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', '1.6.0' );
  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 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @access private
  94          */
  95         # Persistent:
  96         var $mTagHooks;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 100         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102
 103         # Temporary:
 104         var $mOptions,      // ParserOptions object
 105                 $mTitle,        // Title context, used for self-link rendering and similar things
 106                 $mOutputType,   // Output type, one of the OT_xxx constants
 107             $mTemplates,        // cache of already loaded templates, avoids
 108                                 // multiple SQL queries for the same string
 109             $mTemplatePath,     // stores an unsorted hash of all the templates already loaded
 110                                 // in this path. Used for loop detection.
 111                 $mIWTransData = array(),
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 122                 $this->mTemplates = array();
 123                 $this->mTemplatePath = array();
 124                 $this->mTagHooks = array();
 125                 $this->clearState();
 126         }
 127
 128         /**
 129          * Clear Parser state
 130          *
 131          * @access private
 132          */
 133         function clearState() {
 134                 $this->mOutput = new ParserOutput;
 135                 $this->mAutonumber = 0;
 136                 $this->mLastSection = '';
 137                 $this->mDTopen = false;
 138                 $this->mVariables = false;
 139                 $this->mIncludeCount = array();
 140                 $this->mStripState = array();
 141                 $this->mArgStack = array();
 142                 $this->mInPre = false;
 143                 $this->mInterwikiLinkHolders = array(
 144                         'texts' => array(),
 145                         'titles' => array()
 146                 );
 147                 $this->mLinkHolders = array(
 148                         'namespaces' => array(),
 149                         'dbkeys' => array(),
 150                         'queries' => array(),
 151                         'texts' => array(),
 152                         'titles' => array()
 153                 );
 154                 $this->mRevisionId = null;
 155                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 156
 157                 wfRunHooks( 'ParserClearState', array( &$this ) );
 158         }
 159
 160         /**
 161          * Accessor for mUniqPrefix.
 162          *
 163          * @access public
 164          */
 165         function UniqPrefix() {
 166                 return $this->mUniqPrefix;
 167         }
 168
 169         /**
 170          * Convert wikitext to HTML
 171          * Do not call this function recursively.
 172          *
 173          * @access private
 174          * @param string $text Text we want to parse
 175          * @param Title &$title A title object
 176          * @param array $options
 177          * @param boolean $linestart
 178          * @param boolean $clearState
 179          * @param int $revid number to pass in {{REVISIONID}}
 180          * @return ParserOutput a ParserOutput
 181          */
 182         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 183                 /**
 184                  * First pass--just handle <nowiki> sections, pass the rest off
 185                  * to internalParse() which does all the real work.
 186                  */
 187
 188                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 189                 $fname = 'Parser::parse';
 190                 wfProfileIn( $fname );
 191
 192                 if ( $clearState ) {
 193                         $this->clearState();
 194                 }
 195
 196                 $this->mOptions = $options;
 197                 $this->mTitle =& $title;
 198                 $this->mRevisionId = $revid;
 199                 $this->mOutputType = OT_HTML;
 200
 201                 $this->mStripState = NULL;
 202
 203                 //$text = $this->strip( $text, $this->mStripState );
 204                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 205                 $x =& $this->mStripState;
 206
 207                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 208                 $text = $this->strip( $text, $x );
 209                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 210
 211                 # Hook to suspend the parser in this state
 212                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 213                         wfProfileOut( $fname );
 214                         return $text ;
 215                 }
 216
 217                 $text = $this->internalParse( $text );
 218
 219                 $text = $this->unstrip( $text, $this->mStripState );
 220
 221                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 222                 $fixtags = array(
 223                         # french spaces, last one Guillemet-left
 224                         # only if there is something before the space
 225                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 226                         # french spaces, Guillemet-right
 227                         '/(\\302\\253) /' => '\\1&nbsp;',
 228                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 229                 );
 230                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 231
 232                 # only once and last
 233                 $text = $this->doBlockLevels( $text, $linestart );
 234
 235                 $this->replaceLinkHolders( $text );
 236
 237                 # the position of the convert() call should not be changed. it
 238                 # assumes that the links are all replaces and the only thing left
 239                 # is the <nowiki> mark.
 240                 $text = $wgContLang->convert($text);
 241                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 242
 243                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 244
 245                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 246
 247                 $text = Sanitizer::normalizeCharReferences( $text );
 248
 249                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 250                         $text = Parser::tidy($text);
 251                 }
 252
 253                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 254
 255                 $this->mOutput->setText( $text );
 256                 wfProfileOut( $fname );
 257
 258                 return $this->mOutput;
 259         }
 260
 261         /**
 262          * Get a random string
 263          *
 264          * @access private
 265          * @static
 266          */
 267         function getRandomString() {
 268                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 269         }
 270
 271         /**
 272          * Replaces all occurrences of <$tag>content</$tag> in the text
 273          * with a random marker and returns the new text. the output parameter
 274          * $content will be an associative array filled with data on the form
 275          * $unique_marker => content.
 276          *
 277          * If $content is already set, the additional entries will be appended
 278          * If $tag is set to STRIP_COMMENTS, the function will extract
 279          * <!-- HTML comments -->
 280          *
 281          * @access private
 282          * @static
 283          */
 284         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 285                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 286                 if ( !$content ) {
 287                         $content = array( );
 288                 }
 289                 $n = 1;
 290                 $stripped = '';
 291
 292                 if ( !$tags ) {
 293                         $tags = array( );
 294                 }
 295
 296                 if ( !$params ) {
 297                         $params = array( );
 298                 }
 299
 300                 if( $tag == STRIP_COMMENTS ) {
 301                         $start = '/<!--()()/';
 302                         $end   = '/-->/';
 303                 } else {
 304                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 305                         $end   = "/<\\/$tag\\s*>/i";
 306                 }
 307
 308                 while ( '' != $text ) {
 309                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 310                         $stripped .= $p[0];
 311                         if( count( $p ) < 4 ) {
 312                                 break;
 313                         }
 314                         $attributes = $p[1];
 315                         $empty      = $p[2];
 316                         $inside     = $p[3];
 317
 318                         $marker = $rnd . sprintf('%08X', $n++);
 319                         $stripped .= $marker;
 320
 321                         $tags[$marker] = "<$tag$attributes$empty>";
 322                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 323
 324                         if ( $empty === '/' ) {
 325                                 // Empty element tag, <tag />
 326                                 $content[$marker] = null;
 327                                 $text = $inside;
 328                         } else {
 329                                 $q = preg_split( $end, $inside, 2 );
 330                                 $content[$marker] = $q[0];
 331                                 if( count( $q ) < 2 ) {
 332                                         # No end tag -- let it run out to the end of the text.
 333                                         break;
 334                                 } else {
 335                                         $text = $q[1];
 336                                 }
 337                         }
 338                 }
 339                 return $stripped;
 340         }
 341
 342         /**
 343          * Wrapper function for extractTagsAndParams
 344          * for cases where $tags and $params isn't needed
 345          * i.e. where tags will never have params, like <nowiki>
 346          *
 347          * @access private
 348          * @static
 349          */
 350         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 351                 $dummy_tags = array();
 352                 $dummy_params = array();
 353
 354                 return Parser::extractTagsAndParams( $tag, $text, $content,
 355                         $dummy_tags, $dummy_params, $uniq_prefix );
 356         }
 357
 358         /**
 359          * Strips and renders nowiki, pre, math, hiero
 360          * If $render is set, performs necessary rendering operations on plugins
 361          * Returns the text, and fills an array with data needed in unstrip()
 362          * If the $state is already a valid strip state, it adds to the state
 363          *
 364          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 365          *  will be stripped in addition to other tags. This is important
 366          *  for section editing, where these comments cause confusion when
 367          *  counting the sections in the wikisource
 368          *
 369          * @access private
 370          */
 371         function strip( $text, &$state, $stripcomments = false ) {
 372                 $render = ($this->mOutputType == OT_HTML);
 373                 $html_content = array();
 374                 $nowiki_content = array();
 375                 $math_content = array();
 376                 $pre_content = array();
 377                 $comment_content = array();
 378                 $ext_content = array();
 379                 $ext_tags = array();
 380                 $ext_params = array();
 381                 $gallery_content = array();
 382
 383                 # Replace any instances of the placeholders
 384                 $uniq_prefix = $this->mUniqPrefix;
 385                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 386
 387                 # html
 388                 global $wgRawHtml;
 389                 if( $wgRawHtml ) {
 390                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 391                         foreach( $html_content as $marker => $content ) {
 392                                 if ($render ) {
 393                                         # Raw and unchecked for validity.
