includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.5.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # prefix for escaping, used in two functions at least
  47 define( 'UNIQ_PREFIX', 'NaodW29');
  48
  49 # Constants needed for external link processing
  50 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  51 # Everything except bracket, space, or control characters
  52 define( 'EXT_LINK_URL_CLASS', '[^]<>"\\x00-\\x20\\x7F]' );
  53 # Including space
  54 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  55 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  56 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  57 define( 'EXT_LINK_BRACKETED',  '/\[(\b('.$wgUrlProtocols.')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  58 define( 'EXT_IMAGE_REGEX',
  59         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  60         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  61         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  62 );
  63
  64 /**
  65  * PHP Parser
  66  *
  67  * Processes wiki markup
  68  *
  69  * <pre>
  70  * There are three main entry points into the Parser class:
  71  * parse()
  72  *   produces HTML output
  73  * preSaveTransform().
  74  *   produces altered wiki markup.
  75  * transformMsg()
  76  *   performs brace substitution on MediaWiki messages
  77  *
  78  * Globals used:
  79  *    objects:   $wgLang, $wgLinkCache
  80  *
  81  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  82  *
  83  * settings:
  84  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  85  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  86  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @package MediaWiki
  92  */
  93 class Parser
  94 {
  95         /**#@+
  96          * @access private
  97          */
  98         # Persistent:
  99         var $mTagHooks;
 100
 101         # Cleared with clearState():
 102         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 103         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 104         var $mInterwikiLinkHolders, $mLinkHolders;
 105
 106         # Temporary:
 107         var $mOptions, $mTitle, $mOutputType,
 108             $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         var $mIWTransData = array();
 114
 115         /**#@-*/
 116
 117         /**
 118          * Constructor
 119          *
 120          * @access public
 121          */
 122         function Parser() {
 123                 global $wgContLang;
 124                 $this->mTemplates = array();
 125                 $this->mTemplatePath = array();
 126                 $this->mTagHooks = array();
 127                 $this->clearState();
 128         }
 129
 130         /**
 131          * Clear Parser state
 132          *
 133          * @access private
 134          */
 135         function clearState() {
 136                 $this->mOutput = new ParserOutput;
 137                 $this->mAutonumber = 0;
 138                 $this->mLastSection = '';
 139                 $this->mDTopen = false;
 140                 $this->mVariables = false;
 141                 $this->mIncludeCount = array();
 142                 $this->mStripState = array();
 143                 $this->mArgStack = array();
 144                 $this->mInPre = false;
 145                 $this->mInterwikiLinkHolders = array(
 146                         'texts' => array(),
 147                         'titles' => array()
 148                 );
 149                 $this->mLinkHolders = array(
 150                         'namespaces' => array(),
 151                         'dbkeys' => array(),
 152                         'queries' => array(),
 153                         'texts' => array(),
 154                         'titles' => array()
 155                 );
 156         }
 157
 158         /**
 159          * First pass--just handle <nowiki> sections, pass the rest off
 160          * to internalParse() which does all the real work.
 161          *
 162          * @access private
 163          * @param string $text Text we want to parse
 164          * @param Title &$title A title object
 165          * @param array $options
 166          * @param boolean $linestart
 167          * @param boolean $clearState
 168          * @return ParserOutput a ParserOutput
 169          */
 170         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 171                 global $wgUseTidy, $wgContLang;
 172                 $fname = 'Parser::parse';
 173                 wfProfileIn( $fname );
 174
 175                 if ( $clearState ) {
 176                         $this->clearState();
 177                 }
 178
 179                 $this->mOptions = $options;
 180                 $this->mTitle =& $title;
 181                 $this->mOutputType = OT_HTML;
 182
 183                 $this->mStripState = NULL;
 184
 185                 //$text = $this->strip( $text, $this->mStripState );
 186                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 187                 $x =& $this->mStripState;
 188
 189                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 190                 $text = $this->strip( $text, $x );
 191                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 192
 193                 $text = $this->internalParse( $text );
 194
 195                 $text = $this->unstrip( $text, $this->mStripState );
 196
 197                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 198                 $fixtags = array(
 199                         # french spaces, last one Guillemet-left
 200                         # only if there is something before the space
 201                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 202                         # french spaces, Guillemet-right
 203                         '/(\\302\\253) /' => '\\1&nbsp;',
 204                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 205                 );
 206                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 207
 208                 # only once and last
 209                 $text = $this->doBlockLevels( $text, $linestart );
 210
 211                 $this->replaceLinkHolders( $text );
 212
 213                 # the position of the convert() call should not be changed. it
 214                 # assumes that the links are all replaces and the only thing left
 215                 # is the <nowiki> mark.
 216                 $text = $wgContLang->convert($text);
 217                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 218
 219                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 220
 221                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 222
 223                 $text = Sanitizer::normalizeCharReferences( $text );
 224                 global $wgUseTidy;
 225                 if ($wgUseTidy) {
 226                         $text = Parser::tidy($text);
 227                 }
 228
 229                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 230
 231                 $this->mOutput->setText( $text );
 232                 wfProfileOut( $fname );
 233                 return $this->mOutput;
 234         }
 235
 236         /**
 237          * Get a random string
 238          *
 239          * @access private
 240          * @static
 241          */
 242         function getRandomString() {
 243                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 244         }
 245
 246         /**
 247          * Replaces all occurrences of <$tag>content</$tag> in the text
 248          * with a random marker and returns the new text. the output parameter
 249          * $content will be an associative array filled with data on the form
 250          * $unique_marker => content.
 251          *
 252          * If $content is already set, the additional entries will be appended
 253          * If $tag is set to STRIP_COMMENTS, the function will extract
 254          * <!-- HTML comments -->
 255          *
 256          * @access private
 257          * @static
 258          */
 259         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 260                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 261                 if ( !$content ) {
 262                         $content = array( );
 263                 }
 264                 $n = 1;
 265                 $stripped = '';
 266
 267                 if ( !$tags ) {
 268                         $tags = array( );
 269                 }
 270
 271                 if ( !$params ) {
 272                         $params = array( );
 273                 }
 274
 275                 if( $tag == STRIP_COMMENTS ) {
 276                         $start = '/<!--()/';
 277                         $end   = '/-->/';
 278                 } else {
 279                         $start = "/<$tag(\\s+[^>]*|\\s*)>/i";
 280                         $end   = "/<\\/$tag\\s*>/i";
 281                 }
 282
 283                 while ( '' != $text ) {
 284                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 285                         $stripped .= $p[0];
 286                         if( count( $p ) < 3 ) {
 287                                 break;
 288                         }
 289                         $attributes = $p[1];
 290                         $inside     = $p[2];
 291
 292                         $marker = $rnd . sprintf('%08X', $n++);
 293                         $stripped .= $marker;
 294
 295                         $tags[$marker] = "<$tag$attributes>";
 296                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 297
 298                         $q = preg_split( $end, $inside, 2 );
 299                         $content[$marker] = $q[0];
 300                         if( count( $q ) < 2 ) {
 301                                 # No end tag -- let it run out to the end of the text.
 302                                 break;
 303                         } else {
 304                                 $text = $q[1];
 305                         }
 306                 }
 307                 return $stripped;
 308         }
 309
 310         /**
 311          * Wrapper function for extractTagsAndParams
 312          * for cases where $tags and $params isn't needed
 313          * i.e. where tags will never have params, like <nowiki>
 314          *
 315          * @access private
 316          * @static
 317          */
 318         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 319                 $dummy_tags = array();
 320                 $dummy_params = array();
 321
 322                 return Parser::extractTagsAndParams( $tag, $text, $content,
 323                         $dummy_tags, $dummy_params, $uniq_prefix );
 324         }
 325
 326         /**
 327          * Strips and renders nowiki, pre, math, hiero
 328          * If $render is set, performs necessary rendering operations on plugins
 329          * Returns the text, and fills an array with data needed in unstrip()
 330          * If the $state is already a valid strip state, it adds to the state
 331          *
 332          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 333          *  will be stripped in addition to other tags. This is important
 334          *  for section editing, where these comments cause confusion when
 335          *  counting the sections in the wikisource
 336          *
 337          * @access private
 338          */
 339         function strip( $text, &$state, $stripcomments = false ) {
 340                 $render = ($this->mOutputType == OT_HTML);
 341                 $html_content = array();
 342                 $nowiki_content = array();
 343                 $math_content = array();
 344                 $pre_content = array();
 345                 $comment_content = array();
 346                 $ext_content = array();
 347                 $ext_tags = array();
 348                 $ext_params = array();
 349                 $gallery_content = array();
 350
 351                 # Replace any instances of the placeholders
 352                 $uniq_prefix = UNIQ_PREFIX;
 353                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 354
 355                 # html
 356                 global $wgRawHtml;
 357                 if( $wgRawHtml ) {
 358                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 359                         foreach( $html_content as $marker => $content ) {
 360                                 if ($render ) {
 361                                         # Raw and unchecked for validity.
 362                                         $html_content[$marker] = $content;
 363                                 } else {
 364                                         $html_content[$marker] = '<html>'.$content.'</html>';
 365                                 }
 366                         }
 367                 }
 368
 369                 # nowiki
 370                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 371                 foreach( $nowiki_content as $marker => $content ) {
 372                         if( $render ){
 373                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 374                         } else {
 375                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 376                         }
 377                 }
 378
 379                 # math
 380                 if( $this->mOptions->getUseTeX() ) {
 381                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 382                         foreach( $math_content as $marker => $content ){
 383                                 if( $render ) {
 384                                         $math_content[$marker] = renderMath( $content );
 385                                 } else {
 386                                         $math_content[$marker] = '<math>'.$content.'</math>';
 387                                 }
 388                         }
 389                 }
 390
 391                 # pre
 392                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 393                 foreach( $pre_content as $marker => $content ){
 394                         if( $render ){
 395                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 396                         } else {
 397                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 398                         }
 399                 }
 400
 401                 # gallery
 402                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 403                 foreach( $gallery_content as $marker => $content ) {
 404                         require_once( 'ImageGallery.php' );
 405                         if ( $render ) {
 406                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 407                         } else {
 408                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 409                         }
 410                 }
 411
 412                 # Comments
 413                 if($stripcomments) {
 414                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 415                         foreach( $comment_content as $marker => $content ){
 416                                 $comment_content[$marker] = '<!--'.$content.'-->';
 417                         }
 418                 }
 419
 420                 # Extensions
 421                 foreach ( $this->mTagHooks as $tag => $callback ) {
 422                         $ext_content[$tag] = array();
 423                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 424                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 425                         foreach( $ext_content[$tag] as $marker => $content ) {
 426                                 $full_tag = $ext_tags[$tag][$marker];
 427                                 $params = $ext_params[$tag][$marker];
 428                                 if ( $render ) {
 429                                         $ext_content[$tag][$marker] = $callback( $content, $params, $this );
 430                                 } else {
 431                                         $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 432                                 }
 433                         }
 434                 }
 435
 436                 # Merge state with the pre-existing state, if there is one
 437                 if ( $state ) {
 438                         $state['html'] = $state['html'] + $html_content;
 439                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 440                         $state['math'] = $state['math'] + $math_content;
 441                         $state['pre'] = $state['pre'] + $pre_content;
 442                         $state['comment'] = $state['comment'] + $comment_content;
 443                         $state['gallery'] = $state['gallery'] + $gallery_content;
 444
 445                         foreach( $ext_content as $tag => $array ) {
 446                                 if ( array_key_exists( $tag, $state ) ) {
 447                                         $state[$tag] = $state[$tag] + $array;
 448                                 }
 449                         }
 450                 } else {
 451                         $state = array(
 452                           'html' => $html_content,
 453                           'nowiki' => $nowiki_content,
 454                           'math' => $math_content,
 455                           'pre' => $pre_content,
 456                           'comment' => $comment_content,
 457                           'gallery' => $gallery_content,
 458                         ) + $ext_content;
 459                 }
 460                 return $text;
 461         }
 462
 463         /**
 464          * restores pre, math, and hiero removed by strip()
 465          *
 466          * always call unstripNoWiki() after this one
 467          * @access private
 468          */
 469         function unstrip( $text, &$state ) {
 470                 if ( !is_array( $state ) ) {
 471                         return $text;
 472                 }
 473
 474                 # Must expand in reverse order, otherwise nested tags will be corrupted
 475                 $contentDict = end( $state );
 476                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 477                         if( key($state) != 'nowiki' && key($state) != 'html') {
 478                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 479                                         $text = str_replace( key( $contentDict ), $content, $text );
 480                                 }
 481                         }
 482                 }
 483
 484                 return $text;
 485         }
 486
 487         /**
 488          * always call this after unstrip() to preserve the order
 489          *
 490          * @access private
 491          */
 492         function unstripNoWiki( $text, &$state ) {
 493                 if ( !is_array( $state ) ) {
 494                         return $text;
 495                 }
 496
 497                 # Must expand in reverse order, otherwise nested tags will be corrupted
 498                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 499                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 500                 }
 501
 502                 global $wgRawHtml;
 503                 if ($wgRawHtml) {
 504                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 505                                 $text = str_replace( key( $state['html'] ), $content, $text );
 506                         }
 507                 }
 508
 509                 return $text;
 510         }
 511
 512         /**
 513          * Add an item to the strip state
 514          * Returns the unique tag which must be inserted into the stripped text
 515          * The tag will be replaced with the original text in unstrip()
 516          *
 517          * @access private
 518          */
 519         function insertStripItem( $text, &$state ) {
 520                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 521                 if ( !$state ) {
 522                         $state = array(
 523                           'html' => array(),
 524                           'nowiki' => array(),
 525                           'math' => array(),
 526                           'pre' => array(),
 527                           'comment' => array(),
 528                           'gallery' => array(),
 529                         );
 530                 }
 531                 $state['item'][$rnd] = $text;
 532                 return $rnd;
 533         }
 534
 535         /**
 536          * Interface with html tidy, used if $wgUseTidy = true.
