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                         # Escape any control characters introduced by the above step
1150                         $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1151
1152                         # Process the trail (i.e. everything after this link up until start of the next link),
1153                         # replacing any non-bracketed links
1154                         $trail = $this->replaceFreeExternalLinks( $trail );
1155
1156                         # Use the encoded URL
1157                         # This means that users can paste URLs directly into the text
1158                         # Funny characters like &ouml; aren't valid in URLs anyway
1159                         # This was changed in August 2004
1160                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1161
1162                         # Register link in the output object.
1163                         # Replace unnecessary URL escape codes with the referenced character
1164                         # This prevents spammers from hiding links from the filters
1165                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1166                         $this->mOutput->addExternalLink( $pasteurized );
1167                 }
1168
1169                 wfProfileOut( $fname );
1170                 return $s;
1171         }
1172
1173         /**
1174          * Replace anything that looks like a URL with a link
1175          * @private
1176          */
1177         function replaceFreeExternalLinks( $text ) {
1178                 global $wgContLang;
1179                 $fname = 'Parser::replaceFreeExternalLinks';
1180                 wfProfileIn( $fname );
1181
1182                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1183                 $s = array_shift( $bits );
1184                 $i = 0;
1185
1186                 $sk =& $this->mOptions->getSkin();
1187
1188                 while ( $i < count( $bits ) ){
1189                         $protocol = $bits[$i++];
1190                         $remainder = $bits[$i++];
1191
1192                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1193                                 # Found some characters after the protocol that look promising
1194                                 $url = $protocol . $m[1];
1195                                 $trail = $m[2];
1196
1197                                 # special case: handle urls as url args:
1198                                 # http://www.example.com/foo?=http://www.example.com/bar
1199                                 if(strlen($trail) == 0 &&
1200                                         isset($bits[$i]) &&
1201                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1202                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1203                                 {
1204                                         # add protocol, arg
1205                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1206                                         $i += 2;
1207                                         $trail = $m[2];
1208                                 }
1209
1210                                 # The characters '<' and '>' (which were escaped by
1211                                 # removeHTMLtags()) should not be included in
1212                                 # URLs, per RFC 2396.
1213                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1214                                         $trail = substr($url, $m2[0][1]) . $trail;
1215                                         $url = substr($url, 0, $m2[0][1]);
1216                                 }
1217
1218                                 # Move trailing punctuation to $trail
1219                                 $sep = ',;\.:!?';
1220                                 # If there is no left bracket, then consider right brackets fair game too
1221                                 if ( strpos( $url, '(' ) === false ) {
1222                                         $sep .= ')';
1223                                 }
1224
1225                                 $numSepChars = strspn( strrev( $url ), $sep );
1226                                 if ( $numSepChars ) {
1227                                         $trail = substr( $url, -$numSepChars ) . $trail;
1228                                         $url = substr( $url, 0, -$numSepChars );
1229                                 }
1230
1231                                 # Normalize any HTML entities in input. They will be
1232                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1233                                 $url = Sanitizer::decodeCharReferences( $url );
1234
1235                                 # Escape any control characters introduced by the above step
1236                                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1237
1238                                 # Is this an external image?
1239                                 $text = $this->maybeMakeExternalImage( $url );
1240                                 if ( $text === false ) {
1241                                         # Not an image, make a link
1242                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1243                                         # Register it in the output object...
1244                                         # Replace unnecessary URL escape codes with their equivalent characters
1245                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1246                                         $this->mOutput->addExternalLink( $pasteurized );
1247                                 }
1248                                 $s .= $text . $trail;
1249                         } else {
1250                                 $s .= $protocol . $remainder;
1251                         }
1252                 }
1253                 wfProfileOut( $fname );
1254                 return $s;
1255         }
1256
1257         /**
1258          * Replace unusual URL escape codes with their equivalent characters
1259          * @param string
1260          * @return string
1261          * @static
1262          * @fixme This can merge genuinely required bits in the path or query string,
1263          *        breaking legit URLs. A proper fix would treat the various parts of
1264          *        the URL differently; as a workaround, just use the output for
1265          *        statistical records, not for actual linking/output.
1266          */
1267         function replaceUnusualEscapes( $url ) {
1268                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1269                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1270         }
1271
1272         /**
1273          * Callback function used in replaceUnusualEscapes().
1274          * Replaces unusual URL escape codes with their equivalent character
1275          * @static
1276          * @private
1277          */
1278         function replaceUnusualEscapesCallback( $matches ) {
1279                 $char = urldecode( $matches[0] );
1280                 $ord = ord( $char );
1281                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1282                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1283                         // No, shouldn't be escaped
1284                         return $char;
1285                 } else {
1286                         // Yes, leave it escaped
1287                         return $matches[0];
1288                 }
1289         }
1290
1291         /**
1292          * make an image if it's allowed, either through the global
1293          * option or through the exception
1294          * @private
1295          */
1296         function maybeMakeExternalImage( $url ) {
1297                 $sk =& $this->mOptions->getSkin();
1298                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1299                 $imagesexception = !empty($imagesfrom);
1300                 $text = false;
1301                 if ( $this->mOptions->getAllowExternalImages()
1302                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1303                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1304                                 # Image found
1305                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1306                         }
1307                 }
1308                 return $text;
1309         }
1310
1311         /**
1312          * Process [[ ]] wikilinks
1313          *
1314          * @private
1315          */
1316         function replaceInternalLinks( $s ) {
1317                 global $wgContLang;
1318                 static $fname = 'Parser::replaceInternalLinks' ;
1319
1320                 wfProfileIn( $fname );
1321
1322                 wfProfileIn( $fname.'-setup' );
1323                 static $tc = FALSE;
1324                 # the % is needed to support urlencoded titles as well
1325                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1326
1327                 $sk =& $this->mOptions->getSkin();
1328
1329                 #split the entire text string on occurences of [[
1330                 $a = explode( '[[', ' ' . $s );
1331                 #get the first element (all text up to first [[), and remove the space we added
1332                 $s = array_shift( $a );
1333                 $s = substr( $s, 1 );
1334
1335                 # Match a link having the form [[namespace:link|alternate]]trail
1336                 static $e1 = FALSE;
1337                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1338                 # Match cases where there is no "]]", which might still be images
1339                 static $e1_img = FALSE;
1340                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1341                 # Match the end of a line for a word that's not followed by whitespace,
1342                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1343                 $e2 = wfMsgForContent( 'linkprefix' );
1344
1345                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1346
1347                 if( is_null( $this->mTitle ) ) {
1348                         throw new MWException( 'nooo' );
1349                 }
1350                 $nottalk = !$this->mTitle->isTalkPage();
1351
1352                 if ( $useLinkPrefixExtension ) {
1353                         if ( preg_match( $e2, $s, $m ) ) {
1354                                 $first_prefix = $m[2];
1355                         } else {
1356                                 $first_prefix = false;
1357                         }
1358                 } else {
1359                         $prefix = '';
1360                 }
1361
1362                 $selflink = $this->mTitle->getPrefixedText();
1363                 wfProfileOut( $fname.'-setup' );
1364
1365                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1366                 $useSubpages = $this->areSubpagesAllowed();
1367
1368                 # Loop for each link
1369                 for ($k = 0; isset( $a[$k] ); $k++) {
1370                         $line = $a[$k];
1371                         if ( $useLinkPrefixExtension ) {
1372                                 wfProfileIn( $fname.'-prefixhandling' );
1373                                 if ( preg_match( $e2, $s, $m ) ) {
1374                                         $prefix = $m[2];
1375                                         $s = $m[1];
1376                                 } else {
1377                                         $prefix='';
1378                                 }
1379                                 # first link
1380                                 if($first_prefix) {
1381                                         $prefix = $first_prefix;
1382                                         $first_prefix = false;
1383                                 }
1384                                 wfProfileOut( $fname.'-prefixhandling' );
1385                         }
1386
1387                         $might_be_img = false;
1388
1389                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1390                                 $text = $m[2];
1391                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1392                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1393                                 # the real problem is with the $e1 regex
1394                                 # See bug 1300.
1395                                 #
1396                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1397                                 # and no image is in sight. See bug 2095.
1398                                 #
1399                                 if( $text !== '' &&
1400                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1401                                         strpos($text, '[') !== false
1402                                 )
1403                                 {
1404                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1405                                         $m[3] = $n[1];
1406                                 }
1407                                 # fix up urlencoded title texts
1408                                 if(preg_match('/%/', $m[1] ))
1409                                         # Should anchors '#' also be rejected?
1410                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1411                                 $trail = $m[3];
1412                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1413                                 $might_be_img = true;
1414                                 $text = $m[2];
1415                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1416                                 $trail = "";
1417                         } else { # Invalid form; output directly
1418                                 $s .= $prefix . '[[' . $line ;
1419                                 continue;
1420                         }
1421
1422                         # Don't allow internal links to pages containing
1423                         # PROTO: where PROTO is a valid URL protocol; these
1424                         # should be external links.
1425                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1426                                 $s .= $prefix . '[[' . $line ;
1427                                 continue;
1428                         }
1429
1430                         # Make subpage if necessary
1431                         if( $useSubpages ) {
1432                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1433                         } else {
1434                                 $link = $m[1];
1435                         }
1436
1437                         $noforce = (substr($m[1], 0, 1) != ':');
1438                         if (!$noforce) {
1439                                 # Strip off leading ':'
1440                                 $link = substr($link, 1);
1441                         }
1442
1443                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1444                         if( !$nt ) {
1445                                 $s .= $prefix . '[[' . $line;
1446                                 continue;
1447                         }
1448
1449                         #check other language variants of the link
1450                         #if the article does not exist
1451                         if( $checkVariantLink
1452                             && $nt->getArticleID() == 0 ) {
1453                                 $wgContLang->findVariantLink($link, $nt);
1454                         }
1455
1456                         $ns = $nt->getNamespace();
1457                         $iw = $nt->getInterWiki();
1458
1459                         if ($might_be_img) { # if this is actually an invalid link
1460                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1461                                         $found = false;
1462                                         while (isset ($a[$k+1]) ) {
1463                                                 #look at the next 'line' to see if we can close it there
1464                                                 $spliced = array_splice( $a, $k + 1, 1 );
1465                                                 $next_line = array_shift( $spliced );
1466                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1467                                                 # the first ]] closes the inner link, the second the image
1468                                                         $found = true;
1469                                                         $text .= '[[' . $m[1];
1470                                                         $trail = $m[2];
1471                                                         break;
1472                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1473                                                         #if there's exactly one ]] that's fine, we'll keep looking
1474                                                         $text .= '[[' . $m[0];
1475                                                 } else {
1476                                                         #if $next_line is invalid too, we need look no further
1477                                                         $text .= '[[' . $next_line;
1478                                                         break;
1479                                                 }
1480                                         }
1481                                         if ( !$found ) {
1482                                                 # we couldn't find the end of this imageLink, so output it raw
1483                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1484                                                 $text = $this->replaceInternalLinks($text);
1485                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1486                                                 # note: no $trail, because without an end, there *is* no trail
1487                                                 continue;
1488                                         }
1489                                 } else { #it's not an image, so output it raw
1490                                         $s .= $prefix . '[[' . $link . '|' . $text;
1491                                         # note: no $trail, because without an end, there *is* no trail
1492                                         continue;
1493                                 }
1494                         }
1495
1496                         $wasblank = ( '' == $text );
1497                         if( $wasblank ) $text = $link;
1498
1499
1500                         # Link not escaped by : , create the various objects
1501                         if( $noforce ) {
1502
1503                                 # Interwikis
1504                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1505                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1506                                         $s = rtrim($s . "\n");
1507                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1508                                         continue;
1509                                 }
1510
1511                                 if ( $ns == NS_IMAGE ) {
1512                                         wfProfileIn( "$fname-image" );
1513                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1514                                                 # recursively parse links inside the image caption
1515                                                 # actually, this will parse them in any other parameters, too,
1516                                                 # but it might be hard to fix that, and it doesn't matter ATM
1517                                                 $text = $this->replaceExternalLinks($text);
1518                                                 $text = $this->replaceInternalLinks($text);
1519
1520                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1521                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1522                                                 $this->mOutput->addImage( $nt->getDBkey() );
1523
1524                                                 wfProfileOut( "$fname-image" );
1525                                                 continue;
1526                                         }
1527                                         wfProfileOut( "$fname-image" );
1528
1529                                 }
1530
1531                                 if ( $ns == NS_CATEGORY ) {
1532                                         wfProfileIn( "$fname-category" );
1533                                         $s = rtrim($s . "\n"); # bug 87
1534
1535                                         if ( $wasblank ) {
1536                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1537                                                         $sortkey = $this->mTitle->getText();
1538                                                 } else {
1539                                                         $sortkey = $this->mTitle->getPrefixedText();
1540                                                 }
1541                                         } else {
1542                                                 $sortkey = $text;
1543                                         }
1544                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1545                                         $sortkey = str_replace( "\n", '', $sortkey );
1546                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1547                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1548
1549                                         /**
1550                                          * Strip the whitespace Category links produce, see bug 87
1551                                          * @todo We might want to use trim($tmp, "\n") here.
1552                                          */
1553                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1554
1555                                         wfProfileOut( "$fname-category" );
1556                                         continue;
1557                                 }
1558                         }
1559
1560                         if( ( $nt->getPrefixedText() === $selflink ) &&
1561                             ( $nt->getFragment() === '' ) ) {
1562                                 # Self-links are handled specially; generally de-link and change to bold.
1563                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1564                                 continue;
1565                         }
1566
1567                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1568                         if( $ns == NS_MEDIA ) {
1569                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1570                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1571                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1572                                 $this->mOutput->addImage( $nt->getDBkey() );
1573                                 continue;
1574                         } elseif( $ns == NS_SPECIAL ) {
1575                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1576                                 continue;
1577                         } elseif( $ns == NS_IMAGE ) {
1578                                 $img = Image::newFromTitle( $nt );
1579                                 if( $img->exists() ) {
1580                                         // Force a blue link if the file exists; may be a remote
1581                                         // upload on the shared repository, and we want to see its
1582                                         // auto-generated page.
1583                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1584                                         continue;
1585                                 }
1586                         }
1587                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1588                 }
1589                 wfProfileOut( $fname );
1590                 return $s;
1591         }
1592
1593         /**
1594          * Make a link placeholder. The text returned can be later resolved to a real link with
1595          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1596          * parsing of interwiki links, and secondly to allow all extistence checks and
1597          * article length checks (for stub links) to be bundled into a single query.
