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                 $this->stripNoGallery( $text );
 847                 $text = $this->doHeadings( $text );
 848                 if($this->mOptions->getUseDynamicDates()) {
 849                         $df =& DateFormatter::getInstance();
 850                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 851                 }
 852                 $text = $this->doAllQuotes( $text );
 853                 $text = $this->replaceInternalLinks( $text );
 854                 $text = $this->replaceExternalLinks( $text );
 855
 856                 # replaceInternalLinks may sometimes leave behind
 857                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 858                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 859
 860                 $text = $this->doMagicLinks( $text );
 861                 $text = $this->formatHeadings( $text, $isMain );
 862
 863                 wfProfileOut( $fname );
 864                 return $text;
 865         }
 866
 867         /**
 868          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 869          * magic external links.
 870          *
 871          * @private
 872          */
 873         function &doMagicLinks( &$text ) {
 874                 $text = $this->magicISBN( $text );
 875                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 876                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 877                 return $text;
 878         }
 879
 880         /**
 881          * Parse headers and return html
 882          *
 883          * @private
 884          */
 885         function doHeadings( $text ) {
 886                 $fname = 'Parser::doHeadings';
 887                 wfProfileIn( $fname );
 888                 for ( $i = 6; $i >= 1; --$i ) {
 889                         $h = str_repeat( '=', $i );
 890                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 891                           "<h{$i}>\\1</h{$i}>\\2", $text );
 892                 }
 893                 wfProfileOut( $fname );
 894                 return $text;
 895         }
 896
 897         /**
 898          * Replace single quotes with HTML markup
 899          * @private
 900          * @return string the altered text
 901          */
 902         function doAllQuotes( $text ) {
 903                 $fname = 'Parser::doAllQuotes';
 904                 wfProfileIn( $fname );
 905                 $outtext = '';
 906                 $lines = explode( "\n", $text );
 907                 foreach ( $lines as $line ) {
 908                         $outtext .= $this->doQuotes ( $line ) . "\n";
 909                 }
 910                 $outtext = substr($outtext, 0,-1);
 911                 wfProfileOut( $fname );
 912                 return $outtext;
 913         }
 914
 915         /**
 916          * Helper function for doAllQuotes()
 917          * @private
 918          */
 919         function doQuotes( $text ) {
 920                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 921                 if ( count( $arr ) == 1 )
 922                         return $text;
 923                 else
 924                 {
 925                         # First, do some preliminary work. This may shift some apostrophes from
 926                         # being mark-up to being text. It also counts the number of occurrences
 927                         # of bold and italics mark-ups.
 928                         $i = 0;
 929                         $numbold = 0;
 930                         $numitalics = 0;
 931                         foreach ( $arr as $r )
 932                         {
 933                                 if ( ( $i % 2 ) == 1 )
 934                                 {
 935                                         # If there are ever four apostrophes, assume the first is supposed to
 936                                         # be text, and the remaining three constitute mark-up for bold text.
 937                                         if ( strlen( $arr[$i] ) == 4 )
 938                                         {
 939                                                 $arr[$i-1] .= "'";
 940                                                 $arr[$i] = "'''";
 941                                         }
 942                                         # If there are more than 5 apostrophes in a row, assume they're all
 943                                         # text except for the last 5.
 944                                         else if ( strlen( $arr[$i] ) > 5 )
 945                                         {
 946                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 947                                                 $arr[$i] = "'''''";
 948                                         }
 949                                         # Count the number of occurrences of bold and italics mark-ups.
 950                                         # We are not counting sequences of five apostrophes.
 951                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 952                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 953                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 954                                 }
 955                                 $i++;
 956                         }
 957
 958                         # If there is an odd number of both bold and italics, it is likely
 959                         # that one of the bold ones was meant to be an apostrophe followed
 960                         # by italics. Which one we cannot know for certain, but it is more
 961                         # likely to be one that has a single-letter word before it.
 962                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 963                         {
 964                                 $i = 0;
 965                                 $firstsingleletterword = -1;
 966                                 $firstmultiletterword = -1;
 967                                 $firstspace = -1;
 968                                 foreach ( $arr as $r )
 969                                 {
 970                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 971                                         {
 972                                                 $x1 = substr ($arr[$i-1], -1);
 973                                                 $x2 = substr ($arr[$i-1], -2, 1);
 974                                                 if ($x1 == ' ') {
 975                                                         if ($firstspace == -1) $firstspace = $i;
 976                                                 } else if ($x2 == ' ') {
 977                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 978                                                 } else {
 979                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 980                                                 }
 981                                         }
 982                                         $i++;
 983                                 }
 984
 985                                 # If there is a single-letter word, use it!
 986                                 if ($firstsingleletterword > -1)
 987                                 {
 988                                         $arr [ $firstsingleletterword ] = "''";
 989                                         $arr [ $firstsingleletterword-1 ] .= "'";
 990                                 }
 991                                 # If not, but there's a multi-letter word, use that one.
 992                                 else if ($firstmultiletterword > -1)
 993                                 {
 994                                         $arr [ $firstmultiletterword ] = "''";
 995                                         $arr [ $firstmultiletterword-1 ] .= "'";
 996                                 }
 997                                 # ... otherwise use the first one that has neither.
 998                                 # (notice that it is possible for all three to be -1 if, for example,
 999                                 # there is only one pentuple-apostrophe in the line)
1000                                 else if ($firstspace > -1)
1001                                 {
1002                                         $arr [ $firstspace ] = "''";
1003                                         $arr [ $firstspace-1 ] .= "'";
1004                                 }
1005                         }
1006
1007                         # Now let's actually convert our apostrophic mush to HTML!
1008                         $output = '';
1009                         $buffer = '';
1010                         $state = '';
1011                         $i = 0;
1012                         foreach ($arr as $r)
1013                         {
1014                                 if (($i % 2) == 0)
1015                                 {
1016                                         if ($state == 'both')
1017                                                 $buffer .= $r;
1018                                         else
1019                                                 $output .= $r;
1020                                 }
1021                                 else
1022                                 {
1023                                         if (strlen ($r) == 2)
1024                                         {
1025                                                 if ($state == 'i')
1026                                                 { $output .= '</i>'; $state = ''; }
1027                                                 else if ($state == 'bi')
1028                                                 { $output .= '</i>'; $state = 'b'; }
1029                                                 else if ($state == 'ib')
1030                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1031                                                 else if ($state == 'both')
1032                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1033                                                 else # $state can be 'b' or ''
1034                                                 { $output .= '<i>'; $state .= 'i'; }
1035                                         }
1036                                         else if (strlen ($r) == 3)
1037                                         {
1038                                                 if ($state == 'b')
1039                                                 { $output .= '</b>'; $state = ''; }
1040                                                 else if ($state == 'bi')
1041                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1042                                                 else if ($state == 'ib')
1043                                                 { $output .= '</b>'; $state = 'i'; }
1044                                                 else if ($state == 'both')
1045                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1046                                                 else # $state can be 'i' or ''
1047                                                 { $output .= '<b>'; $state .= 'b'; }
1048                                         }
1049                                         else if (strlen ($r) == 5)
1050                                         {
1051                                                 if ($state == 'b')
1052                                                 { $output .= '</b><i>'; $state = 'i'; }
1053                                                 else if ($state == 'i')
1054                                                 { $output .= '</i><b>'; $state = 'b'; }
1055                                                 else if ($state == 'bi')
1056                                                 { $output .= '</i></b>'; $state = ''; }
1057                                                 else if ($state == 'ib')
1058                                                 { $output .= '</b></i>'; $state = ''; }
1059                                                 else if ($state == 'both')
1060                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1061                                                 else # ($state == '')
1062                                                 { $buffer = ''; $state = 'both'; }
1063                                         }
1064                                 }
1065                                 $i++;
1066                         }
1067                         # Now close all remaining tags.  Notice that the order is important.
1068                         if ($state == 'b' || $state == 'ib')
1069                                 $output .= '</b>';
1070                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1071                                 $output .= '</i>';
1072                         if ($state == 'bi')
1073                                 $output .= '</b>';
1074                         if ($state == 'both')
1075                                 $output .= '<b><i>'.$buffer.'</i></b>';
1076                         return $output;
1077                 }
1078         }
1079
1080         /**
1081          * Replace external links
1082          *
1083          * Note: this is all very hackish and the order of execution matters a lot.
1084          * Make sure to run maintenance/parserTests.php if you change this code.
1085          *
1086          * @private
1087          */
1088         function replaceExternalLinks( $text ) {
1089                 global $wgContLang;
1090                 $fname = 'Parser::replaceExternalLinks';
1091                 wfProfileIn( $fname );
1092
1093                 $sk =& $this->mOptions->getSkin();
1094
1095                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1096
1097                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1098
1099                 $i = 0;
1100                 while ( $i<count( $bits ) ) {
1101                         $url = $bits[$i++];
1102                         $protocol = $bits[$i++];
1103                         $text = $bits[$i++];
1104                         $trail = $bits[$i++];
1105
1106                         # The characters '<' and '>' (which were escaped by
1107                         # removeHTMLtags()) should not be included in
1108                         # URLs, per RFC 2396.
1109                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1110                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1111                                 $url = substr($url, 0, $m2[0][1]);
1112                         }
1113
1114                         # If the link text is an image URL, replace it with an <img> tag
1115                         # This happened by accident in the original parser, but some people used it extensively
1116                         $img = $this->maybeMakeExternalImage( $text );
1117                         if ( $img !== false ) {
1118                                 $text = $img;
1119                         }
1120
1121                         $dtrail = '';
1122
1123                         # Set linktype for CSS - if URL==text, link is essentially free
1124                         $linktype = ($text == $url) ? 'free' : 'text';
1125
1126                         # No link text, e.g. [http://domain.tld/some.link]
1127                         if ( $text == '' ) {
1128                                 # Autonumber if allowed. See bug #5918
1129                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1130                                         $text = '[' . ++$this->mAutonumber . ']';
1131                                         $linktype = 'autonumber';
1132                                 } else {
1133                                         # Otherwise just use the URL
1134                                         $text = htmlspecialchars( $url );
1135                                         $linktype = 'free';
1136                                 }
1137                         } else {
1138                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1139                                 # Check for trail
1140                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1141                         }
1142
1143                         $text = $wgContLang->markNoConversion($text);
1144
1145                         # Normalize any HTML entities in input. They will be
1146                         # re-escaped by makeExternalLink().
1147                         $url = Sanitizer::decodeCharReferences( $url );
1148
1149                         # Process the trail (i.e. everything after this link up until start of the next link),
1150                         # replacing any non-bracketed links
1151                         $trail = $this->replaceFreeExternalLinks( $trail );
1152
1153                         # Use the encoded URL
1154                         # This means that users can paste URLs directly into the text
1155                         # Funny characters like &ouml; aren't valid in URLs anyway
1156                         # This was changed in August 2004
1157                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1158
1159                         # Register link in the output object.
1160                         # Replace unnecessary URL escape codes with the referenced character
1161                         # This prevents spammers from hiding links from the filters
1162                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1163                         $this->mOutput->addExternalLink( $pasteurized );
1164                 }
1165
1166                 wfProfileOut( $fname );
1167                 return $s;
1168         }
1169
1170         /**
1171          * Replace anything that looks like a URL with a link
1172          * @private
1173          */
1174         function replaceFreeExternalLinks( $text ) {
1175                 global $wgContLang;
1176                 $fname = 'Parser::replaceFreeExternalLinks';
1177                 wfProfileIn( $fname );
1178
1179                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1180                 $s = array_shift( $bits );
1181                 $i = 0;
1182
1183                 $sk =& $this->mOptions->getSkin();
1184
1185                 while ( $i < count( $bits ) ){
1186                         $protocol = $bits[$i++];
1187                         $remainder = $bits[$i++];
1188
1189                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1190                                 # Found some characters after the protocol that look promising
1191                                 $url = $protocol . $m[1];
1192                                 $trail = $m[2];
1193
1194                                 # special case: handle urls as url args:
1195                                 # http://www.example.com/foo?=http://www.example.com/bar
1196                                 if(strlen($trail) == 0 &&
1197                                         isset($bits[$i]) &&
1198                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1199                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1200                                 {
1201                                         # add protocol, arg
1202                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1203                                         $i += 2;
1204                                         $trail = $m[2];
1205                                 }
1206
1207                                 # The characters '<' and '>' (which were escaped by
1208                                 # removeHTMLtags()) should not be included in
1209                                 # URLs, per RFC 2396.
1210                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1211                                         $trail = substr($url, $m2[0][1]) . $trail;
1212                                         $url = substr($url, 0, $m2[0][1]);
1213                                 }
1214
1215                                 # Move trailing punctuation to $trail
1216                                 $sep = ',;\.:!?';
1217                                 # If there is no left bracket, then consider right brackets fair game too
1218                                 if ( strpos( $url, '(' ) === false ) {
1219                                         $sep .= ')';
1220                                 }
1221
1222                                 $numSepChars = strspn( strrev( $url ), $sep );
1223                                 if ( $numSepChars ) {
1224                                         $trail = substr( $url, -$numSepChars ) . $trail;
1225                                         $url = substr( $url, 0, -$numSepChars );
1226                                 }
1227
1228                                 # Normalize any HTML entities in input. They will be
1229                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1230                                 $url = Sanitizer::decodeCharReferences( $url );
1231
1232                                 # Is this an external image?
1233                                 $text = $this->maybeMakeExternalImage( $url );
1234                                 if ( $text === false ) {
1235                                         # Not an image, make a link
1236                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1237                                         # Register it in the output object...
1238                                         # Replace unnecessary URL escape codes with their equivalent characters
1239                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1240                                         $this->mOutput->addExternalLink( $pasteurized );
1241                                 }
1242                                 $s .= $text . $trail;
1243                         } else {
1244                                 $s .= $protocol . $remainder;
1245                         }
1246                 }
1247                 wfProfileOut( $fname );
1248                 return $s;
1249         }
1250
1251         /**
1252          * Replace unusual URL escape codes with their equivalent characters
1253          * @param string
1254          * @return string
1255          * @static
1256          * @fixme This can merge genuinely required bits in the path or query string,
1257          *        breaking legit URLs. A proper fix would treat the various parts of
1258          *        the URL differently; as a workaround, just use the output for
1259          *        statistical records, not for actual linking/output.
1260          */
1261         function replaceUnusualEscapes( $url ) {
1262                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1263                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1264         }
1265
1266         /**
1267          * Callback function used in replaceUnusualEscapes().