 394                                         $html_content[$marker] = $content;
 395                                 } else {
 396                                         $html_content[$marker] = '<html>'.$content.'</html>';
 397                                 }
 398                         }
 399                 }
 400
 401                 # nowiki
 402                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 403                 foreach( $nowiki_content as $marker => $content ) {
 404                         if( $render ){
 405                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 406                         } else {
 407                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 408                         }
 409                 }
 410
 411                 # math
 412                 if( $this->mOptions->getUseTeX() ) {
 413                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 414                         foreach( $math_content as $marker => $content ){
 415                                 if( $render ) {
 416                                         $math_content[$marker] = renderMath( $content );
 417                                 } else {
 418                                         $math_content[$marker] = '<math>'.$content.'</math>';
 419                                 }
 420                         }
 421                 }
 422
 423                 # pre
 424                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 425                 foreach( $pre_content as $marker => $content ){
 426                         if( $render ){
 427                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 428                         } else {
 429                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 430                         }
 431                 }
 432
 433                 # gallery
 434                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 435                 foreach( $gallery_content as $marker => $content ) {
 436                         require_once( 'ImageGallery.php' );
 437                         if ( $render ) {
 438                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 439                         } else {
 440                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 441                         }
 442                 }
 443
 444                 # Comments
 445                 if($stripcomments) {
 446                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 447                         foreach( $comment_content as $marker => $content ){
 448                                 $comment_content[$marker] = '<!--'.$content.'-->';
 449                         }
 450                 }
 451
 452                 # Extensions
 453                 foreach ( $this->mTagHooks as $tag => $callback ) {
 454                         $ext_content[$tag] = array();
 455                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 456                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 457                         foreach( $ext_content[$tag] as $marker => $content ) {
 458                                 $full_tag = $ext_tags[$tag][$marker];
 459                                 $params = $ext_params[$tag][$marker];
 460                                 if ( $render )
 461                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 462                                 else {
 463                                         if ( is_null( $content ) ) {
 464                                                 // Empty element tag
 465                                                 $ext_content[$tag][$marker] = $full_tag;
 466                                         } else {
 467                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 468                                         }
 469                                 }
 470                         }
 471                 }
 472
 473                 # Merge state with the pre-existing state, if there is one
 474                 if ( $state ) {
 475                         $state['html'] = $state['html'] + $html_content;
 476                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 477                         $state['math'] = $state['math'] + $math_content;
 478                         $state['pre'] = $state['pre'] + $pre_content;
 479                         $state['comment'] = $state['comment'] + $comment_content;
 480                         $state['gallery'] = $state['gallery'] + $gallery_content;
 481
 482                         foreach( $ext_content as $tag => $array ) {
 483                                 if ( array_key_exists( $tag, $state ) ) {
 484                                         $state[$tag] = $state[$tag] + $array;
 485                                 }
 486                         }
 487                 } else {
 488                         $state = array(
 489                           'html' => $html_content,
 490                           'nowiki' => $nowiki_content,
 491                           'math' => $math_content,
 492                           'pre' => $pre_content,
 493                           'comment' => $comment_content,
 494                           'gallery' => $gallery_content,
 495                         ) + $ext_content;
 496                 }
 497                 return $text;
 498         }
 499
 500         /**
 501          * restores pre, math, and hiero removed by strip()
 502          *
 503          * always call unstripNoWiki() after this one
 504          * @access private
 505          */
 506         function unstrip( $text, &$state ) {
 507                 if ( !is_array( $state ) ) {
 508                         return $text;
 509                 }
 510
 511                 # Must expand in reverse order, otherwise nested tags will be corrupted
 512                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 513                         if( $tag != 'nowiki' && $tag != 'html' ) {
 514                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 515                                         $text = str_replace( $uniq, $content, $text );
 516                                 }
 517                         }
 518                 }
 519
 520                 return $text;
 521         }
 522
 523         /**
 524          * always call this after unstrip() to preserve the order
 525          *
 526          * @access private
 527          */
 528         function unstripNoWiki( $text, &$state ) {
 529                 if ( !is_array( $state ) ) {
 530                         return $text;
 531                 }
 532
 533                 # Must expand in reverse order, otherwise nested tags will be corrupted
 534                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 535                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 536                 }
 537
 538                 global $wgRawHtml;
 539                 if ($wgRawHtml) {
 540                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 541                                 $text = str_replace( key( $state['html'] ), $content, $text );
 542                         }
 543                 }
 544
 545                 return $text;
 546         }
 547
 548         /**
 549          * Add an item to the strip state
 550          * Returns the unique tag which must be inserted into the stripped text
 551          * The tag will be replaced with the original text in unstrip()
 552          *
 553          * @access private
 554          */
 555         function insertStripItem( $text, &$state ) {
 556                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 557                 if ( !$state ) {
 558                         $state = array(
 559                           'html' => array(),
 560                           'nowiki' => array(),
 561                           'math' => array(),
 562                           'pre' => array(),
 563                           'comment' => array(),
 564                           'gallery' => array(),
 565                         );
 566                 }
 567                 $state['item'][$rnd] = $text;
 568                 return $rnd;
 569         }
 570
 571         /**
 572          * Interface with html tidy, used if $wgUseTidy = true.
 573          * If tidy isn't able to correct the markup, the original will be
 574          * returned in all its glory with a warning comment appended.
 575          *
 576          * Either the external tidy program or the in-process tidy extension
 577          * will be used depending on availability. Override the default
 578          * $wgTidyInternal setting to disable the internal if it's not working.
 579          *
 580          * @param string $text Hideous HTML input
 581          * @return string Corrected HTML output
 582          * @access public
 583          * @static
 584          */
 585         function tidy( $text ) {
 586                 global $wgTidyInternal;
 587                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 588 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 589 '<head><title>test</title></head><body>'.$text.'</body></html>';
 590                 if( $wgTidyInternal ) {
 591                         $correctedtext = Parser::internalTidy( $wrappedtext );
 592                 } else {
 593                         $correctedtext = Parser::externalTidy( $wrappedtext );
 594                 }
 595                 if( is_null( $correctedtext ) ) {
 596                         wfDebug( "Tidy error detected!\n" );
 597                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 598                 }
 599                 return $correctedtext;
 600         }
 601
 602         /**
 603          * Spawn an external HTML tidy process and get corrected markup back from it.
 604          *
 605          * @access private
 606          * @static
 607          */
 608         function externalTidy( $text ) {
 609                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 610                 $fname = 'Parser::externalTidy';
 611                 wfProfileIn( $fname );
 612
 613                 $cleansource = '';
 614                 $opts = ' -utf8';
 615
 616                 $descriptorspec = array(
 617                         0 => array('pipe', 'r'),
 618                         1 => array('pipe', 'w'),
 619                         2 => array('file', '/dev/null', 'a')
 620                 );
 621                 $pipes = array();
 622                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 623                 if (is_resource($process)) {
 624                         fwrite($pipes[0], $text);
 625                         fclose($pipes[0]);
 626                         while (!feof($pipes[1])) {
 627                                 $cleansource .= fgets($pipes[1], 1024);
 628                         }
 629                         fclose($pipes[1]);
 630                         proc_close($process);
 631                 }
 632
 633                 wfProfileOut( $fname );
 634
 635                 if( $cleansource == '' && $text != '') {
 636                         // Some kind of error happened, so we couldn't get the corrected text.
 637                         // Just give up; we'll use the source text and append a warning.
 638                         return null;
 639                 } else {
 640                         return $cleansource;
 641                 }
 642         }
 643
 644         /**
 645          * Use the HTML tidy PECL extension to use the tidy library in-process,
 646          * saving the overhead of spawning a new process. Currently written to
 647          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 648          *
 649          * 'pear install tidy' should be able to compile the extension module.
 650          *
 651          * @access private
 652          * @static
 653          */
 654         function internalTidy( $text ) {
 655                 global $wgTidyConf;
 656                 $fname = 'Parser::internalTidy';
 657                 wfProfileIn( $fname );
 658
 659                 tidy_load_config( $wgTidyConf );
 660                 tidy_set_encoding( 'utf8' );
 661                 tidy_parse_string( $text );
 662                 tidy_clean_repair();
 663                 if( tidy_get_status() == 2 ) {
 664                         // 2 is magic number for fatal error
 665                         // http://www.php.net/manual/en/function.tidy-get-status.php
 666                         $cleansource = null;
 667                 } else {
 668                         $cleansource = tidy_get_output();
 669                 }
 670                 wfProfileOut( $fname );
 671                 return $cleansource;
 672         }
 673
 674         /**
 675          * parse the wiki syntax used to render tables
 676          *
 677          * @access private
 678          */
 679         function doTableStuff ( $t ) {
 680                 $fname = 'Parser::doTableStuff';
 681                 wfProfileIn( $fname );
 682
 683                 $t = explode ( "\n" , $t ) ;
 684                 $td = array () ; # Is currently a td tag open?
 685                 $ltd = array () ; # Was it TD or TH?
 686                 $tr = array () ; # Is currently a tr tag open?
