includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.6.1' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space, but excluding newlines
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  55         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  56 define( 'EXT_IMAGE_REGEX',
  57         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  58         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  59         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  60 );
  61
  62 // State constants for the definition list colon extraction
  63 define( 'MW_COLON_STATE_TEXT', 0 );
  64 define( 'MW_COLON_STATE_TAG', 1 );
  65 define( 'MW_COLON_STATE_TAGSTART', 2 );
  66 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  67 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  68 define( 'MW_COLON_STATE_COMMENT', 5 );
  69 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  70 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  71
  72 /**
  73  * PHP Parser
  74  *
  75  * Processes wiki markup
  76  *
  77  * <pre>
  78  * There are three main entry points into the Parser class:
  79  * parse()
  80  *   produces HTML output
  81  * preSaveTransform().
  82  *   produces altered wiki markup.
  83  * transformMsg()
  84  *   performs brace substitution on MediaWiki messages
  85  *
  86  * Globals used:
  87  *    objects:   $wgLang, $wgContLang
  88  *
  89  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  90  *
  91  * settings:
  92  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  93  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  94  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  95  *
  96  *  * only within ParserOptions
  97  * </pre>
  98  *
  99  * @package MediaWiki
 100  */
 101 class Parser
 102 {
 103         /**#@+
 104          * @private
 105          */
 106         # Persistent:
 107         var $mTagHooks, $mFunctionHooks;
 108
 109         # Cleared with clearState():
 110         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 111         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 112         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 113         var $mTemplates,        // cache of already loaded templates, avoids
 114                                 // multiple SQL queries for the same string
 115             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 116                                 // in this path. Used for loop detection.
 117
 118         # Temporary
 119         # These are variables reset at least once per parse regardless of $clearState
 120         var $mOptions,      // ParserOptions object
 121                 $mTitle,        // Title context, used for self-link rendering and similar things
 122                 $mOutputType,   // Output type, one of the OT_xxx constants
 123                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 124
 125         /**#@-*/
 126
 127         /**
 128          * Constructor
 129          *
 130          * @public
 131          */
 132         function Parser() {
 133                 $this->mTagHooks = array();
 134                 $this->mFunctionHooks = array();
 135                 $this->clearState();
 136                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 137         }
 138
 139         /**
 140          * Clear Parser state
 141          *
 142          * @private
 143          */
 144         function clearState() {
 145                 $this->mOutput = new ParserOutput;
 146                 $this->mAutonumber = 0;
 147                 $this->mLastSection = '';
 148                 $this->mDTopen = false;
 149                 $this->mVariables = false;
 150                 $this->mIncludeCount = array();
 151                 $this->mStripState = array();
 152                 $this->mArgStack = array();
 153                 $this->mInPre = false;
 154                 $this->mInterwikiLinkHolders = array(
 155                         'texts' => array(),
 156                         'titles' => array()
 157                 );
 158                 $this->mLinkHolders = array(
 159                         'namespaces' => array(),
 160                         'dbkeys' => array(),
 161                         'queries' => array(),
 162                         'texts' => array(),
 163                         'titles' => array()
 164                 );
 165                 $this->mRevisionId = null;
 166
 167                 /**
 168                  * Prefix for temporary replacement strings for the multipass parser.
 169                  * \x07 should never appear in input as it's disallowed in XML.
 170                  * Using it at the front also gives us a little extra robustness
 171                  * since it shouldn't match when butted up against identifier-like
 172                  * string constructs.
 173                  */
 174                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 175
 176                 # Clear these on every parse, bug 4549
 177                 $this->mTemplates = array();
 178                 $this->mTemplatePath = array();
 179
 180                 $this->mShowToc = true;
 181                 $this->mForceTocPosition = false;
 182
 183                 wfRunHooks( 'ParserClearState', array( &$this ) );
 184         }
 185
 186         /**
 187          * Accessor for mUniqPrefix.
 188          *
 189          * @public
 190          */
 191         function UniqPrefix() {
 192                 return $this->mUniqPrefix;
 193         }
 194
 195         /**
 196          * Convert wikitext to HTML
 197          * Do not call this function recursively.
 198          *
 199          * @private
 200          * @param string $text Text we want to parse
 201          * @param Title &$title A title object
 202          * @param array $options
 203          * @param boolean $linestart
 204          * @param boolean $clearState
 205          * @param int $revid number to pass in {{REVISIONID}}
 206          * @return ParserOutput a ParserOutput
 207          */
 208         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 209                 /**
 210                  * First pass--just handle <nowiki> sections, pass the rest off
 211                  * to internalParse() which does all the real work.
 212                  */
 213
 214                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 215                 $fname = 'Parser::parse';
 216                 wfProfileIn( $fname );
 217
 218                 if ( $clearState ) {
 219                         $this->clearState();
 220                 }
 221
 222                 $this->mOptions = $options;
 223                 $this->mTitle =& $title;
 224                 $this->mRevisionId = $revid;
 225                 $this->mOutputType = OT_HTML;
 226
 227                 //$text = $this->strip( $text, $this->mStripState );
 228                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 229                 $x =& $this->mStripState;
 230
 231                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 232                 $text = $this->strip( $text, $x );
 233                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 234
 235                 # Hook to suspend the parser in this state
 236                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 237                         wfProfileOut( $fname );
 238                         return $text ;
 239                 }
 240
 241                 $text = $this->internalParse( $text );
 242
 243                 $text = $this->unstrip( $text, $this->mStripState );
 244
 245                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 246                 $fixtags = array(
 247                         # french spaces, last one Guillemet-left
 248                         # only if there is something before the space
 249                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 250                         # french spaces, Guillemet-right
 251                         '/(\\302\\253) /' => '\\1&nbsp;',
 252                 );
 253                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 254
 255                 # only once and last
 256                 $text = $this->doBlockLevels( $text, $linestart );
 257
 258                 $this->replaceLinkHolders( $text );
 259
 260                 # the position of the parserConvert() call should not be changed. it
 261                 # assumes that the links are all replaced and the only thing left
 262                 # is the <nowiki> mark.
 263                 # Side-effects: this calls $this->mOutput->setTitleText()
 264                 $text = $wgContLang->parserConvert( $text, $this );
 265
 266                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 267
 268                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 269
 270                 $text = Sanitizer::normalizeCharReferences( $text );
 271
 272                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 273                         $text = Parser::tidy($text);
 274                 } else {
 275                         # attempt to sanitize at least some nesting problems
 276                         # (bug #2702 and quite a few others)
 277                         $tidyregs = array(
 278                                 # ''Something [http://www.cool.com cool''] -->
 279                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 280                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 281                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 282                                 # fix up an anchor inside another anchor, only
 283                                 # at least for a single single nested link (bug 3695)
 284                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 285                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 286                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 287                                 # contains a div, so fix it up here; replace
 288                                 # div with escaped text
 289                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 290                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 291                                 # remove empty italic or bold tag pairs, some
 292                                 # introduced by rules above
 293                                 '/<([bi])><\/\\1>/' => ''
 294                         );
 295
 296                         $text = preg_replace(
 297                                 array_keys( $tidyregs ),
 298                                 array_values( $tidyregs ),
 299                                 $text );
 300                 }
 301
 302                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 303
 304                 $this->mOutput->setText( $text );
 305                 wfProfileOut( $fname );
 306
 307                 return $this->mOutput;
 308         }
 309
 310         /**
 311          * Get a random string
 312          *
 313          * @private
 314          * @static
 315          */
 316         function getRandomString() {
 317                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 318         }
 319
 320         function &getTitle() { return $this->mTitle; }
 321         function getOptions() { return $this->mOptions; }
 322
 323         /**
 324          * Replaces all occurrences of HTML-style comments and the given tags
 325          * in the text with a random marker and returns teh next text. The output
 326          * parameter $matches will be an associative array filled with data in
 327          * the form:
 328          *   'UNIQ-xxxxx' => array(
 329          *     'element',
 330          *     'tag content',
 331          *     array( 'param' => 'x' ),
 332          *     '<element param="x">tag content</element>' ) )
 333          *
 334          * @param $elements list of element names. Comments are always extracted.
 335          * @param $text Source text string.
 336          * @param $uniq_prefix
 337          *
 338          * @private
 339          * @static
 340          */
 341         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 342                 $rand = Parser::getRandomString();
 343                 $n = 1;
 344                 $stripped = '';
 345                 $matches = array();
 346
 347                 $taglist = implode( '|', $elements );
 348                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 349
 350                 while ( '' != $text ) {
 351                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 352                         $stripped .= $p[0];
 353                         if( count( $p ) < 5 ) {
 354                                 break;
 355                         }
 356                         if( count( $p ) > 5 ) {
 357                                 // comment
 358                                 $element    = $p[4];
 359                                 $attributes = '';
 360                                 $close      = '';
 361                                 $inside     = $p[5];
 362                         } else {
 363                                 // tag
 364                                 $element    = $p[1];
 365                                 $attributes = $p[2];
 366                                 $close      = $p[3];
 367                                 $inside     = $p[4];
 368                         }
 369
 370                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++) . '-QINU';
 371                         $stripped .= $marker;
 372
 373                         if ( $close === '/>' ) {
 374                                 // Empty element tag, <tag />
 375                                 $content = null;
 376                                 $text = $inside;
 377                                 $tail = null;
 378                         } else {
 379                                 if( $element == '!--' ) {
 380                                         $end = '/(-->)/';
 381                                 } else {
 382                                         $end = "/(<\\/$element\\s*>)/i";
 383                                 }
 384                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 385                                 $content = $q[0];
 386                                 if( count( $q ) < 3 ) {
 387                                         # No end tag -- let it run out to the end of the text.
 388                                         $tail = '';
 389                                         $text = '';
 390                                 } else {
 391                                         $tail = $q[1];
 392                                         $text = $q[2];
 393                                 }
 394                         }
 395
 396                         $matches[$marker] = array( $element,
 397                                 $content,
 398                                 Sanitizer::decodeTagAttributes( $attributes ),
 399                                 "<$element$attributes$close$content$tail" );
 400                 }
 401                 return $stripped;
 402         }
 403
 404         /**
 405          * Strips and renders nowiki, pre, math, hiero
 406          * If $render is set, performs necessary rendering operations on plugins
 407          * Returns the text, and fills an array with data needed in unstrip()
 408          * If the $state is already a valid strip state, it adds to the state
 409          *
 410          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 411          *  will be stripped in addition to other tags. This is important
 412          *  for section editing, where these comments cause confusion when
 413          *  counting the sections in the wikisource
 414          *
 415          * @param array dontstrip contains tags which should not be stripped;
 416          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 417          *
 418          * @private
 419          */
 420         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 421                 $render = ($this->mOutputType == OT_HTML);
 422
 423                 # Replace any instances of the placeholders
 424                 $uniq_prefix = $this->mUniqPrefix;
 425                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 426                 $commentState = array();
 427
 428                 $elements = array_merge(
 429                         array( 'nowiki', 'gallery' ),
 430                         array_keys( $this->mTagHooks ) );
 431                 global $wgRawHtml;
 432                 if( $wgRawHtml ) {
 433                         $elements[] = 'html';
 434                 }
 435                 if( $this->mOptions->getUseTeX() ) {
 436                         $elements[] = 'math';
 437                 }
 438
 439                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 440                 foreach ( $elements AS $k => $v ) {
 441                         if ( !in_array ( $v , $dontstrip ) ) continue;
 442                         unset ( $elements[$k] );
 443                 }
 444
 445                 $matches = array();
 446                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 447
 448                 foreach( $matches as $marker => $data ) {
 449                         list( $element, $content, $params, $tag ) = $data;
 450                         if( $render ) {
 451                                 $tagName = strtolower( $element );
 452                                 switch( $tagName ) {
 453                                 case '!--':
 454                                         // Comment
 455                                         if( substr( $tag, -3 ) == '-->' ) {
 456                                                 $output = $tag;
 457                                         } else {
 458                                                 // Unclosed comment in input.
 459                                                 // Close it so later stripping can remove it
 460                                                 $output = "$tag-->";
 461                                         }
 462                                         break;
 463                                 case 'html':
 464                                         if( $wgRawHtml ) {
 465                                                 $output = $content;
 466                                                 break;
 467                                         }
 468                                         // Shouldn't happen otherwise. :)
 469                                 case 'nowiki':
 470                                         $output = wfEscapeHTMLTagsOnly( $content );
 471                                         break;
 472                                 case 'math':
 473                                         $output = MathRenderer::renderMath( $content );
 474                                         break;
 475                                 case 'gallery':
 476                                         $output = $this->renderImageGallery( $content );
 477                                         break;
 478                                 default:
 479                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 480                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 481                                                         array( $content, $params, $this ) );
 482                                         } else {
 483                                                 throw new MWException( "Invalid call hook $element" );
 484                                         }
 485                                 }
 486                         } else {
 487                                 // Just stripping tags; keep the source
 488                                 $output = $tag;
 489                         }
 490                         if( !$stripcomments && $element == '!--' ) {
 491                                 $commentState[$marker] = $output;
 492                         } else {
 493                                 $state[$element][$marker] = $output;
 494                         }
 495                 }
 496
 497                 # Unstrip comments unless explicitly told otherwise.
 498                 # (The comments are always stripped prior to this point, so as to
 499                 # not invoke any extension tags / parser hooks contained within
 500                 # a comment.)
 501                 if ( !$stripcomments ) {
 502                         // Put them all back and forget them
 503                         $text = strtr( $text, $commentState );
 504                 }
 505
 506                 return $text;
 507         }
 508
 509         /**
 510          * Restores pre, math, and other extensions removed by strip()
 511          *
 512          * always call unstripNoWiki() after this one
 513          * @private
 514          */
 515         function unstrip( $text, &$state ) {
 516                 if ( !is_array( $state ) ) {
 517                         return $text;
 518                 }
 519
 520                 $replacements = array();
 521                 foreach( $state as $tag => $contentDict ) {
 522                         if( $tag != 'nowiki' && $tag != 'html' ) {
 523                                 foreach( $contentDict as $uniq => $content ) {
 524                                         $replacements[$uniq] = $content;
 525                                 }
 526                         }
 527                 }
 528                 $text = strtr( $text, $replacements );
 529
 530                 return $text;
 531         }
 532
 533         /**
 534          * Always call this after unstrip() to preserve the order
 535          *
 536          * @private
 537          */
 538         function unstripNoWiki( $text, &$state ) {
 539                 if ( !is_array( $state ) ) {
 540                         return $text;
 541                 }
 542
 543                 $replacements = array();
 544                 foreach( $state as $tag => $contentDict ) {
 545                         if( $tag == 'nowiki' || $tag == 'html' ) {
 546                                 foreach( $contentDict as $uniq => $content ) {
 547                                         $replacements[$uniq] = $content;
 548                                 }
 549                         }
 550                 }
 551                 $text = strtr( $text, $replacements );
 552
 553                 return $text;
 554         }
 555
 556         /**
 557          * Add an item to the strip state
 558          * Returns the unique tag which must be inserted into the stripped text
 559          * The tag will be replaced with the original text in unstrip()
 560          *
 561          * @private
 562          */
 563         function insertStripItem( $text, &$state ) {
 564                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 565                 if ( !$state ) {
 566                         $state = array();
 567                 }
 568                 $state['item'][$rnd] = $text;
 569                 return $rnd;
 570         }
 571
 572         /**
 573          * Interface with html tidy, used if $wgUseTidy = true.
 574          * If tidy isn't able to correct the markup, the original will be
 575          * returned in all its glory with a warning comment appended.
 576          *
 577          * Either the external tidy program or the in-process tidy extension
 578          * will be used depending on availability. Override the default
 579          * $wgTidyInternal setting to disable the internal if it's not working.
 580          *
 581          * @param string $text Hideous HTML input
 582          * @return string Corrected HTML output
 583          * @public
 584          * @static
 585          */
 586         function tidy( $text ) {
 587                 global $wgTidyInternal;
 588                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 589 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 590 '<head><title>test</title></head><body>'.$text.'</body></html>';
 591                 if( $wgTidyInternal ) {
 592                         $correctedtext = Parser::internalTidy( $wrappedtext );
 593                 } else {
 594                         $correctedtext = Parser::externalTidy( $wrappedtext );
 595                 }
 596                 if( is_null( $correctedtext ) ) {
 597                         wfDebug( "Tidy error detected!\n" );
 598                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 599                 }
 600                 return $correctedtext;
 601         }
 602
 603         /**
 604          * Spawn an external HTML tidy process and get corrected markup back from it.
 605          *
 606          * @private
 607          * @static
 608          */
 609         function externalTidy( $text ) {
 610                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 611                 $fname = 'Parser::externalTidy';
 612                 wfProfileIn( $fname );
 613
 614                 $cleansource = '';
 615                 $opts = ' -utf8';
 616
 617                 $descriptorspec = array(
 618                         0 => array('pipe', 'r'),
 619                         1 => array('pipe', 'w'),
 620                         2 => array('file', '/dev/null', 'a')
 621                 );
 622                 $pipes = array();
 623                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 624                 if (is_resource($process)) {
 625                         // Theoretically, this style of communication could cause a deadlock
 626                         // here. If the stdout buffer fills up, then writes to stdin could
 627                         // block. This doesn't appear to happen with tidy, because tidy only
 628                         // writes to stdout after it's finished reading from stdin. Search
 629                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 630                         fwrite($pipes[0], $text);
 631                         fclose($pipes[0]);
 632                         while (!feof($pipes[1])) {
 633                                 $cleansource .= fgets($pipes[1], 1024);
 634                         }
 635                         fclose($pipes[1]);
 636                         proc_close($process);
 637                 }
 638
 639                 wfProfileOut( $fname );
 640
 641                 if( $cleansource == '' && $text != '') {
 642                         // Some kind of error happened, so we couldn't get the corrected text.
 643                         // Just give up; we'll use the source text and append a warning.
 644                         return null;
 645                 } else {
 646                         return $cleansource;
 647                 }
 648         }
 649
 650         /**
 651          * Use the HTML tidy PECL extension to use the tidy library in-process,
 652          * saving the overhead of spawning a new process. Currently written to
 653          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 654          *
 655          * 'pear install tidy' should be able to compile the extension module.