1268          * Replaces unusual URL escape codes with their equivalent character
1269          * @static
1270          * @private
1271          */
1272         function replaceUnusualEscapesCallback( $matches ) {
1273                 $char = urldecode( $matches[0] );
1274                 $ord = ord( $char );
1275                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1276                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1277                         // No, shouldn't be escaped
1278                         return $char;
1279                 } else {
1280                         // Yes, leave it escaped
1281                         return $matches[0];
1282                 }
1283         }
1284
1285         /**
1286          * make an image if it's allowed, either through the global
1287          * option or through the exception
1288          * @private
1289          */
1290         function maybeMakeExternalImage( $url ) {
1291                 $sk =& $this->mOptions->getSkin();
1292                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1293                 $imagesexception = !empty($imagesfrom);
1294                 $text = false;
1295                 if ( $this->mOptions->getAllowExternalImages()
1296                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1297                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1298                                 # Image found
1299                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1300                         }
1301                 }
1302                 return $text;
1303         }
1304
1305         /**
1306          * Process [[ ]] wikilinks
1307          *
1308          * @private
1309          */
1310         function replaceInternalLinks( $s ) {
1311                 global $wgContLang;
1312                 static $fname = 'Parser::replaceInternalLinks' ;
1313
1314                 wfProfileIn( $fname );
1315
1316                 wfProfileIn( $fname.'-setup' );
1317                 static $tc = FALSE;
1318                 # the % is needed to support urlencoded titles as well
1319                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1320
1321                 $sk =& $this->mOptions->getSkin();
1322
1323                 #split the entire text string on occurences of [[
1324                 $a = explode( '[[', ' ' . $s );
1325                 #get the first element (all text up to first [[), and remove the space we added
1326                 $s = array_shift( $a );
1327                 $s = substr( $s, 1 );
1328
1329                 # Match a link having the form [[namespace:link|alternate]]trail
1330                 static $e1 = FALSE;
1331                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1332                 # Match cases where there is no "]]", which might still be images
1333                 static $e1_img = FALSE;
1334                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1335                 # Match the end of a line for a word that's not followed by whitespace,
1336                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1337                 $e2 = wfMsgForContent( 'linkprefix' );
1338
1339                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1340
1341                 if( is_null( $this->mTitle ) ) {
1342                         throw new MWException( 'nooo' );
1343                 }
1344                 $nottalk = !$this->mTitle->isTalkPage();
1345
1346                 if ( $useLinkPrefixExtension ) {
1347                         if ( preg_match( $e2, $s, $m ) ) {
1348                                 $first_prefix = $m[2];
1349                         } else {
1350                                 $first_prefix = false;
1351                         }
1352                 } else {
1353                         $prefix = '';
1354                 }
1355
1356                 $selflink = $this->mTitle->getPrefixedText();
1357                 wfProfileOut( $fname.'-setup' );
1358
1359                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1360                 $useSubpages = $this->areSubpagesAllowed();
1361
1362                 # Loop for each link
1363                 for ($k = 0; isset( $a[$k] ); $k++) {
1364                         $line = $a[$k];
1365                         if ( $useLinkPrefixExtension ) {
1366                                 wfProfileIn( $fname.'-prefixhandling' );
1367                                 if ( preg_match( $e2, $s, $m ) ) {
1368                                         $prefix = $m[2];
1369                                         $s = $m[1];
1370                                 } else {
1371                                         $prefix='';
1372                                 }
1373                                 # first link
1374                                 if($first_prefix) {
1375                                         $prefix = $first_prefix;
1376                                         $first_prefix = false;
1377                                 }
1378                                 wfProfileOut( $fname.'-prefixhandling' );
1379                         }
1380
1381                         $might_be_img = false;
1382
1383                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1384                                 $text = $m[2];
1385                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1386                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1387                                 # the real problem is with the $e1 regex
1388                                 # See bug 1300.
1389                                 #
1390                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1391                                 # and no image is in sight. See bug 2095.
1392                                 #
1393                                 if( $text !== '' &&
1394                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1395                                         strpos($text, '[') !== false
1396                                 )
1397                                 {
1398                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1399                                         $m[3] = $n[1];
1400                                 }
1401                                 # fix up urlencoded title texts
1402                                 if(preg_match('/%/', $m[1] ))
1403                                         # Should anchors '#' also be rejected?
1404                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1405                                 $trail = $m[3];
1406                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1407                                 $might_be_img = true;
1408                                 $text = $m[2];
1409                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1410                                 $trail = "";
1411                         } else { # Invalid form; output directly
1412                                 $s .= $prefix . '[[' . $line ;
1413                                 continue;
1414                         }
1415
1416                         # Don't allow internal links to pages containing
1417                         # PROTO: where PROTO is a valid URL protocol; these
1418                         # should be external links.
1419                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1420                                 $s .= $prefix . '[[' . $line ;
1421                                 continue;
1422                         }
1423
1424                         # Make subpage if necessary
1425                         if( $useSubpages ) {
1426                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1427                         } else {
1428                                 $link = $m[1];
1429                         }
1430
1431                         $noforce = (substr($m[1], 0, 1) != ':');
1432                         if (!$noforce) {
1433                                 # Strip off leading ':'
1434                                 $link = substr($link, 1);
1435                         }
1436
1437                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1438                         if( !$nt ) {
1439                                 $s .= $prefix . '[[' . $line;
1440                                 continue;
1441                         }
1442
1443                         #check other language variants of the link
1444                         #if the article does not exist
1445                         if( $checkVariantLink
1446                             && $nt->getArticleID() == 0 ) {
1447                                 $wgContLang->findVariantLink($link, $nt);
1448                         }
1449
1450                         $ns = $nt->getNamespace();
1451                         $iw = $nt->getInterWiki();
1452
1453                         if ($might_be_img) { # if this is actually an invalid link
1454                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1455                                         $found = false;
1456                                         while (isset ($a[$k+1]) ) {
1457                                                 #look at the next 'line' to see if we can close it there
1458                                                 $spliced = array_splice( $a, $k + 1, 1 );
1459                                                 $next_line = array_shift( $spliced );
1460                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1461                                                 # the first ]] closes the inner link, the second the image
1462                                                         $found = true;
1463                                                         $text .= '[[' . $m[1];
1464                                                         $trail = $m[2];
1465                                                         break;
1466                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1467                                                         #if there's exactly one ]] that's fine, we'll keep looking
1468                                                         $text .= '[[' . $m[0];
1469                                                 } else {
1470                                                         #if $next_line is invalid too, we need look no further
1471                                                         $text .= '[[' . $next_line;
1472                                                         break;
1473                                                 }
1474                                         }
1475                                         if ( !$found ) {
1476                                                 # we couldn't find the end of this imageLink, so output it raw
1477                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1478                                                 $text = $this->replaceInternalLinks($text);
1479                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1480                                                 # note: no $trail, because without an end, there *is* no trail
1481                                                 continue;
1482                                         }
1483                                 } else { #it's not an image, so output it raw
1484                                         $s .= $prefix . '[[' . $link . '|' . $text;
1485                                         # note: no $trail, because without an end, there *is* no trail
1486                                         continue;
1487                                 }
1488                         }
1489
1490                         $wasblank = ( '' == $text );
1491                         if( $wasblank ) $text = $link;
1492
1493
1494                         # Link not escaped by : , create the various objects
1495                         if( $noforce ) {
1496
1497                                 # Interwikis
1498                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1499                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1500                                         $s = rtrim($s . "\n");
1501                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1502                                         continue;
1503                                 }
1504
1505                                 if ( $ns == NS_IMAGE ) {
1506                                         wfProfileIn( "$fname-image" );
1507                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1508                                                 # recursively parse links inside the image caption
1509                                                 # actually, this will parse them in any other parameters, too,
1510                                                 # but it might be hard to fix that, and it doesn't matter ATM
1511                                                 $text = $this->replaceExternalLinks($text);
1512                                                 $text = $this->replaceInternalLinks($text);
1513
1514                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1515                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1516                                                 $this->mOutput->addImage( $nt->getDBkey() );
1517
1518                                                 wfProfileOut( "$fname-image" );
1519                                                 continue;
1520                                         }
1521                                         wfProfileOut( "$fname-image" );
1522
1523                                 }
1524
1525                                 if ( $ns == NS_CATEGORY ) {
1526                                         wfProfileIn( "$fname-category" );
1527                                         $s = rtrim($s . "\n"); # bug 87
1528
1529                                         if ( $wasblank ) {
1530                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1531                                                         $sortkey = $this->mTitle->getText();
1532                                                 } else {
1533                                                         $sortkey = $this->mTitle->getPrefixedText();
1534                                                 }
1535                                         } else {
1536                                                 $sortkey = $text;
1537                                         }
1538                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1539                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1540                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1541
1542                                         /**
1543                                          * Strip the whitespace Category links produce, see bug 87
1544                                          * @todo We might want to use trim($tmp, "\n") here.
1545                                          */
1546                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1547
1548                                         wfProfileOut( "$fname-category" );
1549                                         continue;
1550                                 }
1551                         }
1552
1553                         if( ( $nt->getPrefixedText() === $selflink ) &&
1554                             ( $nt->getFragment() === '' ) ) {
1555                                 # Self-links are handled specially; generally de-link and change to bold.
1556                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1557                                 continue;
1558                         }
1559
1560                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1561                         if( $ns == NS_MEDIA ) {
1562                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1563                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1564                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1565                                 $this->mOutput->addImage( $nt->getDBkey() );
1566                                 continue;
1567                         } elseif( $ns == NS_SPECIAL ) {
1568                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1569                                 continue;
1570                         } elseif( $ns == NS_IMAGE ) {
1571                                 $img = Image::newFromTitle( $nt );
1572                                 if( $img->exists() ) {
1573                                         // Force a blue link if the file exists; may be a remote
1574                                         // upload on the shared repository, and we want to see its
1575                                         // auto-generated page.
1576                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1577                                         continue;
1578                                 }
1579                         }
1580                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1581                 }
1582                 wfProfileOut( $fname );
1583                 return $s;
1584         }
1585
1586         /**
1587          * Make a link placeholder. The text returned can be later resolved to a real link with
1588          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1589          * parsing of interwiki links, and secondly to allow all extistence checks and
1590          * article length checks (for stub links) to be bundled into a single query.
1591          *
1592          */
1593         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1594                 if ( ! is_object($nt) ) {
1595                         # Fail gracefully
1596                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1597                 } else {
1598                         # Separate the link trail from the rest of the link
1599                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1600
1601                         if ( $nt->isExternal() ) {
1602                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1603                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1604                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1605                         } else {
1606                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1607                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1608                                 $this->mLinkHolders['queries'][] = $query;
1609                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1610                                 $this->mLinkHolders['titles'][] = $nt;
1611
1612                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1613                         }
1614                 }
1615                 return $retVal;
1616         }
1617
1618         /**
1619          * Render a forced-blue link inline; protect against double expansion of
1620          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1621          * Since this little disaster has to split off the trail text to avoid
1622          * breaking URLs in the following text without breaking trails on the
1623          * wiki links, it's been made into a horrible function.
1624          *
1625          * @param Title $nt
1626          * @param string $text
1627          * @param string $query
1628          * @param string $trail
1629          * @param string $prefix
1630          * @return string HTML-wikitext mix oh yuck
1631          */
1632         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1633                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1634                 $sk =& $this->mOptions->getSkin();
1635                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1636                 return $this->armorLinks( $link ) . $trail;
1637         }
1638
1639         /**
1640          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1641          * going to go through further parsing steps before inline URL expansion.
1642          *
1643          * In particular this is important when using action=render, which causes
1644          * full URLs to be included.
1645          *
1646          * Oh man I hate our multi-layer parser!
1647          *
1648          * @param string more-or-less HTML
1649          * @return string less-or-more HTML with NOPARSE bits
1650          */
1651         function armorLinks( $text ) {
1652                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1653                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1654         }
1655
1656         /**
1657          * Return true if subpage links should be expanded on this page.
1658          * @return bool
1659          */
1660         function areSubpagesAllowed() {
1661                 # Some namespaces don't allow subpages
1662                 global $wgNamespacesWithSubpages;
1663                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1664         }
1665
1666         /**
1667          * Handle link to subpage if necessary
1668          * @param string $target the source of the link
1669          * @param string &$text the link text, modified as necessary
1670          * @return string the full name of the link
1671          * @private
1672          */
1673         function maybeDoSubpageLink($target, &$text) {
1674                 # Valid link forms:
1675                 # Foobar -- normal
1676                 # :Foobar -- override special treatment of prefix (images, language links)
1677                 # /Foobar -- convert to CurrentPage/Foobar
1678                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1679                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1680                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1681
1682                 $fname = 'Parser::maybeDoSubpageLink';
1683                 wfProfileIn( $fname );
1684                 $ret = $target; # default return value is no change
1685
1686                 # Some namespaces don't allow subpages,
1687                 # so only perform processing if subpages are allowed
1688                 if( $this->areSubpagesAllowed() ) {
1689                         # Look at the first character
1690                         if( $target != '' && $target{0} == '/' ) {
1691                                 # / at end means we don't want the slash to be shown
1692                                 if( substr( $target, -1, 1 ) == '/' ) {
1693                                         $target = substr( $target, 1, -1 );
1694                                         $noslash = $target;
1695                                 } else {
1696                                         $noslash = substr( $target, 1 );
1697                                 }
1698
1699                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1700                                 if( '' === $text ) {
1701                                         $text = $target;
1702                                 } # this might be changed for ugliness reasons
1703                         } else {
1704                                 # check for .. subpage backlinks
1705                                 $dotdotcount = 0;
1706                                 $nodotdot = $target;
1707                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1708                                         ++$dotdotcount;
1709                                         $nodotdot = substr( $nodotdot, 3 );
1710                                 }
1711                                 if($dotdotcount > 0) {
1712                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1713                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1714                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1715                                                 # / at the end means don't show full path
1716                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1717                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1718                                                         if( '' === $text ) {
1719                                                                 $text = $nodotdot;
1720                                                         }
1721                                                 }
1722                                                 $nodotdot = trim( $nodotdot );
1723                                                 if( $nodotdot != '' ) {
1724                                                         $ret .= '/' . $nodotdot;
1725                                                 }
1726                                         }
1727                                 }
1728                         }
1729                 }
1730
1731                 wfProfileOut( $fname );
1732                 return $ret;
1733         }
1734
1735         /**#@+
1736          * Used by doBlockLevels()
1737          * @private
1738          */
1739         /* private */ function closeParagraph() {
1740                 $result = '';
1741                 if ( '' != $this->mLastSection ) {
1742                         $result = '</' . $this->mLastSection  . ">\n";
1743                 }
1744                 $this->mInPre = false;
1745                 $this->mLastSection = '';
1746                 return $result;
1747         }
1748         # getCommon() returns the length of the longest common substring
1749         # of both arguments, starting at the beginning of both.
1750         #
1751         /* private */ function getCommon( $st1, $st2 ) {
1752                 $fl = strlen( $st1 );
1753                 $shorter = strlen( $st2 );
1754                 if ( $fl < $shorter ) { $shorter = $fl; }
1755
1756                 for ( $i = 0; $i < $shorter; ++$i ) {
1757                         if ( $st1{$i} != $st2{$i} ) { break; }
1758                 }
1759                 return $i;
1760         }
1761         # These next three functions open, continue, and close the list
1762         # element appropriate to the prefix character passed into them.
1763         #
1764         /* private */ function openList( $char ) {
1765                 $result = $this->closeParagraph();
1766
1767                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1768                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1769                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1770                 else if ( ';' == $char ) {
1771                         $result .= '<dl><dt>';
1772                         $this->mDTopen = true;
1773                 }
1774                 else { $result = '<!-- ERR 1 -->'; }
1775
1776                 return $result;
1777         }
1778
1779         /* private */ function nextItem( $char ) {
1780                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1781                 else if ( ':' == $char || ';' == $char ) {
1782                         $close = '</dd>';
1783                         if ( $this->mDTopen ) { $close = '</dt>'; }
1784                         if ( ';' == $char ) {
1785                                 $this->mDTopen = true;
1786                                 return $close . '<dt>';
1787                         } else {
1788                                 $this->mDTopen = false;
1789                                 return $close . '<dd>';
1790                         }
1791                 }
1792                 return '<!-- ERR 2 -->';
1793         }
1794
1795         /* private */ function closeList( $char ) {
1796                 if ( '*' == $char ) { $text = '</li></ul>'; }
1797                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1798                 else if ( ':' == $char ) {
1799                         if ( $this->mDTopen ) {
1800                                 $this->mDTopen = false;
1801                                 $text = '</dt></dl>';
1802                         } else {
1803                                 $text = '</dd></dl>';
1804                         }
1805                 }
1806                 else {  return '<!-- ERR 3 -->'; }
1807                 return $text."\n";
1808         }
1809         /**#@-*/
1810
1811         /**
1812          * Make lists from lines starting with ':', '*', '#', etc.
