includes/Parser.php

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