 687                 $ltr = array () ; # tr attributes
 688                 $indent_level = 0; # indent level of the table
 689                 foreach ( $t AS $k => $x )
 690                 {
 691                         $x = trim ( $x ) ;
 692                         $fc = substr ( $x , 0 , 1 ) ;
 693                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 694                                 $indent_level = strlen( $matches[1] );
 695
 696                                 $attributes = $this->unstripForHTML( $matches[2] );
 697
 698                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 699                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 700                                 array_push ( $td , false ) ;
 701                                 array_push ( $ltd , '' ) ;
 702                                 array_push ( $tr , false ) ;
 703                                 array_push ( $ltr , '' ) ;
 704                         }
 705                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 706                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 707                                 $z = "</table>" . substr ( $x , 2);
 708                                 $l = array_pop ( $ltd ) ;
 709                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 710                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 711                                 array_pop ( $ltr ) ;
 712                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 713                         }
 714                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 715                                 $x = substr ( $x , 1 ) ;
 716                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 717                                 $z = '' ;
 718                                 $l = array_pop ( $ltd ) ;
 719                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 720                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 721                                 array_pop ( $ltr ) ;
 722                                 $t[$k] = $z ;
 723                                 array_push ( $tr , false ) ;
 724                                 array_push ( $td , false ) ;
 725                                 array_push ( $ltd , '' ) ;
 726                                 $attributes = $this->unstripForHTML( $x );
 727                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 728                         }
 729                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 730                                 # $x is a table row
 731                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 732                                         $fc = '+' ;
 733                                         $x = substr ( $x , 1 ) ;
 734                                 }
 735                                 $after = substr ( $x , 1 ) ;
 736                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 737                                 $after = explode ( '||' , $after ) ;
 738                                 $t[$k] = '' ;
 739
 740                                 # Loop through each table cell
 741                                 foreach ( $after AS $theline )
 742                                 {
 743                                         $z = '' ;
 744                                         if ( $fc != '+' )
 745                                         {
 746                                                 $tra = array_pop ( $ltr ) ;
 747                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 748                                                 array_push ( $tr , true ) ;
 749                                                 array_push ( $ltr , '' ) ;
 750                                         }
 751
 752                                         $l = array_pop ( $ltd ) ;
 753                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 754                                         if ( $fc == '|' ) $l = 'td' ;
 755                                         else if ( $fc == '!' ) $l = 'th' ;
 756                                         else if ( $fc == '+' ) $l = 'caption' ;
 757                                         else $l = '' ;
 758                                         array_push ( $ltd , $l ) ;
 759
 760                                         # Cell parameters
 761                                         $y = explode ( '|' , $theline , 2 ) ;
 762                                         # Note that a '|' inside an invalid link should not
 763                                         # be mistaken as delimiting cell parameters
 764                                         if ( strpos( $y[0], '[[' ) !== false ) {
 765                                                 $y = array ($theline);
 766                                         }
 767                                         if ( count ( $y ) == 1 )
 768                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 769                                         else {
 770                                                 $attributes = $this->unstripForHTML( $y[0] );
 771                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 772                                         }
 773                                         $t[$k] .= $y ;
 774                                         array_push ( $td , true ) ;
 775                                 }
 776                         }
 777                 }
 778
 779                 # Closing open td, tr && table
 780                 while ( count ( $td ) > 0 )
 781                 {
 782                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 783                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 784                         $t[] = '</table>' ;
 785                 }
 786
 787                 $t = implode ( "\n" , $t ) ;
 788                 wfProfileOut( $fname );
 789                 return $t ;
 790         }
 791
 792         /**
 793          * Helper function for parse() that transforms wiki markup into
 794          * HTML. Only called for $mOutputType == OT_HTML.
 795          *
 796          * @access private
 797          */
 798         function internalParse( $text ) {
 799                 global $wgContLang;
 800                 $args = array();
 801                 $isMain = true;
 802                 $fname = 'Parser::internalParse';
 803                 wfProfileIn( $fname );
 804
 805                 # Remove <noinclude> tags and <includeonly> sections
 806                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 807                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 808                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 809
 810                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 811                 $text = $this->replaceVariables( $text, $args );
 812
 813                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 814
 815                 $text = $this->doHeadings( $text );
 816                 if($this->mOptions->getUseDynamicDates()) {
 817                         $df =& DateFormatter::getInstance();
 818                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 819                 }
 820                 $text = $this->doAllQuotes( $text );
 821                 $text = $this->replaceInternalLinks( $text );
 822                 $text = $this->replaceExternalLinks( $text );
 823
 824                 # replaceInternalLinks may sometimes leave behind
 825                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 826                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 827
 828                 $text = $this->doMagicLinks( $text );
 829                 $text = $this->doTableStuff( $text );
 830                 $text = $this->formatHeadings( $text, $isMain );
 831
 832                 $regex = '/<!--IW_TRANSCLUDE (\d+)-->/';
 833                 $text = preg_replace_callback($regex, array(&$this, 'scarySubstitution'), $text);
 834
 835                 wfProfileOut( $fname );
 836                 return $text;
 837         }
 838
 839         function scarySubstitution($matches) {
 840 #               return "[[".$matches[0]."]]";
 841                 return $this->mIWTransData[(int)$matches[0]];
 842         }
 843
 844         /**
 845          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 846          * magic external links.
 847          *
 848          * @access private
 849          */
 850         function &doMagicLinks( &$text ) {
 851                 $text = $this->magicISBN( $text );
 852                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 853                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 854                 return $text;
 855         }
 856
 857         /**
 858          * Parse headers and return html
 859          *
 860          * @access private
 861          */
 862         function doHeadings( $text ) {
 863                 $fname = 'Parser::doHeadings';
 864                 wfProfileIn( $fname );
 865                 for ( $i = 6; $i >= 1; --$i ) {
 866                         $h = str_repeat( '=', $i );
 867                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 868                           "<h{$i}>\\1</h{$i}>\\2", $text );
 869                 }
 870                 wfProfileOut( $fname );
 871                 return $text;
 872         }
 873
 874         /**
 875          * Replace single quotes with HTML markup
 876          * @access private
 877          * @return string the altered text
 878          */
 879         function doAllQuotes( $text ) {
 880                 $fname = 'Parser::doAllQuotes';
 881                 wfProfileIn( $fname );
 882                 $outtext = '';
 883                 $lines = explode( "\n", $text );
 884                 foreach ( $lines as $line ) {
 885                         $outtext .= $this->doQuotes ( $line ) . "\n";
 886                 }
 887                 $outtext = substr($outtext, 0,-1);
 888                 wfProfileOut( $fname );
 889                 return $outtext;
 890         }
 891
 892         /**
 893          * Helper function for doAllQuotes()
 894          * @access private
 895          */
 896         function doQuotes( $text ) {
 897                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 898                 if ( count( $arr ) == 1 )
 899                         return $text;
 900                 else
 901                 {
 902                         # First, do some preliminary work. This may shift some apostrophes from
 903                         # being mark-up to being text. It also counts the number of occurrences
 904                         # of bold and italics mark-ups.
 905                         $i = 0;
 906                         $numbold = 0;
 907                         $numitalics = 0;
 908                         foreach ( $arr as $r )
 909                         {
 910                                 if ( ( $i % 2 ) == 1 )
 911                                 {
 912                                         # If there are ever four apostrophes, assume the first is supposed to
 913                                         # be text, and the remaining three constitute mark-up for bold text.
 914                                         if ( strlen( $arr[$i] ) == 4 )
 915                                         {
 916                                                 $arr[$i-1] .= "'";
 917                                                 $arr[$i] = "'''";
 918                                         }
 919                                         # If there are more than 5 apostrophes in a row, assume they're all
 920                                         # text except for the last 5.
 921                                         else if ( strlen( $arr[$i] ) > 5 )
 922                                         {
 923                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 924                                                 $arr[$i] = "'''''";
 925                                         }
 926                                         # Count the number of occurrences of bold and italics mark-ups.
 927                                         # We are not counting sequences of five apostrophes.
 928                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 929                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 930                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 931                                 }
 932                                 $i++;
 933                         }
 934
 935                         # If there is an odd number of both bold and italics, it is likely
 936                         # that one of the bold ones was meant to be an apostrophe followed
 937                         # by italics. Which one we cannot know for certain, but it is more
 938                         # likely to be one that has a single-letter word before it.
 939                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 940                         {
 941                                 $i = 0;
 942                                 $firstsingleletterword = -1;
 943                                 $firstmultiletterword = -1;
 944                                 $firstspace = -1;
 945                                 foreach ( $arr as $r )
 946                                 {
 947                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 948                                         {
 949                                                 $x1 = substr ($arr[$i-1], -1);
 950                                                 $x2 = substr ($arr[$i-1], -2, 1);
 951                                                 if ($x1 == ' ') {
 952                                                         if ($firstspace == -1) $firstspace = $i;
 953                                                 } else if ($x2 == ' ') {
 954                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 955                                                 } else {
 956                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 957                                                 }
 958                                         }
 959                                         $i++;
 960                                 }
 961
 962                                 # If there is a single-letter word, use it!