1598          *
1599          */
1600         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1601                 if ( ! is_object($nt) ) {
1602                         # Fail gracefully
1603                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1604                 } else {
1605                         # Separate the link trail from the rest of the link
1606                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1607
1608                         if ( $nt->isExternal() ) {
1609                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1610                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1611                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1612                         } else {
1613                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1614                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1615                                 $this->mLinkHolders['queries'][] = $query;
1616                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1617                                 $this->mLinkHolders['titles'][] = $nt;
1618
1619                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1620                         }
1621                 }
1622                 return $retVal;
1623         }
1624
1625         /**
1626          * Render a forced-blue link inline; protect against double expansion of
1627          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1628          * Since this little disaster has to split off the trail text to avoid
1629          * breaking URLs in the following text without breaking trails on the
1630          * wiki links, it's been made into a horrible function.
1631          *
1632          * @param Title $nt
1633          * @param string $text
1634          * @param string $query
1635          * @param string $trail
1636          * @param string $prefix
1637          * @return string HTML-wikitext mix oh yuck
1638          */
1639         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1640                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1641                 $sk =& $this->mOptions->getSkin();
1642                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1643                 return $this->armorLinks( $link ) . $trail;
1644         }
1645
1646         /**
1647          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1648          * going to go through further parsing steps before inline URL expansion.
1649          *
1650          * In particular this is important when using action=render, which causes
1651          * full URLs to be included.
1652          *
1653          * Oh man I hate our multi-layer parser!
1654          *
1655          * @param string more-or-less HTML
1656          * @return string less-or-more HTML with NOPARSE bits
1657          */
1658         function armorLinks( $text ) {
1659                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1660                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1661         }
1662
1663         /**
1664          * Return true if subpage links should be expanded on this page.
1665          * @return bool
1666          */
1667         function areSubpagesAllowed() {
1668                 # Some namespaces don't allow subpages
1669                 global $wgNamespacesWithSubpages;
1670                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1671         }
1672
1673         /**
1674          * Handle link to subpage if necessary
1675          * @param string $target the source of the link
1676          * @param string &$text the link text, modified as necessary
1677          * @return string the full name of the link
1678          * @private
1679          */
1680         function maybeDoSubpageLink($target, &$text) {
1681                 # Valid link forms:
1682                 # Foobar -- normal
1683                 # :Foobar -- override special treatment of prefix (images, language links)
1684                 # /Foobar -- convert to CurrentPage/Foobar
1685                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1686                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1687                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1688
1689                 $fname = 'Parser::maybeDoSubpageLink';
1690                 wfProfileIn( $fname );
1691                 $ret = $target; # default return value is no change
1692
1693                 # Some namespaces don't allow subpages,
1694                 # so only perform processing if subpages are allowed
1695                 if( $this->areSubpagesAllowed() ) {
1696                         # Look at the first character
1697                         if( $target != '' && $target{0} == '/' ) {
1698                                 # / at end means we don't want the slash to be shown
1699                                 if( substr( $target, -1, 1 ) == '/' ) {
1700                                         $target = substr( $target, 1, -1 );
1701                                         $noslash = $target;
1702                                 } else {
1703                                         $noslash = substr( $target, 1 );
1704                                 }
1705
1706                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1707                                 if( '' === $text ) {
1708                                         $text = $target;
1709                                 } # this might be changed for ugliness reasons
1710                         } else {
1711                                 # check for .. subpage backlinks
1712                                 $dotdotcount = 0;
1713                                 $nodotdot = $target;
1714                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1715                                         ++$dotdotcount;
1716                                         $nodotdot = substr( $nodotdot, 3 );
1717                                 }
1718                                 if($dotdotcount > 0) {
1719                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1720                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1721                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1722                                                 # / at the end means don't show full path
1723                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1724                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1725                                                         if( '' === $text ) {
1726                                                                 $text = $nodotdot;
1727                                                         }
1728                                                 }
1729                                                 $nodotdot = trim( $nodotdot );
1730                                                 if( $nodotdot != '' ) {
1731                                                         $ret .= '/' . $nodotdot;
1732                                                 }
1733                                         }
1734                                 }
1735                         }
1736                 }
1737
1738                 wfProfileOut( $fname );
1739                 return $ret;
1740         }
1741
1742         /**#@+
1743          * Used by doBlockLevels()
1744          * @private
1745          */
1746         /* private */ function closeParagraph() {
1747                 $result = '';
1748                 if ( '' != $this->mLastSection ) {
1749                         $result = '</' . $this->mLastSection  . ">\n";
1750                 }
1751                 $this->mInPre = false;
1752                 $this->mLastSection = '';
1753                 return $result;
1754         }
1755         # getCommon() returns the length of the longest common substring
1756         # of both arguments, starting at the beginning of both.
1757         #
1758         /* private */ function getCommon( $st1, $st2 ) {
1759                 $fl = strlen( $st1 );
1760                 $shorter = strlen( $st2 );
1761                 if ( $fl < $shorter ) { $shorter = $fl; }
1762
1763                 for ( $i = 0; $i < $shorter; ++$i ) {
1764                         if ( $st1{$i} != $st2{$i} ) { break; }
1765                 }
1766                 return $i;
1767         }
1768         # These next three functions open, continue, and close the list
1769         # element appropriate to the prefix character passed into them.
1770         #
1771         /* private */ function openList( $char ) {
1772                 $result = $this->closeParagraph();
1773
1774                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1775                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1776                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1777                 else if ( ';' == $char ) {
1778                         $result .= '<dl><dt>';
1779                         $this->mDTopen = true;
1780                 }
1781                 else { $result = '<!-- ERR 1 -->'; }
1782
1783                 return $result;
1784         }
1785
1786         /* private */ function nextItem( $char ) {
1787                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1788                 else if ( ':' == $char || ';' == $char ) {
1789                         $close = '</dd>';
1790                         if ( $this->mDTopen ) { $close = '</dt>'; }
1791                         if ( ';' == $char ) {
1792                                 $this->mDTopen = true;
1793                                 return $close . '<dt>';
1794                         } else {
1795                                 $this->mDTopen = false;
1796                                 return $close . '<dd>';
1797                         }
1798                 }
1799                 return '<!-- ERR 2 -->';
1800         }
1801
1802         /* private */ function closeList( $char ) {
1803                 if ( '*' == $char ) { $text = '</li></ul>'; }
1804                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1805                 else if ( ':' == $char ) {
1806                         if ( $this->mDTopen ) {
1807                                 $this->mDTopen = false;
1808                                 $text = '</dt></dl>';
1809                         } else {
1810                                 $text = '</dd></dl>';
1811                         }
1812                 }
1813                 else {  return '<!-- ERR 3 -->'; }
1814                 return $text."\n";
1815         }
1816         /**#@-*/
1817
1818         /**
1819          * Make lists from lines starting with ':', '*', '#', etc.
1820          *
1821          * @private
1822          * @return string the lists rendered as HTML
1823          */
1824         function doBlockLevels( $text, $linestart ) {
1825                 $fname = 'Parser::doBlockLevels';
1826                 wfProfileIn( $fname );
1827
1828                 # Parsing through the text line by line.  The main thing
1829                 # happening here is handling of block-level elements p, pre,
1830                 # and making lists from lines starting with * # : etc.
1831                 #
1832                 $textLines = explode( "\n", $text );
1833
1834                 $lastPrefix = $output = '';
1835                 $this->mDTopen = $inBlockElem = false;
1836                 $prefixLength = 0;
1837                 $paragraphStack = false;
1838
1839                 if ( !$linestart ) {
1840                         $output .= array_shift( $textLines );
1841                 }
1842                 foreach ( $textLines as $oLine ) {
1843                         $lastPrefixLength = strlen( $lastPrefix );
1844                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1845                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1846                         if ( !$this->mInPre ) {
1847                                 # Multiple prefixes may abut each other for nested lists.
1848                                 $prefixLength = strspn( $oLine, '*#:;' );
1849                                 $pref = substr( $oLine, 0, $prefixLength );
1850
1851                                 # eh?
1852                                 $pref2 = str_replace( ';', ':', $pref );
1853                                 $t = substr( $oLine, $prefixLength );
1854                                 $this->mInPre = !empty($preOpenMatch);
1855                         } else {
1856                                 # Don't interpret any other prefixes in preformatted text
1857                                 $prefixLength = 0;
1858                                 $pref = $pref2 = '';
1859                                 $t = $oLine;
1860                         }
1861
1862                         # List generation
1863                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1864                                 # Same as the last item, so no need to deal with nesting or opening stuff
1865                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1866                                 $paragraphStack = false;
1867
1868                                 if ( substr( $pref, -1 ) == ';') {
1869                                         # The one nasty exception: definition lists work like this:
1870                                         # ; title : definition text
1871                                         # So we check for : in the remainder text to split up the
1872                                         # title and definition, without b0rking links.
1873                                         $term = $t2 = '';
1874                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1875                                                 $t = $t2;
1876                                                 $output .= $term . $this->nextItem( ':' );
1877                                         }
1878                                 }
1879                         } elseif( $prefixLength || $lastPrefixLength ) {
1880                                 # Either open or close a level...
1881                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1882                                 $paragraphStack = false;
1883
1884                                 while( $commonPrefixLength < $lastPrefixLength ) {
1885                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1886                                         --$lastPrefixLength;
1887                                 }
1888                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1889                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1890                                 }
1891                                 while ( $prefixLength > $commonPrefixLength ) {
1892                                         $char = substr( $pref, $commonPrefixLength, 1 );
1893                                         $output .= $this->openList( $char );
1894
1895                                         if ( ';' == $char ) {
1896                                                 # FIXME: This is dupe of code above
1897                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1898                                                         $t = $t2;
1899                                                         $output .= $term . $this->nextItem( ':' );
1900                                                 }
1901                                         }
1902                                         ++$commonPrefixLength;
1903                                 }
1904                                 $lastPrefix = $pref2;
1905                         }
1906                         if( 0 == $prefixLength ) {
1907                                 wfProfileIn( "$fname-paragraph" );
1908                                 # No prefix (not in list)--go to paragraph mode
1909                                 // XXX: use a stack for nestable elements like span, table and div
1910                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1911                                 $closematch = preg_match(
1912                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1913                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1914                                 if ( $openmatch or $closematch ) {
1915                                         $paragraphStack = false;
1916                                         # TODO bug 5718: paragraph closed
1917                                         $output .= $this->closeParagraph();
1918                                         if ( $preOpenMatch and !$preCloseMatch ) {
1919                                                 $this->mInPre = true;
1920                                         }
1921                                         if ( $closematch ) {
1922                                                 $inBlockElem = false;
1923                                         } else {
1924                                                 $inBlockElem = true;
1925                                         }
1926                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1927                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1928                                                 // pre
1929                                                 if ($this->mLastSection != 'pre') {
1930                                                         $paragraphStack = false;
1931                                                         $output .= $this->closeParagraph().'<pre>';
1932                                                         $this->mLastSection = 'pre';
1933                                                 }
1934                                                 $t = substr( $t, 1 );
1935                                         } else {
1936                                                 // paragraph
1937                                                 if ( '' == trim($t) ) {
1938                                                         if ( $paragraphStack ) {
1939                                                                 $output .= $paragraphStack.'<br />';
1940                                                                 $paragraphStack = false;
1941                                                                 $this->mLastSection = 'p';
1942                                                         } else {
1943                                                                 if ($this->mLastSection != 'p' ) {
1944                                                                         $output .= $this->closeParagraph();
1945                                                                         $this->mLastSection = '';
1946                                                                         $paragraphStack = '<p>';
1947                                                                 } else {
1948                                                                         $paragraphStack = '</p><p>';
1949                                                                 }
1950                                                         }
1951                                                 } else {
1952                                                         if ( $paragraphStack ) {
1953                                                                 $output .= $paragraphStack;
1954                                                                 $paragraphStack = false;
1955                                                                 $this->mLastSection = 'p';
1956                                                         } else if ($this->mLastSection != 'p') {
1957                                                                 $output .= $this->closeParagraph().'<p>';
1958                                                                 $this->mLastSection = 'p';
1959                                                         }
1960                                                 }
1961                                         }
1962                                 }
1963                                 wfProfileOut( "$fname-paragraph" );
1964                         }
1965                         // somewhere above we forget to get out of pre block (bug 785)
1966                         if($preCloseMatch && $this->mInPre) {
1967                                 $this->mInPre = false;
1968                         }
1969                         if ($paragraphStack === false) {
1970                                 $output .= $t."\n";
1971                         }
1972                 }
1973                 while ( $prefixLength ) {
1974                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1975                         --$prefixLength;
1976                 }
1977                 if ( '' != $this->mLastSection ) {
1978                         $output .= '</' . $this->mLastSection . '>';
1979                         $this->mLastSection = '';
1980                 }
1981
1982                 wfProfileOut( $fname );
1983                 return $output;
1984         }
1985
1986         /**
1987          * Split up a string on ':', ignoring any occurences inside tags
1988          * to prevent illegal overlapping.
1989          * @param string $str the string to split
1990          * @param string &$before set to everything before the ':'
1991          * @param string &$after set to everything after the ':'
1992          * return string the position of the ':', or false if none found
1993          */
1994         function findColonNoLinks($str, &$before, &$after) {
1995                 $fname = 'Parser::findColonNoLinks';
1996                 wfProfileIn( $fname );
1997
1998                 $pos = strpos( $str, ':' );
1999                 if( $pos === false ) {
2000                         // Nothing to find!
2001                         wfProfileOut( $fname );
2002                         return false;
2003                 }
2004
2005                 $lt = strpos( $str, '<' );
2006                 if( $lt === false || $lt > $pos ) {
2007                         // Easy; no tag nesting to worry about
2008                         $before = substr( $str, 0, $pos );
2009                         $after = substr( $str, $pos+1 );
2010                         wfProfileOut( $fname );
2011                         return $pos;
2012                 }
2013
2014                 // Ugly state machine to walk through avoiding tags.
2015                 $state = MW_COLON_STATE_TEXT;
2016                 $stack = 0;
2017                 $len = strlen( $str );
2018                 for( $i = 0; $i < $len; $i++ ) {
2019                         $c = $str{$i};
2020
2021                         switch( $state ) {
2022                         // (Using the number is a performance hack for common cases)
2023                         case 0: // MW_COLON_STATE_TEXT:
2024                                 switch( $c ) {
2025                                 case "<":
2026                                         // Could be either a <start> tag or an </end> tag
2027                                         $state = MW_COLON_STATE_TAGSTART;
2028                                         break;
2029                                 case ":":
2030                                         if( $stack == 0 ) {
2031                                                 // We found it!