 537          * If tidy isn't able to correct the markup, the original will be
 538          * returned in all its glory with a warning comment appended.
 539          *
 540          * Either the external tidy program or the in-process tidy extension
 541          * will be used depending on availability. Override the default
 542          * $wgTidyInternal setting to disable the internal if it's not working.
 543          *
 544          * @param string $text Hideous HTML input
 545          * @return string Corrected HTML output
 546          * @access public
 547          * @static
 548          */
 549         function tidy( $text ) {
 550                 global $wgTidyInternal;
 551                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 552 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 553 '<head><title>test</title></head><body>'.$text.'</body></html>';
 554                 if( $wgTidyInternal ) {
 555                         $correctedtext = Parser::internalTidy( $wrappedtext );
 556                 } else {
 557                         $correctedtext = Parser::externalTidy( $wrappedtext );
 558                 }
 559                 if( is_null( $correctedtext ) ) {
 560                         wfDebug( "Tidy error detected!\n" );
 561                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 562                 }
 563                 return $correctedtext;
 564         }
 565
 566         /**
 567          * Spawn an external HTML tidy process and get corrected markup back from it.
 568          *
 569          * @access private
 570          * @static
 571          */
 572         function externalTidy( $text ) {
 573                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 574                 $fname = 'Parser::externalTidy';
 575                 wfProfileIn( $fname );
 576
 577                 $cleansource = '';
 578                 $opts = ' -utf8';
 579
 580                 $descriptorspec = array(
 581                         0 => array('pipe', 'r'),
 582                         1 => array('pipe', 'w'),
 583                         2 => array('file', '/dev/null', 'a')
 584                 );
 585                 $pipes = array();
 586                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 587                 if (is_resource($process)) {
 588                         fwrite($pipes[0], $text);
 589                         fclose($pipes[0]);
 590                         while (!feof($pipes[1])) {
 591                                 $cleansource .= fgets($pipes[1], 1024);
 592                         }
 593                         fclose($pipes[1]);
 594                         proc_close($process);
 595                 }
 596
 597                 wfProfileOut( $fname );
 598
 599                 if( $cleansource == '' && $text != '') {
 600                         // Some kind of error happened, so we couldn't get the corrected text.
 601                         // Just give up; we'll use the source text and append a warning.
 602                         return null;
 603                 } else {
 604                         return $cleansource;
 605                 }
 606         }
 607
 608         /**
 609          * Use the HTML tidy PECL extension to use the tidy library in-process,
 610          * saving the overhead of spawning a new process. Currently written to
 611          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 612          *
 613          * 'pear install tidy' should be able to compile the extension module.
 614          *
 615          * @access private
 616          * @static
 617          */
 618         function internalTidy( $text ) {
 619                 global $wgTidyConf;
 620                 $fname = 'Parser::internalTidy';
 621                 wfProfileIn( $fname );
 622
 623                 tidy_load_config( $wgTidyConf );
 624                 tidy_set_encoding( 'utf8' );
 625                 tidy_parse_string( $text );
 626                 tidy_clean_repair();
 627                 if( tidy_get_status() == 2 ) {
 628                         // 2 is magic number for fatal error
 629                         // http://www.php.net/manual/en/function.tidy-get-status.php
 630                         $cleansource = null;
 631                 } else {
 632                         $cleansource = tidy_get_output();
 633                 }
 634                 wfProfileOut( $fname );
 635                 return $cleansource;
 636         }
 637
 638         /**
 639          * parse the wiki syntax used to render tables
 640          *
 641          * @access private
 642          */
 643         function doTableStuff ( $t ) {
 644                 $fname = 'Parser::doTableStuff';
 645                 wfProfileIn( $fname );
 646
 647                 $t = explode ( "\n" , $t ) ;
 648                 $td = array () ; # Is currently a td tag open?
 649                 $ltd = array () ; # Was it TD or TH?
 650                 $tr = array () ; # Is currently a tr tag open?
 651                 $ltr = array () ; # tr attributes
 652                 $indent_level = 0; # indent level of the table
 653                 foreach ( $t AS $k => $x )
 654                 {
 655                         $x = trim ( $x ) ;
 656                         $fc = substr ( $x , 0 , 1 ) ;
 657                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 658                                 $indent_level = strlen( $matches[1] );
 659
 660                                 $attributes = $this->unstripForHTML( $matches[2] );
 661
 662                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 663                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 664                                 array_push ( $td , false ) ;
 665                                 array_push ( $ltd , '' ) ;
 666                                 array_push ( $tr , false ) ;
 667                                 array_push ( $ltr , '' ) ;
 668                         }
 669                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 670                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 671                                 $z = "</table>" . substr ( $x , 2);
 672                                 $l = array_pop ( $ltd ) ;
 673                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 674                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 675                                 array_pop ( $ltr ) ;
 676                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 677                         }
 678                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 679                                 $x = substr ( $x , 1 ) ;
 680                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 681                                 $z = '' ;
 682                                 $l = array_pop ( $ltd ) ;
 683                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 684                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 685                                 array_pop ( $ltr ) ;
 686                                 $t[$k] = $z ;
 687                                 array_push ( $tr , false ) ;
 688                                 array_push ( $td , false ) ;
 689                                 array_push ( $ltd , '' ) ;
 690                                 $attributes = $this->unstripForHTML( $x );
 691                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 692                         }
 693                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 694                                 # $x is a table row
 695                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 696                                         $fc = '+' ;
 697                                         $x = substr ( $x , 1 ) ;
 698                                 }
 699                                 $after = substr ( $x , 1 ) ;
 700                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 701                                 $after = explode ( '||' , $after ) ;
 702                                 $t[$k] = '' ;
 703
 704                                 # Loop through each table cell
 705                                 foreach ( $after AS $theline )
 706                                 {
 707                                         $z = '' ;
 708                                         if ( $fc != '+' )
 709                                         {
 710                                                 $tra = array_pop ( $ltr ) ;
 711                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 712                                                 array_push ( $tr , true ) ;
 713                                                 array_push ( $ltr , '' ) ;
 714                                         }
 715
 716                                         $l = array_pop ( $ltd ) ;
 717                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 718                                         if ( $fc == '|' ) $l = 'td' ;
 719                                         else if ( $fc == '!' ) $l = 'th' ;
 720                                         else if ( $fc == '+' ) $l = 'caption' ;
 721                                         else $l = '' ;
 722                                         array_push ( $ltd , $l ) ;
 723
 724                                         # Cell parameters
 725                                         $y = explode ( '|' , $theline , 2 ) ;
 726                                         # Note that a '|' inside an invalid link should not
 727                                         # be mistaken as delimiting cell parameters
 728                                         if ( strpos( $y[0], '[[' ) !== false ) {
 729                                                 $y = array ($theline);
 730                                         }
 731                                         if ( count ( $y ) == 1 )
 732                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 733                                         else {
 734                                                 $attributes = $this->unstripForHTML( $y[0] );
 735                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 736                                         }
 737                                         $t[$k] .= $y ;
 738                                         array_push ( $td , true ) ;
 739                                 }
 740                         }
 741                 }
 742
 743                 # Closing open td, tr && table
 744                 while ( count ( $td ) > 0 )
 745                 {
 746                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 747                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 748                         $t[] = '</table>' ;
 749                 }
 750
 751                 $t = implode ( "\n" , $t ) ;
 752                 wfProfileOut( $fname );
 753                 return $t ;
 754         }
 755
 756         /**
 757          * Helper function for parse() that transforms wiki markup into
 758          * HTML. Only called for $mOutputType == OT_HTML.
 759          *
 760          * @access private
 761          */
 762         function internalParse( $text ) {
 763                 global $wgContLang;
 764                 $args = array();
 765                 $isMain = true;
 766                 $fname = 'Parser::internalParse';
 767                 wfProfileIn( $fname );
 768
 769                 # Remove <noinclude> tags and <includeonly> sections
 770                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 771                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 772
 773                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 774                 $text = $this->replaceVariables( $text, $args );
 775
 776                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 777
 778                 $text = $this->doHeadings( $text );
 779                 if($this->mOptions->getUseDynamicDates()) {
 780                         $df =& DateFormatter::getInstance();
 781                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 782                 }
 783                 $text = $this->doAllQuotes( $text );
 784                 $text = $this->replaceInternalLinks( $text );
 785                 $text = $this->replaceExternalLinks( $text );
 786
 787                 # replaceInternalLinks may sometimes leave behind
 788                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 789                 $text = str_replace(UNIQ_PREFIX."NOPARSE", "", $text);
 790
 791                 $text = $this->doMagicLinks( $text );
 792                 $text = $this->doTableStuff( $text );
 793                 $text = $this->formatHeadings( $text, $isMain );
 794
 795                 $regex = '/<!--IW_TRANSCLUDE (\d+)-->/';
 796                 $text = preg_replace_callback($regex, array(&$this, 'scarySubstitution'), $text);
 797
 798                 wfProfileOut( $fname );
 799                 return $text;
 800         }
 801
 802         function scarySubstitution($matches) {
 803 #               return "[[".$matches[0]."]]";
 804                 return $this->mIWTransData[(int)$matches[0]];
 805         }
 806
 807         /**
 808          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 809          * magic external links.
 810          *
 811          * @access private
 812          */
 813         function &doMagicLinks( &$text ) {
 814                 $text = $this->magicISBN( $text );
 815                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 816                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 817                 return $text;
 818         }
 819
 820         /**
 821          * Parse ^^ tokens and return html
 822          *
 823          * @access private
 824          */
 825         function doExponent( $text ) {
 826                 $fname = 'Parser::doExponent';
 827                 wfProfileIn( $fname );
 828                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 829                 wfProfileOut( $fname );
 830                 return $text;
 831         }
 832
 833         /**
 834          * Parse headers and return html
 835          *
 836          * @access private
 837          */
 838         function doHeadings( $text ) {
 839                 $fname = 'Parser::doHeadings';
 840                 wfProfileIn( $fname );
 841                 for ( $i = 6; $i >= 1; --$i ) {
 842                         $h = substr( '======', 0, $i );
 843                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 844                           "<h{$i}>\\1</h{$i}>\\2", $text );
 845                 }
 846                 wfProfileOut( $fname );
 847                 return $text;
 848         }
 849
 850         /**
 851          * Replace single quotes with HTML markup
 852          * @access private
 853          * @return string the altered text
 854          */
 855         function doAllQuotes( $text ) {
 856                 $fname = 'Parser::doAllQuotes';
 857                 wfProfileIn( $fname );
 858                 $outtext = '';
 859                 $lines = explode( "\n", $text );
 860                 foreach ( $lines as $line ) {
 861                         $outtext .= $this->doQuotes ( $line ) . "\n";
 862                 }
 863                 $outtext = substr($outtext, 0,-1);
 864                 wfProfileOut( $fname );
 865                 return $outtext;
 866         }
 867
 868         /**
 869          * Helper function for doAllQuotes()
 870          * @access private
 871          */
 872         function doQuotes( $text ) {
 873                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 874                 if ( count( $arr ) == 1 )
 875                         return $text;
 876                 else
 877                 {
 878                         # First, do some preliminary work. This may shift some apostrophes from
 879                         # being mark-up to being text. It also counts the number of occurrences
 880                         # of bold and italics mark-ups.
 881                         $i = 0;
 882                         $numbold = 0;
 883                         $numitalics = 0;
 884                         foreach ( $arr as $r )
 885                         {
 886                                 if ( ( $i % 2 ) == 1 )
 887                                 {
 888                                         # If there are ever four apostrophes, assume the first is supposed to
 889                                         # be text, and the remaining three constitute mark-up for bold text.
 890                                         if ( strlen( $arr[$i] ) == 4 )
 891                                         {
 892                                                 $arr[$i-1] .= "'";
 893                                                 $arr[$i] = "'''";
 894                                         }
 895                                         # If there are more than 5 apostrophes in a row, assume they're all
 896                                         # text except for the last 5.
 897                                         else if ( strlen( $arr[$i] ) > 5 )
 898                                         {
 899                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 900                                                 $arr[$i] = "'''''";
 901                                         }
 902                                         # Count the number of occurrences of bold and italics mark-ups.
 903                                         # We are not counting sequences of five apostrophes.
 904                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 905                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 906                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 907                                 }
 908                                 $i++;
 909                         }
 910
 911                         # If there is an odd number of both bold and italics, it is likely
 912                         # that one of the bold ones was meant to be an apostrophe followed
 913                         # by italics. Which one we cannot know for certain, but it is more
 914                         # likely to be one that has a single-letter word before it.
 915                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 916                         {
 917                                 $i = 0;
 918                                 $firstsingleletterword = -1;
 919                                 $firstmultiletterword = -1;
 920                                 $firstspace = -1;
 921                                 foreach ( $arr as $r )
 922                                 {
 923                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 924                                         {
 925                                                 $x1 = substr ($arr[$i-1], -1);
 926                                                 $x2 = substr ($arr[$i-1], -2, 1);
 927                                                 if ($x1 == ' ') {
 928                                                         if ($firstspace == -1) $firstspace = $i;
 929                                                 } else if ($x2 == ' ') {
 930                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 931                                                 } else {
 932                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 933                                                 }
 934                                         }
 935                                         $i++;
 936                                 }
 937
 938                                 # If there is a single-letter word, use it!
 939                                 if ($firstsingleletterword > -1)
 940                                 {
 941                                         $arr [ $firstsingleletterword ] = "''";
 942                                         $arr [ $firstsingleletterword-1 ] .= "'";
 943                                 }
 944                                 # If not, but there's a multi-letter word, use that one.
 945                                 else if ($firstmultiletterword > -1)
 946                                 {
 947                                         $arr [ $firstmultiletterword ] = "''";
 948                                         $arr [ $firstmultiletterword-1 ] .= "'";
 949                                 }
 950                                 # ... otherwise use the first one that has neither.
