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