 963                                 if ($firstsingleletterword > -1)
 964                                 {
 965                                         $arr [ $firstsingleletterword ] = "''";
 966                                         $arr [ $firstsingleletterword-1 ] .= "'";
 967                                 }
 968                                 # If not, but there's a multi-letter word, use that one.
 969                                 else if ($firstmultiletterword > -1)
 970                                 {
 971                                         $arr [ $firstmultiletterword ] = "''";
 972                                         $arr [ $firstmultiletterword-1 ] .= "'";
 973                                 }
 974                                 # ... otherwise use the first one that has neither.
 975                                 # (notice that it is possible for all three to be -1 if, for example,
 976                                 # there is only one pentuple-apostrophe in the line)
 977                                 else if ($firstspace > -1)
 978                                 {
 979                                         $arr [ $firstspace ] = "''";
 980                                         $arr [ $firstspace-1 ] .= "'";
 981                                 }
 982                         }
 983
 984                         # Now let's actually convert our apostrophic mush to HTML!
 985                         $output = '';
 986                         $buffer = '';
 987                         $state = '';
 988                         $i = 0;
 989                         foreach ($arr as $r)
 990                         {
 991                                 if (($i % 2) == 0)
 992                                 {
 993                                         if ($state == 'both')
 994                                                 $buffer .= $r;
 995                                         else
 996                                                 $output .= $r;
 997                                 }
 998                                 else
 999                                 {
1000                                         if (strlen ($r) == 2)
1001                                         {
1002                                                 if ($state == 'i')
1003                                                 { $output .= '</i>'; $state = ''; }
1004                                                 else if ($state == 'bi')
1005                                                 { $output .= '</i>'; $state = 'b'; }
1006                                                 else if ($state == 'ib')
1007                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1008                                                 else if ($state == 'both')
1009                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1010                                                 else # $state can be 'b' or ''
1011                                                 { $output .= '<i>'; $state .= 'i'; }
1012                                         }
1013                                         else if (strlen ($r) == 3)
1014                                         {
1015                                                 if ($state == 'b')
1016                                                 { $output .= '</b>'; $state = ''; }
1017                                                 else if ($state == 'bi')
1018                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1019                                                 else if ($state == 'ib')
1020                                                 { $output .= '</b>'; $state = 'i'; }
1021                                                 else if ($state == 'both')
1022                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1023                                                 else # $state can be 'i' or ''
1024                                                 { $output .= '<b>'; $state .= 'b'; }
1025                                         }
1026                                         else if (strlen ($r) == 5)
1027                                         {
1028                                                 if ($state == 'b')
1029                                                 { $output .= '</b><i>'; $state = 'i'; }
1030                                                 else if ($state == 'i')
1031                                                 { $output .= '</i><b>'; $state = 'b'; }
1032                                                 else if ($state == 'bi')
1033                                                 { $output .= '</i></b>'; $state = ''; }
1034                                                 else if ($state == 'ib')
1035                                                 { $output .= '</b></i>'; $state = ''; }
1036                                                 else if ($state == 'both')
1037                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1038                                                 else # ($state == '')
1039                                                 { $buffer = ''; $state = 'both'; }
1040                                         }
1041                                 }
1042                                 $i++;
1043                         }
1044                         # Now close all remaining tags.  Notice that the order is important.
1045                         if ($state == 'b' || $state == 'ib')
1046                                 $output .= '</b>';
1047                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1048                                 $output .= '</i>';
1049                         if ($state == 'bi')
1050                                 $output .= '</b>';
1051                         if ($state == 'both')
1052                                 $output .= '<b><i>'.$buffer.'</i></b>';
1053                         return $output;
1054                 }
1055         }
1056
1057         /**
1058          * Replace external links
1059          *
1060          * Note: this is all very hackish and the order of execution matters a lot.
1061          * Make sure to run maintenance/parserTests.php if you change this code.
1062          *
1063          * @access private
1064          */
1065         function replaceExternalLinks( $text ) {
1066                 global $wgContLang;
1067                 $fname = 'Parser::replaceExternalLinks';
1068                 wfProfileIn( $fname );
1069
1070                 $sk =& $this->mOptions->getSkin();
1071
1072                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1073
1074                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1075
1076                 $i = 0;
1077                 while ( $i<count( $bits ) ) {
1078                         $url = $bits[$i++];
1079                         $protocol = $bits[$i++];
1080                         $text = $bits[$i++];
1081                         $trail = $bits[$i++];
1082
1083                         # The characters '<' and '>' (which were escaped by
1084                         # removeHTMLtags()) should not be included in
1085                         # URLs, per RFC 2396.
1086                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1087                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1088                                 $url = substr($url, 0, $m2[0][1]);
1089                         }
1090
1091                         # If the link text is an image URL, replace it with an <img> tag
1092                         # This happened by accident in the original parser, but some people used it extensively
1093                         $img = $this->maybeMakeExternalImage( $text );
1094                         if ( $img !== false ) {
1095                                 $text = $img;
1096                         }
1097
1098                         $dtrail = '';
1099
1100                         # Set linktype for CSS - if URL==text, link is essentially free
1101                         $linktype = ($text == $url) ? 'free' : 'text';
1102
1103                         # No link text, e.g. [http://domain.tld/some.link]
1104                         if ( $text == '' ) {
1105                                 # Autonumber if allowed
1106                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1107                                         $text = '[' . ++$this->mAutonumber . ']';
1108                                         $linktype = 'autonumber';
1109                                 } else {
1110                                         # Otherwise just use the URL
1111                                         $text = htmlspecialchars( $url );
1112                                         $linktype = 'free';
1113                                 }
1114                         } else {
1115                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1116                                 # Check for trail
1117                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1118                         }
1119
1120                         $text = $wgContLang->markNoConversion($text);
1121
1122                         # Replace &amp; from obsolete syntax with &.
1123                         # All HTML entities will be escaped by makeExternalLink()
1124                         $url = str_replace( '&amp;', '&', $url );
1125                         # Replace unnecessary URL escape codes with the referenced character
1126                         # This prevents spammers from hiding links from the filters
1127                         $url = Parser::replaceUnusualEscapes( $url );
1128
1129                         # Process the trail (i.e. everything after this link up until start of the next link),
1130                         # replacing any non-bracketed links
1131                         $trail = $this->replaceFreeExternalLinks( $trail );
1132
1133                         # Use the encoded URL
1134                         # This means that users can paste URLs directly into the text
1135                         # Funny characters like &ouml; aren't valid in URLs anyway
1136                         # This was changed in August 2004
1137                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1138
1139                         # Register link in the output object
1140                         $this->mOutput->addExternalLink( $url );
1141                 }
1142
1143                 wfProfileOut( $fname );
1144                 return $s;
1145         }
1146
1147         /**
1148          * Replace anything that looks like a URL with a link
1149          * @access private
1150          */
1151         function replaceFreeExternalLinks( $text ) {
1152                 global $wgContLang;
1153                 $fname = 'Parser::replaceFreeExternalLinks';
1154                 wfProfileIn( $fname );
1155
1156                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1157                 $s = array_shift( $bits );
1158                 $i = 0;
1159
1160                 $sk =& $this->mOptions->getSkin();
1161
1162                 while ( $i < count( $bits ) ){
1163                         $protocol = $bits[$i++];
1164                         $remainder = $bits[$i++];
1165
1166                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1167                                 # Found some characters after the protocol that look promising
1168                                 $url = $protocol . $m[1];
1169                                 $trail = $m[2];
1170
1171                                 # The characters '<' and '>' (which were escaped by
1172                                 # removeHTMLtags()) should not be included in
1173                                 # URLs, per RFC 2396.
1174                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1175                                         $trail = substr($url, $m2[0][1]) . $trail;
1176                                         $url = substr($url, 0, $m2[0][1]);
1177                                 }
1178
1179                                 # Move trailing punctuation to $trail
1180                                 $sep = ',;\.:!?';
1181                                 # If there is no left bracket, then consider right brackets fair game too
1182                                 if ( strpos( $url, '(' ) === false ) {
1183                                         $sep .= ')';
1184                                 }
1185
1186                                 $numSepChars = strspn( strrev( $url ), $sep );
1187                                 if ( $numSepChars ) {
1188                                         $trail = substr( $url, -$numSepChars ) . $trail;
1189                                         $url = substr( $url, 0, -$numSepChars );
1190                                 }
1191
1192                                 # Replace &amp; from obsolete syntax with &.