 951                                 # (notice that it is possible for all three to be -1 if, for example,
 952                                 # there is only one pentuple-apostrophe in the line)
 953                                 else if ($firstspace > -1)
 954                                 {
 955                                         $arr [ $firstspace ] = "''";
 956                                         $arr [ $firstspace-1 ] .= "'";
 957                                 }
 958                         }
 959
 960                         # Now let's actually convert our apostrophic mush to HTML!
 961                         $output = '';
 962                         $buffer = '';
 963                         $state = '';
 964                         $i = 0;
 965                         foreach ($arr as $r)
 966                         {
 967                                 if (($i % 2) == 0)
 968                                 {
 969                                         if ($state == 'both')
 970                                                 $buffer .= $r;
 971                                         else
 972                                                 $output .= $r;
 973                                 }
 974                                 else
 975                                 {
 976                                         if (strlen ($r) == 2)
 977                                         {
 978                                                 if ($state == 'i')
 979                                                 { $output .= '</i>'; $state = ''; }
 980                                                 else if ($state == 'bi')
 981                                                 { $output .= '</i>'; $state = 'b'; }
 982                                                 else if ($state == 'ib')
 983                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 984                                                 else if ($state == 'both')
 985                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 986                                                 else # $state can be 'b' or ''
 987                                                 { $output .= '<i>'; $state .= 'i'; }
 988                                         }
 989                                         else if (strlen ($r) == 3)
 990                                         {
 991                                                 if ($state == 'b')
 992                                                 { $output .= '</b>'; $state = ''; }
 993                                                 else if ($state == 'bi')
 994                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 995                                                 else if ($state == 'ib')
 996                                                 { $output .= '</b>'; $state = 'i'; }
 997                                                 else if ($state == 'both')
 998                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 999                                                 else # $state can be 'i' or ''
1000                                                 { $output .= '<b>'; $state .= 'b'; }
1001                                         }
1002                                         else if (strlen ($r) == 5)
1003                                         {
1004                                                 if ($state == 'b')
1005                                                 { $output .= '</b><i>'; $state = 'i'; }
1006                                                 else if ($state == 'i')
1007                                                 { $output .= '</i><b>'; $state = 'b'; }
1008                                                 else if ($state == 'bi')
1009                                                 { $output .= '</i></b>'; $state = ''; }
1010                                                 else if ($state == 'ib')
1011                                                 { $output .= '</b></i>'; $state = ''; }
1012                                                 else if ($state == 'both')
1013                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1014                                                 else # ($state == '')
1015                                                 { $buffer = ''; $state = 'both'; }
1016                                         }
1017                                 }
1018                                 $i++;
1019                         }
1020                         # Now close all remaining tags.  Notice that the order is important.
1021                         if ($state == 'b' || $state == 'ib')
1022                                 $output .= '</b>';
1023                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1024                                 $output .= '</i>';
1025                         if ($state == 'bi')
1026                                 $output .= '</b>';
1027                         if ($state == 'both')
1028                                 $output .= '<b><i>'.$buffer.'</i></b>';
1029                         return $output;
1030                 }
1031         }
1032
1033         /**
1034          * Replace external links
1035          *
1036          * Note: this is all very hackish and the order of execution matters a lot.
1037          * Make sure to run maintenance/parserTests.php if you change this code.
1038          *
1039          * @access private
1040          */
1041         function replaceExternalLinks( $text ) {
1042                 global $wgContLang;
1043                 $fname = 'Parser::replaceExternalLinks';
1044                 wfProfileIn( $fname );
1045
1046                 $sk =& $this->mOptions->getSkin();
1047
1048                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1049
1050                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1051
1052                 $i = 0;
1053                 while ( $i<count( $bits ) ) {
1054                         $url = $bits[$i++];
1055                         $protocol = $bits[$i++];
1056                         $text = $bits[$i++];
1057                         $trail = $bits[$i++];
1058
1059                         # The characters '<' and '>' (which were escaped by
1060                         # removeHTMLtags()) should not be included in
1061                         # URLs, per RFC 2396.
1062                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1063                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1064                                 $url = substr($url, 0, $m2[0][1]);
1065                         }
1066
1067                         # If the link text is an image URL, replace it with an <img> tag
1068                         # This happened by accident in the original parser, but some people used it extensively
1069                         $img = $this->maybeMakeExternalImage( $text );
1070                         if ( $img !== false ) {
1071                                 $text = $img;
1072                         }
1073
1074                         $dtrail = '';
1075
1076                         # Set linktype for CSS - if URL==text, link is essentially free
1077                         $linktype = ($text == $url) ? 'free' : 'text';
1078
1079                         # No link text, e.g. [http://domain.tld/some.link]
1080                         if ( $text == '' ) {
1081                                 # Autonumber if allowed
1082                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1083                                         $text = '[' . ++$this->mAutonumber . ']';
1084                                         $linktype = 'autonumber';
1085                                 } else {
1086                                         # Otherwise just use the URL
1087                                         $text = htmlspecialchars( $url );
1088                                         $linktype = 'free';
1089                                 }
1090                         } else {
1091                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1092                                 # Check for trail
1093                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1094                         }
1095
1096                         $text = $wgContLang->markNoConversion($text);
1097
1098                         # Replace &amp; from obsolete syntax with &.
1099                         # All HTML entities will be escaped by makeExternalLink()
1100                         # or maybeMakeExternalImage()
1101                         $url = str_replace( '&amp;', '&', $url );
1102
1103                         # Process the trail (i.e. everything after this link up until start of the next link),
1104                         # replacing any non-bracketed links
1105                         $trail = $this->replaceFreeExternalLinks( $trail );
1106
1107
1108                         # Use the encoded URL
1109                         # This means that users can paste URLs directly into the text
1110                         # Funny characters like &ouml; aren't valid in URLs anyway
1111                         # This was changed in August 2004
1112                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1113                 }
1114
1115                 wfProfileOut( $fname );
1116                 return $s;
1117         }
1118
1119         /**
1120          * Replace anything that looks like a URL with a link
1121          * @access private
1122          */
1123         function replaceFreeExternalLinks( $text ) {
1124                 global $wgUrlProtocols;
1125                 global $wgContLang;
1126                 $fname = 'Parser::replaceFreeExternalLinks';
1127                 wfProfileIn( $fname );
1128
1129                 $bits = preg_split( '/(\b(?:'.$wgUrlProtocols.'))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1130                 $s = array_shift( $bits );
1131                 $i = 0;
1132
1133                 $sk =& $this->mOptions->getSkin();
1134
1135                 while ( $i < count( $bits ) ){
1136                         $protocol = $bits[$i++];
1137                         $remainder = $bits[$i++];
1138
1139                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1140                                 # Found some characters after the protocol that look promising
1141                                 $url = $protocol . $m[1];
1142                                 $trail = $m[2];
1143
1144                                 # The characters '<' and '>' (which were escaped by
1145                                 # removeHTMLtags()) should not be included in
1146                                 # URLs, per RFC 2396.
1147                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1148                                         $trail = substr($url, $m2[0][1]) . $trail;
1149                                         $url = substr($url, 0, $m2[0][1]);
1150                                 }
1151
1152                                 # Move trailing punctuation to $trail
1153                                 $sep = ',;\.:!?';
1154                                 # If there is no left bracket, then consider right brackets fair game too
1155                                 if ( strpos( $url, '(' ) === false ) {
1156                                         $sep .= ')';
1157                                 }
1158
1159                                 $numSepChars = strspn( strrev( $url ), $sep );
1160                                 if ( $numSepChars ) {
1161                                         $trail = substr( $url, -$numSepChars ) . $trail;
1162                                         $url = substr( $url, 0, -$numSepChars );
1163                                 }
1164
1165                                 # Replace &amp; from obsolete syntax with &.
1166                                 # All HTML entities will be escaped by makeExternalLink()
1167                                 # or maybeMakeExternalImage()
1168                                 $url = str_replace( '&amp;', '&', $url );
1169
1170                                 # Is this an external image?
1171                                 $text = $this->maybeMakeExternalImage( $url );
1172                                 if ( $text === false ) {
1173                                         # Not an image, make a link
1174                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1175                                 }
1176                                 $s .= $text . $trail;
1177                         } else {
1178                                 $s .= $protocol . $remainder;
1179                         }
1180                 }
1181                 wfProfileOut( $fname );
1182                 return $s;
1183         }
1184
1185         /**
1186          * make an image if it's allowed
1187          * @access private
1188          */
1189         function maybeMakeExternalImage( $url ) {
1190                 $sk =& $this->mOptions->getSkin();
1191                 $text = false;
1192                 if ( $this->mOptions->getAllowExternalImages() ) {
1193                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1194                                 # Image found
1195                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1196                         }
1197                 }
1198                 return $text;
1199         }
1200
1201         /**
1202          * Process [[ ]] wikilinks
1203          *
1204          * @access private
1205          */
1206         function replaceInternalLinks( $s ) {
1207                 global $wgContLang, $wgLinkCache, $wgUrlProtocols;
1208                 static $fname = 'Parser::replaceInternalLinks' ;
1209
1210                 wfProfileIn( $fname );
1211
1212                 wfProfileIn( $fname.'-setup' );
1213                 static $tc = FALSE;
1214                 # the % is needed to support urlencoded titles as well
1215                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1216
1217                 $sk =& $this->mOptions->getSkin();
1218
1219                 #split the entire text string on occurences of [[
1220                 $a = explode( '[[', ' ' . $s );
1221                 #get the first element (all text up to first [[), and remove the space we added
1222                 $s = array_shift( $a );
1223                 $s = substr( $s, 1 );
1224
1225                 # Match a link having the form [[namespace:link|alternate]]trail
1226                 static $e1 = FALSE;
1227                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1228                 # Match cases where there is no "]]", which might still be images
1229                 static $e1_img = FALSE;
1230                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1231                 # Match the end of a line for a word that's not followed by whitespace,
1232                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1233                 $e2 = wfMsgForContent( 'linkprefix' );
1234
1235                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1236
1237                 if( is_null( $this->mTitle ) ) {
1238                         wfDebugDieBacktrace( 'nooo' );
1239                 }
1240                 $nottalk = !$this->mTitle->isTalkPage();
1241
1242                 if ( $useLinkPrefixExtension ) {
1243                         if ( preg_match( $e2, $s, $m ) ) {
1244                                 $first_prefix = $m[2];
1245                         } else {
1246                                 $first_prefix = false;
1247                         }
1248                 } else {
1249                         $prefix = '';
1250                 }
1251
1252                 $selflink = $this->mTitle->getPrefixedText();
1253                 wfProfileOut( $fname.'-setup' );
1254
1255                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1256                 $useSubpages = $this->areSubpagesAllowed();
1257
1258                 # Loop for each link
1259                 for ($k = 0; isset( $a[$k] ); $k++) {
1260                         $line = $a[$k];
1261                         if ( $useLinkPrefixExtension ) {
1262                                 wfProfileIn( $fname.'-prefixhandling' );
1263                                 if ( preg_match( $e2, $s, $m ) ) {
1264                                         $prefix = $m[2];
1265                                         $s = $m[1];
1266                                 } else {
1267                                         $prefix='';
1268                                 }
1269                                 # first link
1270                                 if($first_prefix) {
1271                                         $prefix = $first_prefix;
1272                                         $first_prefix = false;
1273                                 }
1274                                 wfProfileOut( $fname.'-prefixhandling' );
1275                         }
1276
1277                         $might_be_img = false;
1278
1279                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1280                                 $text = $m[2];
1281                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1282                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1283                                 # the real problem is with the $e1 regex
1284                                 # See bug 1300.
1285                                 #
1286                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1287                                 # and no image is in sight. See bug 2095.
1288                                 #
1289                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1290                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1291                                         $m[3] = $n[1];
1292                                 }
1293                                 # fix up urlencoded title texts
1294                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1295                                 $trail = $m[3];
1296                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1297                                 $might_be_img = true;
1298                                 $text = $m[2];
1299                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1300                                 $trail = "";
1301                         } else { # Invalid form; output directly
1302                                 $s .= $prefix . '[[' . $line ;
1303                                 continue;
1304                         }
1305
1306                         # Don't allow internal links to pages containing
1307                         # PROTO: where PROTO is a valid URL protocol; these
1308                         # should be external links.