1813          *
1814          * @private
1815          * @return string the lists rendered as HTML
1816          */
1817         function doBlockLevels( $text, $linestart ) {
1818                 $fname = 'Parser::doBlockLevels';
1819                 wfProfileIn( $fname );
1820
1821                 # Parsing through the text line by line.  The main thing
1822                 # happening here is handling of block-level elements p, pre,
1823                 # and making lists from lines starting with * # : etc.
1824                 #
1825                 $textLines = explode( "\n", $text );
1826
1827                 $lastPrefix = $output = '';
1828                 $this->mDTopen = $inBlockElem = false;
1829                 $prefixLength = 0;
1830                 $paragraphStack = false;
1831
1832                 if ( !$linestart ) {
1833                         $output .= array_shift( $textLines );
1834                 }
1835                 foreach ( $textLines as $oLine ) {
1836                         $lastPrefixLength = strlen( $lastPrefix );
1837                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1838                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1839                         if ( !$this->mInPre ) {
1840                                 # Multiple prefixes may abut each other for nested lists.
1841                                 $prefixLength = strspn( $oLine, '*#:;' );
1842                                 $pref = substr( $oLine, 0, $prefixLength );
1843
1844                                 # eh?
1845                                 $pref2 = str_replace( ';', ':', $pref );
1846                                 $t = substr( $oLine, $prefixLength );
1847                                 $this->mInPre = !empty($preOpenMatch);
1848                         } else {
1849                                 # Don't interpret any other prefixes in preformatted text
1850                                 $prefixLength = 0;
1851                                 $pref = $pref2 = '';
1852                                 $t = $oLine;
1853                         }
1854
1855                         # List generation
1856                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1857                                 # Same as the last item, so no need to deal with nesting or opening stuff
1858                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1859                                 $paragraphStack = false;
1860
1861                                 if ( substr( $pref, -1 ) == ';') {
1862                                         # The one nasty exception: definition lists work like this:
1863                                         # ; title : definition text
1864                                         # So we check for : in the remainder text to split up the
1865                                         # title and definition, without b0rking links.
1866                                         $term = $t2 = '';
1867                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1868                                                 $t = $t2;
1869                                                 $output .= $term . $this->nextItem( ':' );
1870                                         }
1871                                 }
1872                         } elseif( $prefixLength || $lastPrefixLength ) {
1873                                 # Either open or close a level...
1874                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1875                                 $paragraphStack = false;
1876
1877                                 while( $commonPrefixLength < $lastPrefixLength ) {
1878                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1879                                         --$lastPrefixLength;
1880                                 }
1881                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1882                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1883                                 }
1884                                 while ( $prefixLength > $commonPrefixLength ) {
1885                                         $char = substr( $pref, $commonPrefixLength, 1 );
1886                                         $output .= $this->openList( $char );
1887
1888                                         if ( ';' == $char ) {
1889                                                 # FIXME: This is dupe of code above
1890                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1891                                                         $t = $t2;
1892                                                         $output .= $term . $this->nextItem( ':' );
1893                                                 }
1894                                         }
1895                                         ++$commonPrefixLength;
1896                                 }
1897                                 $lastPrefix = $pref2;
1898                         }
1899                         if( 0 == $prefixLength ) {
1900                                 wfProfileIn( "$fname-paragraph" );
1901                                 # No prefix (not in list)--go to paragraph mode
1902                                 // XXX: use a stack for nestable elements like span, table and div
1903                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1904                                 $closematch = preg_match(
1905                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1906                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1907                                 if ( $openmatch or $closematch ) {
1908                                         $paragraphStack = false;
1909                                         # TODO bug 5718: paragraph closed
1910                                         $output .= $this->closeParagraph();
1911                                         if ( $preOpenMatch and !$preCloseMatch ) {
1912                                                 $this->mInPre = true;
1913                                         }
1914                                         if ( $closematch ) {
1915                                                 $inBlockElem = false;
1916                                         } else {
1917                                                 $inBlockElem = true;
1918                                         }
1919                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1920                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1921                                                 // pre
1922                                                 if ($this->mLastSection != 'pre') {
1923                                                         $paragraphStack = false;
1924                                                         $output .= $this->closeParagraph().'<pre>';
1925                                                         $this->mLastSection = 'pre';
1926                                                 }
1927                                                 $t = substr( $t, 1 );
1928                                         } else {
1929                                                 // paragraph
1930                                                 if ( '' == trim($t) ) {
1931                                                         if ( $paragraphStack ) {
1932                                                                 $output .= $paragraphStack.'<br />';
1933                                                                 $paragraphStack = false;
1934                                                                 $this->mLastSection = 'p';
1935                                                         } else {
1936                                                                 if ($this->mLastSection != 'p' ) {
1937                                                                         $output .= $this->closeParagraph();
1938                                                                         $this->mLastSection = '';
1939                                                                         $paragraphStack = '<p>';
1940                                                                 } else {
1941                                                                         $paragraphStack = '</p><p>';
1942                                                                 }
1943                                                         }
1944                                                 } else {
1945                                                         if ( $paragraphStack ) {
1946                                                                 $output .= $paragraphStack;
1947                                                                 $paragraphStack = false;
1948                                                                 $this->mLastSection = 'p';
1949                                                         } else if ($this->mLastSection != 'p') {
1950                                                                 $output .= $this->closeParagraph().'<p>';
1951                                                                 $this->mLastSection = 'p';
1952                                                         }
1953                                                 }
1954                                         }
1955                                 }
1956                                 wfProfileOut( "$fname-paragraph" );
1957                         }
1958                         // somewhere above we forget to get out of pre block (bug 785)
1959                         if($preCloseMatch && $this->mInPre) {
1960                                 $this->mInPre = false;
1961                         }
1962                         if ($paragraphStack === false) {
1963                                 $output .= $t."\n";
1964                         }
1965                 }
1966                 while ( $prefixLength ) {
1967                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1968                         --$prefixLength;
1969                 }
1970                 if ( '' != $this->mLastSection ) {
1971                         $output .= '</' . $this->mLastSection . '>';
1972                         $this->mLastSection = '';
1973                 }
1974
1975                 wfProfileOut( $fname );
1976                 return $output;
1977         }
1978
1979         /**
1980          * Split up a string on ':', ignoring any occurences inside tags
1981          * to prevent illegal overlapping.
1982          * @param string $str the string to split
1983          * @param string &$before set to everything before the ':'
1984          * @param string &$after set to everything after the ':'
1985          * return string the position of the ':', or false if none found
1986          */
1987         function findColonNoLinks($str, &$before, &$after) {
1988                 $fname = 'Parser::findColonNoLinks';
1989                 wfProfileIn( $fname );
1990
1991                 $pos = strpos( $str, ':' );
1992                 if( $pos === false ) {
1993                         // Nothing to find!
1994                         wfProfileOut( $fname );
1995                         return false;
1996                 }
1997
1998                 $lt = strpos( $str, '<' );
1999                 if( $lt === false || $lt > $pos ) {
2000                         // Easy; no tag nesting to worry about
2001                         $before = substr( $str, 0, $pos );
2002                         $after = substr( $str, $pos+1 );
2003                         wfProfileOut( $fname );
2004                         return $pos;
2005                 }
2006
2007                 // Ugly state machine to walk through avoiding tags.
2008                 $state = MW_COLON_STATE_TEXT;
2009                 $stack = 0;
2010                 $len = strlen( $str );
2011                 for( $i = 0; $i < $len; $i++ ) {
2012                         $c = $str{$i};
2013
2014                         switch( $state ) {
2015                         // (Using the number is a performance hack for common cases)
2016                         case 0: // MW_COLON_STATE_TEXT:
2017                                 switch( $c ) {
2018                                 case "<":
2019                                         // Could be either a <start> tag or an </end> tag
2020                                         $state = MW_COLON_STATE_TAGSTART;
2021                                         break;
2022                                 case ":":
2023                                         if( $stack == 0 ) {
2024                                                 // We found it!
2025                                                 $before = substr( $str, 0, $i );
2026                                                 $after = substr( $str, $i + 1 );
2027                                                 wfProfileOut( $fname );
2028                                                 return $i;
2029                                         }
2030                                         // Embedded in a tag; don't break it.
2031                                         break;
2032                                 default:
2033                                         // Skip ahead looking for something interesting
2034                                         $colon = strpos( $str, ':', $i );
2035                                         if( $colon === false ) {
2036                                                 // Nothing else interesting
2037                                                 wfProfileOut( $fname );
2038                                                 return false;
2039                                         }
2040                                         $lt = strpos( $str, '<', $i );
2041                                         if( $stack === 0 ) {
2042                                                 if( $lt === false || $colon < $lt ) {
2043                                                         // We found it!
2044                                                         $before = substr( $str, 0, $colon );
2045                                                         $after = substr( $str, $colon + 1 );
2046                                                         wfProfileOut( $fname );
2047                                                         return $i;
2048                                                 }
2049                                         }
2050                                         if( $lt === false ) {
2051                                                 // Nothing else interesting to find; abort!
2052                                                 // We're nested, but there's no close tags left. Abort!
2053                                                 break 2;
2054                                         }
2055                                         // Skip ahead to next tag start
2056                                         $i = $lt;
2057                                         $state = MW_COLON_STATE_TAGSTART;
2058                                 }
2059                                 break;
2060                         case 1: // MW_COLON_STATE_TAG:
2061                                 // In a <tag>
2062                                 switch( $c ) {
2063                                 case ">":
2064                                         $stack++;
2065                                         $state = MW_COLON_STATE_TEXT;
2066                                         break;
2067                                 case "/":
2068                                         // Slash may be followed by >?
2069                                         $state = MW_COLON_STATE_TAGSLASH;
2070                                         break;
2071                                 default:
2072                                         // ignore
2073                                 }
2074                                 break;
2075                         case 2: // MW_COLON_STATE_TAGSTART:
2076                                 switch( $c ) {
2077                                 case "/":
2078                                         $state = MW_COLON_STATE_CLOSETAG;
2079                                         break;
2080                                 case "!":
2081                                         $state = MW_COLON_STATE_COMMENT;
2082                                         break;
2083                                 case ">":
2084                                         // Illegal early close? This shouldn't happen D:
2085                                         $state = MW_COLON_STATE_TEXT;
2086                                         break;
2087                                 default:
2088                                         $state = MW_COLON_STATE_TAG;
2089                                 }
2090                                 break;
2091                         case 3: // MW_COLON_STATE_CLOSETAG:
2092                                 // In a </tag>
2093                                 if( $c == ">" ) {
2094                                         $stack--;
2095                                         if( $stack < 0 ) {
2096                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2097                                                 wfProfileOut( $fname );
2098                                                 return false;
2099                                         }
2100                                         $state = MW_COLON_STATE_TEXT;
2101                                 }
2102                                 break;
2103                         case MW_COLON_STATE_TAGSLASH:
2104                                 if( $c == ">" ) {
2105                                         // Yes, a self-closed tag <blah/>
2106                                         $state = MW_COLON_STATE_TEXT;
2107                                 } else {
2108                                         // Probably we're jumping the gun, and this is an attribute
2109                                         $state = MW_COLON_STATE_TAG;
2110                                 }
2111                                 break;
2112                         case 5: // MW_COLON_STATE_COMMENT:
2113                                 if( $c == "-" ) {
2114                                         $state = MW_COLON_STATE_COMMENTDASH;
2115                                 }
2116                                 break;
2117                         case MW_COLON_STATE_COMMENTDASH:
2118                                 if( $c == "-" ) {
2119                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2120                                 } else {
2121                                         $state = MW_COLON_STATE_COMMENT;
2122                                 }
2123                                 break;
2124                         case MW_COLON_STATE_COMMENTDASHDASH:
2125                                 if( $c == ">" ) {
2126                                         $state = MW_COLON_STATE_TEXT;
2127                                 } else {
2128                                         $state = MW_COLON_STATE_COMMENT;
2129                                 }
2130                                 break;
2131                         default:
2132                                 throw new MWException( "State machine error in $fname" );
2133                         }
2134                 }
2135                 if( $stack > 0 ) {
2136                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2137                         return false;
2138                 }
2139                 wfProfileOut( $fname );
2140                 return false;
2141         }
2142
2143         /**
2144          * Return value of a magic variable (like PAGENAME)
2145          *
2146          * @private
2147          */
2148         function getVariableValue( $index ) {
2149                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2150
2151                 /**
2152                  * Some of these require message or data lookups and can be
2153                  * expensive to check many times.