 656          *
 657          * @private
 658          * @static
 659          */
 660         function internalTidy( $text ) {
 661                 global $wgTidyConf;
 662                 $fname = 'Parser::internalTidy';
 663                 wfProfileIn( $fname );
 664
 665                 tidy_load_config( $wgTidyConf );
 666                 tidy_set_encoding( 'utf8' );
 667                 tidy_parse_string( $text );
 668                 tidy_clean_repair();
 669                 if( tidy_get_status() == 2 ) {
 670                         // 2 is magic number for fatal error
 671                         // http://www.php.net/manual/en/function.tidy-get-status.php
 672                         $cleansource = null;
 673                 } else {
 674                         $cleansource = tidy_get_output();
 675                 }
 676                 wfProfileOut( $fname );
 677                 return $cleansource;
 678         }
 679
 680         /**
 681          * parse the wiki syntax used to render tables
 682          *
 683          * @private
 684          */
 685         function doTableStuff ( $t ) {
 686                 $fname = 'Parser::doTableStuff';
 687                 wfProfileIn( $fname );
 688
 689                 $t = explode ( "\n" , $t ) ;
 690                 $td = array () ; # Is currently a td tag open?
 691                 $ltd = array () ; # Was it TD or TH?
 692                 $tr = array () ; # Is currently a tr tag open?
 693                 $ltr = array () ; # tr attributes
 694                 $has_opened_tr = array(); # Did this table open a <tr> element?
 695                 $indent_level = 0; # indent level of the table
 696                 foreach ( $t AS $k => $x )
 697                 {
 698                         $x = trim ( $x ) ;
 699                         $fc = substr ( $x , 0 , 1 ) ;
 700                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 701                                 $indent_level = strlen( $matches[1] );
 702
 703                                 $attributes = $this->unstripForHTML( $matches[2] );
 704
 705                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 706                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 707                                 array_push ( $td , false ) ;
 708                                 array_push ( $ltd , '' ) ;
 709                                 array_push ( $tr , false ) ;
 710                                 array_push ( $ltr , '' ) ;
 711                                 array_push ( $has_opened_tr, false );
 712                         }
 713                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 714                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 715                                 $z = "</table>" . substr ( $x , 2);
 716                                 $l = array_pop ( $ltd ) ;
 717                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 718                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 719                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 720                                 array_pop ( $ltr ) ;
 721                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 722                         }
 723                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 724                                 $x = substr ( $x , 1 ) ;
 725                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 726                                 $z = '' ;
 727                                 $l = array_pop ( $ltd ) ;
 728                                 array_pop ( $has_opened_tr );
 729                                 array_push ( $has_opened_tr , true ) ;
 730                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 731                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 732                                 array_pop ( $ltr ) ;
 733                                 $t[$k] = $z ;
 734                                 array_push ( $tr , false ) ;
 735                                 array_push ( $td , false ) ;
 736                                 array_push ( $ltd , '' ) ;
 737                                 $attributes = $this->unstripForHTML( $x );
 738                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 739                         }
 740                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 741                                 # $x is a table row
 742                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 743                                         $fc = '+' ;
 744                                         $x = substr ( $x , 1 ) ;
 745                                 }
 746                                 $after = substr ( $x , 1 ) ;
 747                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 748
 749                                 // Split up multiple cells on the same line.
 750                                 // FIXME: This can result in improper nesting of tags processed
 751                                 // by earlier parser steps, but should avoid splitting up eg
 752                                 // attribute values containing literal "||".
 753                                 $after = wfExplodeMarkup( '||', $after );
 754
 755                                 $t[$k] = '' ;
 756
 757                                 # Loop through each table cell
 758                                 foreach ( $after AS $theline )
 759                                 {
 760                                         $z = '' ;
 761                                         if ( $fc != '+' )
 762                                         {
 763                                                 $tra = array_pop ( $ltr ) ;
 764                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 765                                                 array_push ( $tr , true ) ;
 766                                                 array_push ( $ltr , '' ) ;
 767                                                 array_pop ( $has_opened_tr );
 768                                                 array_push ( $has_opened_tr , true ) ;
 769                                         }
 770
 771                                         $l = array_pop ( $ltd ) ;
 772                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 773                                         if ( $fc == '|' ) $l = 'td' ;
 774                                         else if ( $fc == '!' ) $l = 'th' ;
 775                                         else if ( $fc == '+' ) $l = 'caption' ;
 776                                         else $l = '' ;
 777                                         array_push ( $ltd , $l ) ;
 778
 779                                         # Cell parameters
 780                                         $y = explode ( '|' , $theline , 2 ) ;
 781                                         # Note that a '|' inside an invalid link should not
 782                                         # be mistaken as delimiting cell parameters
 783                                         if ( strpos( $y[0], '[[' ) !== false ) {
 784                                                 $y = array ($theline);
 785                                         }
 786                                         if ( count ( $y ) == 1 )
 787                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 788                                         else {
 789                                                 $attributes = $this->unstripForHTML( $y[0] );
 790                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 791                                         }
 792                                         $t[$k] .= $y ;
 793                                         array_push ( $td , true ) ;
 794                                 }
 795                         }
 796                 }
 797
 798                 # Closing open td, tr && table
 799                 while ( count ( $td ) > 0 )
 800                 {
 801                         $l = array_pop ( $ltd ) ;
 802                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 803                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 804                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 805                         $t[] = '</table>' ;
 806                 }
 807
 808                 $t = implode ( "\n" , $t ) ;
 809                 # special case: don't return empty table
 810                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 811                         $t = '';
 812                 wfProfileOut( $fname );
 813                 return $t ;
 814         }
 815
 816         /**
 817          * Helper function for parse() that transforms wiki markup into
 818          * HTML. Only called for $mOutputType == OT_HTML.
 819          *
 820          * @private
 821          */
 822         function internalParse( $text ) {
 823                 $args = array();
 824                 $isMain = true;
 825                 $fname = 'Parser::internalParse';
 826                 wfProfileIn( $fname );
 827
 828                 # Remove <noinclude> tags and <includeonly> sections
 829                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 830                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 831                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 832
 833                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 834
 835                 $text = $this->replaceVariables( $text, $args );
 836
 837                 // Tables need to come after variable replacement for things to work
 838                 // properly; putting them before other transformations should keep
 839                 // exciting things like link expansions from showing up in surprising
 840                 // places.
 841                 $text = $this->doTableStuff( $text );
 842
 843                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 844
 845                 $text = $this->stripToc( $text );
 846                 $text = $this->doHeadings( $text );
 847                 if($this->mOptions->getUseDynamicDates()) {
 848                         $df =& DateFormatter::getInstance();
 849                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 850                 }
 851                 $text = $this->doAllQuotes( $text );
 852                 $text = $this->replaceInternalLinks( $text );
 853                 $text = $this->replaceExternalLinks( $text );
 854
 855                 # replaceInternalLinks may sometimes leave behind
 856                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 857                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 858
 859                 $text = $this->doMagicLinks( $text );
 860                 $text = $this->formatHeadings( $text, $isMain );
 861
 862                 wfProfileOut( $fname );
 863                 return $text;
 864         }
 865
 866         /**
 867          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 868          * magic external links.
 869          *
 870          * @private
 871          */
 872         function &doMagicLinks( &$text ) {
 873                 $text = $this->magicISBN( $text );
 874                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 875                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 876                 return $text;
 877         }
 878
 879         /**
 880          * Parse headers and return html
 881          *
 882          * @private
 883          */
 884         function doHeadings( $text ) {
 885                 $fname = 'Parser::doHeadings';
 886                 wfProfileIn( $fname );
 887                 for ( $i = 6; $i >= 1; --$i ) {
 888                         $h = str_repeat( '=', $i );
 889                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 890                           "<h{$i}>\\1</h{$i}>\\2", $text );
 891                 }
 892                 wfProfileOut( $fname );
 893                 return $text;
 894         }
 895
 896         /**
 897          * Replace single quotes with HTML markup
 898          * @private
 899          * @return string the altered text
 900          */
 901         function doAllQuotes( $text ) {
 902                 $fname = 'Parser::doAllQuotes';
 903                 wfProfileIn( $fname );
 904                 $outtext = '';
 905                 $lines = explode( "\n", $text );
 906                 foreach ( $lines as $line ) {
 907                         $outtext .= $this->doQuotes ( $line ) . "\n";
 908                 }
 909                 $outtext = substr($outtext, 0,-1);
 910                 wfProfileOut( $fname );
 911                 return $outtext;
 912         }
 913
 914         /**
 915          * Helper function for doAllQuotes()
 916          * @private
 917          */
 918         function doQuotes( $text ) {
 919                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 920                 if ( count( $arr ) == 1 )
 921                         return $text;
 922                 else
 923                 {
 924                         # First, do some preliminary work. This may shift some apostrophes from
 925                         # being mark-up to being text. It also counts the number of occurrences
 926                         # of bold and italics mark-ups.
 927                         $i = 0;
 928                         $numbold = 0;
 929                         $numitalics = 0;
 930                         foreach ( $arr as $r )
 931                         {
 932                                 if ( ( $i % 2 ) == 1 )
 933                                 {
 934                                         # If there are ever four apostrophes, assume the first is supposed to
 935                                         # be text, and the remaining three constitute mark-up for bold text.
 936                                         if ( strlen( $arr[$i] ) == 4 )
 937                                         {
 938                                                 $arr[$i-1] .= "'";
 939                                                 $arr[$i] = "'''";
 940                                         }
 941                                         # If there are more than 5 apostrophes in a row, assume they're all
 942                                         # text except for the last 5.
 943                                         else if ( strlen( $arr[$i] ) > 5 )
 944                                         {
 945                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 946                                                 $arr[$i] = "'''''";
 947                                         }
 948                                         # Count the number of occurrences of bold and italics mark-ups.
 949                                         # We are not counting sequences of five apostrophes.
 950                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 951                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 952                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 953                                 }
 954                                 $i++;
 955                         }
 956
 957                         # If there is an odd number of both bold and italics, it is likely
 958                         # that one of the bold ones was meant to be an apostrophe followed
 959                         # by italics. Which one we cannot know for certain, but it is more
 960                         # likely to be one that has a single-letter word before it.
 961                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 962                         {
 963                                 $i = 0;
 964                                 $firstsingleletterword = -1;
 965                                 $firstmultiletterword = -1;
 966                                 $firstspace = -1;
 967                                 foreach ( $arr as $r )
 968                                 {
 969                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 970                                         {
 971                                                 $x1 = substr ($arr[$i-1], -1);
 972                                                 $x2 = substr ($arr[$i-1], -2, 1);
 973                                                 if ($x1 == ' ') {
 974                                                         if ($firstspace == -1) $firstspace = $i;
 975                                                 } else if ($x2 == ' ') {
 976                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 977                                                 } else {
 978                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 979                                                 }
 980                                         }
 981                                         $i++;
 982                                 }
 983
 984                                 # If there is a single-letter word, use it!
 985                                 if ($firstsingleletterword > -1)
 986                                 {
 987                                         $arr [ $firstsingleletterword ] = "''";
 988                                         $arr [ $firstsingleletterword-1 ] .= "'";
 989                                 }
 990                                 # If not, but there's a multi-letter word, use that one.
 991                                 else if ($firstmultiletterword > -1)
 992                                 {
 993                                         $arr [ $firstmultiletterword ] = "''";
 994                                         $arr [ $firstmultiletterword-1 ] .= "'";
 995                                 }
 996                                 # ... otherwise use the first one that has neither.
 997                                 # (notice that it is possible for all three to be -1 if, for example,
 998                                 # there is only one pentuple-apostrophe in the line)
 999                                 else if ($firstspace > -1)
1000                                 {
1001                                         $arr [ $firstspace ] = "''";
1002                                         $arr [ $firstspace-1 ] .= "'";
1003                                 }
1004                         }
1005
1006                         # Now let's actually convert our apostrophic mush to HTML!
1007                         $output = '';
1008                         $buffer = '';
1009                         $state = '';
1010                         $i = 0;
1011                         foreach ($arr as $r)
1012                         {
1013                                 if (($i % 2) == 0)
1014                                 {
1015                                         if ($state == 'both')
1016                                                 $buffer .= $r;
1017                                         else
1018                                                 $output .= $r;
1019                                 }
1020                                 else
1021                                 {
1022                                         if (strlen ($r) == 2)
1023                                         {
1024                                                 if ($state == 'i')
1025                                                 { $output .= '</i>'; $state = ''; }
1026                                                 else if ($state == 'bi')
1027                                                 { $output .= '</i>'; $state = 'b'; }
1028                                                 else if ($state == 'ib')
1029                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1030                                                 else if ($state == 'both')
1031                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1032                                                 else # $state can be 'b' or ''
1033                                                 { $output .= '<i>'; $state .= 'i'; }
1034                                         }
1035                                         else if (strlen ($r) == 3)
1036                                         {
1037                                                 if ($state == 'b')
1038                                                 { $output .= '</b>'; $state = ''; }
1039                                                 else if ($state == 'bi')
1040                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1041                                                 else if ($state == 'ib')
1042                                                 { $output .= '</b>'; $state = 'i'; }
1043                                                 else if ($state == 'both')
1044                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1045                                                 else # $state can be 'i' or ''
1046                                                 { $output .= '<b>'; $state .= 'b'; }
1047                                         }
1048                                         else if (strlen ($r) == 5)
1049                                         {
1050                                                 if ($state == 'b')
1051                                                 { $output .= '</b><i>'; $state = 'i'; }
1052                                                 else if ($state == 'i')
1053                                                 { $output .= '</i><b>'; $state = 'b'; }
1054                                                 else if ($state == 'bi')
1055                                                 { $output .= '</i></b>'; $state = ''; }
1056                                                 else if ($state == 'ib')
1057                                                 { $output .= '</b></i>'; $state = ''; }
1058                                                 else if ($state == 'both')
1059                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1060                                                 else # ($state == '')
1061                                                 { $buffer = ''; $state = 'both'; }
1062                                         }
1063                                 }
1064                                 $i++;
1065                         }
1066                         # Now close all remaining tags.  Notice that the order is important.
1067                         if ($state == 'b' || $state == 'ib')
1068                                 $output .= '</b>';
1069                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1070                                 $output .= '</i>';
1071                         if ($state == 'bi')
1072                                 $output .= '</b>';
1073                         if ($state == 'both')
1074                                 $output .= '<b><i>'.$buffer.'</i></b>';
1075                         return $output;
1076                 }
1077         }
1078
1079         /**
1080          * Replace external links
1081          *
1082          * Note: this is all very hackish and the order of execution matters a lot.
1083          * Make sure to run maintenance/parserTests.php if you change this code.
1084          *
1085          * @private
1086          */
1087         function replaceExternalLinks( $text ) {
1088                 global $wgContLang;
1089                 $fname = 'Parser::replaceExternalLinks';
1090                 wfProfileIn( $fname );
1091
1092                 $sk =& $this->mOptions->getSkin();
1093
1094                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1095
1096                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1097
1098                 $i = 0;
1099                 while ( $i<count( $bits ) ) {
1100                         $url = $bits[$i++];
1101                         $protocol = $bits[$i++];
1102                         $text = $bits[$i++];
1103                         $trail = $bits[$i++];
1104
1105                         # The characters '<' and '>' (which were escaped by
1106                         # removeHTMLtags()) should not be included in
1107                         # URLs, per RFC 2396.
1108                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1109                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1110                                 $url = substr($url, 0, $m2[0][1]);
1111                         }
1112
1113                         # If the link text is an image URL, replace it with an <img> tag
1114                         # This happened by accident in the original parser, but some people used it extensively
1115                         $img = $this->maybeMakeExternalImage( $text );
1116                         if ( $img !== false ) {
1117                                 $text = $img;
1118                         }
1119
1120                         $dtrail = '';
1121
1122                         # Set linktype for CSS - if URL==text, link is essentially free
1123                         $linktype = ($text == $url) ? 'free' : 'text';
1124
1125                         # No link text, e.g. [http://domain.tld/some.link]
1126                         if ( $text == '' ) {
1127                                 # Autonumber if allowed. See bug #5918
1128                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1129                                         $text = '[' . ++$this->mAutonumber . ']';
1130                                         $linktype = 'autonumber';
1131                                 } else {
1132                                         # Otherwise just use the URL
1133                                         $text = htmlspecialchars( $url );
1134                                         $linktype = 'free';
1135                                 }
1136                         } else {
1137                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1138                                 # Check for trail
1139                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1140                         }
1141
1142                         $text = $wgContLang->markNoConversion($text);
1143
1144                         # Normalize any HTML entities in input. They will be
1145                         # re-escaped by makeExternalLink().
1146                         $url = Sanitizer::decodeCharReferences( $url );
1147
1148                         # Process the trail (i.e. everything after this link up until start of the next link),
1149                         # replacing any non-bracketed links
1150                         $trail = $this->replaceFreeExternalLinks( $trail );
1151
1152                         # Use the encoded URL
1153                         # This means that users can paste URLs directly into the text
1154                         # Funny characters like &ouml; aren't valid in URLs anyway
1155                         # This was changed in August 2004
1156                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1157
1158                         # Register link in the output object.
1159                         # Replace unnecessary URL escape codes with the referenced character
1160                         # This prevents spammers from hiding links from the filters
1161                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1162                         $this->mOutput->addExternalLink( $pasteurized );
1163                 }
1164
1165                 wfProfileOut( $fname );
1166                 return $s;
1167         }
1168
1169         /**
1170          * Replace anything that looks like a URL with a link
1171          * @private
1172          */
1173         function replaceFreeExternalLinks( $text ) {
1174                 global $wgContLang;
1175                 $fname = 'Parser::replaceFreeExternalLinks';
1176                 wfProfileIn( $fname );
1177
1178                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1179                 $s = array_shift( $bits );
1180                 $i = 0;
1181
1182                 $sk =& $this->mOptions->getSkin();
1183
1184                 while ( $i < count( $bits ) ){
1185                         $protocol = $bits[$i++];
1186                         $remainder = $bits[$i++];
1187
1188                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1189                                 # Found some characters after the protocol that look promising
1190                                 $url = $protocol . $m[1];
1191                                 $trail = $m[2];
1192
1193                                 # special case: handle urls as url args:
1194                                 # http://www.example.com/foo?=http://www.example.com/bar
1195                                 if(strlen($trail) == 0 &&
1196                                         isset($bits[$i]) &&
1197                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1198                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1199                                 {
1200                                         # add protocol, arg
1201                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1202                                         $i += 2;
1203                                         $trail = $m[2];
1204                                 }
1205
1206                                 # The characters '<' and '>' (which were escaped by
1207                                 # removeHTMLtags()) should not be included in
1208                                 # URLs, per RFC 2396.
