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
  12 /**
  13  * Update this version number when the ParserOutput format
  14  * changes in an incompatible way, so the parser cache
  15  * can automatically discard old data.
  16  */
  17 define( 'MW_PARSER_VERSION', '1.5.0' );
  18
  19 /**
  20  * Variable substitution O(N^2) attack
  21  *
  22  * Without countermeasures, it would be possible to attack the parser by saving
  23  * a page filled with a large number of inclusions of large pages. The size of
  24  * the generated page would be proportional to the square of the input size.
  25  * Hence, we limit the number of inclusions of any given page, thus bringing any
  26  * attack back to O(N).
  27  */
  28
  29 define( 'MAX_INCLUDE_REPEAT', 100 );
  30 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  31
  32 define( 'RLH_FOR_UPDATE', 1 );
  33
  34 # Allowed values for $mOutputType
  35 define( 'OT_HTML', 1 );
  36 define( 'OT_WIKI', 2 );
  37 define( 'OT_MSG' , 3 );
  38
  39 # string parameter for extractTags which will cause it
  40 # to strip HTML comments in addition to regular
  41 # <XML>-style tags. This should not be anything we
  42 # may want to use in wikisyntax
  43 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  44
  45 # prefix for escaping, used in two functions at least
  46 define( 'UNIQ_PREFIX', 'NaodW29');
  47
  48 # Constants needed for external link processing
  49 define( 'URL_PROTOCOLS', 'http|https|ftp|irc|gopher|news|mailto' );
  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',  '/\[(\b('.URL_PROTOCOLS.'):'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/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
  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         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @access public
 119          */
 120         function Parser() {
 121                 global $wgContLang;
 122                 $this->mTemplates = array();
 123                 $this->mTemplatePath = array();
 124                 $this->mTagHooks = array();
 125                 $this->clearState();
 126         }
 127
 128         /**
 129          * Clear Parser state
 130          *
 131          * @access private
 132          */
 133         function clearState() {
 134                 $this->mOutput = new ParserOutput;
 135                 $this->mAutonumber = 0;
 136                 $this->mLastSection = '';
 137                 $this->mDTopen = false;
 138                 $this->mVariables = false;
 139                 $this->mIncludeCount = array();
 140                 $this->mStripState = array();
 141                 $this->mArgStack = array();
 142                 $this->mInPre = false;
 143                 $this->mInterwikiLinkHolders = array();
 144                 $this->mLinkHolders = array(
 145                         'namespaces' => array(),
 146                         'dbkeys' => array(),
 147                         'queries' => array(),
 148                         'texts' => array(),
 149                         'titles' => array()
 150                 );
 151         }
 152
 153         /**
 154          * First pass--just handle <nowiki> sections, pass the rest off
 155          * to internalParse() which does all the real work.
 156          *
 157          * @access private
 158          * @param string $text Text we want to parse
 159          * @param Title &$title A title object
 160          * @param array $options
 161          * @param boolean $linestart
 162          * @param boolean $clearState
 163          * @return ParserOutput a ParserOutput
 164          */
 165         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 166                 global $wgUseTidy, $wgContLang;
 167                 $fname = 'Parser::parse';
 168                 wfProfileIn( $fname );
 169
 170                 if ( $clearState ) {
 171                         $this->clearState();
 172                 }
 173
 174                 $this->mOptions = $options;
 175                 $this->mTitle =& $title;
 176                 $this->mOutputType = OT_HTML;
 177
 178                 $this->mStripState = NULL;
 179
 180                 //$text = $this->strip( $text, $this->mStripState );
 181                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 182                 $x =& $this->mStripState;
 183                 $text = $this->strip( $text, $x );
 184
 185                 $text = $this->internalParse( $text );
 186
 187                 $dashReplace = array(
 188                         '/ - /' => "&nbsp;&ndash; ", # N dash
 189                         '/(?<=[0-9])-(?=[0-9])/' => "&ndash;", # N dash between numbers
 190                         '/ -- /' => "&nbsp;&mdash; " # M dash
 191                 );
 192                 $text = preg_replace( array_keys($dashReplace), array_values($dashReplace), $text );
 193
 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                         '/<hr *>/i' => '<hr />',
 205                         '/<br *>/i' => '<br />',
 206                         '/<center *>/i' => '<div class="center">',
 207                         '/<\\/center *>/i' => '</div>',
 208                 );
 209                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 210
 211                 # only once and last
 212                 $text = $this->doBlockLevels( $text, $linestart );
 213
 214                 $this->replaceLinkHolders( $text );
 215
 216                 # the position of the convert() call should not be changed. it
 217                 # assumes that the links are all replaces and the only thing left
 218                 # is the <nowiki> mark.
 219                 $text = $wgContLang->convert($text);
 220                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 221
 222                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 223
 224                 $text = Sanitizer::normalizeCharReferences( $text );
 225                 global $wgUseTidy;
 226                 if ($wgUseTidy) {
 227                         $text = Parser::tidy($text);
 228                 }
 229
 230                 $this->mOutput->setText( $text );
 231                 wfProfileOut( $fname );
 232                 return $this->mOutput;
 233         }
 234
 235         /**
 236          * Get a random string
 237          *
 238          * @access private
 239          * @static
 240          */
 241         function getRandomString() {
 242                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 243         }
 244
 245         /**
 246          * Replaces all occurrences of <$tag>content</$tag> in the text
 247          * with a random marker and returns the new text. the output parameter
 248          * $content will be an associative array filled with data on the form
 249          * $unique_marker => content.
 250          *
 251          * If $content is already set, the additional entries will be appended
 252          * If $tag is set to STRIP_COMMENTS, the function will extract
 253          * <!-- HTML comments -->
 254          *
 255          * @access private
 256          * @static
 257          */
 258         function extractTags($tag, $text, &$content, $uniq_prefix = ''){
 259                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 260                 if ( !$content ) {
 261                         $content = array( );
 262                 }
 263                 $n = 1;
 264                 $stripped = '';
 265
 266                 while ( '' != $text ) {
 267                         if($tag==STRIP_COMMENTS) {
 268                                 $p = preg_split( '/<!--/', $text, 2 );
 269                         } else {
 270                                 $p = preg_split( "/<\\s*$tag\\s*>/i", $text, 2 );
 271                         }
 272                         $stripped .= $p[0];
 273                         if ( ( count( $p ) < 2 ) || ( '' == $p[1] ) ) {
 274                                 $text = '';
 275                         } else {
 276                                 if($tag==STRIP_COMMENTS) {
 277                                         $q = preg_split( '/-->/i', $p[1], 2 );
 278                                 } else {
 279                                         $q = preg_split( "/<\\/\\s*$tag\\s*>/i", $p[1], 2 );
 280                                 }
 281                                 $marker = $rnd . sprintf('%08X', $n++);
 282                                 $content[$marker] = $q[0];
 283                                 $stripped .= $marker;
 284                                 $text = $q[1];
 285                         }
 286                 }
 287                 return $stripped;
 288         }
 289
 290         /**
 291          * Strips and renders nowiki, pre, math, hiero
 292          * If $render is set, performs necessary rendering operations on plugins
 293          * Returns the text, and fills an array with data needed in unstrip()
 294          * If the $state is already a valid strip state, it adds to the state
 295          *
 296          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 297          *  will be stripped in addition to other tags. This is important
 298          *  for section editing, where these comments cause confusion when
 299          *  counting the sections in the wikisource
 300          *
 301          * @access private
 302          */
 303         function strip( $text, &$state, $stripcomments = false ) {
 304                 $render = ($this->mOutputType == OT_HTML);
 305                 $html_content = array();
 306                 $nowiki_content = array();
 307                 $math_content = array();
 308                 $pre_content = array();
 309                 $comment_content = array();
 310                 $ext_content = array();
 311                 $gallery_content = array();
 312
 313                 # Replace any instances of the placeholders
 314                 $uniq_prefix = UNIQ_PREFIX;
 315                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 316
 317                 # html
 318                 global $wgRawHtml, $wgWhitelistEdit;
 319                 if( $wgRawHtml && $wgWhitelistEdit ) {
 320                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 321                         foreach( $html_content as $marker => $content ) {
 322                                 if ($render ) {
 323                                         # Raw and unchecked for validity.
 324                                         $html_content[$marker] = $content;
 325                                 } else {
 326                                         $html_content[$marker] = '<html>'.$content.'</html>';
 327                                 }
 328                         }
 329                 }
 330
 331                 # nowiki
 332                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 333                 foreach( $nowiki_content as $marker => $content ) {
 334                         if( $render ){
 335                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 336                         } else {
 337                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 338                         }
 339                 }
 340
 341                 # math
 342                 $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 343                 foreach( $math_content as $marker => $content ){
 344                         if( $render ) {
 345                                 if( $this->mOptions->getUseTeX() ) {
 346                                         $math_content[$marker] = renderMath( $content );
 347                                 } else {
 348                                         $math_content[$marker] = '&lt;math&gt;'.$content.'&lt;math&gt;';
 349                                 }
 350                         } else {
 351                                 $math_content[$marker] = '<math>'.$content.'</math>';
 352                         }
 353                 }
 354
 355                 # pre
 356                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 357                 foreach( $pre_content as $marker => $content ){
 358                         if( $render ){
 359                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 360                         } else {
 361                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 362                         }
 363                 }
 364
 365                 # gallery
 366                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 367                 foreach( $gallery_content as $marker => $content ) {
 368                         require_once( 'ImageGallery.php' );
 369                         if ( $render ) {
 370                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 371                         } else {
 372                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 373                         }
 374                 }
 375
 376                 # Comments
 377                 if($stripcomments) {
 378                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 379                         foreach( $comment_content as $marker => $content ){
 380                                 $comment_content[$marker] = '<!--'.$content.'-->';
 381                         }
 382                 }
 383
 384                 # Extensions
 385                 foreach ( $this->mTagHooks as $tag => $callback ) {
 386                         $ext_content[$tag] = array();
 387                         $text = Parser::extractTags( $tag, $text, $ext_content[$tag], $uniq_prefix );
 388                         foreach( $ext_content[$tag] as $marker => $content ) {
 389                                 if ( $render ) {
 390                                         $ext_content[$tag][$marker] = $callback( $content );
 391                                 } else {
 392                                         $ext_content[$tag][$marker] = "<$tag>$content</$tag>";
 393                                 }
 394                         }
 395                 }
 396
 397                 # Merge state with the pre-existing state, if there is one
 398                 if ( $state ) {
 399                         $state['html'] = $state['html'] + $html_content;
 400                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 401                         $state['math'] = $state['math'] + $math_content;
 402                         $state['pre'] = $state['pre'] + $pre_content;
 403                         $state['comment'] = $state['comment'] + $comment_content;
 404                         $state['gallery'] = $state['gallery'] + $gallery_content;
 405
 406                         foreach( $ext_content as $tag => $array ) {
 407                                 if ( array_key_exists( $tag, $state ) ) {
 408                                         $state[$tag] = $state[$tag] + $array;
 409                                 }
 410                         }
 411                 } else {
 412                         $state = array(
 413                           'html' => $html_content,
 414                           'nowiki' => $nowiki_content,
 415                           'math' => $math_content,
 416                           'pre' => $pre_content,
 417                           'comment' => $comment_content,
 418                           'gallery' => $gallery_content,
 419                         ) + $ext_content;
 420                 }
 421                 return $text;
 422         }
 423
 424         /**
 425          * restores pre, math, and hiero removed by strip()
 426          *
 427          * always call unstripNoWiki() after this one
 428          * @access private
 429          */
 430         function unstrip( $text, &$state ) {
 431                 # Must expand in reverse order, otherwise nested tags will be corrupted
 432                 $contentDict = end( $state );
 433                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 434                         if( key($state) != 'nowiki' && key($state) != 'html') {
 435                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 436                                         $text = str_replace( key( $contentDict ), $content, $text );
 437                                 }
 438                         }
 439                 }
 440
 441                 return $text;
 442         }
 443
 444         /**
 445          * always call this after unstrip() to preserve the order
 446          *
 447          * @access private
 448          */
 449         function unstripNoWiki( $text, &$state ) {
 450                 # Must expand in reverse order, otherwise nested tags will be corrupted
 451                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 452                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 453                 }
 454
 455                 global $wgRawHtml;
 456                 if ($wgRawHtml) {
 457                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 458                                 $text = str_replace( key( $state['html'] ), $content, $text );
 459                         }
 460                 }
 461
 462                 return $text;
 463         }
 464
 465         /**
 466          * Add an item to the strip state
 467          * Returns the unique tag which must be inserted into the stripped text
 468          * The tag will be replaced with the original text in unstrip()
 469          *
 470          * @access private
 471          */
 472         function insertStripItem( $text, &$state ) {
 473                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 474                 if ( !$state ) {
 475                         $state = array(
 476                           'html' => array(),
 477                           'nowiki' => array(),
 478                           'math' => array(),
 479                           'pre' => array()
 480                         );
 481                 }
 482                 $state['item'][$rnd] = $text;
 483                 return $rnd;
 484         }
 485
 486         /**
 487          * Interface with html tidy, used if $wgUseTidy = true.
 488          * If tidy isn't able to correct the markup, the original will be
 489          * returned in all its glory with a warning comment appended.
 490          *
 491          * Either the external tidy program or the in-process tidy extension
 492          * will be used depending on availability. Override the default
 493          * $wgTidyInternal setting to disable the internal if it's not working.
