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', '1.5.0' );
  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 # prefix for escaping, used in two functions at least
  47 define( 'UNIQ_PREFIX', 'NaodW29');
  48
  49 # Constants needed for external link processing
  50 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  51 # Everything except bracket, space, or control characters
  52 define( 'EXT_LINK_URL_CLASS', '[^]<>"\\x00-\\x20\\x7F]' );
  53 # Including space
  54 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  55 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  56 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  57 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  58 define( 'EXT_IMAGE_REGEX',
  59         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  60         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  61         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  62 );
  63
  64 /**
  65  * PHP Parser
  66  *
  67  * Processes wiki markup
  68  *
  69  * <pre>
  70  * There are three main entry points into the Parser class:
  71  * parse()
  72  *   produces HTML output
  73  * preSaveTransform().
  74  *   produces altered wiki markup.
  75  * transformMsg()
  76  *   performs brace substitution on MediaWiki messages
  77  *
  78  * Globals used:
  79  *    objects:   $wgLang, $wgLinkCache
  80  *
  81  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  82  *
  83  * settings:
  84  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  85  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  86  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @package MediaWiki
  92  */
  93 class Parser
  94 {
  95         /**#@+
  96          * @access private
  97          */
  98         # Persistent:
  99         var $mTagHooks;
 100
 101         # Cleared with clearState():
 102         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 103         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 104         var $mInterwikiLinkHolders, $mLinkHolders;
 105
 106         # Temporary:
 107         var $mOptions, $mTitle, $mOutputType,
 108             $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         var $mIWTransData = array();
 114
 115         /**#@-*/
 116
 117         /**
 118          * Constructor
 119          *
 120          * @access public
 121          */
 122         function Parser() {
 123                 $this->mTemplates = array();
 124                 $this->mTemplatePath = array();
 125                 $this->mTagHooks = array();
 126                 $this->clearState();
 127         }
 128
 129         /**
 130          * Clear Parser state
 131          *
 132          * @access private
 133          */
 134         function clearState() {
 135                 $this->mOutput = new ParserOutput;
 136                 $this->mAutonumber = 0;
 137                 $this->mLastSection = '';
 138                 $this->mDTopen = false;
 139                 $this->mVariables = false;
 140                 $this->mIncludeCount = array();
 141                 $this->mStripState = array();
 142                 $this->mArgStack = array();
 143                 $this->mInPre = false;
 144                 $this->mInterwikiLinkHolders = array(
 145                         'texts' => array(),
 146                         'titles' => array()
 147                 );
 148                 $this->mLinkHolders = array(
 149                         'namespaces' => array(),
 150                         'dbkeys' => array(),
 151                         'queries' => array(),
 152                         'texts' => array(),
 153                         'titles' => array()
 154                 );
 155         }
 156
 157         /**
 158          * First pass--just handle <nowiki> sections, pass the rest off
 159          * to internalParse() which does all the real work.
 160          *
 161          * @access private
 162          * @param string $text Text we want to parse
 163          * @param Title &$title A title object
 164          * @param array $options
 165          * @param boolean $linestart
 166          * @param boolean $clearState
 167          * @return ParserOutput a ParserOutput
 168          */
 169         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 170                 global $wgUseTidy, $wgContLang;
 171                 $fname = 'Parser::parse';
 172                 wfProfileIn( $fname );
 173
 174                 if ( $clearState ) {
 175                         $this->clearState();
 176                 }
 177
 178                 $this->mOptions = $options;
 179                 $this->mTitle =& $title;
 180                 $this->mOutputType = OT_HTML;
 181
 182                 $this->mStripState = NULL;
 183
 184                 //$text = $this->strip( $text, $this->mStripState );
 185                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 186                 $x =& $this->mStripState;
 187
 188                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 189                 $text = $this->strip( $text, $x );
 190                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 191
 192                 $text = $this->internalParse( $text );
 193
 194                 $text = $this->unstrip( $text, $this->mStripState );
 195
 196                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 197                 $fixtags = array(
 198                         # french spaces, last one Guillemet-left
 199                         # only if there is something before the space
 200                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 201                         # french spaces, Guillemet-right
 202                         '/(\\302\\253) /' => '\\1&nbsp;',
 203                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 204                 );
 205                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 206
 207                 # only once and last
 208                 $text = $this->doBlockLevels( $text, $linestart );
 209
 210                 $this->replaceLinkHolders( $text );
 211
 212                 # the position of the convert() call should not be changed. it
 213                 # assumes that the links are all replaces and the only thing left
 214                 # is the <nowiki> mark.
 215                 $text = $wgContLang->convert($text);
 216                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 217
 218                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 219
 220                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 221
 222                 $text = Sanitizer::normalizeCharReferences( $text );
 223
 224                 if ($wgUseTidy) {
 225                         $text = Parser::tidy($text);
 226                 }
 227
 228                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 229
 230                 $this->mOutput->setText( $text );
 231                 wfProfileOut( $fname );
 232                 return $this->mOutput;
 233         }
 234
 235         /**
 236          * Get a random string
 237          *
 238          * @access private
 239          * @static
 240          */
 241         function getRandomString() {
 242                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 243         }
 244
 245         /**
 246          * Replaces all occurrences of <$tag>content</$tag> in the text
 247          * with a random marker and returns the new text. the output parameter
 248          * $content will be an associative array filled with data on the form
 249          * $unique_marker => content.
 250          *
 251          * If $content is already set, the additional entries will be appended
 252          * If $tag is set to STRIP_COMMENTS, the function will extract
 253          * <!-- HTML comments -->
 254          *
 255          * @access private
 256          * @static
 257          */
 258         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 259                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 260                 if ( !$content ) {
 261                         $content = array( );
 262                 }
 263                 $n = 1;
 264                 $stripped = '';
 265
 266                 if ( !$tags ) {
 267                         $tags = array( );
 268                 }
 269
 270                 if ( !$params ) {
 271                         $params = array( );
 272                 }
 273
 274                 if( $tag == STRIP_COMMENTS ) {
 275                         $start = '/<!--()()/';
 276                         $end   = '/-->/';
 277                 } else {
 278                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 279                         $end   = "/<\\/$tag\\s*>/i";
 280                 }
 281
 282                 while ( '' != $text ) {
 283                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 284                         $stripped .= $p[0];
 285                         if( count( $p ) < 4 ) {
 286                                 break;
 287                         }
 288                         $attributes = $p[1];
 289                         $empty      = $p[2];
 290                         $inside     = $p[3];
 291
 292                         $marker = $rnd . sprintf('%08X', $n++);
 293                         $stripped .= $marker;
 294
 295                         $tags[$marker] = "<$tag$attributes$empty>";
 296                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 297
 298                         if ( $empty === '/' ) {
 299                                 // Empty element tag, <tag />
 300                                 $content[$marker] = null;
 301                                 $text = $inside;
 302                         } else {
 303                                 $q = preg_split( $end, $inside, 2 );
 304                                 $content[$marker] = $q[0];
 305                                 if( count( $q ) < 2 ) {
 306                                         # No end tag -- let it run out to the end of the text.
 307                                         break;
 308                                 } else {
 309                                         $text = $q[1];
 310                                 }
 311                         }
 312                 }
 313                 return $stripped;
 314         }
 315
 316         /**
 317          * Wrapper function for extractTagsAndParams
 318          * for cases where $tags and $params isn't needed
 319          * i.e. where tags will never have params, like <nowiki>
 320          *
 321          * @access private
 322          * @static
 323          */
 324         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 325                 $dummy_tags = array();
 326                 $dummy_params = array();
 327
 328                 return Parser::extractTagsAndParams( $tag, $text, $content,
 329                         $dummy_tags, $dummy_params, $uniq_prefix );
 330         }
 331
 332         /**
 333          * Strips and renders nowiki, pre, math, hiero
 334          * If $render is set, performs necessary rendering operations on plugins
 335          * Returns the text, and fills an array with data needed in unstrip()
 336          * If the $state is already a valid strip state, it adds to the state
 337          *
 338          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 339          *  will be stripped in addition to other tags. This is important
 340          *  for section editing, where these comments cause confusion when
 341          *  counting the sections in the wikisource
 342          *
 343          * @access private
 344          */
 345         function strip( $text, &$state, $stripcomments = false ) {
 346                 $render = ($this->mOutputType == OT_HTML);
 347                 $html_content = array();
 348                 $nowiki_content = array();
 349                 $math_content = array();
 350                 $pre_content = array();
 351                 $comment_content = array();
 352                 $ext_content = array();
 353                 $ext_tags = array();
 354                 $ext_params = array();
 355                 $gallery_content = array();
 356
 357                 # Replace any instances of the placeholders
 358                 $uniq_prefix = UNIQ_PREFIX;
 359                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 360
 361                 # html
 362                 global $wgRawHtml;
 363                 if( $wgRawHtml ) {
 364                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 365                         foreach( $html_content as $marker => $content ) {
 366                                 if ($render ) {
 367                                         # Raw and unchecked for validity.
 368                                         $html_content[$marker] = $content;
 369                                 } else {
 370                                         $html_content[$marker] = '<html>'.$content.'</html>';
 371                                 }
 372                         }
 373                 }
 374
 375                 # nowiki
 376                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 377                 foreach( $nowiki_content as $marker => $content ) {
 378                         if( $render ){
 379                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 380                         } else {
 381                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 382                         }
 383                 }
 384
 385                 # math
 386                 if( $this->mOptions->getUseTeX() ) {
 387                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 388                         foreach( $math_content as $marker => $content ){
 389                                 if( $render ) {
 390                                         $math_content[$marker] = renderMath( $content );
 391                                 } else {
 392                                         $math_content[$marker] = '<math>'.$content.'</math>';
 393                                 }
 394                         }
 395                 }
 396
 397                 # pre
 398                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 399                 foreach( $pre_content as $marker => $content ){
 400                         if( $render ){
 401                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 402                         } else {
 403                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 404                         }
 405                 }
 406
 407                 # gallery
 408                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 409                 foreach( $gallery_content as $marker => $content ) {
 410                         require_once( 'ImageGallery.php' );
 411                         if ( $render ) {
 412                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 413                         } else {
 414                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 415                         }
 416                 }
 417
 418                 # Comments
 419                 if($stripcomments) {
 420                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 421                         foreach( $comment_content as $marker => $content ){
 422                                 $comment_content[$marker] = '<!--'.$content.'-->';
 423                         }
 424                 }
 425
 426                 # Extensions
 427                 foreach ( $this->mTagHooks as $tag => $callback ) {
 428                         $ext_content[$tag] = array();
 429                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 430                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 431                         foreach( $ext_content[$tag] as $marker => $content ) {
 432                                 $full_tag = $ext_tags[$tag][$marker];
 433                                 $params = $ext_params[$tag][$marker];
 434                                 if ( $render )
 435                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, $this ) );
 436                                 else {
 437                                         if ( is_null( $content ) ) {
 438                                                 // Empty element tag
 439                                                 $ext_content[$tag][$marker] = $full_tag;
 440                                         } else {
 441                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 442                                         }
 443                                 }
 444                         }
 445                 }
 446
 447                 # Merge state with the pre-existing state, if there is one
 448                 if ( $state ) {
 449                         $state['html'] = $state['html'] + $html_content;
 450                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 451                         $state['math'] = $state['math'] + $math_content;
 452                         $state['pre'] = $state['pre'] + $pre_content;
 453                         $state['comment'] = $state['comment'] + $comment_content;
 454                         $state['gallery'] = $state['gallery'] + $gallery_content;
 455
 456                         foreach( $ext_content as $tag => $array ) {
 457                                 if ( array_key_exists( $tag, $state ) ) {
 458                                         $state[$tag] = $state[$tag] + $array;
 459                                 }
 460                         }
 461                 } else {
 462                         $state = array(
 463                           'html' => $html_content,
 464                           'nowiki' => $nowiki_content,
 465                           'math' => $math_content,
 466                           'pre' => $pre_content,
 467                           'comment' => $comment_content,
 468                           'gallery' => $gallery_content,
 469                         ) + $ext_content;
 470                 }
 471                 return $text;
 472         }
 473
 474         /**
 475          * restores pre, math, and hiero removed by strip()
 476          *
 477          * always call unstripNoWiki() after this one
 478          * @access private
 479          */
 480         function unstrip( $text, &$state ) {
 481                 if ( !is_array( $state ) ) {
 482                         return $text;
 483                 }
 484
 485                 # Must expand in reverse order, otherwise nested tags will be corrupted
 486                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 487                         if( $tag != 'nowiki' && $tag != 'html' ) {
 488                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 489                                         $text = str_replace( $uniq, $content, $text );
 490                                 }
 491                         }
 492                 }
 493
 494                 return $text;
 495         }
 496
 497         /**
 498          * always call this after unstrip() to preserve the order
 499          *
 500          * @access private
 501          */
 502         function unstripNoWiki( $text, &$state ) {
 503                 if ( !is_array( $state ) ) {
 504                         return $text;
 505                 }
 506
 507                 # Must expand in reverse order, otherwise nested tags will be corrupted
 508                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 509                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 510                 }
 511
 512                 global $wgRawHtml;
 513                 if ($wgRawHtml) {
 514                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 515                                 $text = str_replace( key( $state['html'] ), $content, $text );
 516                         }
 517                 }
 518
 519                 return $text;
 520         }
 521
 522         /**
 523          * Add an item to the strip state
 524          * Returns the unique tag which must be inserted into the stripped text
 525          * The tag will be replaced with the original text in unstrip()
 526          *
 527          * @access private
 528          */
 529         function insertStripItem( $text, &$state ) {
 530                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 531                 if ( !$state ) {
 532                         $state = array(
 533                           'html' => array(),
 534                           'nowiki' => array(),
 535                           'math' => array(),
 536                           'pre' => array(),
 537                           'comment' => array(),
 538                           'gallery' => array(),
 539                         );
 540                 }
 541                 $state['item'][$rnd] = $text;
 542                 return $rnd;
 543         }
 544
 545         /**
 546          * Interface with html tidy, used if $wgUseTidy = true.
 547          * If tidy isn't able to correct the markup, the original will be
 548          * returned in all its glory with a warning comment appended.
 549          *
 550          * Either the external tidy program or the in-process tidy extension
 551          * will be used depending on availability. Override the default
 552          * $wgTidyInternal setting to disable the internal if it's not working.
 553          *
 554          * @param string $text Hideous HTML input
 555          * @return string Corrected HTML output
 556          * @access public
 557          * @static
 558          */
 559         function tidy( $text ) {
 560                 global $wgTidyInternal;
 561                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 562 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 563 '<head><title>test</title></head><body>'.$text.'</body></html>';
 564                 if( $wgTidyInternal ) {
 565                         $correctedtext = Parser::internalTidy( $wrappedtext );
 566                 } else {
 567                         $correctedtext = Parser::externalTidy( $wrappedtext );
 568                 }
 569                 if( is_null( $correctedtext ) ) {
 570                         wfDebug( "Tidy error detected!\n" );
 571                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 572                 }
 573                 return $correctedtext;
 574         }
 575
 576         /**
 577          * Spawn an external HTML tidy process and get corrected markup back from it.