1209                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1210                                         $trail = substr($url, $m2[0][1]) . $trail;
1211                                         $url = substr($url, 0, $m2[0][1]);
1212                                 }
1213
1214                                 # Move trailing punctuation to $trail
1215                                 $sep = ',;\.:!?';
1216                                 # If there is no left bracket, then consider right brackets fair game too
1217                                 if ( strpos( $url, '(' ) === false ) {
1218                                         $sep .= ')';
1219                                 }
1220
1221                                 $numSepChars = strspn( strrev( $url ), $sep );
1222                                 if ( $numSepChars ) {
1223                                         $trail = substr( $url, -$numSepChars ) . $trail;
1224                                         $url = substr( $url, 0, -$numSepChars );
1225                                 }
1226
1227                                 # Normalize any HTML entities in input. They will be
1228                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1229                                 $url = Sanitizer::decodeCharReferences( $url );
1230
1231                                 # Is this an external image?
1232                                 $text = $this->maybeMakeExternalImage( $url );
1233                                 if ( $text === false ) {
1234                                         # Not an image, make a link
1235                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1236                                         # Register it in the output object...
1237                                         # Replace unnecessary URL escape codes with their equivalent characters
1238                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1239                                         $this->mOutput->addExternalLink( $pasteurized );
1240                                 }
1241                                 $s .= $text . $trail;
1242                         } else {
1243                                 $s .= $protocol . $remainder;
1244                         }
1245                 }
1246                 wfProfileOut( $fname );
1247                 return $s;
1248         }
1249
1250         /**
1251          * Replace unusual URL escape codes with their equivalent characters
1252          * @param string
1253          * @return string
1254          * @static
1255          * @fixme This can merge genuinely required bits in the path or query string,
1256          *        breaking legit URLs. A proper fix would treat the various parts of
1257          *        the URL differently; as a workaround, just use the output for
1258          *        statistical records, not for actual linking/output.
1259          */
1260         function replaceUnusualEscapes( $url ) {
1261                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1262                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1263         }
1264
1265         /**
1266          * Callback function used in replaceUnusualEscapes().
1267          * Replaces unusual URL escape codes with their equivalent character
1268          * @static
1269          * @private
1270          */
1271         function replaceUnusualEscapesCallback( $matches ) {
1272                 $char = urldecode( $matches[0] );
1273                 $ord = ord( $char );
1274                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1275                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1276                         // No, shouldn't be escaped
1277                         return $char;
1278                 } else {
1279                         // Yes, leave it escaped
1280                         return $matches[0];
1281                 }
1282         }
1283
1284         /**
1285          * make an image if it's allowed, either through the global
1286          * option or through the exception
1287          * @private
1288          */
1289         function maybeMakeExternalImage( $url ) {
1290                 $sk =& $this->mOptions->getSkin();
1291                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1292                 $imagesexception = !empty($imagesfrom);
1293                 $text = false;
1294                 if ( $this->mOptions->getAllowExternalImages()
1295                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1296                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1297                                 # Image found
1298                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1299                         }
1300                 }
1301                 return $text;
1302         }
1303
1304         /**
1305          * Process [[ ]] wikilinks
1306          *
1307          * @private
1308          */
1309         function replaceInternalLinks( $s ) {
1310                 global $wgContLang;
1311                 static $fname = 'Parser::replaceInternalLinks' ;
1312
1313                 wfProfileIn( $fname );
1314
1315                 wfProfileIn( $fname.'-setup' );
1316                 static $tc = FALSE;
1317                 # the % is needed to support urlencoded titles as well
1318                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1319
1320                 $sk =& $this->mOptions->getSkin();
1321
1322                 #split the entire text string on occurences of [[
1323                 $a = explode( '[[', ' ' . $s );
1324                 #get the first element (all text up to first [[), and remove the space we added
1325                 $s = array_shift( $a );
1326                 $s = substr( $s, 1 );
1327
1328                 # Match a link having the form [[namespace:link|alternate]]trail
1329                 static $e1 = FALSE;
1330                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1331                 # Match cases where there is no "]]", which might still be images
1332                 static $e1_img = FALSE;
1333                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1334                 # Match the end of a line for a word that's not followed by whitespace,
1335                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1336                 $e2 = wfMsgForContent( 'linkprefix' );
1337
1338                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1339
1340                 if( is_null( $this->mTitle ) ) {
1341                         throw new MWException( 'nooo' );
1342                 }
1343                 $nottalk = !$this->mTitle->isTalkPage();
1344
1345                 if ( $useLinkPrefixExtension ) {
1346                         if ( preg_match( $e2, $s, $m ) ) {
1347                                 $first_prefix = $m[2];
1348                         } else {
1349                                 $first_prefix = false;
1350                         }
1351                 } else {
1352                         $prefix = '';
1353                 }
1354
1355                 $selflink = $this->mTitle->getPrefixedText();
1356                 wfProfileOut( $fname.'-setup' );
1357
1358                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1359                 $useSubpages = $this->areSubpagesAllowed();
1360
1361                 # Loop for each link
1362                 for ($k = 0; isset( $a[$k] ); $k++) {
1363                         $line = $a[$k];
1364                         if ( $useLinkPrefixExtension ) {
1365                                 wfProfileIn( $fname.'-prefixhandling' );
1366                                 if ( preg_match( $e2, $s, $m ) ) {
1367                                         $prefix = $m[2];
1368                                         $s = $m[1];
1369                                 } else {
1370                                         $prefix='';
1371                                 }
1372                                 # first link
1373                                 if($first_prefix) {
1374                                         $prefix = $first_prefix;
1375                                         $first_prefix = false;
1376                                 }
1377                                 wfProfileOut( $fname.'-prefixhandling' );
1378                         }
1379
1380                         $might_be_img = false;
1381
1382                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1383                                 $text = $m[2];
1384                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1385                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1386                                 # the real problem is with the $e1 regex
1387                                 # See bug 1300.
1388                                 #
1389                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1390                                 # and no image is in sight. See bug 2095.
1391                                 #
1392                                 if( $text !== '' &&
1393                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1394                                         strpos($text, '[') !== false
1395                                 )
1396                                 {
1397                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1398                                         $m[3] = $n[1];
1399                                 }
1400                                 # fix up urlencoded title texts
1401                                 if(preg_match('/%/', $m[1] ))
1402                                         # Should anchors '#' also be rejected?
1403                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1404                                 $trail = $m[3];
1405                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1406                                 $might_be_img = true;
1407                                 $text = $m[2];
1408                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1409                                 $trail = "";
1410                         } else { # Invalid form; output directly
1411                                 $s .= $prefix . '[[' . $line ;
1412                                 continue;
1413                         }
1414
1415                         # Don't allow internal links to pages containing
1416                         # PROTO: where PROTO is a valid URL protocol; these
1417                         # should be external links.
1418                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1419                                 $s .= $prefix . '[[' . $line ;
1420                                 continue;
1421                         }
1422
1423                         # Make subpage if necessary
1424                         if( $useSubpages ) {
1425                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1426                         } else {
1427                                 $link = $m[1];
1428                         }
1429
1430                         $noforce = (substr($m[1], 0, 1) != ':');
1431                         if (!$noforce) {
1432                                 # Strip off leading ':'
1433                                 $link = substr($link, 1);
1434                         }
1435
1436                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1437                         if( !$nt ) {
1438                                 $s .= $prefix . '[[' . $line;
1439                                 continue;
1440                         }
1441
1442                         #check other language variants of the link
1443                         #if the article does not exist
1444                         if( $checkVariantLink
1445                             && $nt->getArticleID() == 0 ) {
1446                                 $wgContLang->findVariantLink($link, $nt);
1447                         }
1448
1449                         $ns = $nt->getNamespace();
1450                         $iw = $nt->getInterWiki();
1451
1452                         if ($might_be_img) { # if this is actually an invalid link
1453                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1454                                         $found = false;
1455                                         while (isset ($a[$k+1]) ) {
1456                                                 #look at the next 'line' to see if we can close it there
1457                                                 $spliced = array_splice( $a, $k + 1, 1 );
1458                                                 $next_line = array_shift( $spliced );
1459                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1460                                                 # the first ]] closes the inner link, the second the image
1461                                                         $found = true;
1462                                                         $text .= '[[' . $m[1];
1463                                                         $trail = $m[2];
1464                                                         break;
1465                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1466                                                         #if there's exactly one ]] that's fine, we'll keep looking
1467                                                         $text .= '[[' . $m[0];
1468                                                 } else {
1469                                                         #if $next_line is invalid too, we need look no further
1470                                                         $text .= '[[' . $next_line;
1471                                                         break;
1472                                                 }
1473                                         }
1474                                         if ( !$found ) {
1475                                                 # we couldn't find the end of this imageLink, so output it raw
1476                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1477                                                 $text = $this->replaceInternalLinks($text);
1478                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1479                                                 # note: no $trail, because without an end, there *is* no trail
1480                                                 continue;
1481                                         }
1482                                 } else { #it's not an image, so output it raw
1483                                         $s .= $prefix . '[[' . $link . '|' . $text;
1484                                         # note: no $trail, because without an end, there *is* no trail
1485                                         continue;
1486                                 }
1487                         }
1488
1489                         $wasblank = ( '' == $text );
1490                         if( $wasblank ) $text = $link;
1491
1492
1493                         # Link not escaped by : , create the various objects
1494                         if( $noforce ) {
1495
1496                                 # Interwikis
1497                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1498                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1499                                         $s = rtrim($s . "\n");
1500                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1501                                         continue;
1502                                 }
1503
1504                                 if ( $ns == NS_IMAGE ) {
1505                                         wfProfileIn( "$fname-image" );
1506                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1507                                                 # recursively parse links inside the image caption
1508                                                 # actually, this will parse them in any other parameters, too,
1509                                                 # but it might be hard to fix that, and it doesn't matter ATM
1510                                                 $text = $this->replaceExternalLinks($text);
1511                                                 $text = $this->replaceInternalLinks($text);
1512
1513                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1514                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1515                                                 $this->mOutput->addImage( $nt->getDBkey() );
1516
1517                                                 wfProfileOut( "$fname-image" );
1518                                                 continue;
1519                                         }
1520                                         wfProfileOut( "$fname-image" );
1521
1522                                 }
1523
1524                                 if ( $ns == NS_CATEGORY ) {
1525                                         wfProfileIn( "$fname-category" );
1526                                         $s = rtrim($s . "\n"); # bug 87
1527
1528                                         if ( $wasblank ) {
1529                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1530                                                         $sortkey = $this->mTitle->getText();
1531                                                 } else {
1532                                                         $sortkey = $this->mTitle->getPrefixedText();
1533                                                 }
1534                                         } else {
1535                                                 $sortkey = $text;
1536                                         }
1537                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1538                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1539                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1540
1541                                         /**
1542                                          * Strip the whitespace Category links produce, see bug 87
1543                                          * @todo We might want to use trim($tmp, "\n") here.
1544                                          */
1545                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1546
1547                                         wfProfileOut( "$fname-category" );
1548                                         continue;
1549                                 }
1550                         }
1551
1552                         if( ( $nt->getPrefixedText() === $selflink ) &&
1553                             ( $nt->getFragment() === '' ) ) {
1554                                 # Self-links are handled specially; generally de-link and change to bold.
1555                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1556                                 continue;
1557                         }
1558
1559                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1560                         if( $ns == NS_MEDIA ) {
1561                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1562                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1563                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1564                                 $this->mOutput->addImage( $nt->getDBkey() );
1565                                 continue;
1566                         } elseif( $ns == NS_SPECIAL ) {
1567                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1568                                 continue;
1569                         } elseif( $ns == NS_IMAGE ) {
1570                                 $img = Image::newFromTitle( $nt );
1571                                 if( $img->exists() ) {
1572                                         // Force a blue link if the file exists; may be a remote
1573                                         // upload on the shared repository, and we want to see its
1574                                         // auto-generated page.
1575                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1576                                         continue;
1577                                 }
1578                         }
1579                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1580                 }
1581                 wfProfileOut( $fname );
1582                 return $s;
1583         }
1584
1585         /**
1586          * Make a link placeholder. The text returned can be later resolved to a real link with
1587          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1588          * parsing of interwiki links, and secondly to allow all extistence checks and
1589          * article length checks (for stub links) to be bundled into a single query.
1590          *
1591          */
1592         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1593                 if ( ! is_object($nt) ) {
1594                         # Fail gracefully
1595                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1596                 } else {
1597                         # Separate the link trail from the rest of the link
1598                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1599
1600                         if ( $nt->isExternal() ) {
1601                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1602                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1603                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1604                         } else {
1605                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1606                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1607                                 $this->mLinkHolders['queries'][] = $query;
1608                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1609                                 $this->mLinkHolders['titles'][] = $nt;
1610
1611                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1612                         }
1613                 }
1614                 return $retVal;
1615         }
1616
1617         /**
1618          * Render a forced-blue link inline; protect against double expansion of
1619          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1620          * Since this little disaster has to split off the trail text to avoid
1621          * breaking URLs in the following text without breaking trails on the
1622          * wiki links, it's been made into a horrible function.
1623          *
1624          * @param Title $nt
1625          * @param string $text
1626          * @param string $query
1627          * @param string $trail
1628          * @param string $prefix
1629          * @return string HTML-wikitext mix oh yuck
1630          */
1631         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1632                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1633                 $sk =& $this->mOptions->getSkin();
1634                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1635                 return $this->armorLinks( $link ) . $trail;
1636         }
1637
1638         /**
1639          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1640          * going to go through further parsing steps before inline URL expansion.
1641          *
1642          * In particular this is important when using action=render, which causes
1643          * full URLs to be included.
1644          *
1645          * Oh man I hate our multi-layer parser!
1646          *
1647          * @param string more-or-less HTML
1648          * @return string less-or-more HTML with NOPARSE bits
1649          */
1650         function armorLinks( $text ) {
1651                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1652                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1653         }
1654
1655         /**
1656          * Return true if subpage links should be expanded on this page.
1657          * @return bool
1658          */
1659         function areSubpagesAllowed() {
1660                 # Some namespaces don't allow subpages
1661                 global $wgNamespacesWithSubpages;
1662                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1663         }
1664
1665         /**
1666          * Handle link to subpage if necessary
1667          * @param string $target the source of the link
1668          * @param string &$text the link text, modified as necessary
1669          * @return string the full name of the link
1670          * @private
1671          */
1672         function maybeDoSubpageLink($target, &$text) {
1673                 # Valid link forms:
1674                 # Foobar -- normal
1675                 # :Foobar -- override special treatment of prefix (images, language links)
1676                 # /Foobar -- convert to CurrentPage/Foobar
1677                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1678                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1679                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1680
1681                 $fname = 'Parser::maybeDoSubpageLink';
1682                 wfProfileIn( $fname );
1683                 $ret = $target; # default return value is no change
1684
1685                 # Some namespaces don't allow subpages,
1686                 # so only perform processing if subpages are allowed
1687                 if( $this->areSubpagesAllowed() ) {
1688                         # Look at the first character
1689                         if( $target != '' && $target{0} == '/' ) {
1690                                 # / at end means we don't want the slash to be shown
1691                                 if( substr( $target, -1, 1 ) == '/' ) {
1692                                         $target = substr( $target, 1, -1 );
1693                                         $noslash = $target;
1694                                 } else {
1695                                         $noslash = substr( $target, 1 );
1696                                 }
1697
1698                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1699                                 if( '' === $text ) {
1700                                         $text = $target;
1701                                 } # this might be changed for ugliness reasons
1702                         } else {
1703                                 # check for .. subpage backlinks
1704                                 $dotdotcount = 0;
1705                                 $nodotdot = $target;
1706                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1707                                         ++$dotdotcount;
1708                                         $nodotdot = substr( $nodotdot, 3 );
1709                                 }
1710                                 if($dotdotcount > 0) {
1711                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1712                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1713                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1714                                                 # / at the end means don't show full path
1715                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1716                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1717                                                         if( '' === $text ) {
1718                                                                 $text = $nodotdot;
1719                                                         }
1720                                                 }
1721                                                 $nodotdot = trim( $nodotdot );
1722                                                 if( $nodotdot != '' ) {
1723                                                         $ret .= '/' . $nodotdot;
1724                                                 }
1725                                         }
1726                                 }
1727                         }
1728                 }
1729
1730                 wfProfileOut( $fname );
1731                 return $ret;
1732         }
1733
1734         /**#@+
1735          * Used by doBlockLevels()
1736          * @private
1737          */
1738         /* private */ function closeParagraph() {
1739                 $result = '';
1740                 if ( '' != $this->mLastSection ) {
1741                         $result = '</' . $this->mLastSection  . ">\n";
1742                 }
1743                 $this->mInPre = false;
1744                 $this->mLastSection = '';
1745                 return $result;
1746         }
1747         # getCommon() returns the length of the longest common substring
1748         # of both arguments, starting at the beginning of both.
1749         #
1750         /* private */ function getCommon( $st1, $st2 ) {
1751                 $fl = strlen( $st1 );
1752                 $shorter = strlen( $st2 );
1753                 if ( $fl < $shorter ) { $shorter = $fl; }
1754
1755                 for ( $i = 0; $i < $shorter; ++$i ) {
1756                         if ( $st1{$i} != $st2{$i} ) { break; }
1757                 }
1758                 return $i;
1759         }
1760         # These next three functions open, continue, and close the list
1761         # element appropriate to the prefix character passed into them.
1762         #
1763         /* private */ function openList( $char ) {
1764                 $result = $this->closeParagraph();
1765
1766                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1767                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1768                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1769                 else if ( ';' == $char ) {
1770                         $result .= '<dl><dt>';
1771                         $this->mDTopen = true;
1772                 }
1773                 else { $result = '<!-- ERR 1 -->'; }
1774
1775                 return $result;
1776         }
1777
1778         /* private */ function nextItem( $char ) {
1779                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1780                 else if ( ':' == $char || ';' == $char ) {
1781                         $close = '</dd>';
1782                         if ( $this->mDTopen ) { $close = '</dt>'; }
1783                         if ( ';' == $char ) {
1784                                 $this->mDTopen = true;
1785                                 return $close . '<dt>';
1786                         } else {
1787                                 $this->mDTopen = false;
1788                                 return $close . '<dd>';
1789                         }
1790                 }
1791                 return '<!-- ERR 2 -->';
1792         }
1793
1794         /* private */ function closeList( $char ) {
1795                 if ( '*' == $char ) { $text = '</li></ul>'; }
1796                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1797                 else if ( ':' == $char ) {
1798                         if ( $this->mDTopen ) {
1799                                 $this->mDTopen = false;
1800                                 $text = '</dt></dl>';
1801                         } else {
1802                                 $text = '</dd></dl>';
1803                         }
1804                 }
1805                 else {  return '<!-- ERR 3 -->'; }
1806                 return $text."\n";
1807         }
1808         /**#@-*/
1809
1810         /**
1811          * Make lists from lines starting with ':', '*', '#', etc.