1309                         if (preg_match('/^(\b(?:'.$wgUrlProtocols.'))/', $m[1])) {
1310                                 $s .= $prefix . '[[' . $line ;
1311                                 continue;
1312                         }
1313
1314                         # Make subpage if necessary
1315                         if( $useSubpages ) {
1316                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1317                         } else {
1318                                 $link = $m[1];
1319                         }
1320
1321                         $noforce = (substr($m[1], 0, 1) != ':');
1322                         if (!$noforce) {
1323                                 # Strip off leading ':'
1324                                 $link = substr($link, 1);
1325                         }
1326
1327                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1328                         if( !$nt ) {
1329                                 $s .= $prefix . '[[' . $line;
1330                                 continue;
1331                         }
1332
1333                         #check other language variants of the link
1334                         #if the article does not exist
1335                         if( $checkVariantLink
1336                             && $nt->getArticleID() == 0 ) {
1337                                 $wgContLang->findVariantLink($link, $nt);
1338                         }
1339
1340                         $ns = $nt->getNamespace();
1341                         $iw = $nt->getInterWiki();
1342
1343                         if ($might_be_img) { # if this is actually an invalid link
1344                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1345                                         $found = false;
1346                                         while (isset ($a[$k+1]) ) {
1347                                                 #look at the next 'line' to see if we can close it there
1348                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1349                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1350                                                 # the first ]] closes the inner link, the second the image
1351                                                         $found = true;
1352                                                         $text .= '[[' . $m[1];
1353                                                         $trail = $m[2];
1354                                                         break;
1355                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1356                                                         #if there's exactly one ]] that's fine, we'll keep looking
1357                                                         $text .= '[[' . $m[0];
1358                                                 } else {
1359                                                         #if $next_line is invalid too, we need look no further
1360                                                         $text .= '[[' . $next_line;
1361                                                         break;
1362                                                 }
1363                                         }
1364                                         if ( !$found ) {
1365                                                 # we couldn't find the end of this imageLink, so output it raw
1366                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1367                                                 $text = $this->replaceInternalLinks($text);
1368                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1369                                                 # note: no $trail, because without an end, there *is* no trail
1370                                                 continue;
1371                                         }
1372                                 } else { #it's not an image, so output it raw
1373                                         $s .= $prefix . '[[' . $link . '|' . $text;
1374                                         # note: no $trail, because without an end, there *is* no trail
1375                                         continue;
1376                                 }
1377                         }
1378
1379                         $wasblank = ( '' == $text );
1380                         if( $wasblank ) $text = $link;
1381
1382
1383                         # Link not escaped by : , create the various objects
1384                         if( $noforce ) {
1385
1386                                 # Interwikis
1387                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1388                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1389                                         $s = rtrim($s . "\n");
1390                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1391                                         continue;
1392                                 }
1393
1394                                 if ( $ns == NS_IMAGE ) {
1395                                         wfProfileIn( "$fname-image" );
1396                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1397                                                 # recursively parse links inside the image caption
1398                                                 # actually, this will parse them in any other parameters, too,
1399                                                 # but it might be hard to fix that, and it doesn't matter ATM
1400                                                 $text = $this->replaceExternalLinks($text);
1401                                                 $text = $this->replaceInternalLinks($text);
1402
1403                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1404                                                 $s .= $prefix . preg_replace("/\b($wgUrlProtocols)/", UNIQ_PREFIX."NOPARSE$1", $this->makeImage( $nt, $text) ) . $trail;
1405                                                 $wgLinkCache->addImageLinkObj( $nt );
1406
1407                                                 wfProfileOut( "$fname-image" );
1408                                                 continue;
1409                                         }
1410                                         wfProfileOut( "$fname-image" );
1411
1412                                 }
1413
1414                                 if ( $ns == NS_CATEGORY ) {
1415                                         wfProfileIn( "$fname-category" );
1416                                         $t = $wgContLang->convertHtml( $nt->getText() );
1417                                         $s = rtrim($s . "\n"); # bug 87
1418
1419                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1420                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1421                                         $wgLinkCache->resume();
1422
1423                                         if ( $wasblank ) {
1424                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1425                                                         $sortkey = $this->mTitle->getText();
1426                                                 } else {
1427                                                         $sortkey = $this->mTitle->getPrefixedText();
1428                                                 }
1429                                         } else {
1430                                                 $sortkey = $text;
1431                                         }
1432                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1433                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1434                                         $this->mOutput->addCategoryLink( $t );
1435
1436                                         /**
1437                                          * Strip the whitespace Category links produce, see bug 87
1438                                          * @todo We might want to use trim($tmp, "\n") here.
1439                                          */
1440                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1441
1442                                         wfProfileOut( "$fname-category" );
1443                                         continue;
1444                                 }
1445                         }
1446
1447                         if( ( $nt->getPrefixedText() === $selflink ) &&
1448                             ( $nt->getFragment() === '' ) ) {
1449                                 # Self-links are handled specially; generally de-link and change to bold.
1450                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1451                                 continue;
1452                         }
1453
1454                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1455                         if( $ns == NS_MEDIA ) {
1456                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1457                                 $wgLinkCache->addImageLinkObj( $nt );
1458                                 continue;
1459                         } elseif( $ns == NS_SPECIAL ) {
1460                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1461                                 continue;
1462                         }
1463                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1464                 }
1465                 wfProfileOut( $fname );
1466                 return $s;
1467         }
1468
1469         /**
1470          * Make a link placeholder. The text returned can be later resolved to a real link with
1471          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1472          * parsing of interwiki links, and secondly to allow all extistence checks and
1473          * article length checks (for stub links) to be bundled into a single query.
1474          *
1475          */
1476         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1477                 if ( ! is_object($nt) ) {
1478                         # Fail gracefully
1479                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1480                 } else {
1481                         # Separate the link trail from the rest of the link
1482                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1483
1484                         if ( $nt->isExternal() ) {
1485                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1486                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1487                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1488                         } else {
1489                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1490                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1491                                 $this->mLinkHolders['queries'][] = $query;
1492                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1493                                 $this->mLinkHolders['titles'][] = $nt;
1494
1495                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1496                         }
1497                 }
1498                 return $retVal;
1499         }
1500
1501         /**
1502          * Return true if subpage links should be expanded on this page.
1503          * @return bool
1504          */
1505         function areSubpagesAllowed() {
1506                 # Some namespaces don't allow subpages
1507                 global $wgNamespacesWithSubpages;
1508                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1509         }
1510
1511         /**
1512          * Handle link to subpage if necessary
1513          * @param string $target the source of the link
1514          * @param string &$text the link text, modified as necessary
1515          * @return string the full name of the link
1516          * @access private
1517          */
1518         function maybeDoSubpageLink($target, &$text) {
1519                 # Valid link forms:
1520                 # Foobar -- normal
1521                 # :Foobar -- override special treatment of prefix (images, language links)
1522                 # /Foobar -- convert to CurrentPage/Foobar
1523                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1524                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1525                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1526
1527                 $fname = 'Parser::maybeDoSubpageLink';
1528                 wfProfileIn( $fname );
1529                 $ret = $target; # default return value is no change
1530
1531                 # Some namespaces don't allow subpages,
1532                 # so only perform processing if subpages are allowed
1533                 if( $this->areSubpagesAllowed() ) {
1534                         # Look at the first character
1535                         if( $target != '' && $target{0} == '/' ) {
1536                                 # / at end means we don't want the slash to be shown
1537                                 if( substr( $target, -1, 1 ) == '/' ) {
1538                                         $target = substr( $target, 1, -1 );
1539                                         $noslash = $target;
1540                                 } else {
1541                                         $noslash = substr( $target, 1 );
1542                                 }
1543
1544                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1545                                 if( '' === $text ) {
1546                                         $text = $target;
1547                                 } # this might be changed for ugliness reasons
1548                         } else {
1549                                 # check for .. subpage backlinks
1550                                 $dotdotcount = 0;
1551                                 $nodotdot = $target;
1552                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1553                                         ++$dotdotcount;
1554                                         $nodotdot = substr( $nodotdot, 3 );
1555                                 }
1556                                 if($dotdotcount > 0) {
1557                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1558                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1559                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1560                                                 # / at the end means don't show full path
1561                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1562                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1563                                                         if( '' === $text ) {
1564                                                                 $text = $nodotdot;
1565                                                         }
1566                                                 }
1567                                                 $nodotdot = trim( $nodotdot );
1568                                                 if( $nodotdot != '' ) {
1569                                                         $ret .= '/' . $nodotdot;
1570                                                 }
1571                                         }
1572                                 }
1573                         }
1574                 }
1575
1576                 wfProfileOut( $fname );
1577                 return $ret;
1578         }
1579
1580         /**#@+
1581          * Used by doBlockLevels()
1582          * @access private
1583          */
1584         /* private */ function closeParagraph() {
1585                 $result = '';
1586                 if ( '' != $this->mLastSection ) {
1587                         $result = '</' . $this->mLastSection  . ">\n";
1588                 }
1589                 $this->mInPre = false;
1590                 $this->mLastSection = '';
1591                 return $result;
1592         }
1593         # getCommon() returns the length of the longest common substring
1594         # of both arguments, starting at the beginning of both.
1595         #
1596         /* private */ function getCommon( $st1, $st2 ) {
1597                 $fl = strlen( $st1 );
1598                 $shorter = strlen( $st2 );
1599                 if ( $fl < $shorter ) { $shorter = $fl; }
1600
1601                 for ( $i = 0; $i < $shorter; ++$i ) {
1602                         if ( $st1{$i} != $st2{$i} ) { break; }
1603                 }
1604                 return $i;
1605         }
1606         # These next three functions open, continue, and close the list
1607         # element appropriate to the prefix character passed into them.
1608         #
1609         /* private */ function openList( $char ) {
1610                 $result = $this->closeParagraph();
1611
1612                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1613                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1614                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1615                 else if ( ';' == $char ) {
1616                         $result .= '<dl><dt>';
1617                         $this->mDTopen = true;
1618                 }
1619                 else { $result = '<!-- ERR 1 -->'; }
1620
1621                 return $result;
1622         }
1623
1624         /* private */ function nextItem( $char ) {
1625                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1626                 else if ( ':' == $char || ';' == $char ) {
1627                         $close = '</dd>';
1628                         if ( $this->mDTopen ) { $close = '</dt>'; }
1629                         if ( ';' == $char ) {
1630                                 $this->mDTopen = true;
1631                                 return $close . '<dt>';
1632                         } else {
1633                                 $this->mDTopen = false;
1634                                 return $close . '<dd>';
1635                         }
1636                 }
1637                 return '<!-- ERR 2 -->';
1638         }
1639
1640         /* private */ function closeList( $char ) {
1641                 if ( '*' == $char ) { $text = '</li></ul>'; }
1642                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1643                 else if ( ':' == $char ) {
1644                         if ( $this->mDTopen ) {
1645                                 $this->mDTopen = false;
1646                                 $text = '</dt></dl>';
1647                         } else {
1648                                 $text = '</dd></dl>';
1649                         }
1650                 }
1651                 else {  return '<!-- ERR 3 -->'; }
1652                 return $text."\n";
1653         }
1654         /**#@-*/
1655
1656         /**
1657          * Make lists from lines starting with ':', '*', '#', etc.
1658          *
1659          * @access private
1660          * @return string the lists rendered as HTML
1661          */
1662         function doBlockLevels( $text, $linestart ) {
1663                 $fname = 'Parser::doBlockLevels';
1664                 wfProfileIn( $fname );
1665
1666                 # Parsing through the text line by line.  The main thing
1667                 # happening here is handling of block-level elements p, pre,
1668                 # and making lists from lines starting with * # : etc.
1669                 #
1670                 $textLines = explode( "\n", $text );
1671
1672                 $lastPrefix = $output = '';
1673                 $this->mDTopen = $inBlockElem = false;
1674                 $prefixLength = 0;
1675                 $paragraphStack = false;
1676
1677                 if ( !$linestart ) {
1678                         $output .= array_shift( $textLines );
1679                 }
1680                 foreach ( $textLines as $oLine ) {
1681                         $lastPrefixLength = strlen( $lastPrefix );
1682                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1683                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1684                         if ( !$this->mInPre ) {
1685                                 # Multiple prefixes may abut each other for nested lists.
1686                                 $prefixLength = strspn( $oLine, '*#:;' );
1687                                 $pref = substr( $oLine, 0, $prefixLength );
1688
1689                                 # eh?
1690                                 $pref2 = str_replace( ';', ':', $pref );
1691                                 $t = substr( $oLine, $prefixLength );
1692                                 $this->mInPre = !empty($preOpenMatch);
1693                         } else {
1694                                 # Don't interpret any other prefixes in preformatted text
1695                                 $prefixLength = 0;
1696                                 $pref = $pref2 = '';
1697                                 $t = $oLine;
1698                         }
1699
1700                         # List generation
1701                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1702                                 # Same as the last item, so no need to deal with nesting or opening stuff
1703                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1704                                 $paragraphStack = false;
1705
1706                                 if ( substr( $pref, -1 ) == ';') {
1707                                         # The one nasty exception: definition lists work like this:
1708                                         # ; title : definition text
1709                                         # So we check for : in the remainder text to split up the
1710                                         # title and definition, without b0rking links.
1711                                         $term = $t2 = '';
1712                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1713                                                 $t = $t2;
1714                                                 $output .= $term . $this->nextItem( ':' );
1715                                         }
1716                                 }
1717                         } elseif( $prefixLength || $lastPrefixLength ) {
1718                                 # Either open or close a level...
1719                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1720                                 $paragraphStack = false;
1721
1722                                 while( $commonPrefixLength < $lastPrefixLength ) {
1723                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1724                                         --$lastPrefixLength;
1725                                 }
1726                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1727                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1728                                 }
1729                                 while ( $prefixLength > $commonPrefixLength ) {
1730                                         $char = substr( $pref, $commonPrefixLength, 1 );
1731                                         $output .= $this->openList( $char );
1732
1733                                         if ( ';' == $char ) {
1734                                                 # FIXME: This is dupe of code above
1735                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1736                                                         $t = $t2;
1737                                                         $output .= $term . $this->nextItem( ':' );
1738                                                 }
1739                                         }
1740                                         ++$commonPrefixLength;
1741                                 }
1742                                 $lastPrefix = $pref2;
1743                         }
1744                         if( 0 == $prefixLength ) {
1745                                 wfProfileIn( "$fname-paragraph" );
1746                                 # No prefix (not in list)--go to paragraph mode
1747                                 $uniq_prefix = UNIQ_PREFIX;
1748                                 // XXX: use a stack for nestable elements like span, table and div
1749                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1750                                 $closematch = preg_match(
1751                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1752                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1753                                 if ( $openmatch or $closematch ) {
1754                                         $paragraphStack = false;
1755                                         $output .= $this->closeParagraph();
1756                                         if ( $preOpenMatch and !$preCloseMatch ) {
1757                                                 $this->mInPre = true;
1758                                         }
1759                                         if ( $closematch ) {
1760                                                 $inBlockElem = false;
1761                                         } else {
1762                                                 $inBlockElem = true;
1763                                         }
1764                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1765                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1766                                                 // pre
1767                                                 if ($this->mLastSection != 'pre') {
1768                                                         $paragraphStack = false;
1769                                                         $output .= $this->closeParagraph().'<pre>';
1770                                                         $this->mLastSection = 'pre';
1771                                                 }
1772                                                 $t = substr( $t, 1 );
1773                                         } else {
1774                                                 // paragraph
1775                                                 if ( '' == trim($t) ) {
1776                                                         if ( $paragraphStack ) {
1777                                                                 $output .= $paragraphStack.'<br />';
1778                                                                 $paragraphStack = false;
1779                                                                 $this->mLastSection = 'p';
1780                                                         } else {
1781                                                                 if ($this->mLastSection != 'p' ) {
1782                                                                         $output .= $this->closeParagraph();
1783                                                                         $this->mLastSection = '';
1784                                                                         $paragraphStack = '<p>';
1785                                                                 } else {
1786                                                                         $paragraphStack = '</p><p>';
1787                                                                 }
1788                                                         }
1789                                                 } else {
1790                                                         if ( $paragraphStack ) {
1791                                                                 $output .= $paragraphStack;
1792                                                                 $paragraphStack = false;
1793                                                                 $this->mLastSection = 'p';
1794                                                         } else if ($this->mLastSection != 'p') {
1795                                                                 $output .= $this->closeParagraph().'<p>';
1796                                                                 $this->mLastSection = 'p';
1797                                                         }
1798                                                 }
1799                                         }
1800                                 }
1801                                 wfProfileOut( "$fname-paragraph" );
1802                         }
1803                         // somewhere above we forget to get out of pre block (bug 785)
1804                         if($preCloseMatch && $this->mInPre) {
1805                                 $this->mInPre = false;
1806                         }
1807                         if ($paragraphStack === false) {
1808                                 $output .= $t."\n";
1809                         }
1810                 }
1811                 while ( $prefixLength ) {
1812                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1813                         --$prefixLength;
1814                 }
1815                 if ( '' != $this->mLastSection ) {
1816                         $output .= '</' . $this->mLastSection . '>';
1817                         $this->mLastSection = '';
1818                 }
1819
1820                 wfProfileOut( $fname );
1821                 return $output;
1822         }
1823
1824         /**
1825          * Split up a string on ':', ignoring any occurences inside
1826          * <a>..</a> or <span>...</span>
1827          * @param string $str the string to split
1828          * @param string &$before set to everything before the ':'
1829          * @param string &$after set to everything after the ':'
1830          * return string the position of the ':', or false if none found
1831          */
1832         function findColonNoLinks($str, &$before, &$after) {
1833                 # I wonder if we should make this count all tags, not just <a>
1834                 # and <span>. That would prevent us from matching a ':' that
1835                 # comes in the middle of italics other such formatting....