 578          *
 579          * @access private
 580          * @static
 581          */
 582         function externalTidy( $text ) {
 583                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 584                 $fname = 'Parser::externalTidy';
 585                 wfProfileIn( $fname );
 586
 587                 $cleansource = '';
 588                 $opts = ' -utf8';
 589
 590                 $descriptorspec = array(
 591                         0 => array('pipe', 'r'),
 592                         1 => array('pipe', 'w'),
 593                         2 => array('file', '/dev/null', 'a')
 594                 );
 595                 $pipes = array();
 596                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 597                 if (is_resource($process)) {
 598                         fwrite($pipes[0], $text);
 599                         fclose($pipes[0]);
 600                         while (!feof($pipes[1])) {
 601                                 $cleansource .= fgets($pipes[1], 1024);
 602                         }
 603                         fclose($pipes[1]);
 604                         proc_close($process);
 605                 }
 606
 607                 wfProfileOut( $fname );
 608
 609                 if( $cleansource == '' && $text != '') {
 610                         // Some kind of error happened, so we couldn't get the corrected text.
 611                         // Just give up; we'll use the source text and append a warning.
 612                         return null;
 613                 } else {
 614                         return $cleansource;
 615                 }
 616         }
 617
 618         /**
 619          * Use the HTML tidy PECL extension to use the tidy library in-process,
 620          * saving the overhead of spawning a new process. Currently written to
 621          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 622          *
 623          * 'pear install tidy' should be able to compile the extension module.
 624          *
 625          * @access private
 626          * @static
 627          */
 628         function internalTidy( $text ) {
 629                 global $wgTidyConf;
 630                 $fname = 'Parser::internalTidy';
 631                 wfProfileIn( $fname );
 632
 633                 tidy_load_config( $wgTidyConf );
 634                 tidy_set_encoding( 'utf8' );
 635                 tidy_parse_string( $text );
 636                 tidy_clean_repair();
 637                 if( tidy_get_status() == 2 ) {
 638                         // 2 is magic number for fatal error
 639                         // http://www.php.net/manual/en/function.tidy-get-status.php
 640                         $cleansource = null;
 641                 } else {
 642                         $cleansource = tidy_get_output();
 643                 }
 644                 wfProfileOut( $fname );
 645                 return $cleansource;
 646         }
 647
 648         /**
 649          * parse the wiki syntax used to render tables
 650          *
 651          * @access private
 652          */
 653         function doTableStuff ( $t ) {
 654                 $fname = 'Parser::doTableStuff';
 655                 wfProfileIn( $fname );
 656
 657                 $t = explode ( "\n" , $t ) ;
 658                 $td = array () ; # Is currently a td tag open?
 659                 $ltd = array () ; # Was it TD or TH?
 660                 $tr = array () ; # Is currently a tr tag open?
 661                 $ltr = array () ; # tr attributes
 662                 $indent_level = 0; # indent level of the table
 663                 foreach ( $t AS $k => $x )
 664                 {
 665                         $x = trim ( $x ) ;
 666                         $fc = substr ( $x , 0 , 1 ) ;
 667                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 668                                 $indent_level = strlen( $matches[1] );
 669
 670                                 $attributes = $this->unstripForHTML( $matches[2] );
 671
 672                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 673                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 674                                 array_push ( $td , false ) ;
 675                                 array_push ( $ltd , '' ) ;
 676                                 array_push ( $tr , false ) ;
 677                                 array_push ( $ltr , '' ) ;
 678                         }
 679                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 680                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 681                                 $z = "</table>" . substr ( $x , 2);
 682                                 $l = array_pop ( $ltd ) ;
 683                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 684                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 685                                 array_pop ( $ltr ) ;
 686                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 687                         }
 688                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 689                                 $x = substr ( $x , 1 ) ;
 690                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 691                                 $z = '' ;
 692                                 $l = array_pop ( $ltd ) ;
 693                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 694                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 695                                 array_pop ( $ltr ) ;
 696                                 $t[$k] = $z ;
 697                                 array_push ( $tr , false ) ;
 698                                 array_push ( $td , false ) ;
 699                                 array_push ( $ltd , '' ) ;
 700                                 $attributes = $this->unstripForHTML( $x );
 701                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 702                         }
 703                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 704                                 # $x is a table row
 705                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 706                                         $fc = '+' ;
 707                                         $x = substr ( $x , 1 ) ;
 708                                 }
 709                                 $after = substr ( $x , 1 ) ;
 710                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 711                                 $after = explode ( '||' , $after ) ;
 712                                 $t[$k] = '' ;
 713
 714                                 # Loop through each table cell
 715                                 foreach ( $after AS $theline )
 716                                 {
 717                                         $z = '' ;
 718                                         if ( $fc != '+' )
 719                                         {
 720                                                 $tra = array_pop ( $ltr ) ;
 721                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 722                                                 array_push ( $tr , true ) ;
 723                                                 array_push ( $ltr , '' ) ;
 724                                         }
 725
 726                                         $l = array_pop ( $ltd ) ;
 727                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 728                                         if ( $fc == '|' ) $l = 'td' ;
 729                                         else if ( $fc == '!' ) $l = 'th' ;
 730                                         else if ( $fc == '+' ) $l = 'caption' ;
 731                                         else $l = '' ;
 732                                         array_push ( $ltd , $l ) ;
 733
 734                                         # Cell parameters
 735                                         $y = explode ( '|' , $theline , 2 ) ;
 736                                         # Note that a '|' inside an invalid link should not
 737                                         # be mistaken as delimiting cell parameters
 738                                         if ( strpos( $y[0], '[[' ) !== false ) {
 739                                                 $y = array ($theline);
 740                                         }
 741                                         if ( count ( $y ) == 1 )
 742                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 743                                         else {
 744                                                 $attributes = $this->unstripForHTML( $y[0] );
 745                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 746                                         }
 747                                         $t[$k] .= $y ;
 748                                         array_push ( $td , true ) ;
 749                                 }
 750                         }
 751                 }
 752
 753                 # Closing open td, tr && table
 754                 while ( count ( $td ) > 0 )
 755                 {
 756                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 757                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 758                         $t[] = '</table>' ;
 759                 }
 760
 761                 $t = implode ( "\n" , $t ) ;
 762                 wfProfileOut( $fname );
 763                 return $t ;
 764         }
 765
 766         /**
 767          * Helper function for parse() that transforms wiki markup into
 768          * HTML. Only called for $mOutputType == OT_HTML.
 769          *
 770          * @access private
 771          */
 772         function internalParse( $text ) {
 773                 global $wgContLang;
 774                 $args = array();
 775                 $isMain = true;
 776                 $fname = 'Parser::internalParse';
 777                 wfProfileIn( $fname );
 778
 779                 # Remove <noinclude> tags and <includeonly> sections
 780                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 781                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 782                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 783
 784                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 785                 $text = $this->replaceVariables( $text, $args );
 786
 787                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 788
 789                 $text = $this->doHeadings( $text );
 790                 if($this->mOptions->getUseDynamicDates()) {
 791                         $df =& DateFormatter::getInstance();
 792                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 793                 }
 794                 $text = $this->doAllQuotes( $text );
 795                 $text = $this->replaceInternalLinks( $text );
 796                 $text = $this->replaceExternalLinks( $text );
 797
 798                 # replaceInternalLinks may sometimes leave behind
 799                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 800                 $text = str_replace(UNIQ_PREFIX."NOPARSE", "", $text);
 801
 802                 $text = $this->doMagicLinks( $text );
 803                 $text = $this->doTableStuff( $text );
 804                 $text = $this->formatHeadings( $text, $isMain );
 805
 806                 $regex = '/<!--IW_TRANSCLUDE (\d+)-->/';
 807                 $text = preg_replace_callback($regex, array(&$this, 'scarySubstitution'), $text);
 808
 809                 wfProfileOut( $fname );
 810                 return $text;
 811         }
 812
 813         function scarySubstitution($matches) {
 814 #               return "[[".$matches[0]."]]";
 815                 return $this->mIWTransData[(int)$matches[0]];
 816         }
 817
 818         /**
 819          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 820          * magic external links.
 821          *
 822          * @access private
 823          */
 824         function &doMagicLinks( &$text ) {
 825                 $text = $this->magicISBN( $text );
 826                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 827                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 828                 return $text;
 829         }
 830
 831         /**
 832          * Parse ^^ tokens and return html
 833          *
 834          * @access private
 835          */
 836         function doExponent( $text ) {
 837                 $fname = 'Parser::doExponent';
 838                 wfProfileIn( $fname );
 839                 $text = preg_replace('/\^\^(.*?)\^\^/','<small><sup>\\1</sup></small>', $text);
 840                 wfProfileOut( $fname );
 841                 return $text;
 842         }
 843
 844         /**
 845          * Parse headers and return html
 846          *
 847          * @access private
 848          */
 849         function doHeadings( $text ) {
 850                 $fname = 'Parser::doHeadings';
 851                 wfProfileIn( $fname );
 852                 for ( $i = 6; $i >= 1; --$i ) {
 853                         $h = substr( '======', 0, $i );
 854                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 855                           "<h{$i}>\\1</h{$i}>\\2", $text );
 856                 }
 857                 wfProfileOut( $fname );
 858                 return $text;
 859         }
 860
 861         /**
 862          * Replace single quotes with HTML markup
 863          * @access private
 864          * @return string the altered text
 865          */
 866         function doAllQuotes( $text ) {
 867                 $fname = 'Parser::doAllQuotes';
 868                 wfProfileIn( $fname );
 869                 $outtext = '';
 870                 $lines = explode( "\n", $text );
 871                 foreach ( $lines as $line ) {
 872                         $outtext .= $this->doQuotes ( $line ) . "\n";
 873                 }
 874                 $outtext = substr($outtext, 0,-1);
 875                 wfProfileOut( $fname );
 876                 return $outtext;
 877         }
 878
 879         /**
 880          * Helper function for doAllQuotes()
 881          * @access private
 882          */
 883         function doQuotes( $text ) {
 884                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 885                 if ( count( $arr ) == 1 )
 886                         return $text;
 887                 else
 888                 {
 889                         # First, do some preliminary work. This may shift some apostrophes from
 890                         # being mark-up to being text. It also counts the number of occurrences
 891                         # of bold and italics mark-ups.
 892                         $i = 0;
 893                         $numbold = 0;
 894                         $numitalics = 0;
 895                         foreach ( $arr as $r )
 896                         {
 897                                 if ( ( $i % 2 ) == 1 )
 898                                 {
 899                                         # If there are ever four apostrophes, assume the first is supposed to
 900                                         # be text, and the remaining three constitute mark-up for bold text.
 901                                         if ( strlen( $arr[$i] ) == 4 )
 902                                         {
 903                                                 $arr[$i-1] .= "'";
 904                                                 $arr[$i] = "'''";
 905                                         }
 906                                         # If there are more than 5 apostrophes in a row, assume they're all
 907                                         # text except for the last 5.
 908                                         else if ( strlen( $arr[$i] ) > 5 )
 909                                         {
 910                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 911                                                 $arr[$i] = "'''''";
 912                                         }
 913                                         # Count the number of occurrences of bold and italics mark-ups.
 914                                         # We are not counting sequences of five apostrophes.
 915                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 916                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 917                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 918                                 }
 919                                 $i++;
 920                         }
 921
 922                         # If there is an odd number of both bold and italics, it is likely
 923                         # that one of the bold ones was meant to be an apostrophe followed
 924                         # by italics. Which one we cannot know for certain, but it is more
 925                         # likely to be one that has a single-letter word before it.
 926                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 927                         {
 928                                 $i = 0;
 929                                 $firstsingleletterword = -1;
 930                                 $firstmultiletterword = -1;
 931                                 $firstspace = -1;
 932                                 foreach ( $arr as $r )
 933                                 {
 934                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 935                                         {
 936                                                 $x1 = substr ($arr[$i-1], -1);
 937                                                 $x2 = substr ($arr[$i-1], -2, 1);
 938                                                 if ($x1 == ' ') {
 939                                                         if ($firstspace == -1) $firstspace = $i;
 940                                                 } else if ($x2 == ' ') {
 941                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 942                                                 } else {
 943                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 944                                                 }
 945                                         }
 946                                         $i++;
 947                                 }
 948
 949                                 # If there is a single-letter word, use it!
 950                                 if ($firstsingleletterword > -1)
 951                                 {
 952                                         $arr [ $firstsingleletterword ] = "''";
 953                                         $arr [ $firstsingleletterword-1 ] .= "'";
 954                                 }
 955                                 # If not, but there's a multi-letter word, use that one.
 956                                 else if ($firstmultiletterword > -1)
 957                                 {
 958                                         $arr [ $firstmultiletterword ] = "''";
 959                                         $arr [ $firstmultiletterword-1 ] .= "'";
 960                                 }
 961                                 # ... otherwise use the first one that has neither.
 962                                 # (notice that it is possible for all three to be -1 if, for example,
 963                                 # there is only one pentuple-apostrophe in the line)
 964                                 else if ($firstspace > -1)
 965                                 {
 966                                         $arr [ $firstspace ] = "''";
 967                                         $arr [ $firstspace-1 ] .= "'";
 968                                 }
 969                         }
 970
 971                         # Now let's actually convert our apostrophic mush to HTML!
 972                         $output = '';
 973                         $buffer = '';
 974                         $state = '';
 975                         $i = 0;
 976                         foreach ($arr as $r)
 977                         {
 978                                 if (($i % 2) == 0)
 979                                 {
 980                                         if ($state == 'both')
 981                                                 $buffer .= $r;
 982                                         else
 983                                                 $output .= $r;
 984                                 }
 985                                 else
 986                                 {
 987                                         if (strlen ($r) == 2)
 988                                         {
 989                                                 if ($state == 'i')
 990                                                 { $output .= '</i>'; $state = ''; }
 991                                                 else if ($state == 'bi')
 992                                                 { $output .= '</i>'; $state = 'b'; }
 993                                                 else if ($state == 'ib')
 994                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 995                                                 else if ($state == 'both')
 996                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 997                                                 else # $state can be 'b' or ''
 998                                                 { $output .= '<i>'; $state .= 'i'; }
 999                                         }
1000                                         else if (strlen ($r) == 3)
1001                                         {
1002                                                 if ($state == 'b')
1003                                                 { $output .= '</b>'; $state = ''; }
1004                                                 else if ($state == 'bi')
1005                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1006                                                 else if ($state == 'ib')
1007                                                 { $output .= '</b>'; $state = 'i'; }
1008                                                 else if ($state == 'both')
1009                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1010                                                 else # $state can be 'i' or ''
1011                                                 { $output .= '<b>'; $state .= 'b'; }
1012                                         }
1013                                         else if (strlen ($r) == 5)
1014                                         {
1015                                                 if ($state == 'b')
1016                                                 { $output .= '</b><i>'; $state = 'i'; }
1017                                                 else if ($state == 'i')
1018                                                 { $output .= '</i><b>'; $state = 'b'; }
1019                                                 else if ($state == 'bi')
1020                                                 { $output .= '</i></b>'; $state = ''; }
1021                                                 else if ($state == 'ib')
1022                                                 { $output .= '</b></i>'; $state = ''; }
1023                                                 else if ($state == 'both')
1024                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1025                                                 else # ($state == '')
1026                                                 { $buffer = ''; $state = 'both'; }
1027                                         }
1028                                 }
1029                                 $i++;
1030                         }
1031                         # Now close all remaining tags.  Notice that the order is important.