1812          *
1813          * @private
1814          * @return string the lists rendered as HTML
1815          */
1816         function doBlockLevels( $text, $linestart ) {
1817                 $fname = 'Parser::doBlockLevels';
1818                 wfProfileIn( $fname );
1819
1820                 # Parsing through the text line by line.  The main thing
1821                 # happening here is handling of block-level elements p, pre,
1822                 # and making lists from lines starting with * # : etc.
1823                 #
1824                 $textLines = explode( "\n", $text );
1825
1826                 $lastPrefix = $output = '';
1827                 $this->mDTopen = $inBlockElem = false;
1828                 $prefixLength = 0;
1829                 $paragraphStack = false;
1830
1831                 if ( !$linestart ) {
1832                         $output .= array_shift( $textLines );
1833                 }
1834                 foreach ( $textLines as $oLine ) {
1835                         $lastPrefixLength = strlen( $lastPrefix );
1836                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1837                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1838                         if ( !$this->mInPre ) {
1839                                 # Multiple prefixes may abut each other for nested lists.
1840                                 $prefixLength = strspn( $oLine, '*#:;' );
1841                                 $pref = substr( $oLine, 0, $prefixLength );
1842
1843                                 # eh?
1844                                 $pref2 = str_replace( ';', ':', $pref );
1845                                 $t = substr( $oLine, $prefixLength );
1846                                 $this->mInPre = !empty($preOpenMatch);
1847                         } else {
1848                                 # Don't interpret any other prefixes in preformatted text
1849                                 $prefixLength = 0;
1850                                 $pref = $pref2 = '';
1851                                 $t = $oLine;
1852                         }
1853
1854                         # List generation
1855                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1856                                 # Same as the last item, so no need to deal with nesting or opening stuff
1857                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1858                                 $paragraphStack = false;
1859
1860                                 if ( substr( $pref, -1 ) == ';') {
1861                                         # The one nasty exception: definition lists work like this:
1862                                         # ; title : definition text
1863                                         # So we check for : in the remainder text to split up the
1864                                         # title and definition, without b0rking links.
1865                                         $term = $t2 = '';
1866                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1867                                                 $t = $t2;
1868                                                 $output .= $term . $this->nextItem( ':' );
1869                                         }
1870                                 }
1871                         } elseif( $prefixLength || $lastPrefixLength ) {
1872                                 # Either open or close a level...
1873                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1874                                 $paragraphStack = false;
1875
1876                                 while( $commonPrefixLength < $lastPrefixLength ) {
1877                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1878                                         --$lastPrefixLength;
1879                                 }
1880                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1881                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1882                                 }
1883                                 while ( $prefixLength > $commonPrefixLength ) {
1884                                         $char = substr( $pref, $commonPrefixLength, 1 );
1885                                         $output .= $this->openList( $char );
1886
1887                                         if ( ';' == $char ) {
1888                                                 # FIXME: This is dupe of code above
1889                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1890                                                         $t = $t2;
1891                                                         $output .= $term . $this->nextItem( ':' );
1892                                                 }
1893                                         }
1894                                         ++$commonPrefixLength;
1895                                 }
1896                                 $lastPrefix = $pref2;
1897                         }
1898                         if( 0 == $prefixLength ) {
1899                                 wfProfileIn( "$fname-paragraph" );
1900                                 # No prefix (not in list)--go to paragraph mode
1901                                 // XXX: use a stack for nestable elements like span, table and div
1902                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1903                                 $closematch = preg_match(
1904                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1905                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1906                                 if ( $openmatch or $closematch ) {
1907                                         $paragraphStack = false;
1908                                         # TODO bug 5718: paragraph closed
1909                                         $output .= $this->closeParagraph();
1910                                         if ( $preOpenMatch and !$preCloseMatch ) {
1911                                                 $this->mInPre = true;
1912                                         }
1913                                         if ( $closematch ) {
1914                                                 $inBlockElem = false;
1915                                         } else {
1916                                                 $inBlockElem = true;
1917                                         }
1918                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1919                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1920                                                 // pre
1921                                                 if ($this->mLastSection != 'pre') {
1922                                                         $paragraphStack = false;
1923                                                         $output .= $this->closeParagraph().'<pre>';
1924                                                         $this->mLastSection = 'pre';
1925                                                 }
1926                                                 $t = substr( $t, 1 );
1927                                         } else {
1928                                                 // paragraph
1929                                                 if ( '' == trim($t) ) {
1930                                                         if ( $paragraphStack ) {
1931                                                                 $output .= $paragraphStack.'<br />';
1932                                                                 $paragraphStack = false;
1933                                                                 $this->mLastSection = 'p';
1934                                                         } else {
1935                                                                 if ($this->mLastSection != 'p' ) {
1936                                                                         $output .= $this->closeParagraph();
1937                                                                         $this->mLastSection = '';
1938                                                                         $paragraphStack = '<p>';
1939                                                                 } else {
1940                                                                         $paragraphStack = '</p><p>';
1941                                                                 }
1942                                                         }
1943                                                 } else {
1944                                                         if ( $paragraphStack ) {
1945                                                                 $output .= $paragraphStack;
1946                                                                 $paragraphStack = false;
1947                                                                 $this->mLastSection = 'p';
1948                                                         } else if ($this->mLastSection != 'p') {
1949                                                                 $output .= $this->closeParagraph().'<p>';
1950                                                                 $this->mLastSection = 'p';
1951                                                         }
1952                                                 }
1953                                         }
1954                                 }
1955                                 wfProfileOut( "$fname-paragraph" );
1956                         }
1957                         // somewhere above we forget to get out of pre block (bug 785)
1958                         if($preCloseMatch && $this->mInPre) {
1959                                 $this->mInPre = false;
1960                         }
1961                         if ($paragraphStack === false) {
1962                                 $output .= $t."\n";
1963                         }
1964                 }
1965                 while ( $prefixLength ) {
1966                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1967                         --$prefixLength;
1968                 }
1969                 if ( '' != $this->mLastSection ) {
1970                         $output .= '</' . $this->mLastSection . '>';
1971                         $this->mLastSection = '';
1972                 }
1973
1974                 wfProfileOut( $fname );
1975                 return $output;
1976         }
1977
1978         /**
1979          * Split up a string on ':', ignoring any occurences inside tags
1980          * to prevent illegal overlapping.
1981          * @param string $str the string to split
1982          * @param string &$before set to everything before the ':'
1983          * @param string &$after set to everything after the ':'
1984          * return string the position of the ':', or false if none found
1985          */
1986         function findColonNoLinks($str, &$before, &$after) {
1987                 $fname = 'Parser::findColonNoLinks';
1988                 wfProfileIn( $fname );
1989
1990                 $pos = strpos( $str, ':' );
1991                 if( $pos === false ) {
1992                         // Nothing to find!
1993                         wfProfileOut( $fname );
1994                         return false;
1995                 }
1996
1997                 $lt = strpos( $str, '<' );
1998                 if( $lt === false || $lt > $pos ) {
1999                         // Easy; no tag nesting to worry about
2000                         $before = substr( $str, 0, $pos );
2001                         $after = substr( $str, $pos+1 );
2002                         wfProfileOut( $fname );
2003                         return $pos;
2004                 }
2005
2006                 // Ugly state machine to walk through avoiding tags.
2007                 $state = MW_COLON_STATE_TEXT;
2008                 $stack = 0;
2009                 $len = strlen( $str );
2010                 for( $i = 0; $i < $len; $i++ ) {
2011                         $c = $str{$i};
2012
2013                         switch( $state ) {
2014                         // (Using the number is a performance hack for common cases)
2015                         case 0: // MW_COLON_STATE_TEXT:
2016                                 switch( $c ) {
2017                                 case "<":
2018                                         // Could be either a <start> tag or an </end> tag
2019                                         $state = MW_COLON_STATE_TAGSTART;
2020                                         break;
2021                                 case ":":
2022                                         if( $stack == 0 ) {
2023                                                 // We found it!
2024                                                 $before = substr( $str, 0, $i );
2025                                                 $after = substr( $str, $i + 1 );
2026                                                 wfProfileOut( $fname );
2027                                                 return $i;
2028                                         }
2029                                         // Embedded in a tag; don't break it.
2030                                         break;
2031                                 default:
2032                                         // Skip ahead looking for something interesting
2033                                         $colon = strpos( $str, ':', $i );
2034                                         if( $colon === false ) {
2035                                                 // Nothing else interesting
2036                                                 wfProfileOut( $fname );
2037                                                 return false;
2038                                         }
2039                                         $lt = strpos( $str, '<', $i );
2040                                         if( $stack === 0 ) {
2041                                                 if( $lt === false || $colon < $lt ) {
2042                                                         // We found it!
2043                                                         $before = substr( $str, 0, $colon );
2044                                                         $after = substr( $str, $colon + 1 );
2045                                                         wfProfileOut( $fname );
2046                                                         return $i;
2047                                                 }
2048                                         }
2049                                         if( $lt === false ) {
2050                                                 // Nothing else interesting to find; abort!
2051                                                 // We're nested, but there's no close tags left. Abort!
2052                                                 break 2;
2053                                         }
2054                                         // Skip ahead to next tag start
2055                                         $i = $lt;
2056                                         $state = MW_COLON_STATE_TAGSTART;
2057                                 }
2058                                 break;
2059                         case 1: // MW_COLON_STATE_TAG:
2060                                 // In a <tag>
2061                                 switch( $c ) {
2062                                 case ">":
2063                                         $stack++;
2064                                         $state = MW_COLON_STATE_TEXT;
2065                                         break;
2066                                 case "/":
2067                                         // Slash may be followed by >?
2068                                         $state = MW_COLON_STATE_TAGSLASH;
2069                                         break;
2070                                 default:
2071                                         // ignore
2072                                 }
2073                                 break;
2074                         case 2: // MW_COLON_STATE_TAGSTART:
2075                                 switch( $c ) {
2076                                 case "/":
2077                                         $state = MW_COLON_STATE_CLOSETAG;
2078                                         break;
2079                                 case "!":
2080                                         $state = MW_COLON_STATE_COMMENT;
2081                                         break;
2082                                 case ">":
2083                                         // Illegal early close? This shouldn't happen D:
2084                                         $state = MW_COLON_STATE_TEXT;
2085                                         break;
2086                                 default:
2087                                         $state = MW_COLON_STATE_TAG;
2088                                 }
2089                                 break;
2090                         case 3: // MW_COLON_STATE_CLOSETAG:
2091                                 // In a </tag>
2092                                 if( $c == ">" ) {
2093                                         $stack--;
2094                                         if( $stack < 0 ) {
2095                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2096                                                 wfProfileOut( $fname );
2097                                                 return false;
2098                                         }
2099                                         $state = MW_COLON_STATE_TEXT;
2100                                 }
2101                                 break;
2102                         case MW_COLON_STATE_TAGSLASH:
2103                                 if( $c == ">" ) {
2104                                         // Yes, a self-closed tag <blah/>
2105                                         $state = MW_COLON_STATE_TEXT;
2106                                 } else {
2107                                         // Probably we're jumping the gun, and this is an attribute
2108                                         $state = MW_COLON_STATE_TAG;
2109                                 }
2110                                 break;
2111                         case 5: // MW_COLON_STATE_COMMENT:
2112                                 if( $c == "-" ) {
2113                                         $state = MW_COLON_STATE_COMMENTDASH;
2114                                 }
2115                                 break;
2116                         case MW_COLON_STATE_COMMENTDASH:
2117                                 if( $c == "-" ) {
2118                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2119                                 } else {
2120                                         $state = MW_COLON_STATE_COMMENT;
2121                                 }
2122                                 break;
2123                         case MW_COLON_STATE_COMMENTDASHDASH:
2124                                 if( $c == ">" ) {
2125                                         $state = MW_COLON_STATE_TEXT;
2126                                 } else {
2127                                         $state = MW_COLON_STATE_COMMENT;
2128                                 }
2129                                 break;
2130                         default:
2131                                 throw new MWException( "State machine error in $fname" );
2132                         }
2133                 }
2134                 if( $stack > 0 ) {
2135                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2136                         return false;
2137                 }
2138                 wfProfileOut( $fname );
2139                 return false;
2140         }
2141
2142         /**
2143          * Return value of a magic variable (like PAGENAME)
2144          *
2145          * @private
2146          */
2147         function getVariableValue( $index ) {
2148                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2149
2150                 /**
2151                  * Some of these require message or data lookups and can be
2152                  * expensive to check many times.
2153                  */
2154                 static $varCache = array();
2155                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2156                         if ( isset( $varCache[$index] ) )
2157                                 return $varCache[$index];
2158
2159                 $ts = time();
2160                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2161
2162                 switch ( $index ) {
2163                         case MAG_CURRENTMONTH:
2164                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2165                         case MAG_CURRENTMONTHNAME:
2166                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2167                         case MAG_CURRENTMONTHNAMEGEN:
2168                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2169                         case MAG_CURRENTMONTHABBREV:
2170                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2171                         case MAG_CURRENTDAY:
2172                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2173                         case MAG_CURRENTDAY2:
2174                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2175                         case MAG_PAGENAME:
2176                                 return $this->mTitle->getText();
2177                         case MAG_PAGENAMEE:
2178                                 return $this->mTitle->getPartialURL();
2179                         case MAG_FULLPAGENAME:
2180                                 return $this->mTitle->getPrefixedText();
2181                         case MAG_FULLPAGENAMEE:
2182                                 return $this->mTitle->getPrefixedURL();
2183                         case MAG_SUBPAGENAME:
2184                                 return $this->mTitle->getSubpageText();
2185                         case MAG_SUBPAGENAMEE:
2186                                 return $this->mTitle->getSubpageUrlForm();
2187                         case MAG_BASEPAGENAME:
2188                                 return $this->mTitle->getBaseText();
2189                         case MAG_BASEPAGENAMEE:
2190                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2191                         case MAG_TALKPAGENAME:
2192                                 if( $this->mTitle->canTalk() ) {
2193                                         $talkPage = $this->mTitle->getTalkPage();
2194                                         return $talkPage->getPrefixedText();
2195                                 } else {
2196                                         return '';
2197                                 }
2198                         case MAG_TALKPAGENAMEE:
2199                                 if( $this->mTitle->canTalk() ) {
2200                                         $talkPage = $this->mTitle->getTalkPage();
2201                                         return $talkPage->getPrefixedUrl();
2202                                 } else {
2203                                         return '';
2204                                 }
2205                         case MAG_SUBJECTPAGENAME:
2206                                 $subjPage = $this->mTitle->getSubjectPage();
2207                                 return $subjPage->getPrefixedText();
2208                         case MAG_SUBJECTPAGENAMEE:
2209                                 $subjPage = $this->mTitle->getSubjectPage();
2210                                 return $subjPage->getPrefixedUrl();
2211                         case MAG_REVISIONID:
2212                                 return $this->mRevisionId;
2213                         case MAG_NAMESPACE:
2214                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2215                         case MAG_NAMESPACEE:
2216                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2217                         case MAG_TALKSPACE:
2218                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2219                         case MAG_TALKSPACEE:
2220                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2221                         case MAG_SUBJECTSPACE:
2222                                 return $this->mTitle->getSubjectNsText();
2223                         case MAG_SUBJECTSPACEE:
2224                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2225                         case MAG_CURRENTDAYNAME:
2226                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2227                         case MAG_CURRENTYEAR:
2228                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2229                         case MAG_CURRENTTIME:
2230                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2231                         case MAG_CURRENTWEEK:
2232                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2233                                 // int to remove the padding
2234                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2235                         case MAG_CURRENTDOW:
2236                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2237                         case MAG_NUMBEROFARTICLES:
2238                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2239                         case MAG_NUMBEROFFILES:
2240                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2241                         case MAG_NUMBEROFUSERS:
2242                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2243                         case MAG_NUMBEROFPAGES:
2244                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2245                         case MAG_CURRENTTIMESTAMP:
2246                                 return $varCache[$index] = wfTimestampNow();
2247                         case MAG_CURRENTVERSION:
2248                                 global $wgVersion;
2249                                 return $wgVersion;
2250                         case MAG_SITENAME:
2251                                 return $wgSitename;
2252                         case MAG_SERVER:
2253                                 return $wgServer;
2254                         case MAG_SERVERNAME:
2255                                 return $wgServerName;
2256                         case MAG_SCRIPTPATH:
2257                                 return $wgScriptPath;
2258                         case MAG_DIRECTIONMARK:
2259                                 return $wgContLang->getDirMark();
2260                         case MAG_CONTENTLANGUAGE:
2261                                 global $wgContLanguageCode;
2262                                 return $wgContLanguageCode;
2263                         default:
2264                                 $ret = null;
2265                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2266                                         return $ret;
2267                                 else
2268                                         return null;
2269                 }
2270         }
2271
2272         /**
2273          * initialise the magic variables (like CURRENTMONTHNAME)
2274          *
2275          * @private
2276          */
2277         function initialiseVariables() {
2278                 $fname = 'Parser::initialiseVariables';
2279                 wfProfileIn( $fname );
2280                 global $wgVariableIDs;
2281                 $this->mVariables = array();
2282                 foreach ( $wgVariableIDs as $id ) {
2283                         $mw =& MagicWord::get( $id );
2284                         $mw->addToArray( $this->mVariables, $id );
2285                 }
2286                 wfProfileOut( $fname );
2287         }
2288
2289         /**
2290          * parse any parentheses in format ((title|part|part))
2291          * and call callbacks to get a replacement text for any found piece
2292          *
2293          * @param string $text The text to parse
2294          * @param array $callbacks rules in form:
2295          *     '{' => array(                            # opening parentheses
2296          *                                      'end' => '}',   # closing parentheses
2297          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2298          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2299          *                                                                )
2300          *                                      )
2301          * @private
2302          */
2303         function replace_callback ($text, $callbacks) {
2304                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2305                 $lastOpeningBrace = -1;                 # last not closed parentheses
2306
2307                 for ($i = 0; $i < strlen($text); $i++) {
2308                         # check for any opening brace
2309                         $rule = null;
2310                         $nextPos = -1;
2311                         foreach ($callbacks as $key => $value) {
2312                                 $pos = strpos ($text, $key, $i);
2313                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2314                                         $rule = $value;
2315                                         $nextPos = $pos;
2316                                 }
2317                         }
2318
2319                         if ($lastOpeningBrace >= 0) {
2320                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2321
2322                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2323                                         $rule = null;
2324                                         $nextPos = $pos;
2325                                 }
2326
2327                                 $pos = strpos ($text, '|', $i);
2328
2329                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2330                                         $rule = null;
2331                                         $nextPos = $pos;
2332                                 }
2333                         }
2334
2335                         if ($nextPos == -1)
2336                                 break;
2337
2338                         $i = $nextPos;
2339
2340                         # found openning brace, lets add it to parentheses stack
2341                         if (null != $rule) {
2342                                 $piece = array('brace' => $text[$i],
2343                                                            'braceEnd' => $rule['end'],
2344                                                            'count' => 1,
2345                                                            'title' => '',
2346                                                            'parts' => null);
2347
2348                                 # count openning brace characters
2349                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2350                                         $piece['count']++;
2351                                         $i++;
2352                                 }
2353
2354                                 $piece['startAt'] = $i+1;
2355                                 $piece['partStart'] = $i+1;
2356
2357                                 # we need to add to stack only if openning brace count is enough for any given rule
2358                                 foreach ($rule['cb'] as $cnt => $fn) {
2359                                         if ($piece['count'] >= $cnt) {
2360                                                 $lastOpeningBrace ++;
2361                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2362                                                 break;
2363                                         }
2364                                 }
2365
2366                                 continue;
2367                         }
2368                         else if ($lastOpeningBrace >= 0) {
2369                                 # first check if it is a closing brace
2370                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2371                                         # lets check if it is enough characters for closing brace
2372                                         $count = 1;
2373                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2374                                                 $count++;
2375
2376                                         # if there are more closing parentheses than opening ones, we parse less
2377                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2378                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2379
2380                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2381                                         $matchingCount = 0;
2382                                         $matchingCallback = null;
2383                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2384                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2385                                                         $matchingCount = $cnt;
2386                                                         $matchingCallback = $fn;
2387                                                 }
2388                                         }
2389
2390                                         if ($matchingCount == 0) {
2391                                                 $i += $count - 1;
2392                                                 continue;
2393                                         }
2394
2395                                         # lets set a title or last part (if '|' was found)
2396                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2397                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2398                                         else
2399                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2400
2401                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2402                                         $pieceEnd = $i + $matchingCount;
2403
2404                                         if( is_callable( $matchingCallback ) ) {
2405                                                 $cbArgs = array (
2406                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2407                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2408                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2409                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2410                                                                                  );
2411                                                 # finally we can call a user callback and replace piece of text
2412                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2413                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2414                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2415                                         }
2416                                         else {
2417                                                 # null value for callback means that parentheses should be parsed, but not replaced
2418                                                 $i += $matchingCount - 1;
2419                                         }
2420
2421                                         # reset last openning parentheses, but keep it in case there are unused characters
2422                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2423                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2424                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2425                                                                    'title' => '',
2426                                                                    'parts' => null,
2427                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2428                                         $openingBraceStack[$lastOpeningBrace--] = null;
2429
2430                                         if ($matchingCount < $piece['count']) {
2431                                                 $piece['count'] -= $matchingCount;
2432                                                 $piece['startAt'] -= $matchingCount;
2433                                                 $piece['partStart'] = $piece['startAt'];
2434                                                 # do we still qualify for any callback with remaining count?