 494          *
 495          * @param string $text Hideous HTML input
 496          * @return string Corrected HTML output
 497          * @access public
 498          * @static
 499          */
 500         function tidy( $text ) {
 501                 global $wgTidyInternal;
 502                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 503 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 504 '<head><title>test</title></head><body>'.$text.'</body></html>';
 505                 if( $wgTidyInternal ) {
 506                         $correctedtext = Parser::internalTidy( $wrappedtext );
 507                 } else {
 508                         $correctedtext = Parser::externalTidy( $wrappedtext );
 509                 }
 510                 if( is_null( $correctedtext ) ) {
 511                         wfDebug( "Tidy error detected!\n" );
 512                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 513                 }
 514                 return $correctedtext;
 515         }
 516
 517         /**
 518          * Spawn an external HTML tidy process and get corrected markup back from it.
 519          *
 520          * @access private
 521          * @static
 522          */
 523         function externalTidy( $text ) {
 524                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 525                 $fname = 'Parser::externalTidy';
 526                 wfProfileIn( $fname );
 527
 528                 $cleansource = '';
 529                 $opts = ' -utf8';
 530
 531                 $descriptorspec = array(
 532                         0 => array('pipe', 'r'),
 533                         1 => array('pipe', 'w'),
 534                         2 => array('file', '/dev/null', 'a')
 535                 );
 536                 $pipes = array();
 537                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 538                 if (is_resource($process)) {
 539                         fwrite($pipes[0], $text);
 540                         fclose($pipes[0]);
 541                         while (!feof($pipes[1])) {
 542                                 $cleansource .= fgets($pipes[1], 1024);
 543                         }
 544                         fclose($pipes[1]);
 545                         proc_close($process);
 546                 }
 547
 548                 wfProfileOut( $fname );
 549
 550                 if( $cleansource == '' && $text != '') {
 551                         // Some kind of error happened, so we couldn't get the corrected text.
 552                         // Just give up; we'll use the source text and append a warning.
 553                         return null;
 554                 } else {
 555                         return $cleansource;
 556                 }
 557         }
 558
 559         /**
 560          * Use the HTML tidy PECL extension to use the tidy library in-process,
 561          * saving the overhead of spawning a new process. Currently written to
 562          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 563          *
 564          * 'pear install tidy' should be able to compile the extension module.
 565          *
 566          * @access private
 567          * @static
 568          */
 569         function internalTidy( $text ) {
 570                 global $wgTidyConf;
 571                 $fname = 'Parser::internalTidy';
 572                 wfProfileIn( $fname );
 573
 574                 tidy_load_config( $wgTidyConf );
 575                 tidy_set_encoding( 'utf8' );
 576                 tidy_parse_string( $text );
 577                 tidy_clean_repair();
 578                 if( tidy_get_status() == 2 ) {
 579                         // 2 is magic number for fatal error
 580                         // http://www.php.net/manual/en/function.tidy-get-status.php
 581                         $cleansource = null;
 582                 } else {
 583                         $cleansource = tidy_get_output();
 584                 }
 585                 wfProfileOut( $fname );
 586                 return $cleansource;
 587         }
 588
 589         /**
 590          * parse the wiki syntax used to render tables
 591          *
 592          * @access private
 593          */
 594         function doTableStuff ( $t ) {
 595                 $fname = 'Parser::doTableStuff';
 596                 wfProfileIn( $fname );
 597
 598                 $t = explode ( "\n" , $t ) ;
 599                 $td = array () ; # Is currently a td tag open?
 600                 $ltd = array () ; # Was it TD or TH?
 601                 $tr = array () ; # Is currently a tr tag open?
 602                 $ltr = array () ; # tr attributes
 603                 $indent_level = 0; # indent level of the table
 604                 foreach ( $t AS $k => $x )
 605                 {
 606                         $x = trim ( $x ) ;
 607                         $fc = substr ( $x , 0 , 1 ) ;
 608                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 609                                 $indent_level = strlen( $matches[1] );
 610                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 611                                         '<table' . Sanitizer::fixTagAttributes ( $matches[2], 'table' ) . '>' ;
 612                                 array_push ( $td , false ) ;
 613                                 array_push ( $ltd , '' ) ;
 614                                 array_push ( $tr , false ) ;
 615                                 array_push ( $ltr , '' ) ;
 616                         }
 617                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 618                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 619                                 $z = "</table>" . substr ( $x , 2);
 620                                 $l = array_pop ( $ltd ) ;
 621                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 622                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 623                                 array_pop ( $ltr ) ;
 624                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 625                         }
 626                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 627                                 $x = substr ( $x , 1 ) ;
 628                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 629                                 $z = '' ;
 630                                 $l = array_pop ( $ltd ) ;
 631                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 632                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 633                                 array_pop ( $ltr ) ;
 634                                 $t[$k] = $z ;
 635                                 array_push ( $tr , false ) ;
 636                                 array_push ( $td , false ) ;
 637                                 array_push ( $ltd , '' ) ;
 638                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $x, 'tr' ) ) ;
 639                         }
 640                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 641                                 # $x is a table row
 642                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 643                                         $fc = '+' ;
 644                                         $x = substr ( $x , 1 ) ;
 645                                 }
 646                                 $after = substr ( $x , 1 ) ;
 647                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 648                                 $after = explode ( '||' , $after ) ;
 649                                 $t[$k] = '' ;
 650
 651                                 # Loop through each table cell
 652                                 foreach ( $after AS $theline )
 653                                 {
 654                                         $z = '' ;
 655                                         if ( $fc != '+' )
 656                                         {
 657                                                 $tra = array_pop ( $ltr ) ;
 658                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 659                                                 array_push ( $tr , true ) ;
 660                                                 array_push ( $ltr , '' ) ;
 661                                         }
 662
 663                                         $l = array_pop ( $ltd ) ;
 664                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 665                                         if ( $fc == '|' ) $l = 'td' ;
 666                                         else if ( $fc == '!' ) $l = 'th' ;
 667                                         else if ( $fc == '+' ) $l = 'caption' ;
 668                                         else $l = '' ;
 669                                         array_push ( $ltd , $l ) ;
 670
 671                                         # Cell parameters
 672                                         $y = explode ( '|' , $theline , 2 ) ;
 673                                         # Note that a '|' inside an invalid link should not
 674                                         # be mistaken as delimiting cell parameters
 675                                         if ( strpos( $y[0], '[[' ) !== false ) {
 676                                                 $y = array ($theline);
 677                                         }
 678                                         if ( count ( $y ) == 1 )
 679                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 680                                         else $y = $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($y[0], $l).">{$y[1]}" ;
 681                                         $t[$k] .= $y ;
 682                                         array_push ( $td , true ) ;
 683                                 }
 684                         }
 685                 }
 686
 687                 # Closing open td, tr && table
 688                 while ( count ( $td ) > 0 )
 689                 {
 690                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 691                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 692                         $t[] = '</table>' ;
 693                 }
 694
 695                 $t = implode ( "\n" , $t ) ;
 696                 wfProfileOut( $fname );
 697                 return $t ;
 698         }
 699
 700         /**
 701          * Helper function for parse() that transforms wiki markup into
 702          * HTML. Only called for $mOutputType == OT_HTML.
 703          *
 704          * @access private
 705          */
 706         function internalParse( $text ) {
 707                 global $wgContLang;
 708                 $args = array();
 709                 $isMain = true;
 710                 $fname = 'Parser::internalParse';
 711                 wfProfileIn( $fname );
 712
 713                 $text = Sanitizer::removeHTMLtags( $text );
 714                 $text = $this->replaceVariables( $text, $args );
 715
 716                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 717
 718                 $text = $this->doHeadings( $text );
 719                 if($this->mOptions->getUseDynamicDates()) {
 720                         $df =& DateFormatter::getInstance();
 721                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 722                 }
 723                 $text = $this->doAllQuotes( $text );
 724                 $text = $this->replaceInternalLinks( $text );
 725                 $text = $this->replaceExternalLinks( $text );
 726
 727                 # replaceInternalLinks may sometimes leave behind
 728                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 729                 $text = str_replace("http-noparse://","http://",$text);
 730
 731                 $text = $this->doMagicLinks( $text );
 732                 $text = $this->doTableStuff( $text );
 733                 $text = $this->formatHeadings( $text, $isMain );
 734
 735                 wfProfileOut( $fname );
 736                 return $text;
 737         }
 738
 739         /**
 740          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 741          * magic external links.
 742          *
 743          * @access private
 744          */
 745         function &doMagicLinks( &$text ) {
 746                 $text = $this->magicISBN( $text );
 747                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 748                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 749                 return $text;
 750         }
 751
 752         /**
 753          * Parse ^^ tokens and return html
 754          *
 755          * @access private
 756          */
 757         function doExponent( $text ) {
 758                 $fname = 'Parser::doExponent';
 759                 wfProfileIn( $fname );
 760                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 761                 wfProfileOut( $fname );
 762                 return $text;
 763         }
 764
 765         /**
 766          * Parse headers and return html
 767          *
 768          * @access private
 769          */
 770         function doHeadings( $text ) {
 771                 $fname = 'Parser::doHeadings';
 772                 wfProfileIn( $fname );
 773                 for ( $i = 6; $i >= 1; --$i ) {
 774                         $h = substr( '======', 0, $i );
 775                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 776                           "<h{$i}>\\1</h{$i}>\\2", $text );
 777                 }
 778                 wfProfileOut( $fname );
 779                 return $text;
 780         }
 781
 782         /**
 783          * Replace single quotes with HTML markup
 784          * @access private
 785          * @return string the altered text
 786          */
 787         function doAllQuotes( $text ) {
 788                 $fname = 'Parser::doAllQuotes';
 789                 wfProfileIn( $fname );
 790                 $outtext = '';
 791                 $lines = explode( "\n", $text );
 792                 foreach ( $lines as $line ) {
 793                         $outtext .= $this->doQuotes ( $line ) . "\n";
 794                 }
 795                 $outtext = substr($outtext, 0,-1);
 796                 wfProfileOut( $fname );
 797                 return $outtext;
 798         }
 799
 800         /**
 801          * Helper function for doAllQuotes()
 802          * @access private
 803          */
 804         function doQuotes( $text ) {
 805                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 806                 if ( count( $arr ) == 1 )
 807                         return $text;
 808                 else
 809                 {
 810                         # First, do some preliminary work. This may shift some apostrophes from
 811                         # being mark-up to being text. It also counts the number of occurrences
 812                         # of bold and italics mark-ups.
 813                         $i = 0;
 814                         $numbold = 0;
 815                         $numitalics = 0;
 816                         foreach ( $arr as $r )
 817                         {
 818                                 if ( ( $i % 2 ) == 1 )
 819                                 {
 820                                         # If there are ever four apostrophes, assume the first is supposed to
 821                                         # be text, and the remaining three constitute mark-up for bold text.
 822                                         if ( strlen( $arr[$i] ) == 4 )
 823                                         {
 824                                                 $arr[$i-1] .= "'";
 825                                                 $arr[$i] = "'''";
 826                                         }
 827                                         # If there are more than 5 apostrophes in a row, assume they're all
 828                                         # text except for the last 5.
 829                                         else if ( strlen( $arr[$i] ) > 5 )
 830                                         {
 831                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 832                                                 $arr[$i] = "'''''";
 833                                         }
 834                                         # Count the number of occurrences of bold and italics mark-ups.
 835                                         # We are not counting sequences of five apostrophes.
 836                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 837                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 838                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 839                                 }
 840                                 $i++;
 841                         }
 842
 843                         # If there is an odd number of both bold and italics, it is likely
 844                         # that one of the bold ones was meant to be an apostrophe followed
 845                         # by italics. Which one we cannot know for certain, but it is more
 846                         # likely to be one that has a single-letter word before it.
 847                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 848                         {
 849                                 $i = 0;
 850                                 $firstsingleletterword = -1;
 851                                 $firstmultiletterword = -1;
 852                                 $firstspace = -1;
 853                                 foreach ( $arr as $r )
 854                                 {
 855                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 856                                         {
 857                                                 $x1 = substr ($arr[$i-1], -1);
 858                                                 $x2 = substr ($arr[$i-1], -2, 1);
 859                                                 if ($x1 == ' ') {
 860                                                         if ($firstspace == -1) $firstspace = $i;
 861                                                 } else if ($x2 == ' ') {
 862                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 863                                                 } else {
 864                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 865                                                 }
 866                                         }
 867                                         $i++;
 868                                 }
 869
 870                                 # If there is a single-letter word, use it!
 871                                 if ($firstsingleletterword > -1)
 872                                 {
 873                                         $arr [ $firstsingleletterword ] = "''";
 874                                         $arr [ $firstsingleletterword-1 ] .= "'";
 875                                 }
 876                                 # If not, but there's a multi-letter word, use that one.
 877                                 else if ($firstmultiletterword > -1)
 878                                 {
 879                                         $arr [ $firstmultiletterword ] = "''";
 880                                         $arr [ $firstmultiletterword-1 ] .= "'";
 881                                 }
 882                                 # ... otherwise use the first one that has neither.