1032                         if ($state == 'b' || $state == 'ib')
1033                                 $output .= '</b>';
1034                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1035                                 $output .= '</i>';
1036                         if ($state == 'bi')
1037                                 $output .= '</b>';
1038                         if ($state == 'both')
1039                                 $output .= '<b><i>'.$buffer.'</i></b>';
1040                         return $output;
1041                 }
1042         }
1043
1044         /**
1045          * Replace external links
1046          *
1047          * Note: this is all very hackish and the order of execution matters a lot.
1048          * Make sure to run maintenance/parserTests.php if you change this code.
1049          *
1050          * @access private
1051          */
1052         function replaceExternalLinks( $text ) {
1053                 global $wgContLang;
1054                 $fname = 'Parser::replaceExternalLinks';
1055                 wfProfileIn( $fname );
1056
1057                 $sk =& $this->mOptions->getSkin();
1058
1059                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1060
1061                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1062
1063                 $i = 0;
1064                 while ( $i<count( $bits ) ) {
1065                         $url = $bits[$i++];
1066                         $protocol = $bits[$i++];
1067                         $text = $bits[$i++];
1068                         $trail = $bits[$i++];
1069
1070                         # The characters '<' and '>' (which were escaped by
1071                         # removeHTMLtags()) should not be included in
1072                         # URLs, per RFC 2396.
1073                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1074                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1075                                 $url = substr($url, 0, $m2[0][1]);
1076                         }
1077
1078                         # If the link text is an image URL, replace it with an <img> tag
1079                         # This happened by accident in the original parser, but some people used it extensively
1080                         $img = $this->maybeMakeExternalImage( $text );
1081                         if ( $img !== false ) {
1082                                 $text = $img;
1083                         }
1084
1085                         $dtrail = '';
1086
1087                         # Set linktype for CSS - if URL==text, link is essentially free
1088                         $linktype = ($text == $url) ? 'free' : 'text';
1089
1090                         # No link text, e.g. [http://domain.tld/some.link]
1091                         if ( $text == '' ) {
1092                                 # Autonumber if allowed
1093                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1094                                         $text = '[' . ++$this->mAutonumber . ']';
1095                                         $linktype = 'autonumber';
1096                                 } else {
1097                                         # Otherwise just use the URL
1098                                         $text = htmlspecialchars( $url );
1099                                         $linktype = 'free';
1100                                 }
1101                         } else {
1102                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1103                                 # Check for trail
1104                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1105                         }
1106
1107                         $text = $wgContLang->markNoConversion($text);
1108
1109                         # Replace &amp; from obsolete syntax with &.
1110                         # All HTML entities will be escaped by makeExternalLink()
1111                         # or maybeMakeExternalImage()
1112                         $url = str_replace( '&amp;', '&', $url );
1113
1114                         # Process the trail (i.e. everything after this link up until start of the next link),
1115                         # replacing any non-bracketed links
1116                         $trail = $this->replaceFreeExternalLinks( $trail );
1117
1118
1119                         # Use the encoded URL
1120                         # This means that users can paste URLs directly into the text
1121                         # Funny characters like &ouml; aren't valid in URLs anyway
1122                         # This was changed in August 2004
1123                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1124                 }
1125
1126                 wfProfileOut( $fname );
1127                 return $s;
1128         }
1129
1130         /**
1131          * Replace anything that looks like a URL with a link
1132          * @access private
1133          */
1134         function replaceFreeExternalLinks( $text ) {
1135                 global $wgContLang;
1136                 $fname = 'Parser::replaceFreeExternalLinks';
1137                 wfProfileIn( $fname );
1138
1139                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1140                 $s = array_shift( $bits );
1141                 $i = 0;
1142
1143                 $sk =& $this->mOptions->getSkin();
1144
1145                 while ( $i < count( $bits ) ){
1146                         $protocol = $bits[$i++];
1147                         $remainder = $bits[$i++];
1148
1149                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1150                                 # Found some characters after the protocol that look promising
1151                                 $url = $protocol . $m[1];
1152                                 $trail = $m[2];
1153
1154                                 # The characters '<' and '>' (which were escaped by
1155                                 # removeHTMLtags()) should not be included in
1156                                 # URLs, per RFC 2396.
1157                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1158                                         $trail = substr($url, $m2[0][1]) . $trail;
1159                                         $url = substr($url, 0, $m2[0][1]);
1160                                 }
1161
1162                                 # Move trailing punctuation to $trail
1163                                 $sep = ',;\.:!?';
1164                                 # If there is no left bracket, then consider right brackets fair game too
1165                                 if ( strpos( $url, '(' ) === false ) {
1166                                         $sep .= ')';
1167                                 }
1168
1169                                 $numSepChars = strspn( strrev( $url ), $sep );
1170                                 if ( $numSepChars ) {
1171                                         $trail = substr( $url, -$numSepChars ) . $trail;
1172                                         $url = substr( $url, 0, -$numSepChars );
1173                                 }
1174
1175                                 # Replace &amp; from obsolete syntax with &.
1176                                 # All HTML entities will be escaped by makeExternalLink()
1177                                 # or maybeMakeExternalImage()
1178                                 $url = str_replace( '&amp;', '&', $url );
1179
1180                                 # Is this an external image?
1181                                 $text = $this->maybeMakeExternalImage( $url );
1182                                 if ( $text === false ) {
1183                                         # Not an image, make a link
1184                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1185                                 }
1186                                 $s .= $text . $trail;
1187                         } else {
1188                                 $s .= $protocol . $remainder;
1189                         }
1190                 }
1191                 wfProfileOut( $fname );
1192                 return $s;
1193         }
1194
1195         /**
1196          * make an image if it's allowed, either through the global
1197          * option or through the exception
1198          * @access private
1199          */
1200         function maybeMakeExternalImage( $url ) {
1201                 $sk =& $this->mOptions->getSkin();
1202                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1203                 $imagesexception = !empty($imagesfrom);
1204                 $text = false;
1205                 if ( $this->mOptions->getAllowExternalImages()
1206                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1207                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1208                                 # Image found
1209                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1210                         }
1211                 }
1212                 return $text;
1213         }
1214
1215         /**
1216          * Process [[ ]] wikilinks
1217          *
1218          * @access private
1219          */
1220         function replaceInternalLinks( $s ) {
1221                 global $wgContLang, $wgLinkCache;
1222                 static $fname = 'Parser::replaceInternalLinks' ;
1223
1224                 wfProfileIn( $fname );
1225
1226                 wfProfileIn( $fname.'-setup' );
1227                 static $tc = FALSE;
1228                 # the % is needed to support urlencoded titles as well
1229                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1230
1231                 $sk =& $this->mOptions->getSkin();
1232
1233                 #split the entire text string on occurences of [[
1234                 $a = explode( '[[', ' ' . $s );
1235                 #get the first element (all text up to first [[), and remove the space we added
1236                 $s = array_shift( $a );
1237                 $s = substr( $s, 1 );
1238
1239                 # Match a link having the form [[namespace:link|alternate]]trail
1240                 static $e1 = FALSE;
1241                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1242                 # Match cases where there is no "]]", which might still be images
1243                 static $e1_img = FALSE;
1244                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1245                 # Match the end of a line for a word that's not followed by whitespace,
1246                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1247                 $e2 = wfMsgForContent( 'linkprefix' );
1248
1249                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1250
1251                 if( is_null( $this->mTitle ) ) {
1252                         wfDebugDieBacktrace( 'nooo' );
1253                 }
1254                 $nottalk = !$this->mTitle->isTalkPage();
1255
1256                 if ( $useLinkPrefixExtension ) {
1257                         if ( preg_match( $e2, $s, $m ) ) {
1258                                 $first_prefix = $m[2];
1259                         } else {
1260                                 $first_prefix = false;
1261                         }
1262                 } else {
1263                         $prefix = '';
1264                 }
1265
1266                 $selflink = $this->mTitle->getPrefixedText();
1267                 wfProfileOut( $fname.'-setup' );
1268
1269                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1270                 $useSubpages = $this->areSubpagesAllowed();
1271
1272                 # Loop for each link
1273                 for ($k = 0; isset( $a[$k] ); $k++) {
1274                         $line = $a[$k];
1275                         if ( $useLinkPrefixExtension ) {
1276                                 wfProfileIn( $fname.'-prefixhandling' );
1277                                 if ( preg_match( $e2, $s, $m ) ) {
1278                                         $prefix = $m[2];
1279                                         $s = $m[1];
1280                                 } else {
1281                                         $prefix='';
1282                                 }
1283                                 # first link
1284                                 if($first_prefix) {
1285                                         $prefix = $first_prefix;
1286                                         $first_prefix = false;
1287                                 }
1288                                 wfProfileOut( $fname.'-prefixhandling' );
1289                         }
1290
1291                         $might_be_img = false;
1292
1293                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1294                                 $text = $m[2];
1295                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1296                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1297                                 # the real problem is with the $e1 regex
1298                                 # See bug 1300.
1299                                 #
1300                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1301                                 # and no image is in sight. See bug 2095.
1302                                 #
1303                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1304                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1305                                         $m[3] = $n[1];
1306                                 }
1307                                 # fix up urlencoded title texts
1308                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1309                                 $trail = $m[3];
1310                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1311                                 $might_be_img = true;
1312                                 $text = $m[2];
1313                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1314                                 $trail = "";
1315                         } else { # Invalid form; output directly
1316                                 $s .= $prefix . '[[' . $line ;
1317                                 continue;
1318                         }
1319
1320                         # Don't allow internal links to pages containing
1321                         # PROTO: where PROTO is a valid URL protocol; these
1322                         # should be external links.
1323                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1324                                 $s .= $prefix . '[[' . $line ;
1325                                 continue;
1326                         }
1327
1328                         # Make subpage if necessary
1329                         if( $useSubpages ) {
1330                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1331                         } else {
1332                                 $link = $m[1];
1333                         }
1334
1335                         $noforce = (substr($m[1], 0, 1) != ':');
1336                         if (!$noforce) {
1337                                 # Strip off leading ':'
1338                                 $link = substr($link, 1);
1339                         }
1340
1341                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1342                         if( !$nt ) {
1343                                 $s .= $prefix . '[[' . $line;
1344                                 continue;
1345                         }
1346
1347                         #check other language variants of the link
1348                         #if the article does not exist
1349                         if( $checkVariantLink
1350                             && $nt->getArticleID() == 0 ) {
1351                                 $wgContLang->findVariantLink($link, $nt);
1352                         }
1353
1354                         $ns = $nt->getNamespace();
1355                         $iw = $nt->getInterWiki();
1356
1357                         if ($might_be_img) { # if this is actually an invalid link
1358                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1359                                         $found = false;
1360                                         while (isset ($a[$k+1]) ) {
1361                                                 #look at the next 'line' to see if we can close it there
1362                                                 $spliced = array_splice( $a, $k + 1, 1 );
1363                                                 $next_line = array_shift( $spliced );
1364                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1365                                                 # the first ]] closes the inner link, the second the image
1366                                                         $found = true;
1367                                                         $text .= '[[' . $m[1];
1368                                                         $trail = $m[2];
1369                                                         break;
1370                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1371                                                         #if there's exactly one ]] that's fine, we'll keep looking
1372                                                         $text .= '[[' . $m[0];
1373                                                 } else {
1374                                                         #if $next_line is invalid too, we need look no further
1375                                                         $text .= '[[' . $next_line;
1376                                                         break;
1377                                                 }
1378                                         }
1379                                         if ( !$found ) {
1380                                                 # we couldn't find the end of this imageLink, so output it raw
1381                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1382                                                 $text = $this->replaceInternalLinks($text);
1383                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1384                                                 # note: no $trail, because without an end, there *is* no trail
1385                                                 continue;
1386                                         }
1387                                 } else { #it's not an image, so output it raw
1388                                         $s .= $prefix . '[[' . $link . '|' . $text;
1389                                         # note: no $trail, because without an end, there *is* no trail
1390                                         continue;
1391                                 }
1392                         }
1393
1394                         $wasblank = ( '' == $text );
1395                         if( $wasblank ) $text = $link;
1396
1397
1398                         # Link not escaped by : , create the various objects
1399                         if( $noforce ) {
1400
1401                                 # Interwikis
1402                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1403                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1404                                         $s = rtrim($s . "\n");
1405                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1406                                         continue;
1407                                 }
1408
1409                                 if ( $ns == NS_IMAGE ) {
1410                                         wfProfileIn( "$fname-image" );
1411                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1412                                                 # recursively parse links inside the image caption
1413                                                 # actually, this will parse them in any other parameters, too,
1414                                                 # but it might be hard to fix that, and it doesn't matter ATM
1415                                                 $text = $this->replaceExternalLinks($text);
1416                                                 $text = $this->replaceInternalLinks($text);
1417
1418                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1419                                                 $s .= $prefix . preg_replace( "/\b(" . wfUrlProtocols() . ')/', UNIQ_PREFIX."NOPARSE$1", $this->makeImage( $nt, $text) ) . $trail;
1420                                                 $wgLinkCache->addImageLinkObj( $nt );
1421
1422                                                 wfProfileOut( "$fname-image" );
1423                                                 continue;
1424                                         }
1425                                         wfProfileOut( "$fname-image" );
1426
1427                                 }
1428
1429                                 if ( $ns == NS_CATEGORY ) {
1430                                         wfProfileIn( "$fname-category" );
1431                                         $t = $wgContLang->convertHtml( $nt->getText() );
1432                                         $s = rtrim($s . "\n"); # bug 87
1433
1434                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1435                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1436                                         $wgLinkCache->resume();
1437
1438                                         if ( $wasblank ) {
1439                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1440                                                         $sortkey = $this->mTitle->getText();
1441                                                 } else {
1442                                                         $sortkey = $this->mTitle->getPrefixedText();
1443                                                 }
1444                                         } else {
1445                                                 $sortkey = $text;
1446                                         }
1447                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1448                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1449                                         $this->mOutput->addCategoryLink( $t );
1450
1451                                         /**
1452                                          * Strip the whitespace Category links produce, see bug 87
1453                                          * @todo We might want to use trim($tmp, "\n") here.
1454                                          */
1455                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1456
1457                                         wfProfileOut( "$fname-category" );
1458                                         continue;
1459                                 }
1460                         }
1461
1462                         if( ( $nt->getPrefixedText() === $selflink ) &&
1463                             ( $nt->getFragment() === '' ) ) {
1464                                 # Self-links are handled specially; generally de-link and change to bold.