2435                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2436                                                         if ($piece['count'] >= $cnt) {
2437                                                                 $lastOpeningBrace ++;
2438                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2439                                                                 break;
2440                                                         }
2441                                                 }
2442                                         }
2443                                         continue;
2444                                 }
2445
2446                                 # lets set a title if it is a first separator, or next part otherwise
2447                                 if ($text[$i] == '|') {
2448                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2449                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2450                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2451                                         }
2452                                         else
2453                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2454
2455                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2456                                 }
2457                         }
2458                 }
2459
2460                 return $text;
2461         }
2462
2463         /**
2464          * Replace magic variables, templates, and template arguments
2465          * with the appropriate text. Templates are substituted recursively,
2466          * taking care to avoid infinite loops.
2467          *
2468          * Note that the substitution depends on value of $mOutputType:
2469          *  OT_WIKI: only {{subst:}} templates
2470          *  OT_MSG: only magic variables
2471          *  OT_HTML: all templates and magic variables
2472          *
2473          * @param string $tex The text to transform
2474          * @param array $args Key-value pairs representing template parameters to substitute
2475          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2476          * @private
2477          */
2478         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2479                 # Prevent too big inclusions
2480                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2481                         return $text;
2482                 }
2483
2484                 $fname = 'Parser::replaceVariables';
2485                 wfProfileIn( $fname );
2486
2487                 # This function is called recursively. To keep track of arguments we need a stack:
2488                 array_push( $this->mArgStack, $args );
2489
2490                 $braceCallbacks = array();
2491                 if ( !$argsOnly ) {
2492                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2493                 }
2494                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2495                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2496                 }
2497                 $callbacks = array();
2498                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2499                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2500                 $text = $this->replace_callback ($text, $callbacks);
2501
2502                 array_pop( $this->mArgStack );
2503
2504                 wfProfileOut( $fname );
2505                 return $text;
2506         }
2507
2508         /**
2509          * Replace magic variables
2510          * @private
2511          */
2512         function variableSubstitution( $matches ) {
2513                 $fname = 'Parser::variableSubstitution';
2514                 $varname = $matches[1];
2515                 wfProfileIn( $fname );
2516                 if ( !$this->mVariables ) {
2517                         $this->initialiseVariables();
2518                 }
2519                 $skip = false;
2520                 if ( $this->mOutputType == OT_WIKI ) {
2521                         # Do only magic variables prefixed by SUBST
2522                         $mwSubst =& MagicWord::get( MAG_SUBST );
2523                         if (!$mwSubst->matchStartAndRemove( $varname ))
2524                                 $skip = true;
2525                         # Note that if we don't substitute the variable below,
2526                         # we don't remove the {{subst:}} magic word, in case
2527                         # it is a template rather than a magic variable.
2528                 }
2529                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2530                         $id = $this->mVariables[$varname];
2531                         $text = $this->getVariableValue( $id );
2532                         $this->mOutput->mContainsOldMagic = true;
2533                 } else {
2534                         $text = $matches[0];
2535                 }
2536                 wfProfileOut( $fname );
2537                 return $text;
2538         }
2539
2540         # Split template arguments
2541         function getTemplateArgs( $argsString ) {
2542                 if ( $argsString === '' ) {
2543                         return array();
2544                 }
2545
2546                 $args = explode( '|', substr( $argsString, 1 ) );
2547
2548                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2549                 # merged with the next arg because the '|' character between belongs
2550                 # to the link syntax and not the template parameter syntax.
2551                 $argc = count($args);
2552
2553                 for ( $i = 0; $i < $argc-1; $i++ ) {
2554                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2555                                 $args[$i] .= '|'.$args[$i+1];
2556                                 array_splice($args, $i+1, 1);
2557                                 $i--;
2558                                 $argc--;
2559                         }
2560                 }
2561
2562                 return $args;
2563         }
2564
2565         /**
2566          * Return the text of a template, after recursively
2567          * replacing any variables or templates within the template.
2568          *
2569          * @param array $piece The parts of the template
2570          *  $piece['text']: matched text
2571          *  $piece['title']: the title, i.e. the part before the |
2572          *  $piece['parts']: the parameter array
2573          * @return string the text of the template
2574          * @private
2575          */
2576         function braceSubstitution( $piece ) {
2577                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2578                 $fname = 'Parser::braceSubstitution';
2579                 wfProfileIn( $fname );
2580
2581                 # Flags
2582                 $found = false;             # $text has been filled
2583                 $nowiki = false;            # wiki markup in $text should be escaped
2584                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2585                 $noargs = false;            # Don't replace triple-brace arguments in $text
2586                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2587                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2588                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2589
2590                 # Title object, where $text came from
2591                 $title = NULL;
2592
2593                 $linestart = '';
2594
2595                 # $part1 is the bit before the first |, and must contain only title characters
2596                 # $args is a list of arguments, starting from index 0, not including $part1
2597
2598                 $part1 = $piece['title'];
2599                 # If the third subpattern matched anything, it will start with |
2600
2601                 if (null == $piece['parts']) {
2602                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2603                         if ($replaceWith != $piece['text']) {
2604                                 $text = $replaceWith;
2605                                 $found = true;
2606                                 $noparse = true;
2607                                 $noargs = true;
2608                         }
2609                 }
2610
2611                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2612                 $argc = count( $args );
2613
2614                 # SUBST
2615                 if ( !$found ) {
2616                         $mwSubst =& MagicWord::get( MAG_SUBST );
2617                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2618                                 # One of two possibilities is true:
2619                                 # 1) Found SUBST but not in the PST phase
2620                                 # 2) Didn't find SUBST and in the PST phase
2621                                 # In either case, return without further processing
2622                                 $text = $piece['text'];
2623                                 $found = true;
2624                                 $noparse = true;
2625                                 $noargs = true;
2626                         }
2627                 }
2628
2629                 # MSG, MSGNW, INT and RAW
2630                 if ( !$found ) {
2631                         # Check for MSGNW:
2632                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2633                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2634                                 $nowiki = true;
2635                         } else {
2636                                 # Remove obsolete MSG:
2637                                 $mwMsg =& MagicWord::get( MAG_MSG );
2638                                 $mwMsg->matchStartAndRemove( $part1 );
2639                         }
2640
2641                         # Check for RAW:
2642                         $mwRaw =& MagicWord::get( MAG_RAW );
2643                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2644                                 $forceRawInterwiki = true;
2645                         }
2646
2647                         # Check if it is an internal message
2648                         $mwInt =& MagicWord::get( MAG_INT );
2649                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2650                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2651                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2652                                         $found = true;
2653                                 }
2654                         }
2655                 }
2656
2657                 # NS
2658                 if ( !$found ) {
2659                         # Check for NS: (namespace expansion)
2660                         $mwNs = MagicWord::get( MAG_NS );
2661                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2662                                 if ( intval( $part1 ) || $part1 == "0" ) {
2663                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2664                                         $found = true;
2665                                 } else {
2666                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2667                                         if ( !is_null( $index ) ) {
2668                                                 $text = $linestart . $wgContLang->getNsText( $index );
2669                                                 $found = true;
2670                                         }
2671                                 }
2672                         }
2673                 }
2674
2675                 # URLENCODE
2676                 if( !$found ) {
2677                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2678                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2679                                 $text = $linestart . urlencode( $part1 );
2680                                 $found = true;
2681                         }
2682                 }
2683
2684                 # LCFIRST, UCFIRST, LC and UC
2685                 if ( !$found ) {
2686                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2687                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2688                         $lc =& MagicWord::get( MAG_LC );
2689                         $uc =& MagicWord::get( MAG_UC );
2690                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2691                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2692                                 $found = true;
2693                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2694                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2695                                 $found = true;
2696                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2697                                 $text = $linestart . $wgContLang->lc( $part1 );
2698                                 $found = true;
2699                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2700                                  $text = $linestart . $wgContLang->uc( $part1 );
2701                                  $found = true;
2702                         }
2703                 }
2704
2705                 # LOCALURL and FULLURL
2706                 if ( !$found ) {
2707                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2708                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2709                         $mwFull =& MagicWord::get( MAG_FULLURL );
2710                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2711
2712
2713                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2714                                 $func = 'getLocalURL';
2715                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2716                                 $func = 'escapeLocalURL';
2717                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2718                                 $func = 'getFullURL';
2719                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2720                                 $func = 'escapeFullURL';
2721                         } else {
2722                                 $func = false;
2723                         }
2724
2725                         if ( $func !== false ) {
2726                                 $title = Title::newFromText( $part1 );
2727                                 # Due to order of execution of a lot of bits, the values might be encoded
2728                                 # before arriving here; if that's true, then the title can't be created
2729                                 # and the variable will fail. If we can't get a decent title from the first
2730                                 # attempt, url-decode and try for a second.
2731                                 if( is_null( $title ) )
2732                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2733                                 if ( !is_null( $title ) ) {
2734                                         if ( $argc > 0 ) {
2735                                                 $text = $linestart . $title->$func( $args[0] );
2736                                         } else {
2737                                                 $text = $linestart . $title->$func();
2738                                         }
2739                                         $found = true;
2740                                 }
2741                         }
2742                 }
2743
2744                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2745                 # GRAMMAR
2746                 if ( !$found && $argc == 1 ) {
2747                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2748                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2749                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2750                                 $found = true;
2751                         }
2752                 }
2753
2754                 # PLURAL
2755                 if ( !$found && $argc >= 2 ) {
2756                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2757                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2758                                 while ( count($args) < 5 ) { $args[] = $args[count($args)-1]; }
2759                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1],
2760                                         $args[2], $args[3], $args[4]);
2761                                 $found = true;
2762                         }
2763                 }
2764
2765                 # DISPLAYTITLE
2766                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2767                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2768                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2769
2770                                 # Set title in parser output object
2771                                 $param = $args[0];
2772                                 $parserOptions = new ParserOptions;
2773                                 $local_parser = new Parser ();
2774                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2775                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2776
2777                                 # Add subtitle
2778                                 $t = $this->mTitle->getPrefixedText();
2779                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2780                                 $text = "" ;
2781                                 $found = true ;
2782                         }
2783                 }
2784
2785                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2786                 if( !$found ) {
2787                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2788                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2789                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2790                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2791                         foreach( $mwWordsToCheck as $word => $func ) {
2792                                 $mwCurrentWord =& MagicWord::get( $word );
2793                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2794                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2795                                         if( $mwRawSuffix->match( $args[0] ) ) {
2796                                                 # Raw and unformatted
2797                                                 $text = $linestart . call_user_func( $func );
2798                                         } else {
2799                                                 # Formatted according to the content default
2800                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2801                                         }
2802                                         $found = true;
2803                                 }
2804                         }
2805                 }
2806
2807                 # PAGESINNAMESPACE
2808                 if( !$found ) {
2809                         $mwPagesInNs =& MagicWord::get( MAG_PAGESINNAMESPACE );
2810                         if( $mwPagesInNs->matchStartAndRemove( $part1 ) ) {
2811                                 $found = true;
2812                                 $count = wfPagesInNs( intval( $part1 ) );
2813                                 $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2814                                 if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2815                                         $text = $linestart . $count;
2816                                 } else {
2817                                         $text = $linestart . $wgContLang->formatNum( $count );
2818                                 }
2819                         }
2820                 }
2821
2822                 # #LANGUAGE:
2823                 if( !$found ) {
2824                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2825                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2826                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2827                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2828                                 $found = true;
2829                         }
2830                 }
2831
2832                 # Extensions
2833                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2834                         $colonPos = strpos( $part1, ':' );
2835                         if ( $colonPos !== false ) {
2836                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2837                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2838                                         $funcArgs = array_map( 'trim', $args );
2839                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2840                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2841                                         $found = true;
2842
2843                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2844                                         //$noargs = true;
2845                                         //$noparse = true;
2846
2847                                         if ( is_array( $result ) ) {
2848                                                 $text = $linestart . $result[0];
2849                                                 unset( $result[0] );
2850
2851                                                 // Extract flags into the local scope
2852                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2853                                                 extract( $result );
2854                                         } else {
2855                                                 $text = $linestart . $result;
2856                                         }
2857                                 }
2858                         }
2859                 }
2860
2861                 # Template table test
2862
2863                 # Did we encounter this template already? If yes, it is in the cache
2864                 # and we need to check for loops.
2865                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2866                         $found = true;
2867
2868                         # Infinite loop test
2869                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2870                                 $noparse = true;
2871                                 $noargs = true;
2872                                 $found = true;
2873                                 $text = $linestart .
2874                                         '{{' . $part1 . '}}' .
2875                                         '<!-- WARNING: template loop detected -->';
2876                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2877                         } else {
2878                                 # set $text to cached message.
2879                                 $text = $linestart . $this->mTemplates[$piece['title']];
2880                         }
2881                 }
2882
2883                 # Load from database
2884                 $lastPathLevel = $this->mTemplatePath;
2885                 if ( !$found ) {
2886                         $ns = NS_TEMPLATE;
2887                         # declaring $subpage directly in the function call
2888                         # does not work correctly with references and breaks
2889                         # {{/subpage}}-style inclusions
2890                         $subpage = '';
2891                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2892                         if ($subpage !== '') {
2893                                 $ns = $this->mTitle->getNamespace();
2894                         }
2895                         $title = Title::newFromText( $part1, $ns );
2896
2897                         if ( !is_null( $title ) ) {
2898                                 if ( !$title->isExternal() ) {
2899                                         # Check for excessive inclusion
2900                                         $dbk = $title->getPrefixedDBkey();
2901                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2902                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2903                                                         $text = SpecialPage::capturePath( $title );
2904                                                         if ( is_string( $text ) ) {
2905                                                                 $found = true;
2906                                                                 $noparse = true;
2907                                                                 $noargs = true;
2908                                                                 $isHTML = true;
2909                                                                 $this->disableCache();
2910                                                         }
2911                                                 } else {
2912                                                         $articleContent = $this->fetchTemplate( $title );
2913                                                         if ( $articleContent !== false ) {
2914                                                                 $found = true;
2915                                                                 $text = $articleContent;
2916                                                                 $replaceHeadings = true;
2917                                                         }
2918                                                 }
2919                                         }
2920
2921                                         # If the title is valid but undisplayable, make a link to it
2922                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2923                                                 $text = '[['.$title->getPrefixedText().']]';
2924                                                 $found = true;
2925                                         }
2926                                 } elseif ( $title->isTrans() ) {
2927                                         // Interwiki transclusion
2928                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2929                                                 $text = $this->interwikiTransclude( $title, 'render' );
2930                                                 $isHTML = true;
2931                                                 $noparse = true;
2932                                         } else {
2933                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2934                                                 $replaceHeadings = true;
2935                                         }
2936                                         $found = true;
2937                                 }
2938
2939                                 # Template cache array insertion
2940                                 # Use the original $piece['title'] not the mangled $part1, so that
2941                                 # modifiers such as RAW: produce separate cache entries
2942                                 if( $found ) {
2943                                         if( $isHTML ) {
2944                                                 // A special page; don't store it in the template cache.