 883                                 # (notice that it is possible for all three to be -1 if, for example,
 884                                 # there is only one pentuple-apostrophe in the line)
 885                                 else if ($firstspace > -1)
 886                                 {
 887                                         $arr [ $firstspace ] = "''";
 888                                         $arr [ $firstspace-1 ] .= "'";
 889                                 }
 890                         }
 891
 892                         # Now let's actually convert our apostrophic mush to HTML!
 893                         $output = '';
 894                         $buffer = '';
 895                         $state = '';
 896                         $i = 0;
 897                         foreach ($arr as $r)
 898                         {
 899                                 if (($i % 2) == 0)
 900                                 {
 901                                         if ($state == 'both')
 902                                                 $buffer .= $r;
 903                                         else
 904                                                 $output .= $r;
 905                                 }
 906                                 else
 907                                 {
 908                                         if (strlen ($r) == 2)
 909                                         {
 910                                                 if ($state == 'i')
 911                                                 { $output .= '</i>'; $state = ''; }
 912                                                 else if ($state == 'bi')
 913                                                 { $output .= '</i>'; $state = 'b'; }
 914                                                 else if ($state == 'ib')
 915                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 916                                                 else if ($state == 'both')
 917                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 918                                                 else # $state can be 'b' or ''
 919                                                 { $output .= '<i>'; $state .= 'i'; }
 920                                         }
 921                                         else if (strlen ($r) == 3)
 922                                         {
 923                                                 if ($state == 'b')
 924                                                 { $output .= '</b>'; $state = ''; }
 925                                                 else if ($state == 'bi')
 926                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 927                                                 else if ($state == 'ib')
 928                                                 { $output .= '</b>'; $state = 'i'; }
 929                                                 else if ($state == 'both')
 930                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 931                                                 else # $state can be 'i' or ''
 932                                                 { $output .= '<b>'; $state .= 'b'; }
 933                                         }
 934                                         else if (strlen ($r) == 5)
 935                                         {
 936                                                 if ($state == 'b')
 937                                                 { $output .= '</b><i>'; $state = 'i'; }
 938                                                 else if ($state == 'i')
 939                                                 { $output .= '</i><b>'; $state = 'b'; }
 940                                                 else if ($state == 'bi')
 941                                                 { $output .= '</i></b>'; $state = ''; }
 942                                                 else if ($state == 'ib')
 943                                                 { $output .= '</b></i>'; $state = ''; }
 944                                                 else if ($state == 'both')
 945                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
 946                                                 else # ($state == '')
 947                                                 { $buffer = ''; $state = 'both'; }
 948                                         }
 949                                 }
 950                                 $i++;
 951                         }
 952                         # Now close all remaining tags.  Notice that the order is important.
 953                         if ($state == 'b' || $state == 'ib')
 954                                 $output .= '</b>';
 955                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
 956                                 $output .= '</i>';
 957                         if ($state == 'bi')
 958                                 $output .= '</b>';
 959                         if ($state == 'both')
 960                                 $output .= '<b><i>'.$buffer.'</i></b>';
 961                         return $output;
 962                 }
 963         }
 964
 965         /**
 966          * Replace external links
 967          *
 968          * Note: this is all very hackish and the order of execution matters a lot.
 969          * Make sure to run maintenance/parserTests.php if you change this code.
 970          *
 971          * @access private
 972          */
 973         function replaceExternalLinks( $text ) {
 974                 $fname = 'Parser::replaceExternalLinks';
 975                 wfProfileIn( $fname );
 976
 977                 $sk =& $this->mOptions->getSkin();
 978
 979                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 980
 981                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
 982
 983                 $i = 0;
 984                 while ( $i<count( $bits ) ) {
 985                         $url = $bits[$i++];
 986                         $protocol = $bits[$i++];
 987                         $text = $bits[$i++];
 988                         $trail = $bits[$i++];
 989
 990                         # The characters '<' and '>' (which were escaped by
 991                         # removeHTMLtags()) should not be included in
 992                         # URLs, per RFC 2396.
 993                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
 994                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
 995                                 $url = substr($url, 0, $m2[0][1]);
 996                         }
 997
 998                         # If the link text is an image URL, replace it with an <img> tag
 999                         # This happened by accident in the original parser, but some people used it extensively
1000                         $img = $this->maybeMakeExternalImage( $text );
1001                         if ( $img !== false ) {
1002                                 $text = $img;
1003                         }
1004
1005                         $dtrail = '';
1006
1007                         # Set linktype for CSS - if URL==text, link is essentially free
1008                         $linktype = ($text == $url) ? 'free' : 'text';
1009
1010                         # No link text, e.g. [http://domain.tld/some.link]
1011                         if ( $text == '' ) {
1012                                 # Autonumber if allowed
1013                                 if ( strpos( HTTP_PROTOCOLS, $protocol ) !== false ) {
1014                                         $text = '[' . ++$this->mAutonumber . ']';
1015                                         $linktype = 'autonumber';
1016                                 } else {
1017                                         # Otherwise just use the URL
1018                                         $text = htmlspecialchars( $url );
1019                                         $linktype = 'free';
1020                                 }
1021                         } else {
1022                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1023                                 # Check for trail
1024                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1025                         }
1026
1027                         $text = $wgContLang->markNoConversion($text);
1028
1029                         # Replace &amp; from obsolete syntax with &.
1030                         # All HTML entities will be escaped by makeExternalLink()
1031                         # or maybeMakeExternalImage()
1032                         $url = str_replace( '&amp;', '&', $url );
1033
1034                         # Process the trail (i.e. everything after this link up until start of the next link),
1035                         # replacing any non-bracketed links
1036                         $trail = $this->replaceFreeExternalLinks( $trail );
1037
1038
1039                         # Use the encoded URL
1040                         # This means that users can paste URLs directly into the text
1041                         # Funny characters like &ouml; aren't valid in URLs anyway
1042                         # This was changed in August 2004
1043                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1044                 }
1045
1046                 wfProfileOut( $fname );
1047                 return $s;
1048         }
1049
1050         /**
1051          * Replace anything that looks like a URL with a link
1052          * @access private
1053          */
1054         function replaceFreeExternalLinks( $text ) {
1055                 global $wgContLang;
1056                 $fname = 'Parser::replaceFreeExternalLinks';
1057                 wfProfileIn( $fname );
1058
1059                 $bits = preg_split( '/(\b(?:'.URL_PROTOCOLS.'):)/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1060                 $s = array_shift( $bits );
1061                 $i = 0;
1062
1063                 $sk =& $this->mOptions->getSkin();
1064
1065                 while ( $i < count( $bits ) ){
1066                         $protocol = $bits[$i++];
1067                         $remainder = $bits[$i++];
1068
1069                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1070                                 # Found some characters after the protocol that look promising
1071                                 $url = $protocol . $m[1];
1072                                 $trail = $m[2];
1073
1074                                 # The characters '<' and '>' (which were escaped by
1075                                 # removeHTMLtags()) should not be included in
1076                                 # URLs, per RFC 2396.
1077                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1078                                         $trail = substr($url, $m2[0][1]) . $trail;
1079                                         $url = substr($url, 0, $m2[0][1]);
1080                                 }
1081
1082                                 # Move trailing punctuation to $trail
1083                                 $sep = ',;\.:!?';
1084                                 # If there is no left bracket, then consider right brackets fair game too
1085                                 if ( strpos( $url, '(' ) === false ) {
1086                                         $sep .= ')';
1087                                 }
1088
1089                                 $numSepChars = strspn( strrev( $url ), $sep );
1090                                 if ( $numSepChars ) {
1091                                         $trail = substr( $url, -$numSepChars ) . $trail;
1092                                         $url = substr( $url, 0, -$numSepChars );
1093                                 }
1094
1095                                 # Replace &amp; from obsolete syntax with &.
1096                                 # All HTML entities will be escaped by makeExternalLink()
1097                                 # or maybeMakeExternalImage()
1098                                 $url = str_replace( '&amp;', '&', $url );
1099
1100                                 # Is this an external image?
1101                                 $text = $this->maybeMakeExternalImage( $url );
1102                                 if ( $text === false ) {
1103                                         # Not an image, make a link
1104                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1105                                 }
1106                                 $s .= $text . $trail;
1107                         } else {
1108                                 $s .= $protocol . $remainder;
1109                         }
1110                 }
1111                 wfProfileOut();
1112                 return $s;
1113         }
1114
1115         /**
1116          * make an image if it's allowed
1117          * @access private
1118          */
1119         function maybeMakeExternalImage( $url ) {
1120                 $sk =& $this->mOptions->getSkin();
1121                 $text = false;
1122                 if ( $this->mOptions->getAllowExternalImages() ) {
1123                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1124                                 # Image found
1125                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1126                         }
1127                 }
1128                 return $text;
1129         }
1130
1131         /**
1132          * Process [[ ]] wikilinks
1133          *
1134          * @access private
1135          */
1136         function replaceInternalLinks( $s ) {
1137                 global $wgContLang, $wgLinkCache;
1138                 static $fname = 'Parser::replaceInternalLinks' ;
1139
1140                 wfProfileIn( $fname );
1141
1142                 wfProfileIn( $fname.'-setup' );
1143                 static $tc = FALSE;
1144                 # the % is needed to support urlencoded titles as well
1145                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1146
1147                 $sk =& $this->mOptions->getSkin();
1148
1149                 #split the entire text string on occurences of [[
1150                 $a = explode( '[[', ' ' . $s );
1151                 #get the first element (all text up to first [[), and remove the space we added
1152                 $s = array_shift( $a );
1153                 $s = substr( $s, 1 );
1154
1155                 # Match a link having the form [[namespace:link|alternate]]trail
1156                 static $e1 = FALSE;
1157                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1158                 # Match cases where there is no "]]", which might still be images
1159                 static $e1_img = FALSE;
1160                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1161                 # Match the end of a line for a word that's not followed by whitespace,
1162                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1163                 static $e2 = '/^(.*?)([a-zA-Z\x80-\xff]+)$/sD';
1164
1165                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1166
1167                 if( is_null( $this->mTitle ) ) {
1168                         wfDebugDieBacktrace( 'nooo' );
1169                 }
1170                 $nottalk = !$this->mTitle->isTalkPage();
1171
1172                 if ( $useLinkPrefixExtension ) {
1173                         if ( preg_match( $e2, $s, $m ) ) {
1174                                 $first_prefix = $m[2];
1175                                 $s = $m[1];
1176                         } else {
1177                                 $first_prefix = false;
1178                         }
1179                 } else {
1180                         $prefix = '';
1181                 }
1182
1183                 $selflink = $this->mTitle->getPrefixedText();
1184                 wfProfileOut( $fname.'-setup' );
1185
1186                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1187                 $useSubpages = $this->areSubpagesAllowed();
1188
1189                 # Loop for each link
1190                 for ($k = 0; isset( $a[$k] ); $k++) {
1191                         $line = $a[$k];
1192                         if ( $useLinkPrefixExtension ) {
1193                                 wfProfileIn( $fname.'-prefixhandling' );
1194                                 if ( preg_match( $e2, $s, $m ) ) {
1195                                         $prefix = $m[2];
1196                                         $s = $m[1];
1197                                 } else {
1198                                         $prefix='';
1199                                 }
1200                                 # first link
1201                                 if($first_prefix) {
1202                                         $prefix = $first_prefix;
1203                                         $first_prefix = false;
1204                                 }
1205                                 wfProfileOut( $fname.'-prefixhandling' );
1206                         }
1207
1208                         $might_be_img = false;
1209
1210                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1211                                 $text = $m[2];
1212                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1213                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1214                                 # the real problem is with the $e1 regex
1215                                 # See bug 1300.
1216                                 if (preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1217                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1218                                         $m[3] = $n[1];
1219                                 }
1220                                 # fix up urlencoded title texts
1221                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1222                                 $trail = $m[3];
1223                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1224                                 $might_be_img = true;
1225                                 $text = $m[2];
1226                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1227                                 $trail = "";
1228                         } else { # Invalid form; output directly
1229                                 $s .= $prefix . '[[' . $line ;
1230                                 continue;
1231                         }
1232
1233                         # Don't allow internal links to pages containing
1234                         # PROTO: where PROTO is a valid URL protocol; these
1235                         # should be external links.