1465                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1466                                 continue;
1467                         }
1468
1469                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1470                         if( $ns == NS_MEDIA ) {
1471                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1472                                 $wgLinkCache->addImageLinkObj( $nt );
1473                                 continue;
1474                         } elseif( $ns == NS_SPECIAL ) {
1475                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1476                                 continue;
1477                         }
1478                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1479                 }
1480                 wfProfileOut( $fname );
1481                 return $s;
1482         }
1483
1484         /**
1485          * Make a link placeholder. The text returned can be later resolved to a real link with
1486          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1487          * parsing of interwiki links, and secondly to allow all extistence checks and
1488          * article length checks (for stub links) to be bundled into a single query.
1489          *
1490          */
1491         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1492                 if ( ! is_object($nt) ) {
1493                         # Fail gracefully
1494                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1495                 } else {
1496                         # Separate the link trail from the rest of the link
1497                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1498
1499                         if ( $nt->isExternal() ) {
1500                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1501                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1502                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1503                         } else {
1504                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1505                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1506                                 $this->mLinkHolders['queries'][] = $query;
1507                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1508                                 $this->mLinkHolders['titles'][] = $nt;
1509
1510                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1511                         }
1512                 }
1513                 return $retVal;
1514         }
1515
1516         /**
1517          * Return true if subpage links should be expanded on this page.
1518          * @return bool
1519          */
1520         function areSubpagesAllowed() {
1521                 # Some namespaces don't allow subpages
1522                 global $wgNamespacesWithSubpages;
1523                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1524         }
1525
1526         /**
1527          * Handle link to subpage if necessary
1528          * @param string $target the source of the link
1529          * @param string &$text the link text, modified as necessary
1530          * @return string the full name of the link
1531          * @access private
1532          */
1533         function maybeDoSubpageLink($target, &$text) {
1534                 # Valid link forms:
1535                 # Foobar -- normal
1536                 # :Foobar -- override special treatment of prefix (images, language links)
1537                 # /Foobar -- convert to CurrentPage/Foobar
1538                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1539                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1540                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1541
1542                 $fname = 'Parser::maybeDoSubpageLink';
1543                 wfProfileIn( $fname );
1544                 $ret = $target; # default return value is no change
1545
1546                 # Some namespaces don't allow subpages,
1547                 # so only perform processing if subpages are allowed
1548                 if( $this->areSubpagesAllowed() ) {
1549                         # Look at the first character
1550                         if( $target != '' && $target{0} == '/' ) {
1551                                 # / at end means we don't want the slash to be shown
1552                                 if( substr( $target, -1, 1 ) == '/' ) {
1553                                         $target = substr( $target, 1, -1 );
1554                                         $noslash = $target;
1555                                 } else {
1556                                         $noslash = substr( $target, 1 );
1557                                 }
1558
1559                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1560                                 if( '' === $text ) {
1561                                         $text = $target;
1562                                 } # this might be changed for ugliness reasons
1563                         } else {
1564                                 # check for .. subpage backlinks
1565                                 $dotdotcount = 0;
1566                                 $nodotdot = $target;
1567                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1568                                         ++$dotdotcount;
1569                                         $nodotdot = substr( $nodotdot, 3 );
1570                                 }
1571                                 if($dotdotcount > 0) {
1572                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1573                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1574                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1575                                                 # / at the end means don't show full path
1576                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1577                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1578                                                         if( '' === $text ) {
1579                                                                 $text = $nodotdot;
1580                                                         }
1581                                                 }
1582                                                 $nodotdot = trim( $nodotdot );
1583                                                 if( $nodotdot != '' ) {
1584                                                         $ret .= '/' . $nodotdot;
1585                                                 }
1586                                         }
1587                                 }
1588                         }
1589                 }
1590
1591                 wfProfileOut( $fname );
1592                 return $ret;
1593         }
1594
1595         /**#@+
1596          * Used by doBlockLevels()
1597          * @access private
1598          */
1599         /* private */ function closeParagraph() {
1600                 $result = '';
1601                 if ( '' != $this->mLastSection ) {
1602                         $result = '</' . $this->mLastSection  . ">\n";
1603                 }
1604                 $this->mInPre = false;
1605                 $this->mLastSection = '';
1606                 return $result;
1607         }
1608         # getCommon() returns the length of the longest common substring
1609         # of both arguments, starting at the beginning of both.
1610         #
1611         /* private */ function getCommon( $st1, $st2 ) {
1612                 $fl = strlen( $st1 );
1613                 $shorter = strlen( $st2 );
1614                 if ( $fl < $shorter ) { $shorter = $fl; }
1615
1616                 for ( $i = 0; $i < $shorter; ++$i ) {
1617                         if ( $st1{$i} != $st2{$i} ) { break; }
1618                 }
1619                 return $i;
1620         }
1621         # These next three functions open, continue, and close the list
1622         # element appropriate to the prefix character passed into them.
1623         #
1624         /* private */ function openList( $char ) {
1625                 $result = $this->closeParagraph();
1626
1627                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1628                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1629                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1630                 else if ( ';' == $char ) {
1631                         $result .= '<dl><dt>';
1632                         $this->mDTopen = true;
1633                 }
1634                 else { $result = '<!-- ERR 1 -->'; }
1635
1636                 return $result;
1637         }
1638
1639         /* private */ function nextItem( $char ) {
1640                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1641                 else if ( ':' == $char || ';' == $char ) {
1642                         $close = '</dd>';
1643                         if ( $this->mDTopen ) { $close = '</dt>'; }
1644                         if ( ';' == $char ) {
1645                                 $this->mDTopen = true;
1646                                 return $close . '<dt>';
1647                         } else {
1648                                 $this->mDTopen = false;
1649                                 return $close . '<dd>';
1650                         }
1651                 }
1652                 return '<!-- ERR 2 -->';
1653         }
1654
1655         /* private */ function closeList( $char ) {
1656                 if ( '*' == $char ) { $text = '</li></ul>'; }
1657                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1658                 else if ( ':' == $char ) {
1659                         if ( $this->mDTopen ) {
1660                                 $this->mDTopen = false;
1661                                 $text = '</dt></dl>';
1662                         } else {
1663                                 $text = '</dd></dl>';
1664                         }
1665                 }
1666                 else {  return '<!-- ERR 3 -->'; }
1667                 return $text."\n";
1668         }
1669         /**#@-*/
1670
1671         /**
1672          * Make lists from lines starting with ':', '*', '#', etc.
1673          *
1674          * @access private
1675          * @return string the lists rendered as HTML
1676          */
1677         function doBlockLevels( $text, $linestart ) {
1678                 $fname = 'Parser::doBlockLevels';
1679                 wfProfileIn( $fname );
1680
1681                 # Parsing through the text line by line.  The main thing
1682                 # happening here is handling of block-level elements p, pre,
1683                 # and making lists from lines starting with * # : etc.
1684                 #
1685                 $textLines = explode( "\n", $text );
1686
1687                 $lastPrefix = $output = '';
1688                 $this->mDTopen = $inBlockElem = false;
1689                 $prefixLength = 0;
1690                 $paragraphStack = false;
1691
1692                 if ( !$linestart ) {
1693                         $output .= array_shift( $textLines );
1694                 }
1695                 foreach ( $textLines as $oLine ) {
1696                         $lastPrefixLength = strlen( $lastPrefix );
1697                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1698                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1699                         if ( !$this->mInPre ) {
1700                                 # Multiple prefixes may abut each other for nested lists.
1701                                 $prefixLength = strspn( $oLine, '*#:;' );
1702                                 $pref = substr( $oLine, 0, $prefixLength );
1703
1704                                 # eh?
1705                                 $pref2 = str_replace( ';', ':', $pref );
1706                                 $t = substr( $oLine, $prefixLength );
1707                                 $this->mInPre = !empty($preOpenMatch);
1708                         } else {
1709                                 # Don't interpret any other prefixes in preformatted text
1710                                 $prefixLength = 0;
1711                                 $pref = $pref2 = '';
1712                                 $t = $oLine;
1713                         }
1714
1715                         # List generation
1716                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1717                                 # Same as the last item, so no need to deal with nesting or opening stuff
1718                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1719                                 $paragraphStack = false;
1720
1721                                 if ( substr( $pref, -1 ) == ';') {
1722                                         # The one nasty exception: definition lists work like this:
1723                                         # ; title : definition text
1724                                         # So we check for : in the remainder text to split up the
1725                                         # title and definition, without b0rking links.
1726                                         $term = $t2 = '';
1727                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1728                                                 $t = $t2;
1729                                                 $output .= $term . $this->nextItem( ':' );
1730                                         }
1731                                 }
1732                         } elseif( $prefixLength || $lastPrefixLength ) {
1733                                 # Either open or close a level...
1734                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1735                                 $paragraphStack = false;
1736
1737                                 while( $commonPrefixLength < $lastPrefixLength ) {
1738                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1739                                         --$lastPrefixLength;
1740                                 }
1741                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1742                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1743                                 }
1744                                 while ( $prefixLength > $commonPrefixLength ) {
1745                                         $char = substr( $pref, $commonPrefixLength, 1 );
1746                                         $output .= $this->openList( $char );
1747
1748                                         if ( ';' == $char ) {
1749                                                 # FIXME: This is dupe of code above
1750                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1751                                                         $t = $t2;
1752                                                         $output .= $term . $this->nextItem( ':' );
1753                                                 }
1754                                         }
1755                                         ++$commonPrefixLength;
1756                                 }
1757                                 $lastPrefix = $pref2;
1758                         }
1759                         if( 0 == $prefixLength ) {
1760                                 wfProfileIn( "$fname-paragraph" );
1761                                 # No prefix (not in list)--go to paragraph mode
1762                                 $uniq_prefix = UNIQ_PREFIX;
1763                                 // XXX: use a stack for nestable elements like span, table and div
1764                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1765                                 $closematch = preg_match(
1766                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1767                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1768                                 if ( $openmatch or $closematch ) {
1769                                         $paragraphStack = false;
1770                                         $output .= $this->closeParagraph();
1771                                         if ( $preOpenMatch and !$preCloseMatch ) {
1772                                                 $this->mInPre = true;
1773                                         }
1774                                         if ( $closematch ) {
1775                                                 $inBlockElem = false;
1776                                         } else {
1777                                                 $inBlockElem = true;
1778                                         }
1779                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1780                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1781                                                 // pre
1782                                                 if ($this->mLastSection != 'pre') {
1783                                                         $paragraphStack = false;
1784                                                         $output .= $this->closeParagraph().'<pre>';
1785                                                         $this->mLastSection = 'pre';
1786                                                 }
1787                                                 $t = substr( $t, 1 );
1788                                         } else {
1789                                                 // paragraph
1790                                                 if ( '' == trim($t) ) {
1791                                                         if ( $paragraphStack ) {
1792                                                                 $output .= $paragraphStack.'<br />';
1793                                                                 $paragraphStack = false;
1794                                                                 $this->mLastSection = 'p';
1795                                                         } else {
1796                                                                 if ($this->mLastSection != 'p' ) {
1797                                                                         $output .= $this->closeParagraph();
1798                                                                         $this->mLastSection = '';
1799                                                                         $paragraphStack = '<p>';
1800                                                                 } else {
1801                                                                         $paragraphStack = '</p><p>';
1802                                                                 }
1803                                                         }
1804                                                 } else {
1805                                                         if ( $paragraphStack ) {
1806                                                                 $output .= $paragraphStack;
1807                                                                 $paragraphStack = false;
1808                                                                 $this->mLastSection = 'p';
1809                                                         } else if ($this->mLastSection != 'p') {
1810                                                                 $output .= $this->closeParagraph().'<p>';
1811                                                                 $this->mLastSection = 'p';
1812                                                         }
1813                                                 }
1814                                         }
1815                                 }
1816                                 wfProfileOut( "$fname-paragraph" );
1817                         }
1818                         // somewhere above we forget to get out of pre block (bug 785)
1819                         if($preCloseMatch && $this->mInPre) {
1820                                 $this->mInPre = false;
1821                         }
1822                         if ($paragraphStack === false) {
1823                                 $output .= $t."\n";
1824                         }
1825                 }
1826                 while ( $prefixLength ) {
1827                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1828                         --$prefixLength;
1829                 }
1830                 if ( '' != $this->mLastSection ) {
1831                         $output .= '</' . $this->mLastSection . '>';
1832                         $this->mLastSection = '';
1833                 }
1834
1835                 wfProfileOut( $fname );
1836                 return $output;
1837         }
1838
1839         /**
1840          * Split up a string on ':', ignoring any occurences inside
1841          * <a>..</a> or <span>...</span>
1842          * @param string $str the string to split
1843          * @param string &$before set to everything before the ':'
1844          * @param string &$after set to everything after the ':'
1845          * return string the position of the ':', or false if none found
1846          */
1847         function findColonNoLinks($str, &$before, &$after) {
1848                 # I wonder if we should make this count all tags, not just <a>
1849                 # and <span>. That would prevent us from matching a ':' that
1850                 # comes in the middle of italics other such formatting....
1851                 # -- Wil
1852                 $fname = 'Parser::findColonNoLinks';
1853                 wfProfileIn( $fname );
1854                 $pos = 0;
1855                 do {
1856                         $colon = strpos($str, ':', $pos);
1857
1858                         if ($colon !== false) {
1859                                 $before = substr($str, 0, $colon);
1860                                 $after = substr($str, $colon + 1);
1861
1862                                 # Skip any ':' within <a> or <span> pairs
1863                                 $a = substr_count($before, '<a');
1864                                 $s = substr_count($before, '<span');
1865                                 $ca = substr_count($before, '</a>');
1866                                 $cs = substr_count($before, '</span>');
1867
1868                                 if ($a <= $ca and $s <= $cs) {
1869                                         # Tags are balanced before ':'; ok
1870                                         break;
1871                                 }
1872                                 $pos = $colon + 1;
1873                         }
1874                 } while ($colon !== false);
1875                 wfProfileOut( $fname );
1876                 return $colon;
1877         }
1878
1879         /**
1880          * Return value of a magic variable (like PAGENAME)
1881          *
1882          * @access private
1883          */
1884         function getVariableValue( $index ) {
1885                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgArticle, $wgScriptPath;
1886
1887                 /**
1888                  * Some of these require message or data lookups and can be
1889                  * expensive to check many times.