2154                  */
2155                 static $varCache = array();
2156                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2157                         if ( isset( $varCache[$index] ) )
2158                                 return $varCache[$index];
2159
2160                 $ts = time();
2161                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2162
2163                 switch ( $index ) {
2164                         case MAG_CURRENTMONTH:
2165                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2166                         case MAG_CURRENTMONTHNAME:
2167                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2168                         case MAG_CURRENTMONTHNAMEGEN:
2169                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2170                         case MAG_CURRENTMONTHABBREV:
2171                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2172                         case MAG_CURRENTDAY:
2173                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2174                         case MAG_CURRENTDAY2:
2175                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2176                         case MAG_PAGENAME:
2177                                 return $this->mTitle->getText();
2178                         case MAG_PAGENAMEE:
2179                                 return $this->mTitle->getPartialURL();
2180                         case MAG_FULLPAGENAME:
2181                                 return $this->mTitle->getPrefixedText();
2182                         case MAG_FULLPAGENAMEE:
2183                                 return $this->mTitle->getPrefixedURL();
2184                         case MAG_SUBPAGENAME:
2185                                 return $this->mTitle->getSubpageText();
2186                         case MAG_SUBPAGENAMEE:
2187                                 return $this->mTitle->getSubpageUrlForm();
2188                         case MAG_BASEPAGENAME:
2189                                 return $this->mTitle->getBaseText();
2190                         case MAG_BASEPAGENAMEE:
2191                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2192                         case MAG_TALKPAGENAME:
2193                                 if( $this->mTitle->canTalk() ) {
2194                                         $talkPage = $this->mTitle->getTalkPage();
2195                                         return $talkPage->getPrefixedText();
2196                                 } else {
2197                                         return '';
2198                                 }
2199                         case MAG_TALKPAGENAMEE:
2200                                 if( $this->mTitle->canTalk() ) {
2201                                         $talkPage = $this->mTitle->getTalkPage();
2202                                         return $talkPage->getPrefixedUrl();
2203                                 } else {
2204                                         return '';
2205                                 }
2206                         case MAG_SUBJECTPAGENAME:
2207                                 $subjPage = $this->mTitle->getSubjectPage();
2208                                 return $subjPage->getPrefixedText();
2209                         case MAG_SUBJECTPAGENAMEE:
2210                                 $subjPage = $this->mTitle->getSubjectPage();
2211                                 return $subjPage->getPrefixedUrl();
2212                         case MAG_REVISIONID:
2213                                 return $this->mRevisionId;
2214                         case MAG_NAMESPACE:
2215                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2216                         case MAG_NAMESPACEE:
2217                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2218                         case MAG_TALKSPACE:
2219                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2220                         case MAG_TALKSPACEE:
2221                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2222                         case MAG_SUBJECTSPACE:
2223                                 return $this->mTitle->getSubjectNsText();
2224                         case MAG_SUBJECTSPACEE:
2225                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2226                         case MAG_CURRENTDAYNAME:
2227                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2228                         case MAG_CURRENTYEAR:
2229                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2230                         case MAG_CURRENTTIME:
2231                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2232                         case MAG_CURRENTWEEK:
2233                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2234                                 // int to remove the padding
2235                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2236                         case MAG_CURRENTDOW:
2237                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2238                         case MAG_NUMBEROFARTICLES:
2239                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2240                         case MAG_NUMBEROFFILES:
2241                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2242                         case MAG_NUMBEROFUSERS:
2243                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2244                         case MAG_NUMBEROFPAGES:
2245                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2246                         case MAG_CURRENTTIMESTAMP:
2247                                 return $varCache[$index] = wfTimestampNow();
2248                         case MAG_CURRENTVERSION:
2249                                 global $wgVersion;
2250                                 return $wgVersion;
2251                         case MAG_SITENAME:
2252                                 return $wgSitename;
2253                         case MAG_SERVER:
2254                                 return $wgServer;
2255                         case MAG_SERVERNAME:
2256                                 return $wgServerName;
2257                         case MAG_SCRIPTPATH:
2258                                 return $wgScriptPath;
2259                         case MAG_DIRECTIONMARK:
2260                                 return $wgContLang->getDirMark();
2261                         case MAG_CONTENTLANGUAGE:
2262                                 global $wgContLanguageCode;
2263                                 return $wgContLanguageCode;
2264                         default:
2265                                 $ret = null;
2266                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2267                                         return $ret;
2268                                 else
2269                                         return null;
2270                 }
2271         }
2272
2273         /**
2274          * initialise the magic variables (like CURRENTMONTHNAME)
2275          *
2276          * @private
2277          */
2278         function initialiseVariables() {
2279                 $fname = 'Parser::initialiseVariables';
2280                 wfProfileIn( $fname );
2281                 global $wgVariableIDs;
2282                 $this->mVariables = array();
2283                 foreach ( $wgVariableIDs as $id ) {
2284                         $mw =& MagicWord::get( $id );
2285                         $mw->addToArray( $this->mVariables, $id );
2286                 }
2287                 wfProfileOut( $fname );
2288         }
2289
2290         /**
2291          * parse any parentheses in format ((title|part|part))
2292          * and call callbacks to get a replacement text for any found piece
2293          *
2294          * @param string $text The text to parse
2295          * @param array $callbacks rules in form:
2296          *     '{' => array(                            # opening parentheses
2297          *                                      'end' => '}',   # closing parentheses
2298          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2299          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2300          *                                                                )
2301          *                                      )
2302          * @private
2303          */
2304         function replace_callback ($text, $callbacks) {
2305                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2306                 $lastOpeningBrace = -1;                 # last not closed parentheses
2307
2308                 for ($i = 0; $i < strlen($text); $i++) {
2309                         # check for any opening brace
2310                         $rule = null;
2311                         $nextPos = -1;
2312                         foreach ($callbacks as $key => $value) {
2313                                 $pos = strpos ($text, $key, $i);
2314                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2315                                         $rule = $value;
2316                                         $nextPos = $pos;
2317                                 }
2318                         }
2319
2320                         if ($lastOpeningBrace >= 0) {
2321                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2322
2323                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2324                                         $rule = null;
2325                                         $nextPos = $pos;
2326                                 }
2327
2328                                 $pos = strpos ($text, '|', $i);
2329
2330                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2331                                         $rule = null;
2332                                         $nextPos = $pos;
2333                                 }
2334                         }
2335
2336                         if ($nextPos == -1)
2337                                 break;
2338
2339                         $i = $nextPos;
2340
2341                         # found openning brace, lets add it to parentheses stack
2342                         if (null != $rule) {
2343                                 $piece = array('brace' => $text[$i],
2344                                                            'braceEnd' => $rule['end'],
2345                                                            'count' => 1,
2346                                                            'title' => '',
2347                                                            'parts' => null);
2348
2349                                 # count openning brace characters
2350                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2351                                         $piece['count']++;
2352                                         $i++;
2353                                 }
2354
2355                                 $piece['startAt'] = $i+1;
2356                                 $piece['partStart'] = $i+1;
2357
2358                                 # we need to add to stack only if openning brace count is enough for any given rule
2359                                 foreach ($rule['cb'] as $cnt => $fn) {
2360                                         if ($piece['count'] >= $cnt) {
2361                                                 $lastOpeningBrace ++;
2362                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2363                                                 break;
2364                                         }
2365                                 }
2366
2367                                 continue;
2368                         }
2369                         else if ($lastOpeningBrace >= 0) {
2370                                 # first check if it is a closing brace
2371                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2372                                         # lets check if it is enough characters for closing brace
2373                                         $count = 1;
2374                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2375                                                 $count++;
2376
2377                                         # if there are more closing parentheses than opening ones, we parse less
2378                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2379                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2380
2381                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2382                                         $matchingCount = 0;
2383                                         $matchingCallback = null;
2384                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2385                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2386                                                         $matchingCount = $cnt;
2387                                                         $matchingCallback = $fn;
2388                                                 }
2389                                         }
2390
2391                                         if ($matchingCount == 0) {
2392                                                 $i += $count - 1;
2393                                                 continue;
2394                                         }
2395
2396                                         # lets set a title or last part (if '|' was found)
2397                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2398                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2399                                         else
2400                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2401
2402                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2403                                         $pieceEnd = $i + $matchingCount;
2404
2405                                         if( is_callable( $matchingCallback ) ) {
2406                                                 $cbArgs = array (
2407                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2408                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2409                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2410                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2411                                                                                  );
2412                                                 # finally we can call a user callback and replace piece of text
2413                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2414                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2415                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2416                                         }
2417                                         else {
2418                                                 # null value for callback means that parentheses should be parsed, but not replaced
2419                                                 $i += $matchingCount - 1;
2420                                         }
2421
2422                                         # reset last openning parentheses, but keep it in case there are unused characters
2423                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2424                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2425                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2426                                                                    'title' => '',
2427                                                                    'parts' => null,
2428                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2429                                         $openingBraceStack[$lastOpeningBrace--] = null;
2430
2431                                         if ($matchingCount < $piece['count']) {
2432                                                 $piece['count'] -= $matchingCount;
2433                                                 $piece['startAt'] -= $matchingCount;
2434                                                 $piece['partStart'] = $piece['startAt'];
2435                                                 # do we still qualify for any callback with remaining count?
2436                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2437                                                         if ($piece['count'] >= $cnt) {
2438                                                                 $lastOpeningBrace ++;
2439                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2440                                                                 break;
2441                                                         }
2442                                                 }
2443                                         }
2444                                         continue;
2445                                 }
2446
2447                                 # lets set a title if it is a first separator, or next part otherwise
2448                                 if ($text[$i] == '|') {
2449                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2450                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2451                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2452                                         }
2453                                         else
2454                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2455
2456                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2457                                 }
2458                         }
2459                 }
2460
2461                 return $text;
2462         }
2463
2464         /**
2465          * Replace magic variables, templates, and template arguments
2466          * with the appropriate text. Templates are substituted recursively,
2467          * taking care to avoid infinite loops.
2468          *
2469          * Note that the substitution depends on value of $mOutputType:
2470          *  OT_WIKI: only {{subst:}} templates
2471          *  OT_MSG: only magic variables
2472          *  OT_HTML: all templates and magic variables
2473          *
2474          * @param string $tex The text to transform
2475          * @param array $args Key-value pairs representing template parameters to substitute
2476          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2477          * @private
2478          */
2479         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2480                 # Prevent too big inclusions
2481                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2482                         return $text;
2483                 }
2484
2485                 $fname = 'Parser::replaceVariables';
2486                 wfProfileIn( $fname );
2487
2488                 # This function is called recursively. To keep track of arguments we need a stack:
2489                 array_push( $this->mArgStack, $args );
2490
2491                 $braceCallbacks = array();
2492                 if ( !$argsOnly ) {
2493                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2494                 }
2495                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2496                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2497                 }
2498                 $callbacks = array();
2499                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2500                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2501                 $text = $this->replace_callback ($text, $callbacks);
2502
2503                 array_pop( $this->mArgStack );
2504
2505                 wfProfileOut( $fname );
2506                 return $text;
2507         }
2508
2509         /**
2510          * Replace magic variables
2511          * @private
2512          */
2513         function variableSubstitution( $matches ) {
2514                 $fname = 'Parser::variableSubstitution';
2515                 $varname = $matches[1];
2516                 wfProfileIn( $fname );
2517                 if ( !$this->mVariables ) {
2518                         $this->initialiseVariables();
2519                 }
2520                 $skip = false;
2521                 if ( $this->mOutputType == OT_WIKI ) {
2522                         # Do only magic variables prefixed by SUBST
2523                         $mwSubst =& MagicWord::get( MAG_SUBST );
2524                         if (!$mwSubst->matchStartAndRemove( $varname ))
2525                                 $skip = true;
2526                         # Note that if we don't substitute the variable below,
2527                         # we don't remove the {{subst:}} magic word, in case
2528                         # it is a template rather than a magic variable.
2529                 }
2530                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2531                         $id = $this->mVariables[$varname];
2532                         $text = $this->getVariableValue( $id );
2533                         $this->mOutput->mContainsOldMagic = true;
2534                 } else {
2535                         $text = $matches[0];
2536                 }
2537                 wfProfileOut( $fname );
2538                 return $text;
2539         }
2540
2541         # Split template arguments
2542         function getTemplateArgs( $argsString ) {
2543                 if ( $argsString === '' ) {
2544                         return array();
2545                 }
2546
2547                 $args = explode( '|', substr( $argsString, 1 ) );
2548
2549                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2550                 # merged with the next arg because the '|' character between belongs
2551                 # to the link syntax and not the template parameter syntax.
2552                 $argc = count($args);
2553
2554                 for ( $i = 0; $i < $argc-1; $i++ ) {
2555                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2556                                 $args[$i] .= '|'.$args[$i+1];
2557                                 array_splice($args, $i+1, 1);
2558                                 $i--;
2559                                 $argc--;
2560                         }
2561                 }
2562
2563                 return $args;
2564         }
2565
2566         /**
2567          * Return the text of a template, after recursively
2568          * replacing any variables or templates within the template.
2569          *
2570          * @param array $piece The parts of the template
2571          *  $piece['text']: matched text
2572          *  $piece['title']: the title, i.e. the part before the |
2573          *  $piece['parts']: the parameter array
2574          * @return string the text of the template
2575          * @private
2576          */
2577         function braceSubstitution( $piece ) {
2578                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2579                 $fname = 'Parser::braceSubstitution';
2580                 wfProfileIn( $fname );
2581
2582                 # Flags
2583                 $found = false;             # $text has been filled
2584                 $nowiki = false;            # wiki markup in $text should be escaped
2585                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2586                 $noargs = false;            # Don't replace triple-brace arguments in $text
2587                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2588                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2589                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2590
2591                 # Title object, where $text came from
2592                 $title = NULL;
2593
2594                 $linestart = '';
2595
2596                 # $part1 is the bit before the first |, and must contain only title characters
2597                 # $args is a list of arguments, starting from index 0, not including $part1
2598
2599                 $part1 = $piece['title'];
2600                 # If the third subpattern matched anything, it will start with |
2601
2602                 if (null == $piece['parts']) {
2603                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2604                         if ($replaceWith != $piece['text']) {
2605                                 $text = $replaceWith;
2606                                 $found = true;
2607                                 $noparse = true;
2608                                 $noargs = true;
2609                         }
2610                 }
2611
2612                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2613                 $argc = count( $args );
2614
2615                 # SUBST
2616                 if ( !$found ) {
2617                         $mwSubst =& MagicWord::get( MAG_SUBST );
2618                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2619                                 # One of two possibilities is true:
2620                                 # 1) Found SUBST but not in the PST phase
2621                                 # 2) Didn't find SUBST and in the PST phase
2622                                 # In either case, return without further processing
2623                                 $text = $piece['text'];
2624                                 $found = true;
2625                                 $noparse = true;
2626                                 $noargs = true;
2627                         }
2628                 }
2629
2630                 # MSG, MSGNW, INT and RAW
2631                 if ( !$found ) {
2632                         # Check for MSGNW:
2633                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2634                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2635                                 $nowiki = true;
2636                         } else {
2637                                 # Remove obsolete MSG:
2638                                 $mwMsg =& MagicWord::get( MAG_MSG );
2639                                 $mwMsg->matchStartAndRemove( $part1 );
2640                         }
2641
2642                         # Check for RAW:
2643                         $mwRaw =& MagicWord::get( MAG_RAW );
2644                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2645                                 $forceRawInterwiki = true;
2646                         }
2647
2648                         # Check if it is an internal message
2649                         $mwInt =& MagicWord::get( MAG_INT );
2650                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2651                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2652                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2653                                         $found = true;
2654                                 }
2655                         }
2656                 }
2657
2658                 # NS
2659                 if ( !$found ) {
2660                         # Check for NS: (namespace expansion)
2661                         $mwNs = MagicWord::get( MAG_NS );
2662                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2663                                 if ( intval( $part1 ) || $part1 == "0" ) {
2664                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2665                                         $found = true;
2666                                 } else {
2667                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2668                                         if ( !is_null( $index ) ) {
2669                                                 $text = $linestart . $wgContLang->getNsText( $index );
2670                                                 $found = true;
2671                                         }
2672                                 }
2673                         }
2674                 }
2675
2676                 # URLENCODE
2677                 if( !$found ) {
2678                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2679                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2680                                 $text = $linestart . urlencode( $part1 );
2681                                 $found = true;
2682                         }
2683                 }
2684
2685                 # LCFIRST, UCFIRST, LC and UC
2686                 if ( !$found ) {
2687                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2688                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2689                         $lc =& MagicWord::get( MAG_LC );
2690                         $uc =& MagicWord::get( MAG_UC );
2691                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2692                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2693                                 $found = true;
2694                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2695                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2696                                 $found = true;
2697                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2698                                 $text = $linestart . $wgContLang->lc( $part1 );
2699                                 $found = true;
2700                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2701                                  $text = $linestart . $wgContLang->uc( $part1 );
2702                                  $found = true;
2703                         }
2704                 }
2705
2706                 # LOCALURL and FULLURL
2707                 if ( !$found ) {
2708                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2709                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2710                         $mwFull =& MagicWord::get( MAG_FULLURL );
2711                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2712
2713
2714                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2715                                 $func = 'getLocalURL';
2716                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2717                                 $func = 'escapeLocalURL';
2718                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2719                                 $func = 'getFullURL';
2720                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2721                                 $func = 'escapeFullURL';
2722                         } else {
2723                                 $func = false;
2724                         }
2725
2726                         if ( $func !== false ) {
2727                                 $title = Title::newFromText( $part1 );
2728                                 # Due to order of execution of a lot of bits, the values might be encoded
2729                                 # before arriving here; if that's true, then the title can't be created
2730                                 # and the variable will fail. If we can't get a decent title from the first
2731                                 # attempt, url-decode and try for a second.