1236                         if (preg_match('/^(\b(?:'.URL_PROTOCOLS.'):)/', $m[1])) {
1237                                 $s .= $prefix . '[[' . $line ;
1238                                 continue;
1239                         }
1240
1241                         # Make subpage if necessary
1242                         if( $useSubpages ) {
1243                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1244                         } else {
1245                                 $link = $m[1];
1246                         }
1247
1248                         $noforce = (substr($m[1], 0, 1) != ':');
1249                         if (!$noforce) {
1250                                 # Strip off leading ':'
1251                                 $link = substr($link, 1);
1252                         }
1253
1254                         $nt =& Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1255                         if( !$nt ) {
1256                                 $s .= $prefix . '[[' . $line;
1257                                 continue;
1258                         }
1259
1260                         #check other language variants of the link
1261                         #if the article does not exist
1262                         if( $checkVariantLink
1263                             && $nt->getArticleID() == 0 ) {
1264                                 $wgContLang->findVariantLink($link, $nt);
1265                         }
1266
1267                         $ns = $nt->getNamespace();
1268                         $iw = $nt->getInterWiki();
1269
1270                         if ($might_be_img) { # if this is actually an invalid link
1271                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1272                                         $found = false;
1273                                         while (isset ($a[$k+1]) ) {
1274                                                 #look at the next 'line' to see if we can close it there
1275                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1276                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1277                                                 # the first ]] closes the inner link, the second the image
1278                                                         $found = true;
1279                                                         $text .= '[[' . $m[1];
1280                                                         $trail = $m[2];
1281                                                         break;
1282                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1283                                                         #if there's exactly one ]] that's fine, we'll keep looking
1284                                                         $text .= '[[' . $m[0];
1285                                                 } else {
1286                                                         #if $next_line is invalid too, we need look no further
1287                                                         $text .= '[[' . $next_line;
1288                                                         break;
1289                                                 }
1290                                         }
1291                                         if ( !$found ) {
1292                                                 # we couldn't find the end of this imageLink, so output it raw
1293                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1294                                                 $text = $this->replaceInternalLinks($text);
1295                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1296                                                 # note: no $trail, because without an end, there *is* no trail
1297                                                 continue;
1298                                         }
1299                                 } else { #it's not an image, so output it raw
1300                                         $s .= $prefix . '[[' . $link . '|' . $text;
1301                                         # note: no $trail, because without an end, there *is* no trail
1302                                         continue;
1303                                 }
1304                         }
1305
1306                         $wasblank = ( '' == $text );
1307                         if( $wasblank ) $text = $link;
1308
1309
1310                         # Link not escaped by : , create the various objects
1311                         if( $noforce ) {
1312
1313                                 # Interwikis
1314                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1315                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1316                                         $s = rtrim($s . "\n");
1317                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1318                                         continue;
1319                                 }
1320
1321                                 if ( $ns == NS_IMAGE ) {
1322                                         wfProfileIn( "$fname-image" );
1323                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1324                                                 # recursively parse links inside the image caption
1325                                                 # actually, this will parse them in any other parameters, too,
1326                                                 # but it might be hard to fix that, and it doesn't matter ATM
1327                                                 $text = $this->replaceExternalLinks($text);
1328                                                 $text = $this->replaceInternalLinks($text);
1329
1330                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1331                                                 $s .= $prefix . str_replace('http://', 'http-noparse://', $this->makeImage( $nt, $text ) ) . $trail;
1332                                                 $wgLinkCache->addImageLinkObj( $nt );
1333
1334                                                 wfProfileOut( "$fname-image" );
1335                                                 continue;
1336                                         }
1337                                         wfProfileOut( "$fname-image" );
1338
1339                                 }
1340
1341                                 if ( $ns == NS_CATEGORY ) {
1342                                         wfProfileIn( "$fname-category" );
1343                                         $t = $nt->getText();
1344                                         $s = rtrim($s . "\n"); # bug 87
1345
1346                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1347                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1348                                         $wgLinkCache->resume();
1349
1350                                         if ( $wasblank ) {
1351                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1352                                                         $sortkey = $this->mTitle->getText();
1353                                                 } else {
1354                                                         $sortkey = $this->mTitle->getPrefixedText();
1355                                                 }
1356                                         } else {
1357                                                 $sortkey = $text;
1358                                         }
1359                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1360                                         $this->mOutput->addCategoryLink( $t );
1361
1362                                         /**
1363                                          * Strip the whitespace Category links produce, see bug 87
1364                                          * @todo We might want to use trim($tmp, "\n") here.
1365                                          */
1366                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1367
1368                                         wfProfileOut( "$fname-category" );
1369                                         continue;
1370                                 }
1371                         }
1372
1373                         if( ( $nt->getPrefixedText() === $selflink ) &&
1374                             ( $nt->getFragment() === '' ) ) {
1375                                 # Self-links are handled specially; generally de-link and change to bold.
1376                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1377                                 continue;
1378                         }
1379
1380                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1381                         if( $ns == NS_MEDIA ) {
1382                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1383                                 $wgLinkCache->addImageLinkObj( $nt );
1384                                 continue;
1385                         } elseif( $ns == NS_SPECIAL ) {
1386                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1387                                 continue;
1388                         }
1389                         if ( $nt->isAlwaysKnown() ) {
1390                                 $s .= $sk->makeKnownLinkObj( $nt, $text, '', $trail, $prefix );
1391                         } else {
1392                                 /**
1393                                  * Add a link placeholder
1394                                  * Later, this will be replaced by a real link, after the existence or
1395                                  * non-existence of all the links is known
1396                                  */
1397                                 $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1398                         }
1399                 }
1400                 wfProfileOut( $fname );
1401                 return $s;
1402         }
1403
1404         /**
1405          * Make a link placeholder. The text returned can be later resolved to a real link with
1406          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1407          * parsing of interwiki links, and secondly to allow all extistence checks and
1408          * article length checks (for stub links) to be bundled into a single query.
1409          *
1410          */
1411         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1412                 if ( ! is_object($nt) ) {
1413                         # Fail gracefully
1414                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1415                 } else {
1416                         # Separate the link trail from the rest of the link
1417                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1418
1419                         if ( $nt->isExternal() ) {
1420                                 $iwRecord = array( $nt->getPrefixedDBkey(), $prefix.$text.$inside );
1421                                 $nr = array_push($this->mInterwikiLinkHolders, $iwRecord);
1422                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1423                         } else {
1424                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1425                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1426                                 $this->mLinkHolders['queries'][] = $query;
1427                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1428                                 $this->mLinkHolders['titles'][] =& $nt;
1429
1430                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1431                         }
1432                 }
1433                 return $retVal;
1434         }
1435
1436         /**
1437          * Return true if subpage links should be expanded on this page.
1438          * @return bool
1439          */
1440         function areSubpagesAllowed() {
1441                 # Some namespaces don't allow subpages
1442                 global $wgNamespacesWithSubpages;
1443                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1444         }
1445
1446         /**
1447          * Handle link to subpage if necessary
1448          * @param string $target the source of the link
1449          * @param string &$text the link text, modified as necessary
1450          * @return string the full name of the link
1451          * @access private
1452          */
1453         function maybeDoSubpageLink($target, &$text) {
1454                 # Valid link forms:
1455                 # Foobar -- normal
1456                 # :Foobar -- override special treatment of prefix (images, language links)
1457                 # /Foobar -- convert to CurrentPage/Foobar
1458                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1459                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1460                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1461
1462                 $fname = 'Parser::maybeDoSubpageLink';
1463                 wfProfileIn( $fname );
1464                 $ret = $target; # default return value is no change
1465
1466                 # Some namespaces don't allow subpages,
1467                 # so only perform processing if subpages are allowed
1468                 if( $this->areSubpagesAllowed() ) {
1469                         # Look at the first character
1470                         if( $target != '' && $target{0} == '/' ) {
1471                                 # / at end means we don't want the slash to be shown
1472                                 if( substr( $target, -1, 1 ) == '/' ) {
1473                                         $target = substr( $target, 1, -1 );
1474                                         $noslash = $target;
1475                                 } else {
1476                                         $noslash = substr( $target, 1 );
1477                                 }
1478
1479                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1480                                 if( '' === $text ) {
1481                                         $text = $target;
1482                                 } # this might be changed for ugliness reasons
1483                         } else {
1484                                 # check for .. subpage backlinks
1485                                 $dotdotcount = 0;
1486                                 $nodotdot = $target;
1487                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1488                                         ++$dotdotcount;
1489                                         $nodotdot = substr( $nodotdot, 3 );
1490                                 }
1491                                 if($dotdotcount > 0) {
1492                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1493                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1494                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1495                                                 # / at the end means don't show full path
1496                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1497                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1498                                                         if( '' === $text ) {
1499                                                                 $text = $nodotdot;
1500                                                         }
1501                                                 }
1502                                                 $nodotdot = trim( $nodotdot );
1503                                                 if( $nodotdot != '' ) {
1504                                                         $ret .= '/' . $nodotdot;
1505                                                 }
1506                                         }
1507                                 }
1508                         }
1509                 }
1510
1511                 wfProfileOut( $fname );
1512                 return $ret;
1513         }
1514
1515         /**#@+
1516          * Used by doBlockLevels()
1517          * @access private
1518          */
1519         /* private */ function closeParagraph() {
1520                 $result = '';
1521                 if ( '' != $this->mLastSection ) {
1522                         $result = '</' . $this->mLastSection  . ">\n";
1523                 }
1524                 $this->mInPre = false;
1525                 $this->mLastSection = '';
1526                 return $result;
1527         }
1528         # getCommon() returns the length of the longest common substring
1529         # of both arguments, starting at the beginning of both.
1530         #
1531         /* private */ function getCommon( $st1, $st2 ) {
1532                 $fl = strlen( $st1 );
1533                 $shorter = strlen( $st2 );
1534                 if ( $fl < $shorter ) { $shorter = $fl; }
1535
1536                 for ( $i = 0; $i < $shorter; ++$i ) {
1537                         if ( $st1{$i} != $st2{$i} ) { break; }
1538                 }
1539                 return $i;
1540         }
1541         # These next three functions open, continue, and close the list
1542         # element appropriate to the prefix character passed into them.
1543         #
1544         /* private */ function openList( $char ) {
1545                 $result = $this->closeParagraph();
1546
1547                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1548                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1549                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1550                 else if ( ';' == $char ) {
1551                         $result .= '<dl><dt>';
1552                         $this->mDTopen = true;
1553                 }
1554                 else { $result = '<!-- ERR 1 -->'; }
1555
1556                 return $result;
1557         }
1558
1559         /* private */ function nextItem( $char ) {
1560                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1561                 else if ( ':' == $char || ';' == $char ) {
1562                         $close = '</dd>';
1563                         if ( $this->mDTopen ) { $close = '</dt>'; }
1564                         if ( ';' == $char ) {
1565                                 $this->mDTopen = true;
1566                                 return $close . '<dt>';
1567                         } else {
1568                                 $this->mDTopen = false;
1569                                 return $close . '<dd>';
1570                         }
1571                 }
1572                 return '<!-- ERR 2 -->';
1573         }
1574
1575         /* private */ function closeList( $char ) {
1576                 if ( '*' == $char ) { $text = '</li></ul>'; }
1577                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1578                 else if ( ':' == $char ) {
1579                         if ( $this->mDTopen ) {
1580                                 $this->mDTopen = false;
1581                                 $text = '</dt></dl>';
1582                         } else {
1583                                 $text = '</dd></dl>';
1584                         }
1585                 }
1586                 else {  return '<!-- ERR 3 -->'; }
1587                 return $text."\n";
1588         }
1589         /**#@-*/
1590
1591         /**
1592          * Make lists from lines starting with ':', '*', '#', etc.
1593          *
1594          * @access private
1595          * @return string the lists rendered as HTML
1596          */
1597         function doBlockLevels( $text, $linestart ) {
1598                 $fname = 'Parser::doBlockLevels';
1599                 wfProfileIn( $fname );
1600
1601                 # Parsing through the text line by line.  The main thing
1602                 # happening here is handling of block-level elements p, pre,
1603                 # and making lists from lines starting with * # : etc.
1604                 #
1605                 $textLines = explode( "\n", $text );
1606
1607                 $lastPrefix = $output = '';
1608                 $this->mDTopen = $inBlockElem = false;
1609                 $prefixLength = 0;
1610                 $paragraphStack = false;
1611
1612                 if ( !$linestart ) {
1613                         $output .= array_shift( $textLines );
1614                 }
1615                 foreach ( $textLines as $oLine ) {
1616                         $lastPrefixLength = strlen( $lastPrefix );
1617                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1618                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1619                         if ( !$this->mInPre ) {
1620                                 # Multiple prefixes may abut each other for nested lists.
1621                                 $prefixLength = strspn( $oLine, '*#:;' );
1622                                 $pref = substr( $oLine, 0, $prefixLength );
1623
1624                                 # eh?
1625                                 $pref2 = str_replace( ';', ':', $pref );
1626                                 $t = substr( $oLine, $prefixLength );
1627                                 $this->mInPre = !empty($preOpenMatch);
1628                         } else {
1629                                 # Don't interpret any other prefixes in preformatted text
1630                                 $prefixLength = 0;
1631                                 $pref = $pref2 = '';
1632                                 $t = $oLine;
1633                         }
1634
1635                         # List generation
1636                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1637                                 # Same as the last item, so no need to deal with nesting or opening stuff
1638                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1639                                 $paragraphStack = false;
1640
1641                                 if ( substr( $pref, -1 ) == ';') {
1642                                         # The one nasty exception: definition lists work like this:
1643                                         # ; title : definition text
1644                                         # So we check for : in the remainder text to split up the
1645                                         # title and definition, without b0rking links.
1646                                         $term = $t2 = '';
1647                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1648                                                 $t = $t2;
1649                                                 $output .= $term . $this->nextItem( ':' );
1650                                         }
1651                                 }
1652                         } elseif( $prefixLength || $lastPrefixLength ) {
1653                                 # Either open or close a level...