1890                  */
1891                 static $varCache = array();
1892                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1893                         if ( isset( $varCache[$index] ) )
1894                                 return $varCache[$index];
1895
1896                 $ts = time();
1897                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
1898                 $revid = $wgArticle->getRevIdFetched();
1899                 wfRunHooks( 'ParserGetVariableValueRevid', array( &$this, &$revid ) );
1900
1901                 switch ( $index ) {
1902                         case MAG_CURRENTMONTH:
1903                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
1904                         case MAG_CURRENTMONTHNAME:
1905                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
1906                         case MAG_CURRENTMONTHNAMEGEN:
1907                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
1908                         case MAG_CURRENTMONTHABBREV:
1909                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
1910                         case MAG_CURRENTDAY:
1911                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
1912                         case MAG_PAGENAME:
1913                                 return $this->mTitle->getText();
1914                         case MAG_PAGENAMEE:
1915                                 return $this->mTitle->getPartialURL();
1916                         case MAG_FULLPAGENAME:
1917                                 return $this->mTitle->getPrefixedText();
1918                         case MAG_FULLPAGENAMEE:
1919                                 return wfUrlencode( $this->mTitle->getPrefixedText() );
1920                         case MAG_REVISIONID:
1921                                 return $revid;
1922                         case MAG_NAMESPACE:
1923                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
1924                         case MAG_NAMESPACEE:
1925                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
1926                         case MAG_CURRENTDAYNAME:
1927                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
1928                         case MAG_CURRENTYEAR:
1929                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
1930                         case MAG_CURRENTTIME:
1931                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
1932                         case MAG_CURRENTWEEK:
1933                                 return $varCache[$index] = $wgContLang->formatNum( date( 'W', $ts ) );
1934                         case MAG_CURRENTDOW:
1935                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
1936                         case MAG_NUMBEROFARTICLES:
1937                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1938                         case MAG_NUMBEROFFILES:
1939                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1940                         case MAG_SITENAME:
1941                                 return $wgSitename;
1942                         case MAG_SERVER:
1943                                 return $wgServer;
1944                         case MAG_SERVERNAME:
1945                                 return $wgServerName;
1946                         case MAG_SCRIPTPATH:
1947                                 return $wgScriptPath;
1948                         default:
1949                                 $ret = null;
1950                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
1951                                         return $ret;
1952                                 else
1953                                         return null;
1954                 }
1955         }
1956
1957         /**
1958          * initialise the magic variables (like CURRENTMONTHNAME)
1959          *
1960          * @access private
1961          */
1962         function initialiseVariables() {
1963                 $fname = 'Parser::initialiseVariables';
1964                 wfProfileIn( $fname );
1965                 global $wgVariableIDs;
1966                 $this->mVariables = array();
1967                 foreach ( $wgVariableIDs as $id ) {
1968                         $mw =& MagicWord::get( $id );
1969                         $mw->addToArray( $this->mVariables, $id );
1970                 }
1971                 wfProfileOut( $fname );
1972         }
1973
1974         /**
1975          * parse any parentheses in format ((title|part|part))
1976          * and call callbacks to get a replacement text for any found piece
1977          *
1978          * @param string $text The text to parse
1979          * @param array $callbacks rules in form:
1980          *     '{' => array(                            # opening parentheses
1981          *                                      'end' => '}',   # closing parentheses
1982          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
1983          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
1984          *                                                                )
1985          *                                      )
1986          * @access private
1987          */
1988         function replace_callback ($text, $callbacks) {
1989                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
1990                 $lastOpeningBrace = -1;                 # last not closed parentheses
1991
1992                 for ($i = 0; $i < strlen($text); $i++) {
1993                         # check for any opening brace
1994                         $rule = null;
1995                         $nextPos = -1;
1996                         foreach ($callbacks as $key => $value) {
1997                                 $pos = strpos ($text, $key, $i);
1998                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
1999                                         $rule = $value;
2000                                         $nextPos = $pos;
2001                                 }
2002                         }
2003
2004                         if ($lastOpeningBrace >= 0) {
2005                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2006
2007                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2008                                         $rule = null;
2009                                         $nextPos = $pos;
2010                                 }
2011
2012                                 $pos = strpos ($text, '|', $i);
2013
2014                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2015                                         $rule = null;
2016                                         $nextPos = $pos;
2017                                 }
2018                         }
2019
2020                         if ($nextPos == -1)
2021                                 break;
2022
2023                         $i = $nextPos;
2024
2025                         # found openning brace, lets add it to parentheses stack
2026                         if (null != $rule) {
2027                                 $piece = array('brace' => $text[$i],
2028                                                            'braceEnd' => $rule['end'],
2029                                                            'count' => 1,
2030                                                            'title' => '',
2031                                                            'parts' => null);
2032
2033                                 # count openning brace characters
2034                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2035                                         $piece['count']++;
2036                                         $i++;
2037                                 }
2038
2039                                 $piece['startAt'] = $i+1;
2040                                 $piece['partStart'] = $i+1;
2041
2042                                 # we need to add to stack only if openning brace count is enough for any given rule
2043                                 foreach ($rule['cb'] as $cnt => $fn) {
2044                                         if ($piece['count'] >= $cnt) {
2045                                                 $lastOpeningBrace ++;
2046                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2047                                                 break;
2048                                         }
2049                                 }
2050
2051                                 continue;
2052                         }
2053                         else if ($lastOpeningBrace >= 0) {
2054                                 # first check if it is a closing brace
2055                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2056                                         # lets check if it is enough characters for closing brace
2057                                         $count = 1;
2058                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2059                                                 $count++;
2060
2061                                         # if there are more closing parentheses than opening ones, we parse less
2062                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2063                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2064
2065                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2066                                         $matchingCount = 0;
2067                                         $matchingCallback = null;
2068                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2069                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2070                                                         $matchingCount = $cnt;
2071                                                         $matchingCallback = $fn;
2072                                                 }
2073                                         }
2074
2075                                         if ($matchingCount == 0) {
2076                                                 $i += $count - 1;
2077                                                 continue;
2078                                         }
2079
2080                                         # lets set a title or last part (if '|' was found)
2081                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2082                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2083                                         else
2084                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2085
2086                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2087                                         $pieceEnd = $i + $matchingCount;
2088
2089                                         if( is_callable( $matchingCallback ) ) {
2090                                                 $cbArgs = array (
2091                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2092                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2093                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2094                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2095                                                                                  );
2096                                                 # finally we can call a user callback and replace piece of text
2097                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2098                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2099                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2100                                         }
2101                                         else {
2102                                                 # null value for callback means that parentheses should be parsed, but not replaced
2103                                                 $i += $matchingCount - 1;
2104                                         }
2105
2106                                         # reset last openning parentheses, but keep it in case there are unused characters
2107                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2108                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2109                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2110                                                                    'title' => '',
2111                                                                    'parts' => null,
2112                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2113                                         $openingBraceStack[$lastOpeningBrace--] = null;
2114
2115                                         if ($matchingCount < $piece['count']) {
2116                                                 $piece['count'] -= $matchingCount;
2117                                                 $piece['startAt'] -= $matchingCount;
2118                                                 $piece['partStart'] = $piece['startAt'];
2119                                                 # do we still qualify for any callback with remaining count?
2120                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2121                                                         if ($piece['count'] >= $cnt) {
2122                                                                 $lastOpeningBrace ++;
2123                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2124                                                                 break;
2125                                                         }
2126                                                 }
2127                                         }
2128                                         continue;
2129                                 }
2130
2131                                 # lets set a title if it is a first separator, or next part otherwise
2132                                 if ($text[$i] == '|') {
2133                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2134                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2135                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2136                                         }
2137                                         else
2138                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2139
2140                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2141                                 }
2142                         }
2143                 }
2144
2145                 return $text;
2146         }
2147
2148         /**
2149          * Replace magic variables, templates, and template arguments
2150          * with the appropriate text. Templates are substituted recursively,
2151          * taking care to avoid infinite loops.
2152          *
2153          * Note that the substitution depends on value of $mOutputType:
2154          *  OT_WIKI: only {{subst:}} templates
2155          *  OT_MSG: only magic variables
2156          *  OT_HTML: all templates and magic variables
2157          *
2158          * @param string $tex The text to transform
2159          * @param array $args Key-value pairs representing template parameters to substitute
2160          * @access private
2161          */
2162         function replaceVariables( $text, $args = array() ) {
2163                 # Prevent too big inclusions
2164                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2165                         return $text;
2166                 }
2167
2168                 $fname = 'Parser::replaceVariables';
2169                 wfProfileIn( $fname );
2170
2171                 # This function is called recursively. To keep track of arguments we need a stack:
2172                 array_push( $this->mArgStack, $args );
2173
2174                 $braceCallbacks = array();
2175                 $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2176                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2177                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2178                 }
2179                 $callbacks = array();
2180                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2181                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2182                 $text = $this->replace_callback ($text, $callbacks);
2183
2184                 array_pop( $this->mArgStack );
2185
2186                 wfProfileOut( $fname );
2187                 return $text;
2188         }
2189
2190         /**
2191          * Replace magic variables
2192          * @access private
2193          */
2194         function variableSubstitution( $matches ) {
2195                 $fname = 'parser::variableSubstitution';
2196                 $varname = $matches[1];
2197                 wfProfileIn( $fname );
2198                 if ( !$this->mVariables ) {
2199                         $this->initialiseVariables();
2200                 }
2201                 $skip = false;
2202                 if ( $this->mOutputType == OT_WIKI ) {
2203                         # Do only magic variables prefixed by SUBST
2204                         $mwSubst =& MagicWord::get( MAG_SUBST );
2205                         if (!$mwSubst->matchStartAndRemove( $varname ))
2206                                 $skip = true;
2207                         # Note that if we don't substitute the variable below,
2208                         # we don't remove the {{subst:}} magic word, in case
2209                         # it is a template rather than a magic variable.
2210                 }
2211                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2212                         $id = $this->mVariables[$varname];
2213                         $text = $this->getVariableValue( $id );
2214                         $this->mOutput->mContainsOldMagic = true;
2215                 } else {
2216                         $text = $matches[0];
2217                 }
2218                 wfProfileOut( $fname );
2219                 return $text;
2220         }
2221
2222         # Split template arguments
2223         function getTemplateArgs( $argsString ) {
2224                 if ( $argsString === '' ) {
2225                         return array();
2226                 }
2227
2228                 $args = explode( '|', substr( $argsString, 1 ) );
2229
2230                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2231                 # merged with the next arg because the '|' character between belongs
2232                 # to the link syntax and not the template parameter syntax.
2233                 $argc = count($args);
2234
2235                 for ( $i = 0; $i < $argc-1; $i++ ) {
2236                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2237                                 $args[$i] .= '|'.$args[$i+1];
2238                                 array_splice($args, $i+1, 1);
2239                                 $i--;
2240                                 $argc--;
2241                         }
2242                 }
2243
2244                 return $args;
2245         }
2246
2247         /**
2248          * Return the text of a template, after recursively
2249          * replacing any variables or templates within the template.
2250          *
2251          * @param array $piece The parts of the template
2252          *  $piece['text']: matched text
2253          *  $piece['title']: the title, i.e. the part before the |
2254          *  $piece['parts']: the parameter array
2255          * @return string the text of the template
2256          * @access private
2257          */
2258         function braceSubstitution( $piece ) {
2259                 global $wgLinkCache, $wgContLang;
2260                 $fname = 'Parser::braceSubstitution';
2261                 wfProfileIn( $fname );
2262
2263                 $found = false;
2264                 $nowiki = false;
2265                 $noparse = false;
2266
2267                 $title = NULL;
2268
2269                 $linestart = '';
2270
2271                 # $part1 is the bit before the first |, and must contain only title characters
2272                 # $args is a list of arguments, starting from index 0, not including $part1
2273
2274                 $part1 = $piece['title'];
2275                 # If the third subpattern matched anything, it will start with |
2276
2277                 if (null == $piece['parts']) {
2278                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2279                         if ($replaceWith != $piece['text']) {
2280                                 $text = $replaceWith;
2281                                 $found = true;
2282                                 $noparse = true;
2283                         }
2284                 }
2285
2286                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2287                 $argc = count( $args );
2288
2289                 # SUBST
2290                 if ( !$found ) {
2291                         $mwSubst =& MagicWord::get( MAG_SUBST );
2292                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2293                                 # One of two possibilities is true:
2294                                 # 1) Found SUBST but not in the PST phase
2295                                 # 2) Didn't find SUBST and in the PST phase
2296                                 # In either case, return without further processing
2297                                 $text = $piece['text'];
2298                                 $found = true;
2299                                 $noparse = true;
2300                         }
2301                 }
2302
2303                 # MSG, MSGNW and INT
2304                 if ( !$found ) {
2305                         # Check for MSGNW:
2306                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2307                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2308                                 $nowiki = true;
2309                         } else {
2310                                 # Remove obsolete MSG:
2311                                 $mwMsg =& MagicWord::get( MAG_MSG );
2312                                 $mwMsg->matchStartAndRemove( $part1 );
2313                         }
2314
2315                         # Check if it is an internal message
2316                         $mwInt =& MagicWord::get( MAG_INT );
2317                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2318                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2319                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2320                                         $found = true;
2321                                 }
2322                         }
2323                 }
2324
2325                 # NS
2326                 if ( !$found ) {
2327                         # Check for NS: (namespace expansion)
2328                         $mwNs = MagicWord::get( MAG_NS );
2329                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2330                                 if ( intval( $part1 ) ) {
2331                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2332                                         $found = true;
2333                                 } else {
2334                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2335                                         if ( !is_null( $index ) ) {
2336                                                 $text = $linestart . $wgContLang->getNsText( $index );
2337                                                 $found = true;
2338                                         }
2339                                 }
2340                         }
2341                 }
2342
2343                 # LCFIRST, UCFIRST, LC and UC
2344                 if ( !$found ) {
2345                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2346                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2347                         $lc =& MagicWord::get( MAG_LC );
2348                         $uc =& MagicWord::get( MAG_UC );
2349                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2350                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2351                                 $found = true;
2352                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2353                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2354                                 $found = true;
2355                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2356                                 $text = $linestart . $wgContLang->lc( $part1 );
2357                                 $found = true;
2358                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2359                                  $text = $linestart . $wgContLang->uc( $part1 );
2360                                  $found = true;
2361                         }
2362                 }
2363
2364                 # LOCALURL and FULLURL
2365                 if ( !$found ) {
2366                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2367                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2368                         $mwFull =& MagicWord::get( MAG_FULLURL );
2369                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2370
2371
2372                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2373                                 $func = 'getLocalURL';
2374                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2375                                 $func = 'escapeLocalURL';
2376                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2377                                 $func = 'getFullURL';
2378                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2379                                 $func = 'escapeFullURL';
2380                         } else {
2381                                 $func = false;
2382                         }
2383
2384                         if ( $func !== false ) {
2385                                 $title = Title::newFromText( $part1 );
2386                                 if ( !is_null( $title ) ) {
2387                                         if ( $argc > 0 ) {
2388                                                 $text = $linestart . $title->$func( $args[0] );
2389                                         } else {
2390                                                 $text = $linestart . $title->$func();
2391                                         }
2392                                         $found = true;
2393                                 }
2394                         }
2395                 }
2396
2397                 # GRAMMAR
2398                 if ( !$found && $argc == 1 ) {
2399                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2400                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2401                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2402                                 $found = true;
2403                         }
2404                 }
2405
2406                 # PLURAL
2407                 if ( !$found && $argc >= 2 ) {
2408                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2409                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2410                                 if ($argc==2) {$args[2]=$args[1];}
2411                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2412                                 $found = true;
2413                         }
2414                 }
2415
2416                 # Template table test
2417
2418                 # Did we encounter this template already? If yes, it is in the cache
2419                 # and we need to check for loops.
2420                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2421                         $found = true;
2422
2423                         # Infinite loop test
2424                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2425                                 $noparse = true;
2426                                 $found = true;
2427                                 $text = $linestart .
2428                                         "\{\{$part1}}" .
