 if( $count == $section ) {
4478                                         $rv = $headerLine . $content;
4479                                         $sectionLevel = $headerLevel;
4480                                 } elseif( $count > $section ) {
4481                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4482                                                 $rv .= $headerLine . $content;
4483                                         } else {
4484                                                 // Broke out to a higher-level section
4485                                                 break;
4486                                         }
4487                                 }
4488                         } elseif( $mode == "replace" ) {
4489                                 if( $count < $section ) {
4490                                         $rv .= $headerLine . $content;
4491                                 } elseif( $count == $section ) {
4492                                         $rv .= $newtext . "\n\n";
4493                                         $sectionLevel = $headerLevel;
4494                                 } elseif( $count > $section ) {
4495                                         if( $headerLevel <= $sectionLevel ) {
4496                                                 // Passed the section's sub-parts.
4497                                                 $remainder = true;
4498                                         }
4499                                         if( $remainder ) {
4500                                                 $rv .= $headerLine . $content;
4501                                         }
4502                                 }
4503                         }
4504                 }
4505                 # reinsert stripped tags
4506                 $rv = trim( $stripState->unstripBoth( $rv ) );
4507                 return $rv;
4508         }
4509
4510         /**
4511          * This function returns the text of a section, specified by a number ($section).
4512          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4513          * the first section before any such heading (section 0).
4514          *
4515          * If a section contains subsections, these are also returned.
4516          *
4517          * @param $text String: text to look in
4518          * @param $section Integer: section number
4519          * @return string text of the requested section
4520          */
4521         function getSection( $text, $section ) {
4522                 return $this->extractSections( $text, $section, "get" );
4523         }
4524
4525         function replaceSection( $oldtext, $section, $text ) {
4526                 return $this->extractSections( $oldtext, $section, "replace", $text );
4527         }
4528
4529         /**
4530          * Get the timestamp associated with the current revision, adjusted for
4531          * the default server-local timestamp
4532          */
4533         function getRevisionTimestamp() {
4534                 if ( is_null( $this->mRevisionTimestamp ) ) {
4535                         wfProfileIn( __METHOD__ );
4536                         global $wgContLang;
4537                         $dbr =& wfGetDB( DB_SLAVE );
4538                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4539                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4540
4541                         // Normalize timestamp to internal MW format for timezone processing.
4542                         // This has the added side-effect of replacing a null value with
4543                         // the current time, which gives us more sensible behavior for
4544                         // previews.
4545                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4546
4547                         // The cryptic '' timezone parameter tells to use the site-default
4548                         // timezone offset instead of the user settings.
4549                         //
4550                         // Since this value will be saved into the parser cache, served
4551                         // to other users, and potentially even used inside links and such,
4552                         // it needs to be consistent for all visitors.
4553                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4554
4555                         wfProfileOut( __METHOD__ );
4556                 }
4557                 return $this->mRevisionTimestamp;
4558         }
4559 }
4560
4561 /**
4562  * @todo document
4563  * @package MediaWiki
4564  */
4565 class ParserOutput
4566 {
4567         var $mText,             # The output text
4568                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4569                 $mCategories,       # Map of category names to sort keys
4570                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4571                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4572                 $mVersion,          # Compatibility check
4573                 $mTitleText,        # title text of the chosen language variant
4574                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4575                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4576                 $mImages,           # DB keys of the images used, in the array key only
4577                 $mExternalLinks,    # External link URLs, in the key only
4578                 $mHTMLtitle,            # Display HTML title
4579                 $mSubtitle,                     # Additional subtitle
4580                 $mNewSection,           # Show a new section link?