1836                 # -- Wil
1837                 $fname = 'Parser::findColonNoLinks';
1838                 wfProfileIn( $fname );
1839                 $pos = 0;
1840                 do {
1841                         $colon = strpos($str, ':', $pos);
1842
1843                         if ($colon !== false) {
1844                                 $before = substr($str, 0, $colon);
1845                                 $after = substr($str, $colon + 1);
1846
1847                                 # Skip any ':' within <a> or <span> pairs
1848                                 $a = substr_count($before, '<a');
1849                                 $s = substr_count($before, '<span');
1850                                 $ca = substr_count($before, '</a>');
1851                                 $cs = substr_count($before, '</span>');
1852
1853                                 if ($a <= $ca and $s <= $cs) {
1854                                         # Tags are balanced before ':'; ok
1855                                         break;
1856                                 }
1857                                 $pos = $colon + 1;
1858                         }
1859                 } while ($colon !== false);
1860                 wfProfileOut( $fname );
1861                 return $colon;
1862         }
1863
1864         /**
1865          * Return value of a magic variable (like PAGENAME)
1866          *
1867          * @access private
1868          */
1869         function getVariableValue( $index ) {
1870                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgArticle, $wgScriptPath;
1871
1872                 /**
1873                  * Some of these require message or data lookups and can be
1874                  * expensive to check many times.
1875                  */
1876                 static $varCache = array();
1877                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1878
1879                 switch ( $index ) {
1880                         case MAG_CURRENTMONTH:
1881                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1882                         case MAG_CURRENTMONTHNAME:
1883                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1884                         case MAG_CURRENTMONTHNAMEGEN:
1885                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1886                         case MAG_CURRENTMONTHABBREV:
1887                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1888                         case MAG_CURRENTDAY:
1889                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1890                         case MAG_PAGENAME:
1891                                 return $this->mTitle->getText();
1892                         case MAG_PAGENAMEE:
1893                                 return $this->mTitle->getPartialURL();
1894                         case MAG_REVISIONID:
1895                                 return $wgArticle->getRevIdFetched();
1896                         case MAG_NAMESPACE:
1897                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1898                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1899                         case MAG_CURRENTDAYNAME:
1900                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1901                         case MAG_CURRENTYEAR:
1902                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1903                         case MAG_CURRENTTIME:
1904                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1905                         case MAG_CURRENTWEEK:
1906                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1907                         case MAG_CURRENTDOW:
1908                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1909                         case MAG_NUMBEROFARTICLES:
1910                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1911                         case MAG_NUMBEROFFILES:
1912                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1913                         case MAG_SITENAME:
1914                                 return $wgSitename;
1915                         case MAG_SERVER:
1916                                 return $wgServer;
1917                         case MAG_SERVERNAME:
1918                                 return $wgServerName;
1919                         case MAG_SCRIPTPATH:
1920                                 return $wgScriptPath;
1921                         default:
1922                                 return NULL;
1923                 }
1924         }
1925
1926         /**
1927          * initialise the magic variables (like CURRENTMONTHNAME)
1928          *
1929          * @access private
1930          */
1931         function initialiseVariables() {
1932                 $fname = 'Parser::initialiseVariables';
1933                 wfProfileIn( $fname );
1934                 global $wgVariableIDs;
1935                 $this->mVariables = array();
1936                 foreach ( $wgVariableIDs as $id ) {
1937                         $mw =& MagicWord::get( $id );
1938                         $mw->addToArray( $this->mVariables, $id );
1939                 }
1940                 wfProfileOut( $fname );
1941         }
1942
1943         /**
1944          * Replace magic variables, templates, and template arguments
1945          * with the appropriate text. Templates are substituted recursively,
1946          * taking care to avoid infinite loops.
1947          *
1948          * Note that the substitution depends on value of $mOutputType:
1949          *  OT_WIKI: only {{subst:}} templates
1950          *  OT_MSG: only magic variables
1951          *  OT_HTML: all templates and magic variables
1952          *
1953          * @param string $tex The text to transform
1954          * @param array $args Key-value pairs representing template parameters to substitute
1955          * @access private
1956          */
1957         function replaceVariables( $text, $args = array() ) {
1958
1959                 # Prevent too big inclusions
1960                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1961                         return $text;
1962                 }
1963
1964                 $fname = 'Parser::replaceVariables';
1965                 wfProfileIn( $fname );
1966
1967                 $titleChars = Title::legalChars();
1968
1969                 # This function is called recursively. To keep track of arguments we need a stack:
1970                 array_push( $this->mArgStack, $args );
1971
1972                 # Variable substitution
1973                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1974
1975                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1976                         # Argument substitution
1977                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1978                 }
1979                 # Template substitution
1980                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1981                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1982
1983                 array_pop( $this->mArgStack );
1984
1985                 wfProfileOut( $fname );
1986                 return $text;
1987         }
1988
1989         /**
1990          * Replace magic variables
1991          * @access private
1992          */
1993         function variableSubstitution( $matches ) {
1994                 $fname = 'parser::variableSubstitution';
1995                 $varname = $matches[1];
1996                 wfProfileIn( $fname );
1997                 if ( !$this->mVariables ) {
1998                         $this->initialiseVariables();
1999                 }
2000                 $skip = false;
2001                 if ( $this->mOutputType == OT_WIKI ) {
2002                         # Do only magic variables prefixed by SUBST
2003                         $mwSubst =& MagicWord::get( MAG_SUBST );
2004                         if (!$mwSubst->matchStartAndRemove( $varname ))
2005                                 $skip = true;
2006                         # Note that if we don't substitute the variable below,
2007                         # we don't remove the {{subst:}} magic word, in case
2008                         # it is a template rather than a magic variable.
2009                 }
2010                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2011                         $id = $this->mVariables[$varname];
2012                         $text = $this->getVariableValue( $id );
2013                         $this->mOutput->mContainsOldMagic = true;
2014                 } else {
2015                         $text = $matches[0];
2016                 }
2017                 wfProfileOut( $fname );
2018                 return $text;
2019         }
2020
2021         # Split template arguments
2022         function getTemplateArgs( $argsString ) {
2023                 if ( $argsString === '' ) {
2024                         return array();
2025                 }
2026
2027                 $args = explode( '|', substr( $argsString, 1 ) );
2028
2029                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2030                 # merged with the next arg because the '|' character between belongs
2031                 # to the link syntax and not the template parameter syntax.
2032                 $argc = count($args);
2033
2034                 for ( $i = 0; $i < $argc-1; $i++ ) {
2035                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2036                                 $args[$i] .= '|'.$args[$i+1];
2037                                 array_splice($args, $i+1, 1);
2038                                 $i--;
2039                                 $argc--;
2040                         }
2041                 }
2042
2043                 return $args;
2044         }
2045
2046         /**
2047          * Return the text of a template, after recursively
2048          * replacing any variables or templates within the template.
2049          *
2050          * @param array $matches The parts of the template
2051          *  $matches[1]: the title, i.e. the part before the |
2052          *  $matches[2]: the parameters (including a leading |), if  any
2053          * @return string the text of the template
2054          * @access private
2055          */
2056         function braceSubstitution( $matches ) {
2057                 global $wgLinkCache, $wgContLang;
2058                 $fname = 'Parser::braceSubstitution';
2059                 wfProfileIn( $fname );
2060
2061                 $found = false;
2062                 $nowiki = false;
2063                 $noparse = false;
2064
2065                 $title = NULL;
2066
2067                 # Need to know if the template comes at the start of a line,
2068                 # to treat the beginning of the template like the beginning
2069                 # of a line for tables and block-level elements.
2070                 $linestart = $matches[1];
2071
2072                 # $part1 is the bit before the first |, and must contain only title characters
2073                 # $args is a list of arguments, starting from index 0, not including $part1
2074
2075                 $part1 = $matches[2];
2076                 # If the third subpattern matched anything, it will start with |
2077
2078                 $args = $this->getTemplateArgs($matches[3]);
2079                 $argc = count( $args );
2080
2081                 # Don't parse {{{}}} because that's only for template arguments
2082                 if ( $linestart === '{' ) {
2083                         $text = $matches[0];
2084                         $found = true;
2085                         $noparse = true;
2086                 }
2087
2088                 # SUBST
2089                 if ( !$found ) {
2090                         $mwSubst =& MagicWord::get( MAG_SUBST );
2091                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2092                                 # One of two possibilities is true:
2093                                 # 1) Found SUBST but not in the PST phase
2094                                 # 2) Didn't find SUBST and in the PST phase
2095                                 # In either case, return without further processing
2096                                 $text = $matches[0];
2097                                 $found = true;
2098                                 $noparse = true;
2099                         }
2100                 }
2101
2102                 # MSG, MSGNW and INT
2103                 if ( !$found ) {
2104                         # Check for MSGNW:
2105                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2106                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2107                                 $nowiki = true;
2108                         } else {
2109                                 # Remove obsolete MSG:
2110                                 $mwMsg =& MagicWord::get( MAG_MSG );
2111                                 $mwMsg->matchStartAndRemove( $part1 );
2112                         }
2113
2114                         # Check if it is an internal message
2115                         $mwInt =& MagicWord::get( MAG_INT );
2116                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2117                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2118                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2119                                         $found = true;
2120                                 }
2121                         }
2122                 }
2123
2124                 # NS
2125                 if ( !$found ) {
2126                         # Check for NS: (namespace expansion)
2127                         $mwNs = MagicWord::get( MAG_NS );
2128                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2129                                 if ( intval( $part1 ) ) {
2130                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2131                                         $found = true;
2132                                 } else {
2133                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2134                                         if ( !is_null( $index ) ) {
2135                                                 $text = $linestart . $wgContLang->getNsText( $index );
2136                                                 $found = true;
2137                                         }
2138                                 }
2139                         }
2140                 }
2141
2142                 # LOCALURL and LOCALURLE
2143                 if ( !$found ) {
2144                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2145                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2146
2147                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2148                                 $func = 'getLocalURL';
2149                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2150                                 $func = 'escapeLocalURL';
2151                         } else {
2152                                 $func = '';
2153                         }
2154
2155                         if ( $func !== '' ) {
2156                                 $title = Title::newFromText( $part1 );
2157                                 if ( !is_null( $title ) ) {
2158                                         if ( $argc > 0 ) {
2159                                                 $text = $linestart . $title->$func( $args[0] );
2160                                         } else {
2161                                                 $text = $linestart . $title->$func();
2162                                         }
2163                                         $found = true;
2164                                 }
2165                         }
2166                 }
2167
2168                 # GRAMMAR
2169                 if ( !$found && $argc == 1 ) {
2170                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2171                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2172                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2173                                 $found = true;
2174                         }
2175                 }
2176
2177                 # PLURAL
2178                 if ( !$found && $argc >= 2 ) {
2179                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2180                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2181                                 if ($argc==2) {$args[2]=$args[1];}
2182                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2183                                 $found = true;
2184                         }
2185                 }
2186
2187                 # Template table test
2188
2189                 # Did we encounter this template already? If yes, it is in the cache
2190                 # and we need to check for loops.
2191                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2192                         $found = true;
2193
2194                         # Infinite loop test
2195                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2196                                 $noparse = true;
2197                                 $found = true;
2198                                 $text = $linestart .
2199                                         "\{\{$part1}}" .
2200                                         '<!-- WARNING: template loop detected -->';
2201                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2202                         } else {
2203                                 # set $text to cached message.