2032                                                 $before = substr( $str, 0, $i );
2033                                                 $after = substr( $str, $i + 1 );
2034                                                 wfProfileOut( $fname );
2035                                                 return $i;
2036                                         }
2037                                         // Embedded in a tag; don't break it.
2038                                         break;
2039                                 default:
2040                                         // Skip ahead looking for something interesting
2041                                         $colon = strpos( $str, ':', $i );
2042                                         if( $colon === false ) {
2043                                                 // Nothing else interesting
2044                                                 wfProfileOut( $fname );
2045                                                 return false;
2046                                         }
2047                                         $lt = strpos( $str, '<', $i );
2048                                         if( $stack === 0 ) {
2049                                                 if( $lt === false || $colon < $lt ) {
2050                                                         // We found it!
2051                                                         $before = substr( $str, 0, $colon );
2052                                                         $after = substr( $str, $colon + 1 );
2053                                                         wfProfileOut( $fname );
2054                                                         return $i;
2055                                                 }
2056                                         }
2057                                         if( $lt === false ) {
2058                                                 // Nothing else interesting to find; abort!
2059                                                 // We're nested, but there's no close tags left. Abort!
2060                                                 break 2;
2061                                         }
2062                                         // Skip ahead to next tag start
2063                                         $i = $lt;
2064                                         $state = MW_COLON_STATE_TAGSTART;
2065                                 }
2066                                 break;
2067                         case 1: // MW_COLON_STATE_TAG:
2068                                 // In a <tag>
2069                                 switch( $c ) {
2070                                 case ">":
2071                                         $stack++;
2072                                         $state = MW_COLON_STATE_TEXT;
2073                                         break;
2074                                 case "/":
2075                                         // Slash may be followed by >?
2076                                         $state = MW_COLON_STATE_TAGSLASH;
2077                                         break;
2078                                 default:
2079                                         // ignore
2080                                 }
2081                                 break;
2082                         case 2: // MW_COLON_STATE_TAGSTART:
2083                                 switch( $c ) {
2084                                 case "/":
2085                                         $state = MW_COLON_STATE_CLOSETAG;
2086                                         break;
2087                                 case "!":
2088                                         $state = MW_COLON_STATE_COMMENT;
2089                                         break;
2090                                 case ">":
2091                                         // Illegal early close? This shouldn't happen D:
2092                                         $state = MW_COLON_STATE_TEXT;
2093                                         break;
2094                                 default:
2095                                         $state = MW_COLON_STATE_TAG;
2096                                 }
2097                                 break;
2098                         case 3: // MW_COLON_STATE_CLOSETAG:
2099                                 // In a </tag>
2100                                 if( $c == ">" ) {
2101                                         $stack--;
2102                                         if( $stack < 0 ) {
2103                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2104                                                 wfProfileOut( $fname );
2105                                                 return false;
2106                                         }
2107                                         $state = MW_COLON_STATE_TEXT;
2108                                 }
2109                                 break;
2110                         case MW_COLON_STATE_TAGSLASH:
2111                                 if( $c == ">" ) {
2112                                         // Yes, a self-closed tag <blah/>
2113                                         $state = MW_COLON_STATE_TEXT;
2114                                 } else {
2115                                         // Probably we're jumping the gun, and this is an attribute
2116                                         $state = MW_COLON_STATE_TAG;
2117                                 }
2118                                 break;
2119                         case 5: // MW_COLON_STATE_COMMENT:
2120                                 if( $c == "-" ) {
2121                                         $state = MW_COLON_STATE_COMMENTDASH;
2122                                 }
2123                                 break;
2124                         case MW_COLON_STATE_COMMENTDASH:
2125                                 if( $c == "-" ) {
2126                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2127                                 } else {
2128                                         $state = MW_COLON_STATE_COMMENT;
2129                                 }
2130                                 break;
2131                         case MW_COLON_STATE_COMMENTDASHDASH:
2132                                 if( $c == ">" ) {
2133                                         $state = MW_COLON_STATE_TEXT;
2134                                 } else {
2135                                         $state = MW_COLON_STATE_COMMENT;
2136                                 }
2137                                 break;
2138                         default:
2139                                 throw new MWException( "State machine error in $fname" );
2140                         }
2141                 }
2142                 if( $stack > 0 ) {
2143                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2144                         return false;
2145                 }
2146                 wfProfileOut( $fname );
2147                 return false;
2148         }
2149
2150         /**
2151          * Return value of a magic variable (like PAGENAME)
2152          *
2153          * @private
2154          */
2155         function getVariableValue( $index ) {
2156                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2157
2158                 /**
2159                  * Some of these require message or data lookups and can be
2160                  * expensive to check many times.
2161                  */
2162                 static $varCache = array();
2163                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2164                         if ( isset( $varCache[$index] ) )
2165                                 return $varCache[$index];
2166
2167                 $ts = time();
2168                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2169
2170                 switch ( $index ) {
2171                         case MAG_CURRENTMONTH:
2172                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2173                         case MAG_CURRENTMONTHNAME:
2174                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2175                         case MAG_CURRENTMONTHNAMEGEN:
2176                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2177                         case MAG_CURRENTMONTHABBREV:
2178                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2179                         case MAG_CURRENTDAY:
2180                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2181                         case MAG_CURRENTDAY2:
2182                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2183                         case MAG_PAGENAME:
2184                                 return $this->mTitle->getText();
2185                         case MAG_PAGENAMEE:
2186                                 return $this->mTitle->getPartialURL();
2187                         case MAG_FULLPAGENAME:
2188                                 return $this->mTitle->getPrefixedText();
2189                         case MAG_FULLPAGENAMEE:
2190                                 return $this->mTitle->getPrefixedURL();
2191                         case MAG_SUBPAGENAME:
2192                                 return $this->mTitle->getSubpageText();
2193                         case MAG_SUBPAGENAMEE:
2194                                 return $this->mTitle->getSubpageUrlForm();
2195                         case MAG_BASEPAGENAME:
2196                                 return $this->mTitle->getBaseText();
2197                         case MAG_BASEPAGENAMEE:
2198                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2199                         case MAG_TALKPAGENAME:
2200                                 if( $this->mTitle->canTalk() ) {
2201                                         $talkPage = $this->mTitle->getTalkPage();
2202                                         return $talkPage->getPrefixedText();
2203                                 } else {
2204                                         return '';
2205                                 }
2206                         case MAG_TALKPAGENAMEE:
2207                                 if( $this->mTitle->canTalk() ) {
2208                                         $talkPage = $this->mTitle->getTalkPage();
2209                                         return $talkPage->getPrefixedUrl();
2210                                 } else {
2211                                         return '';
2212                                 }
2213                         case MAG_SUBJECTPAGENAME:
2214                                 $subjPage = $this->mTitle->getSubjectPage();
2215                                 return $subjPage->getPrefixedText();
2216                         case MAG_SUBJECTPAGENAMEE:
2217                                 $subjPage = $this->mTitle->getSubjectPage();
2218                                 return $subjPage->getPrefixedUrl();
2219                         case MAG_REVISIONID:
2220                                 return $this->mRevisionId;
2221                         case MAG_NAMESPACE:
2222                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2223                         case MAG_NAMESPACEE:
2224                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2225                         case MAG_TALKSPACE:
2226                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2227                         case MAG_TALKSPACEE:
2228                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2229                         case MAG_SUBJECTSPACE:
2230                                 return $this->mTitle->getSubjectNsText();
2231                         case MAG_SUBJECTSPACEE:
2232                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2233                         case MAG_CURRENTDAYNAME:
2234                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2235                         case MAG_CURRENTYEAR:
2236                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2237                         case MAG_CURRENTTIME:
2238                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2239                         case MAG_CURRENTWEEK:
2240                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2241                                 // int to remove the padding
2242                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2243                         case MAG_CURRENTDOW:
2244                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2245                         case MAG_NUMBEROFARTICLES:
2246                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2247                         case MAG_NUMBEROFFILES:
2248                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2249                         case MAG_NUMBEROFUSERS:
2250                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2251                         case MAG_NUMBEROFPAGES:
2252                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2253                         case MAG_NUMBEROFADMINS:
2254                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2255                         case MAG_CURRENTTIMESTAMP:
2256                                 return $varCache[$index] = wfTimestampNow();
2257                         case MAG_CURRENTVERSION:
2258                                 global $wgVersion;
2259                                 return $wgVersion;
2260                         case MAG_SITENAME:
2261                                 return $wgSitename;
2262                         case MAG_SERVER:
2263                                 return $wgServer;
2264                         case MAG_SERVERNAME:
2265                                 return $wgServerName;
2266                         case MAG_SCRIPTPATH:
2267                                 return $wgScriptPath;
2268                         case MAG_DIRECTIONMARK:
2269                                 return $wgContLang->getDirMark();
2270                         case MAG_CONTENTLANGUAGE:
2271                                 global $wgContLanguageCode;
2272                                 return $wgContLanguageCode;
2273                         default:
2274                                 $ret = null;
2275                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2276                                         return $ret;
2277                                 else
2278                                         return null;
2279                 }
2280         }
2281
2282         /**
2283          * initialise the magic variables (like CURRENTMONTHNAME)
2284          *
2285          * @private
2286          */
2287         function initialiseVariables() {
2288                 $fname = 'Parser::initialiseVariables';
2289                 wfProfileIn( $fname );
2290                 global $wgVariableIDs;
2291                 $this->mVariables = array();
2292                 foreach ( $wgVariableIDs as $id ) {
2293                         $mw =& MagicWord::get( $id );
2294                         $mw->addToArray( $this->mVariables, $id );
2295                 }
2296                 wfProfileOut( $fname );
2297         }
2298
2299         /**
2300          * parse any parentheses in format ((title|part|part))
2301          * and call callbacks to get a replacement text for any found piece
2302          *
2303          * @param string $text The text to parse
2304          * @param array $callbacks rules in form:
2305          *     '{' => array(                            # opening parentheses
2306          *                                      'end' => '}',   # closing parentheses
2307          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2308          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2309          *                                                                )
2310          *                                      )
2311          * @private
2312          */
2313         function replace_callback ($text, $callbacks) {
2314                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2315                 $lastOpeningBrace = -1;                 # last not closed parentheses
2316
2317                 for ($i = 0; $i < strlen($text); $i++) {
2318                         # check for any opening brace
2319                         $rule = null;
2320                         $nextPos = -1;
2321                         foreach ($callbacks as $key => $value) {
2322                                 $pos = strpos ($text, $key, $i);
2323                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2324                                         $rule = $value;
2325                                         $nextPos = $pos;
2326                                 }
2327                         }
2328
2329                         if ($lastOpeningBrace >= 0) {
2330                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2331
2332                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2333                                         $rule = null;
2334                                         $nextPos = $pos;
2335                                 }
2336
2337                                 $pos = strpos ($text, '|', $i);
2338
2339                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2340                                         $rule = null;
2341                                         $nextPos = $pos;
2342                                 }
2343                         }
2344
2345                         if ($nextPos == -1)
2346                                 break;
2347
2348                         $i = $nextPos;
2349
2350                         # found openning brace, lets add it to parentheses stack
2351                         if (null != $rule) {
2352                                 $piece = array('brace' => $text[$i],
2353                                                            'braceEnd' => $rule['end'],
2354                                                            'count' => 1,
2355                                                            'title' => '',
2356                                                            'parts' => null);
2357
2358                                 # count openning brace characters
2359                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2360                                         $piece['count']++;
2361                                         $i++;
2362                                 }
2363
2364                                 $piece['startAt'] = $i+1;
2365                                 $piece['partStart'] = $i+1;
2366
2367                                 # we need to add to stack only if openning brace count is enough for any given rule
2368                                 foreach ($rule['cb'] as $cnt => $fn) {
2369                                         if ($piece['count'] >= $cnt) {
2370                                                 $lastOpeningBrace ++;
2371                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2372                                                 break;
2373                                         }
2374                                 }
2375
2376                                 continue;
2377                         }
2378                         else if ($lastOpeningBrace >= 0) {
2379                                 # first check if it is a closing brace
2380                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2381                                         # lets check if it is enough characters for closing brace
2382                                         $count = 1;
2383                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2384                                                 $count++;
2385
2386                                         # if there are more closing parentheses than opening ones, we parse less
2387                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2388                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2389
2390                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2391                                         $matchingCount = 0;
2392                                         $matchingCallback = null;
2393                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2394                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2395                                                         $matchingCount = $cnt;
2396                                                         $matchingCallback = $fn;
2397                                                 }
2398                                         }
2399
2400                                         if ($matchingCount == 0) {
2401                                                 $i += $count - 1;
2402                                                 continue;
2403                                         }
2404
2405                                         # lets set a title or last part (if '|' was found)
2406                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2407                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2408                                         else
2409                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2410
2411                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2412                                         $pieceEnd = $i + $matchingCount;
2413
2414                                         if( is_callable( $matchingCallback ) ) {
2415                                                 $cbArgs = array (
2416                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2417                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2418                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2419                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2420                                                                                  );
2421                                                 # finally we can call a user callback and replace piece of text
2422                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2423                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2424                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2425                                         }
2426                                         else {
2427                                                 # null value for callback means that parentheses should be parsed, but not replaced
2428                                                 $i += $matchingCount - 1;
2429                                         }
2430
2431                                         # reset last openning parentheses, but keep it in case there are unused characters
2432                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2433                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2434                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2435                                                                    'title' => '',
2436                                                                    'parts' => null,
2437                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2438                                         $openingBraceStack[$lastOpeningBrace--] = null;
2439
2440                                         if ($matchingCount < $piece['count']) {
2441                                                 $piece['count'] -= $matchingCount;
2442                                                 $piece['startAt'] -= $matchingCount;
2443                                                 $piece['partStart'] = $piece['startAt'];
2444                                                 # do we still qualify for any callback with remaining count?