1193                                 # All HTML entities will be escaped by makeExternalLink()
1194                                 # or maybeMakeExternalImage()
1195                                 $url = str_replace( '&amp;', '&', $url );
1196                                 # Replace unnecessary URL escape codes with their equivalent characters
1197                                 $url = Parser::replaceUnusualEscapes( $url );
1198
1199                                 # Is this an external image?
1200                                 $text = $this->maybeMakeExternalImage( $url );
1201                                 if ( $text === false ) {
1202                                         # Not an image, make a link
1203                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1204                                         # Register it in the output object
1205                                         $this->mOutput->addExternalLink( $url );
1206                                 }
1207                                 $s .= $text . $trail;
1208                         } else {
1209                                 $s .= $protocol . $remainder;
1210                         }
1211                 }
1212                 wfProfileOut( $fname );
1213                 return $s;
1214         }
1215
1216         /**
1217          * Replace unusual URL escape codes with their equivalent characters
1218          * @param string
1219          * @return string
1220          * @static
1221          */
1222         function replaceUnusualEscapes( $url ) {
1223                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1224                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1225         }
1226
1227         /**
1228          * Callback function used in replaceUnusualEscapes().
1229          * Replaces unusual URL escape codes with their equivalent character
1230          * @static
1231          * @access private
1232          */
1233         function replaceUnusualEscapesCallback( $matches ) {
1234                 $char = urldecode( $matches[0] );
1235                 $ord = ord( $char );
1236                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1237                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1238                         // No, shouldn't be escaped
1239                         return $char;
1240                 } else {
1241                         // Yes, leave it escaped
1242                         return $matches[0];
1243                 }
1244         }
1245
1246         /**
1247          * make an image if it's allowed, either through the global
1248          * option or through the exception
1249          * @access private
1250          */
1251         function maybeMakeExternalImage( $url ) {
1252                 $sk =& $this->mOptions->getSkin();
1253                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1254                 $imagesexception = !empty($imagesfrom);
1255                 $text = false;
1256                 if ( $this->mOptions->getAllowExternalImages()
1257                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1258                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1259                                 # Image found
1260                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1261                         }
1262                 }
1263                 return $text;
1264         }
1265
1266         /**
1267          * Process [[ ]] wikilinks
1268          *
1269          * @access private
1270          */
1271         function replaceInternalLinks( $s ) {
1272                 global $wgContLang;
1273                 static $fname = 'Parser::replaceInternalLinks' ;
1274
1275                 wfProfileIn( $fname );
1276
1277                 wfProfileIn( $fname.'-setup' );
1278                 static $tc = FALSE;
1279                 # the % is needed to support urlencoded titles as well
1280                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1281
1282                 $sk =& $this->mOptions->getSkin();
1283
1284                 #split the entire text string on occurences of [[
1285                 $a = explode( '[[', ' ' . $s );
1286                 #get the first element (all text up to first [[), and remove the space we added
1287                 $s = array_shift( $a );
1288                 $s = substr( $s, 1 );
1289
1290                 # Match a link having the form [[namespace:link|alternate]]trail
1291                 static $e1 = FALSE;
1292                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1293                 # Match cases where there is no "]]", which might still be images
1294                 static $e1_img = FALSE;
1295                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1296                 # Match the end of a line for a word that's not followed by whitespace,
1297                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1298                 $e2 = wfMsgForContent( 'linkprefix' );
1299
1300                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1301
1302                 if( is_null( $this->mTitle ) ) {
1303                         wfDebugDieBacktrace( 'nooo' );
1304                 }
1305                 $nottalk = !$this->mTitle->isTalkPage();
1306
1307                 if ( $useLinkPrefixExtension ) {
1308                         if ( preg_match( $e2, $s, $m ) ) {
1309                                 $first_prefix = $m[2];
1310                         } else {
1311                                 $first_prefix = false;
1312                         }
1313                 } else {
1314                         $prefix = '';
1315                 }
1316
1317                 $selflink = $this->mTitle->getPrefixedText();
1318                 wfProfileOut( $fname.'-setup' );
1319
1320                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1321                 $useSubpages = $this->areSubpagesAllowed();
1322
1323                 # Loop for each link
1324                 for ($k = 0; isset( $a[$k] ); $k++) {
1325                         $line = $a[$k];
1326                         if ( $useLinkPrefixExtension ) {
1327                                 wfProfileIn( $fname.'-prefixhandling' );
1328                                 if ( preg_match( $e2, $s, $m ) ) {
1329                                         $prefix = $m[2];
1330                                         $s = $m[1];
1331                                 } else {
1332                                         $prefix='';
1333                                 }
1334                                 # first link
1335                                 if($first_prefix) {
1336                                         $prefix = $first_prefix;
1337                                         $first_prefix = false;
1338                                 }
1339                                 wfProfileOut( $fname.'-prefixhandling' );
1340                         }
1341
1342                         $might_be_img = false;
1343
1344                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1345                                 $text = $m[2];
1346                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1347                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1348                                 # the real problem is with the $e1 regex
1349                                 # See bug 1300.
1350                                 #
1351                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1352                                 # and no image is in sight. See bug 2095.
1353                                 #
1354                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1355                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1356                                         $m[3] = $n[1];
1357                                 }
1358                                 # fix up urlencoded title texts
1359                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1360                                 $trail = $m[3];
1361                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1362                                 $might_be_img = true;
1363                                 $text = $m[2];
1364                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1365                                 $trail = "";
1366                         } else { # Invalid form; output directly
1367                                 $s .= $prefix . '[[' . $line ;
1368                                 continue;
1369                         }
1370
1371                         # Don't allow internal links to pages containing
1372                         # PROTO: where PROTO is a valid URL protocol; these
1373                         # should be external links.
1374                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1375                                 $s .= $prefix . '[[' . $line ;
1376                                 continue;
1377                         }
1378
1379                         # Make subpage if necessary
1380                         if( $useSubpages ) {
1381                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1382                         } else {
1383                                 $link = $m[1];
1384                         }
1385
1386                         $noforce = (substr($m[1], 0, 1) != ':');
1387                         if (!$noforce) {
1388                                 # Strip off leading ':'
1389                                 $link = substr($link, 1);
1390                         }
1391
1392                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1393                         if( !$nt ) {
1394                                 $s .= $prefix . '[[' . $line;
1395                                 continue;
1396                         }
1397
1398                         #check other language variants of the link
1399                         #if the article does not exist
1400                         if( $checkVariantLink
1401                             && $nt->getArticleID() == 0 ) {
1402                                 $wgContLang->findVariantLink($link, $nt);
1403                         }
1404
1405                         $ns = $nt->getNamespace();
1406                         $iw = $nt->getInterWiki();
1407
1408                         if ($might_be_img) { # if this is actually an invalid link
1409                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1410                                         $found = false;
1411                                         while (isset ($a[$k+1]) ) {
1412                                                 #look at the next 'line' to see if we can close it there
1413                                                 $spliced = array_splice( $a, $k + 1, 1 );
1414                                                 $next_line = array_shift( $spliced );
1415                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1416                                                 # the first ]] closes the inner link, the second the image
1417                                                         $found = true;
1418                                                         $text .= '[[' . $m[1];
1419                                                         $trail = $m[2];
1420                                                         break;
1421                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1422                                                         #if there's exactly one ]] that's fine, we'll keep looking
1423                                                         $text .= '[[' . $m[0];
1424                                                 } else {
1425                                                         #if $next_line is invalid too, we need look no further
1426                                                         $text .= '[[' . $next_line;
1427                                                         break;
1428                                                 }
1429                                         }
1430                                         if ( !$found ) {
1431                                                 # we couldn't find the end of this imageLink, so output it raw
1432                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1433                                                 $text = $this->replaceInternalLinks($text);
1434                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1435                                                 # note: no $trail, because without an end, there *is* no trail
1436                                                 continue;
1437                                         }
1438                                 } else { #it's not an image, so output it raw
1439                                         $s .= $prefix . '[[' . $link . '|' . $text;
1440                                         # note: no $trail, because without an end, there *is* no trail
1441                                         continue;
1442                                 }
1443                         }
1444
1445                         $wasblank = ( '' == $text );
1446                         if( $wasblank ) $text = $link;
1447
1448
1449                         # Link not escaped by : , create the various objects
1450                         if( $noforce ) {
1451
1452                                 # Interwikis
1453                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1454                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1455                                         $s = rtrim($s . "\n");
1456                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1457                                         continue;
1458                                 }
1459
1460                                 if ( $ns == NS_IMAGE ) {
1461                                         wfProfileIn( "$fname-image" );
1462                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1463                                                 # recursively parse links inside the image caption
1464                                                 # actually, this will parse them in any other parameters, too,
1465                                                 # but it might be hard to fix that, and it doesn't matter ATM
1466                                                 $text = $this->replaceExternalLinks($text);
1467                                                 $text = $this->replaceInternalLinks($text);
1468
1469                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1470                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1471                                                 $this->mOutput->addImage( $nt->getDBkey() );
1472
1473                                                 wfProfileOut( "$fname-image" );
1474                                                 continue;
1475                                         }
1476                                         wfProfileOut( "$fname-image" );
1477
1478                                 }
1479
1480                                 if ( $ns == NS_CATEGORY ) {
1481                                         wfProfileIn( "$fname-category" );
1482                                         $s = rtrim($s . "\n"); # bug 87
1483
1484                                         if ( $wasblank ) {
1485                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1486                                                         $sortkey = $this->mTitle->getText();
1487                                                 } else {
1488                                                         $sortkey = $this->mTitle->getPrefixedText();
1489                                                 }
1490                                         } else {
1491                                                 $sortkey = $text;
1492                                         }
1493                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1494                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1495                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1496
1497                                         /**
1498                                          * Strip the whitespace Category links produce, see bug 87
1499                                          * @todo We might want to use trim($tmp, "\n") here.