2945                                         } else {
2946                                                 $this->mTemplates[$piece['title']] = $text;
2947                                         }
2948                                         $text = $linestart . $text;
2949                                 }
2950                         }
2951                 }
2952
2953                 # Recursive parsing, escaping and link table handling
2954                 # Only for HTML output
2955                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2956                         $text = wfEscapeWikiText( $text );
2957                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2958                         if ( $noargs ) {
2959                                 $assocArgs = array();
2960                         } else {
2961                                 # Clean up argument array
2962                                 $assocArgs = array();
2963                                 $index = 1;
2964                                 foreach( $args as $arg ) {
2965                                         $eqpos = strpos( $arg, '=' );
2966                                         if ( $eqpos === false ) {
2967                                                 $assocArgs[$index++] = $arg;
2968                                         } else {
2969                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2970                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2971                                                 if ( $value === false ) {
2972                                                         $value = '';
2973                                                 }
2974                                                 if ( $name !== false ) {
2975                                                         $assocArgs[$name] = $value;
2976                                                 }
2977                                         }
2978                                 }
2979
2980                                 # Add a new element to the templace recursion path
2981                                 $this->mTemplatePath[$part1] = 1;
2982                         }
2983
2984                         if ( !$noparse ) {
2985                                 # If there are any <onlyinclude> tags, only include them
2986                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2987                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2988                                         $text = '';
2989                                         foreach ($m[1] as $piece)
2990                                                 $text .= $piece;
2991                                 }
2992                                 # Remove <noinclude> sections and <includeonly> tags
2993                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2994                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2995
2996                                 if( $this->mOutputType == OT_HTML ) {
2997                                         # Strip <nowiki>, <pre>, etc.
2998                                         $text = $this->strip( $text, $this->mStripState );
2999                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3000                                 }
3001                                 $text = $this->replaceVariables( $text, $assocArgs );
3002
3003                                 # If the template begins with a table or block-level
3004                                 # element, it should be treated as beginning a new line.
3005                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3006                                         $text = "\n" . $text;
3007                                 }
3008                         } elseif ( !$noargs ) {
3009                                 # $noparse and !$noargs
3010                                 # Just replace the arguments, not any double-brace items
3011                                 # This is used for rendered interwiki transclusion
3012                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3013                         }
3014                 }
3015                 # Prune lower levels off the recursion check path
3016                 $this->mTemplatePath = $lastPathLevel;
3017
3018                 if ( !$found ) {
3019                         wfProfileOut( $fname );
3020                         return $piece['text'];
3021                 } else {
3022                         if ( $isHTML ) {
3023                                 # Replace raw HTML by a placeholder
3024                                 # Add a blank line preceding, to prevent it from mucking up
3025                                 # immediately preceding headings
3026                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3027                         } else {
3028                                 # replace ==section headers==
3029                                 # XXX this needs to go away once we have a better parser.
3030                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3031                                         if( !is_null( $title ) )
3032                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3033                                         else
3034                                                 $encodedname = base64_encode("");
3035                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3036                                                 PREG_SPLIT_DELIM_CAPTURE);
3037                                         $text = '';
3038                                         $nsec = 0;
3039                                         for( $i = 0; $i < count($m); $i += 2 ) {
3040                                                 $text .= $m[$i];
3041                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3042                                                 $hl = $m[$i + 1];
3043                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3044                                                         $text .= $hl;
3045                                                         continue;
3046                                                 }
3047                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3048                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3049                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3050
3051                                                 $nsec++;
3052                                         }
3053                                 }
3054                         }
3055                 }
3056
3057                 # Prune lower levels off the recursion check path
3058                 $this->mTemplatePath = $lastPathLevel;
3059
3060                 if ( !$found ) {
3061                         wfProfileOut( $fname );
3062                         return $piece['text'];
3063                 } else {
3064                         wfProfileOut( $fname );
3065                         return $text;
3066                 }
3067         }
3068
3069         /**
3070          * Fetch the unparsed text of a template and register a reference to it.
3071          */
3072         function fetchTemplate( $title ) {
3073                 $text = false;
3074                 // Loop to fetch the article, with up to 1 redirect
3075                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3076                         $rev = Revision::newFromTitle( $title );
3077                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3078                         if ( !$rev ) {
3079                                 break;
3080                         }
3081                         $text = $rev->getText();
3082                         if ( $text === false ) {
3083                                 break;
3084                         }
3085                         // Redirect?
3086                         $title = Title::newFromRedirect( $text );
3087                 }
3088                 return $text;
3089         }
3090
3091         /**
3092          * Transclude an interwiki link.
3093          */
3094         function interwikiTransclude( $title, $action ) {
3095                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3096
3097                 if (!$wgEnableScaryTranscluding)
3098                         return wfMsg('scarytranscludedisabled');
3099
3100                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3101                 // But we'll handle it generally anyway
3102                 if ( $title->getNamespace() ) {
3103                         // Use the canonical namespace, which should work anywhere
3104                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3105                 } else {
3106                         $articleName = $title->getDBkey();
3107                 }
3108
3109                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3110                 $url .= "?action=$action";
3111                 if (strlen($url) > 255)
3112                         return wfMsg('scarytranscludetoolong');
3113                 return $this->fetchScaryTemplateMaybeFromCache($url);
3114         }
3115
3116         function fetchScaryTemplateMaybeFromCache($url) {
3117                 global $wgTranscludeCacheExpiry;
3118                 $dbr =& wfGetDB(DB_SLAVE);
3119                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3120                                 array('tc_url' => $url));
3121                 if ($obj) {
3122                         $time = $obj->tc_time;
3123                         $text = $obj->tc_contents;
3124                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3125                                 return $text;
3126                         }
3127                 }
3128
3129                 $text = wfGetHTTP($url);
3130                 if (!$text)
3131                         return wfMsg('scarytranscludefailed', $url);
3132
3133                 $dbw =& wfGetDB(DB_MASTER);
3134                 $dbw->replace('transcache', array('tc_url'), array(
3135                         'tc_url' => $url,
3136                         'tc_time' => time(),
3137                         'tc_contents' => $text));
3138                 return $text;
3139         }
3140
3141
3142         /**
3143          * Triple brace replacement -- used for template arguments
3144          * @private
3145          */
3146         function argSubstitution( $matches ) {
3147                 $arg = trim( $matches['title'] );
3148                 $text = $matches['text'];
3149                 $inputArgs = end( $this->mArgStack );
3150
3151                 if ( array_key_exists( $arg, $inputArgs ) ) {
3152                         $text = $inputArgs[$arg];
3153                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3154                         $text = $matches['parts'][0];
3155                 }
3156
3157                 return $text;
3158         }
3159
3160         /**
3161          * Returns true if the function is allowed to include this entity
3162          * @private
3163          */
3164         function incrementIncludeCount( $dbk ) {
3165                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3166                         $this->mIncludeCount[$dbk] = 0;
3167                 }
3168                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3169                         return true;
3170                 } else {
3171                         return false;
3172                 }
3173         }
3174
3175         /**
3176          * Detect __TOC__ magic word and set a placeholder
3177          */
3178         function stripToc( $text ) {
3179                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3180                 # do not add TOC
3181                 $mw = MagicWord::get( MAG_NOTOC );
3182                 if( $mw->matchAndRemove( $text ) ) {
3183                         $this->mShowToc = false;
3184                 }
3185
3186                 $mw = MagicWord::get( MAG_TOC );
3187                 if( $mw->match( $text ) ) {
3188                         $this->mShowToc = true;
3189                         $this->mForceTocPosition = true;
3190
3191                         // Set a placeholder. At the end we'll fill it in with the TOC.
3192                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3193
3194                         // Only keep the first one.
3195                         $text = $mw->replace( '', $text );
3196                 }
3197                 return $text;
3198         }
3199
3200         /**
3201          * This function accomplishes several tasks:
3202          * 1) Auto-number headings if that option is enabled
3203          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3204          * 3) Add a Table of contents on the top for users who have enabled the option
3205          * 4) Auto-anchor headings
3206          *
3207          * It loops through all headlines, collects the necessary data, then splits up the
3208          * string and re-inserts the newly formatted headlines.
3209          *
3210          * @param string $text
3211          * @param boolean $isMain
3212          * @private
3213          */
3214         function formatHeadings( $text, $isMain=true ) {
3215                 global $wgMaxTocLevel, $wgContLang;
3216
3217                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3218                 if( !$this->mTitle->userCanEdit() ) {
3219                         $showEditLink = 0;
3220                 } else {
3221                         $showEditLink = $this->mOptions->getEditSection();
3222                 }
3223
3224                 # Inhibit editsection links if requested in the page
3225                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3226                 if( $esw->matchAndRemove( $text ) ) {
3227                         $showEditLink = 0;
3228                 }
3229
3230                 # Get all headlines for numbering them and adding funky stuff like [edit]
3231                 # links - this is for later, but we need the number of headlines right now
3232                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3233
3234                 # if there are fewer than 4 headlines in the article, do not show TOC
3235                 # unless it's been explicitly enabled.
3236                 $enoughToc = $this->mShowToc &&
3237                         (($numMatches >= 4) || $this->mForceTocPosition);
3238
3239                 # Allow user to stipulate that a page should have a "new section"
3240                 # link added via __NEWSECTIONLINK__
3241                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3242                 if( $mw->matchAndRemove( $text ) )
3243                         $this->mOutput->setNewSection( true );
3244
3245                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3246                 # override above conditions and always show TOC above first header
3247                 $mw =& MagicWord::get( MAG_FORCETOC );
3248                 if ($mw->matchAndRemove( $text ) ) {
3249                         $this->mShowToc = true;
3250                         $enoughToc = true;
3251                 }
3252
3253                 # Never ever show TOC if no headers
3254                 if( $numMatches < 1 ) {
3255                         $enoughToc = false;
3256                 }
3257
3258                 # We need this to perform operations on the HTML
3259                 $sk =& $this->mOptions->getSkin();
3260
3261                 # headline counter
3262                 $headlineCount = 0;
3263                 $sectionCount = 0; # headlineCount excluding template sections
3264
3265                 # Ugh .. the TOC should have neat indentation levels which can be
3266                 # passed to the skin functions. These are determined here
3267                 $toc = '';
3268                 $full = '';
3269                 $head = array();
3270                 $sublevelCount = array();
3271                 $levelCount = array();
3272                 $toclevel = 0;
3273                 $level = 0;
3274                 $prevlevel = 0;
3275                 $toclevel = 0;
3276                 $prevtoclevel = 0;
3277
3278                 foreach( $matches[3] as $headline ) {
3279                         $istemplate = 0;
3280                         $templatetitle = '';
3281                         $templatesection = 0;
3282                         $numbering = '';
3283
3284                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3285                                 $istemplate = 1;
3286                                 $templatetitle = base64_decode($mat[1]);
3287                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3288                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3289                         }
3290
3291                         if( $toclevel ) {
3292                                 $prevlevel = $level;
3293                                 $prevtoclevel = $toclevel;
3294                         }
3295                         $level = $matches[1][$headlineCount];
3296
3297                         if( $doNumberHeadings || $enoughToc ) {
3298
3299                                 if ( $level > $prevlevel ) {
3300                                         # Increase TOC level
3301                                         $toclevel++;
3302                                         $sublevelCount[$toclevel] = 0;
3303                                         if( $toclevel<$wgMaxTocLevel ) {
3304                                                 $toc .= $sk->tocIndent();
3305                                         }
3306                                 }
3307                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3308                                         # Decrease TOC level, find level to jump to
3309
3310                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3311                                                 # Can only go down to level 1
3312                                                 $toclevel = 1;
3313                                         } else {
3314                                                 for ($i = $toclevel; $i > 0; $i--) {
3315                                                         if ( $levelCount[$i] == $level ) {
3316                                                                 # Found last matching level
3317                                                                 $toclevel = $i;
3318                                                                 break;
3319                                                         }
3320                                                         elseif ( $levelCount[$i] < $level ) {
3321                                                                 # Found first matching level below current level
3322                                                                 $toclevel = $i + 1;
3323                                                                 break;
3324                                                         }
3325                                                 }
3326                                         }
3327                                         if( $toclevel<$wgMaxTocLevel ) {
3328                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3329                                         }
3330                                 }
3331                                 else {
3332                                         # No change in level, end TOC line
3333                                         if( $toclevel<$wgMaxTocLevel ) {
3334                                                 $toc .= $sk->tocLineEnd();
3335                                         }
3336                                 }
3337
3338                                 $levelCount[$toclevel] = $level;
3339
3340                                 # count number of headlines for each level
3341                                 @$sublevelCount[$toclevel]++;
3342                                 $dot = 0;
3343                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3344                                         if( !empty( $sublevelCount[$i] ) ) {
3345                                                 if( $dot ) {
3346                                                         $numbering .= '.';
3347                                                 }
3348                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3349                                                 $dot = 1;
3350                                         }
3351                                 }
3352                         }
3353
3354                         # The canonized header is a version of the header text safe to use for links
3355                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3356                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3357                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3358
3359                         # Remove link placeholders by the link text.
3360                         #     <!--LINK number-->
3361                         # turns into
3362                         #     link text with suffix
3363                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3364                                                             "\$this->mLinkHolders['texts'][\$1]",
3365                                                             $canonized_headline );
3366                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3367                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3368                                                             $canonized_headline );
3369
3370                         # strip out HTML
3371                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3372                         $tocline = trim( $canonized_headline );
3373                         # Save headline for section edit hint before it's escaped
3374                         $headline_hint = trim( $canonized_headline );
3375                         $canonized_headline = Sanitizer::escapeId( $tocline );
3376                         $refers[$headlineCount] = $canonized_headline;
3377
3378                         # count how many in assoc. array so we can track dupes in anchors
3379                         @$refers[$canonized_headline]++;
3380                         $refcount[$headlineCount]=$refers[$canonized_headline];
3381
3382                         # Don't number the heading if it is the only one (looks silly)
3383                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3384                                 # the two are different if the line contains a link
3385                                 $headline=$numbering . ' ' . $headline;
3386                         }
3387
3388                         # Create the anchor for linking from the TOC to the section
3389                         $anchor = $canonized_headline;
3390                         if($refcount[$headlineCount] > 1 ) {
3391                                 $anchor .= '_' . $refcount[$headlineCount];
3392                         }
3393                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3394                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3395                         }
3396                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3397                                 if ( empty( $head[$headlineCount] ) ) {
3398                                         $head[$headlineCount] = '';
3399                                 }
3400                                 if( $istemplate )
3401                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3402                                 else
3403                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3404                         }
3405
3406                         # give headline the correct <h#> tag
3407                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3408
3409                         $headlineCount++;
3410                         if( !$istemplate )
3411                                 $sectionCount++;
3412                 }
3413
3414                 if( $enoughToc ) {
3415                         if( $toclevel<$wgMaxTocLevel ) {
3416                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3417                         }
3418                         $toc = $sk->tocList( $toc );
3419                 }
3420
3421                 # split up and insert constructed headlines
3422
3423                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3424                 $i = 0;
3425
3426                 foreach( $blocks as $block ) {
3427                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3428                                 # This is the [edit] link that appears for the top block of text when
3429                                 # section editing is enabled
3430
3431                                 # Disabled because it broke block formatting
3432                                 # For example, a bullet point in the top line
3433                                 # $full .= $sk->editSectionLink(0);
3434                         }
3435                         $full .= $block;
3436                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3437                                 # Top anchor now in skin
3438                                 $full = $full.$toc;
3439                         }
3440
3441                         if( !empty( $head[$i] ) ) {
3442                                 $full .= $head[$i];
3443                         }
3444                         $i++;
3445                 }
3446                 if( $this->mForceTocPosition ) {
3447                         return str_replace( '<!--MWTOC-->', $toc, $full );
3448                 } else {
3449                         return $full;
3450                 }
3451         }
3452
3453         /**
3454          * Return an HTML link for the "ISBN 123456" text
3455          * @private
3456          */
3457         function magicISBN( $text ) {
3458                 $fname = 'Parser::magicISBN';
3459                 wfProfileIn( $fname );
3460
3461                 $a = split( 'ISBN ', ' '.$text );
3462                 if ( count ( $a ) < 2 ) {
3463                         wfProfileOut( $fname );
3464                         return $text;
3465                 }
3466                 $text = substr( array_shift( $a ), 1);
3467                 $valid = '0123456789-Xx';
3468
3469                 foreach ( $a as $x ) {
3470                         # hack: don't replace inside thumbnail title/alt
3471                         # attributes
3472                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3473                                 $text .= "ISBN $x";
3474                                 continue;
3475                         }
3476
3477                         $isbn = $blank = '' ;
3478                         while ( ' ' == $x{0} ) {
3479                                 $blank .= ' ';
3480                                 $x = substr( $x, 1 );
3481                         }
3482                         if ( $x == '' ) { # blank isbn
3483                                 $text .= "ISBN $blank";
3484                                 continue;
3485                         }
3486                         while ( strstr( $valid, $x{0} ) != false ) {
3487                                 $isbn .= $x{0};
3488                                 $x = substr( $x, 1 );
3489                         }
3490                         $num = str_replace( '-', '', $isbn );
3491                         $num = str_replace( ' ', '', $num );
3492                         $num = str_replace( 'x', 'X', $num );
3493
3494                         if ( '' == $num ) {
3495                                 $text .= "ISBN $blank$x";
3496                         } else {
3497                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3498                                 $text .= '<a href="' .
3499                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3500                                         "\" class=\"internal\">ISBN $isbn</a>";
3501                                 $text .= $x;
3502                         }
3503                 }
3504                 wfProfileOut( $fname );
3505                 return $text;
3506         }
3507
3508         /**
3509          * Return an HTML link for the "RFC 1234" text
3510          *
3511          * @private
3512          * @param string $text     Text to be processed
3513          * @param string $keyword  Magic keyword to use (default RFC)
3514          * @param string $urlmsg   Interface message to use (default rfcurl)
3515          * @return string
3516          */
3517         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3518
3519                 $valid = '0123456789';
3520                 $internal = false;
3521
3522                 $a = split( $keyword, ' '.$text );
3523                 if ( count ( $a ) < 2 ) {
3524                         return $text;
3525                 }
3526                 $text = substr( array_shift( $a ), 1);
3527
3528                 /* Check if keyword is preceed by [[.
3529                  * This test is made here cause of the array_shift above
3530                  * that prevent the test to be done in the foreach.