2445                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2446                                                         if ($piece['count'] >= $cnt) {
2447                                                                 $lastOpeningBrace ++;
2448                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2449                                                                 break;
2450                                                         }
2451                                                 }
2452                                         }
2453                                         continue;
2454                                 }
2455
2456                                 # lets set a title if it is a first separator, or next part otherwise
2457                                 if ($text[$i] == '|') {
2458                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2459                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2460                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2461                                         }
2462                                         else
2463                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2464
2465                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2466                                 }
2467                         }
2468                 }
2469
2470                 return $text;
2471         }
2472
2473         /**
2474          * Replace magic variables, templates, and template arguments
2475          * with the appropriate text. Templates are substituted recursively,
2476          * taking care to avoid infinite loops.
2477          *
2478          * Note that the substitution depends on value of $mOutputType:
2479          *  OT_WIKI: only {{subst:}} templates
2480          *  OT_MSG: only magic variables
2481          *  OT_HTML: all templates and magic variables
2482          *
2483          * @param string $tex The text to transform
2484          * @param array $args Key-value pairs representing template parameters to substitute
2485          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2486          * @private
2487          */
2488         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2489                 # Prevent too big inclusions
2490                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2491                         return $text;
2492                 }
2493
2494                 $fname = 'Parser::replaceVariables';
2495                 wfProfileIn( $fname );
2496
2497                 # This function is called recursively. To keep track of arguments we need a stack:
2498                 array_push( $this->mArgStack, $args );
2499
2500                 $braceCallbacks = array();
2501                 if ( !$argsOnly ) {
2502                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2503                 }
2504                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2505                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2506                 }
2507                 $callbacks = array();
2508                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2509                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2510                 $text = $this->replace_callback ($text, $callbacks);
2511
2512                 array_pop( $this->mArgStack );
2513
2514                 wfProfileOut( $fname );
2515                 return $text;
2516         }
2517
2518         /**
2519          * Replace magic variables
2520          * @private
2521          */
2522         function variableSubstitution( $matches ) {
2523                 $fname = 'Parser::variableSubstitution';
2524                 $varname = $matches[1];
2525                 wfProfileIn( $fname );
2526                 if ( !$this->mVariables ) {
2527                         $this->initialiseVariables();
2528                 }
2529                 $skip = false;
2530                 if ( $this->mOutputType == OT_WIKI ) {
2531                         # Do only magic variables prefixed by SUBST
2532                         $mwSubst =& MagicWord::get( MAG_SUBST );
2533                         if (!$mwSubst->matchStartAndRemove( $varname ))
2534                                 $skip = true;
2535                         # Note that if we don't substitute the variable below,
2536                         # we don't remove the {{subst:}} magic word, in case
2537                         # it is a template rather than a magic variable.
2538                 }
2539                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2540                         $id = $this->mVariables[$varname];
2541                         $text = $this->getVariableValue( $id );
2542                         $this->mOutput->mContainsOldMagic = true;
2543                 } else {
2544                         $text = $matches[0];
2545                 }
2546                 wfProfileOut( $fname );
2547                 return $text;
2548         }
2549
2550         # Split template arguments
2551         function getTemplateArgs( $argsString ) {
2552                 if ( $argsString === '' ) {
2553                         return array();
2554                 }
2555
2556                 $args = explode( '|', substr( $argsString, 1 ) );
2557
2558                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2559                 # merged with the next arg because the '|' character between belongs
2560                 # to the link syntax and not the template parameter syntax.
2561                 $argc = count($args);
2562
2563                 for ( $i = 0; $i < $argc-1; $i++ ) {
2564                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2565                                 $args[$i] .= '|'.$args[$i+1];
2566                                 array_splice($args, $i+1, 1);
2567                                 $i--;
2568                                 $argc--;
2569                         }
2570                 }
2571
2572                 return $args;
2573         }
2574
2575         /**
2576          * Return the text of a template, after recursively
2577          * replacing any variables or templates within the template.
2578          *
2579          * @param array $piece The parts of the template
2580          *  $piece['text']: matched text
2581          *  $piece['title']: the title, i.e. the part before the |
2582          *  $piece['parts']: the parameter array
2583          * @return string the text of the template
2584          * @private
2585          */
2586         function braceSubstitution( $piece ) {
2587                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2588                 $fname = 'Parser::braceSubstitution';
2589                 wfProfileIn( $fname );
2590
2591                 # Flags
2592                 $found = false;             # $text has been filled
2593                 $nowiki = false;            # wiki markup in $text should be escaped
2594                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2595                 $noargs = false;            # Don't replace triple-brace arguments in $text
2596                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2597                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2598                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2599
2600                 # Title object, where $text came from
2601                 $title = NULL;
2602
2603                 $linestart = '';
2604
2605                 # $part1 is the bit before the first |, and must contain only title characters
2606                 # $args is a list of arguments, starting from index 0, not including $part1
2607
2608                 $part1 = $piece['title'];
2609                 # If the third subpattern matched anything, it will start with |
2610
2611                 if (null == $piece['parts']) {
2612                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2613                         if ($replaceWith != $piece['text']) {
2614                                 $text = $replaceWith;
2615                                 $found = true;
2616                                 $noparse = true;
2617                                 $noargs = true;
2618                         }
2619                 }
2620
2621                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2622                 $argc = count( $args );
2623
2624                 # SUBST
2625                 if ( !$found ) {
2626                         $mwSubst =& MagicWord::get( MAG_SUBST );
2627                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2628                                 # One of two possibilities is true:
2629                                 # 1) Found SUBST but not in the PST phase
2630                                 # 2) Didn't find SUBST and in the PST phase
2631                                 # In either case, return without further processing
2632                                 $text = $piece['text'];
2633                                 $found = true;
2634                                 $noparse = true;
2635                                 $noargs = true;
2636                         }
2637                 }
2638
2639                 # MSG, MSGNW, INT and RAW
2640                 if ( !$found ) {
2641                         # Check for MSGNW:
2642                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2643                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2644                                 $nowiki = true;
2645                         } else {
2646                                 # Remove obsolete MSG:
2647                                 $mwMsg =& MagicWord::get( MAG_MSG );
2648                                 $mwMsg->matchStartAndRemove( $part1 );
2649                         }
2650
2651                         # Check for RAW:
2652                         $mwRaw =& MagicWord::get( MAG_RAW );
2653                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2654                                 $forceRawInterwiki = true;
2655                         }
2656
2657                         # Check if it is an internal message
2658                         $mwInt =& MagicWord::get( MAG_INT );
2659                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2660                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2661                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2662                                         $found = true;
2663                                 }
2664                         }
2665                 }
2666
2667                 # NS
2668                 if ( !$found ) {
2669                         # Check for NS: (namespace expansion)
2670                         $mwNs = MagicWord::get( MAG_NS );
2671                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2672                                 if ( intval( $part1 ) || $part1 == "0" ) {
2673                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2674                                         $found = true;
2675                                 } else {
2676                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2677                                         if ( !is_null( $index ) ) {
2678                                                 $text = $linestart . $wgContLang->getNsText( $index );
2679                                                 $found = true;
2680                                         }
2681                                 }
2682                         }
2683                 }
2684
2685                 # URLENCODE
2686                 if( !$found ) {
2687                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2688                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2689                                 $text = $linestart . urlencode( $part1 );
2690                                 $found = true;
2691                         }
2692                 }
2693
2694                 # LCFIRST, UCFIRST, LC and UC
2695                 if ( !$found ) {
2696                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2697                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2698                         $lc =& MagicWord::get( MAG_LC );
2699                         $uc =& MagicWord::get( MAG_UC );
2700                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2701                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2702                                 $found = true;
2703                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2704                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2705                                 $found = true;
2706                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2707                                 $text = $linestart . $wgContLang->lc( $part1 );
2708                                 $found = true;
2709                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2710                                  $text = $linestart . $wgContLang->uc( $part1 );
2711                                  $found = true;
2712                         }
2713                 }
2714
2715                 # LOCALURL and FULLURL
2716                 if ( !$found ) {
2717                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2718                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2719                         $mwFull =& MagicWord::get( MAG_FULLURL );
2720                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2721
2722
2723                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2724                                 $func = 'getLocalURL';
2725                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2726                                 $func = 'escapeLocalURL';
2727                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2728                                 $func = 'getFullURL';
2729                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2730                                 $func = 'escapeFullURL';
2731                         } else {
2732                                 $func = false;
2733                         }
2734
2735                         if ( $func !== false ) {
2736                                 $title = Title::newFromText( $part1 );
2737                                 # Due to order of execution of a lot of bits, the values might be encoded
2738                                 # before arriving here; if that's true, then the title can't be created
2739                                 # and the variable will fail. If we can't get a decent title from the first
2740                                 # attempt, url-decode and try for a second.
2741                                 if( is_null( $title ) )
2742                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2743                                 if ( !is_null( $title ) ) {
2744                                         if ( $argc > 0 ) {
2745                                                 $text = $linestart . $title->$func( $args[0] );
2746                                         } else {
2747                                                 $text = $linestart . $title->$func();
2748                                         }
2749                                         $found = true;
2750                                 }
2751                         }
2752                 }
2753
2754                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2755                 # GRAMMAR
2756                 if ( !$found && $argc == 1 ) {
2757                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2758                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2759                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2760                                 $found = true;
2761                         }
2762                 }
2763
2764                 # PLURAL
2765                 if ( !$found && $argc >= 2 ) {
2766                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2767                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2768                                 while ( count($args) < 5 ) { $args[] = $args[count($args)-1]; }
2769                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1],
2770                                         $args[2], $args[3], $args[4]);
2771                                 $found = true;
2772                         }
2773                 }
2774
2775                 # DISPLAYTITLE
2776                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2777                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2778                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2779
2780                                 # Set title in parser output object
2781                                 $param = $args[0];
2782                                 $parserOptions = new ParserOptions;
2783                                 $local_parser = new Parser ();
2784                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2785                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2786
2787                                 # Add subtitle
2788                                 $t = $this->mTitle->getPrefixedText();
2789                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2790                                 $text = "" ;
2791                                 $found = true ;
2792                         }
2793                 }
2794
2795                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2796                 if( !$found ) {
2797                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2798                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2799                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2800                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles',
2801                                                                          MAG_NUMBEROFADMINS => 'wfNumberOfAdmins' );
2802                         foreach( $mwWordsToCheck as $word => $func ) {
2803                                 $mwCurrentWord =& MagicWord::get( $word );
2804                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2805                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2806                                         if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2807                                                 # Raw and unformatted
2808                                                 $text = $linestart . call_user_func( $func );
2809                                         } else {
2810                                                 # Formatted according to the content default
2811                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2812                                         }
2813                                         $found = true;
2814                                 }
2815                         }
2816                 }
2817
2818                 # PAGESINNAMESPACE
2819                 if( !$found ) {
2820                         $mwPagesInNs =& MagicWord::get( MAG_PAGESINNAMESPACE );
2821                         if( $mwPagesInNs->matchStartAndRemove( $part1 ) ) {
2822                                 $found = true;
2823                                 $count = wfPagesInNs( intval( $part1 ) );
2824                                 $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2825                                 if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2826                                         $text = $linestart . $count;
2827                                 } else {
2828                                         $text = $linestart . $wgContLang->formatNum( $count );
2829                                 }
2830                         }
2831                 }
2832
2833                 # #LANGUAGE:
2834                 if( !$found ) {
2835                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2836                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2837                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2838                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2839                                 $found = true;
2840                         }
2841                 }
2842
2843                 # Extensions
2844                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2845                         $colonPos = strpos( $part1, ':' );
2846                         if ( $colonPos !== false ) {
2847                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2848                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2849                                         $funcArgs = array_map( 'trim', $args );
2850                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2851                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2852                                         $found = true;
2853
2854                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2855                                         //$noargs = true;
2856                                         //$noparse = true;
2857
2858                                         if ( is_array( $result ) ) {
2859                                                 $text = $linestart . $result[0];
2860                                                 unset( $result[0] );
2861
2862                                                 // Extract flags into the local scope
2863                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2864                                                 extract( $result );
2865                                         } else {
2866                                                 $text = $linestart . $result;
2867                                         }
2868                                 }
2869                         }
2870                 }
2871
2872                 # Template table test
2873
2874                 # Did we encounter this template already? If yes, it is in the cache
2875                 # and we need to check for loops.
2876                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2877                         $found = true;
2878
2879                         # Infinite loop test
2880                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2881                                 $noparse = true;
2882                                 $noargs = true;
2883                                 $found = true;
2884                                 $text = $linestart .
2885                                         '{{' . $part1 . '}}' .
2886                                         '<!-- WARNING: template loop detected -->';
2887                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2888                         } else {
2889                                 # set $text to cached message.
2890                                 $text = $linestart . $this->mTemplates[$piece['title']];
2891                         }
2892                 }
2893
2894                 # Load from database
2895                 $lastPathLevel = $this->mTemplatePath;
2896                 if ( !$found ) {
2897                         $ns = NS_TEMPLATE;
2898                         # declaring $subpage directly in the function call
2899                         # does not work correctly with references and breaks
2900                         # {{/subpage}}-style inclusions
2901                         $subpage = '';
2902                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2903                         if ($subpage !== '') {
2904                                 $ns = $this->mTitle->getNamespace();
2905                         }
2906                         $title = Title::newFromText( $part1, $ns );
2907
2908
2909                         if ( !is_null( $title ) ) {
2910                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2911                                 # Check for language variants if the template is not found
2912                                 if($checkVariantLink && $title->getArticleID() == 0){
2913                                         $wgContLang->findVariantLink($part1, $title);
2914                                 }
2915
2916                                 if ( !$title->isExternal() ) {
2917                                         # Check for excessive inclusion
2918                                         $dbk = $title->getPrefixedDBkey();
2919                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2920                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2921                                                         $text = SpecialPage::capturePath( $title );
2922                                                         if ( is_string( $text ) ) {
2923                                                                 $found = true;
2924                                                                 $noparse = true;
2925                                                                 $noargs = true;
2926                                                                 $isHTML = true;
2927                                                                 $this->disableCache();
2928                                                         }
2929                                                 } else {
2930                                                         $articleContent = $this->fetchTemplate( $title );
2931                                                         if ( $articleContent !== false ) {
2932                                                                 $found = true;
2933                                                                 $text = $articleContent;
2934                                                                 $replaceHeadings = true;
2935                                                         }
2936                                                 }
2937                                         }
2938
2939                                         # If the title is valid but undisplayable, make a link to it
2940                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2941                                                 $text = '[['.$title->getPrefixedText().']]';
2942                                                 $found = true;
2943                                         }
2944                                 } elseif ( $title->isTrans() ) {
2945                                         // Interwiki transclusion
2946                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2947                                                 $text = $this->interwikiTransclude( $title, 'render' );
2948                                                 $isHTML = true;
2949                                                 $noparse = true;
2950                                         } else {
2951                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2952                                                 $replaceHeadings = true;
2953                                         }
2954                                         $found = true;
2955                                 }
2956
2957                                 # Template cache array insertion
2958                                 # Use the original $piece['title'] not the mangled $part1, so that
2959                                 # modifiers such as RAW: produce separate cache entries
2960                                 if( $found ) {
2961                                         if( $isHTML ) {
2962                                                 // A special page; don't store it in the template cache.