1500                                          */
1501                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1502
1503                                         wfProfileOut( "$fname-category" );
1504                                         continue;
1505                                 }
1506                         }
1507
1508                         if( ( $nt->getPrefixedText() === $selflink ) &&
1509                             ( $nt->getFragment() === '' ) ) {
1510                                 # Self-links are handled specially; generally de-link and change to bold.
1511                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1512                                 continue;
1513                         }
1514
1515                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1516                         if( $ns == NS_MEDIA ) {
1517                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1518                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1519                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1520                                 $this->mOutput->addImage( $nt->getDBkey() );
1521                                 continue;
1522                         } elseif( $ns == NS_SPECIAL ) {
1523                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1524                                 continue;
1525                         } elseif( $ns == NS_IMAGE ) {
1526                                 $img = Image::newFromTitle( $nt );
1527                                 if( $img->exists() ) {
1528                                         // Force a blue link if the file exists; may be a remote
1529                                         // upload on the shared repository, and we want to see its
1530                                         // auto-generated page.
1531                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1532                                         continue;
1533                                 }
1534                         }
1535                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1536                 }
1537                 wfProfileOut( $fname );
1538                 return $s;
1539         }
1540
1541         /**
1542          * Make a link placeholder. The text returned can be later resolved to a real link with
1543          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1544          * parsing of interwiki links, and secondly to allow all extistence checks and
1545          * article length checks (for stub links) to be bundled into a single query.
1546          *
1547          */
1548         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1549                 if ( ! is_object($nt) ) {
1550                         # Fail gracefully
1551                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1552                 } else {
1553                         # Separate the link trail from the rest of the link
1554                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1555
1556                         if ( $nt->isExternal() ) {
1557                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1558                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1559                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1560                         } else {
1561                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1562                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1563                                 $this->mLinkHolders['queries'][] = $query;
1564                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1565                                 $this->mLinkHolders['titles'][] = $nt;
1566
1567                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1568                         }
1569                 }
1570                 return $retVal;
1571         }
1572
1573         /**
1574          * Render a forced-blue link inline; protect against double expansion of
1575          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1576          * Since this little disaster has to split off the trail text to avoid
1577          * breaking URLs in the following text without breaking trails on the
1578          * wiki links, it's been made into a horrible function.
1579          *
1580          * @param Title $nt
1581          * @param string $text
1582          * @param string $query
1583          * @param string $trail
1584          * @param string $prefix
1585          * @return string HTML-wikitext mix oh yuck
1586          */
1587         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1588                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1589                 $sk =& $this->mOptions->getSkin();
1590                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1591                 return $this->armorLinks( $link ) . $trail;
1592         }
1593
1594         /**
1595          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1596          * going to go through further parsing steps before inline URL expansion.
1597          *
1598          * In particular this is important when using action=render, which causes
1599          * full URLs to be included.
1600          *
1601          * Oh man I hate our multi-layer parser!
1602          *
1603          * @param string more-or-less HTML
1604          * @return string less-or-more HTML with NOPARSE bits
1605          */
1606         function armorLinks( $text ) {
1607                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1608                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1609         }
1610
1611         /**
1612          * Return true if subpage links should be expanded on this page.
1613          * @return bool
1614          */
1615         function areSubpagesAllowed() {
1616                 # Some namespaces don't allow subpages
1617                 global $wgNamespacesWithSubpages;
1618                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1619         }
1620
1621         /**
1622          * Handle link to subpage if necessary
1623          * @param string $target the source of the link
1624          * @param string &$text the link text, modified as necessary
1625          * @return string the full name of the link
1626          * @access private
1627          */
1628         function maybeDoSubpageLink($target, &$text) {
1629                 # Valid link forms:
1630                 # Foobar -- normal
1631                 # :Foobar -- override special treatment of prefix (images, language links)
1632                 # /Foobar -- convert to CurrentPage/Foobar
1633                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1634                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1635                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1636
1637                 $fname = 'Parser::maybeDoSubpageLink';
1638                 wfProfileIn( $fname );
1639                 $ret = $target; # default return value is no change
1640
1641                 # Some namespaces don't allow subpages,
1642                 # so only perform processing if subpages are allowed
1643                 if( $this->areSubpagesAllowed() ) {
1644                         # Look at the first character
1645                         if( $target != '' && $target{0} == '/' ) {
1646                                 # / at end means we don't want the slash to be shown
1647                                 if( substr( $target, -1, 1 ) == '/' ) {
1648                                         $target = substr( $target, 1, -1 );
1649                                         $noslash = $target;
1650                                 } else {
1651                                         $noslash = substr( $target, 1 );
1652                                 }
1653
1654                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1655                                 if( '' === $text ) {
1656                                         $text = $target;
1657                                 } # this might be changed for ugliness reasons
1658                         } else {
1659                                 # check for .. subpage backlinks
1660                                 $dotdotcount = 0;
1661                                 $nodotdot = $target;
1662                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1663                                         ++$dotdotcount;
1664                                         $nodotdot = substr( $nodotdot, 3 );
1665                                 }
1666                                 if($dotdotcount > 0) {
1667                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1668                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1669                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1670                                                 # / at the end means don't show full path
1671                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1672                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1673                                                         if( '' === $text ) {
1674                                                                 $text = $nodotdot;
1675                                                         }
1676                                                 }
1677                                                 $nodotdot = trim( $nodotdot );
1678                                                 if( $nodotdot != '' ) {
1679                                                         $ret .= '/' . $nodotdot;
1680                                                 }
1681                                         }
1682                                 }
1683                         }
1684                 }
1685
1686                 wfProfileOut( $fname );
1687                 return $ret;
1688         }
1689
1690         /**#@+
1691          * Used by doBlockLevels()
1692          * @access private
1693          */
1694         /* private */ function closeParagraph() {
1695                 $result = '';
1696                 if ( '' != $this->mLastSection ) {
1697                         $result = '</' . $this->mLastSection  . ">\n";
1698                 }
1699                 $this->mInPre = false;
1700                 $this->mLastSection = '';
1701                 return $result;
1702         }
1703         # getCommon() returns the length of the longest common substring
1704         # of both arguments, starting at the beginning of both.
1705         #
1706         /* private */ function getCommon( $st1, $st2 ) {
1707                 $fl = strlen( $st1 );
1708                 $shorter = strlen( $st2 );
1709                 if ( $fl < $shorter ) { $shorter = $fl; }
1710
1711                 for ( $i = 0; $i < $shorter; ++$i ) {
1712                         if ( $st1{$i} != $st2{$i} ) { break; }
1713                 }
1714                 return $i;
1715         }
1716         # These next three functions open, continue, and close the list
1717         # element appropriate to the prefix character passed into them.
1718         #
1719         /* private */ function openList( $char ) {
1720                 $result = $this->closeParagraph();
1721
1722                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1723                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1724                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1725                 else if ( ';' == $char ) {
1726                         $result .= '<dl><dt>';
1727                         $this->mDTopen = true;
1728                 }
1729                 else { $result = '<!-- ERR 1 -->'; }
1730
1731                 return $result;
1732         }
1733
1734         /* private */ function nextItem( $char ) {
1735                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1736                 else if ( ':' == $char || ';' == $char ) {
1737                         $close = '</dd>';
1738                         if ( $this->mDTopen ) { $close = '</dt>'; }
1739                         if ( ';' == $char ) {
1740                                 $this->mDTopen = true;
1741                                 return $close . '<dt>';
1742                         } else {
1743                                 $this->mDTopen = false;
1744                                 return $close . '<dd>';
1745                         }
1746                 }
1747                 return '<!-- ERR 2 -->';
1748         }
1749
1750         /* private */ function closeList( $char ) {
1751                 if ( '*' == $char ) { $text = '</li></ul>'; }
1752                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1753                 else if ( ':' == $char ) {
1754                         if ( $this->mDTopen ) {
1755                                 $this->mDTopen = false;
1756                                 $text = '</dt></dl>';
1757                         } else {
1758                                 $text = '</dd></dl>';
1759                         }
1760                 }
1761                 else {  return '<!-- ERR 3 -->'; }
1762                 return $text."\n";
1763         }
1764         /**#@-*/
1765
1766         /**
1767          * Make lists from lines starting with ':', '*', '#', etc.