3531                  */
3532                 if ( substr( $text, -2 ) == '[[' ) {
3533                         $internal = true;
3534                 }
3535
3536                 foreach ( $a as $x ) {
3537                         /* token might be empty if we have RFC RFC 1234 */
3538                         if ( $x=='' ) {
3539                                 $text.=$keyword;
3540                                 continue;
3541                                 }
3542
3543                         # hack: don't replace inside thumbnail title/alt
3544                         # attributes
3545                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3546                                 $text .= $keyword . $x;
3547                                 continue;
3548                         }
3549
3550                         $id = $blank = '' ;
3551
3552                         /** remove and save whitespaces in $blank */
3553                         while ( $x{0} == ' ' ) {
3554                                 $blank .= ' ';
3555                                 $x = substr( $x, 1 );
3556                         }
3557
3558                         /** remove and save the rfc number in $id */
3559                         while ( strstr( $valid, $x{0} ) != false ) {
3560                                 $id .= $x{0};
3561                                 $x = substr( $x, 1 );
3562                         }
3563
3564                         if ( $id == '' ) {
3565                                 /* call back stripped spaces*/
3566                                 $text .= $keyword.$blank.$x;
3567                         } elseif( $internal ) {
3568                                 /* normal link */
3569                                 $text .= $keyword.$id.$x;
3570                         } else {
3571                                 /* build the external link*/
3572                                 $url = wfMsg( $urlmsg, $id);
3573                                 $sk =& $this->mOptions->getSkin();
3574                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3575                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3576                         }
3577
3578                         /* Check if the next RFC keyword is preceed by [[ */
3579                         $internal = ( substr($x,-2) == '[[' );
3580                 }
3581                 return $text;
3582         }
3583
3584         /**
3585          * Transform wiki markup when saving a page by doing \r\n -> \n
3586          * conversion, substitting signatures, {{subst:}} templates, etc.
3587          *
3588          * @param string $text the text to transform
3589          * @param Title &$title the Title object for the current article
3590          * @param User &$user the User object describing the current user
3591          * @param ParserOptions $options parsing options
3592          * @param bool $clearState whether to clear the parser state first
3593          * @return string the altered wiki markup
3594          * @public
3595          */
3596         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3597                 $this->mOptions = $options;
3598                 $this->mTitle =& $title;
3599                 $this->mOutputType = OT_WIKI;
3600
3601                 if ( $clearState ) {
3602                         $this->clearState();
3603                 }
3604
3605                 $stripState = false;
3606                 $pairs = array(
3607                         "\r\n" => "\n",
3608                 );
3609                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3610                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3611                 $text = $this->pstPass2( $text, $stripState, $user );
3612                 $text = $this->unstrip( $text, $stripState );
3613                 $text = $this->unstripNoWiki( $text, $stripState );
3614                 return $text;
3615         }
3616
3617         /**
3618          * Pre-save transform helper function
3619          * @private
3620          */
3621         function pstPass2( $text, &$stripState, &$user ) {
3622                 global $wgContLang, $wgLocaltimezone;
3623
3624                 /* Note: This is the timestamp saved as hardcoded wikitext to
3625                  * the database, we use $wgContLang here in order to give
3626                  * everyone the same signature and use the default one rather
3627                  * than the one selected in each user's preferences.
3628                  */
3629                 if ( isset( $wgLocaltimezone ) ) {
3630                         $oldtz = getenv( 'TZ' );
3631                         putenv( 'TZ='.$wgLocaltimezone );
3632                 }
3633                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3634                   ' (' . date( 'T' ) . ')';
3635                 if ( isset( $wgLocaltimezone ) ) {
3636                         putenv( 'TZ='.$oldtz );
3637                 }
3638
3639                 # Variable replacement
3640                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3641                 $text = $this->replaceVariables( $text );
3642
3643                 # Strip out <nowiki> etc. added via replaceVariables
3644                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3645
3646                 # Signatures
3647                 $sigText = $this->getUserSig( $user );
3648                 $text = strtr( $text, array(
3649                         '~~~~~' => $d,
3650                         '~~~~' => "$sigText $d",
3651                         '~~~' => $sigText
3652                 ) );
3653
3654                 # Context links: [[|name]] and [[name (context)|]]
3655                 #
3656                 global $wgLegalTitleChars;
3657                 $tc = "[$wgLegalTitleChars]";
3658                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3659
3660                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3661                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3662
3663                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3664                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3665                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3666                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3667                 $context = '';
3668                 $t = $this->mTitle->getText();
3669                 if ( preg_match( $conpat, $t, $m ) ) {
3670                         $context = $m[2];
3671                 }
3672                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3673                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3674                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3675
3676                 if ( '' == $context ) {
3677                         $text = preg_replace( $p2, '[[\\1]]', $text );
3678                 } else {
3679                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3680                 }
3681
3682                 # Trim trailing whitespace
3683                 # MAG_END (__END__) tag allows for trailing
3684                 # whitespace to be deliberately included
3685                 $text = rtrim( $text );
3686                 $mw =& MagicWord::get( MAG_END );
3687                 $mw->matchAndRemove( $text );
3688
3689                 return $text;
3690         }
3691
3692         /**
3693          * Fetch the user's signature text, if any, and normalize to
3694          * validated, ready-to-insert wikitext.
3695          *
3696          * @param User $user
3697          * @return string
3698          * @private
3699          */
3700         function getUserSig( &$user ) {
3701                 $username = $user->getName();
3702                 $nickname = $user->getOption( 'nickname' );
3703                 $nickname = $nickname === '' ? $username : $nickname;
3704
3705                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3706                         # Sig. might contain markup; validate this
3707                         if( $this->validateSig( $nickname ) !== false ) {
3708                                 # Validated; clean up (if needed) and return it
3709                                 return $this->cleanSig( $nickname, true );
3710                         } else {
3711                                 # Failed to validate; fall back to the default
3712                                 $nickname = $username;
3713                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3714                         }
3715                 }
3716
3717                 # If we're still here, make it a link to the user page
3718                 $userpage = $user->getUserPage();
3719                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3720         }
3721
3722         /**
3723          * Check that the user's signature contains no bad XML
3724          *
3725          * @param string $text
3726          * @return mixed An expanded string, or false if invalid.
3727          */
3728         function validateSig( $text ) {
3729                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3730         }
3731
3732         /**
3733          * Clean up signature text
3734          *
3735          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3736          * 2) Substitute all transclusions
3737          *
3738          * @param string $text
3739          * @param $parsing Whether we're cleaning (preferences save) or parsing
3740          * @return string Signature text
3741          */
3742         function cleanSig( $text, $parsing = false ) {
3743                 global $wgTitle;
3744                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3745
3746                 $substWord = MagicWord::get( MAG_SUBST );
3747                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3748                 $substText = '{{' . $substWord->getSynonym( 0 );
3749
3750                 $text = preg_replace( $substRegex, $substText, $text );
3751                 $text = preg_replace( '/~{3,5}/', '', $text );
3752                 $text = $this->replaceVariables( $text );
3753
3754                 $this->clearState();
3755                 return $text;
3756         }
3757
3758         /**
3759          * Set up some variables which are usually set up in parse()
3760          * so that an external function can call some class members with confidence
3761          * @public
3762          */
3763         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3764                 $this->mTitle =& $title;
3765                 $this->mOptions = $options;
3766                 $this->mOutputType = $outputType;
3767                 if ( $clearState ) {
3768                         $this->clearState();
3769                 }
3770         }
3771
3772         /**
3773          * Transform a MediaWiki message by replacing magic variables.
3774          *
3775          * @param string $text the text to transform
3776          * @param ParserOptions $options  options
3777          * @return string the text with variables substituted
3778          * @public
3779          */
3780         function transformMsg( $text, $options ) {
3781                 global $wgTitle;
3782                 static $executing = false;
3783
3784                 $fname = "Parser::transformMsg";
3785
3786                 # Guard against infinite recursion
3787                 if ( $executing ) {
3788                         return $text;
3789                 }
3790                 $executing = true;
3791
3792                 wfProfileIn($fname);
3793
3794                 $this->mTitle = $wgTitle;
3795                 $this->mOptions = $options;
3796                 $this->mOutputType = OT_MSG;
3797                 $this->clearState();
3798                 $text = $this->replaceVariables( $text );
3799
3800                 $executing = false;
3801                 wfProfileOut($fname);
3802                 return $text;
3803         }
3804
3805         /**
3806          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3807          * The callback should have the following form:
3808          *    function myParserHook( $text, $params, &$parser ) { ... }
3809          *
3810          * Transform and return $text. Use $parser for any required context, e.g. use
3811          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3812          *
3813          * @public
3814          *
3815          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3816          * @param mixed $callback The callback function (and object) to use for the tag
3817          *
3818          * @return The old value of the mTagHooks array associated with the hook
3819          */
3820         function setHook( $tag, $callback ) {
3821                 $tag = strtolower( $tag );
3822                 $oldVal = @$this->mTagHooks[$tag];
3823                 $this->mTagHooks[$tag] = $callback;
3824
3825                 return $oldVal;
3826         }
3827
3828         /**
3829          * Create a function, e.g. {{sum:1|2|3}}
3830          * The callback function should have the form:
3831          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3832          *
3833          * The callback may either return the text result of the function, or an array with the text
3834          * in element 0, and a number of flags in the other elements. The names of the flags are
3835          * specified in the keys. Valid flags are:
3836          *   found                     The text returned is valid, stop processing the template. This
3837          *                             is on by default.
3838          *   nowiki                    Wiki markup in the return value should be escaped
3839          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3840          *   noargs                    Don't replace triple-brace arguments in the return value
3841          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3842          *
3843          * @public
3844          *
3845          * @param string $name The function name. Function names are case-insensitive.
3846          * @param mixed $callback The callback function (and object) to use
3847          *
3848          * @return The old callback function for this name, if any
3849          */
3850         function setFunctionHook( $name, $callback ) {
3851                 $name = strtolower( $name );
3852                 $oldVal = @$this->mFunctionHooks[$name];
3853                 $this->mFunctionHooks[$name] = $callback;
3854                 return $oldVal;
3855         }
3856
3857         /**
3858          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3859          * Placeholders created in Skin::makeLinkObj()
3860          * Returns an array of links found, indexed by PDBK:
3861          *  0 - broken
3862          *  1 - normal link
3863          *  2 - stub
3864          * $options is a bit field, RLH_FOR_UPDATE to select for update
3865          */
3866         function replaceLinkHolders( &$text, $options = 0 ) {
3867                 global $wgUser;
3868                 global $wgOutputReplace;
3869
3870                 $fname = 'Parser::replaceLinkHolders';
3871                 wfProfileIn( $fname );
3872
3873                 $pdbks = array();
3874                 $colours = array();
3875                 $sk =& $this->mOptions->getSkin();
3876                 $linkCache =& LinkCache::singleton();
3877
3878                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3879                         wfProfileIn( $fname.'-check' );
3880                         $dbr =& wfGetDB( DB_SLAVE );
3881                         $page = $dbr->tableName( 'page' );
3882                         $threshold = $wgUser->getOption('stubthreshold');
3883
3884                         # Sort by namespace
3885                         asort( $this->mLinkHolders['namespaces'] );
3886
3887                         # Generate query
3888                         $query = false;
3889                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3890                                 # Make title object
3891                                 $title = $this->mLinkHolders['titles'][$key];
3892
3893                                 # Skip invalid entries.
3894                                 # Result will be ugly, but prevents crash.
3895                                 if ( is_null( $title ) ) {
3896                                         continue;
3897                                 }
3898                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3899
3900                                 # Check if it's a static known link, e.g. interwiki
3901                                 if ( $title->isAlwaysKnown() ) {
3902                                         $colours[$pdbk] = 1;
3903                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3904                                         $colours[$pdbk] = 1;
3905                                         $this->mOutput->addLink( $title, $id );
3906                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3907                                         $colours[$pdbk] = 0;
3908                                 } else {
3909                                         # Not in the link cache, add it to the query
3910                                         if ( !isset( $current ) ) {
3911                                                 $current = $ns;
3912                                                 $query =  "SELECT page_id, page_namespace, page_title";
3913                                                 if ( $threshold > 0 ) {
3914                                                         $query .= ', page_len, page_is_redirect';
3915                                                 }
3916                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3917                                         } elseif ( $current != $ns ) {
3918                                                 $current = $ns;
3919                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3920                                         } else {
3921                                                 $query .= ', ';
3922                                         }
3923
3924                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3925                                 }
3926                         }
3927                         if ( $query ) {
3928                                 $query .= '))';
3929                                 if ( $options & RLH_FOR_UPDATE ) {
3930                                         $query .= ' FOR UPDATE';
3931                                 }
3932
3933                                 $res = $dbr->query( $query, $fname );
3934
3935                                 # Fetch data and form into an associative array
3936                                 # non-existent = broken
3937                                 # 1 = known
3938                                 # 2 = stub
3939                                 while ( $s = $dbr->fetchObject($res) ) {
3940                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3941                                         $pdbk = $title->getPrefixedDBkey();
3942                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3943                                         $this->mOutput->addLink( $title, $s->page_id );
3944
3945                                         if ( $threshold >  0 ) {
3946                                                 $size = $s->page_len;
3947                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3948                                                         $colours[$pdbk] = 1;
3949                                                 } else {
3950                                                         $colours[$pdbk] = 2;
3951                                                 }
3952                                         } else {
3953                                                 $colours[$pdbk] = 1;
3954                                         }
3955                                 }
3956                         }
3957                         wfProfileOut( $fname.'-check' );
3958
3959                         # Construct search and replace arrays
3960                         wfProfileIn( $fname.'-construct' );
3961                         $wgOutputReplace = array();
3962                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3963                                 $pdbk = $pdbks[$key];
3964                                 $searchkey = "<!--LINK $key-->";
3965                                 $title = $this->mLinkHolders['titles'][$key];
3966                                 if ( empty( $colours[$pdbk] ) ) {
3967                                         $linkCache->addBadLinkObj( $title );
3968                                         $colours[$pdbk] = 0;
3969                                         $this->mOutput->addLink( $title, 0 );
3970                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3971                                                                         $this->mLinkHolders['texts'][$key],
3972                                                                         $this->mLinkHolders['queries'][$key] );
3973                                 } elseif ( $colours[$pdbk] == 1 ) {
3974                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3975                                                                         $this->mLinkHolders['texts'][$key],
3976                                                                         $this->mLinkHolders['queries'][$key] );
3977                                 } elseif ( $colours[$pdbk] == 2 ) {
3978                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3979                                                                         $this->mLinkHolders['texts'][$key],
3980                                                                         $this->mLinkHolders['queries'][$key] );
3981                                 }
3982                         }
3983                         wfProfileOut( $fname.'-construct' );
3984
3985                         # Do the thing
3986                         wfProfileIn( $fname.'-replace' );
3987
3988                         $text = preg_replace_callback(
3989                                 '/(<!--LINK .*?-->)/',
3990                                 "wfOutputReplaceMatches",
3991                                 $text);
3992
3993                         wfProfileOut( $fname.'-replace' );
3994                 }
3995
3996                 # Now process interwiki link holders
3997                 # This is quite a bit simpler than internal links
3998                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3999                         wfProfileIn( $fname.'-interwiki' );
4000                         # Make interwiki link HTML
4001                         $wgOutputReplace = array();
4002                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4003                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4004                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4005                         }
4006
4007                         $text = preg_replace_callback(
4008                                 '/<!--IWLINK (.*?)-->/',
4009                                 "wfOutputReplaceMatches",
4010                                 $text );
4011                         wfProfileOut( $fname.'-interwiki' );
4012                 }
4013
4014                 wfProfileOut( $fname );
4015                 return $colours;
4016         }
4017
4018         /**
4019          * Replace <!--LINK--> link placeholders with plain text of links
4020          * (not HTML-formatted).
4021          * @param string $text
4022          * @return string
4023          */
4024         function replaceLinkHoldersText( $text ) {
4025                 $fname = 'Parser::replaceLinkHoldersText';
4026                 wfProfileIn( $fname );
4027
4028                 $text = preg_replace_callback(
4029                         '/<!--(LINK|IWLINK) (.*?)-->/',
4030                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4031                         $text );
4032
4033                 wfProfileOut( $fname );
4034                 return $text;
4035         }
4036
4037         /**
4038          * @param array $matches
4039          * @return string
4040          * @private
4041          */
4042         function replaceLinkHoldersTextCallback( $matches ) {
4043                 $type = $matches[1];
4044                 $key  = $matches[2];
4045                 if( $type == 'LINK' ) {
4046                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4047                                 return $this->mLinkHolders['texts'][$key];
4048                         }
4049                 } elseif( $type == 'IWLINK' ) {
4050                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4051                                 return $this->mInterwikiLinkHolders['texts'][$key];
4052                         }
4053                 }
4054                 return $matches[0];
4055         }
4056
4057         /**
4058          * Tag hook handler for 'pre'.
4059          */
4060         function renderPreTag( $text, $attribs, $parser ) {
4061                 // Backwards-compatibility hack
4062                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4063
4064                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4065                 return wfOpenElement( 'pre', $attribs ) .
4066                         wfEscapeHTMLTagsOnly( $content ) .
4067                         '</pre>';
4068         }
4069
4070         /**
4071          * Renders an image gallery from a text with one line per image.
4072          * text labels may be given by using |-style alternative text. E.g.
4073          *   Image:one.jpg|The number "1"
4074          *   Image:tree.jpg|A tree
4075          * given as text will return the HTML of a gallery with two images,
4076          * labeled 'The number "1"' and
4077          * 'A tree'.
4078          */
4079         function renderImageGallery( $text ) {
4080                 $ig = new ImageGallery();
4081                 $ig->setShowBytes( false );
4082                 $ig->setShowFilename( false );
4083                 $ig->setParsing();
4084                 $lines = explode( "\n", $text );
4085
4086                 foreach ( $lines as $line ) {
4087                         # match lines like these:
4088                         # Image:someimage.jpg|This is some image
4089                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4090                         # Skip empty lines
4091                         if ( count( $matches ) == 0 ) {
4092                                 continue;
4093                         }
4094                         $nt =& Title::newFromText( $matches[1] );
4095                         if( is_null( $nt ) ) {
4096                                 # Bogus title. Ignore these so we don't bomb out later.
4097                                 continue;
4098                         }
4099                         if ( isset( $matches[3] ) ) {
4100                                 $label = $matches[3];
4101                         } else {
4102                                 $label = '';
4103                         }
4104
4105                         $pout = $this->parse( $label,
4106                                 $this->mTitle,
4107                                 $this->mOptions,
4108                                 false, // Strip whitespace...?
4109                                 false  // Don't clear state!
4110                         );
4111                         $html = $pout->getText();
4112
4113                         $ig->add( new Image( $nt ), $html );
4114
4115                         # Only add real images (bug #5586)
4116                         if ( $nt->getNamespace() == NS_IMAGE ) {
4117                                 $this->mOutput->addImage( $nt->getDBkey() );
4118                         }
4119                 }
4120                 return $ig->toHTML();
4121         }
4122
4123         /**
4124          * Parse image options text and use it to make an image
4125          */
4126         function makeImage( &$nt, $options ) {
4127                 global $wgUseImageResize;
4128
4129                 $align = '';
4130
4131                 # Check if the options text is of the form "options|alt text"
4132                 # Options are:
4133                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4134                 #  * left               no resizing, just left align. label is used for alt= only
4135                 #  * right              same, but right aligned
4136                 #  * none               same, but not aligned
4137                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4138                 #  * center             center the image
4139                 #  * framed             Keep original image size, no magnify-button.