2963                                         } else {
2964                                                 $this->mTemplates[$piece['title']] = $text;
2965                                         }
2966                                         $text = $linestart . $text;
2967                                 }
2968                         }
2969                 }
2970
2971                 # Recursive parsing, escaping and link table handling
2972                 # Only for HTML output
2973                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2974                         $text = wfEscapeWikiText( $text );
2975                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2976                         if ( $noargs ) {
2977                                 $assocArgs = array();
2978                         } else {
2979                                 # Clean up argument array
2980                                 $assocArgs = array();
2981                                 $index = 1;
2982                                 foreach( $args as $arg ) {
2983                                         $eqpos = strpos( $arg, '=' );
2984                                         if ( $eqpos === false ) {
2985                                                 $assocArgs[$index++] = $arg;
2986                                         } else {
2987                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2988                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2989                                                 if ( $value === false ) {
2990                                                         $value = '';
2991                                                 }
2992                                                 if ( $name !== false ) {
2993                                                         $assocArgs[$name] = $value;
2994                                                 }
2995                                         }
2996                                 }
2997
2998                                 # Add a new element to the templace recursion path
2999                                 $this->mTemplatePath[$part1] = 1;
3000                         }
3001
3002                         if ( !$noparse ) {
3003                                 # If there are any <onlyinclude> tags, only include them
3004                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3005                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
3006                                         $text = '';
3007                                         foreach ($m[1] as $piece)
3008                                                 $text .= $piece;
3009                                 }
3010                                 # Remove <noinclude> sections and <includeonly> tags
3011                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
3012                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3013
3014                                 if( $this->mOutputType == OT_HTML ) {
3015                                         # Strip <nowiki>, <pre>, etc.
3016                                         $text = $this->strip( $text, $this->mStripState );
3017                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3018                                 }
3019                                 $text = $this->replaceVariables( $text, $assocArgs );
3020
3021                                 # If the template begins with a table or block-level
3022                                 # element, it should be treated as beginning a new line.
3023                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3024                                         $text = "\n" . $text;
3025                                 }
3026                         } elseif ( !$noargs ) {
3027                                 # $noparse and !$noargs
3028                                 # Just replace the arguments, not any double-brace items
3029                                 # This is used for rendered interwiki transclusion
3030                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3031                         }
3032                 }
3033                 # Prune lower levels off the recursion check path
3034                 $this->mTemplatePath = $lastPathLevel;
3035
3036                 if ( !$found ) {
3037                         wfProfileOut( $fname );
3038                         return $piece['text'];
3039                 } else {
3040                         if ( $isHTML ) {
3041                                 # Replace raw HTML by a placeholder
3042                                 # Add a blank line preceding, to prevent it from mucking up
3043                                 # immediately preceding headings
3044                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3045                         } else {
3046                                 # replace ==section headers==
3047                                 # XXX this needs to go away once we have a better parser.
3048                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3049                                         if( !is_null( $title ) )
3050                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3051                                         else
3052                                                 $encodedname = base64_encode("");
3053                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3054                                                 PREG_SPLIT_DELIM_CAPTURE);
3055                                         $text = '';
3056                                         $nsec = 0;
3057                                         for( $i = 0; $i < count($m); $i += 2 ) {
3058                                                 $text .= $m[$i];
3059                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3060                                                 $hl = $m[$i + 1];
3061                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3062                                                         $text .= $hl;
3063                                                         continue;
3064                                                 }
3065                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3066                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3067                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3068
3069                                                 $nsec++;
3070                                         }
3071                                 }
3072                         }
3073                 }
3074
3075                 # Prune lower levels off the recursion check path
3076                 $this->mTemplatePath = $lastPathLevel;
3077
3078                 if ( !$found ) {
3079                         wfProfileOut( $fname );
3080                         return $piece['text'];
3081                 } else {
3082                         wfProfileOut( $fname );
3083                         return $text;
3084                 }
3085         }
3086
3087         /**
3088          * Fetch the unparsed text of a template and register a reference to it.
3089          */
3090         function fetchTemplate( $title ) {
3091                 $text = false;
3092                 // Loop to fetch the article, with up to 1 redirect
3093                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3094                         $rev = Revision::newFromTitle( $title );
3095                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3096                         if ( !$rev ) {
3097                                 break;
3098                         }
3099                         $text = $rev->getText();
3100                         if ( $text === false ) {
3101                                 break;
3102                         }
3103                         // Redirect?
3104                         $title = Title::newFromRedirect( $text );
3105                 }
3106                 return $text;
3107         }
3108
3109         /**
3110          * Transclude an interwiki link.
3111          */
3112         function interwikiTransclude( $title, $action ) {
3113                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3114
3115                 if (!$wgEnableScaryTranscluding)
3116                         return wfMsg('scarytranscludedisabled');
3117
3118                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3119                 // But we'll handle it generally anyway
3120                 if ( $title->getNamespace() ) {
3121                         // Use the canonical namespace, which should work anywhere
3122                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3123                 } else {
3124                         $articleName = $title->getDBkey();
3125                 }
3126
3127                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3128                 $url .= "?action=$action";
3129                 if (strlen($url) > 255)
3130                         return wfMsg('scarytranscludetoolong');
3131                 return $this->fetchScaryTemplateMaybeFromCache($url);
3132         }
3133
3134         function fetchScaryTemplateMaybeFromCache($url) {
3135                 global $wgTranscludeCacheExpiry;
3136                 $dbr =& wfGetDB(DB_SLAVE);
3137                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3138                                 array('tc_url' => $url));
3139                 if ($obj) {
3140                         $time = $obj->tc_time;
3141                         $text = $obj->tc_contents;
3142                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3143                                 return $text;
3144                         }
3145                 }
3146
3147                 $text = wfGetHTTP($url);
3148                 if (!$text)
3149                         return wfMsg('scarytranscludefailed', $url);
3150
3151                 $dbw =& wfGetDB(DB_MASTER);
3152                 $dbw->replace('transcache', array('tc_url'), array(
3153                         'tc_url' => $url,
3154                         'tc_time' => time(),
3155                         'tc_contents' => $text));
3156                 return $text;
3157         }
3158
3159
3160         /**
3161          * Triple brace replacement -- used for template arguments
3162          * @private
3163          */
3164         function argSubstitution( $matches ) {
3165                 $arg = trim( $matches['title'] );
3166                 $text = $matches['text'];
3167                 $inputArgs = end( $this->mArgStack );
3168
3169                 if ( array_key_exists( $arg, $inputArgs ) ) {
3170                         $text = $inputArgs[$arg];
3171                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3172                         $text = $matches['parts'][0];
3173                 }
3174
3175                 return $text;
3176         }
3177
3178         /**
3179          * Returns true if the function is allowed to include this entity
3180          * @private
3181          */
3182         function incrementIncludeCount( $dbk ) {
3183                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3184                         $this->mIncludeCount[$dbk] = 0;
3185                 }
3186                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3187                         return true;
3188                 } else {
3189                         return false;
3190                 }
3191         }
3192
3193         /**
3194          * Detect __NOGALLERY__ magic word and set a placeholder
3195          */
3196         function stripNoGallery( &$text ) {
3197                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3198                 # do not add TOC
3199                 $mw = MagicWord::get( MAG_NOGALLERY );
3200                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3201         }
3202
3203         /**
3204          * Detect __TOC__ magic word and set a placeholder
3205          */
3206         function stripToc( $text ) {
3207                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3208                 # do not add TOC
3209                 $mw = MagicWord::get( MAG_NOTOC );
3210                 if( $mw->matchAndRemove( $text ) ) {
3211                         $this->mShowToc = false;
3212                 }
3213
3214                 $mw = MagicWord::get( MAG_TOC );
3215                 if( $mw->match( $text ) ) {
3216                         $this->mShowToc = true;
3217                         $this->mForceTocPosition = true;
3218
3219                         // Set a placeholder. At the end we'll fill it in with the TOC.
3220                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3221
3222                         // Only keep the first one.
3223                         $text = $mw->replace( '', $text );
3224                 }
3225                 return $text;
3226         }
3227
3228         /**
3229          * This function accomplishes several tasks:
3230          * 1) Auto-number headings if that option is enabled
3231          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3232          * 3) Add a Table of contents on the top for users who have enabled the option
3233          * 4) Auto-anchor headings
3234          *
3235          * It loops through all headlines, collects the necessary data, then splits up the
3236          * string and re-inserts the newly formatted headlines.
3237          *
3238          * @param string $text
3239          * @param boolean $isMain
3240          * @private
3241          */
3242         function formatHeadings( $text, $isMain=true ) {
3243                 global $wgMaxTocLevel, $wgContLang;
3244
3245                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3246                 if( !$this->mTitle->userCanEdit() ) {
3247                         $showEditLink = 0;
3248                 } else {
3249                         $showEditLink = $this->mOptions->getEditSection();
3250                 }
3251
3252                 # Inhibit editsection links if requested in the page
3253                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3254                 if( $esw->matchAndRemove( $text ) ) {
3255                         $showEditLink = 0;
3256                 }
3257
3258                 # Get all headlines for numbering them and adding funky stuff like [edit]
3259                 # links - this is for later, but we need the number of headlines right now
3260                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3261
3262                 # if there are fewer than 4 headlines in the article, do not show TOC
3263                 # unless it's been explicitly enabled.
3264                 $enoughToc = $this->mShowToc &&
3265                         (($numMatches >= 4) || $this->mForceTocPosition);
3266
3267                 # Allow user to stipulate that a page should have a "new section"
3268                 # link added via __NEWSECTIONLINK__
3269                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3270                 if( $mw->matchAndRemove( $text ) )
3271                         $this->mOutput->setNewSection( true );
3272
3273                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3274                 # override above conditions and always show TOC above first header
3275                 $mw =& MagicWord::get( MAG_FORCETOC );
3276                 if ($mw->matchAndRemove( $text ) ) {
3277                         $this->mShowToc = true;
3278                         $enoughToc = true;
3279                 }
3280
3281                 # Never ever show TOC if no headers
3282                 if( $numMatches < 1 ) {
3283                         $enoughToc = false;
3284                 }
3285
3286                 # We need this to perform operations on the HTML
3287                 $sk =& $this->mOptions->getSkin();
3288
3289                 # headline counter
3290                 $headlineCount = 0;
3291                 $sectionCount = 0; # headlineCount excluding template sections
3292
3293                 # Ugh .. the TOC should have neat indentation levels which can be
3294                 # passed to the skin functions. These are determined here
3295                 $toc = '';
3296                 $full = '';
3297                 $head = array();
3298                 $sublevelCount = array();
3299                 $levelCount = array();
3300                 $toclevel = 0;
3301                 $level = 0;
3302                 $prevlevel = 0;
3303                 $toclevel = 0;
3304                 $prevtoclevel = 0;
3305
3306                 foreach( $matches[3] as $headline ) {
3307                         $istemplate = 0;
3308                         $templatetitle = '';
3309                         $templatesection = 0;
3310                         $numbering = '';
3311
3312                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3313                                 $istemplate = 1;
3314                                 $templatetitle = base64_decode($mat[1]);
3315                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3316                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3317                         }
3318
3319                         if( $toclevel ) {
3320                                 $prevlevel = $level;
3321                                 $prevtoclevel = $toclevel;
3322                         }
3323                         $level = $matches[1][$headlineCount];
3324
3325                         if( $doNumberHeadings || $enoughToc ) {
3326
3327                                 if ( $level > $prevlevel ) {
3328                                         # Increase TOC level
3329                                         $toclevel++;
3330                                         $sublevelCount[$toclevel] = 0;
3331                                         if( $toclevel<$wgMaxTocLevel ) {
3332                                                 $toc .= $sk->tocIndent();
3333                                         }
3334                                 }
3335                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3336                                         # Decrease TOC level, find level to jump to
3337
3338                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3339                                                 # Can only go down to level 1
3340                                                 $toclevel = 1;
3341                                         } else {
3342                                                 for ($i = $toclevel; $i > 0; $i--) {
3343                                                         if ( $levelCount[$i] == $level ) {
3344                                                                 # Found last matching level
3345                                                                 $toclevel = $i;
3346                                                                 break;
3347                                                         }
3348                                                         elseif ( $levelCount[$i] < $level ) {
3349                                                                 # Found first matching level below current level
3350                                                                 $toclevel = $i + 1;
3351                                                                 break;
3352                                                         }
3353                                                 }
3354                                         }
3355                                         if( $toclevel<$wgMaxTocLevel ) {
3356                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3357                                         }
3358                                 }
3359                                 else {
3360                                         # No change in level, end TOC line
3361                                         if( $toclevel<$wgMaxTocLevel ) {
3362                                                 $toc .= $sk->tocLineEnd();
3363                                         }
3364                                 }
3365
3366                                 $levelCount[$toclevel] = $level;
3367
3368                                 # count number of headlines for each level
3369                                 @$sublevelCount[$toclevel]++;
3370                                 $dot = 0;
3371                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3372                                         if( !empty( $sublevelCount[$i] ) ) {
3373                                                 if( $dot ) {
3374                                                         $numbering .= '.';
3375                                                 }
3376                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3377                                                 $dot = 1;
3378                                         }
3379                                 }
3380                         }
3381
3382                         # The canonized header is a version of the header text safe to use for links
3383                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3384                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3385                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3386
3387                         # Remove link placeholders by the link text.