1654                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1655                                 $paragraphStack = false;
1656
1657                                 while( $commonPrefixLength < $lastPrefixLength ) {
1658                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1659                                         --$lastPrefixLength;
1660                                 }
1661                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1662                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1663                                 }
1664                                 while ( $prefixLength > $commonPrefixLength ) {
1665                                         $char = substr( $pref, $commonPrefixLength, 1 );
1666                                         $output .= $this->openList( $char );
1667
1668                                         if ( ';' == $char ) {
1669                                                 # FIXME: This is dupe of code above
1670                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1671                                                         $t = $t2;
1672                                                         $output .= $term . $this->nextItem( ':' );
1673                                                 }
1674                                         }
1675                                         ++$commonPrefixLength;
1676                                 }
1677                                 $lastPrefix = $pref2;
1678                         }
1679                         if( 0 == $prefixLength ) {
1680                                 wfProfileIn( "$fname-paragraph" );
1681                                 # No prefix (not in list)--go to paragraph mode
1682                                 $uniq_prefix = UNIQ_PREFIX;
1683                                 // XXX: use a stack for nestable elements like span, table and div
1684                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1685                                 $closematch = preg_match(
1686                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1687                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1688                                 if ( $openmatch or $closematch ) {
1689                                         $paragraphStack = false;
1690                                         $output .= $this->closeParagraph();
1691                                         if($preOpenMatch and !$preCloseMatch) {
1692                                                 $this->mInPre = true;
1693                                         }
1694                                         if ( $closematch ) {
1695                                                 $inBlockElem = false;
1696                                         } else {
1697                                                 $inBlockElem = true;
1698                                         }
1699                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1700                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1701                                                 // pre
1702                                                 if ($this->mLastSection != 'pre') {
1703                                                         $paragraphStack = false;
1704                                                         $output .= $this->closeParagraph().'<pre>';
1705                                                         $this->mLastSection = 'pre';
1706                                                 }
1707                                                 $t = substr( $t, 1 );
1708                                         } else {
1709                                                 // paragraph
1710                                                 if ( '' == trim($t) ) {
1711                                                         if ( $paragraphStack ) {
1712                                                                 $output .= $paragraphStack.'<br />';
1713                                                                 $paragraphStack = false;
1714                                                                 $this->mLastSection = 'p';
1715                                                         } else {
1716                                                                 if ($this->mLastSection != 'p' ) {
1717                                                                         $output .= $this->closeParagraph();
1718                                                                         $this->mLastSection = '';
1719                                                                         $paragraphStack = '<p>';
1720                                                                 } else {
1721                                                                         $paragraphStack = '</p><p>';
1722                                                                 }
1723                                                         }
1724                                                 } else {
1725                                                         if ( $paragraphStack ) {
1726                                                                 $output .= $paragraphStack;
1727                                                                 $paragraphStack = false;
1728                                                                 $this->mLastSection = 'p';
1729                                                         } else if ($this->mLastSection != 'p') {
1730                                                                 $output .= $this->closeParagraph().'<p>';
1731                                                                 $this->mLastSection = 'p';
1732                                                         }
1733                                                 }
1734                                         }
1735                                 }
1736                                 wfProfileOut( "$fname-paragraph" );
1737                         }
1738                         if ($paragraphStack === false) {
1739                                 $output .= $t."\n";
1740                         }
1741                 }
1742                 while ( $prefixLength ) {
1743                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1744                         --$prefixLength;
1745                 }
1746                 if ( '' != $this->mLastSection ) {
1747                         $output .= '</' . $this->mLastSection . '>';
1748                         $this->mLastSection = '';
1749                 }
1750
1751                 wfProfileOut( $fname );
1752                 return $output;
1753         }
1754
1755         /**
1756          * Split up a string on ':', ignoring any occurences inside
1757          * <a>..</a> or <span>...</span>
1758          * @param string $str the string to split
1759          * @param string &$before set to everything before the ':'
1760          * @param string &$after set to everything after the ':'
1761          * return string the position of the ':', or false if none found
1762          */
1763         function findColonNoLinks($str, &$before, &$after) {
1764                 # I wonder if we should make this count all tags, not just <a>
1765                 # and <span>. That would prevent us from matching a ':' that
1766                 # comes in the middle of italics other such formatting....
1767                 # -- Wil
1768                 $fname = 'Parser::findColonNoLinks';
1769                 wfProfileIn( $fname );
1770                 $pos = 0;
1771                 do {
1772                         $colon = strpos($str, ':', $pos);
1773
1774                         if ($colon !== false) {
1775                                 $before = substr($str, 0, $colon);
1776                                 $after = substr($str, $colon + 1);
1777
1778                                 # Skip any ':' within <a> or <span> pairs
1779                                 $a = substr_count($before, '<a');
1780                                 $s = substr_count($before, '<span');
1781                                 $ca = substr_count($before, '</a>');
1782                                 $cs = substr_count($before, '</span>');
1783
1784                                 if ($a <= $ca and $s <= $cs) {
1785                                         # Tags are balanced before ':'; ok
1786                                         break;
1787                                 }
1788                                 $pos = $colon + 1;
1789                         }
1790                 } while ($colon !== false);
1791                 wfProfileOut( $fname );
1792                 return $colon;
1793         }
1794
1795         /**
1796          * Return value of a magic variable (like PAGENAME)
1797          *
1798          * @access private
1799          */
1800         function getVariableValue( $index ) {
1801                 global $wgContLang, $wgSitename, $wgServer, $wgArticle;
1802
1803                 /**
1804                  * Some of these require message or data lookups and can be
1805                  * expensive to check many times.
1806                  */
1807                 static $varCache = array();
1808                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1809
1810                 switch ( $index ) {
1811                         case MAG_CURRENTMONTH:
1812                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1813                         case MAG_CURRENTMONTHNAME:
1814                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1815                         case MAG_CURRENTMONTHNAMEGEN:
1816                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1817                         case MAG_CURRENTMONTHABBREV:
1818                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1819                         case MAG_CURRENTDAY:
1820                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1821                         case MAG_PAGENAME:
1822                                 return $this->mTitle->getText();
1823                         case MAG_PAGENAMEE:
1824                                 return $this->mTitle->getPartialURL();
1825                         case MAG_REVISIONID:
1826                                 return $wgArticle->getRevIdFetched();
1827                         case MAG_NAMESPACE:
1828                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1829                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1830                         case MAG_CURRENTDAYNAME:
1831                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1832                         case MAG_CURRENTYEAR:
1833                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1834                         case MAG_CURRENTTIME:
1835                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1836                         case MAG_CURRENTWEEK:
1837                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1838                         case MAG_CURRENTDOW:
1839                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1840                         case MAG_NUMBEROFARTICLES:
1841                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1842                         case MAG_SITENAME:
1843                                 return $wgSitename;
1844                         case MAG_SERVER:
1845                                 return $wgServer;
1846                         default:
1847                                 return NULL;
1848                 }
1849         }
1850
1851         /**
1852          * initialise the magic variables (like CURRENTMONTHNAME)
1853          *
1854          * @access private
1855          */
1856         function initialiseVariables() {
1857                 $fname = 'Parser::initialiseVariables';
1858                 wfProfileIn( $fname );
1859                 global $wgVariableIDs;
1860                 $this->mVariables = array();
1861                 foreach ( $wgVariableIDs as $id ) {
1862                         $mw =& MagicWord::get( $id );
1863                         $mw->addToArray( $this->mVariables, $id );
1864                 }
1865                 wfProfileOut( $fname );
1866         }
1867
1868         /**
1869          * Replace magic variables, templates, and template arguments
1870          * with the appropriate text. Templates are substituted recursively,
1871          * taking care to avoid infinite loops.
1872          *
1873          * Note that the substitution depends on value of $mOutputType:
1874          *  OT_WIKI: only {{subst:}} templates
1875          *  OT_MSG: only magic variables
1876          *  OT_HTML: all templates and magic variables
1877          *
1878          * @param string $tex The text to transform
1879          * @param array $args Key-value pairs representing template parameters to substitute
1880          * @access private
1881          */
1882         function replaceVariables( $text, $args = array() ) {
1883
1884                 # Prevent too big inclusions
1885                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1886                         return $text;
1887                 }
1888
1889                 $fname = 'Parser::replaceVariables';
1890                 wfProfileIn( $fname );
1891
1892                 $titleChars = Title::legalChars();
1893
1894                 # This function is called recursively. To keep track of arguments we need a stack:
1895                 array_push( $this->mArgStack, $args );
1896
1897                 # Variable substitution
1898                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1899
1900                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1901                         # Argument substitution
1902                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1903                 }
1904                 # Template substitution
1905                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1906                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1907
1908                 array_pop( $this->mArgStack );
1909
1910                 wfProfileOut( $fname );
1911                 return $text;
1912         }
1913
1914         /**
1915          * Replace magic variables
1916          * @access private
1917          */
1918         function variableSubstitution( $matches ) {
1919                 $fname = 'parser::variableSubstitution';
1920                 $varname = $matches[1];
1921                 wfProfileIn( $fname );
1922                 if ( !$this->mVariables ) {
1923                         $this->initialiseVariables();
1924                 }
1925                 $skip = false;
1926                 if ( $this->mOutputType == OT_WIKI ) {
1927                         # Do only magic variables prefixed by SUBST
1928                         $mwSubst =& MagicWord::get( MAG_SUBST );
1929                         if (!$mwSubst->matchStartAndRemove( $varname ))
1930                                 $skip = true;
1931                         # Note that if we don't substitute the variable below,
1932                         # we don't remove the {{subst:}} magic word, in case
1933                         # it is a template rather than a magic variable.
1934                 }
1935                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
1936                         $id = $this->mVariables[$varname];
1937                         $text = $this->getVariableValue( $id );
1938                         $this->mOutput->mContainsOldMagic = true;
1939                 } else {
1940                         $text = $matches[0];
1941                 }
1942                 wfProfileOut( $fname );
1943                 return $text;
1944         }
1945
1946         # Split template arguments
1947         function getTemplateArgs( $argsString ) {
1948                 if ( $argsString === '' ) {
1949                         return array();
1950                 }
1951
1952                 $args = explode( '|', substr( $argsString, 1 ) );
1953
1954                 # If any of the arguments contains a '[[' but no ']]', it needs to be
1955                 # merged with the next arg because the '|' character between belongs
1956                 # to the link syntax and not the template parameter syntax.
1957                 $argc = count($args);
1958
1959                 for ( $i = 0; $i < $argc-1; $i++ ) {
1960                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
1961                                 $args[$i] .= '|'.$args[$i+1];
1962                                 array_splice($args, $i+1, 1);
1963                                 $i--;
1964                                 $argc--;
1965                         }
1966                 }
1967
1968                 return $args;
1969         }
1970
1971         /**
1972          * Return the text of a template, after recursively
1973          * replacing any variables or templates within the template.
1974          *
1975          * @param array $matches The parts of the template
1976          *  $matches[1]: the title, i.e. the part before the |
1977          *  $matches[2]: the parameters (including a leading |), if  any
1978          * @return string the text of the template
1979          * @access private
1980          */
1981         function braceSubstitution( $matches ) {
1982                 global $wgLinkCache, $wgContLang;
1983                 $fname = 'Parser::braceSubstitution';
1984                 wfProfileIn( $fname );
1985
1986                 $found = false;
1987                 $nowiki = false;
1988                 $noparse = false;
1989
1990                 $title = NULL;
1991
1992                 # Need to know if the template comes at the start of a line,
1993                 # to treat the beginning of the template like the beginning
1994                 # of a line for tables and block-level elements.
1995                 $linestart = $matches[1];
1996
1997                 # $part1 is the bit before the first |, and must contain only title characters
1998                 # $args is a list of arguments, starting from index 0, not including $part1
1999
2000                 $part1 = $matches[2];
2001                 # If the third subpattern matched anything, it will start with |
2002
2003                 $args = $this->getTemplateArgs($matches[3]);
2004                 $argc = count( $args );
2005
2006                 # Don't parse {{{}}} because that's only for template arguments
2007                 if ( $linestart === '{' ) {
2008                         $text = $matches[0];
2009                         $found = true;
2010                         $noparse = true;
2011                 }
2012
2013                 # SUBST
2014                 if ( !$found ) {
2015                         $mwSubst =& MagicWord::get( MAG_SUBST );
2016                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2017                                 # One of two possibilities is true:
2018                                 # 1) Found SUBST but not in the PST phase
2019                                 # 2) Didn't find SUBST and in the PST phase
2020                                 # In either case, return without further processing
2021                                 $text = $matches[0];
2022                                 $found = true;
2023                                 $noparse = true;
2024                         }
2025                 }
2026
2027                 # MSG, MSGNW and INT
2028                 if ( !$found ) {
2029                         # Check for MSGNW:
2030                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2031                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2032                                 $nowiki = true;
2033                         } else {
2034                                 # Remove obsolete MSG:
2035                                 $mwMsg =& MagicWord::get( MAG_MSG );
2036                                 $mwMsg->matchStartAndRemove( $part1 );
2037                         }
2038
2039                         # Check if it is an internal message
2040                         $mwInt =& MagicWord::get( MAG_INT );
2041                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2042                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2043                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2044                                         $found = true;
2045                                 }
2046                         }
2047                 }
2048
2049                 # NS
2050                 if ( !$found ) {
2051                         # Check for NS: (namespace expansion)
2052                         $mwNs = MagicWord::get( MAG_NS );
2053                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2054                                 if ( intval( $part1 ) ) {
2055                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2056                                         $found = true;
2057                                 } else {
2058                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2059                                         if ( !is_null( $index ) ) {
2060                                                 $text = $linestart . $wgContLang->getNsText( $index );
2061                                                 $found = true;
2062                                         }
2063                                 }
2064                         }
2065                 }
2066
2067                 # LOCALURL and LOCALURLE
2068                 if ( !$found ) {
2069                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2070                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2071
2072                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2073                                 $func = 'getLocalURL';
2074                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2075                                 $func = 'escapeLocalURL';
2076                         } else {
2077                                 $func = '';
2078                         }
2079
2080                         if ( $func !== '' ) {
2081                                 $title = Title::newFromText( $part1 );
2082                                 if ( !is_null( $title ) ) {
2083                                         if ( $argc > 0 ) {
2084                                                 $text = $linestart . $title->$func( $args[0] );
2085                                         } else {
2086                                                 $text = $linestart . $title->$func();
2087                                         }
2088                                         $found = true;
2089                                 }
2090                         }
2091                 }
2092
2093                 # GRAMMAR
2094                 if ( !$found && $argc == 1 ) {
2095                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2096                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2097                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2098                                 $found = true;
2099                         }
2100                 }
2101
2102                 # Template table test
2103
2104                 # Did we encounter this template already? If yes, it is in the cache
2105                 # and we need to check for loops.