2429                                         '<!-- WARNING: template loop detected -->';
2430                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2431                         } else {
2432                                 # set $text to cached message.
2433                                 $text = $linestart . $this->mTemplates[$part1];
2434                         }
2435                 }
2436
2437                 # Load from database
2438                 $replaceHeadings = false;
2439                 $isHTML = false;
2440                 $lastPathLevel = $this->mTemplatePath;
2441                 if ( !$found ) {
2442                         $ns = NS_TEMPLATE;
2443                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2444                         if ($subpage !== '') {
2445                                 $ns = $this->mTitle->getNamespace();
2446                         }
2447                         $title = Title::newFromText( $part1, $ns );
2448
2449                         if ($title) {
2450                                 $interwiki = Title::getInterwikiLink($title->getInterwiki());
2451                                 if ($interwiki != '' && $title->isTrans()) {
2452                                         return $this->scarytransclude($title, $interwiki);
2453                                 }
2454                         }
2455
2456                         if ( !is_null( $title ) && !$title->isExternal() ) {
2457                                 # Check for excessive inclusion
2458                                 $dbk = $title->getPrefixedDBkey();
2459                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2460                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2461                                                 # Capture special page output
2462                                                 $text = SpecialPage::capturePath( $title );
2463                                                 if ( is_string( $text ) ) {
2464                                                         $found = true;
2465                                                         $noparse = true;
2466                                                         $isHTML = true;
2467                                                         $this->disableCache();
2468                                                 }
2469                                         } else {
2470                                                 $article = new Article( $title );
2471                                                 $articleContent = $article->fetchContent(0, false);
2472                                                 if ( $articleContent !== false ) {
2473                                                         $found = true;
2474                                                         $text = $articleContent;
2475                                                         $replaceHeadings = true;
2476                                                 }
2477                                         }
2478                                 }
2479
2480                                 # If the title is valid but undisplayable, make a link to it
2481                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2482                                         $text = '[['.$title->getPrefixedText().']]';
2483                                         $found = true;
2484                                 }
2485
2486                                 # Template cache array insertion
2487                                 if( $found ) {
2488                                         $this->mTemplates[$part1] = $text;
2489                                         $text = $linestart . $text;
2490                                 }
2491                         }
2492                 }
2493
2494                 # Recursive parsing, escaping and link table handling
2495                 # Only for HTML output
2496                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2497                         $text = wfEscapeWikiText( $text );
2498                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2499                         # Clean up argument array
2500                         $assocArgs = array();
2501                         $index = 1;
2502                         foreach( $args as $arg ) {
2503                                 $eqpos = strpos( $arg, '=' );
2504                                 if ( $eqpos === false ) {
2505                                         $assocArgs[$index++] = $arg;
2506                                 } else {
2507                                         $name = trim( substr( $arg, 0, $eqpos ) );
2508                                         $value = trim( substr( $arg, $eqpos+1 ) );
2509                                         if ( $value === false ) {
2510                                                 $value = '';
2511                                         }
2512                                         if ( $name !== false ) {
2513                                                 $assocArgs[$name] = $value;
2514                                         }
2515                                 }
2516                         }
2517
2518                         # Add a new element to the templace recursion path
2519                         $this->mTemplatePath[$part1] = 1;
2520
2521                         # If there are any <onlyinclude> tags, only include them
2522                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2523                                 preg_match_all( '/<onlyinclude>(.*?)<\/onlyinclude>/s', $text, $m );
2524                                 $text = '';
2525                                 foreach ($m[1] as $piece)
2526                                         $text .= $this->trimOnlyinclude( $piece );
2527                         }
2528                         # Remove <noinclude> sections and <includeonly> tags
2529                         $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2530                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2531
2532                         if( $this->mOutputType == OT_HTML ) {
2533                                 # Strip <nowiki>, <pre>, etc.
2534                                 $text = $this->strip( $text, $this->mStripState );
2535                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2536                         }
2537                         $text = $this->replaceVariables( $text, $assocArgs );
2538
2539                         # Resume the link cache and register the inclusion as a link
2540                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2541                                 $wgLinkCache->addLinkObj( $title );
2542                         }
2543
2544                         # If the template begins with a table or block-level
2545                         # element, it should be treated as beginning a new line.
2546                         if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2547                                 $text = "\n" . $text;
2548                         }
2549                 }
2550                 # Prune lower levels off the recursion check path
2551                 $this->mTemplatePath = $lastPathLevel;
2552
2553                 if ( !$found ) {
2554                         wfProfileOut( $fname );
2555                         return $piece['text'];
2556                 } else {
2557                         if ( $isHTML ) {
2558                                 # Replace raw HTML by a placeholder
2559                                 # Add a blank line preceding, to prevent it from mucking up
2560                                 # immediately preceding headings
2561                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2562                         } else {
2563                                 # replace ==section headers==
2564                                 # XXX this needs to go away once we have a better parser.
2565                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2566                                         if( !is_null( $title ) )
2567                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2568                                         else
2569                                                 $encodedname = base64_encode("");
2570                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2571                                                 PREG_SPLIT_DELIM_CAPTURE);
2572                                         $text = '';
2573                                         $nsec = 0;
2574                                         for( $i = 0; $i < count($m); $i += 2 ) {
2575                                                 $text .= $m[$i];
2576                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2577                                                 $hl = $m[$i + 1];
2578                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2579                                                         $text .= $hl;
2580                                                         continue;
2581                                                 }
2582                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2583                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2584                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2585
2586                                                 $nsec++;
2587                                         }
2588                                 }
2589                         }
2590                 }
2591
2592                 # Prune lower levels off the recursion check path
2593                 $this->mTemplatePath = $lastPathLevel;
2594
2595                 if ( !$found ) {
2596                         wfProfileOut( $fname );
2597                         return $piece['text'];
2598                 } else {
2599                         wfProfileOut( $fname );
2600                         return $text;
2601                 }
2602         }
2603
2604         /**
2605          * Trim the first and last newlines of a string, this is not equivalent
2606          * to trim( $str, "\n" ) which would trim them all.
2607          *
2608          * @param string $str The string to trim
2609          * @return string
2610          */
2611         function trimOnlyinclude( $str ) {
2612                 $str = preg_replace( "/^\n/", '', $str );
2613                 $str = preg_replace( "/\n$/", '', $str );
2614                 return $str;
2615         }
2616
2617         /**
2618          * Translude an interwiki link.
2619          */
2620         function scarytransclude($title, $interwiki) {
2621                 global $wgEnableScaryTranscluding;
2622
2623                 if (!$wgEnableScaryTranscluding)
2624                         return wfMsg('scarytranscludedisabled');
2625
2626                 $articlename = "Template:" . $title->getDBkey();
2627                 $url = str_replace('$1', urlencode($articlename), $interwiki);
2628                 if (strlen($url) > 255)
2629                         return wfMsg('scarytranscludetoolong');
2630                 $text = $this->fetchScaryTemplateMaybeFromCache($url);
2631                 $this->mIWTransData[] = $text;
2632                 return "<!--IW_TRANSCLUDE ".(count($this->mIWTransData) - 1)."-->";
2633         }
2634
2635         function fetchScaryTemplateMaybeFromCache($url) {
2636                 $dbr =& wfGetDB(DB_SLAVE);
2637                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2638                                 array('tc_url' => $url));
2639                 if ($obj) {
2640                         $time = $obj->tc_time;
2641                         $text = $obj->tc_contents;
2642                         if ($time && $time < (time() + (60*60))) {
2643                                 return $text;
2644                         }
2645                 }
2646
2647                 $text = wfGetHTTP($url . '?action=render');
2648                 if (!$text)
2649                         return wfMsg('scarytranscludefailed', $url);
2650
2651                 $dbw =& wfGetDB(DB_MASTER);
2652                 $dbw->replace('transcache', array(), array(
2653                         'tc_url' => $url,
2654                         'tc_time' => time(),
2655                         'tc_contents' => $text));
2656                 return $text;
2657         }
2658
2659
2660         /**
2661          * Triple brace replacement -- used for template arguments
2662          * @access private
2663          */
2664         function argSubstitution( $matches ) {
2665                 $arg = trim( $matches['title'] );
2666                 $text = $matches['text'];
2667                 $inputArgs = end( $this->mArgStack );
2668
2669                 if ( array_key_exists( $arg, $inputArgs ) ) {
2670                         $text = $inputArgs[$arg];
2671                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2672                         $text = $matches['parts'][0];
2673                 }
2674
2675                 return $text;
2676         }
2677
2678         /**
2679          * Returns true if the function is allowed to include this entity
2680          * @access private
2681          */
2682         function incrementIncludeCount( $dbk ) {
2683                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2684                         $this->mIncludeCount[$dbk] = 0;
2685                 }
2686                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2687                         return true;
2688                 } else {
2689                         return false;
2690                 }
2691         }
2692
2693         /**
2694          * This function accomplishes several tasks:
2695          * 1) Auto-number headings if that option is enabled
2696          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2697          * 3) Add a Table of contents on the top for users who have enabled the option
2698          * 4) Auto-anchor headings
2699          *
2700          * It loops through all headlines, collects the necessary data, then splits up the
2701          * string and re-inserts the newly formatted headlines.
2702          *
2703          * @param string $text
2704          * @param boolean $isMain
2705          * @access private
2706          */
2707         function formatHeadings( $text, $isMain=true ) {
2708                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2709
2710                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2711                 $doShowToc = true;
2712                 $forceTocHere = false;
2713                 if( !$this->mTitle->userCanEdit() ) {
2714                         $showEditLink = 0;
2715                 } else {
2716                         $showEditLink = $this->mOptions->getEditSection();
2717                 }
2718
2719                 # Inhibit editsection links if requested in the page
2720                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2721                 if( $esw->matchAndRemove( $text ) ) {
2722                         $showEditLink = 0;
2723                 }
2724                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2725                 # do not add TOC
2726                 $mw =& MagicWord::get( MAG_NOTOC );
2727                 if( $mw->matchAndRemove( $text ) ) {
2728                         $doShowToc = false;
2729                 }
2730
2731                 # Get all headlines for numbering them and adding funky stuff like [edit]
2732                 # links - this is for later, but we need the number of headlines right now
2733                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2734
2735                 # if there are fewer than 4 headlines in the article, do not show TOC
2736                 if( $numMatches < 4 ) {
2737                         $doShowToc = false;
2738                 }
2739
2740                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2741                 # override above conditions and always show TOC at that place
2742
2743                 $mw =& MagicWord::get( MAG_TOC );
2744                 if($mw->match( $text ) ) {
2745                         $doShowToc = true;
2746                         $forceTocHere = true;
2747                 } else {
2748                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2749                         # override above conditions and always show TOC above first header
2750                         $mw =& MagicWord::get( MAG_FORCETOC );
2751                         if ($mw->matchAndRemove( $text ) ) {
2752                                 $doShowToc = true;
2753                         }
2754                 }
2755
2756                 # Never ever show TOC if no headers
2757                 if( $numMatches < 1 ) {
2758                         $doShowToc = false;
2759                 }
2760
2761                 # We need this to perform operations on the HTML
2762                 $sk =& $this->mOptions->getSkin();
2763
2764                 # headline counter
2765                 $headlineCount = 0;
2766                 $sectionCount = 0; # headlineCount excluding template sections
2767
2768                 # Ugh .. the TOC should have neat indentation levels which can be
2769                 # passed to the skin functions. These are determined here
2770                 $toc = '';
2771                 $full = '';
2772                 $head = array();
2773                 $sublevelCount = array();
2774                 $levelCount = array();
2775                 $toclevel = 0;
2776                 $level = 0;
2777                 $prevlevel = 0;
2778                 $toclevel = 0;
2779                 $prevtoclevel = 0;
2780
2781                 foreach( $matches[3] as $headline ) {
2782                         $istemplate = 0;
2783                         $templatetitle = '';
2784                         $templatesection = 0;
2785                         $numbering = '';
2786
2787                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2788                                 $istemplate = 1;
2789                                 $templatetitle = base64_decode($mat[1]);
2790                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2791                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2792                         }
2793
2794                         if( $toclevel ) {
2795                                 $prevlevel = $level;
2796                                 $prevtoclevel = $toclevel;
2797                         }
2798                         $level = $matches[1][$headlineCount];
2799
2800                         if( $doNumberHeadings || $doShowToc ) {
2801
2802                                 if ( $level > $prevlevel ) {
2803                                         # Increase TOC level
2804                                         $toclevel++;
2805                                         $sublevelCount[$toclevel] = 0;
2806                                         $toc .= $sk->tocIndent();
2807                                 }
2808                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2809                                         # Decrease TOC level, find level to jump to
2810
2811                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2812                                                 # Can only go down to level 1
2813                                                 $toclevel = 1;
2814                                         } else {
2815                                                 for ($i = $toclevel; $i > 0; $i--) {
2816                                                         if ( $levelCount[$i] == $level ) {
2817                                                                 # Found last matching level
2818                                                                 $toclevel = $i;
2819                                                                 break;
2820                                                         }
2821                                                         elseif ( $levelCount[$i] < $level ) {
2822                                                                 # Found first matching level below current level
2823                                                                 $toclevel = $i + 1;
2824                                                                 break;
2825                                                         }
2826                                                 }
2827                                         }
2828
2829                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2830                                 }
2831                                 else {
2832                                         # No change in level, end TOC line
2833                                         $toc .= $sk->tocLineEnd();
2834                                 }
2835
2836                                 $levelCount[$toclevel] = $level;
2837
2838                                 # count number of headlines for each level
2839                                 @$sublevelCount[$toclevel]++;
2840                                 $dot = 0;
2841                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2842                                         if( !empty( $sublevelCount[$i] ) ) {
2843                                                 if( $dot ) {
2844                                                         $numbering .= '.';
2845                                                 }
2846                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2847                                                 $dot = 1;
2848                                         }
2849                                 }
2850                         }
2851
2852                         # The canonized header is a version of the header text safe to use for links
2853                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2854                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2855                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2856
2857                         # Remove link placeholders by the link text.
2858                         #     <!--LINK number-->
2859                         # turns into
2860                         #     link text with suffix
2861                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2862                                                             "\$this->mLinkHolders['texts'][\$1]",
2863                                                             $canonized_headline );
2864                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2865                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2866                                                             $canonized_headline );
2867
2868                         # strip out HTML
2869                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2870                         $tocline = trim( $canonized_headline );
2871                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2872                         $replacearray = array(
2873                                 '%3A' => ':',
2874                                 '%' => '.'