2204                                 $text = $linestart . $this->mTemplates[$part1];
2205                         }
2206                 }
2207
2208                 # Load from database
2209                 $replaceHeadings = false;
2210                 $isHTML = false;
2211                 $lastPathLevel = $this->mTemplatePath;
2212                 if ( !$found ) {
2213                         $ns = NS_TEMPLATE;
2214                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2215                         if ($subpage !== '') {
2216                                 $ns = $this->mTitle->getNamespace();
2217                         }
2218                         $title = Title::newFromText( $part1, $ns );
2219
2220                         if ($title) {
2221                                 $interwiki = Title::getInterwikiLink($title->getInterwiki());
2222                                 if ($interwiki != '' && $title->isTrans()) {
2223                                         return $this->scarytransclude($title, $interwiki);
2224                                 }
2225                         }
2226
2227                         if ( !is_null( $title ) && !$title->isExternal() ) {
2228                                 # Check for excessive inclusion
2229                                 $dbk = $title->getPrefixedDBkey();
2230                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2231                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2232                                                 # Capture special page output
2233                                                 $text = SpecialPage::capturePath( $title );
2234                                                 if ( is_string( $text ) ) {
2235                                                         $found = true;
2236                                                         $noparse = true;
2237                                                         $isHTML = true;
2238                                                         $this->disableCache();
2239                                                 }
2240                                         } else {
2241                                                 $article = new Article( $title );
2242                                                 $articleContent = $article->fetchContent(0, false);
2243                                                 if ( $articleContent !== false ) {
2244                                                         $found = true;
2245                                                         $text = $articleContent;
2246                                                         $replaceHeadings = true;
2247                                                 }
2248                                         }
2249                                 }
2250
2251                                 # If the title is valid but undisplayable, make a link to it
2252                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2253                                         $text = '[['.$title->getPrefixedText().']]';
2254                                         $found = true;
2255                                 }
2256
2257                                 # Template cache array insertion
2258                                 if( $found ) {
2259                                         $this->mTemplates[$part1] = $text;
2260                                         $text = $linestart . $text;
2261                                 }
2262                         }
2263                 }
2264
2265                 # Recursive parsing, escaping and link table handling
2266                 # Only for HTML output
2267                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2268                         $text = wfEscapeWikiText( $text );
2269                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2270                         # Clean up argument array
2271                         $assocArgs = array();
2272                         $index = 1;
2273                         foreach( $args as $arg ) {
2274                                 $eqpos = strpos( $arg, '=' );
2275                                 if ( $eqpos === false ) {
2276                                         $assocArgs[$index++] = $arg;
2277                                 } else {
2278                                         $name = trim( substr( $arg, 0, $eqpos ) );
2279                                         $value = trim( substr( $arg, $eqpos+1 ) );
2280                                         if ( $value === false ) {
2281                                                 $value = '';
2282                                         }
2283                                         if ( $name !== false ) {
2284                                                 $assocArgs[$name] = $value;
2285                                         }
2286                                 }
2287                         }
2288
2289                         # Add a new element to the templace recursion path
2290                         $this->mTemplatePath[$part1] = 1;
2291
2292                         if( $this->mOutputType == OT_HTML ) {
2293                                 # Remove <noinclude> sections and <includeonly> tags
2294                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2295                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2296                                 # Strip <nowiki>, <pre>, etc.
2297                                 $text = $this->strip( $text, $this->mStripState );
2298                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2299                         }
2300                         $text = $this->replaceVariables( $text, $assocArgs );
2301
2302                         # Resume the link cache and register the inclusion as a link
2303                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2304                                 $wgLinkCache->addLinkObj( $title );
2305                         }
2306
2307                         # If the template begins with a table or block-level
2308                         # element, it should be treated as beginning a new line.
2309                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2310                                 $text = "\n" . $text;
2311                         }
2312                 }
2313                 # Prune lower levels off the recursion check path
2314                 $this->mTemplatePath = $lastPathLevel;
2315
2316                 if ( !$found ) {
2317                         wfProfileOut( $fname );
2318                         return $matches[0];
2319                 } else {
2320                         if ( $isHTML ) {
2321                                 # Replace raw HTML by a placeholder
2322                                 # Add a blank line preceding, to prevent it from mucking up
2323                                 # immediately preceding headings
2324                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2325                         } else {
2326                                 # replace ==section headers==
2327                                 # XXX this needs to go away once we have a better parser.
2328                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2329                                         if( !is_null( $title ) )
2330                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2331                                         else
2332                                                 $encodedname = base64_encode("");
2333                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2334                                                 PREG_SPLIT_DELIM_CAPTURE);
2335                                         $text = '';
2336                                         $nsec = 0;
2337                                         for( $i = 0; $i < count($m); $i += 2 ) {
2338                                                 $text .= $m[$i];
2339                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2340                                                 $hl = $m[$i + 1];
2341                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2342                                                         $text .= $hl;
2343                                                         continue;
2344                                                 }
2345                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2346                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2347                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2348
2349                                                 $nsec++;
2350                                         }
2351                                 }
2352                         }
2353                 }
2354
2355                 # Prune lower levels off the recursion check path
2356                 $this->mTemplatePath = $lastPathLevel;
2357
2358                 if ( !$found ) {
2359                         wfProfileOut( $fname );
2360                         return $matches[0];
2361                 } else {
2362                         wfProfileOut( $fname );
2363                         return $text;
2364                 }
2365         }
2366
2367         /**
2368          * Translude an interwiki link.
2369          */
2370         function scarytransclude($title, $interwiki) {
2371                 global $wgEnableScaryTranscluding;
2372
2373                 if (!$wgEnableScaryTranscluding)
2374                         return wfMsg('scarytranscludedisabled');
2375
2376                 $articlename = "Template:" . $title->getDBkey();
2377                 $url = str_replace('$1', urlencode($articlename), $interwiki);
2378                 if (strlen($url) > 255)
2379                         return wfMsg('scarytranscludetoolong');
2380                 $text = $this->fetchScaryTemplateMaybeFromCache($url);
2381                 $this->mIWTransData[] = $text;
2382                 return "<!--IW_TRANSCLUDE ".(count($this->mIWTransData) - 1)."-->";
2383         }
2384
2385         function fetchScaryTemplateMaybeFromCache($url) {
2386                 $dbr =& wfGetDB(DB_SLAVE);
2387                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2388                                 array('tc_url' => $url));
2389                 if ($obj) {
2390                         $time = $obj->tc_time;
2391                         $text = $obj->tc_contents;
2392                         if ($time && $time < (time() + (60*60))) {
2393                                 return $text;
2394                         }
2395                 }
2396
2397                 $text = wfGetHTTP($url . '?action=render');
2398                 if (!$text)
2399                         return wfMsg('scarytranscludefailed', $url);
2400
2401                 $dbw =& wfGetDB(DB_MASTER);
2402                 $dbw->replace('transcache', array(), array(
2403                         'tc_url' => $url,
2404                         'tc_time' => time(),
2405                         'tc_contents' => $text));
2406                 return $text;
2407         }
2408
2409
2410         /**
2411          * Triple brace replacement -- used for template arguments
2412          * @access private
2413          */
2414         function argSubstitution( $matches ) {
2415                 $arg = trim( $matches[1] );
2416                 $text = $matches[0];
2417                 $inputArgs = end( $this->mArgStack );
2418
2419                 if ( array_key_exists( $arg, $inputArgs ) ) {
2420                         $text = $inputArgs[$arg];
2421                 }
2422
2423                 return $text;
2424         }
2425
2426         /**
2427          * Returns true if the function is allowed to include this entity
2428          * @access private
2429          */
2430         function incrementIncludeCount( $dbk ) {
2431                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2432                         $this->mIncludeCount[$dbk] = 0;
2433                 }
2434                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2435                         return true;
2436                 } else {
2437                         return false;
2438                 }
2439         }
2440
2441         /**
2442          * This function accomplishes several tasks:
2443          * 1) Auto-number headings if that option is enabled
2444          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2445          * 3) Add a Table of contents on the top for users who have enabled the option
2446          * 4) Auto-anchor headings
2447          *
2448          * It loops through all headlines, collects the necessary data, then splits up the
2449          * string and re-inserts the newly formatted headlines.
2450          *
2451          * @param string $text
2452          * @param boolean $isMain
2453          * @access private
2454          */
2455         function formatHeadings( $text, $isMain=true ) {
2456                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2457
2458                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2459                 $doShowToc = true;
2460                 $forceTocHere = false;
2461                 if( !$this->mTitle->userCanEdit() ) {
2462                         $showEditLink = 0;
2463                 } else {
2464                         $showEditLink = $this->mOptions->getEditSection();
2465                 }
2466
2467                 # Inhibit editsection links if requested in the page
2468                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2469                 if( $esw->matchAndRemove( $text ) ) {
2470                         $showEditLink = 0;
2471                 }
2472                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2473                 # do not add TOC
2474                 $mw =& MagicWord::get( MAG_NOTOC );
2475                 if( $mw->matchAndRemove( $text ) ) {
2476                         $doShowToc = false;
2477                 }
2478
2479                 # Get all headlines for numbering them and adding funky stuff like [edit]
2480                 # links - this is for later, but we need the number of headlines right now
2481                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2482
2483                 # if there are fewer than 4 headlines in the article, do not show TOC
2484                 if( $numMatches < 4 ) {
2485                         $doShowToc = false;
2486                 }
2487
2488                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2489                 # override above conditions and always show TOC at that place
2490
2491                 $mw =& MagicWord::get( MAG_TOC );
2492                 if($mw->match( $text ) ) {
2493                         $doShowToc = true;
2494                         $forceTocHere = true;
2495                 } else {
2496                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2497                         # override above conditions and always show TOC above first header
2498                         $mw =& MagicWord::get( MAG_FORCETOC );
2499                         if ($mw->matchAndRemove( $text ) ) {
2500                                 $doShowToc = true;
2501                         }
2502                 }
2503
2504                 # Never ever show TOC if no headers
2505                 if( $numMatches < 1 ) {
2506                         $doShowToc = false;
2507                 }
2508
2509                 # We need this to perform operations on the HTML
2510                 $sk =& $this->mOptions->getSkin();
2511
2512                 # headline counter
2513                 $headlineCount = 0;
2514                 $sectionCount = 0; # headlineCount excluding template sections
2515
2516                 # Ugh .. the TOC should have neat indentation levels which can be
2517                 # passed to the skin functions. These are determined here
2518                 $toc = '';
2519                 $full = '';
2520                 $head = array();
2521                 $sublevelCount = array();
2522                 $levelCount = array();
2523                 $toclevel = 0;
2524                 $level = 0;
2525                 $prevlevel = 0;
2526                 $toclevel = 0;
2527                 $prevtoclevel = 0;
2528
2529                 foreach( $matches[3] as $headline ) {
2530                         $istemplate = 0;
2531                         $templatetitle = '';
2532                         $templatesection = 0;
2533                         $numbering = '';
2534
2535                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2536                                 $istemplate = 1;
2537                                 $templatetitle = base64_decode($mat[1]);
2538                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2539                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2540                         }
2541
2542                         if( $toclevel ) {
2543                                 $prevlevel = $level;
2544                                 $prevtoclevel = $toclevel;
2545                         }
2546                         $level = $matches[1][$headlineCount];
2547
2548                         if( $doNumberHeadings || $doShowToc ) {
2549
2550                                 if ( $level > $prevlevel ) {
2551                                         # Increase TOC level
2552                                         $toclevel++;
2553                                         $sublevelCount[$toclevel] = 0;
2554                                         $toc .= $sk->tocIndent();
2555                                 }
2556                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2557                                         # Decrease TOC level, find level to jump to
2558
2559                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2560                                                 # Can only go down to level 1
2561                                                 $toclevel = 1;
2562                                         } else {
2563                                                 for ($i = $toclevel; $i > 0; $i--) {
2564                                                         if ( $levelCount[$i] == $level ) {
2565                                                                 # Found last matching level
2566                                                                 $toclevel = $i;
2567                                                                 break;
2568                                                         }
2569                                                         elseif ( $levelCount[$i] < $level ) {
2570                                                                 # Found first matching level below current level
2571                                                                 $toclevel = $i + 1;
2572                                                                 break;
2573                                                         }
2574                                                 }
2575                                         }
2576
2577                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2578                                 }
2579                                 else {
2580                                         # No change in level, end TOC line
2581                                         $toc .= $sk->tocLineEnd();
2582                                 }
2583
2584                                 $levelCount[$toclevel] = $level;
2585
2586                                 # count number of headlines for each level
2587                                 @$sublevelCount[$toclevel]++;
2588                                 $dot = 0;
2589                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2590                                         if( !empty( $sublevelCount[$i] ) ) {
2591                                                 if( $dot ) {
2592                                                         $numbering .= '.';
2593                                                 }
2594                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2595                                                 $dot = 1;
2596                                         }
2597                                 }
2598                         }
2599
2600                         # The canonized header is a version of the header text safe to use for links
2601                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2602                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2603                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2604
2605                         # Remove link placeholders by the link text.
2606                         #     <!--LINK number-->
2607                         # turns into
2608                         #     link text with suffix
2609                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2610                                                             "\$this->mLinkHolders['texts'][\$1]",
2611                                                             $canonized_headline );
2612                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2613                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2614                                                             $canonized_headline );
2615
2616                         # strip out HTML
2617                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2618                         $tocline = trim( $canonized_headline );
2619                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2620                         $replacearray = array(
2621                                 '%3A' => ':',
2622                                 '%' => '.'