4581                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4582
4583         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4584                 $containsOldMagic = false, $titletext = '' )
4585         {
4586                 $this->mText = $text;
4587                 $this->mLanguageLinks = $languageLinks;
4588                 $this->mCategories = $categoryLinks;
4589                 $this->mContainsOldMagic = $containsOldMagic;
4590                 $this->mCacheTime = '';
4591                 $this->mVersion = MW_PARSER_VERSION;
4592                 $this->mTitleText = $titletext;
4593                 $this->mLinks = array();
4594                 $this->mTemplates = array();
4595                 $this->mImages = array();
4596                 $this->mExternalLinks = array();
4597                 $this->mHTMLtitle = "" ;
4598                 $this->mSubtitle = "" ;
4599                 $this->mNewSection = false;
4600                 $this->mNoGallery = false;
4601         }
4602
4603         function getText()                   { return $this->mText; }
4604         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4605         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4606         function &getCategories()            { return $this->mCategories; }
4607         function getCacheTime()              { return $this->mCacheTime; }
4608         function getTitleText()              { return $this->mTitleText; }
4609         function &getLinks()                 { return $this->mLinks; }
4610         function &getTemplates()             { return $this->mTemplates; }
4611         function &getImages()                { return $this->mImages; }
4612         function &getExternalLinks()         { return $this->mExternalLinks; }
4613         function getNoGallery()              { return $this->mNoGallery; }
4614         function getSubtitle()               { return $this->mSubtitle; }
4615
4616         function containsOldMagic()          { return $this->mContainsOldMagic; }
4617         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4618         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4619         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4620         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4621         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4622         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4623         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4624
4625         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4626         function addImage( $name )           { $this->mImages[$name] = 1; }
4627         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4628         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4629
4630         function setNewSection( $value ) {
4631                 $this->mNewSection = (bool)$value;
4632         }
4633         function getNewSection() {
4634                 return (bool)$this->mNewSection;
4635         }
4636
4637         function addLink( $title, $id = null ) {
4638                 $ns = $title->getNamespace();
4639                 $dbk = $title->getDBkey();
4640                 if ( !isset( $this->mLinks[$ns] ) ) {
4641                         $this->mLinks[$ns] = array();
4642                 }
4643                 if ( is_null( $id ) ) {
4644                         $id = $title->getArticleID();
4645                 }
4646                 $this->mLinks[$ns][$dbk] = $id;
4647         }
4648
4649         function addTemplate( $title, $id ) {
4650                 $ns = $title->getNamespace();
4651                 $dbk = $title->getDBkey();
4652                 if ( !isset( $this->mTemplates[$ns] ) ) {
4653                         $this->mTemplates[$ns] = array();
4654                 }
4655                 $this->mTemplates[$ns][$dbk] = $id;
4656         }
4657
4658         /**
4659          * Return true if this cached output object predates the global or
4660          * per-article cache invalidation timestamps, or if it comes from
4661          * an incompatible older version.
4662          *
4663          * @param string $touched the affected article's last touched timestamp
4664          * @return bool
4665          * @public
4666          */
4667         function expired( $touched ) {
4668                 global $wgCacheEpoch;
4669                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4670                        $this->getCacheTime() < $touched ||
4671                        $this->getCacheTime() <= $wgCacheEpoch ||
4672                        !isset( $this->mVersion ) ||
4673                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4674         }
4675 }
4676
4677 /**
4678  * Set options of the Parser
4679  * @todo document
4680  * @package MediaWiki
4681  */
4682 class ParserOptions
4683 {
4684         # All variables are supposed to be private in theory, although in practise this is not the case.
4685         var $mUseTeX;                    # Use texvc to expand <math> tags
4686         var $mUseDynamicDates;           # Use DateFormatter to format dates
4687         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4688         var $mAllowExternalImages;       # Allow external images inline
4689         var $mAllowExternalImagesFrom;   # If not, any exception?
