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