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