4140
4141                 $part = explode( '|', $options);
4142
4143                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4144                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4145                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4146                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4147                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4148                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4149                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4150                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4151                 $caption = '';
4152
4153                 $width = $height = $framed = $thumb = false;
4154                 $manual_thumb = '' ;
4155
4156                 foreach( $part as $key => $val ) {
4157                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4158                                 $thumb=true;
4159                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4160                                 # use manually specified thumbnail
4161                                 $thumb=true;
4162                                 $manual_thumb = $match;
4163                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4164                                 # remember to set an alignment, don't render immediately
4165                                 $align = 'right';
4166                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4167                                 # remember to set an alignment, don't render immediately
4168                                 $align = 'left';
4169                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4170                                 # remember to set an alignment, don't render immediately
4171                                 $align = 'center';
4172                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4173                                 # remember to set an alignment, don't render immediately
4174                                 $align = 'none';
4175                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4176                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4177                                 # $match is the image width in pixels
4178                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4179                                         $width = intval( $m[1] );
4180                                         $height = intval( $m[2] );
4181                                 } else {
4182                                         $width = intval($match);
4183                                 }
4184                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4185                                 $framed=true;
4186                         } else {
4187                                 $caption = $val;
4188                         }
4189                 }
4190                 # Strip bad stuff out of the alt text
4191                 $alt = $this->replaceLinkHoldersText( $caption );
4192
4193                 # make sure there are no placeholders in thumbnail attributes
4194                 # that are later expanded to html- so expand them now and
4195                 # remove the tags
4196                 $alt = $this->unstrip($alt, $this->mStripState);
4197                 $alt = Sanitizer::stripAllTags( $alt );
4198
4199                 # Linker does the rest
4200                 $sk =& $this->mOptions->getSkin();
4201                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4202         }
4203
4204         /**
4205          * Set a flag in the output object indicating that the content is dynamic and
4206          * shouldn't be cached.
4207          */
4208         function disableCache() {
4209                 wfDebug( "Parser output marked as uncacheable.\n" );
4210                 $this->mOutput->mCacheTime = -1;
4211         }
4212
4213         /**#@+
4214          * Callback from the Sanitizer for expanding items found in HTML attribute
4215          * values, so they can be safely tested and escaped.
4216          * @param string $text
4217          * @param array $args
4218          * @return string
4219          * @private
4220          */
4221         function attributeStripCallback( &$text, $args ) {
4222                 $text = $this->replaceVariables( $text, $args );
4223                 $text = $this->unstripForHTML( $text );
4224                 return $text;
4225         }
4226
4227         function unstripForHTML( $text ) {
4228                 $text = $this->unstrip( $text, $this->mStripState );
4229                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4230                 return $text;
4231         }
4232         /**#@-*/
4233
4234         /**#@+
4235          * Accessor/mutator
4236          */
4237         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4238         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4239         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4240         /**#@-*/
4241
4242         /**#@+
4243          * Accessor
4244          */
4245         function getTags() { return array_keys( $this->mTagHooks ); }
4246         /**#@-*/
4247
4248
4249         /**
4250          * Break wikitext input into sections, and either pull or replace
4251          * some particular section's text.
4252          *
4253          * External callers should use the getSection and replaceSection methods.
4254          *
4255          * @param $text Page wikitext
4256          * @param $section Numbered section. 0 pulls the text before the first
4257          *                 heading; other numbers will pull the given section
4258          *                 along with its lower-level subsections.
4259          * @param $mode One of "get" or "replace"
4260          * @param $newtext Replacement text for section data.
4261          * @return string for "get", the extracted section text.
4262          *                for "replace", the whole page with the section replaced.
4263          */
4264         private function extractSections( $text, $section, $mode, $newtext='' ) {
4265                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4266                 # comments to be stripped as well)
4267                 $striparray = array();
4268
4269                 $oldOutputType = $this->mOutputType;
4270                 $oldOptions = $this->mOptions;
4271                 $this->mOptions = new ParserOptions();
4272                 $this->mOutputType = OT_WIKI;
4273
4274                 $striptext = $this->strip( $text, $striparray, true );
4275
4276                 $this->mOutputType = $oldOutputType;
4277                 $this->mOptions = $oldOptions;
4278
4279                 # now that we can be sure that no pseudo-sections are in the source,
4280                 # split it up by section
4281                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4282                 $comment = "(?:$uniq-!--.*?QINU)";
4283                 $secs = preg_split(
4284                 /*
4285                         "/
4286                         ^(
4287                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4288                         (?:
4289                                 (=+) # Should this be limited to 6?
4290                                 .+?  # Section title...
4291                                 \\2  # Ending = count must match start
4292                         |
4293                                 ^
4294                                 <h([1-6])\b.*?>
4295                                 .*?
4296                                 <\/h\\3\s*>
4297                         )
4298                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4299                         )$
4300                         /mix",
4301                 */
4302                         "/
4303                         (
4304                                 ^
4305                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4306                                 (=+) # Should this be limited to 6?
4307                                 .+?  # Section title...
4308                                 \\2  # Ending = count must match start
4309                                 (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4310                                 $
4311                         |
4312                                 <h([1-6])\b.*?>
4313                                 .*?
4314                                 <\/h\\3\s*>
4315                         )
4316                         /mix",
4317                         $striptext, -1,
4318                         PREG_SPLIT_DELIM_CAPTURE);
4319
4320                 if( $mode == "get" ) {
4321                         if( $section == 0 ) {
4322                                 // "Section 0" returns the content before any other section.
4323                                 $rv = $secs[0];
4324                         } else {
4325                                 $rv = "";
4326                         }
4327                 } elseif( $mode == "replace" ) {
4328                         if( $section == 0 ) {
4329                                 $rv = $newtext . "\n\n";
4330                                 $remainder = true;
4331                         } else {
4332                                 $rv = $secs[0];
4333                                 $remainder = false;
4334                         }
4335                 }
4336                 $count = 0;
4337                 $sectionLevel = 0;
4338                 for( $index = 1; $index < count( $secs ); ) {
4339                         $headerLine = $secs[$index++];
4340                         if( $secs[$index] ) {
4341                                 // A wiki header
4342                                 $headerLevel = strlen( $secs[$index++] );
4343                         } else {
4344                                 // An HTML header
4345                                 $index++;
4346                                 $headerLevel = intval( $secs[$index++] );
4347                         }
4348                         $content = $secs[$index++];
4349
4350                         $count++;
4351                         if( $mode == "get" ) {
4352                                 if( $count == $section ) {
4353                                         $rv = $headerLine . $content;
4354                                         $sectionLevel = $headerLevel;
4355                                 } elseif( $count > $section ) {
4356                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4357                                                 $rv .= $headerLine . $content;
4358                                         } else {
4359                                                 // Broke out to a higher-level section
4360                                                 break;
4361                                         }
4362                                 }
4363                         } elseif( $mode == "replace" ) {
4364                                 if( $count < $section ) {
4365                                         $rv .= $headerLine . $content;
4366                                 } elseif( $count == $section ) {
4367                                         $rv .= $newtext . "\n\n";
4368                                         $sectionLevel = $headerLevel;
4369                                 } elseif( $count > $section ) {
4370                                         if( $headerLevel <= $sectionLevel ) {
4371                                                 // Passed the section's sub-parts.
4372                                                 $remainder = true;
4373                                         }
4374                                         if( $remainder ) {
4375                                                 $rv .= $headerLine . $content;
4376                                         }
4377                                 }
4378                         }
4379                 }
4380                 # reinsert stripped tags
4381                 $rv = $this->unstrip( $rv, $striparray );
4382                 $rv = $this->unstripNoWiki( $rv, $striparray );
4383                 $rv = trim( $rv );
4384                 return $rv;
4385         }
4386
4387         /**
4388          * This function returns the text of a section, specified by a number ($section).
4389          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4390          * the first section before any such heading (section 0).
4391          *
4392          * If a section contains subsections, these are also returned.
4393          *
4394          * @param $text String: text to look in
4395          * @param $section Integer: section number
4396          * @return string text of the requested section
4397          */
4398         function getSection( $text, $section ) {
4399                 return $this->extractSections( $text, $section, "get" );
4400         }
4401
4402         function replaceSection( $oldtext, $section, $text ) {
4403                 return $this->extractSections( $oldtext, $section, "replace", $text );
4404         }
4405
4406 }
4407
4408 /**
4409  * @todo document
4410  * @package MediaWiki
4411  */
4412 class ParserOutput
4413 {
4414         var $mText,             # The output text
4415                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4416                 $mCategories,       # Map of category names to sort keys
4417                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4418                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4419                 $mVersion,          # Compatibility check
4420                 $mTitleText,        # title text of the chosen language variant
4421                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4422                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4423                 $mImages,           # DB keys of the images used, in the array key only
4424                 $mExternalLinks,    # External link URLs, in the key only
4425                 $mHTMLtitle,            # Display HTML title
4426                 $mSubtitle,                     # Additional subtitle
4427                 $mNewSection;           # Show a new section link?
4428
4429         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4430                 $containsOldMagic = false, $titletext = '' )
4431         {
4432                 $this->mText = $text;
4433                 $this->mLanguageLinks = $languageLinks;
4434                 $this->mCategories = $categoryLinks;
4435                 $this->mContainsOldMagic = $containsOldMagic;
4436                 $this->mCacheTime = '';
4437                 $this->mVersion = MW_PARSER_VERSION;
4438                 $this->mTitleText = $titletext;
4439                 $this->mLinks = array();
4440                 $this->mTemplates = array();
4441                 $this->mImages = array();
4442                 $this->mExternalLinks = array();
4443                 $this->mHTMLtitle = "" ;
4444                 $this->mSubtitle = "" ;
4445                 $this->mNewSection = false;
4446         }
4447
4448         function getText()                   { return $this->mText; }
4449         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4450         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4451         function &getCategories()            { return $this->mCategories; }
4452         function getCacheTime()              { return $this->mCacheTime; }
4453         function getTitleText()              { return $this->mTitleText; }
4454         function &getLinks()                 { return $this->mLinks; }
4455         function &getTemplates()             { return $this->mTemplates; }
4456         function &getImages()                { return $this->mImages; }
4457         function &getExternalLinks()         { return $this->mExternalLinks; }
4458
4459         function containsOldMagic()          { return $this->mContainsOldMagic; }
4460         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4461         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4462         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4463         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4464         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4465         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4466
4467         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4468         function addImage( $name )           { $this->mImages[$name] = 1; }
4469         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4470         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4471
4472         function setNewSection( $value ) {
4473                 $this->mNewSection = (bool)$value;
4474         }
4475         function getNewSection() {
4476                 return (bool)$this->mNewSection;
4477         }
4478
4479         function addLink( $title, $id ) {
4480                 $ns = $title->getNamespace();
4481                 $dbk = $title->getDBkey();
4482                 if ( !isset( $this->mLinks[$ns] ) ) {
4483                         $this->mLinks[$ns] = array();
4484                 }
4485                 $this->mLinks[$ns][$dbk] = $id;
4486         }
4487
4488         function addTemplate( $title, $id ) {
4489                 $ns = $title->getNamespace();
4490                 $dbk = $title->getDBkey();
4491                 if ( !isset( $this->mTemplates[$ns] ) ) {
4492                         $this->mTemplates[$ns] = array();
4493                 }
4494                 $this->mTemplates[$ns][$dbk] = $id;
4495         }
4496
4497         /**
4498          * Return true if this cached output object predates the global or
4499          * per-article cache invalidation timestamps, or if it comes from
4500          * an incompatible older version.
4501          *
4502          * @param string $touched the affected article's last touched timestamp
4503          * @return bool
4504          * @public
4505          */
4506         function expired( $touched ) {
4507                 global $wgCacheEpoch;
4508                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4509                        $this->getCacheTime() < $touched ||
4510                        $this->getCacheTime() <= $wgCacheEpoch ||
4511                        !isset( $this->mVersion ) ||
4512                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4513         }
4514 }
4515
4516 /**
4517  * Set options of the Parser
4518  * @todo document
4519  * @package MediaWiki
4520  */
4521 class ParserOptions
4522 {
4523         # All variables are private
4524         var $mUseTeX;                    # Use texvc to expand <math> tags
4525         var $mUseDynamicDates;           # Use DateFormatter to format dates
4526         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4527         var $mAllowExternalImages;       # Allow external images inline
4528         var $mAllowExternalImagesFrom;   # If not, any exception?
4529         var $mSkin;                      # Reference to the preferred skin
4530         var $mDateFormat;                # Date format index
4531         var $mEditSection;               # Create "edit section" links
4532         var $mNumberHeadings;            # Automatically number headings
4533         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4534         var $mTidy;                      # Ask for tidy cleanup
4535         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4536
4537         function getUseTeX()                        { return $this->mUseTeX; }
4538         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4539         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4540         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4541         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4542         function &getSkin()                         { return $this->mSkin; }
4543         function getDateFormat()                    { return $this->mDateFormat; }
4544         function getEditSection()                   { return $this->mEditSection; }
4545         function getNumberHeadings()                { return $this->mNumberHeadings; }
4546         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4547         function getTidy()                          { return $this->mTidy; }
4548         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4549
4550         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4551         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4552         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4553         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4554         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4555         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4556         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4557         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4558         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4559         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4560         function setSkin( &$x ) { $this->mSkin =& $x; }
4561         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4562
4563         function ParserOptions() {
4564                 global $wgUser;
4565                 $this->initialiseFromUser( $wgUser );
4566         }
4567
4568         /**
4569          * Get parser options
4570          * @static
4571          */
4572         function newFromUser( &$user ) {
4573                 $popts = new ParserOptions;
4574                 $popts->initialiseFromUser( $user );
4575                 return $popts;
4576         }
4577
4578         /** Get user options */
4579         function initialiseFromUser( &$userInput ) {
4580                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4581                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4582                 $fname = 'ParserOptions::initialiseFromUser';
4583                 wfProfileIn( $fname );
4584                 if ( !$userInput ) {
4585                         $user = new User;
4586                         $user->setLoaded( true );
4587                 } else {
4588                         $user =& $userInput;
4589                 }
4590
4591                 $this->mUseTeX = $wgUseTeX;
4592                 $this->mUseDynamicDates = $wgUseDynamicDates;
4593                 $this->mInterwikiMagic = $wgInterwikiMagic;
4594                 $this->mAllowExternalImages = $wgAllowExternalImages;
4595                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4596                 wfProfileIn( $fname.'-skin' );
4597                 $this->mSkin =& $user->getSkin();
4598                 wfProfileOut( $fname.'-skin' );
4599                 $this->mDateFormat = $user->getOption( 'date' );
4600                 $this->mEditSection = true;
4601                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4602                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4603                 $this->mTidy = false;
4604                 $this->mInterfaceMessage = false;
4605                 wfProfileOut( $fname );
4606         }
4607 }
4608
4609 /**
4610  * Callback function used by Parser::replaceLinkHolders()
4611  * to substitute link placeholders.
4612  */
4613 function &wfOutputReplaceMatches( $matches ) {
4614         global $wgOutputReplace;
4615         return $wgOutputReplace[$matches[1]];
4616 }
4617
4618 /**
4619  * Return the total number of articles
4620  */
4621 function wfNumberOfArticles() {
4622         global $wgNumberOfArticles;
4623
4624         wfLoadSiteStats();
4625         return $wgNumberOfArticles;
4626 }
4627
4628 /**
4629  * Return the number of files
4630  */
4631 function wfNumberOfFiles() {
4632         $fname = 'wfNumberOfFiles';
4633
4634         wfProfileIn( $fname );
4635         $dbr =& wfGetDB( DB_SLAVE );
4636         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4637         wfProfileOut( $fname );
4638
4639         return $numImages;
4640 }
4641
4642 /**
4643  * Return the number of user accounts
4644  * @return integer
4645  */
4646 function wfNumberOfUsers() {
4647         wfProfileIn( 'wfNumberOfUsers' );
4648         $dbr =& wfGetDB( DB_SLAVE );
4649         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4650         wfProfileOut( 'wfNumberOfUsers' );
4651         return (int)$count;
4652 }
4653
4654 /**
4655  * Return the total number of pages
4656  * @return integer
4657  */
4658 function wfNumberOfPages() {
4659         wfProfileIn( 'wfNumberOfPages' );
4660         $dbr =& wfGetDB( DB_SLAVE );
4661         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4662         wfProfileOut( 'wfNumberOfPages' );
4663         return (int)$count;
4664 }
4665
4666 /**
4667  * Count the number of pages in a particular namespace
4668  *
4669  * @param $ns Namespace
4670  * @return integer
4671  */
4672 function wfPagesInNs( $ns ) {
4673         static $pageCount = array();
4674         wfProfileIn( 'wfPagesInNs' );
4675         if( !isset( $pageCount[$ns] ) ) {
4676                 $dbr =& wfGetDB( DB_SLAVE );
4677                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4678         }
4679         wfProfileOut( 'wfPagesInNs' );
4680         return (int)$pageCount[$ns];
4681 }
4682
4683 /**
4684  * Get various statistics from the database
4685  * @private
4686  */
4687 function wfLoadSiteStats() {
4688         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4689         $fname = 'wfLoadSiteStats';
4690
4691         if ( -1 != $wgNumberOfArticles ) return;
4692         $dbr =& wfGetDB( DB_SLAVE );
4693         $s = $dbr->selectRow( 'site_stats',
4694                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4695                 array( 'ss_row_id' => 1 ), $fname
4696         );
4697
4698         if ( $s === false ) {
4699                 return;
4700         } else {
4701                 $wgTotalViews = $s->ss_total_views;
4702                 $wgTotalEdits = $s->ss_total_edits;
4703                 $wgNumberOfArticles = $s->ss_good_articles;
4704         }
4705 }
4706
4707 /**
4708  * Escape html tags
4709  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4710  *
4711  * @param $in String: text that might contain HTML tags.
4712  * @return string Escaped string
4713  */
4714 function wfEscapeHTMLTagsOnly( $in ) {
4715         return str_replace(
4716                 array( '"', '>', '<' ),
4717                 array( '&quot;', '&gt;', '&lt;' ),
4718                 $in );
4719 }
4720
4721 ?>