3388                         #     <!--LINK number-->
3389                         # turns into
3390                         #     link text with suffix
3391                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3392                                                             "\$this->mLinkHolders['texts'][\$1]",
3393                                                             $canonized_headline );
3394                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3395                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3396                                                             $canonized_headline );
3397
3398                         # strip out HTML
3399                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3400                         $tocline = trim( $canonized_headline );
3401                         # Save headline for section edit hint before it's escaped
3402                         $headline_hint = trim( $canonized_headline );
3403                         $canonized_headline = Sanitizer::escapeId( $tocline );
3404                         $refers[$headlineCount] = $canonized_headline;
3405
3406                         # count how many in assoc. array so we can track dupes in anchors
3407                         @$refers[$canonized_headline]++;
3408                         $refcount[$headlineCount]=$refers[$canonized_headline];
3409
3410                         # Don't number the heading if it is the only one (looks silly)
3411                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3412                                 # the two are different if the line contains a link
3413                                 $headline=$numbering . ' ' . $headline;
3414                         }
3415
3416                         # Create the anchor for linking from the TOC to the section
3417                         $anchor = $canonized_headline;
3418                         if($refcount[$headlineCount] > 1 ) {
3419                                 $anchor .= '_' . $refcount[$headlineCount];
3420                         }
3421                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3422                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3423                         }
3424                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3425                                 if ( empty( $head[$headlineCount] ) ) {
3426                                         $head[$headlineCount] = '';
3427                                 }
3428                                 if( $istemplate )
3429                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3430                                 else
3431                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3432                         }
3433
3434                         # give headline the correct <h#> tag
3435                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3436
3437                         $headlineCount++;
3438                         if( !$istemplate )
3439                                 $sectionCount++;
3440                 }
3441
3442                 if( $enoughToc ) {
3443                         if( $toclevel<$wgMaxTocLevel ) {
3444                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3445                         }
3446                         $toc = $sk->tocList( $toc );
3447                 }
3448
3449                 # split up and insert constructed headlines
3450
3451                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3452                 $i = 0;
3453
3454                 foreach( $blocks as $block ) {
3455                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3456                                 # This is the [edit] link that appears for the top block of text when
3457                                 # section editing is enabled
3458
3459                                 # Disabled because it broke block formatting
3460                                 # For example, a bullet point in the top line
3461                                 # $full .= $sk->editSectionLink(0);
3462                         }
3463                         $full .= $block;
3464                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3465                                 # Top anchor now in skin
3466                                 $full = $full.$toc;
3467                         }
3468
3469                         if( !empty( $head[$i] ) ) {
3470                                 $full .= $head[$i];
3471                         }
3472                         $i++;
3473                 }
3474                 if( $this->mForceTocPosition ) {
3475                         return str_replace( '<!--MWTOC-->', $toc, $full );
3476                 } else {
3477                         return $full;
3478                 }
3479         }
3480
3481         /**
3482          * Return an HTML link for the "ISBN 123456" text
3483          * @private
3484          */
3485         function magicISBN( $text ) {
3486                 $fname = 'Parser::magicISBN';
3487                 wfProfileIn( $fname );
3488
3489                 $a = split( 'ISBN ', ' '.$text );
3490                 if ( count ( $a ) < 2 ) {
3491                         wfProfileOut( $fname );
3492                         return $text;
3493                 }
3494                 $text = substr( array_shift( $a ), 1);
3495                 $valid = '0123456789-Xx';
3496
3497                 foreach ( $a as $x ) {
3498                         # hack: don't replace inside thumbnail title/alt
3499                         # attributes
3500                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3501                                 $text .= "ISBN $x";
3502                                 continue;
3503                         }
3504
3505                         $isbn = $blank = '' ;
3506                         while ( ' ' == $x{0} ) {
3507                                 $blank .= ' ';
3508                                 $x = substr( $x, 1 );
3509                         }
3510                         if ( $x == '' ) { # blank isbn
3511                                 $text .= "ISBN $blank";
3512                                 continue;
3513                         }
3514                         while ( strstr( $valid, $x{0} ) != false ) {
3515                                 $isbn .= $x{0};
3516                                 $x = substr( $x, 1 );
3517                         }
3518                         $num = str_replace( '-', '', $isbn );
3519                         $num = str_replace( ' ', '', $num );
3520                         $num = str_replace( 'x', 'X', $num );
3521
3522                         if ( '' == $num ) {
3523                                 $text .= "ISBN $blank$x";
3524                         } else {
3525                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3526                                 $text .= '<a href="' .
3527                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3528                                         "\" class=\"internal\">ISBN $isbn</a>";
3529                                 $text .= $x;
3530                         }
3531                 }
3532                 wfProfileOut( $fname );
3533                 return $text;
3534         }
3535
3536         /**
3537          * Return an HTML link for the "RFC 1234" text
3538          *
3539          * @private
3540          * @param string $text     Text to be processed
3541          * @param string $keyword  Magic keyword to use (default RFC)
3542          * @param string $urlmsg   Interface message to use (default rfcurl)
3543          * @return string
3544          */
3545         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3546
3547                 $valid = '0123456789';
3548                 $internal = false;
3549
3550                 $a = split( $keyword, ' '.$text );
3551                 if ( count ( $a ) < 2 ) {
3552                         return $text;
3553                 }
3554                 $text = substr( array_shift( $a ), 1);
3555
3556                 /* Check if keyword is preceed by [[.
3557                  * This test is made here cause of the array_shift above
3558                  * that prevent the test to be done in the foreach.
3559                  */
3560                 if ( substr( $text, -2 ) == '[[' ) {
3561                         $internal = true;
3562                 }
3563
3564                 foreach ( $a as $x ) {
3565                         /* token might be empty if we have RFC RFC 1234 */
3566                         if ( $x=='' ) {
3567                                 $text.=$keyword;
3568                                 continue;
3569                                 }
3570
3571                         # hack: don't replace inside thumbnail title/alt
3572                         # attributes
3573                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3574                                 $text .= $keyword . $x;
3575                                 continue;
3576                         }
3577
3578                         $id = $blank = '' ;
3579
3580                         /** remove and save whitespaces in $blank */
3581                         while ( $x{0} == ' ' ) {
3582                                 $blank .= ' ';
3583                                 $x = substr( $x, 1 );
3584                         }
3585
3586                         /** remove and save the rfc number in $id */
3587                         while ( strstr( $valid, $x{0} ) != false ) {
3588                                 $id .= $x{0};
3589                                 $x = substr( $x, 1 );
3590                         }
3591
3592                         if ( $id == '' ) {
3593                                 /* call back stripped spaces*/
3594                                 $text .= $keyword.$blank.$x;
3595                         } elseif( $internal ) {
3596                                 /* normal link */
3597                                 $text .= $keyword.$id.$x;
3598                         } else {
3599                                 /* build the external link*/
3600                                 $url = wfMsg( $urlmsg, $id);
3601                                 $sk =& $this->mOptions->getSkin();
3602                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3603                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3604                         }
3605
3606                         /* Check if the next RFC keyword is preceed by [[ */
3607                         $internal = ( substr($x,-2) == '[[' );
3608                 }
3609                 return $text;
3610         }
3611
3612         /**
3613          * Transform wiki markup when saving a page by doing \r\n -> \n
3614          * conversion, substitting signatures, {{subst:}} templates, etc.
3615          *
3616          * @param string $text the text to transform
3617          * @param Title &$title the Title object for the current article
3618          * @param User &$user the User object describing the current user
3619          * @param ParserOptions $options parsing options
3620          * @param bool $clearState whether to clear the parser state first
3621          * @return string the altered wiki markup
3622          * @public
3623          */
3624         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3625                 $this->mOptions = $options;
3626                 $this->mTitle =& $title;
3627                 $this->mOutputType = OT_WIKI;
3628
3629                 if ( $clearState ) {
3630                         $this->clearState();
3631                 }
3632
3633                 $stripState = false;
3634                 $pairs = array(
3635                         "\r\n" => "\n",
3636                 );
3637                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3638                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3639                 $text = $this->pstPass2( $text, $stripState, $user );
3640                 $text = $this->unstrip( $text, $stripState );
3641                 $text = $this->unstripNoWiki( $text, $stripState );
3642                 return $text;
3643         }
3644
3645         /**
3646          * Pre-save transform helper function
3647          * @private
3648          */
3649         function pstPass2( $text, &$stripState, &$user ) {
3650                 global $wgContLang, $wgLocaltimezone;
3651
3652                 /* Note: This is the timestamp saved as hardcoded wikitext to
3653                  * the database, we use $wgContLang here in order to give
3654                  * everyone the same signature and use the default one rather
3655                  * than the one selected in each user's preferences.
3656                  */
3657                 if ( isset( $wgLocaltimezone ) ) {
3658                         $oldtz = getenv( 'TZ' );
3659                         putenv( 'TZ='.$wgLocaltimezone );
3660                 }
3661                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3662                   ' (' . date( 'T' ) . ')';
3663                 if ( isset( $wgLocaltimezone ) ) {
3664                         putenv( 'TZ='.$oldtz );
3665                 }
3666
3667                 # Variable replacement
3668                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3669                 $text = $this->replaceVariables( $text );
3670
3671                 # Strip out <nowiki> etc. added via replaceVariables
3672                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3673
3674                 # Signatures
3675                 $sigText = $this->getUserSig( $user );
3676                 $text = strtr( $text, array(
3677                         '~~~~~' => $d,
3678                         '~~~~' => "$sigText $d",
3679                         '~~~' => $sigText
3680                 ) );
3681
3682                 # Context links: [[|name]] and [[name (context)|]]
3683                 #
3684                 global $wgLegalTitleChars;
3685                 $tc = "[$wgLegalTitleChars]";
3686                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3687
3688                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3689                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3690
3691                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3692                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3693                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3694                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3695                 $context = '';
3696                 $t = $this->mTitle->getText();
3697                 if ( preg_match( $conpat, $t, $m ) ) {
3698                         $context = $m[2];
3699                 }
3700                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3701                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3702                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3703
3704                 if ( '' == $context ) {
3705                         $text = preg_replace( $p2, '[[\\1]]', $text );
3706                 } else {
3707                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3708                 }
3709
3710                 # Trim trailing whitespace
3711                 # MAG_END (__END__) tag allows for trailing
3712                 # whitespace to be deliberately included
3713                 $text = rtrim( $text );
3714                 $mw =& MagicWord::get( MAG_END );
3715                 $mw->matchAndRemove( $text );
3716
3717                 return $text;
3718         }
3719
3720         /**
3721          * Fetch the user's signature text, if any, and normalize to
3722          * validated, ready-to-insert wikitext.
3723          *
3724          * @param User $user
3725          * @return string
3726          * @private
3727          */
3728         function getUserSig( &$user ) {
3729                 $username = $user->getName();
3730                 $nickname = $user->getOption( 'nickname' );
3731                 $nickname = $nickname === '' ? $username : $nickname;
3732
3733                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3734                         # Sig. might contain markup; validate this
3735                         if( $this->validateSig( $nickname ) !== false ) {
3736                                 # Validated; clean up (if needed) and return it
3737                                 return $this->cleanSig( $nickname, true );
3738                         } else {
3739                                 # Failed to validate; fall back to the default
3740                                 $nickname = $username;
3741                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3742                         }
3743                 }
3744
3745                 # If we're still here, make it a link to the user page
3746                 $userpage = $user->getUserPage();
3747                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3748         }
3749
3750         /**
3751          * Check that the user's signature contains no bad XML
3752          *
3753          * @param string $text
3754          * @return mixed An expanded string, or false if invalid.
3755          */
3756         function validateSig( $text ) {
3757                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3758         }
3759
3760         /**
3761          * Clean up signature text
3762          *
3763          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3764          * 2) Substitute all transclusions
3765          *
3766          * @param string $text
3767          * @param $parsing Whether we're cleaning (preferences save) or parsing
3768          * @return string Signature text
3769          */
3770         function cleanSig( $text, $parsing = false ) {
3771                 global $wgTitle;
3772                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3773
3774                 $substWord = MagicWord::get( MAG_SUBST );
3775                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3776                 $substText = '{{' . $substWord->getSynonym( 0 );
3777
3778                 $text = preg_replace( $substRegex, $substText, $text );
3779                 $text = preg_replace( '/~{3,5}/', '', $text );
3780                 $text = $this->replaceVariables( $text );
3781
3782                 $this->clearState();
3783                 return $text;
3784         }
3785
3786         /**
3787          * Set up some variables which are usually set up in parse()
3788          * so that an external function can call some class members with confidence
3789          * @public
3790          */
3791         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3792                 $this->mTitle =& $title;
3793                 $this->mOptions = $options;
3794                 $this->mOutputType = $outputType;
3795                 if ( $clearState ) {
3796                         $this->clearState();
3797                 }
3798         }
3799
3800         /**
3801          * Transform a MediaWiki message by replacing magic variables.
3802          *
3803          * @param string $text the text to transform
3804          * @param ParserOptions $options  options
3805          * @return string the text with variables substituted
3806          * @public
3807          */
3808         function transformMsg( $text, $options ) {
3809                 global $wgTitle;
3810                 static $executing = false;
3811
3812                 $fname = "Parser::transformMsg";
3813
3814                 # Guard against infinite recursion
3815                 if ( $executing ) {
3816                         return $text;
3817                 }
3818                 $executing = true;
3819
3820                 wfProfileIn($fname);
3821
3822                 $this->mTitle = $wgTitle;
3823                 $this->mOptions = $options;
3824                 $this->mOutputType = OT_MSG;
3825                 $this->clearState();
3826                 $text = $this->replaceVariables( $text );
3827
3828                 $executing = false;
3829                 wfProfileOut($fname);
3830                 return $text;
3831         }
3832
3833         /**
3834          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3835          * The callback should have the following form:
3836          *    function myParserHook( $text, $params, &$parser ) { ... }
3837          *
3838          * Transform and return $text. Use $parser for any required context, e.g. use
3839          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3840          *
3841          * @public
3842          *
3843          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3844          * @param mixed $callback The callback function (and object) to use for the tag
3845          *
3846          * @return The old value of the mTagHooks array associated with the hook
3847          */
3848         function setHook( $tag, $callback ) {
3849                 $tag = strtolower( $tag );
3850                 $oldVal = @$this->mTagHooks[$tag];
3851                 $this->mTagHooks[$tag] = $callback;
3852
3853                 return $oldVal;
3854         }
3855
3856         /**
3857          * Create a function, e.g. {{sum:1|2|3}}
3858          * The callback function should have the form:
3859          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3860          *
3861          * The callback may either return the text result of the function, or an array with the text
3862          * in element 0, and a number of flags in the other elements. The names of the flags are
3863          * specified in the keys. Valid flags are:
3864          *   found                     The text returned is valid, stop processing the template. This
3865          *                             is on by default.
3866          *   nowiki                    Wiki markup in the return value should be escaped
3867          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3868          *   noargs                    Don't replace triple-brace arguments in the return value
3869          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3870          *
3871          * @public
3872          *
3873          * @param string $name The function name. Function names are case-insensitive.