2106                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2107                         $found = true;
2108
2109                         # Infinite loop test
2110                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2111                                 $noparse = true;
2112                                 $found = true;
2113                                 $text = $linestart .
2114                                         "\{\{$part1}}" .
2115                                         '<!-- WARNING: template loop detected -->';
2116                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2117                         } else {
2118                                 # set $text to cached message.
2119                                 $text = $linestart . $this->mTemplates[$part1];
2120                         }
2121                 }
2122
2123                 # Load from database
2124                 $itcamefromthedatabase = false;
2125                 $lastPathLevel = $this->mTemplatePath;
2126                 if ( !$found ) {
2127                         $ns = NS_TEMPLATE;
2128                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2129                         if ($subpage !== '') {
2130                                 $ns = $this->mTitle->getNamespace();
2131                         }
2132                         $title = Title::newFromText( $part1, $ns );
2133                         if ( !is_null( $title ) && !$title->isExternal() ) {
2134                                 # Check for excessive inclusion
2135                                 $dbk = $title->getPrefixedDBkey();
2136                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2137                                         # This should never be reached.
2138                                         $article = new Article( $title );
2139                                         $articleContent = $article->getContentWithoutUsingSoManyDamnGlobals();
2140                                         if ( $articleContent !== false ) {
2141                                                 $found = true;
2142                                                 $text = $linestart . $articleContent;
2143                                                 $itcamefromthedatabase = true;
2144                                         }
2145                                 }
2146
2147                                 # If the title is valid but undisplayable, make a link to it
2148                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2149                                         $text = $linestart . '[['.$title->getPrefixedText().']]';
2150                                         $found = true;
2151                                 }
2152
2153                                 # Template cache array insertion
2154                                 if( $found ) {
2155                                         $this->mTemplates[$part1] = $text;
2156                                 }
2157                         }
2158                 }
2159
2160                 # Recursive parsing, escaping and link table handling
2161                 # Only for HTML output
2162                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2163                         $text = wfEscapeWikiText( $text );
2164                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2165                         # Clean up argument array
2166                         $assocArgs = array();
2167                         $index = 1;
2168                         foreach( $args as $arg ) {
2169                                 $eqpos = strpos( $arg, '=' );
2170                                 if ( $eqpos === false ) {
2171                                         $assocArgs[$index++] = $arg;
2172                                 } else {
2173                                         $name = trim( substr( $arg, 0, $eqpos ) );
2174                                         $value = trim( substr( $arg, $eqpos+1 ) );
2175                                         if ( $value === false ) {
2176                                                 $value = '';
2177                                         }
2178                                         if ( $name !== false ) {
2179                                                 $assocArgs[$name] = $value;
2180                                         }
2181                                 }
2182                         }
2183
2184                         # Add a new element to the templace recursion path
2185                         $this->mTemplatePath[$part1] = 1;
2186
2187                         if( $this->mOutputType == OT_HTML ) {
2188                                 $text = $this->strip( $text, $this->mStripState );
2189                                 $text = Sanitizer::removeHTMLtags( $text );
2190                         }
2191                         $text = $this->replaceVariables( $text, $assocArgs );
2192
2193                         # Resume the link cache and register the inclusion as a link
2194                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2195                                 $wgLinkCache->addLinkObj( $title );
2196                         }
2197
2198                         # If the template begins with a table or block-level
2199                         # element, it should be treated as beginning a new line.
2200                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2201                                 $text = "\n" . $text;
2202                         }
2203                 }
2204                 # Prune lower levels off the recursion check path
2205                 $this->mTemplatePath = $lastPathLevel;
2206
2207                 if ( !$found ) {
2208                         wfProfileOut( $fname );
2209                         return $matches[0];
2210                 } else {
2211                         # replace ==section headers==
2212                         # XXX this needs to go away once we have a better parser.
2213                         if ( $this->mOutputType != OT_WIKI && $itcamefromthedatabase ) {
2214                                 if( !is_null( $title ) )
2215                                         $encodedname = base64_encode($title->getPrefixedDBkey());
2216                                 else
2217                                         $encodedname = base64_encode("");
2218                                 $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2219                                         PREG_SPLIT_DELIM_CAPTURE);
2220                                 $text = '';
2221                                 $nsec = 0;
2222                                 for( $i = 0; $i < count($m); $i += 2 ) {
2223                                         $text .= $m[$i];
2224                                         if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2225                                         $hl = $m[$i + 1];
2226                                         if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2227                                                 $text .= $hl;
2228                                                 continue;
2229                                         }
2230                                         preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2231                                         $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2232                                                 . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2233
2234                                         $nsec++;
2235                                 }
2236                         }
2237                 }
2238                 # Prune lower levels off the recursion check path
2239                 $this->mTemplatePath = $lastPathLevel;
2240
2241                 if ( !$found ) {
2242                         wfProfileOut( $fname );
2243                         return $matches[0];
2244                 } else {
2245                         wfProfileOut( $fname );
2246                         return $text;
2247                 }
2248         }
2249
2250         /**
2251          * Triple brace replacement -- used for template arguments
2252          * @access private
2253          */
2254         function argSubstitution( $matches ) {
2255                 $arg = trim( $matches[1] );
2256                 $text = $matches[0];
2257                 $inputArgs = end( $this->mArgStack );
2258
2259                 if ( array_key_exists( $arg, $inputArgs ) ) {
2260                         $text = $inputArgs[$arg];
2261                 }
2262
2263                 return $text;
2264         }
2265
2266         /**
2267          * Returns true if the function is allowed to include this entity
2268          * @access private
2269          */
2270         function incrementIncludeCount( $dbk ) {
2271                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2272                         $this->mIncludeCount[$dbk] = 0;
2273                 }
2274                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2275                         return true;
2276                 } else {
2277                         return false;
2278                 }
2279         }
2280
2281         /**
2282          * This function accomplishes several tasks:
2283          * 1) Auto-number headings if that option is enabled
2284          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2285          * 3) Add a Table of contents on the top for users who have enabled the option
2286          * 4) Auto-anchor headings
2287          *
2288          * It loops through all headlines, collects the necessary data, then splits up the
2289          * string and re-inserts the newly formatted headlines.
2290          *
2291          * @param string $text
2292          * @param boolean $isMain
2293          * @access private
2294          */
2295         function formatHeadings( $text, $isMain=true ) {
2296                 global $wgInputEncoding, $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2297
2298                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2299                 $doShowToc = true;
2300                 $forceTocHere = false;
2301                 if( !$this->mTitle->userCanEdit() ) {
2302                         $showEditLink = 0;
2303                 } else {
2304                         $showEditLink = $this->mOptions->getEditSection();
2305                 }
2306
2307                 # Inhibit editsection links if requested in the page
2308                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2309                 if( $esw->matchAndRemove( $text ) ) {
2310                         $showEditLink = 0;
2311                 }
2312                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2313                 # do not add TOC
2314                 $mw =& MagicWord::get( MAG_NOTOC );
2315                 if( $mw->matchAndRemove( $text ) ) {
2316                         $doShowToc = false;
2317                 }
2318
2319                 # Get all headlines for numbering them and adding funky stuff like [edit]
2320                 # links - this is for later, but we need the number of headlines right now
2321                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2322
2323                 # if there are fewer than 4 headlines in the article, do not show TOC
2324                 if( $numMatches < 4 ) {
2325                         $doShowToc = false;
2326                 }
2327
2328                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2329                 # override above conditions and always show TOC at that place
2330
2331                 $mw =& MagicWord::get( MAG_TOC );
2332                 if($mw->match( $text ) ) {
2333                         $doShowToc = true;
2334                         $forceTocHere = true;
2335                 } else {
2336                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2337                         # override above conditions and always show TOC above first header
2338                         $mw =& MagicWord::get( MAG_FORCETOC );
2339                         if ($mw->matchAndRemove( $text ) ) {
2340                                 $doShowToc = true;
2341                         }
2342                 }
2343
2344                 # Never ever show TOC if no headers
2345                 if( $numMatches < 1 ) {
2346                         $doShowToc = false;
2347                 }
2348
2349                 # We need this to perform operations on the HTML
2350                 $sk =& $this->mOptions->getSkin();
2351
2352                 # headline counter
2353                 $headlineCount = 0;
2354                 $sectionCount = 0; # headlineCount excluding template sections
2355
2356                 # Ugh .. the TOC should have neat indentation levels which can be
2357                 # passed to the skin functions. These are determined here
2358                 $toc = '';
2359                 $full = '';
2360                 $head = array();
2361                 $sublevelCount = array();
2362                 $levelCount = array();
2363                 $toclevel = 0;
2364                 $level = 0;
2365                 $prevlevel = 0;
2366                 $toclevel = 0;
2367                 $prevtoclevel = 0;
2368
2369                 foreach( $matches[3] as $headline ) {
2370                         $istemplate = 0;
2371                         $templatetitle = '';
2372                         $templatesection = 0;
2373                         $numbering = '';
2374
2375                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2376                                 $istemplate = 1;
2377                                 $templatetitle = base64_decode($mat[1]);
2378                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2379                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2380                         }
2381
2382                         if( $toclevel ) {
2383                                 $prevlevel = $level;
2384                                 $prevtoclevel = $toclevel;
2385                         }
2386                         $level = $matches[1][$headlineCount];
2387
2388                         if( $doNumberHeadings || $doShowToc ) {
2389
2390                                 if ( $level > $prevlevel ) {
2391                                         # Increase TOC level
2392                                         $toclevel++;
2393                                         $sublevelCount[$toclevel] = 0;
2394                                         $toc .= $sk->tocIndent();
2395                                 }
2396                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2397                                         # Decrease TOC level, find level to jump to
2398
2399                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2400                                                 # Can only go down to level 1
2401                                                 $toclevel = 1;
2402                                         } else {
2403                                                 for ($i = $toclevel; $i > 0; $i--) {
2404                                                         if ( $levelCount[$i] == $level ) {
2405                                                                 # Found last matching level
2406                                                                 $toclevel = $i;
2407                                                                 break;
2408                                                         }
2409                                                         elseif ( $levelCount[$i] < $level ) {
2410                                                                 # Found first matching level below current level
2411                                                                 $toclevel = $i + 1;
2412                                                                 break;
2413                                                         }
2414                                                 }
2415                                         }
2416
2417                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2418                                 }
2419                                 else {
2420                                         # No change in level, end TOC line
2421                                         $toc .= $sk->tocLineEnd();
2422                                 }
2423
2424                                 $levelCount[$toclevel] = $level;
2425
2426                                 # count number of headlines for each level
2427                                 @$sublevelCount[$toclevel]++;
2428                                 $dot = 0;
2429                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2430                                         if( !empty( $sublevelCount[$i] ) ) {
2431                                                 if( $dot ) {
2432                                                         $numbering .= '.';
2433                                                 }
2434                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2435                                                 $dot = 1;
2436                                         }
2437                                 }
2438                         }
2439
2440                         # The canonized header is a version of the header text safe to use for links
2441                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2442                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2443                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2444
2445                         # Remove link placeholders by the link text.
2446                         #     <!--LINK number-->
2447                         # turns into
2448                         #     link text with suffix
2449                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2450                                                             "\$this->mLinkHolders['texts'][\$1]",
2451                                                             $canonized_headline );
2452                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2453                                                             "\$this->mInterwikiLinkHolders[\$1][1]",
2454                                                             $canonized_headline );
2455
2456                         # strip out HTML
2457                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2458                         $tocline = trim( $canonized_headline );
2459                         $canonized_headline = urlencode( do_html_entity_decode( str_replace(' ', '_', $tocline), ENT_COMPAT, $wgInputEncoding ) );
2460                         $replacearray = array(
2461                                 '%3A' => ':',
2462                                 '%' => '.'