2623                         );
2624                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2625                         $refers[$headlineCount] = $canonized_headline;
2626
2627                         # count how many in assoc. array so we can track dupes in anchors
2628                         @$refers[$canonized_headline]++;
2629                         $refcount[$headlineCount]=$refers[$canonized_headline];
2630
2631                         # Don't number the heading if it is the only one (looks silly)
2632                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2633                                 # the two are different if the line contains a link
2634                                 $headline=$numbering . ' ' . $headline;
2635                         }
2636
2637                         # Create the anchor for linking from the TOC to the section
2638                         $anchor = $canonized_headline;
2639                         if($refcount[$headlineCount] > 1 ) {
2640                                 $anchor .= '_' . $refcount[$headlineCount];
2641                         }
2642                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2643                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2644                         }
2645                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2646                                 if ( empty( $head[$headlineCount] ) ) {
2647                                         $head[$headlineCount] = '';
2648                                 }
2649                                 if( $istemplate )
2650                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2651                                 else
2652                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2653                         }
2654
2655                         # give headline the correct <h#> tag
2656                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2657
2658                         $headlineCount++;
2659                         if( !$istemplate )
2660                                 $sectionCount++;
2661                 }
2662
2663                 if( $doShowToc ) {
2664                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2665                         $toc = $sk->tocList( $toc );
2666                 }
2667
2668                 # split up and insert constructed headlines
2669
2670                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2671                 $i = 0;
2672
2673                 foreach( $blocks as $block ) {
2674                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2675                                 # This is the [edit] link that appears for the top block of text when
2676                                 # section editing is enabled
2677
2678                                 # Disabled because it broke block formatting
2679                                 # For example, a bullet point in the top line
2680                                 # $full .= $sk->editSectionLink(0);
2681                         }
2682                         $full .= $block;
2683                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2684                         # Top anchor now in skin
2685                                 $full = $full.$toc;
2686                         }
2687
2688                         if( !empty( $head[$i] ) ) {
2689                                 $full .= $head[$i];
2690                         }
2691                         $i++;
2692                 }
2693                 if($forceTocHere) {
2694                         $mw =& MagicWord::get( MAG_TOC );
2695                         return $mw->replace( $toc, $full );
2696                 } else {
2697                         return $full;
2698                 }
2699         }
2700
2701         /**
2702          * Return an HTML link for the "ISBN 123456" text
2703          * @access private
2704          */
2705         function magicISBN( $text ) {
2706                 $fname = 'Parser::magicISBN';
2707                 wfProfileIn( $fname );
2708
2709                 $a = split( 'ISBN ', ' '.$text );
2710                 if ( count ( $a ) < 2 ) {
2711                         wfProfileOut( $fname );
2712                         return $text;
2713                 }
2714                 $text = substr( array_shift( $a ), 1);
2715                 $valid = '0123456789-Xx';
2716
2717                 foreach ( $a as $x ) {
2718                         $isbn = $blank = '' ;
2719                         while ( ' ' == $x{0} ) {
2720                                 $blank .= ' ';
2721                                 $x = substr( $x, 1 );
2722                         }
2723                         if ( $x == '' ) { # blank isbn
2724                                 $text .= "ISBN $blank";
2725                                 continue;
2726                         }
2727                         while ( strstr( $valid, $x{0} ) != false ) {
2728                                 $isbn .= $x{0};
2729                                 $x = substr( $x, 1 );
2730                         }
2731                         $num = str_replace( '-', '', $isbn );
2732                         $num = str_replace( ' ', '', $num );
2733                         $num = str_replace( 'x', 'X', $num );
2734
2735                         if ( '' == $num ) {
2736                                 $text .= "ISBN $blank$x";
2737                         } else {
2738                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2739                                 $text .= '<a href="' .
2740                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2741                                         "\" class=\"internal\">ISBN $isbn</a>";
2742                                 $text .= $x;
2743                         }
2744                 }
2745                 wfProfileOut( $fname );
2746                 return $text;
2747         }
2748
2749         /**
2750          * Return an HTML link for the "RFC 1234" text
2751          *
2752          * @access private
2753          * @param string $text     Text to be processed
2754          * @param string $keyword  Magic keyword to use (default RFC)
2755          * @param string $urlmsg   Interface message to use (default rfcurl)
2756          * @return string
2757          */
2758         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2759
2760                 $valid = '0123456789';
2761                 $internal = false;
2762
2763                 $a = split( $keyword, ' '.$text );
2764                 if ( count ( $a ) < 2 ) {
2765                         return $text;
2766                 }
2767                 $text = substr( array_shift( $a ), 1);
2768
2769                 /* Check if keyword is preceed by [[.
2770                  * This test is made here cause of the array_shift above
2771                  * that prevent the test to be done in the foreach.
2772                  */
2773                 if ( substr( $text, -2 ) == '[[' ) {
2774                         $internal = true;
2775                 }
2776
2777                 foreach ( $a as $x ) {
2778                         /* token might be empty if we have RFC RFC 1234 */
2779                         if ( $x=='' ) {
2780                                 $text.=$keyword;
2781                                 continue;
2782                                 }
2783
2784                         $id = $blank = '' ;
2785
2786                         /** remove and save whitespaces in $blank */
2787                         while ( $x{0} == ' ' ) {
2788                                 $blank .= ' ';
2789                                 $x = substr( $x, 1 );
2790                         }
2791
2792                         /** remove and save the rfc number in $id */
2793                         while ( strstr( $valid, $x{0} ) != false ) {
2794                                 $id .= $x{0};
2795                                 $x = substr( $x, 1 );
2796                         }
2797
2798                         if ( $id == '' ) {
2799                                 /* call back stripped spaces*/
2800                                 $text .= $keyword.$blank.$x;
2801                         } elseif( $internal ) {
2802                                 /* normal link */
2803                                 $text .= $keyword.$id.$x;
2804                         } else {
2805                                 /* build the external link*/
2806                                 $url = wfMsg( $urlmsg, $id);
2807                                 $sk =& $this->mOptions->getSkin();
2808                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2809                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2810                         }
2811
2812                         /* Check if the next RFC keyword is preceed by [[ */
2813                         $internal = ( substr($x,-2) == '[[' );
2814                 }
2815                 return $text;
2816         }
2817
2818         /**
2819          * Transform wiki markup when saving a page by doing \r\n -> \n
2820          * conversion, substitting signatures, {{subst:}} templates, etc.
2821          *
2822          * @param string $text the text to transform
2823          * @param Title &$title the Title object for the current article
2824          * @param User &$user the User object describing the current user
2825          * @param ParserOptions $options parsing options
2826          * @param bool $clearState whether to clear the parser state first
2827          * @return string the altered wiki markup
2828          * @access public
2829          */
2830         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2831                 $this->mOptions = $options;
2832                 $this->mTitle =& $title;
2833                 $this->mOutputType = OT_WIKI;
2834
2835                 if ( $clearState ) {
2836                         $this->clearState();
2837                 }
2838
2839                 $stripState = false;
2840                 $pairs = array(
2841                         "\r\n" => "\n",
2842                 );
2843                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2844                 $text = $this->strip( $text, $stripState, false );
2845                 $text = $this->pstPass2( $text, $user );
2846                 $text = $this->unstrip( $text, $stripState );
2847                 $text = $this->unstripNoWiki( $text, $stripState );
2848                 return $text;
2849         }
2850
2851         /**
2852          * Pre-save transform helper function
2853          * @access private
2854          */
2855         function pstPass2( $text, &$user ) {
2856                 global $wgContLang, $wgLocaltimezone;
2857
2858                 # Variable replacement
2859                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2860                 $text = $this->replaceVariables( $text );
2861
2862                 # Signatures
2863                 #
2864                 $n = $user->getName();
2865                 $k = $user->getOption( 'nickname' );
2866                 if ( '' == $k ) { $k = $n; }
2867                 if ( isset( $wgLocaltimezone ) ) {
2868                         $oldtz = getenv( 'TZ' );
2869                         putenv( 'TZ='.$wgLocaltimezone );
2870                 }
2871
2872                 /* Note: This is the timestamp saved as hardcoded wikitext to
2873                  * the database, we use $wgContLang here in order to give
2874                  * everyone the same signiture and use the default one rather
2875                  * than the one selected in each users preferences.
2876                  */
2877                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
2878                   ' (' . date( 'T' ) . ')';
2879                 if ( isset( $wgLocaltimezone ) ) {
2880                         putenv( 'TZ='.$oldtz );
2881                 }
2882
2883                 if( $user->getOption( 'fancysig' ) ) {
2884                         $sigText = $k;
2885                 } else {
2886                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2887                 }
2888                 $text = preg_replace( '/~~~~~/', $d, $text );
2889                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2890                 $text = preg_replace( '/~~~/', $sigText, $text );
2891
2892                 # Context links: [[|name]] and [[name (context)|]]
2893                 #
2894                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2895                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2896                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2897                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2898
2899                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2900                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2901                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2902                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2903                 $context = '';
2904                 $t = $this->mTitle->getText();
2905                 if ( preg_match( $conpat, $t, $m ) ) {
2906                         $context = $m[2];
2907                 }
2908                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2909                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2910                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2911
2912                 if ( '' == $context ) {
2913                         $text = preg_replace( $p2, '[[\\1]]', $text );
2914                 } else {
2915                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2916                 }
2917
2918                 # Trim trailing whitespace
2919                 # MAG_END (__END__) tag allows for trailing
2920                 # whitespace to be deliberately included
2921                 $text = rtrim( $text );
2922                 $mw =& MagicWord::get( MAG_END );
2923                 $mw->matchAndRemove( $text );
2924
2925                 return $text;
2926         }
2927
2928         /**
2929          * Set up some variables which are usually set up in parse()
2930          * so that an external function can call some class members with confidence
2931          * @access public
2932          */
2933         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2934                 $this->mTitle =& $title;
2935                 $this->mOptions = $options;
2936                 $this->mOutputType = $outputType;
2937                 if ( $clearState ) {
2938                         $this->clearState();
2939                 }
2940         }
2941
2942         /**
2943          * Transform a MediaWiki message by replacing magic variables.
2944          *
2945          * @param string $text the text to transform
2946          * @param ParserOptions $options  options
2947          * @return string the text with variables substituted
2948          * @access public
2949          */
2950         function transformMsg( $text, $options ) {
2951                 global $wgTitle;
2952                 static $executing = false;
2953
2954                 # Guard against infinite recursion
2955                 if ( $executing ) {
2956                         return $text;
2957                 }
2958                 $executing = true;
2959
2960                 $this->mTitle = $wgTitle;
2961                 $this->mOptions = $options;
2962                 $this->mOutputType = OT_MSG;
2963                 $this->clearState();
2964                 $text = $this->replaceVariables( $text );
2965
2966                 $executing = false;
2967                 return $text;
2968         }
2969
2970         /**
2971          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2972          * Callback will be called with the text within
2973          * Transform and return the text within
2974          * @access public
2975          */
2976         function setHook( $tag, $callback ) {
2977                 $oldVal = @$this->mTagHooks[$tag];
2978                 $this->mTagHooks[$tag] = $callback;
2979                 return $oldVal;
2980         }
2981
2982         /**
2983          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2984          * Placeholders created in Skin::makeLinkObj()
2985          * Returns an array of links found, indexed by PDBK:
2986          *  0 - broken
2987          *  1 - normal link
2988          *  2 - stub
2989          * $options is a bit field, RLH_FOR_UPDATE to select for update
2990          */
2991         function replaceLinkHolders( &$text, $options = 0 ) {
2992                 global $wgUser, $wgLinkCache;
2993                 global $wgOutputReplace;
2994
2995                 $fname = 'Parser::replaceLinkHolders';
2996                 wfProfileIn( $fname );
2997
2998                 $pdbks = array();
2999                 $colours = array();
3000                 $sk = $this->mOptions->getSkin();
3001
3002                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3003                         wfProfileIn( $fname.'-check' );
3004                         $dbr =& wfGetDB( DB_SLAVE );
3005                         $page = $dbr->tableName( 'page' );
3006                         $threshold = $wgUser->getOption('stubthreshold');
3007
3008                         # Sort by namespace
3009                         asort( $this->mLinkHolders['namespaces'] );
3010
3011                         # Generate query
3012                         $query = false;
3013                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
3014                                 # Make title object
3015                                 $title = $this->mLinkHolders['titles'][$key];
3016
3017                                 # Skip invalid entries.
3018                                 # Result will be ugly, but prevents crash.
3019                                 if ( is_null( $title ) ) {
3020                                         continue;
3021                                 }
3022                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3023
3024                                 # Check if it's in the link cache already
3025                                 if ( $title->isAlwaysKnown() || $wgLinkCache->getGoodLinkID( $pdbk ) ) {
3026                                         $colours[$pdbk] = 1;
3027                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
3028                                         $colours[$pdbk] = 0;
3029                                 } else {
3030                                         # Not in the link cache, add it to the query
3031                                         if ( !isset( $current ) ) {
3032                                                 $current = $val;
3033                                                 $query =  "SELECT page_id, page_namespace, page_title";
3034                                                 if ( $threshold > 0 ) {
3035                                                         $query .= ', page_len, page_is_redirect';
3036                                                 }
3037                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
3038                                         } elseif ( $current != $val ) {
3039                                                 $current = $val;
3040                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
3041                                         } else {
3042                                                 $query .= ', ';
3043                                         }
3044
3045                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3046                                 }
3047                         }
3048                         if ( $query ) {
3049                                 $query .= '))';
3050                                 if ( $options & RLH_FOR_UPDATE ) {
3051                                         $query .= ' FOR UPDATE';
3052                                 }
3053
3054                                 $res = $dbr->query( $query, $fname );
3055
3056                                 # Fetch data and form into an associative array
3057                                 # non-existent = broken
3058                                 # 1 = known
3059                                 # 2 = stub
3060                                 while ( $s = $dbr->fetchObject($res) ) {
3061                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3062                                         $pdbk = $title->getPrefixedDBkey();
3063                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
3064
3065                                         if ( $threshold >  0 ) {
3066                                                 $size = $s->page_len;
3067                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3068                                                         $colours[$pdbk] = 1;
3069                                                 } else {
3070                                                         $colours[$pdbk] = 2;
3071                                                 }
3072                                         } else {
3073                                                 $colours[$pdbk] = 1;
3074                                         }
3075                                 }
3076                         }
3077                         wfProfileOut( $fname.'-check' );
3078
3079                         # Construct search and replace arrays
3080                         wfProfileIn( $fname.'-construct' );
3081                         $wgOutputReplace = array();
3082                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3083                                 $pdbk = $pdbks[$key];
3084                                 $searchkey = "<!--LINK $key-->";
3085                                 $title = $this->mLinkHolders['titles'][$key];
3086                                 if ( empty( $colours[$pdbk] ) ) {
3087                                         $wgLinkCache->addBadLinkObj( $title );
3088                                         $colours[$pdbk] = 0;
3089                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3090                                                                         $this->mLinkHolders['texts'][$key],
3091                                                                         $this->mLinkHolders['queries'][$key] );
3092                                 } elseif ( $colours[$pdbk] == 1 ) {
3093                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3094                                                                         $this->mLinkHolders['texts'][$key],
3095                                                                         $this->mLinkHolders['queries'][$key] );
3096                                 } elseif ( $colours[$pdbk] == 2 ) {
3097                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3098                                                                         $this->mLinkHolders['texts'][$key],
3099                                                                         $this->mLinkHolders['queries'][$key] );
3100                                 }
3101                         }
3102                         wfProfileOut( $fname.'-construct' );
3103
3104                         # Do the thing
3105                         wfProfileIn( $fname.'-replace' );
3106
3107                         $text = preg_replace_callback(
3108                                 '/(<!--LINK .*?-->)/',
3109                                 "wfOutputReplaceMatches",
3110                                 $text);
3111
3112                         wfProfileOut( $fname.'-replace' );
3113                 }
3114
3115                 # Now process interwiki link holders
3116                 # This is quite a bit simpler than internal links
3117                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3118                         wfProfileIn( $fname.'-interwiki' );
3119                         # Make interwiki link HTML
3120                         $wgOutputReplace = array();
3121                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3122                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3123                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3124                         }
3125
3126                         $text = preg_replace_callback(
3127                                 '/<!--IWLINK (.*?)-->/',
3128                                 "wfOutputReplaceMatches",
3129                                 $text );
3130                         wfProfileOut( $fname.'-interwiki' );
3131                 }
3132
3133                 wfProfileOut( $fname );
3134                 return $colours;
3135         }
3136
3137         /**
3138          * Replace <!--LINK--> link placeholders with plain text of links
3139          * (not HTML-formatted).