3874          * @param mixed $callback The callback function (and object) to use
3875          *
3876          * @return The old callback function for this name, if any
3877          */
3878         function setFunctionHook( $name, $callback ) {
3879                 $name = strtolower( $name );
3880                 $oldVal = @$this->mFunctionHooks[$name];
3881                 $this->mFunctionHooks[$name] = $callback;
3882                 return $oldVal;
3883         }
3884
3885         /**
3886          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3887          * Placeholders created in Skin::makeLinkObj()
3888          * Returns an array of links found, indexed by PDBK:
3889          *  0 - broken
3890          *  1 - normal link
3891          *  2 - stub
3892          * $options is a bit field, RLH_FOR_UPDATE to select for update
3893          */
3894         function replaceLinkHolders( &$text, $options = 0 ) {
3895                 global $wgUser;
3896                 global $wgOutputReplace;
3897
3898                 $fname = 'Parser::replaceLinkHolders';
3899                 wfProfileIn( $fname );
3900
3901                 $pdbks = array();
3902                 $colours = array();
3903                 $sk =& $this->mOptions->getSkin();
3904                 $linkCache =& LinkCache::singleton();
3905
3906                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3907                         wfProfileIn( $fname.'-check' );
3908                         $dbr =& wfGetDB( DB_SLAVE );
3909                         $page = $dbr->tableName( 'page' );
3910                         $threshold = $wgUser->getOption('stubthreshold');
3911
3912                         # Sort by namespace
3913                         asort( $this->mLinkHolders['namespaces'] );
3914
3915                         # Generate query
3916                         $query = false;
3917                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3918                                 # Make title object
3919                                 $title = $this->mLinkHolders['titles'][$key];
3920
3921                                 # Skip invalid entries.
3922                                 # Result will be ugly, but prevents crash.
3923                                 if ( is_null( $title ) ) {
3924                                         continue;
3925                                 }
3926                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3927
3928                                 # Check if it's a static known link, e.g. interwiki
3929                                 if ( $title->isAlwaysKnown() ) {
3930                                         $colours[$pdbk] = 1;
3931                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3932                                         $colours[$pdbk] = 1;
3933                                         $this->mOutput->addLink( $title, $id );
3934                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3935                                         $colours[$pdbk] = 0;
3936                                 } else {
3937                                         # Not in the link cache, add it to the query
3938                                         if ( !isset( $current ) ) {
3939                                                 $current = $ns;
3940                                                 $query =  "SELECT page_id, page_namespace, page_title";
3941                                                 if ( $threshold > 0 ) {
3942                                                         $query .= ', page_len, page_is_redirect';
3943                                                 }
3944                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3945                                         } elseif ( $current != $ns ) {
3946                                                 $current = $ns;
3947                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3948                                         } else {
3949                                                 $query .= ', ';
3950                                         }
3951
3952                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3953                                 }
3954                         }
3955                         if ( $query ) {
3956                                 $query .= '))';
3957                                 if ( $options & RLH_FOR_UPDATE ) {
3958                                         $query .= ' FOR UPDATE';
3959                                 }
3960
3961                                 $res = $dbr->query( $query, $fname );
3962
3963                                 # Fetch data and form into an associative array
3964                                 # non-existent = broken
3965                                 # 1 = known
3966                                 # 2 = stub
3967                                 while ( $s = $dbr->fetchObject($res) ) {
3968                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3969                                         $pdbk = $title->getPrefixedDBkey();
3970                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3971                                         $this->mOutput->addLink( $title, $s->page_id );
3972
3973                                         if ( $threshold >  0 ) {
3974                                                 $size = $s->page_len;
3975                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3976                                                         $colours[$pdbk] = 1;
3977                                                 } else {
3978                                                         $colours[$pdbk] = 2;
3979                                                 }
3980                                         } else {
3981                                                 $colours[$pdbk] = 1;
3982                                         }
3983                                 }
3984                         }
3985                         wfProfileOut( $fname.'-check' );
3986
3987                         # Construct search and replace arrays
3988                         wfProfileIn( $fname.'-construct' );
3989                         $wgOutputReplace = array();
3990                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3991                                 $pdbk = $pdbks[$key];
3992                                 $searchkey = "<!--LINK $key-->";
3993                                 $title = $this->mLinkHolders['titles'][$key];
3994                                 if ( empty( $colours[$pdbk] ) ) {
3995                                         $linkCache->addBadLinkObj( $title );
3996                                         $colours[$pdbk] = 0;
3997                                         $this->mOutput->addLink( $title, 0 );
3998                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3999                                                                         $this->mLinkHolders['texts'][$key],
4000                                                                         $this->mLinkHolders['queries'][$key] );
4001                                 } elseif ( $colours[$pdbk] == 1 ) {
4002                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
4003                                                                         $this->mLinkHolders['texts'][$key],
4004                                                                         $this->mLinkHolders['queries'][$key] );
4005                                 } elseif ( $colours[$pdbk] == 2 ) {
4006                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
4007                                                                         $this->mLinkHolders['texts'][$key],
4008                                                                         $this->mLinkHolders['queries'][$key] );
4009                                 }
4010                         }
4011                         wfProfileOut( $fname.'-construct' );
4012
4013                         # Do the thing
4014                         wfProfileIn( $fname.'-replace' );
4015
4016                         $text = preg_replace_callback(
4017                                 '/(<!--LINK .*?-->)/',
4018                                 "wfOutputReplaceMatches",
4019                                 $text);
4020
4021                         wfProfileOut( $fname.'-replace' );
4022                 }
4023
4024                 # Now process interwiki link holders
4025                 # This is quite a bit simpler than internal links
4026                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4027                         wfProfileIn( $fname.'-interwiki' );
4028                         # Make interwiki link HTML
4029                         $wgOutputReplace = array();
4030                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4031                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4032                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4033                         }
4034
4035                         $text = preg_replace_callback(
4036                                 '/<!--IWLINK (.*?)-->/',
4037                                 "wfOutputReplaceMatches",
4038                                 $text );
4039                         wfProfileOut( $fname.'-interwiki' );
4040                 }
4041
4042                 wfProfileOut( $fname );
4043                 return $colours;
4044         }
4045
4046         /**
4047          * Replace <!--LINK--> link placeholders with plain text of links
4048          * (not HTML-formatted).
4049          * @param string $text
4050          * @return string
4051          */
4052         function replaceLinkHoldersText( $text ) {
4053                 $fname = 'Parser::replaceLinkHoldersText';
4054                 wfProfileIn( $fname );
4055
4056                 $text = preg_replace_callback(
4057                         '/<!--(LINK|IWLINK) (.*?)-->/',
4058                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4059                         $text );
4060
4061                 wfProfileOut( $fname );
4062                 return $text;
4063         }
4064
4065         /**
4066          * @param array $matches
4067          * @return string
4068          * @private
4069          */
4070         function replaceLinkHoldersTextCallback( $matches ) {
4071                 $type = $matches[1];
4072                 $key  = $matches[2];
4073                 if( $type == 'LINK' ) {
4074                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4075                                 return $this->mLinkHolders['texts'][$key];
4076                         }
4077                 } elseif( $type == 'IWLINK' ) {
4078                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4079                                 return $this->mInterwikiLinkHolders['texts'][$key];
4080                         }
4081                 }
4082                 return $matches[0];
4083         }
4084
4085         /**
4086          * Tag hook handler for 'pre'.
4087          */
4088         function renderPreTag( $text, $attribs, $parser ) {
4089                 // Backwards-compatibility hack
4090                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4091
4092                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4093                 return wfOpenElement( 'pre', $attribs ) .
4094                         wfEscapeHTMLTagsOnly( $content ) .
4095                         '</pre>';
4096         }
4097
4098         /**
4099          * Renders an image gallery from a text with one line per image.
4100          * text labels may be given by using |-style alternative text. E.g.
4101          *   Image:one.jpg|The number "1"
4102          *   Image:tree.jpg|A tree
4103          * given as text will return the HTML of a gallery with two images,
4104          * labeled 'The number "1"' and
4105          * 'A tree'.
4106          */
4107         function renderImageGallery( $text ) {
4108                 $ig = new ImageGallery();
4109                 $ig->setShowBytes( false );
4110                 $ig->setShowFilename( false );
4111                 $ig->setParsing();
4112                 $lines = explode( "\n", $text );
4113
4114                 foreach ( $lines as $line ) {
4115                         # match lines like these:
4116                         # Image:someimage.jpg|This is some image
4117                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4118                         # Skip empty lines
4119                         if ( count( $matches ) == 0 ) {
4120                                 continue;
4121                         }
4122                         $nt =& Title::newFromText( $matches[1] );
4123                         if( is_null( $nt ) ) {
4124                                 # Bogus title. Ignore these so we don't bomb out later.
4125                                 continue;
4126                         }
4127                         if ( isset( $matches[3] ) ) {
4128                                 $label = $matches[3];
4129                         } else {
4130                                 $label = '';
4131                         }
4132
4133                         $pout = $this->parse( $label,
4134                                 $this->mTitle,
4135                                 $this->mOptions,
4136                                 false, // Strip whitespace...?
4137                                 false  // Don't clear state!
4138                         );
4139                         $html = $pout->getText();
4140
4141                         $ig->add( new Image( $nt ), $html );
4142
4143                         # Only add real images (bug #5586)
4144                         if ( $nt->getNamespace() == NS_IMAGE ) {
4145                                 $this->mOutput->addImage( $nt->getDBkey() );
4146                         }
4147                 }
4148                 return $ig->toHTML();
4149         }
4150
4151         /**
4152          * Parse image options text and use it to make an image
4153          */
4154         function makeImage( &$nt, $options ) {
4155                 global $wgUseImageResize;
4156
4157                 $align = '';
4158
4159                 # Check if the options text is of the form "options|alt text"
4160                 # Options are:
4161                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4162                 #  * left               no resizing, just left align. label is used for alt= only
4163                 #  * right              same, but right aligned
4164                 #  * none               same, but not aligned
4165                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4166                 #  * center             center the image
4167                 #  * framed             Keep original image size, no magnify-button.
4168
4169                 $part = explode( '|', $options);
4170
4171                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4172                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4173                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4174                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4175                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4176                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4177                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4178                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4179                 $caption = '';
4180
4181                 $width = $height = $framed = $thumb = false;
4182                 $manual_thumb = '' ;
4183
4184                 foreach( $part as $key => $val ) {
4185                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4186                                 $thumb=true;
4187                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4188                                 # use manually specified thumbnail
4189                                 $thumb=true;
4190                                 $manual_thumb = $match;
4191                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4192                                 # remember to set an alignment, don't render immediately
4193                                 $align = 'right';
4194                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4195                                 # remember to set an alignment, don't render immediately
4196                                 $align = 'left';
4197                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4198                                 # remember to set an alignment, don't render immediately
4199                                 $align = 'center';
4200                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4201                                 # remember to set an alignment, don't render immediately
4202                                 $align = 'none';
4203                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4204                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4205                                 # $match is the image width in pixels
4206                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4207                                         $width = intval( $m[1] );
4208                                         $height = intval( $m[2] );
4209                                 } else {
4210                                         $width = intval($match);
4211                                 }
4212                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4213                                 $framed=true;
4214                         } else {
4215                                 $caption = $val;
4216                         }
4217                 }
4218                 # Strip bad stuff out of the alt text
4219                 $alt = $this->replaceLinkHoldersText( $caption );
4220
4221                 # make sure there are no placeholders in thumbnail attributes
4222                 # that are later expanded to html- so expand them now and
4223                 # remove the tags
4224                 $alt = $this->unstrip($alt, $this->mStripState);
4225                 $alt = Sanitizer::stripAllTags( $alt );
4226
4227                 # Linker does the rest
4228                 $sk =& $this->mOptions->getSkin();
4229                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4230         }
4231
4232         /**
4233          * Set a flag in the output object indicating that the content is dynamic and
4234          * shouldn't be cached.
4235          */
4236         function disableCache() {
4237                 wfDebug( "Parser output marked as uncacheable.\n" );
4238                 $this->mOutput->mCacheTime = -1;
4239         }
4240
4241         /**#@+
4242          * Callback from the Sanitizer for expanding items found in HTML attribute
4243          * values, so they can be safely tested and escaped.
4244          * @param string $text
4245          * @param array $args
4246          * @return string
4247          * @private
4248          */
4249         function attributeStripCallback( &$text, $args ) {
4250                 $text = $this->replaceVariables( $text, $args );
4251                 $text = $this->unstripForHTML( $text );
4252                 return $text;
4253         }
4254
4255         function unstripForHTML( $text ) {
4256                 $text = $this->unstrip( $text, $this->mStripState );
4257                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4258                 return $text;
4259         }
4260         /**#@-*/
4261
4262         /**#@+
4263          * Accessor/mutator
4264          */
4265         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4266         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4267         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4268         /**#@-*/
4269
4270         /**#@+
4271          * Accessor
4272          */
4273         function getTags() { return array_keys( $this->mTagHooks ); }
4274         /**#@-*/
4275
4276
4277         /**
4278          * Break wikitext input into sections, and either pull or replace
4279          * some particular section's text.
4280          *
4281          * External callers should use the getSection and replaceSection methods.
4282          *
4283          * @param $text Page wikitext
4284          * @param $section Numbered section. 0 pulls the text before the first
4285          *                 heading; other numbers will pull the given section
4286          *                 along with its lower-level subsections.
4287          * @param $mode One of "get" or "replace"
4288          * @param $newtext Replacement text for section data.
4289          * @return string for "get", the extracted section text.
4290          *                for "replace", the whole page with the section replaced.
4291          */
4292         private function extractSections( $text, $section, $mode, $newtext='' ) {
4293                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4294                 # comments to be stripped as well)
4295                 $striparray = array();
4296
4297                 $oldOutputType = $this->mOutputType;
4298                 $oldOptions = $this->mOptions;
4299                 $this->mOptions = new ParserOptions();
4300                 $this->mOutputType = OT_WIKI;
4301
4302                 $striptext = $this->strip( $text, $striparray, true );
4303
4304                 $this->mOutputType = $oldOutputType;
4305                 $this->mOptions = $oldOptions;
4306
4307                 # now that we can be sure that no pseudo-sections are in the source,
4308                 # split it up by section
4309                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4310                 $comment = "(?:$uniq-!--.*?QINU)";
4311                 $secs = preg_split(
4312                 /*
4313                         "/
4314                         ^(
4315                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4316                         (?:
4317                                 (=+) # Should this be limited to 6?