4690         var $mSkin;                      # Reference to the preferred skin
4691         var $mDateFormat;                # Date format index
4692         var $mEditSection;               # Create "edit section" links
4693         var $mNumberHeadings;            # Automatically number headings
4694         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4695         var $mTidy;                      # Ask for tidy cleanup
4696         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4697         var $mMaxIncludeSize;            # Maximum size of template expansions, in bytes
4698         var $mRemoveComments;            # Remove HTML comments. ONLY APPLIES TO PREPROCESS OPERATIONS
4699
4700         var $mUser;                      # Stored user object, just used to initialise the skin
4701
4702         function getUseTeX()                        { return $this->mUseTeX; }
4703         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4704         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4705         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4706         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4707         function getEditSection()                   { return $this->mEditSection; }
4708         function getNumberHeadings()                { return $this->mNumberHeadings; }
4709         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4710         function getTidy()                          { return $this->mTidy; }
4711         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4712         function getMaxIncludeSize()                { return $this->mMaxIncludeSize; }
4713         function getRemoveComments()                { return $this->mRemoveComments; }
4714
4715         function &getSkin() {
4716                 if ( !isset( $this->mSkin ) ) {
4717                         $this->mSkin = $this->mUser->getSkin();
4718                 }
4719                 return $this->mSkin;
4720         }
4721
4722         function getDateFormat() {
4723                 if ( !isset( $this->mDateFormat ) ) {
4724                         $this->mDateFormat = $this->mUser->getDatePreference();
4725                 }
4726                 return $this->mDateFormat;
4727         }
4728
4729         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4730         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4731         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4732         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4733         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4734         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4735         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4736         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4737         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4738         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4739         function setSkin( $x )                      { $this->mSkin = $x; }
4740         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4741         function setMaxIncludeSize( $x )            { return wfSetVar( $this->mMaxIncludeSize, $x ); }
4742         function setRemoveComments( $x )            { return wfSetVar( $this->mRemoveComments, $x ); }
4743
4744         function ParserOptions( $user = null ) {
4745                 $this->initialiseFromUser( $user );
4746         }
4747
4748         /**
4749          * Get parser options
4750          * @static
4751          */
4752         static function newFromUser( $user ) {
4753                 return new ParserOptions( $user );
4754         }
4755
4756         /** Get user options */
4757         function initialiseFromUser( $userInput ) {
4758                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4759                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
4760                 $fname = 'ParserOptions::initialiseFromUser';
4761                 wfProfileIn( $fname );
4762                 if ( !$userInput ) {
4763                         global $wgUser;
4764                         if ( isset( $wgUser ) ) {
4765                                 $user = $wgUser;
4766                         } else {
4767                                 $user = new User;
4768                         }
4769                 } else {
4770                         $user =& $userInput;
4771                 }
4772
4773                 $this->mUser = $user;
4774
4775                 $this->mUseTeX = $wgUseTeX;
4776                 $this->mUseDynamicDates = $wgUseDynamicDates;
4777                 $this->mInterwikiMagic = $wgInterwikiMagic;
4778                 $this->mAllowExternalImages = $wgAllowExternalImages;
4779                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4780                 $this->mSkin = null; # Deferred
4781                 $this->mDateFormat = null; # Deferred
4782                 $this->mEditSection = true;
4783                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4784                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4785                 $this->mTidy = false;
4786                 $this->mInterfaceMessage = false;
4787                 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
4788                 $this->mRemoveComments = true;
4789                 wfProfileOut( $fname );
4790         }
4791 }
4792
4793 class OnlyIncludeReplacer {
4794         var $output = '';
4795
4796         function replace( $matches ) {
4797                 if ( substr( $matches[1], -1 ) == "\n" ) {
4798                         $this->output .= substr( $matches[1], 0, -1 );
4799                 } else {
4800                         $this->output .= $matches[1];
4801                 }
4802         }
4803 }
4804
4805 class StripState {
4806         var $general, $nowiki;
4807
4808         function __construct() {
4809                 $this->general = new ReplacementArray;
4810                 $this->nowiki = new ReplacementArray;
4811         }
4812
4813         function unstripGeneral( $text ) {
4814                 wfProfileIn( __METHOD__ );
4815                 $text = $this->general->replace( $text );
4816                 wfProfileOut( __METHOD__ );
4817                 return $text;
4818         }
4819
4820         function unstripNoWiki( $text ) {
4821                 wfProfileIn( __METHOD__ );
4822                 $text = $this->nowiki->replace( $text );
4823                 wfProfileOut( __METHOD__ );
4824                 return $text;
4825         }
4826
4827         function unstripBoth( $text ) {
4828                 wfProfileIn( __METHOD__ );
4829                 $text = $this->general->replace( $text );
4830                 $text = $this->nowiki->replace( $text );
4831                 wfProfileOut( __METHOD__ );
4832                 return $text;
4833         }
4834 }
4835
4836 ?>