3140          * @param string $text
3141          * @return string
3142          */
3143         function replaceLinkHoldersText( $text ) {
3144                 global $wgUser, $wgLinkCache;
3145                 global $wgOutputReplace;
3146
3147                 $fname = 'Parser::replaceLinkHoldersText';
3148                 wfProfileIn( $fname );
3149
3150                 $text = preg_replace_callback(
3151                         '/<!--(LINK|IWLINK) (.*?)-->/',
3152                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3153                         $text );
3154
3155                 wfProfileOut( $fname );
3156                 return $text;
3157         }
3158
3159         /**
3160          * @param array $matches
3161          * @return string
3162          * @access private
3163          */
3164         function replaceLinkHoldersTextCallback( $matches ) {
3165                 $type = $matches[1];
3166                 $key  = $matches[2];
3167                 if( $type == 'LINK' ) {
3168                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3169                                 return $this->mLinkHolders['texts'][$key];
3170                         }
3171                 } elseif( $type == 'IWLINK' ) {
3172                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3173                                 return $this->mInterwikiLinkHolders['texts'][$key];
3174                         }
3175                 }
3176                 return $matches[0];
3177         }
3178
3179         /**
3180          * Renders an image gallery from a text with one line per image.
3181          * text labels may be given by using |-style alternative text. E.g.
3182          *   Image:one.jpg|The number "1"
3183          *   Image:tree.jpg|A tree
3184          * given as text will return the HTML of a gallery with two images,
3185          * labeled 'The number "1"' and
3186          * 'A tree'.
3187          *
3188          * @static
3189          */
3190         function renderImageGallery( $text ) {
3191                 # Setup the parser
3192                 global $wgUser, $wgTitle;
3193                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3194                 $localParser = new Parser();
3195
3196                 global $wgLinkCache;
3197                 $ig = new ImageGallery();
3198                 $ig->setShowBytes( false );
3199                 $ig->setShowFilename( false );
3200                 $lines = explode( "\n", $text );
3201
3202                 foreach ( $lines as $line ) {
3203                         # match lines like these:
3204                         # Image:someimage.jpg|This is some image
3205                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3206                         # Skip empty lines
3207                         if ( count( $matches ) == 0 ) {
3208                                 continue;
3209                         }
3210                         $nt = Title::newFromURL( $matches[1] );
3211                         if( is_null( $nt ) ) {
3212                                 # Bogus title. Ignore these so we don't bomb out later.
3213                                 continue;
3214                         }
3215                         if ( isset( $matches[3] ) ) {
3216                                 $label = $matches[3];
3217                         } else {
3218                                 $label = '';
3219                         }
3220
3221                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3222                         $html = $html->mText;
3223
3224                         $ig->add( new Image( $nt ), $html );
3225                         $wgLinkCache->addImageLinkObj( $nt );
3226                 }
3227                 return $ig->toHTML();
3228         }
3229
3230         /**
3231          * Parse image options text and use it to make an image
3232          */
3233         function makeImage( &$nt, $options ) {
3234                 global $wgContLang, $wgUseImageResize;
3235                 global $wgUser, $wgThumbLimits;
3236
3237                 $align = '';
3238
3239                 # Check if the options text is of the form "options|alt text"
3240                 # Options are:
3241                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3242                 #  * left               no resizing, just left align. label is used for alt= only
3243                 #  * right              same, but right aligned
3244                 #  * none               same, but not aligned
3245                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3246                 #  * center             center the image
3247                 #  * framed             Keep original image size, no magnify-button.
3248
3249                 $part = explode( '|', $options);
3250
3251                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3252                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3253                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3254                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3255                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3256                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3257                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3258                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3259                 $caption = '';
3260
3261                 $width = $height = $framed = $thumb = false;
3262                 $manual_thumb = '' ;
3263
3264                 foreach( $part as $key => $val ) {
3265                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3266                                 $thumb=true;
3267                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3268                                 # use manually specified thumbnail
3269                                 $thumb=true;
3270                                 $manual_thumb = $match;
3271                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3272                                 # remember to set an alignment, don't render immediately
3273                                 $align = 'right';
3274                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3275                                 # remember to set an alignment, don't render immediately
3276                                 $align = 'left';
3277                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3278                                 # remember to set an alignment, don't render immediately
3279                                 $align = 'center';
3280                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3281                                 # remember to set an alignment, don't render immediately
3282                                 $align = 'none';
3283                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3284                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3285                                 # $match is the image width in pixels
3286                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3287                                         $width = intval( $m[1] );
3288                                         $height = intval( $m[2] );
3289                                 } else {
3290                                         $width = intval($match);
3291                                 }
3292                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3293                                 $framed=true;
3294                         } else {
3295                                 $caption = $val;
3296                         }
3297                 }
3298                 # Strip bad stuff out of the alt text
3299                 $alt = $this->replaceLinkHoldersText( $caption );
3300                 $alt = Sanitizer::stripAllTags( $alt );
3301
3302                 # Linker does the rest
3303                 $sk =& $this->mOptions->getSkin();
3304                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3305         }
3306
3307         /**
3308          * Set a flag in the output object indicating that the content is dynamic and
3309          * shouldn't be cached.
3310          */
3311         function disableCache() {
3312                 $this->mOutput->mCacheTime = -1;
3313         }
3314
3315         /**
3316          * Callback from the Sanitizer for expanding items found in HTML attribute
3317          * values, so they can be safely tested and escaped.
3318          * @param string $text
3319          * @param array $args
3320          * @return string
3321          * @access private
3322          */
3323         function attributeStripCallback( &$text, $args ) {
3324                 $text = $this->replaceVariables( $text, $args );
3325                 $text = $this->unstripForHTML( $text );
3326                 return $text;
3327         }
3328
3329         function unstripForHTML( $text ) {
3330                 $text = $this->unstrip( $text, $this->mStripState );
3331                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3332                 return $text;
3333         }
3334 }
3335
3336 /**
3337  * @todo document
3338  * @package MediaWiki
3339  */
3340 class ParserOutput
3341 {
3342         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3343         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3344         var $mVersion;   # Compatibility check
3345         var $mTitleText; # title text of the chosen language variant
3346
3347         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3348                 $containsOldMagic = false, $titletext = '' )
3349         {
3350                 $this->mText = $text;
3351                 $this->mLanguageLinks = $languageLinks;
3352                 $this->mCategoryLinks = $categoryLinks;
3353                 $this->mContainsOldMagic = $containsOldMagic;
3354                 $this->mCacheTime = '';
3355                 $this->mVersion = MW_PARSER_VERSION;
3356                 $this->mTitleText = $titletext;
3357         }
3358
3359         function getText()                   { return $this->mText; }
3360         function getLanguageLinks()          { return $this->mLanguageLinks; }
3361         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3362         function getCacheTime()              { return $this->mCacheTime; }
3363         function getTitleText()              { return $this->mTitleText; }
3364         function containsOldMagic()          { return $this->mContainsOldMagic; }
3365         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3366         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3367         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3368         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3369         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3370         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3371
3372         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3373
3374         function merge( $other ) {
3375                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3376                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3377                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3378         }
3379
3380         /**
3381          * Return true if this cached output object predates the global or
3382          * per-article cache invalidation timestamps, or if it comes from
3383          * an incompatible older version.
3384          *
3385          * @param string $touched the affected article's last touched timestamp
3386          * @return bool
3387          * @access public
3388          */
3389         function expired( $touched ) {
3390                 global $wgCacheEpoch;
3391                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3392                        $this->getCacheTime() <= $touched ||
3393                        $this->getCacheTime() <= $wgCacheEpoch ||
3394                        !isset( $this->mVersion ) ||
3395                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3396         }
3397 }
3398
3399 /**
3400  * Set options of the Parser
3401  * @todo document
3402  * @package MediaWiki
3403  */
3404 class ParserOptions
3405 {
3406         # All variables are private
3407         var $mUseTeX;                    # Use texvc to expand <math> tags
3408         var $mUseDynamicDates;           # Use DateFormatter to format dates
3409         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3410         var $mAllowExternalImages;       # Allow external images inline
3411         var $mSkin;                      # Reference to the preferred skin
3412         var $mDateFormat;                # Date format index
3413         var $mEditSection;               # Create "edit section" links
3414         var $mNumberHeadings;            # Automatically number headings
3415         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3416
3417         function getUseTeX()                        { return $this->mUseTeX; }
3418         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3419         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3420         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3421         function &getSkin()                         { return $this->mSkin; }
3422         function getDateFormat()                    { return $this->mDateFormat; }
3423         function getEditSection()                   { return $this->mEditSection; }
3424         function getNumberHeadings()                { return $this->mNumberHeadings; }
3425         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3426
3427
3428         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3429         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3430         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3431         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3432         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3433         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3434         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3435         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3436
3437         function setSkin( &$x ) { $this->mSkin =& $x; }
3438
3439         function ParserOptions() {
3440                 global $wgUser;
3441                 $this->initialiseFromUser( $wgUser );
3442         }
3443
3444         /**
3445          * Get parser options
3446          * @static
3447          */
3448         function newFromUser( &$user ) {
3449                 $popts = new ParserOptions;
3450                 $popts->initialiseFromUser( $user );
3451                 return $popts;
3452         }
3453
3454         /** Get user options */
3455         function initialiseFromUser( &$userInput ) {
3456                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3457                        $wgAllowSpecialInclusion;
3458                 $fname = 'ParserOptions::initialiseFromUser';
3459                 wfProfileIn( $fname );
3460                 if ( !$userInput ) {
3461                         $user = new User;
3462                         $user->setLoaded( true );
3463                 } else {
3464                         $user =& $userInput;
3465                 }
3466
3467                 $this->mUseTeX = $wgUseTeX;
3468                 $this->mUseDynamicDates = $wgUseDynamicDates;
3469                 $this->mInterwikiMagic = $wgInterwikiMagic;
3470                 $this->mAllowExternalImages = $wgAllowExternalImages;
3471                 wfProfileIn( $fname.'-skin' );
3472                 $this->mSkin =& $user->getSkin();
3473                 wfProfileOut( $fname.'-skin' );
3474                 $this->mDateFormat = $user->getOption( 'date' );
3475                 $this->mEditSection = true;
3476                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3477                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3478                 wfProfileOut( $fname );
3479         }
3480 }
3481
3482 /**
3483  * Callback function used by Parser::replaceLinkHolders()
3484  * to substitute link placeholders.
3485  */
3486 function &wfOutputReplaceMatches( $matches ) {
3487         global $wgOutputReplace;
3488         return $wgOutputReplace[$matches[1]];
3489 }
3490
3491 /**
3492  * Return the total number of articles
3493  */
3494 function wfNumberOfArticles() {
3495         global $wgNumberOfArticles;
3496
3497         wfLoadSiteStats();
3498         return $wgNumberOfArticles;
3499 }
3500
3501 /**
3502  * Return the number of files
3503  */
3504 function wfNumberOfFiles() {
3505         $fname = 'Parser::wfNumberOfFiles';
3506
3507         wfProfileIn( $fname );
3508         $dbr =& wfGetDB( DB_SLAVE );
3509         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3510         wfProfileOut( $fname );
3511
3512         return $res;
3513 }
3514
3515 /**
3516  * Get various statistics from the database
3517  * @private
3518  */
3519 function wfLoadSiteStats() {
3520         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3521         $fname = 'wfLoadSiteStats';
3522
3523         if ( -1 != $wgNumberOfArticles ) return;
3524         $dbr =& wfGetDB( DB_SLAVE );
3525         $s = $dbr->selectRow( 'site_stats',
3526                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3527                 array( 'ss_row_id' => 1 ), $fname
3528         );
3529
3530         if ( $s === false ) {
3531                 return;
3532         } else {
3533                 $wgTotalViews = $s->ss_total_views;
3534                 $wgTotalEdits = $s->ss_total_edits;
3535                 $wgNumberOfArticles = $s->ss_good_articles;
3536         }
3537 }
3538
3539 /**
3540  * Escape html tags
3541  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3542  *
3543  * @param string $in Text that might contain HTML tags
3544  * @return string Escaped string
3545  */
3546 function wfEscapeHTMLTagsOnly( $in ) {
3547         return str_replace(
3548                 array( '"', '>', '<' ),
3549                 array( '&quot;', '&gt;', '&lt;' ),
3550                 $in );
3551 }
3552
3553 ?>