2875                         );
2876                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2877                         $refers[$headlineCount] = $canonized_headline;
2878
2879                         # count how many in assoc. array so we can track dupes in anchors
2880                         @$refers[$canonized_headline]++;
2881                         $refcount[$headlineCount]=$refers[$canonized_headline];
2882
2883                         # Don't number the heading if it is the only one (looks silly)
2884                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2885                                 # the two are different if the line contains a link
2886                                 $headline=$numbering . ' ' . $headline;
2887                         }
2888
2889                         # Create the anchor for linking from the TOC to the section
2890                         $anchor = $canonized_headline;
2891                         if($refcount[$headlineCount] > 1 ) {
2892                                 $anchor .= '_' . $refcount[$headlineCount];
2893                         }
2894                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2895                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2896                         }
2897                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2898                                 if ( empty( $head[$headlineCount] ) ) {
2899                                         $head[$headlineCount] = '';
2900                                 }
2901                                 if( $istemplate )
2902                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2903                                 else
2904                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2905                         }
2906
2907                         # give headline the correct <h#> tag
2908                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2909
2910                         $headlineCount++;
2911                         if( !$istemplate )
2912                                 $sectionCount++;
2913                 }
2914
2915                 if( $doShowToc ) {
2916                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2917                         $toc = $sk->tocList( $toc );
2918                 }
2919
2920                 # split up and insert constructed headlines
2921
2922                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2923                 $i = 0;
2924
2925                 foreach( $blocks as $block ) {
2926                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2927                                 # This is the [edit] link that appears for the top block of text when
2928                                 # section editing is enabled
2929
2930                                 # Disabled because it broke block formatting
2931                                 # For example, a bullet point in the top line
2932                                 # $full .= $sk->editSectionLink(0);
2933                         }
2934                         $full .= $block;
2935                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2936                         # Top anchor now in skin
2937                                 $full = $full.$toc;
2938                         }
2939
2940                         if( !empty( $head[$i] ) ) {
2941                                 $full .= $head[$i];
2942                         }
2943                         $i++;
2944                 }
2945                 if($forceTocHere) {
2946                         $mw =& MagicWord::get( MAG_TOC );
2947                         return $mw->replace( $toc, $full );
2948                 } else {
2949                         return $full;
2950                 }
2951         }
2952
2953         /**
2954          * Return an HTML link for the "ISBN 123456" text
2955          * @access private
2956          */
2957         function magicISBN( $text ) {
2958                 $fname = 'Parser::magicISBN';
2959                 wfProfileIn( $fname );
2960
2961                 $a = split( 'ISBN ', ' '.$text );
2962                 if ( count ( $a ) < 2 ) {
2963                         wfProfileOut( $fname );
2964                         return $text;
2965                 }
2966                 $text = substr( array_shift( $a ), 1);
2967                 $valid = '0123456789-Xx';
2968
2969                 foreach ( $a as $x ) {
2970                         $isbn = $blank = '' ;
2971                         while ( ' ' == $x{0} ) {
2972                                 $blank .= ' ';
2973                                 $x = substr( $x, 1 );
2974                         }
2975                         if ( $x == '' ) { # blank isbn
2976                                 $text .= "ISBN $blank";
2977                                 continue;
2978                         }
2979                         while ( strstr( $valid, $x{0} ) != false ) {
2980                                 $isbn .= $x{0};
2981                                 $x = substr( $x, 1 );
2982                         }
2983                         $num = str_replace( '-', '', $isbn );
2984                         $num = str_replace( ' ', '', $num );
2985                         $num = str_replace( 'x', 'X', $num );
2986
2987                         if ( '' == $num ) {
2988                                 $text .= "ISBN $blank$x";
2989                         } else {
2990                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2991                                 $text .= '<a href="' .
2992                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2993                                         "\" class=\"internal\">ISBN $isbn</a>";
2994                                 $text .= $x;
2995                         }
2996                 }
2997                 wfProfileOut( $fname );
2998                 return $text;
2999         }
3000
3001         /**
3002          * Return an HTML link for the "RFC 1234" text
3003          *
3004          * @access private
3005          * @param string $text     Text to be processed
3006          * @param string $keyword  Magic keyword to use (default RFC)
3007          * @param string $urlmsg   Interface message to use (default rfcurl)
3008          * @return string
3009          */
3010         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3011
3012                 $valid = '0123456789';
3013                 $internal = false;
3014
3015                 $a = split( $keyword, ' '.$text );
3016                 if ( count ( $a ) < 2 ) {
3017                         return $text;
3018                 }
3019                 $text = substr( array_shift( $a ), 1);
3020
3021                 /* Check if keyword is preceed by [[.
3022                  * This test is made here cause of the array_shift above
3023                  * that prevent the test to be done in the foreach.
3024                  */
3025                 if ( substr( $text, -2 ) == '[[' ) {
3026                         $internal = true;
3027                 }
3028
3029                 foreach ( $a as $x ) {
3030                         /* token might be empty if we have RFC RFC 1234 */
3031                         if ( $x=='' ) {
3032                                 $text.=$keyword;
3033                                 continue;
3034                                 }
3035
3036                         $id = $blank = '' ;
3037
3038                         /** remove and save whitespaces in $blank */
3039                         while ( $x{0} == ' ' ) {
3040                                 $blank .= ' ';
3041                                 $x = substr( $x, 1 );
3042                         }
3043
3044                         /** remove and save the rfc number in $id */
3045                         while ( strstr( $valid, $x{0} ) != false ) {
3046                                 $id .= $x{0};
3047                                 $x = substr( $x, 1 );
3048                         }
3049
3050                         if ( $id == '' ) {
3051                                 /* call back stripped spaces*/
3052                                 $text .= $keyword.$blank.$x;
3053                         } elseif( $internal ) {
3054                                 /* normal link */
3055                                 $text .= $keyword.$id.$x;
3056                         } else {
3057                                 /* build the external link*/
3058                                 $url = wfMsg( $urlmsg, $id);
3059                                 $sk =& $this->mOptions->getSkin();
3060                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3061                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3062                         }
3063
3064                         /* Check if the next RFC keyword is preceed by [[ */
3065                         $internal = ( substr($x,-2) == '[[' );
3066                 }
3067                 return $text;
3068         }
3069
3070         /**
3071          * Transform wiki markup when saving a page by doing \r\n -> \n
3072          * conversion, substitting signatures, {{subst:}} templates, etc.
3073          *
3074          * @param string $text the text to transform
3075          * @param Title &$title the Title object for the current article
3076          * @param User &$user the User object describing the current user
3077          * @param ParserOptions $options parsing options
3078          * @param bool $clearState whether to clear the parser state first
3079          * @return string the altered wiki markup
3080          * @access public
3081          */
3082         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3083                 $this->mOptions = $options;
3084                 $this->mTitle =& $title;
3085                 $this->mOutputType = OT_WIKI;
3086
3087                 if ( $clearState ) {
3088                         $this->clearState();
3089                 }
3090
3091                 $stripState = false;
3092                 $pairs = array(
3093                         "\r\n" => "\n",
3094                 );
3095                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3096                 $text = $this->strip( $text, $stripState, true );
3097                 $text = $this->pstPass2( $text, $user );
3098                 $text = $this->unstrip( $text, $stripState );
3099                 $text = $this->unstripNoWiki( $text, $stripState );
3100                 return $text;
3101         }
3102
3103         /**
3104          * Pre-save transform helper function
3105          * @access private
3106          */
3107         function pstPass2( $text, &$user ) {
3108                 global $wgContLang, $wgLocaltimezone;
3109
3110                 # Variable replacement
3111                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3112                 $text = $this->replaceVariables( $text );
3113
3114                 # Signatures
3115                 #
3116                 $sigText = $this->getUserSig( $user );
3117
3118                 /* Note: This is the timestamp saved as hardcoded wikitext to
3119                  * the database, we use $wgContLang here in order to give
3120                  * everyone the same signiture and use the default one rather
3121                  * than the one selected in each users preferences.
3122                  */
3123                 if ( isset( $wgLocaltimezone ) ) {
3124                         $oldtz = getenv( 'TZ' );
3125                         putenv( 'TZ='.$wgLocaltimezone );
3126                 }
3127                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3128                   ' (' . date( 'T' ) . ')';
3129                 if ( isset( $wgLocaltimezone ) ) {
3130                         putenv( 'TZ='.$oldtz );
3131                 }
3132
3133                 $text = preg_replace( '/~~~~~/', $d, $text );
3134                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
3135                 $text = preg_replace( '/~~~/', $sigText, $text );
3136
3137                 # Context links: [[|name]] and [[name (context)|]]
3138                 #
3139                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
3140                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
3141                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3142                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3143
3144                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3145                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3146                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3147                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3148                 $context = '';
3149                 $t = $this->mTitle->getText();
3150                 if ( preg_match( $conpat, $t, $m ) ) {
3151                         $context = $m[2];
3152                 }
3153                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3154                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3155                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3156
3157                 if ( '' == $context ) {
3158                         $text = preg_replace( $p2, '[[\\1]]', $text );
3159                 } else {
3160                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3161                 }
3162
3163                 # Trim trailing whitespace
3164                 # MAG_END (__END__) tag allows for trailing
3165                 # whitespace to be deliberately included
3166                 $text = rtrim( $text );
3167                 $mw =& MagicWord::get( MAG_END );
3168                 $mw->matchAndRemove( $text );
3169
3170                 return $text;
3171         }
3172
3173         /**
3174          * Fetch the user's signature text, if any, and normalize to
3175          * validated, ready-to-insert wikitext.
3176          *
3177          * @param User $user
3178          * @return string
3179          * @access private
3180          */
3181         function getUserSig( &$user ) {
3182                 $name = $user->getName();
3183                 $nick = trim( $user->getOption( 'nickname' ) );
3184                 if ( '' == $nick ) {
3185                         $nick = $name;
3186                 }
3187
3188                 if( $user->getOption( 'fancysig' ) ) {
3189                         // A wikitext signature.
3190                         $valid = $this->validateSig( $nick );
3191                         if( $valid === false ) {
3192                                 // Fall back to default sig
3193                                 $nick = $name;
3194                                 wfDebug( "Parser::getUserSig: $name has bad XML tags in signature.\n" );
3195                         } else {
3196                                 return $nick;
3197                         }
3198                 }
3199
3200                 // Plain text linking to the user's homepage
3201                 global $wgContLang;
3202                 $page = $user->getUserPage();
3203                 return '[[' .
3204                         $page->getPrefixedText() .
3205                         "|" .
3206                         wfEscapeWikIText( $nick ) .
3207                         "]]";
3208         }
3209
3210         /**
3211          * We want to enforce two rules on wikitext sigs here:
3212          * 1) Expand any templates at save time (forced subst:)
3213          * 2) Check for unbalanced XML tags, and reject if so.
3214          *
3215          * @param string $text
3216          * @return mixed An expanded string, or false if invalid.
3217          *
3218          * @todo Run brace substitutions
3219          * @todo ?? Check for unbalanced '' and ''' quotes, etc
3220          */
3221         function validateSig( $text ) {
3222                 if( wfIsWellFormedXmlFragment( $text ) ) {
3223                         return $text;
3224                 } else {
3225                         return false;
3226                 }
3227         }
3228
3229         /**
3230          * Set up some variables which are usually set up in parse()
3231          * so that an external function can call some class members with confidence
3232          * @access public
3233          */
3234         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3235                 $this->mTitle =& $title;
3236                 $this->mOptions = $options;
3237                 $this->mOutputType = $outputType;
3238                 if ( $clearState ) {
3239                         $this->clearState();
3240                 }
3241         }
3242
3243         /**
3244          * Transform a MediaWiki message by replacing magic variables.
3245          *
3246          * @param string $text the text to transform
3247          * @param ParserOptions $options  options
3248          * @return string the text with variables substituted
3249          * @access public
3250          */
3251         function transformMsg( $text, $options ) {
3252                 global $wgTitle;
3253                 static $executing = false;
3254
3255                 # Guard against infinite recursion
3256                 if ( $executing ) {
3257                         return $text;
3258                 }
3259                 $executing = true;
3260
3261                 $this->mTitle = $wgTitle;
3262                 $this->mOptions = $options;
3263                 $this->mOutputType = OT_MSG;
3264                 $this->clearState();
3265                 $text = $this->replaceVariables( $text );
3266
3267                 $executing = false;
3268                 return $text;
3269         }
3270
3271         /**
3272          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3273          * Callback will be called with the text within
3274          * Transform and return the text within
3275          *
3276          * @access public
3277          *
3278          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3279          * @param mixed $callback The callback function (and object) to use for the tag
3280          *
3281          * @return The old value of the mTagHooks array associated with the hook
3282          */
3283         function setHook( $tag, $callback ) {
3284                 $oldVal = @$this->mTagHooks[$tag];
3285                 $this->mTagHooks[$tag] = $callback;
3286
3287                 return $oldVal;
3288         }
3289
3290         /**
3291          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3292          * Placeholders created in Skin::makeLinkObj()
3293          * Returns an array of links found, indexed by PDBK:
3294          *  0 - broken
3295          *  1 - normal link
3296          *  2 - stub
3297          * $options is a bit field, RLH_FOR_UPDATE to select for update
3298          */
3299         function replaceLinkHolders( &$text, $options = 0 ) {
3300                 global $wgUser, $wgLinkCache;
3301                 global $wgOutputReplace;
3302
3303                 $fname = 'Parser::replaceLinkHolders';
3304                 wfProfileIn( $fname );
3305
3306                 $pdbks = array();
3307                 $colours = array();
3308                 $sk = $this->mOptions->getSkin();
3309
3310                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3311                         wfProfileIn( $fname.'-check' );
3312                         $dbr =& wfGetDB( DB_SLAVE );
3313                         $page = $dbr->tableName( 'page' );
3314                         $threshold = $wgUser->getOption('stubthreshold');
3315
3316                         # Sort by namespace
3317                         asort( $this->mLinkHolders['namespaces'] );
3318
3319                         # Generate query
3320                         $query = false;
3321                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
3322                                 # Make title object
3323                                 $title = $this->mLinkHolders['titles'][$key];
3324
3325                                 # Skip invalid entries.
3326                                 # Result will be ugly, but prevents crash.