1768          *
1769          * @access private
1770          * @return string the lists rendered as HTML
1771          */
1772         function doBlockLevels( $text, $linestart ) {
1773                 $fname = 'Parser::doBlockLevels';
1774                 wfProfileIn( $fname );
1775
1776                 # Parsing through the text line by line.  The main thing
1777                 # happening here is handling of block-level elements p, pre,
1778                 # and making lists from lines starting with * # : etc.
1779                 #
1780                 $textLines = explode( "\n", $text );
1781
1782                 $lastPrefix = $output = '';
1783                 $this->mDTopen = $inBlockElem = false;
1784                 $prefixLength = 0;
1785                 $paragraphStack = false;
1786
1787                 if ( !$linestart ) {
1788                         $output .= array_shift( $textLines );
1789                 }
1790                 foreach ( $textLines as $oLine ) {
1791                         $lastPrefixLength = strlen( $lastPrefix );
1792                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1793                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1794                         if ( !$this->mInPre ) {
1795                                 # Multiple prefixes may abut each other for nested lists.
1796                                 $prefixLength = strspn( $oLine, '*#:;' );
1797                                 $pref = substr( $oLine, 0, $prefixLength );
1798
1799                                 # eh?
1800                                 $pref2 = str_replace( ';', ':', $pref );
1801                                 $t = substr( $oLine, $prefixLength );
1802                                 $this->mInPre = !empty($preOpenMatch);
1803                         } else {
1804                                 # Don't interpret any other prefixes in preformatted text
1805                                 $prefixLength = 0;
1806                                 $pref = $pref2 = '';
1807                                 $t = $oLine;
1808                         }
1809
1810                         # List generation
1811                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1812                                 # Same as the last item, so no need to deal with nesting or opening stuff
1813                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1814                                 $paragraphStack = false;
1815
1816                                 if ( substr( $pref, -1 ) == ';') {
1817                                         # The one nasty exception: definition lists work like this:
1818                                         # ; title : definition text
1819                                         # So we check for : in the remainder text to split up the
1820                                         # title and definition, without b0rking links.
1821                                         $term = $t2 = '';
1822                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1823                                                 $t = $t2;
1824                                                 $output .= $term . $this->nextItem( ':' );
1825                                         }
1826                                 }
1827                         } elseif( $prefixLength || $lastPrefixLength ) {
1828                                 # Either open or close a level...
1829                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1830                                 $paragraphStack = false;
1831
1832                                 while( $commonPrefixLength < $lastPrefixLength ) {
1833                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1834                                         --$lastPrefixLength;
1835                                 }
1836                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1837                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1838                                 }
1839                                 while ( $prefixLength > $commonPrefixLength ) {
1840                                         $char = substr( $pref, $commonPrefixLength, 1 );
1841                                         $output .= $this->openList( $char );
1842
1843                                         if ( ';' == $char ) {
1844                                                 # FIXME: This is dupe of code above
1845                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1846                                                         $t = $t2;
1847                                                         $output .= $term . $this->nextItem( ':' );
1848                                                 }
1849                                         }
1850                                         ++$commonPrefixLength;
1851                                 }
1852                                 $lastPrefix = $pref2;
1853                         }
1854                         if( 0 == $prefixLength ) {
1855                                 wfProfileIn( "$fname-paragraph" );
1856                                 # No prefix (not in list)--go to paragraph mode
1857                                 // XXX: use a stack for nestable elements like span, table and div
1858                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1859                                 $closematch = preg_match(
1860                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1861                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1862                                 if ( $openmatch or $closematch ) {
1863                                         $paragraphStack = false;
1864                                         $output .= $this->closeParagraph();
1865                                         if ( $preOpenMatch and !$preCloseMatch ) {
1866                                                 $this->mInPre = true;
1867                                         }
1868                                         if ( $closematch ) {
1869                                                 $inBlockElem = false;
1870                                         } else {
1871                                                 $inBlockElem = true;
1872                                         }
1873                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1874                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1875                                                 // pre
1876                                                 if ($this->mLastSection != 'pre') {
1877                                                         $paragraphStack = false;
1878                                                         $output .= $this->closeParagraph().'<pre>';
1879                                                         $this->mLastSection = 'pre';
1880                                                 }
1881                                                 $t = substr( $t, 1 );
1882                                         } else {
1883                                                 // paragraph
1884                                                 if ( '' == trim($t) ) {
1885                                                         if ( $paragraphStack ) {
1886                                                                 $output .= $paragraphStack.'<br />';
1887                                                                 $paragraphStack = false;
1888                                                                 $this->mLastSection = 'p';
1889                                                         } else {
1890                                                                 if ($this->mLastSection != 'p' ) {
1891                                                                         $output .= $this->closeParagraph();
1892                                                                         $this->mLastSection = '';
1893                                                                         $paragraphStack = '<p>';
1894                                                                 } else {
1895                                                                         $paragraphStack = '</p><p>';
1896                                                                 }
1897                                                         }
1898                                                 } else {
1899                                                         if ( $paragraphStack ) {
1900                                                                 $output .= $paragraphStack;
1901                                                                 $paragraphStack = false;
1902                                                                 $this->mLastSection = 'p';
1903                                                         } else if ($this->mLastSection != 'p') {
1904                                                                 $output .= $this->closeParagraph().'<p>';
1905                                                                 $this->mLastSection = 'p';
1906                                                         }
1907                                                 }
1908                                         }
1909                                 }
1910                                 wfProfileOut( "$fname-paragraph" );
1911                         }
1912                         // somewhere above we forget to get out of pre block (bug 785)
1913                         if($preCloseMatch && $this->mInPre) {
1914                                 $this->mInPre = false;
1915                         }
1916                         if ($paragraphStack === false) {
1917                                 $output .= $t."\n";
1918                         }
1919                 }
1920                 while ( $prefixLength ) {
1921                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1922                         --$prefixLength;
1923                 }
1924                 if ( '' != $this->mLastSection ) {
1925                         $output .= '</' . $this->mLastSection . '>';
1926                         $this->mLastSection = '';
1927                 }
1928
1929                 wfProfileOut( $fname );
1930                 return $output;
1931         }
1932
1933         /**
1934          * Split up a string on ':', ignoring any occurences inside
1935          * <a>..</a> or <span>...</span>
1936          * @param string $str the string to split
1937          * @param string &$before set to everything before the ':'
1938          * @param string &$after set to everything after the ':'
1939          * return string the position of the ':', or false if none found
1940          */
1941         function findColonNoLinks($str, &$before, &$after) {
1942                 # I wonder if we should make this count all tags, not just <a>
1943                 # and <span>. That would prevent us from matching a ':' that
1944                 # comes in the middle of italics other such formatting....
1945                 # -- Wil
1946                 $fname = 'Parser::findColonNoLinks';
1947                 wfProfileIn( $fname );
1948                 $pos = 0;
1949                 do {
1950                         $colon = strpos($str, ':', $pos);
1951
1952                         if ($colon !== false) {
1953                                 $before = substr($str, 0, $colon);
1954                                 $after = substr($str, $colon + 1);
1955
1956                                 # Skip any ':' within <a> or <span> pairs
1957                                 $a = substr_count($before, '<a');
1958                                 $s = substr_count($before, '<span');
1959                                 $ca = substr_count($before, '</a>');
1960                                 $cs = substr_count($before, '</span>');
1961
1962                                 if ($a <= $ca and $s <= $cs) {
1963                                         # Tags are balanced before ':'; ok
1964                                         break;
1965                                 }
1966                                 $pos = $colon + 1;
1967                         }
1968                 } while ($colon !== false);
1969                 wfProfileOut( $fname );
1970                 return $colon;
1971         }
1972
1973         /**
1974          * Return value of a magic variable (like PAGENAME)
1975          *
1976          * @access private
1977          */
1978         function getVariableValue( $index ) {
1979                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1980
1981                 /**
1982                  * Some of these require message or data lookups and can be
1983                  * expensive to check many times.
