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