3327                                 if ( is_null( $title ) ) {
3328                                         continue;
3329                                 }
3330                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3331
3332                                 # Check if it's in the link cache already
3333                                 if ( $title->isAlwaysKnown() || $wgLinkCache->getGoodLinkID( $pdbk ) ) {
3334                                         $colours[$pdbk] = 1;
3335                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
3336                                         $colours[$pdbk] = 0;
3337                                 } else {
3338                                         # Not in the link cache, add it to the query
3339                                         if ( !isset( $current ) ) {
3340                                                 $current = $val;
3341                                                 $query =  "SELECT page_id, page_namespace, page_title";
3342                                                 if ( $threshold > 0 ) {
3343                                                         $query .= ', page_len, page_is_redirect';
3344                                                 }
3345                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
3346                                         } elseif ( $current != $val ) {
3347                                                 $current = $val;
3348                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
3349                                         } else {
3350                                                 $query .= ', ';
3351                                         }
3352
3353                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3354                                 }
3355                         }
3356                         if ( $query ) {
3357                                 $query .= '))';
3358                                 if ( $options & RLH_FOR_UPDATE ) {
3359                                         $query .= ' FOR UPDATE';
3360                                 }
3361
3362                                 $res = $dbr->query( $query, $fname );
3363
3364                                 # Fetch data and form into an associative array
3365                                 # non-existent = broken
3366                                 # 1 = known
3367                                 # 2 = stub
3368                                 while ( $s = $dbr->fetchObject($res) ) {
3369                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3370                                         $pdbk = $title->getPrefixedDBkey();
3371                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
3372
3373                                         if ( $threshold >  0 ) {
3374                                                 $size = $s->page_len;
3375                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3376                                                         $colours[$pdbk] = 1;
3377                                                 } else {
3378                                                         $colours[$pdbk] = 2;
3379                                                 }
3380                                         } else {
3381                                                 $colours[$pdbk] = 1;
3382                                         }
3383                                 }
3384                         }
3385                         wfProfileOut( $fname.'-check' );
3386
3387                         # Construct search and replace arrays
3388                         wfProfileIn( $fname.'-construct' );
3389                         $wgOutputReplace = array();
3390                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3391                                 $pdbk = $pdbks[$key];
3392                                 $searchkey = "<!--LINK $key-->";
3393                                 $title = $this->mLinkHolders['titles'][$key];
3394                                 if ( empty( $colours[$pdbk] ) ) {
3395                                         $wgLinkCache->addBadLinkObj( $title );
3396                                         $colours[$pdbk] = 0;
3397                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3398                                                                         $this->mLinkHolders['texts'][$key],
3399                                                                         $this->mLinkHolders['queries'][$key] );
3400                                 } elseif ( $colours[$pdbk] == 1 ) {
3401                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3402                                                                         $this->mLinkHolders['texts'][$key],
3403                                                                         $this->mLinkHolders['queries'][$key] );
3404                                 } elseif ( $colours[$pdbk] == 2 ) {
3405                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3406                                                                         $this->mLinkHolders['texts'][$key],
3407                                                                         $this->mLinkHolders['queries'][$key] );
3408                                 }
3409                         }
3410                         wfProfileOut( $fname.'-construct' );
3411
3412                         # Do the thing
3413                         wfProfileIn( $fname.'-replace' );
3414
3415                         $text = preg_replace_callback(
3416                                 '/(<!--LINK .*?-->)/',
3417                                 "wfOutputReplaceMatches",
3418                                 $text);
3419
3420                         wfProfileOut( $fname.'-replace' );
3421                 }
3422
3423                 # Now process interwiki link holders
3424                 # This is quite a bit simpler than internal links
3425                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3426                         wfProfileIn( $fname.'-interwiki' );
3427                         # Make interwiki link HTML
3428                         $wgOutputReplace = array();
3429                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3430                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3431                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3432                         }
3433
3434                         $text = preg_replace_callback(
3435                                 '/<!--IWLINK (.*?)-->/',
3436                                 "wfOutputReplaceMatches",
3437                                 $text );
3438                         wfProfileOut( $fname.'-interwiki' );
3439                 }
3440
3441                 wfProfileOut( $fname );
3442                 return $colours;
3443         }
3444
3445         /**
3446          * Replace <!--LINK--> link placeholders with plain text of links
3447          * (not HTML-formatted).
3448          * @param string $text
3449          * @return string
3450          */
3451         function replaceLinkHoldersText( $text ) {
3452                 global $wgUser, $wgLinkCache;
3453                 global $wgOutputReplace;
3454
3455                 $fname = 'Parser::replaceLinkHoldersText';
3456                 wfProfileIn( $fname );
3457
3458                 $text = preg_replace_callback(
3459                         '/<!--(LINK|IWLINK) (.*?)-->/',
3460                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3461                         $text );
3462
3463                 wfProfileOut( $fname );
3464                 return $text;
3465         }
3466
3467         /**
3468          * @param array $matches
3469          * @return string
3470          * @access private
3471          */
3472         function replaceLinkHoldersTextCallback( $matches ) {
3473                 $type = $matches[1];
3474                 $key  = $matches[2];
3475                 if( $type == 'LINK' ) {
3476                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3477                                 return $this->mLinkHolders['texts'][$key];
3478                         }
3479                 } elseif( $type == 'IWLINK' ) {
3480                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3481                                 return $this->mInterwikiLinkHolders['texts'][$key];
3482                         }
3483                 }
3484                 return $matches[0];
3485         }
3486
3487         /**
3488          * Renders an image gallery from a text with one line per image.
3489          * text labels may be given by using |-style alternative text. E.g.
3490          *   Image:one.jpg|The number "1"
3491          *   Image:tree.jpg|A tree
3492          * given as text will return the HTML of a gallery with two images,
3493          * labeled 'The number "1"' and
3494          * 'A tree'.
3495          *
3496          * @static
3497          */
3498         function renderImageGallery( $text ) {
3499                 # Setup the parser
3500                 global $wgUser, $wgTitle;
3501                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3502                 $localParser = new Parser();
3503
3504                 global $wgLinkCache;
3505                 $ig = new ImageGallery();
3506                 $ig->setShowBytes( false );
3507                 $ig->setShowFilename( false );
3508                 $lines = explode( "\n", $text );
3509
3510                 foreach ( $lines as $line ) {
3511                         # match lines like these:
3512                         # Image:someimage.jpg|This is some image
3513                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3514                         # Skip empty lines
3515                         if ( count( $matches ) == 0 ) {
3516                                 continue;
3517                         }
3518                         $nt = Title::newFromURL( $matches[1] );
3519                         if( is_null( $nt ) ) {
3520                                 # Bogus title. Ignore these so we don't bomb out later.
3521                                 continue;
3522                         }
3523                         if ( isset( $matches[3] ) ) {
3524                                 $label = $matches[3];
3525                         } else {
3526                                 $label = '';
3527                         }
3528
3529                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3530                         $html = $html->mText;
3531
3532                         $ig->add( new Image( $nt ), $html );
3533                         $wgLinkCache->addImageLinkObj( $nt );
3534                 }
3535                 return $ig->toHTML();
3536         }
3537
3538         /**
3539          * Parse image options text and use it to make an image
3540          */
3541         function makeImage( &$nt, $options ) {
3542                 global $wgContLang, $wgUseImageResize;
3543                 global $wgUser, $wgThumbLimits;
3544
3545                 $align = '';
3546
3547                 # Check if the options text is of the form "options|alt text"
3548                 # Options are:
3549                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3550                 #  * left               no resizing, just left align. label is used for alt= only
3551                 #  * right              same, but right aligned
3552                 #  * none               same, but not aligned
3553                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3554                 #  * center             center the image
3555                 #  * framed             Keep original image size, no magnify-button.
3556
3557                 $part = explode( '|', $options);
3558
3559                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3560                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3561                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3562                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3563                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3564                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3565                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3566                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3567                 $caption = '';
3568
3569                 $width = $height = $framed = $thumb = false;
3570                 $manual_thumb = '' ;
3571
3572                 foreach( $part as $key => $val ) {
3573                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3574                                 $thumb=true;
3575                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3576                                 # use manually specified thumbnail
3577                                 $thumb=true;
3578                                 $manual_thumb = $match;
3579                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3580                                 # remember to set an alignment, don't render immediately
3581                                 $align = 'right';
3582                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3583                                 # remember to set an alignment, don't render immediately
3584                                 $align = 'left';
3585                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3586                                 # remember to set an alignment, don't render immediately
3587                                 $align = 'center';
3588                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3589                                 # remember to set an alignment, don't render immediately
3590                                 $align = 'none';
3591                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3592                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3593                                 # $match is the image width in pixels
3594                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3595                                         $width = intval( $m[1] );
3596                                         $height = intval( $m[2] );
3597                                 } else {
3598                                         $width = intval($match);
3599                                 }
3600                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3601                                 $framed=true;
3602                         } else {
3603                                 $caption = $val;
3604                         }
3605                 }
3606                 # Strip bad stuff out of the alt text
3607                 $alt = $this->replaceLinkHoldersText( $caption );
3608                 $alt = Sanitizer::stripAllTags( $alt );
3609
3610                 # Linker does the rest
3611                 $sk =& $this->mOptions->getSkin();
3612                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3613         }
3614
3615         /**
3616          * Set a flag in the output object indicating that the content is dynamic and
3617          * shouldn't be cached.
3618          */
3619         function disableCache() {
3620                 $this->mOutput->mCacheTime = -1;
3621         }
3622
3623         /**
3624          * Callback from the Sanitizer for expanding items found in HTML attribute
3625          * values, so they can be safely tested and escaped.
3626          * @param string $text
3627          * @param array $args
3628          * @return string
3629          * @access private
3630          */
3631         function attributeStripCallback( &$text, $args ) {
3632                 $text = $this->replaceVariables( $text, $args );
3633                 $text = $this->unstripForHTML( $text );
3634                 return $text;
3635         }
3636
3637         function unstripForHTML( $text ) {
3638                 $text = $this->unstrip( $text, $this->mStripState );
3639                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3640                 return $text;
3641         }
3642 }
3643
3644 /**
3645  * @todo document
3646  * @package MediaWiki
3647  */
3648 class ParserOutput
3649 {
3650         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3651         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3652         var $mVersion;   # Compatibility check
3653         var $mTitleText; # title text of the chosen language variant
3654
3655         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3656                 $containsOldMagic = false, $titletext = '' )
3657         {
3658                 $this->mText = $text;
3659                 $this->mLanguageLinks = $languageLinks;
3660                 $this->mCategoryLinks = $categoryLinks;
3661                 $this->mContainsOldMagic = $containsOldMagic;
3662                 $this->mCacheTime = '';
3663                 $this->mVersion = MW_PARSER_VERSION;
3664                 $this->mTitleText = $titletext;
3665         }
3666
3667         function getText()                   { return $this->mText; }
3668         function getLanguageLinks()          { return $this->mLanguageLinks; }
3669         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3670         function getCacheTime()              { return $this->mCacheTime; }
3671         function getTitleText()              { return $this->mTitleText; }
3672         function containsOldMagic()          { return $this->mContainsOldMagic; }
3673         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3674         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3675         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3676         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3677         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3678         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3679
3680         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3681
3682         function merge( $other ) {
3683                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3684                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3685                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3686         }
3687
3688         /**
3689          * Return true if this cached output object predates the global or
3690          * per-article cache invalidation timestamps, or if it comes from
3691          * an incompatible older version.
3692          *
3693          * @param string $touched the affected article's last touched timestamp
3694          * @return bool
3695          * @access public
3696          */
3697         function expired( $touched ) {
3698                 global $wgCacheEpoch;
3699                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3700                        $this->getCacheTime() < $touched ||
3701                        $this->getCacheTime() <= $wgCacheEpoch ||
3702                        !isset( $this->mVersion ) ||
3703                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3704         }
3705 }
3706
3707 /**
3708  * Set options of the Parser
3709  * @todo document
3710  * @package MediaWiki
3711  */
3712 class ParserOptions
3713 {
3714         # All variables are private
3715         var $mUseTeX;                    # Use texvc to expand <math> tags
3716         var $mUseDynamicDates;           # Use DateFormatter to format dates
3717         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3718         var $mAllowExternalImages;       # Allow external images inline
3719         var $mAllowExternalImagesFrom;   # If not, any exception?
3720         var $mSkin;                      # Reference to the preferred skin
3721         var $mDateFormat;                # Date format index
3722         var $mEditSection;               # Create "edit section" links
3723         var $mNumberHeadings;            # Automatically number headings
3724         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3725
3726         function getUseTeX()                        { return $this->mUseTeX; }
3727         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3728         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3729         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3730         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3731         function &getSkin()                         { return $this->mSkin; }
3732         function getDateFormat()                    { return $this->mDateFormat; }
3733         function getEditSection()                   { return $this->mEditSection; }
3734         function getNumberHeadings()                { return $this->mNumberHeadings; }
3735         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3736
3737
3738         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3739         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3740         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3741         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3742         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3743         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3744         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3745         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3746         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3747
3748         function setSkin( &$x ) { $this->mSkin =& $x; }
3749
3750         function ParserOptions() {
3751                 global $wgUser;
3752                 $this->initialiseFromUser( $wgUser );
3753         }
3754
3755         /**
3756          * Get parser options
3757          * @static
3758          */
3759         function newFromUser( &$user ) {
3760                 $popts = new ParserOptions;
3761                 $popts->initialiseFromUser( $user );
3762                 return $popts;
3763         }
3764
3765         /** Get user options */
3766         function initialiseFromUser( &$userInput ) {
3767                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3768                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3769                 $fname = 'ParserOptions::initialiseFromUser';
3770                 wfProfileIn( $fname );
3771                 if ( !$userInput ) {
3772                         $user = new User;
3773                         $user->setLoaded( true );
3774                 } else {
3775                         $user =& $userInput;
3776                 }
3777
3778                 $this->mUseTeX = $wgUseTeX;
3779                 $this->mUseDynamicDates = $wgUseDynamicDates;
3780                 $this->mInterwikiMagic = $wgInterwikiMagic;
3781                 $this->mAllowExternalImages = $wgAllowExternalImages;
3782                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
3783                 wfProfileIn( $fname.'-skin' );
3784                 $this->mSkin =& $user->getSkin();
3785                 wfProfileOut( $fname.'-skin' );
3786                 $this->mDateFormat = $user->getOption( 'date' );
3787                 $this->mEditSection = true;
3788                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3789                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3790                 wfProfileOut( $fname );
3791         }
3792 }
3793
3794 /**
3795  * Callback function used by Parser::replaceLinkHolders()
3796  * to substitute link placeholders.
3797  */
3798 function &wfOutputReplaceMatches( $matches ) {
3799         global $wgOutputReplace;
3800         return $wgOutputReplace[$matches[1]];
3801 }
3802
3803 /**
3804  * Return the total number of articles
3805  */
3806 function wfNumberOfArticles() {
3807         global $wgNumberOfArticles;
3808
3809         wfLoadSiteStats();
3810         return $wgNumberOfArticles;
3811 }
3812
3813 /**
3814  * Return the number of files
3815  */
3816 function wfNumberOfFiles() {
3817         $fname = 'Parser::wfNumberOfFiles';
3818
3819         wfProfileIn( $fname );
3820         $dbr =& wfGetDB( DB_SLAVE );
3821         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3822         wfProfileOut( $fname );
3823
3824         return $res;
3825 }
3826
3827 /**
3828  * Get various statistics from the database
3829  * @private
3830  */
3831 function wfLoadSiteStats() {
3832         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3833         $fname = 'wfLoadSiteStats';
3834
3835         if ( -1 != $wgNumberOfArticles ) return;
3836         $dbr =& wfGetDB( DB_SLAVE );
3837         $s = $dbr->selectRow( 'site_stats',
3838                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3839                 array( 'ss_row_id' => 1 ), $fname
3840         );
3841
3842         if ( $s === false ) {
3843                 return;
3844         } else {
3845                 $wgTotalViews = $s->ss_total_views;
3846                 $wgTotalEdits = $s->ss_total_edits;
3847                 $wgNumberOfArticles = $s->ss_good_articles;
3848         }
3849 }
3850
3851 /**
3852  * Escape html tags
3853  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3854  *
3855  * @param string $in Text that might contain HTML tags
3856  * @return string Escaped string
3857  */
3858 function wfEscapeHTMLTagsOnly( $in ) {
3859         return str_replace(
3860                 array( '"', '>', '<' ),
3861                 array( '&quot;', '&gt;', '&lt;' ),
3862                 $in );
3863 }
3864
3865 ?>