4318                                 .+?  # Section title...
4319                                 \\2  # Ending = count must match start
4320                         |
4321                                 ^
4322                                 <h([1-6])\b.*?>
4323                                 .*?
4324                                 <\/h\\3\s*>
4325                         )
4326                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4327                         )$
4328                         /mix",
4329                 */
4330                         "/
4331                         (
4332                                 ^
4333                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4334                                 (=+) # Should this be limited to 6?
4335                                 .+?  # Section title...
4336                                 \\2  # Ending = count must match start
4337                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4338                                 $
4339                         |
4340                                 <h([1-6])\b.*?>
4341                                 .*?
4342                                 <\/h\\3\s*>
4343                         )
4344                         /mix",
4345                         $striptext, -1,
4346                         PREG_SPLIT_DELIM_CAPTURE);
4347
4348                 if( $mode == "get" ) {
4349                         if( $section == 0 ) {
4350                                 // "Section 0" returns the content before any other section.
4351                                 $rv = $secs[0];
4352                         } else {
4353                                 $rv = "";
4354                         }
4355                 } elseif( $mode == "replace" ) {
4356                         if( $section == 0 ) {
4357                                 $rv = $newtext . "\n\n";
4358                                 $remainder = true;
4359                         } else {
4360                                 $rv = $secs[0];
4361                                 $remainder = false;
4362                         }
4363                 }
4364                 $count = 0;
4365                 $sectionLevel = 0;
4366                 for( $index = 1; $index < count( $secs ); ) {
4367                         $headerLine = $secs[$index++];
4368                         if( $secs[$index] ) {
4369                                 // A wiki header
4370                                 $headerLevel = strlen( $secs[$index++] );
4371                         } else {
4372                                 // An HTML header
4373                                 $index++;
4374                                 $headerLevel = intval( $secs[$index++] );
4375                         }
4376                         $content = $secs[$index++];
4377
4378                         $count++;
4379                         if( $mode == "get" ) {
4380                                 if( $count == $section ) {
4381                                         $rv = $headerLine . $content;
4382                                         $sectionLevel = $headerLevel;
4383                                 } elseif( $count > $section ) {
4384                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4385                                                 $rv .= $headerLine . $content;
4386                                         } else {
4387                                                 // Broke out to a higher-level section
4388                                                 break;
4389                                         }
4390                                 }
4391                         } elseif( $mode == "replace" ) {
4392                                 if( $count < $section ) {
4393                                         $rv .= $headerLine . $content;
4394                                 } elseif( $count == $section ) {
4395                                         $rv .= $newtext . "\n\n";
4396                                         $sectionLevel = $headerLevel;
4397                                 } elseif( $count > $section ) {
4398                                         if( $headerLevel <= $sectionLevel ) {
4399                                                 // Passed the section's sub-parts.
4400                                                 $remainder = true;
4401                                         }
4402                                         if( $remainder ) {
4403                                                 $rv .= $headerLine . $content;
4404                                         }
4405                                 }
4406                         }
4407                 }
4408                 # reinsert stripped tags
4409                 $rv = $this->unstrip( $rv, $striparray );
4410                 $rv = $this->unstripNoWiki( $rv, $striparray );
4411                 $rv = trim( $rv );
4412                 return $rv;
4413         }
4414
4415         /**
4416          * This function returns the text of a section, specified by a number ($section).
4417          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4418          * the first section before any such heading (section 0).
4419          *
4420          * If a section contains subsections, these are also returned.
4421          *
4422          * @param $text String: text to look in
4423          * @param $section Integer: section number
4424          * @return string text of the requested section
4425          */
4426         function getSection( $text, $section ) {
4427                 return $this->extractSections( $text, $section, "get" );
4428         }
4429
4430         function replaceSection( $oldtext, $section, $text ) {
4431                 return $this->extractSections( $oldtext, $section, "replace", $text );
4432         }
4433
4434 }
4435
4436 /**
4437  * @todo document
4438  * @package MediaWiki
4439  */
4440 class ParserOutput
4441 {
4442         var $mText,             # The output text
4443                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4444                 $mCategories,       # Map of category names to sort keys
4445                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4446                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4447                 $mVersion,          # Compatibility check
4448                 $mTitleText,        # title text of the chosen language variant
4449                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4450                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4451                 $mImages,           # DB keys of the images used, in the array key only
4452                 $mExternalLinks,    # External link URLs, in the key only
4453                 $mHTMLtitle,            # Display HTML title
4454                 $mSubtitle,                     # Additional subtitle
4455                 $mNewSection,           # Show a new section link?
4456                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4457
4458         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4459                 $containsOldMagic = false, $titletext = '' )
4460         {
4461                 $this->mText = $text;
4462                 $this->mLanguageLinks = $languageLinks;
4463                 $this->mCategories = $categoryLinks;
4464                 $this->mContainsOldMagic = $containsOldMagic;
4465                 $this->mCacheTime = '';
4466                 $this->mVersion = MW_PARSER_VERSION;
4467                 $this->mTitleText = $titletext;
4468                 $this->mLinks = array();
4469                 $this->mTemplates = array();
4470                 $this->mImages = array();
4471                 $this->mExternalLinks = array();
4472                 $this->mHTMLtitle = "" ;
4473                 $this->mSubtitle = "" ;
4474                 $this->mNewSection = false;
4475                 $this->mNoGallery = false;
4476         }
4477
4478         function getText()                   { return $this->mText; }
4479         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4480         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4481         function &getCategories()            { return $this->mCategories; }
4482         function getCacheTime()              { return $this->mCacheTime; }
4483         function getTitleText()              { return $this->mTitleText; }
4484         function &getLinks()                 { return $this->mLinks; }
4485         function &getTemplates()             { return $this->mTemplates; }
4486         function &getImages()                { return $this->mImages; }
4487         function &getExternalLinks()         { return $this->mExternalLinks; }
4488         function getNoGallery()              { return $this->mNoGallery; }
4489
4490         function containsOldMagic()          { return $this->mContainsOldMagic; }
4491         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4492         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4493         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4494         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4495         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4496         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4497
4498         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4499         function addImage( $name )           { $this->mImages[$name] = 1; }
4500         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4501         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4502
4503         function setNewSection( $value ) {
4504                 $this->mNewSection = (bool)$value;
4505         }
4506         function getNewSection() {
4507                 return (bool)$this->mNewSection;
4508         }
4509
4510         function addLink( $title, $id ) {
4511                 $ns = $title->getNamespace();
4512                 $dbk = $title->getDBkey();
4513                 if ( !isset( $this->mLinks[$ns] ) ) {
4514                         $this->mLinks[$ns] = array();
4515                 }
4516                 $this->mLinks[$ns][$dbk] = $id;
4517         }
4518
4519         function addTemplate( $title, $id ) {
4520                 $ns = $title->getNamespace();
4521                 $dbk = $title->getDBkey();
4522                 if ( !isset( $this->mTemplates[$ns] ) ) {
4523                         $this->mTemplates[$ns] = array();
4524                 }
4525                 $this->mTemplates[$ns][$dbk] = $id;
4526         }
4527
4528         /**
4529          * Return true if this cached output object predates the global or
4530          * per-article cache invalidation timestamps, or if it comes from
4531          * an incompatible older version.
4532          *
4533          * @param string $touched the affected article's last touched timestamp
4534          * @return bool
4535          * @public
4536          */
4537         function expired( $touched ) {
4538                 global $wgCacheEpoch;
4539                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4540                        $this->getCacheTime() < $touched ||
4541                        $this->getCacheTime() <= $wgCacheEpoch ||
4542                        !isset( $this->mVersion ) ||
4543                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4544         }
4545 }
4546
4547 /**
4548  * Set options of the Parser
4549  * @todo document
4550  * @package MediaWiki
4551  */
4552 class ParserOptions
4553 {
4554         # All variables are private
4555         var $mUseTeX;                    # Use texvc to expand <math> tags
4556         var $mUseDynamicDates;           # Use DateFormatter to format dates
4557         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4558         var $mAllowExternalImages;       # Allow external images inline
4559         var $mAllowExternalImagesFrom;   # If not, any exception?
4560         var $mSkin;                      # Reference to the preferred skin
4561         var $mDateFormat;                # Date format index
4562         var $mEditSection;               # Create "edit section" links
4563         var $mNumberHeadings;            # Automatically number headings
4564         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4565         var $mTidy;                      # Ask for tidy cleanup
4566         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4567
4568         function getUseTeX()                        { return $this->mUseTeX; }
4569         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4570         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4571         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4572         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4573         function &getSkin()                         { return $this->mSkin; }
4574         function getDateFormat()                    { return $this->mDateFormat; }
4575         function getEditSection()                   { return $this->mEditSection; }
4576         function getNumberHeadings()                { return $this->mNumberHeadings; }
4577         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4578         function getTidy()                          { return $this->mTidy; }
4579         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4580
4581         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4582         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4583         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4584         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4585         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4586         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4587         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4588         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4589         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4590         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4591         function setSkin( &$x ) { $this->mSkin =& $x; }
4592         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4593
4594         function ParserOptions() {
4595                 global $wgUser;
4596                 $this->initialiseFromUser( $wgUser );
4597         }
4598
4599         /**
4600          * Get parser options
4601          * @static
4602          */
4603         function newFromUser( &$user ) {
4604                 $popts = new ParserOptions;
4605                 $popts->initialiseFromUser( $user );
4606                 return $popts;
4607         }
4608
4609         /** Get user options */
4610         function initialiseFromUser( &$userInput ) {
4611                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4612                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4613                 $fname = 'ParserOptions::initialiseFromUser';
4614                 wfProfileIn( $fname );
4615                 if ( !$userInput ) {
4616                         $user = new User;
4617                         $user->setLoaded( true );
4618                 } else {
4619                         $user =& $userInput;
4620                 }
4621
4622                 $this->mUseTeX = $wgUseTeX;
4623                 $this->mUseDynamicDates = $wgUseDynamicDates;
4624                 $this->mInterwikiMagic = $wgInterwikiMagic;
4625                 $this->mAllowExternalImages = $wgAllowExternalImages;
4626                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4627                 wfProfileIn( $fname.'-skin' );
4628                 $this->mSkin =& $user->getSkin();
4629                 wfProfileOut( $fname.'-skin' );
4630                 $this->mDateFormat = $user->getOption( 'date' );
4631                 $this->mEditSection = true;
4632                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4633                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4634                 $this->mTidy = false;
4635                 $this->mInterfaceMessage = false;
4636                 wfProfileOut( $fname );
4637         }
4638 }
4639
4640 /**
4641  * Callback function used by Parser::replaceLinkHolders()
4642  * to substitute link placeholders.
4643  */
4644 function &wfOutputReplaceMatches( $matches ) {
4645         global $wgOutputReplace;
4646         return $wgOutputReplace[$matches[1]];
4647 }
4648
4649 /**
4650  * Return the total number of articles
4651  */
4652 function wfNumberOfArticles() {
4653         global $wgNumberOfArticles;
4654
4655         wfLoadSiteStats();
4656         return $wgNumberOfArticles;
4657 }
4658
4659 /**
4660  * Return the number of files
4661  */
4662 function wfNumberOfFiles() {
4663         $fname = 'wfNumberOfFiles';
4664
4665         wfProfileIn( $fname );
4666         $dbr =& wfGetDB( DB_SLAVE );
4667         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4668         wfProfileOut( $fname );
4669
4670         return $numImages;
4671 }
4672
4673 /**
4674  * Return the number of user accounts
4675  * @return integer
4676  */
4677 function wfNumberOfUsers() {
4678         wfProfileIn( 'wfNumberOfUsers' );
4679         $dbr =& wfGetDB( DB_SLAVE );
4680         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4681         wfProfileOut( 'wfNumberOfUsers' );
4682         return (int)$count;
4683 }
4684
4685 /**
4686  * Return the total number of pages
4687  * @return integer
4688  */
4689 function wfNumberOfPages() {
4690         wfProfileIn( 'wfNumberOfPages' );
4691         $dbr =& wfGetDB( DB_SLAVE );
4692         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4693         wfProfileOut( 'wfNumberOfPages' );
4694         return (int)$count;
4695 }
4696
4697 /**
4698  * Return the total number of admins
4699  *
4700  * @return integer
4701  */
4702 function wfNumberOfAdmins() {
4703         static $admins = -1;
4704         wfProfileIn( 'wfNumberOfAdmins' );
4705         if( $admins == -1 ) {
4706                 $dbr =& wfGetDB( DB_SLAVE );
4707                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4708         }
4709         wfProfileOut( 'wfNumberOfAdmins' );
4710         return (int)$admins;
4711 }
4712
4713 /**
4714  * Count the number of pages in a particular namespace
4715  *
4716  * @param $ns Namespace
4717  * @return integer
4718  */
4719 function wfPagesInNs( $ns ) {
4720         static $pageCount = array();
4721         wfProfileIn( 'wfPagesInNs' );
4722         if( !isset( $pageCount[$ns] ) ) {
4723                 $dbr =& wfGetDB( DB_SLAVE );
4724                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4725         }
4726         wfProfileOut( 'wfPagesInNs' );
4727         return (int)$pageCount[$ns];
4728 }
4729
4730 /**
4731  * Get various statistics from the database
4732  * @private
4733  */
4734 function wfLoadSiteStats() {
4735         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4736         $fname = 'wfLoadSiteStats';
4737
4738         if ( -1 != $wgNumberOfArticles ) return;
4739         $dbr =& wfGetDB( DB_SLAVE );
4740         $s = $dbr->selectRow( 'site_stats',
4741                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4742                 array( 'ss_row_id' => 1 ), $fname
4743         );
4744
4745         if ( $s === false ) {
4746                 return;
4747         } else {
4748                 $wgTotalViews = $s->ss_total_views;
4749                 $wgTotalEdits = $s->ss_total_edits;
4750                 $wgNumberOfArticles = $s->ss_good_articles;
4751         }
4752 }
4753
4754 /**
4755  * Escape html tags
4756  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4757  *
4758  * @param $in String: text that might contain HTML tags.
4759  * @return string Escaped string
4760  */
4761 function wfEscapeHTMLTagsOnly( $in ) {
4762         return str_replace(
4763                 array( '"', '>', '<' ),
4764                 array( '&quot;', '&gt;', '&lt;' ),
4765                 $in );
4766 }
4767
4768 ?>