2732                                 if( is_null( $title ) )
2733                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2734                                 if ( !is_null( $title ) ) {
2735                                         if ( $argc > 0 ) {
2736                                                 $text = $linestart . $title->$func( $args[0] );
2737                                         } else {
2738                                                 $text = $linestart . $title->$func();
2739                                         }
2740                                         $found = true;
2741                                 }
2742                         }
2743                 }
2744
2745                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2746                 # GRAMMAR
2747                 if ( !$found && $argc == 1 ) {
2748                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2749                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2750                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2751                                 $found = true;
2752                         }
2753                 }
2754
2755                 # PLURAL
2756                 if ( !$found && $argc >= 2 ) {
2757                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2758                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2759                                 while ( count($args) < 5 ) { $args[] = $args[count($args)-1]; }
2760                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1],
2761                                         $args[2], $args[3], $args[4]);
2762                                 $found = true;
2763                         }
2764                 }
2765
2766                 # DISPLAYTITLE
2767                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2768                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2769                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2770
2771                                 # Set title in parser output object
2772                                 $param = $args[0];
2773                                 $parserOptions = new ParserOptions;
2774                                 $local_parser = new Parser ();
2775                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2776                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2777
2778                                 # Add subtitle
2779                                 $t = $this->mTitle->getPrefixedText();
2780                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2781                                 $text = "" ;
2782                                 $found = true ;
2783                         }
2784                 }
2785
2786                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2787                 if( !$found ) {
2788                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2789                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2790                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2791                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2792                         foreach( $mwWordsToCheck as $word => $func ) {
2793                                 $mwCurrentWord =& MagicWord::get( $word );
2794                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2795                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2796                                         if( $mwRawSuffix->match( $args[0] ) ) {
2797                                                 # Raw and unformatted
2798                                                 $text = $linestart . call_user_func( $func );
2799                                         } else {
2800                                                 # Formatted according to the content default
2801                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2802                                         }
2803                                         $found = true;
2804                                 }
2805                         }
2806                 }
2807
2808                 # PAGESINNAMESPACE
2809                 if( !$found ) {
2810                         $mwPagesInNs =& MagicWord::get( MAG_PAGESINNAMESPACE );
2811                         if( $mwPagesInNs->matchStartAndRemove( $part1 ) ) {
2812                                 $found = true;
2813                                 $count = wfPagesInNs( intval( $part1 ) );
2814                                 $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2815                                 if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2816                                         $text = $linestart . $count;
2817                                 } else {
2818                                         $text = $linestart . $wgContLang->formatNum( $count );
2819                                 }
2820                         }
2821                 }
2822
2823                 # #LANGUAGE:
2824                 if( !$found ) {
2825                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2826                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2827                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2828                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2829                                 $found = true;
2830                         }
2831                 }
2832
2833                 # Extensions
2834                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2835                         $colonPos = strpos( $part1, ':' );
2836                         if ( $colonPos !== false ) {
2837                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2838                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2839                                         $funcArgs = array_map( 'trim', $args );
2840                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2841                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2842                                         $found = true;
2843
2844                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2845                                         //$noargs = true;
2846                                         //$noparse = true;
2847
2848                                         if ( is_array( $result ) ) {
2849                                                 $text = $linestart . $result[0];
2850                                                 unset( $result[0] );
2851
2852                                                 // Extract flags into the local scope
2853                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2854                                                 extract( $result );
2855                                         } else {
2856                                                 $text = $linestart . $result;
2857                                         }
2858                                 }
2859                         }
2860                 }
2861
2862                 # Template table test
2863
2864                 # Did we encounter this template already? If yes, it is in the cache
2865                 # and we need to check for loops.
2866                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2867                         $found = true;
2868
2869                         # Infinite loop test
2870                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2871                                 $noparse = true;
2872                                 $noargs = true;
2873                                 $found = true;
2874                                 $text = $linestart .
2875                                         '{{' . $part1 . '}}' .
2876                                         '<!-- WARNING: template loop detected -->';
2877                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2878                         } else {
2879                                 # set $text to cached message.
2880                                 $text = $linestart . $this->mTemplates[$piece['title']];
2881                         }
2882                 }
2883
2884                 # Load from database
2885                 $lastPathLevel = $this->mTemplatePath;
2886                 if ( !$found ) {
2887                         $ns = NS_TEMPLATE;
2888                         # declaring $subpage directly in the function call
2889                         # does not work correctly with references and breaks
2890                         # {{/subpage}}-style inclusions
2891                         $subpage = '';
2892                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2893                         if ($subpage !== '') {
2894                                 $ns = $this->mTitle->getNamespace();
2895                         }
2896                         $title = Title::newFromText( $part1, $ns );
2897
2898                         if ( !is_null( $title ) ) {
2899                                 if ( !$title->isExternal() ) {
2900                                         # Check for excessive inclusion
2901                                         $dbk = $title->getPrefixedDBkey();
2902                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2903                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2904                                                         $text = SpecialPage::capturePath( $title );
2905                                                         if ( is_string( $text ) ) {
2906                                                                 $found = true;
2907                                                                 $noparse = true;
2908                                                                 $noargs = true;
2909                                                                 $isHTML = true;
2910                                                                 $this->disableCache();
2911                                                         }
2912                                                 } else {
2913                                                         $articleContent = $this->fetchTemplate( $title );
2914                                                         if ( $articleContent !== false ) {
2915                                                                 $found = true;
2916                                                                 $text = $articleContent;
2917                                                                 $replaceHeadings = true;
2918                                                         }
2919                                                 }
2920                                         }
2921
2922                                         # If the title is valid but undisplayable, make a link to it
2923                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2924                                                 $text = '[['.$title->getPrefixedText().']]';
2925                                                 $found = true;
2926                                         }
2927                                 } elseif ( $title->isTrans() ) {
2928                                         // Interwiki transclusion
2929                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2930                                                 $text = $this->interwikiTransclude( $title, 'render' );
2931                                                 $isHTML = true;
2932                                                 $noparse = true;
2933                                         } else {
2934                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2935                                                 $replaceHeadings = true;
2936                                         }
2937                                         $found = true;
2938                                 }
2939
2940                                 # Template cache array insertion
2941                                 # Use the original $piece['title'] not the mangled $part1, so that
2942                                 # modifiers such as RAW: produce separate cache entries
2943                                 if( $found ) {
2944                                         if( $isHTML ) {
2945                                                 // A special page; don't store it in the template cache.
2946                                         } else {
2947                                                 $this->mTemplates[$piece['title']] = $text;
2948                                         }
2949                                         $text = $linestart . $text;
2950                                 }
2951                         }
2952                 }
2953
2954                 # Recursive parsing, escaping and link table handling
2955                 # Only for HTML output
2956                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2957                         $text = wfEscapeWikiText( $text );
2958                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2959                         if ( $noargs ) {
2960                                 $assocArgs = array();
2961                         } else {
2962                                 # Clean up argument array
2963                                 $assocArgs = array();
2964                                 $index = 1;
2965                                 foreach( $args as $arg ) {
2966                                         $eqpos = strpos( $arg, '=' );
2967                                         if ( $eqpos === false ) {
2968                                                 $assocArgs[$index++] = $arg;
2969                                         } else {
2970                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2971                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2972                                                 if ( $value === false ) {
2973                                                         $value = '';
2974                                                 }
2975                                                 if ( $name !== false ) {
2976                                                         $assocArgs[$name] = $value;
2977                                                 }
2978                                         }
2979                                 }
2980
2981                                 # Add a new element to the templace recursion path
2982                                 $this->mTemplatePath[$part1] = 1;
2983                         }
2984
2985                         if ( !$noparse ) {
2986                                 # If there are any <onlyinclude> tags, only include them
2987                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2988                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2989                                         $text = '';
2990                                         foreach ($m[1] as $piece)
2991                                                 $text .= $piece;
2992                                 }
2993                                 # Remove <noinclude> sections and <includeonly> tags
2994                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2995                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2996
2997                                 if( $this->mOutputType == OT_HTML ) {
2998                                         # Strip <nowiki>, <pre>, etc.
2999                                         $text = $this->strip( $text, $this->mStripState );
3000                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3001                                 }
3002                                 $text = $this->replaceVariables( $text, $assocArgs );
3003
3004                                 # If the template begins with a table or block-level
3005                                 # element, it should be treated as beginning a new line.
3006                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3007                                         $text = "\n" . $text;
3008                                 }
3009                         } elseif ( !$noargs ) {
3010                                 # $noparse and !$noargs
3011                                 # Just replace the arguments, not any double-brace items
3012                                 # This is used for rendered interwiki transclusion
3013                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3014                         }
3015                 }
3016                 # Prune lower levels off the recursion check path
3017                 $this->mTemplatePath = $lastPathLevel;
3018
3019                 if ( !$found ) {
3020                         wfProfileOut( $fname );
3021                         return $piece['text'];
3022                 } else {
3023                         if ( $isHTML ) {
3024                                 # Replace raw HTML by a placeholder
3025                                 # Add a blank line preceding, to prevent it from mucking up
3026                                 # immediately preceding headings
3027                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3028                         } else {
3029                                 # replace ==section headers==
3030                                 # XXX this needs to go away once we have a better parser.
3031                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3032                                         if( !is_null( $title ) )
3033                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3034                                         else
3035                                                 $encodedname = base64_encode("");
3036                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3037                                                 PREG_SPLIT_DELIM_CAPTURE);
3038                                         $text = '';
3039                                         $nsec = 0;
3040                                         for( $i = 0; $i < count($m); $i += 2 ) {
3041                                                 $text .= $m[$i];
3042                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3043                                                 $hl = $m[$i + 1];
3044                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3045                                                         $text .= $hl;
3046                                                         continue;
3047                                                 }
3048                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3049                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3050                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3051
3052                                                 $nsec++;
3053                                         }
3054                                 }
3055                         }
3056                 }
3057
3058                 # Prune lower levels off the recursion check path
3059                 $this->mTemplatePath = $lastPathLevel;
3060
3061                 if ( !$found ) {
3062                         wfProfileOut( $fname );
3063                         return $piece['text'];
3064                 } else {
3065                         wfProfileOut( $fname );
3066                         return $text;
3067                 }
3068         }
3069
3070         /**
3071          * Fetch the unparsed text of a template and register a reference to it.
3072          */
3073         function fetchTemplate( $title ) {
3074                 $text = false;
3075                 // Loop to fetch the article, with up to 1 redirect
3076                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3077                         $rev = Revision::newFromTitle( $title );
3078                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3079                         if ( !$rev ) {
3080                                 break;
3081                         }
3082                         $text = $rev->getText();
3083                         if ( $text === false ) {
3084                                 break;
3085                         }
3086                         // Redirect?
3087                         $title = Title::newFromRedirect( $text );
3088                 }
3089                 return $text;
3090         }
3091
3092         /**
3093          * Transclude an interwiki link.
3094          */
3095         function interwikiTransclude( $title, $action ) {
3096                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3097
3098                 if (!$wgEnableScaryTranscluding)
3099                         return wfMsg('scarytranscludedisabled');
3100
3101                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3102                 // But we'll handle it generally anyway
3103                 if ( $title->getNamespace() ) {
3104                         // Use the canonical namespace, which should work anywhere
3105                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3106                 } else {
3107                         $articleName = $title->getDBkey();
3108                 }
3109
3110                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3111                 $url .= "?action=$action";
3112                 if (strlen($url) > 255)
3113                         return wfMsg('scarytranscludetoolong');
3114                 return $this->fetchScaryTemplateMaybeFromCache($url);
3115         }
3116
3117         function fetchScaryTemplateMaybeFromCache($url) {
3118                 global $wgTranscludeCacheExpiry;
3119                 $dbr =& wfGetDB(DB_SLAVE);
3120                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3121                                 array('tc_url' => $url));
3122                 if ($obj) {
3123                         $time = $obj->tc_time;
3124                         $text = $obj->tc_contents;
3125                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3126                                 return $text;
3127                         }
3128                 }
3129
3130                 $text = wfGetHTTP($url);
3131                 if (!$text)
3132                         return wfMsg('scarytranscludefailed', $url);
3133
3134                 $dbw =& wfGetDB(DB_MASTER);
3135                 $dbw->replace('transcache', array('tc_url'), array(
3136                         'tc_url' => $url,
3137                         'tc_time' => time(),
3138                         'tc_contents' => $text));
3139                 return $text;
3140         }
3141
3142
3143         /**
3144          * Triple brace replacement -- used for template arguments
3145          * @private
3146          */
3147         function argSubstitution( $matches ) {
3148                 $arg = trim( $matches['title'] );
3149                 $text = $matches['text'];
3150                 $inputArgs = end( $this->mArgStack );
3151
3152                 if ( array_key_exists( $arg, $inputArgs ) ) {
3153                         $text = $inputArgs[$arg];
3154                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3155                         $text = $matches['parts'][0];
3156                 }
3157
3158                 return $text;
3159         }
3160
3161         /**
3162          * Returns true if the function is allowed to include this entity
3163          * @private
3164          */
3165         function incrementIncludeCount( $dbk ) {
3166                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3167                         $this->mIncludeCount[$dbk] = 0;
3168                 }
3169                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3170                         return true;
3171                 } else {
3172                         return false;
3173                 }
3174         }
3175
3176         /**
3177          * Detect __NOGALLERY__ magic word and set a placeholder
3178          */
3179         function stripNoGallery( &$text ) {
3180                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3181                 # do not add TOC
3182                 $mw = MagicWord::get( MAG_NOGALLERY );
3183                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3184         }
3185
3186         /**
3187          * Detect __TOC__ magic word and set a placeholder
3188          */
3189         function stripToc( $text ) {
3190                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3191                 # do not add TOC
3192                 $mw = MagicWord::get( MAG_NOTOC );
3193                 if( $mw->matchAndRemove( $text ) ) {
3194                         $this->mShowToc = false;
3195                 }
3196
3197                 $mw = MagicWord::get( MAG_TOC );
3198                 if( $mw->match( $text ) ) {
3199                         $this->mShowToc = true;
3200                         $this->mForceTocPosition = true;
3201
3202                         // Set a placeholder. At the end we'll fill it in with the TOC.
3203                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3204
3205                         // Only keep the first one.
3206                         $text = $mw->replace( '', $text );
3207                 }
3208                 return $text;
3209         }
3210
3211         /**
3212          * This function accomplishes several tasks:
3213          * 1) Auto-number headings if that option is enabled
3214          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3215          * 3) Add a Table of contents on the top for users who have enabled the option
3216          * 4) Auto-anchor headings
3217          *
3218          * It loops through all headlines, collects the necessary data, then splits up the
3219          * string and re-inserts the newly formatted headlines.
3220          *
3221          * @param string $text
3222          * @param boolean $isMain
3223          * @private
3224          */
3225         function formatHeadings( $text, $isMain=true ) {
3226                 global $wgMaxTocLevel, $wgContLang;
3227
3228                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3229                 if( !$this->mTitle->userCanEdit() ) {
3230                         $showEditLink = 0;
3231                 } else {
3232                         $showEditLink = $this->mOptions->getEditSection();
3233                 }
3234
3235                 # Inhibit editsection links if requested in the page
3236                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3237                 if( $esw->matchAndRemove( $text ) ) {
3238                         $showEditLink = 0;
3239                 }
3240
3241                 # Get all headlines for numbering them and adding funky stuff like [edit]
3242                 # links - this is for later, but we need the number of headlines right now
3243                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3244
3245                 # if there are fewer than 4 headlines in the article, do not show TOC
3246                 # unless it's been explicitly enabled.