2463                         );
2464                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2465                         $refers[$headlineCount] = $canonized_headline;
2466
2467                         # count how many in assoc. array so we can track dupes in anchors
2468                         @$refers[$canonized_headline]++;
2469                         $refcount[$headlineCount]=$refers[$canonized_headline];
2470
2471                         # Don't number the heading if it is the only one (looks silly)
2472                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2473                                 # the two are different if the line contains a link
2474                                 $headline=$numbering . ' ' . $headline;
2475                         }
2476
2477                         # Create the anchor for linking from the TOC to the section
2478                         $anchor = $canonized_headline;
2479                         if($refcount[$headlineCount] > 1 ) {
2480                                 $anchor .= '_' . $refcount[$headlineCount];
2481                         }
2482                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2483                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2484                         }
2485                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2486                                 if ( empty( $head[$headlineCount] ) ) {
2487                                         $head[$headlineCount] = '';
2488                                 }
2489                                 if( $istemplate )
2490                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2491                                 else
2492                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2493                         }
2494
2495                         # give headline the correct <h#> tag
2496                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2497
2498                         $headlineCount++;
2499                         if( !$istemplate )
2500                                 $sectionCount++;
2501                 }
2502
2503                 if( $doShowToc ) {
2504                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2505                         $toc = $sk->tocList( $toc );
2506                 }
2507
2508                 # split up and insert constructed headlines
2509
2510                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2511                 $i = 0;
2512
2513                 foreach( $blocks as $block ) {
2514                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2515                                 # This is the [edit] link that appears for the top block of text when
2516                                 # section editing is enabled
2517
2518                                 # Disabled because it broke block formatting
2519                                 # For example, a bullet point in the top line
2520                                 # $full .= $sk->editSectionLink(0);
2521                         }
2522                         $full .= $block;
2523                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2524                         # Top anchor now in skin
2525                                 $full = $full.$toc;
2526                         }
2527
2528                         if( !empty( $head[$i] ) ) {
2529                                 $full .= $head[$i];
2530                         }
2531                         $i++;
2532                 }
2533                 if($forceTocHere) {
2534                         $mw =& MagicWord::get( MAG_TOC );
2535                         return $mw->replace( $toc, $full );
2536                 } else {
2537                         return $full;
2538                 }
2539         }
2540
2541         /**
2542          * Return an HTML link for the "ISBN 123456" text
2543          * @access private
2544          */
2545         function magicISBN( $text ) {
2546                 $fname = 'Parser::magicISBN';
2547                 wfProfileIn( $fname );
2548
2549                 $a = split( 'ISBN ', ' '.$text );
2550                 if ( count ( $a ) < 2 ) {
2551                         wfProfileOut( $fname );
2552                         return $text;
2553                 }
2554                 $text = substr( array_shift( $a ), 1);
2555                 $valid = '0123456789-Xx';
2556
2557                 foreach ( $a as $x ) {
2558                         $isbn = $blank = '' ;
2559                         while ( ' ' == $x{0} ) {
2560                                 $blank .= ' ';
2561                                 $x = substr( $x, 1 );
2562                         }
2563                         if ( $x == '' ) { # blank isbn
2564                                 $text .= "ISBN $blank";
2565                                 continue;
2566                         }
2567                         while ( strstr( $valid, $x{0} ) != false ) {
2568                                 $isbn .= $x{0};
2569                                 $x = substr( $x, 1 );
2570                         }
2571                         $num = str_replace( '-', '', $isbn );
2572                         $num = str_replace( ' ', '', $num );
2573                         $num = str_replace( 'x', 'X', $num );
2574
2575                         if ( '' == $num ) {
2576                                 $text .= "ISBN $blank$x";
2577                         } else {
2578                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2579                                 $text .= '<a href="' .
2580                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2581                                         "\" class=\"internal\">ISBN $isbn</a>";
2582                                 $text .= $x;
2583                         }
2584                 }
2585                 wfProfileOut( $fname );
2586                 return $text;
2587         }
2588
2589         /**
2590          * Return an HTML link for the "RFC 1234" text
2591          *
2592          * @access private
2593          * @param string $text     Text to be processed
2594          * @param string $keyword  Magic keyword to use (default RFC)
2595          * @param string $urlmsg   Interface message to use (default rfcurl)
2596          * @return string
2597          */
2598         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2599
2600                 $valid = '0123456789';
2601                 $internal = false;
2602
2603                 $a = split( $keyword, ' '.$text );
2604                 if ( count ( $a ) < 2 ) {
2605                         return $text;
2606                 }
2607                 $text = substr( array_shift( $a ), 1);
2608
2609                 /* Check if keyword is preceed by [[.
2610                  * This test is made here cause of the array_shift above
2611                  * that prevent the test to be done in the foreach.
2612                  */
2613                 if ( substr( $text, -2 ) == '[[' ) {
2614                         $internal = true;
2615                 }
2616
2617                 foreach ( $a as $x ) {
2618                         /* token might be empty if we have RFC RFC 1234 */
2619                         if ( $x=='' ) {
2620                                 $text.=$keyword;
2621                                 continue;
2622                                 }
2623
2624                         $id = $blank = '' ;
2625
2626                         /** remove and save whitespaces in $blank */
2627                         while ( $x{0} == ' ' ) {
2628                                 $blank .= ' ';
2629                                 $x = substr( $x, 1 );
2630                         }
2631
2632                         /** remove and save the rfc number in $id */
2633                         while ( strstr( $valid, $x{0} ) != false ) {
2634                                 $id .= $x{0};
2635                                 $x = substr( $x, 1 );
2636                         }
2637
2638                         if ( $id == '' ) {
2639                                 /* call back stripped spaces*/
2640                                 $text .= $keyword.$blank.$x;
2641                         } elseif( $internal ) {
2642                                 /* normal link */
2643                                 $text .= $keyword.$id.$x;
2644                         } else {
2645                                 /* build the external link*/
2646                                 $url = wfMsg( $urlmsg, $id);
2647                                 $sk =& $this->mOptions->getSkin();
2648                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2649                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2650                         }
2651
2652                         /* Check if the next RFC keyword is preceed by [[ */
2653                         $internal = ( substr($x,-2) == '[[' );
2654                 }
2655                 return $text;
2656         }
2657
2658         /**
2659          * Transform wiki markup when saving a page by doing \r\n -> \n
2660          * conversion, substitting signatures, {{subst:}} templates, etc.
2661          *
2662          * @param string $text the text to transform
2663          * @param Title &$title the Title object for the current article
2664          * @param User &$user the User object describing the current user
2665          * @param ParserOptions $options parsing options
2666          * @param bool $clearState whether to clear the parser state first
2667          * @return string the altered wiki markup
2668          * @access public
2669          */
2670         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2671                 $this->mOptions = $options;
2672                 $this->mTitle =& $title;
2673                 $this->mOutputType = OT_WIKI;
2674
2675                 if ( $clearState ) {
2676                         $this->clearState();
2677                 }
2678
2679                 $stripState = false;
2680                 $pairs = array(
2681                         "\r\n" => "\n",
2682                 );
2683                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2684                 $text = $this->strip( $text, $stripState, false );
2685                 $text = $this->pstPass2( $text, $user );
2686                 $text = $this->unstrip( $text, $stripState );
2687                 $text = $this->unstripNoWiki( $text, $stripState );
2688                 return $text;
2689         }
2690
2691         /**
2692          * Pre-save transform helper function
2693          * @access private
2694          */
2695         function pstPass2( $text, &$user ) {
2696                 global $wgContLang, $wgLocaltimezone;
2697
2698                 # Variable replacement
2699                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2700                 $text = $this->replaceVariables( $text );
2701
2702                 # Signatures
2703                 #
2704                 $n = $user->getName();
2705                 $k = $user->getOption( 'nickname' );
2706                 if ( '' == $k ) { $k = $n; }
2707                 if ( isset( $wgLocaltimezone ) ) {
2708                         $oldtz = getenv( 'TZ' );
2709                         putenv( 'TZ='.$wgLocaltimezone );
2710                 }
2711
2712                 /* Note: This is the timestamp saved as hardcoded wikitext to
2713                  * the database, we use $wgContLang here in order to give
2714                  * everyone the same signiture and use the default one rather
2715                  * than the one selected in each users preferences.
2716                  */
2717                 $d = $wgContLang->timeanddate( wfTimestampNow(), false, false) .
2718                   ' (' . date( 'T' ) . ')';
2719                 if ( isset( $wgLocaltimezone ) ) {
2720                         putenv( 'TZ='.$oldtz );
2721                 }
2722
2723                 if( $user->getOption( 'fancysig' ) ) {
2724                         $sigText = $k;
2725                 } else {
2726                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2727                 }
2728                 $text = preg_replace( '/~~~~~/', $d, $text );
2729                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2730                 $text = preg_replace( '/~~~/', $sigText, $text );
2731
2732                 # Context links: [[|name]] and [[name (context)|]]
2733                 #
2734                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2735                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2736                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2737                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2738
2739                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2740                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2741                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2742                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2743                 $context = '';
2744                 $t = $this->mTitle->getText();
2745                 if ( preg_match( $conpat, $t, $m ) ) {
2746                         $context = $m[2];
2747                 }
2748                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2749                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2750                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2751
2752                 if ( '' == $context ) {
2753                         $text = preg_replace( $p2, '[[\\1]]', $text );
2754                 } else {
2755                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2756                 }
2757
2758                 # Trim trailing whitespace
2759                 # MAG_END (__END__) tag allows for trailing
2760                 # whitespace to be deliberately included
2761                 $text = rtrim( $text );
2762                 $mw =& MagicWord::get( MAG_END );
2763                 $mw->matchAndRemove( $text );
2764
2765                 return $text;
2766         }
2767
2768         /**
2769          * Set up some variables which are usually set up in parse()
2770          * so that an external function can call some class members with confidence
2771          * @access public
2772          */
2773         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2774                 $this->mTitle =& $title;
2775                 $this->mOptions = $options;
2776                 $this->mOutputType = $outputType;
2777                 if ( $clearState ) {
2778                         $this->clearState();
2779                 }
2780         }
2781
2782         /**
2783          * Transform a MediaWiki message by replacing magic variables.
2784          *
2785          * @param string $text the text to transform
2786          * @param ParserOptions $options  options
2787          * @return string the text with variables substituted
2788          * @access public
2789          */
2790         function transformMsg( $text, $options ) {
2791                 global $wgTitle;
2792                 static $executing = false;
2793
2794                 # Guard against infinite recursion
2795                 if ( $executing ) {
2796                         return $text;
2797                 }
2798                 $executing = true;
2799
2800                 $this->mTitle = $wgTitle;
2801                 $this->mOptions = $options;
2802                 $this->mOutputType = OT_MSG;
2803                 $this->clearState();
2804                 $text = $this->replaceVariables( $text );
2805
2806                 $executing = false;
2807                 return $text;
2808         }
2809
2810         /**
2811          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2812          * Callback will be called with the text within
2813          * Transform and return the text within
2814          * @access public
2815          */
2816         function setHook( $tag, $callback ) {
2817                 $oldVal = @$this->mTagHooks[$tag];
2818                 $this->mTagHooks[$tag] = $callback;
2819                 return $oldVal;
2820         }
2821
2822         /**
2823          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2824          * Placeholders created in Skin::makeLinkObj()
2825          * Returns an array of links found, indexed by PDBK:
2826          *  0 - broken
2827          *  1 - normal link
2828          *  2 - stub
2829          * $options is a bit field, RLH_FOR_UPDATE to select for update
2830          */
2831         function replaceLinkHolders( &$text, $options = 0 ) {
2832                 global $wgUser, $wgLinkCache;
2833                 global $wgOutputReplace;
2834
2835                 $fname = 'Parser::replaceLinkHolders';
2836                 wfProfileIn( $fname );
2837
2838                 $pdbks = array();
2839                 $colours = array();
2840                 $sk = $this->mOptions->getSkin();
2841
2842                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
2843                         wfProfileIn( $fname.'-check' );
2844                         $dbr =& wfGetDB( DB_SLAVE );
2845                         $page = $dbr->tableName( 'page' );
2846                         $threshold = $wgUser->getOption('stubthreshold');
2847
2848                         # Sort by namespace
2849                         asort( $this->mLinkHolders['namespaces'] );
2850
2851                         # Generate query
2852                         $query = false;
2853                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
2854                                 # Make title object
2855                                 $title = $this->mLinkHolders['titles'][$key];
2856
2857                                 # Skip invalid entries.
2858                                 # Result will be ugly, but prevents crash.