3247                 $enoughToc = $this->mShowToc &&
3248                         (($numMatches >= 4) || $this->mForceTocPosition);
3249
3250                 # Allow user to stipulate that a page should have a "new section"
3251                 # link added via __NEWSECTIONLINK__
3252                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3253                 if( $mw->matchAndRemove( $text ) )
3254                         $this->mOutput->setNewSection( true );
3255
3256                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3257                 # override above conditions and always show TOC above first header
3258                 $mw =& MagicWord::get( MAG_FORCETOC );
3259                 if ($mw->matchAndRemove( $text ) ) {
3260                         $this->mShowToc = true;
3261                         $enoughToc = true;
3262                 }
3263
3264                 # Never ever show TOC if no headers
3265                 if( $numMatches < 1 ) {
3266                         $enoughToc = false;
3267                 }
3268
3269                 # We need this to perform operations on the HTML
3270                 $sk =& $this->mOptions->getSkin();
3271
3272                 # headline counter
3273                 $headlineCount = 0;
3274                 $sectionCount = 0; # headlineCount excluding template sections
3275
3276                 # Ugh .. the TOC should have neat indentation levels which can be
3277                 # passed to the skin functions. These are determined here
3278                 $toc = '';
3279                 $full = '';
3280                 $head = array();
3281                 $sublevelCount = array();
3282                 $levelCount = array();
3283                 $toclevel = 0;
3284                 $level = 0;
3285                 $prevlevel = 0;
3286                 $toclevel = 0;
3287                 $prevtoclevel = 0;
3288
3289                 foreach( $matches[3] as $headline ) {
3290                         $istemplate = 0;
3291                         $templatetitle = '';
3292                         $templatesection = 0;
3293                         $numbering = '';
3294
3295                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3296                                 $istemplate = 1;
3297                                 $templatetitle = base64_decode($mat[1]);
3298                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3299                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3300                         }
3301
3302                         if( $toclevel ) {
3303                                 $prevlevel = $level;
3304                                 $prevtoclevel = $toclevel;
3305                         }
3306                         $level = $matches[1][$headlineCount];
3307
3308                         if( $doNumberHeadings || $enoughToc ) {
3309
3310                                 if ( $level > $prevlevel ) {
3311                                         # Increase TOC level
3312                                         $toclevel++;
3313                                         $sublevelCount[$toclevel] = 0;
3314                                         if( $toclevel<$wgMaxTocLevel ) {
3315                                                 $toc .= $sk->tocIndent();
3316                                         }
3317                                 }
3318                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3319                                         # Decrease TOC level, find level to jump to
3320
3321                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3322                                                 # Can only go down to level 1
3323                                                 $toclevel = 1;
3324                                         } else {
3325                                                 for ($i = $toclevel; $i > 0; $i--) {
3326                                                         if ( $levelCount[$i] == $level ) {
3327                                                                 # Found last matching level
3328                                                                 $toclevel = $i;
3329                                                                 break;
3330                                                         }
3331                                                         elseif ( $levelCount[$i] < $level ) {
3332                                                                 # Found first matching level below current level
3333                                                                 $toclevel = $i + 1;
3334                                                                 break;
3335                                                         }
3336                                                 }
3337                                         }
3338                                         if( $toclevel<$wgMaxTocLevel ) {
3339                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3340                                         }
3341                                 }
3342                                 else {
3343                                         # No change in level, end TOC line
3344                                         if( $toclevel<$wgMaxTocLevel ) {
3345                                                 $toc .= $sk->tocLineEnd();
3346                                         }
3347                                 }
3348
3349                                 $levelCount[$toclevel] = $level;
3350
3351                                 # count number of headlines for each level
3352                                 @$sublevelCount[$toclevel]++;
3353                                 $dot = 0;
3354                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3355                                         if( !empty( $sublevelCount[$i] ) ) {
3356                                                 if( $dot ) {
3357                                                         $numbering .= '.';
3358                                                 }
3359                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3360                                                 $dot = 1;
3361                                         }
3362                                 }
3363                         }
3364
3365                         # The canonized header is a version of the header text safe to use for links
3366                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3367                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3368                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3369
3370                         # Remove link placeholders by the link text.
3371                         #     <!--LINK number-->
3372                         # turns into
3373                         #     link text with suffix
3374                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3375                                                             "\$this->mLinkHolders['texts'][\$1]",
3376                                                             $canonized_headline );
3377                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3378                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3379                                                             $canonized_headline );
3380
3381                         # strip out HTML
3382                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3383                         $tocline = trim( $canonized_headline );
3384                         # Save headline for section edit hint before it's escaped
3385                         $headline_hint = trim( $canonized_headline );
3386                         $canonized_headline = Sanitizer::escapeId( $tocline );
3387                         $refers[$headlineCount] = $canonized_headline;
3388
3389                         # count how many in assoc. array so we can track dupes in anchors
3390                         @$refers[$canonized_headline]++;
3391                         $refcount[$headlineCount]=$refers[$canonized_headline];
3392
3393                         # Don't number the heading if it is the only one (looks silly)
3394                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3395                                 # the two are different if the line contains a link
3396                                 $headline=$numbering . ' ' . $headline;
3397                         }
3398
3399                         # Create the anchor for linking from the TOC to the section
3400                         $anchor = $canonized_headline;
3401                         if($refcount[$headlineCount] > 1 ) {
3402                                 $anchor .= '_' . $refcount[$headlineCount];
3403                         }
3404                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3405                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3406                         }
3407                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3408                                 if ( empty( $head[$headlineCount] ) ) {
3409                                         $head[$headlineCount] = '';
3410                                 }
3411                                 if( $istemplate )
3412                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3413                                 else
3414                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3415                         }
3416
3417                         # give headline the correct <h#> tag
3418                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3419
3420                         $headlineCount++;
3421                         if( !$istemplate )
3422                                 $sectionCount++;
3423                 }
3424
3425                 if( $enoughToc ) {
3426                         if( $toclevel<$wgMaxTocLevel ) {
3427                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3428                         }
3429                         $toc = $sk->tocList( $toc );
3430                 }
3431
3432                 # split up and insert constructed headlines
3433
3434                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3435                 $i = 0;
3436
3437                 foreach( $blocks as $block ) {
3438                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3439                                 # This is the [edit] link that appears for the top block of text when
3440                                 # section editing is enabled
3441
3442                                 # Disabled because it broke block formatting
3443                                 # For example, a bullet point in the top line
3444                                 # $full .= $sk->editSectionLink(0);
3445                         }
3446                         $full .= $block;
3447                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3448                                 # Top anchor now in skin
3449                                 $full = $full.$toc;
3450                         }
3451
3452                         if( !empty( $head[$i] ) ) {
3453                                 $full .= $head[$i];
3454                         }
3455                         $i++;
3456                 }
3457                 if( $this->mForceTocPosition ) {
3458                         return str_replace( '<!--MWTOC-->', $toc, $full );
3459                 } else {
3460                         return $full;
3461                 }
3462         }
3463
3464         /**
3465          * Return an HTML link for the "ISBN 123456" text
3466          * @private
3467          */
3468         function magicISBN( $text ) {
3469                 $fname = 'Parser::magicISBN';
3470                 wfProfileIn( $fname );
3471
3472                 $a = split( 'ISBN ', ' '.$text );
3473                 if ( count ( $a ) < 2 ) {
3474                         wfProfileOut( $fname );
3475                         return $text;
3476                 }
3477                 $text = substr( array_shift( $a ), 1);
3478                 $valid = '0123456789-Xx';
3479
3480                 foreach ( $a as $x ) {
3481                         # hack: don't replace inside thumbnail title/alt
3482                         # attributes
3483                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3484                                 $text .= "ISBN $x";
3485                                 continue;
3486                         }
3487
3488                         $isbn = $blank = '' ;
3489                         while ( ' ' == $x{0} ) {
3490                                 $blank .= ' ';
3491                                 $x = substr( $x, 1 );
3492                         }
3493                         if ( $x == '' ) { # blank isbn
3494                                 $text .= "ISBN $blank";
3495                                 continue;
3496                         }
3497                         while ( strstr( $valid, $x{0} ) != false ) {
3498                                 $isbn .= $x{0};
3499                                 $x = substr( $x, 1 );
3500                         }
3501                         $num = str_replace( '-', '', $isbn );
3502                         $num = str_replace( ' ', '', $num );
3503                         $num = str_replace( 'x', 'X', $num );
3504
3505                         if ( '' == $num ) {
3506                                 $text .= "ISBN $blank$x";
3507                         } else {
3508                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3509                                 $text .= '<a href="' .
3510                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3511                                         "\" class=\"internal\">ISBN $isbn</a>";
3512                                 $text .= $x;
3513                         }
3514                 }
3515                 wfProfileOut( $fname );
3516                 return $text;
3517         }
3518
3519         /**
3520          * Return an HTML link for the "RFC 1234" text
3521          *
3522          * @private
3523          * @param string $text     Text to be processed
3524          * @param string $keyword  Magic keyword to use (default RFC)
3525          * @param string $urlmsg   Interface message to use (default rfcurl)
3526          * @return string
3527          */
3528         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3529
3530                 $valid = '0123456789';
3531                 $internal = false;
3532
3533                 $a = split( $keyword, ' '.$text );
3534                 if ( count ( $a ) < 2 ) {
3535                         return $text;
3536                 }
3537                 $text = substr( array_shift( $a ), 1);
3538
3539                 /* Check if keyword is preceed by [[.
3540                  * This test is made here cause of the array_shift above
3541                  * that prevent the test to be done in the foreach.
3542                  */
3543                 if ( substr( $text, -2 ) == '[[' ) {
3544                         $internal = true;
3545                 }
3546
3547                 foreach ( $a as $x ) {
3548                         /* token might be empty if we have RFC RFC 1234 */
3549                         if ( $x=='' ) {
3550                                 $text.=$keyword;
3551                                 continue;
3552                                 }
3553
3554                         # hack: don't replace inside thumbnail title/alt
3555                         # attributes
3556                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3557                                 $text .= $keyword . $x;
3558                                 continue;
3559                         }
3560
3561                         $id = $blank = '' ;
3562
3563                         /** remove and save whitespaces in $blank */
3564                         while ( $x{0} == ' ' ) {
3565                                 $blank .= ' ';
3566                                 $x = substr( $x, 1 );
3567                         }
3568
3569                         /** remove and save the rfc number in $id */
3570                         while ( strstr( $valid, $x{0} ) != false ) {
3571                                 $id .= $x{0};
3572                                 $x = substr( $x, 1 );
3573                         }
3574
3575                         if ( $id == '' ) {
3576                                 /* call back stripped spaces*/
3577                                 $text .= $keyword.$blank.$x;
3578                         } elseif( $internal ) {
3579                                 /* normal link */
3580                                 $text .= $keyword.$id.$x;
3581                         } else {
3582                                 /* build the external link*/
3583                                 $url = wfMsg( $urlmsg, $id);
3584                                 $sk =& $this->mOptions->getSkin();
3585                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3586                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3587                         }
3588
3589                         /* Check if the next RFC keyword is preceed by [[ */
3590                         $internal = ( substr($x,-2) == '[[' );
3591                 }
3592                 return $text;
3593         }
3594
3595         /**
3596          * Transform wiki markup when saving a page by doing \r\n -> \n
3597          * conversion, substitting signatures, {{subst:}} templates, etc.
3598          *
3599          * @param string $text the text to transform
3600          * @param Title &$title the Title object for the current article
3601          * @param User &$user the User object describing the current user
3602          * @param ParserOptions $options parsing options
3603          * @param bool $clearState whether to clear the parser state first
3604          * @return string the altered wiki markup
3605          * @public
3606          */
3607         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3608                 $this->mOptions = $options;
3609                 $this->mTitle =& $title;
3610                 $this->mOutputType = OT_WIKI;
3611
3612                 if ( $clearState ) {
3613                         $this->clearState();
3614                 }
3615
3616                 $stripState = false;
3617                 $pairs = array(
3618                         "\r\n" => "\n",
3619                 );
3620                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3621                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3622                 $text = $this->pstPass2( $text, $stripState, $user );
3623                 $text = $this->unstrip( $text, $stripState );
3624                 $text = $this->unstripNoWiki( $text, $stripState );
3625                 return $text;
3626         }
3627
3628         /**
3629          * Pre-save transform helper function
3630          * @private
3631          */
3632         function pstPass2( $text, &$stripState, &$user ) {
3633                 global $wgContLang, $wgLocaltimezone;
3634
3635                 /* Note: This is the timestamp saved as hardcoded wikitext to
3636                  * the database, we use $wgContLang here in order to give
3637                  * everyone the same signature and use the default one rather
3638                  * than the one selected in each user's preferences.
3639                  */
3640                 if ( isset( $wgLocaltimezone ) ) {
3641                         $oldtz = getenv( 'TZ' );
3642                         putenv( 'TZ='.$wgLocaltimezone );
3643                 }
3644                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3645                   ' (' . date( 'T' ) . ')';
3646                 if ( isset( $wgLocaltimezone ) ) {
3647                         putenv( 'TZ='.$oldtz );
3648                 }
3649
3650                 # Variable replacement
3651                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3652                 $text = $this->replaceVariables( $text );
3653
3654                 # Strip out <nowiki> etc. added via replaceVariables
3655                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3656
3657                 # Signatures
3658                 $sigText = $this->getUserSig( $user );
3659                 $text = strtr( $text, array(
3660                         '~~~~~' => $d,
3661                         '~~~~' => "$sigText $d",
3662                         '~~~' => $sigText
3663                 ) );
3664
3665                 # Context links: [[|name]] and [[name (context)|]]
3666                 #
3667                 global $wgLegalTitleChars;
3668                 $tc = "[$wgLegalTitleChars]";
3669                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3670
3671                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3672                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3673
3674                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3675                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3676                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3677                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3678                 $context = '';
3679                 $t = $this->mTitle->getText();
3680                 if ( preg_match( $conpat, $t, $m ) ) {
3681                         $context = $m[2];
3682                 }
3683                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3684                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3685                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3686
3687                 if ( '' == $context ) {
3688                         $text = preg_replace( $p2, '[[\\1]]', $text );
3689                 } else {
3690                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3691                 }
3692
3693                 # Trim trailing whitespace
3694                 # MAG_END (__END__) tag allows for trailing
3695                 # whitespace to be deliberately included
3696                 $text = rtrim( $text );
3697                 $mw =& MagicWord::get( MAG_END );
3698                 $mw->matchAndRemove( $text );
3699
3700                 return $text;
3701         }
3702
3703         /**
3704          * Fetch the user's signature text, if any, and normalize to
3705          * validated, ready-to-insert wikitext.
3706          *
3707          * @param User $user
3708          * @return string
3709          * @private
3710          */
3711         function getUserSig( &$user ) {
3712                 $username = $user->getName();
3713                 $nickname = $user->getOption( 'nickname' );
3714                 $nickname = $nickname === '' ? $username : $nickname;
3715
3716                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3717                         # Sig. might contain markup; validate this
3718                         if( $this->validateSig( $nickname ) !== false ) {
3719                                 # Validated; clean up (if needed) and return it
3720                                 return $this->cleanSig( $nickname, true );
3721                         } else {
3722                                 # Failed to validate; fall back to the default
3723                                 $nickname = $username;
3724                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3725                         }
3726                 }
3727
3728                 # If we're still here, make it a link to the user page
3729                 $userpage = $user->getUserPage();
3730                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3731         }
3732
3733         /**
3734          * Check that the user's signature contains no bad XML
3735          *
3736          * @param string $text
3737          * @return mixed An expanded string, or false if invalid.
3738          */
3739         function validateSig( $text ) {
3740                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3741         }
3742
3743         /**
3744          * Clean up signature text
3745          *
3746          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3747          * 2) Substitute all transclusions
3748          *
3749          * @param string $text
3750          * @param $parsing Whether we're cleaning (preferences save) or parsing
3751          * @return string Signature text
3752          */
3753         function cleanSig( $text, $parsing = false ) {
3754                 global $wgTitle;
3755                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3756
3757                 $substWord = MagicWord::get( MAG_SUBST );
3758                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3759                 $substText = '{{' . $substWord->getSynonym( 0 );
3760
3761                 $text = preg_replace( $substRegex, $substText, $text );
3762                 $text = preg_replace( '/~{3,5}/', '', $text );
3763                 $text = $this->replaceVariables( $text );
3764
3765                 $this->clearState();
3766                 return $text;
3767         }
3768
3769         /**
3770          * Set up some variables which are usually set up in parse()
3771          * so that an external function can call some class members with confidence
3772          * @public
3773          */
3774         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3775                 $this->mTitle =& $title;
3776                 $this->mOptions = $options;
3777                 $this->mOutputType = $outputType;
3778                 if ( $clearState ) {
3779                         $this->clearState();
3780                 }
3781         }
3782
3783         /**
3784          * Transform a MediaWiki message by replacing magic variables.