2859                                 if ( is_null( $title ) ) {
2860                                         continue;
2861                                 }
2862                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
2863
2864                                 # Check if it's in the link cache already
2865                                 if ( $wgLinkCache->getGoodLinkID( $pdbk ) ) {
2866                                         $colours[$pdbk] = 1;
2867                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
2868                                         $colours[$pdbk] = 0;
2869                                 } else {
2870                                         # Not in the link cache, add it to the query
2871                                         if ( !isset( $current ) ) {
2872                                                 $current = $val;
2873                                                 $query =  "SELECT page_id, page_namespace, page_title";
2874                                                 if ( $threshold > 0 ) {
2875                                                         $query .= ', page_len, page_is_redirect';
2876                                                 }
2877                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
2878                                         } elseif ( $current != $val ) {
2879                                                 $current = $val;
2880                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
2881                                         } else {
2882                                                 $query .= ', ';
2883                                         }
2884
2885                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
2886                                 }
2887                         }
2888                         if ( $query ) {
2889                                 $query .= '))';
2890                                 if ( $options & RLH_FOR_UPDATE ) {
2891                                         $query .= ' FOR UPDATE';
2892                                 }
2893
2894                                 $res = $dbr->query( $query, $fname );
2895
2896                                 # Fetch data and form into an associative array
2897                                 # non-existent = broken
2898                                 # 1 = known
2899                                 # 2 = stub
2900                                 while ( $s = $dbr->fetchObject($res) ) {
2901                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
2902                                         $pdbk = $title->getPrefixedDBkey();
2903                                         $wgLinkCache->addGoodLink( $s->page_id, $pdbk );
2904
2905                                         if ( $threshold >  0 ) {
2906                                                 $size = $s->page_len;
2907                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
2908                                                         $colours[$pdbk] = 1;
2909                                                 } else {
2910                                                         $colours[$pdbk] = 2;
2911                                                 }
2912                                         } else {
2913                                                 $colours[$pdbk] = 1;
2914                                         }
2915                                 }
2916                         }
2917                         wfProfileOut( $fname.'-check' );
2918
2919                         # Construct search and replace arrays
2920                         wfProfileIn( $fname.'-construct' );
2921                         $wgOutputReplace = array();
2922                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
2923                                 $pdbk = $pdbks[$key];
2924                                 $searchkey = "<!--LINK $key-->";
2925                                 $title = $this->mLinkHolders['titles'][$key];
2926                                 if ( empty( $colours[$pdbk] ) ) {
2927                                         $wgLinkCache->addBadLink( $pdbk );
2928                                         $colours[$pdbk] = 0;
2929                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
2930                                                                         $this->mLinkHolders['texts'][$key],
2931                                                                         $this->mLinkHolders['queries'][$key] );
2932                                 } elseif ( $colours[$pdbk] == 1 ) {
2933                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
2934                                                                         $this->mLinkHolders['texts'][$key],
2935                                                                         $this->mLinkHolders['queries'][$key] );
2936                                 } elseif ( $colours[$pdbk] == 2 ) {
2937                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
2938                                                                         $this->mLinkHolders['texts'][$key],
2939                                                                         $this->mLinkHolders['queries'][$key] );
2940                                 }
2941                         }
2942                         wfProfileOut( $fname.'-construct' );
2943
2944                         # Do the thing
2945                         wfProfileIn( $fname.'-replace' );
2946
2947                         $text = preg_replace_callback(
2948                                 '/(<!--LINK .*?-->)/',
2949                                 "wfOutputReplaceMatches",
2950                                 $text);
2951
2952                         wfProfileOut( $fname.'-replace' );
2953                 }
2954
2955                 # Now process interwiki link holders
2956                 # This is quite a bit simpler than internal links
2957                 if ( !empty( $this->mInterwikiLinkHolders ) ) {
2958                         wfProfileIn( $fname.'-interwiki' );
2959                         # Make interwiki link HTML
2960                         $wgOutputReplace = array();
2961                         foreach( $this->mInterwikiLinkHolders as $i => $lh ) {
2962                                 $s = $sk->makeLink( $lh[0], $lh[1] );
2963                                 $wgOutputReplace[] = $s;
2964                         }
2965
2966                         $text = preg_replace_callback(
2967                                 '/<!--IWLINK (.*?)-->/',
2968                                 "wfOutputReplaceMatches",
2969                                 $text );
2970                         wfProfileOut( $fname.'-interwiki' );
2971                 }
2972
2973                 wfProfileOut( $fname );
2974                 return $colours;
2975         }
2976
2977         /**
2978          * Renders an image gallery from a text with one line per image.
2979          * text labels may be given by using |-style alternative text. E.g.
2980          *   Image:one.jpg|The number "1"
2981          *   Image:tree.jpg|A tree
2982          * given as text will return the HTML of a gallery with two images,
2983          * labeled 'The number "1"' and
2984          * 'A tree'.
2985          *
2986          * @static
2987          */
2988         function renderImageGallery( $text ) {
2989                 # Setup the parser
2990                 global $wgUser, $wgParser, $wgTitle;
2991                 $parserOptions = ParserOptions::newFromUser( $wgUser );
2992
2993                 global $wgLinkCache;
2994                 $ig = new ImageGallery();
2995                 $ig->setShowBytes( false );
2996                 $ig->setShowFilename( false );
2997                 $lines = explode( "\n", $text );
2998
2999                 foreach ( $lines as $line ) {
3000                         # match lines like these:
3001                         # Image:someimage.jpg|This is some image
3002                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3003                         # Skip empty lines
3004                         if ( count( $matches ) == 0 ) {
3005                                 continue;
3006                         }
3007                         $nt = Title::newFromURL( $matches[1] );
3008                         if( is_null( $nt ) ) {
3009                                 # Bogus title. Ignore these so we don't bomb out later.
3010                                 continue;
3011                         }
3012                         if ( isset( $matches[3] ) ) {
3013                                 $label = $matches[3];
3014                         } else {
3015                                 $label = '';
3016                         }
3017
3018                         $html = $wgParser->parse( $label , $wgTitle, $parserOptions );
3019                         $html = $html->mText;
3020
3021                         $ig->add( new Image( $nt ), $html );
3022                         $wgLinkCache->addImageLinkObj( $nt );
3023                 }
3024                 return $ig->toHTML();
3025         }
3026
3027         /**
3028          * Parse image options text and use it to make an image
3029          */
3030         function makeImage( &$nt, $options ) {
3031                 global $wgContLang, $wgUseImageResize;
3032                 global $wgUser, $wgThumbLimits;
3033
3034                 $align = '';
3035
3036                 # Check if the options text is of the form "options|alt text"
3037                 # Options are:
3038                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3039                 #  * left               no resizing, just left align. label is used for alt= only
3040                 #  * right              same, but right aligned
3041                 #  * none               same, but not aligned
3042                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3043                 #  * center             center the image
3044                 #  * framed             Keep original image size, no magnify-button.
3045
3046                 $part = explode( '|', $options);
3047
3048                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3049                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3050                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3051                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3052                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3053                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3054                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3055                 $caption = '';
3056
3057                 $width = $height = $framed = $thumb = false;
3058                 $manual_thumb = "" ;
3059
3060                 foreach( $part as $key => $val ) {
3061                         $val_parts = explode ( "=" , $val , 2 ) ;
3062                         $left_part = array_shift ( $val_parts ) ;
3063                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3064                                 $thumb=true;
3065                         } elseif ( $wgUseImageResize && count ( $val_parts ) == 1 && ! is_null( $mwThumb->matchVariableStartToEnd($left_part) ) ) {
3066                                 # use manually specified thumbnail
3067                                 $thumb=true;
3068                                 $manual_thumb = array_shift ( $val_parts ) ;
3069                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3070                                 # remember to set an alignment, don't render immediately
3071                                 $align = 'right';
3072                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3073                                 # remember to set an alignment, don't render immediately
3074                                 $align = 'left';
3075                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3076                                 # remember to set an alignment, don't render immediately
3077                                 $align = 'center';
3078                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3079                                 # remember to set an alignment, don't render immediately
3080                                 $align = 'none';
3081                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3082                                 # $match is the image width in pixels
3083                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3084                                         $width = intval( $m[1] );
3085                                         $height = intval( $m[2] );
3086                                 } else {
3087                                         $width = intval($match);
3088                                 }
3089                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3090                                 $framed=true;
3091                         } else {
3092                                 $caption = $val;
3093                         }
3094                 }
3095                 # Strip bad stuff out of the alt text
3096                 $alt = $caption;
3097                 $this->replaceLinkHolders( $alt );
3098                 $alt = Sanitizer::stripAllTags( $alt );
3099
3100                 # Linker does the rest
3101                 $sk =& $this->mOptions->getSkin();
3102                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3103         }
3104 }
3105
3106 /**
3107  * @todo document
3108  * @package MediaWiki
3109  */
3110 class ParserOutput
3111 {
3112         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3113         var $mCacheTime; # Used in ParserCache
3114         var $mVersion;   # Compatibility check
3115         var $mTitleText; # title text of the chosen language variant
3116
3117         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3118                 $containsOldMagic = false, $titletext = '' )
3119         {
3120                 $this->mText = $text;
3121                 $this->mLanguageLinks = $languageLinks;
3122                 $this->mCategoryLinks = $categoryLinks;
3123                 $this->mContainsOldMagic = $containsOldMagic;
3124                 $this->mCacheTime = '';
3125                 $this->mVersion = MW_PARSER_VERSION;
3126                 $this->mTitleText = $titletext;
3127         }
3128
3129         function getText()                   { return $this->mText; }
3130         function getLanguageLinks()          { return $this->mLanguageLinks; }
3131         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3132         function getCacheTime()              { return $this->mCacheTime; }
3133         function getTitleText()              { return $this->mTitleText; }
3134         function containsOldMagic()          { return $this->mContainsOldMagic; }
3135         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3136         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3137         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3138         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3139         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3140         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3141
3142         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3143
3144         function merge( $other ) {
3145                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3146                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3147                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3148         }
3149
3150         /**
3151          * Return true if this cached output object predates the global or
3152          * per-article cache invalidation timestamps, or if it comes from
3153          * an incompatible older version.
3154          *
3155          * @param string $touched the affected article's last touched timestamp
3156          * @return bool
3157          * @access public
3158          */
3159         function expired( $touched ) {
3160                 global $wgCacheEpoch;
3161                 return $this->getCacheTime() <= $touched ||
3162                        $this->getCacheTime() <= $wgCacheEpoch ||
3163                        !isset( $this->mVersion ) ||
3164                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3165         }
3166 }
3167
3168 /**
3169  * Set options of the Parser
3170  * @todo document
3171  * @package MediaWiki
3172  */
3173 class ParserOptions
3174 {
3175         # All variables are private
3176         var $mUseTeX;                    # Use texvc to expand <math> tags
3177         var $mUseDynamicDates;           # Use DateFormatter to format dates
3178         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3179         var $mAllowExternalImages;       # Allow external images inline
3180         var $mSkin;                      # Reference to the preferred skin
3181         var $mDateFormat;                # Date format index
3182         var $mEditSection;               # Create "edit section" links
3183         var $mNumberHeadings;            # Automatically number headings
3184
3185         function getUseTeX()                        { return $this->mUseTeX; }
3186         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3187         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3188         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3189         function getSkin()                          { return $this->mSkin; }
3190         function getDateFormat()                    { return $this->mDateFormat; }
3191         function getEditSection()                   { return $this->mEditSection; }
3192         function getNumberHeadings()                { return $this->mNumberHeadings; }
3193
3194         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3195         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3196         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3197         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3198         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3199         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3200         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3201
3202         function setSkin( &$x ) { $this->mSkin =& $x; }
3203
3204         /**
3205          * Get parser options
3206          * @static
3207          */
3208         function newFromUser( &$user ) {
3209                 $popts = new ParserOptions;
3210                 $popts->initialiseFromUser( $user );
3211                 return $popts;
3212         }
3213
3214         /** Get user options */
3215         function initialiseFromUser( &$userInput ) {
3216                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
3217                 $fname = 'ParserOptions::initialiseFromUser';
3218                 wfProfileIn( $fname );
3219                 if ( !$userInput ) {
3220                         $user = new User;
3221                         $user->setLoaded( true );
3222                 } else {
3223                         $user =& $userInput;
3224                 }
3225
3226                 $this->mUseTeX = $wgUseTeX;
3227                 $this->mUseDynamicDates = $wgUseDynamicDates;
3228                 $this->mInterwikiMagic = $wgInterwikiMagic;
3229                 $this->mAllowExternalImages = $wgAllowExternalImages;
3230                 wfProfileIn( $fname.'-skin' );
3231                 $this->mSkin =& $user->getSkin();
3232                 wfProfileOut( $fname.'-skin' );
3233                 $this->mDateFormat = $user->getOption( 'date' );
3234                 $this->mEditSection = $user->getOption( 'editsection' );
3235                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3236                 wfProfileOut( $fname );
3237         }
3238 }
3239
3240 /**
3241  * Callback function used by Parser::replaceLinkHolders()
3242  * to substitute link placeholders.
3243  */
3244 function &wfOutputReplaceMatches( $matches ) {
3245         global $wgOutputReplace;
3246         return $wgOutputReplace[$matches[1]];
3247 }
3248
3249 /**
3250  * Return the total number of articles
3251  */
3252 function wfNumberOfArticles() {
3253         global $wgNumberOfArticles;
3254
3255         wfLoadSiteStats();
3256         return $wgNumberOfArticles;
3257 }
3258
3259 /**
3260  * Get various statistics from the database
3261  * @private
3262  */
3263 function wfLoadSiteStats() {
3264         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3265         $fname = 'wfLoadSiteStats';
3266
3267         if ( -1 != $wgNumberOfArticles ) return;
3268         $dbr =& wfGetDB( DB_SLAVE );
3269         $s = $dbr->selectRow( 'site_stats',
3270                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3271                 array( 'ss_row_id' => 1 ), $fname
3272         );
3273
3274         if ( $s === false ) {
3275                 return;
3276         } else {
3277                 $wgTotalViews = $s->ss_total_views;
3278                 $wgTotalEdits = $s->ss_total_edits;
3279                 $wgNumberOfArticles = $s->ss_good_articles;
3280         }
3281 }
3282
3283 /**
3284  * Escape html tags
3285  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3286  *
3287  * @param string $in Text that might contain HTML tags
3288  * @return string Escaped string
3289  */
3290 function wfEscapeHTMLTagsOnly( $in ) {
3291         return str_replace(
3292                 array( '"', '>', '<' ),
3293                 array( '&quot;', '&gt;', '&lt;' ),
3294                 $in );
3295 }
3296
3297 ?>