3785          *
3786          * @param string $text the text to transform
3787          * @param ParserOptions $options  options
3788          * @return string the text with variables substituted
3789          * @public
3790          */
3791         function transformMsg( $text, $options ) {
3792                 global $wgTitle;
3793                 static $executing = false;
3794
3795                 $fname = "Parser::transformMsg";
3796
3797                 # Guard against infinite recursion
3798                 if ( $executing ) {
3799                         return $text;
3800                 }
3801                 $executing = true;
3802
3803                 wfProfileIn($fname);
3804
3805                 $this->mTitle = $wgTitle;
3806                 $this->mOptions = $options;
3807                 $this->mOutputType = OT_MSG;
3808                 $this->clearState();
3809                 $text = $this->replaceVariables( $text );
3810
3811                 $executing = false;
3812                 wfProfileOut($fname);
3813                 return $text;
3814         }
3815
3816         /**
3817          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3818          * The callback should have the following form:
3819          *    function myParserHook( $text, $params, &$parser ) { ... }
3820          *
3821          * Transform and return $text. Use $parser for any required context, e.g. use
3822          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3823          *
3824          * @public
3825          *
3826          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3827          * @param mixed $callback The callback function (and object) to use for the tag
3828          *
3829          * @return The old value of the mTagHooks array associated with the hook
3830          */
3831         function setHook( $tag, $callback ) {
3832                 $tag = strtolower( $tag );
3833                 $oldVal = @$this->mTagHooks[$tag];
3834                 $this->mTagHooks[$tag] = $callback;
3835
3836                 return $oldVal;
3837         }
3838
3839         /**
3840          * Create a function, e.g. {{sum:1|2|3}}
3841          * The callback function should have the form:
3842          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3843          *
3844          * The callback may either return the text result of the function, or an array with the text
3845          * in element 0, and a number of flags in the other elements. The names of the flags are
3846          * specified in the keys. Valid flags are:
3847          *   found                     The text returned is valid, stop processing the template. This
3848          *                             is on by default.
3849          *   nowiki                    Wiki markup in the return value should be escaped
3850          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3851          *   noargs                    Don't replace triple-brace arguments in the return value
3852          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3853          *
3854          * @public
3855          *
3856          * @param string $name The function name. Function names are case-insensitive.
3857          * @param mixed $callback The callback function (and object) to use
3858          *
3859          * @return The old callback function for this name, if any
3860          */
3861         function setFunctionHook( $name, $callback ) {
3862                 $name = strtolower( $name );
3863                 $oldVal = @$this->mFunctionHooks[$name];
3864                 $this->mFunctionHooks[$name] = $callback;
3865                 return $oldVal;
3866         }
3867
3868         /**
3869          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3870          * Placeholders created in Skin::makeLinkObj()
3871          * Returns an array of links found, indexed by PDBK:
3872          *  0 - broken
3873          *  1 - normal link
3874          *  2 - stub
3875          * $options is a bit field, RLH_FOR_UPDATE to select for update
3876          */
3877         function replaceLinkHolders( &$text, $options = 0 ) {
3878                 global $wgUser;
3879                 global $wgOutputReplace;
3880
3881                 $fname = 'Parser::replaceLinkHolders';
3882                 wfProfileIn( $fname );
3883
3884                 $pdbks = array();
3885                 $colours = array();
3886                 $sk =& $this->mOptions->getSkin();
3887                 $linkCache =& LinkCache::singleton();
3888
3889                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3890                         wfProfileIn( $fname.'-check' );
3891                         $dbr =& wfGetDB( DB_SLAVE );
3892                         $page = $dbr->tableName( 'page' );
3893                         $threshold = $wgUser->getOption('stubthreshold');
3894
3895                         # Sort by namespace
3896                         asort( $this->mLinkHolders['namespaces'] );
3897
3898                         # Generate query
3899                         $query = false;
3900                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3901                                 # Make title object
3902                                 $title = $this->mLinkHolders['titles'][$key];
3903
3904                                 # Skip invalid entries.
3905                                 # Result will be ugly, but prevents crash.
3906                                 if ( is_null( $title ) ) {
3907                                         continue;
3908                                 }
3909                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3910
3911                                 # Check if it's a static known link, e.g. interwiki
3912                                 if ( $title->isAlwaysKnown() ) {
3913                                         $colours[$pdbk] = 1;
3914                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3915                                         $colours[$pdbk] = 1;
3916                                         $this->mOutput->addLink( $title, $id );
3917                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3918                                         $colours[$pdbk] = 0;
3919                                 } else {
3920                                         # Not in the link cache, add it to the query
3921                                         if ( !isset( $current ) ) {
3922                                                 $current = $ns;
3923                                                 $query =  "SELECT page_id, page_namespace, page_title";
3924                                                 if ( $threshold > 0 ) {
3925                                                         $query .= ', page_len, page_is_redirect';
3926                                                 }
3927                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3928                                         } elseif ( $current != $ns ) {
3929                                                 $current = $ns;
3930                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3931                                         } else {
3932                                                 $query .= ', ';
3933                                         }
3934
3935                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3936                                 }
3937                         }
3938                         if ( $query ) {
3939                                 $query .= '))';
3940                                 if ( $options & RLH_FOR_UPDATE ) {
3941                                         $query .= ' FOR UPDATE';
3942                                 }
3943
3944                                 $res = $dbr->query( $query, $fname );
3945
3946                                 # Fetch data and form into an associative array
3947                                 # non-existent = broken
3948                                 # 1 = known
3949                                 # 2 = stub
3950                                 while ( $s = $dbr->fetchObject($res) ) {
3951                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3952                                         $pdbk = $title->getPrefixedDBkey();
3953                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3954                                         $this->mOutput->addLink( $title, $s->page_id );
3955
3956                                         if ( $threshold >  0 ) {
3957                                                 $size = $s->page_len;
3958                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3959                                                         $colours[$pdbk] = 1;
3960                                                 } else {
3961                                                         $colours[$pdbk] = 2;
3962                                                 }
3963                                         } else {
3964                                                 $colours[$pdbk] = 1;
3965                                         }
3966                                 }
3967                         }
3968                         wfProfileOut( $fname.'-check' );
3969
3970                         # Construct search and replace arrays
3971                         wfProfileIn( $fname.'-construct' );
3972                         $wgOutputReplace = array();
3973                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3974                                 $pdbk = $pdbks[$key];
3975                                 $searchkey = "<!--LINK $key-->";
3976                                 $title = $this->mLinkHolders['titles'][$key];
3977                                 if ( empty( $colours[$pdbk] ) ) {
3978                                         $linkCache->addBadLinkObj( $title );
3979                                         $colours[$pdbk] = 0;
3980                                         $this->mOutput->addLink( $title, 0 );
3981                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3982                                                                         $this->mLinkHolders['texts'][$key],
3983                                                                         $this->mLinkHolders['queries'][$key] );
3984                                 } elseif ( $colours[$pdbk] == 1 ) {
3985                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3986                                                                         $this->mLinkHolders['texts'][$key],
3987                                                                         $this->mLinkHolders['queries'][$key] );
3988                                 } elseif ( $colours[$pdbk] == 2 ) {
3989                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3990                                                                         $this->mLinkHolders['texts'][$key],
3991                                                                         $this->mLinkHolders['queries'][$key] );
3992                                 }
3993                         }
3994                         wfProfileOut( $fname.'-construct' );
3995
3996                         # Do the thing
3997                         wfProfileIn( $fname.'-replace' );
3998
3999                         $text = preg_replace_callback(
4000                                 '/(<!--LINK .*?-->)/',
4001                                 "wfOutputReplaceMatches",
4002                                 $text);
4003
4004                         wfProfileOut( $fname.'-replace' );
4005                 }
4006
4007                 # Now process interwiki link holders
4008                 # This is quite a bit simpler than internal links
4009                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4010                         wfProfileIn( $fname.'-interwiki' );
4011                         # Make interwiki link HTML
4012                         $wgOutputReplace = array();
4013                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4014                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4015                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4016                         }
4017
4018                         $text = preg_replace_callback(
4019                                 '/<!--IWLINK (.*?)-->/',
4020                                 "wfOutputReplaceMatches",
4021                                 $text );
4022                         wfProfileOut( $fname.'-interwiki' );
4023                 }
4024
4025                 wfProfileOut( $fname );
4026                 return $colours;
4027         }
4028
4029         /**
4030          * Replace <!--LINK--> link placeholders with plain text of links
4031          * (not HTML-formatted).
4032          * @param string $text
4033          * @return string
4034          */
4035         function replaceLinkHoldersText( $text ) {
4036                 $fname = 'Parser::replaceLinkHoldersText';
4037                 wfProfileIn( $fname );
4038
4039                 $text = preg_replace_callback(
4040                         '/<!--(LINK|IWLINK) (.*?)-->/',
4041                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4042                         $text );
4043
4044                 wfProfileOut( $fname );
4045                 return $text;
4046         }
4047
4048         /**
4049          * @param array $matches
4050          * @return string
4051          * @private
4052          */
4053         function replaceLinkHoldersTextCallback( $matches ) {
4054                 $type = $matches[1];
4055                 $key  = $matches[2];
4056                 if( $type == 'LINK' ) {
4057                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4058                                 return $this->mLinkHolders['texts'][$key];
4059                         }
4060                 } elseif( $type == 'IWLINK' ) {
4061                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4062                                 return $this->mInterwikiLinkHolders['texts'][$key];
4063                         }
4064                 }
4065                 return $matches[0];
4066         }
4067
4068         /**
4069          * Tag hook handler for 'pre'.
4070          */
4071         function renderPreTag( $text, $attribs, $parser ) {
4072                 // Backwards-compatibility hack
4073                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4074
4075                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4076                 return wfOpenElement( 'pre', $attribs ) .
4077                         wfEscapeHTMLTagsOnly( $content ) .
4078                         '</pre>';
4079         }
4080
4081         /**
4082          * Renders an image gallery from a text with one line per image.
4083          * text labels may be given by using |-style alternative text. E.g.
4084          *   Image:one.jpg|The number "1"
4085          *   Image:tree.jpg|A tree
4086          * given as text will return the HTML of a gallery with two images,
4087          * labeled 'The number "1"' and
4088          * 'A tree'.
4089          */
4090         function renderImageGallery( $text ) {
4091                 $ig = new ImageGallery();
4092                 $ig->setShowBytes( false );
4093                 $ig->setShowFilename( false );
4094                 $ig->setParsing();
4095                 $lines = explode( "\n", $text );
4096
4097                 foreach ( $lines as $line ) {
4098                         # match lines like these:
4099                         # Image:someimage.jpg|This is some image
4100                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4101                         # Skip empty lines
4102                         if ( count( $matches ) == 0 ) {
4103                                 continue;
4104                         }
4105                         $nt =& Title::newFromText( $matches[1] );
4106                         if( is_null( $nt ) ) {
4107                                 # Bogus title. Ignore these so we don't bomb out later.
4108                                 continue;
4109                         }
4110                         if ( isset( $matches[3] ) ) {
4111                                 $label = $matches[3];
4112                         } else {
4113                                 $label = '';
4114                         }
4115
4116                         $pout = $this->parse( $label,
4117                                 $this->mTitle,
4118                                 $this->mOptions,
4119                                 false, // Strip whitespace...?
4120                                 false  // Don't clear state!
4121                         );
4122                         $html = $pout->getText();
4123
4124                         $ig->add( new Image( $nt ), $html );
4125
4126                         # Only add real images (bug #5586)
4127                         if ( $nt->getNamespace() == NS_IMAGE ) {
4128                                 $this->mOutput->addImage( $nt->getDBkey() );
4129                         }
4130                 }
4131                 return $ig->toHTML();
4132         }
4133
4134         /**
4135          * Parse image options text and use it to make an image
4136          */
4137         function makeImage( &$nt, $options ) {
4138                 global $wgUseImageResize;
4139
4140                 $align = '';
4141
4142                 # Check if the options text is of the form "options|alt text"
4143                 # Options are:
4144                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4145                 #  * left               no resizing, just left align. label is used for alt= only
4146                 #  * right              same, but right aligned
4147                 #  * none               same, but not aligned
4148                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4149                 #  * center             center the image
4150                 #  * framed             Keep original image size, no magnify-button.
4151
4152                 $part = explode( '|', $options);
4153
4154                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4155                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4156                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4157                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4158                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4159                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4160                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4161                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4162                 $caption = '';
4163
4164                 $width = $height = $framed = $thumb = false;
4165                 $manual_thumb = '' ;
4166
4167                 foreach( $part as $key => $val ) {
4168                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4169                                 $thumb=true;
4170                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4171                                 # use manually specified thumbnail
4172                                 $thumb=true;
4173                                 $manual_thumb = $match;
4174                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4175                                 # remember to set an alignment, don't render immediately
4176                                 $align = 'right';
4177                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4178                                 # remember to set an alignment, don't render immediately
4179                                 $align = 'left';
4180                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4181                                 # remember to set an alignment, don't render immediately
4182                                 $align = 'center';
4183                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4184                                 # remember to set an alignment, don't render immediately
4185                                 $align = 'none';
4186                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4187                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4188                                 # $match is the image width in pixels
4189                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4190                                         $width = intval( $m[1] );
4191                                         $height = intval( $m[2] );
4192                                 } else {
4193                                         $width = intval($match);
4194                                 }
4195                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4196                                 $framed=true;
4197                         } else {
4198                                 $caption = $val;
4199                         }
4200                 }
4201                 # Strip bad stuff out of the alt text
4202                 $alt = $this->replaceLinkHoldersText( $caption );
4203
4204                 # make sure there are no placeholders in thumbnail attributes
4205                 # that are later expanded to html- so expand them now and
4206                 # remove the tags
4207                 $alt = $this->unstrip($alt, $this->mStripState);
4208                 $alt = Sanitizer::stripAllTags( $alt );
4209
4210                 # Linker does the rest
4211                 $sk =& $this->mOptions->getSkin();
4212                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4213         }
4214
4215         /**
4216          * Set a flag in the output object indicating that the content is dynamic and
4217          * shouldn't be cached.
4218          */
4219         function disableCache() {
4220                 wfDebug( "Parser output marked as uncacheable.\n" );
4221                 $this->mOutput->mCacheTime = -1;
4222         }
4223
4224         /**#@+
4225          * Callback from the Sanitizer for expanding items found in HTML attribute
4226          * values, so they can be safely tested and escaped.
4227          * @param string $text
4228          * @param array $args
4229          * @return string
4230          * @private
4231          */
4232         function attributeStripCallback( &$text, $args ) {
4233                 $text = $this->replaceVariables( $text, $args );
4234                 $text = $this->unstripForHTML( $text );
4235                 return $text;
4236         }
4237
4238         function unstripForHTML( $text ) {
4239                 $text = $this->unstrip( $text, $this->mStripState );
4240                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4241                 return $text;
4242         }
4243         /**#@-*/
4244
4245         /**#@+
4246          * Accessor/mutator
4247          */
4248         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4249         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4250         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4251         /**#@-*/
4252
4253         /**#@+
4254          * Accessor
4255          */
4256         function getTags() { return array_keys( $this->mTagHooks ); }
4257         /**#@-*/
4258
4259
4260         /**
4261          * Break wikitext input into sections, and either pull or replace
4262          * some particular section's text.
4263          *
4264          * External callers should use the getSection and replaceSection methods.
4265          *
4266          * @param $text Page wikitext
4267          * @param $section Numbered section. 0 pulls the text before the first
4268          *                 heading; other numbers will pull the given section
4269          *                 along with its lower-level subsections.
4270          * @param $mode One of "get" or "replace"
4271          * @param $newtext Replacement text for section data.
4272          * @return string for "get", the extracted section text.
4273          *                for "replace", the whole page with the section replaced.
4274          */
4275         private function extractSections( $text, $section, $mode, $newtext='' ) {
4276                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4277                 # comments to be stripped as well)
4278                 $striparray = array();
4279
4280                 $oldOutputType = $this->mOutputType;
4281                 $oldOptions = $this->mOptions;
4282                 $this->mOptions = new ParserOptions();
4283                 $this->mOutputType = OT_WIKI;
4284
4285                 $striptext = $this->strip( $text, $striparray, true );
4286
4287                 $this->mOutputType = $oldOutputType;
4288                 $this->mOptions = $oldOptions;
4289
4290                 # now that we can be sure that no pseudo-sections are in the source,
4291                 # split it up by section
4292                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4293                 $comment = "(?:$uniq-!--.*?QINU)";
4294                 $secs = preg_split(
4295                 /*
4296                         "/
4297                         ^(
4298                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4299                         (?:
4300                                 (=+) # Should this be limited to 6?
4301                                 .+?  # Section title...
4302                                 \\2  # Ending = count must match start
4303                         |
4304                                 ^
4305                                 <h([1-6])\b.*?>
4306                                 .*?
4307                                 <\/h\\3\s*>
4308                         )
4309                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4310                         )$
4311                         /mix",
4312                 */
4313                         "/
4314                         (
4315                                 ^
4316                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4317                                 (=+) # Should this be limited to 6?
4318                                 .+?  # Section title...
4319                                 \\2  # Ending = count must match start
4320                                 (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4321                                 $
4322                         |
4323                                 <h([1-6])\b.*?>
4324                                 .*?
4325                                 <\/h\\3\s*>
4326                         )
4327                         /mix",
4328                         $striptext, -1,
4329                         PREG_SPLIT_DELIM_CAPTURE);
4330
4331                 if( $mode == "get" ) {
4332                         if( $section == 0 ) {
4333                                 // "Section 0" returns the content before any other section.
4334                                 $rv = $secs[0];
4335                         } else {
4336                                 $rv = "";
4337                         }
4338                 } elseif( $mode == "replace" ) {
4339                         if( $section == 0 ) {
4340                                 $rv = $newtext . "\n\n";
4341                                 $remainder = true;
4342                         } else {
4343                                 $rv = $secs[0];
4344                                 $remainder = false;
4345                         }
4346                 }
4347                 $count = 0;
4348                 $sectionLevel = 0;
4349                 for( $index = 1; $index < count( $secs ); ) {
4350                         $headerLine = $secs[$index++];
4351                         if( $secs[$index] ) {
4352                                 // A wiki header
4353                                 $headerLevel = strlen( $secs[$index++] );
4354                         } else {
4355                                 // An HTML header
4356                                 $index++;
4357                                 $headerLevel = intval( $secs[$index++] );
4358                         }
4359                         $content = $secs[$index++];
4360
4361                         $count++;
4362                         if( $mode == "get" ) {
4363                                 if( $count == $section ) {
4364                                         $rv = $headerLine . $content;
4365                                         $sectionLevel = $headerLevel;
4366                                 } elseif( $count > $section ) {
4367                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4368                                                 $rv .= $headerLine . $content;
4369                                         } else {
4370                                                 // Broke out to a higher-level section
4371                                                 break;
4372                                         }
4373                                 }
4374                         } elseif( $mode == "replace" ) {
4375                                 if( $count < $section ) {
4376                                         $rv .= $headerLine . $content;
4377                                 } elseif( $count == $section ) {
4378                                         $rv .= $newtext . "\n\n";
4379                                         $sectionLevel = $headerLevel;
4380                                 } elseif( $count > $section ) {
4381                                         if( $headerLevel <= $sectionLevel ) {
4382                                                 // Passed the section's sub-parts.
4383                                                 $remainder = true;
4384                                         }
4385                                         if( $remainder ) {
4386                                                 $rv .= $headerLine . $content;
4387                                         }
4388                                 }
4389                         }
4390                 }
4391                 # reinsert stripped tags
4392                 $rv = $this->unstrip( $rv, $striparray );
4393                 $rv = $this->unstripNoWiki( $rv, $striparray );
4394                 $rv = trim( $rv );
4395                 return $rv;
4396         }
4397
4398         /**
4399          * This function returns the text of a section, specified by a number ($section).
4400          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4401          * the first section before any such heading (section 0).
4402          *
4403          * If a section contains subsections, these are also returned.
4404          *
4405          * @param $text String: text to look in
4406          * @param $section Integer: section number
4407          * @return string text of the requested section
4408          */
4409         function getSection( $text, $section ) {
4410                 return $this->extractSections( $text, $section, "get" );
4411         }
4412
4413         function replaceSection( $oldtext, $section, $text ) {
4414                 return $this->extractSections( $oldtext, $section, "replace", $text );
4415         }
4416
4417 }
4418
4419 /**
4420  * @todo document
4421  * @package MediaWiki
4422  */
4423 class ParserOutput
4424 {
4425         var $mText,             # The output text
4426                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4427                 $mCategories,       # Map of category names to sort keys
4428                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4429                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4430                 $mVersion,          # Compatibility check
4431                 $mTitleText,        # title text of the chosen language variant
4432                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4433                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4434                 $mImages,           # DB keys of the images used, in the array key only
4435                 $mExternalLinks,    # External link URLs, in the key only
4436                 $mHTMLtitle,            # Display HTML title
4437                 $mSubtitle,                     # Additional subtitle
4438                 $mNewSection,           # Show a new section link?
4439                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4440
4441         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4442                 $containsOldMagic = false, $titletext = '' )
4443         {
4444                 $this->mText = $text;
4445                 $this->mLanguageLinks = $languageLinks;
4446                 $this->mCategories = $categoryLinks;
4447                 $this->mContainsOldMagic = $containsOldMagic;
4448                 $this->mCacheTime = '';
4449                 $this->mVersion = MW_PARSER_VERSION;
4450                 $this->mTitleText = $titletext;
4451                 $this->mLinks = array();
4452                 $this->mTemplates = array();
4453                 $this->mImages = array();
4454                 $this->mExternalLinks = array();
4455                 $this->mHTMLtitle = "" ;
4456                 $this->mSubtitle = "" ;
4457                 $this->mNewSection = false;
4458                 $this->mNoGallery = false;
4459         }
4460
4461         function getText()                   { return $this->mText; }
4462         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4463         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4464         function &getCategories()            { return $this->mCategories; }
4465         function getCacheTime()              { return $this->mCacheTime; }
4466         function getTitleText()              { return $this->mTitleText; }
4467         function &getLinks()                 { return $this->mLinks; }
4468         function &getTemplates()             { return $this->mTemplates; }
4469         function &getImages()                { return $this->mImages; }
4470         function &getExternalLinks()         { return $this->mExternalLinks; }
4471         function getNoGallery()              { return $this->mNoGallery; }
4472
4473         function containsOldMagic()          { return $this->mContainsOldMagic; }
4474         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4475         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4476         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4477         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4478         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4479         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4480
4481         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4482         function addImage( $name )           { $this->mImages[$name] = 1; }
4483         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4484         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4485
4486         function setNewSection( $value ) {
4487                 $this->mNewSection = (bool)$value;
4488         }
4489         function getNewSection() {
4490                 return (bool)$this->mNewSection;
4491         }
4492
4493         function addLink( $title, $id ) {
4494                 $ns = $title->getNamespace();
4495                 $dbk = $title->getDBkey();
4496                 if ( !isset( $this->mLinks[$ns] ) ) {
4497                         $this->mLinks[$ns] = array();
4498                 }
4499                 $this->mLinks[$ns][$dbk] = $id;
4500         }
4501
4502         function addTemplate( $title, $id ) {
4503                 $ns = $title->getNamespace();
4504                 $dbk = $title->getDBkey();
4505                 if ( !isset( $this->mTemplates[$ns] ) ) {
4506                         $this->mTemplates[$ns] = array();
4507                 }
4508                 $this->mTemplates[$ns][$dbk] = $id;
4509         }
4510
4511         /**
4512          * Return true if this cached output object predates the global or
4513          * per-article cache invalidation timestamps, or if it comes from
4514          * an incompatible older version.
4515          *
4516          * @param string $touched the affected article's last touched timestamp
4517          * @return bool
4518          * @public
4519          */
4520         function expired( $touched ) {
4521                 global $wgCacheEpoch;
4522                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4523                        $this->getCacheTime() < $touched ||
4524                        $this->getCacheTime() <= $wgCacheEpoch ||
4525                        !isset( $this->mVersion ) ||
4526                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4527         }
4528 }
4529
4530 /**
4531  * Set options of the Parser
4532  * @todo document
4533  * @package MediaWiki
4534  */
4535 class ParserOptions
4536 {
4537         # All variables are private
4538         var $mUseTeX;                    # Use texvc to expand <math> tags
4539         var $mUseDynamicDates;           # Use DateFormatter to format dates
4540         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4541         var $mAllowExternalImages;       # Allow external images inline
4542         var $mAllowExternalImagesFrom;   # If not, any exception?
4543         var $mSkin;                      # Reference to the preferred skin
4544         var $mDateFormat;                # Date format index
4545         var $mEditSection;               # Create "edit section" links
4546         var $mNumberHeadings;            # Automatically number headings
4547         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4548         var $mTidy;                      # Ask for tidy cleanup
4549         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4550
4551         function getUseTeX()                        { return $this->mUseTeX; }
4552         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4553         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4554         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4555         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4556         function &getSkin()                         { return $this->mSkin; }
4557         function getDateFormat()                    { return $this->mDateFormat; }
4558         function getEditSection()                   { return $this->mEditSection; }
4559         function getNumberHeadings()                { return $this->mNumberHeadings; }
4560         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4561         function getTidy()                          { return $this->mTidy; }
4562         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4563
4564         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4565         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4566         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4567         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4568         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4569         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4570         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4571         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4572         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4573         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4574         function setSkin( &$x ) { $this->mSkin =& $x; }
4575         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4576
4577         function ParserOptions() {
4578                 global $wgUser;
4579                 $this->initialiseFromUser( $wgUser );
4580         }
4581
4582         /**
4583          * Get parser options
4584          * @static
4585          */
4586         function newFromUser( &$user ) {
4587                 $popts = new ParserOptions;
4588                 $popts->initialiseFromUser( $user );
4589                 return $popts;
4590         }
4591
4592         /** Get user options */
4593         function initialiseFromUser( &$userInput ) {
4594                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4595                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4596                 $fname = 'ParserOptions::initialiseFromUser';
4597                 wfProfileIn( $fname );
4598                 if ( !$userInput ) {
4599                         $user = new User;
4600                         $user->setLoaded( true );
4601                 } else {
4602                         $user =& $userInput;
4603                 }
4604
4605                 $this->mUseTeX = $wgUseTeX;
4606                 $this->mUseDynamicDates = $wgUseDynamicDates;
4607                 $this->mInterwikiMagic = $wgInterwikiMagic;
4608                 $this->mAllowExternalImages = $wgAllowExternalImages;
4609                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4610                 wfProfileIn( $fname.'-skin' );
4611                 $this->mSkin =& $user->getSkin();
4612                 wfProfileOut( $fname.'-skin' );
4613                 $this->mDateFormat = $user->getOption( 'date' );
4614                 $this->mEditSection = true;
4615                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4616                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4617                 $this->mTidy = false;
4618                 $this->mInterfaceMessage = false;
4619                 wfProfileOut( $fname );
4620         }
4621 }
4622
4623 /**
4624  * Callback function used by Parser::replaceLinkHolders()
4625  * to substitute link placeholders.
4626  */
4627 function &wfOutputReplaceMatches( $matches ) {
4628         global $wgOutputReplace;
4629         return $wgOutputReplace[$matches[1]];
4630 }
4631
4632 /**
4633  * Return the total number of articles
4634  */
4635 function wfNumberOfArticles() {
4636         global $wgNumberOfArticles;
4637
4638         wfLoadSiteStats();
4639         return $wgNumberOfArticles;
4640 }
4641
4642 /**
4643  * Return the number of files
4644  */
4645 function wfNumberOfFiles() {
4646         $fname = 'wfNumberOfFiles';
4647
4648         wfProfileIn( $fname );
4649         $dbr =& wfGetDB( DB_SLAVE );
4650         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4651         wfProfileOut( $fname );
4652
4653         return $numImages;
4654 }
4655
4656 /**
4657  * Return the number of user accounts
4658  * @return integer
4659  */
4660 function wfNumberOfUsers() {
4661         wfProfileIn( 'wfNumberOfUsers' );
4662         $dbr =& wfGetDB( DB_SLAVE );
4663         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4664         wfProfileOut( 'wfNumberOfUsers' );
4665         return (int)$count;
4666 }
4667
4668 /**
4669  * Return the total number of pages
4670  * @return integer
4671  */
4672 function wfNumberOfPages() {
4673         wfProfileIn( 'wfNumberOfPages' );
4674         $dbr =& wfGetDB( DB_SLAVE );
4675         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4676         wfProfileOut( 'wfNumberOfPages' );
4677         return (int)$count;
4678 }
4679
4680 /**
4681  * Count the number of pages in a particular namespace
4682  *
4683  * @param $ns Namespace
4684  * @return integer
4685  */
4686 function wfPagesInNs( $ns ) {
4687         static $pageCount = array();
4688         wfProfileIn( 'wfPagesInNs' );
4689         if( !isset( $pageCount[$ns] ) ) {
4690                 $dbr =& wfGetDB( DB_SLAVE );
4691                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4692         }
4693         wfProfileOut( 'wfPagesInNs' );
4694         return (int)$pageCount[$ns];
4695 }
4696
4697 /**
4698  * Get various statistics from the database
4699  * @private
4700  */
4701 function wfLoadSiteStats() {
4702         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4703         $fname = 'wfLoadSiteStats';
4704
4705         if ( -1 != $wgNumberOfArticles ) return;
4706         $dbr =& wfGetDB( DB_SLAVE );
4707         $s = $dbr->selectRow( 'site_stats',
4708                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4709                 array( 'ss_row_id' => 1 ), $fname
4710         );
4711
4712         if ( $s === false ) {
4713                 return;
4714         } else {
4715                 $wgTotalViews = $s->ss_total_views;
4716                 $wgTotalEdits = $s->ss_total_edits;
4717                 $wgNumberOfArticles = $s->ss_good_articles;
4718         }
4719 }
4720
4721 /**
4722  * Escape html tags
4723  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4724  *
4725  * @param $in String: text that might contain HTML tags.
4726  * @return string Escaped string
4727  */
4728 function wfEscapeHTMLTagsOnly( $in ) {
4729         return str_replace(
4730                 array( '"', '>', '<' ),
4731                 array( '&quot;', '&gt;', '&lt;' ),
4732                 $in );
4733 }
4734
4735 ?>