1984                  */
1985                 static $varCache = array();
1986                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1987                         if ( isset( $varCache[$index] ) )
1988                                 return $varCache[$index];
1989
1990                 $ts = time();
1991                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
1992
1993                 switch ( $index ) {
1994                         case MAG_CURRENTMONTH:
1995                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
1996                         case MAG_CURRENTMONTHNAME:
1997                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
1998                         case MAG_CURRENTMONTHNAMEGEN:
1999                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2000                         case MAG_CURRENTMONTHABBREV:
2001                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2002                         case MAG_CURRENTDAY:
2003                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2004                         case MAG_CURRENTDAY2:
2005                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2006                         case MAG_PAGENAME:
2007                                 return $this->mTitle->getText();
2008                         case MAG_PAGENAMEE:
2009                                 return $this->mTitle->getPartialURL();
2010                         case MAG_FULLPAGENAME:
2011                                 return $this->mTitle->getPrefixedText();
2012                         case MAG_FULLPAGENAMEE:
2013                                 return $this->mTitle->getPrefixedURL();
2014                         case MAG_REVISIONID:
2015                                 return $this->mRevisionId;
2016                         case MAG_NAMESPACE:
2017                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2018                         case MAG_NAMESPACEE:
2019                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2020                         case MAG_CURRENTDAYNAME:
2021                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2022                         case MAG_CURRENTYEAR:
2023                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2024                         case MAG_CURRENTTIME:
2025                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2026                         case MAG_CURRENTWEEK:
2027                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2028                                 // int to remove the padding
2029                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2030                         case MAG_CURRENTDOW:
2031                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2032                         case MAG_NUMBEROFARTICLES:
2033                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2034                         case MAG_NUMBEROFFILES:
2035                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2036                         case MAG_SITENAME:
2037                                 return $wgSitename;
2038                         case MAG_SERVER:
2039                                 return $wgServer;
2040                         case MAG_SERVERNAME:
2041                                 return $wgServerName;
2042                         case MAG_SCRIPTPATH:
2043                                 return $wgScriptPath;
2044                         default:
2045                                 $ret = null;
2046                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2047                                         return $ret;
2048                                 else
2049                                         return null;
2050                 }
2051         }
2052
2053         /**
2054          * initialise the magic variables (like CURRENTMONTHNAME)
2055          *
2056          * @access private
2057          */
2058         function initialiseVariables() {
2059                 $fname = 'Parser::initialiseVariables';
2060                 wfProfileIn( $fname );
2061                 global $wgVariableIDs;
2062                 $this->mVariables = array();
2063                 foreach ( $wgVariableIDs as $id ) {
2064                         $mw =& MagicWord::get( $id );
2065                         $mw->addToArray( $this->mVariables, $id );
2066                 }
2067                 wfProfileOut( $fname );
2068         }
2069
2070         /**
2071          * parse any parentheses in format ((title|part|part))
2072          * and call callbacks to get a replacement text for any found piece
2073          *
2074          * @param string $text The text to parse
2075          * @param array $callbacks rules in form:
2076          *     '{' => array(                            # opening parentheses
2077          *                                      'end' => '}',   # closing parentheses
2078          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2079          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2080          *                                                                )
2081          *                                      )
2082          * @access private
2083          */
2084         function replace_callback ($text, $callbacks) {
2085                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2086                 $lastOpeningBrace = -1;                 # last not closed parentheses
2087
2088                 for ($i = 0; $i < strlen($text); $i++) {
2089                         # check for any opening brace
2090                         $rule = null;
2091                         $nextPos = -1;
2092                         foreach ($callbacks as $key => $value) {
2093                                 $pos = strpos ($text, $key, $i);
2094                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2095                                         $rule = $value;
2096                                         $nextPos = $pos;
2097                                 }
2098                         }
2099
2100                         if ($lastOpeningBrace >= 0) {
2101                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2102
2103                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2104                                         $rule = null;
2105                                         $nextPos = $pos;
2106                                 }
2107
2108                                 $pos = strpos ($text, '|', $i);
2109
2110                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2111                                         $rule = null;
2112                                         $nextPos = $pos;
2113                                 }
2114                         }
2115
2116                         if ($nextPos == -1)
2117                                 break;
2118
2119                         $i = $nextPos;
2120
2121                         # found openning brace, lets add it to parentheses stack
2122                         if (null != $rule) {
2123                                 $piece = array('brace' => $text[$i],
2124                                                            'braceEnd' => $rule['end'],
2125                                                            'count' => 1,
2126                                                            'title' => '',
2127                                                            'parts' => null);
2128
2129                                 # count openning brace characters
2130                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2131                                         $piece['count']++;
2132                                         $i++;
2133                                 }
2134
2135                                 $piece['startAt'] = $i+1;
2136                                 $piece['partStart'] = $i+1;
2137
2138                                 # we need to add to stack only if openning brace count is enough for any given rule
2139                                 foreach ($rule['cb'] as $cnt => $fn) {
2140                                         if ($piece['count'] >= $cnt) {
2141                                                 $lastOpeningBrace ++;
2142                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2143                                                 break;
2144                                         }
2145                                 }
2146
2147                                 continue;
2148                         }
2149                         else if ($lastOpeningBrace >= 0) {
2150                                 # first check if it is a closing brace
2151                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2152                                         # lets check if it is enough characters for closing brace
2153                                         $count = 1;
2154                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2155                                                 $count++;
2156
2157                                         # if there are more closing parentheses than opening ones, we parse less
2158                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2159                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2160
2161                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2162                                         $matchingCount = 0;
2163                                         $matchingCallback = null;
2164                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2165                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2166                                                         $matchingCount = $cnt;
2167                                                         $matchingCallback = $fn;
2168                                                 }
2169                                         }
2170
2171                                         if ($matchingCount == 0) {
2172                                                 $i += $count - 1;
2173                                                 continue;
2174                                         }
2175
2176                                         # lets set a title or last part (if '|' was found)
2177                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2178                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2179                                         else
2180                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2181
2182                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2183                                         $pieceEnd = $i + $matchingCount;
2184
2185                                         if( is_callable( $matchingCallback ) ) {
2186                                                 $cbArgs = array (
2187                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2188                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2189                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2190                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2191                                                                                  );
2192                                                 # finally we can call a user callback and replace piece of text
2193                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2194                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2195                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2196                                         }
2197                                         else {
2198                                                 # null value for callback means that parentheses should be parsed, but not replaced
2199                                                 $i += $matchingCount - 1;
2200                                         }
2201
2202                                         # reset last openning parentheses, but keep it in case there are unused characters
2203                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2204                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2205                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2206                                                                    'title' => '',
2207                                                                    'parts' => null,
2208                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2209                                         $openingBraceStack[$lastOpeningBrace--] = null;
2210
2211                                         if ($matchingCount < $piece['count']) {
2212                                                 $piece['count'] -= $matchingCount;
2213                                                 $piece['startAt'] -= $matchingCount;
2214                                                 $piece['partStart'] = $piece['startAt'];
2215                                                 # do we still qualify for any callback with remaining count?
2216                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2217                                                         if ($piece['count'] >= $cnt) {
2218                                                                 $lastOpeningBrace ++;
2219                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2220                                                                 break;
2221                                                         }
2222                                                 }
2223                                         }
2224                                         continue;
2225                                 }
2226
2227                                 # lets set a title if it is a first separator, or next part otherwise
2228                                 if ($text[$i] == '|') {
2229                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2230                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2231                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2232                                         }
2233                                         else
2234                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2235
2236                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2237                                 }
2238                         }
2239                 }
2240
2241                 return $text;
2242         }
2243
2244         /**
2245          * Replace magic variables, templates, and template arguments
2246          * with the appropriate text. Templates are substituted recursively,
2247          * taking care to avoid infinite loops.
2248          *
2249          * Note that the substitution depends on value of $mOutputType:
2250          *  OT_WIKI: only {{subst:}} templates
2251          *  OT_MSG: only magic variables
2252          *  OT_HTML: all templates and magic variables
2253          *
2254          * @param string $tex The text to transform
2255          * @param array $args Key-value pairs representing template parameters to substitute
2256          * @access private
2257          */
2258         function replaceVariables( $text, $args = array() ) {
2259                 # Prevent too big inclusions
2260                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2261                         return $text;
2262                 }
2263
2264                 $fname = 'Parser::replaceVariables';
2265                 wfProfileIn( $fname );
2266
2267                 # This function is called recursively. To keep track of arguments we need a stack:
2268                 array_push( $this->mArgStack, $args );
2269
2270                 $braceCallbacks = array();
2271                 $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2272                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2273                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2274                 }
2275                 $callbacks = array();
2276                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2277                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2278                 $text = $this->replace_callback ($text, $callbacks);
2279
2280                 array_pop( $this->mArgStack );
2281
2282                 wfProfileOut( $fname );
2283                 return $text;
2284         }
2285
2286         /**
2287          * Replace magic variables
2288          * @access private
2289          */
2290         function variableSubstitution( $matches ) {
2291                 $fname = 'Parser::variableSubstitution';
2292                 $varname = $matches[1];
2293                 wfProfileIn( $fname );
2294                 if ( !$this->mVariables ) {
2295                         $this->initialiseVariables();
2296                 }
2297                 $skip = false;
2298                 if ( $this->mOutputType == OT_WIKI ) {
2299                         # Do only magic variables prefixed by SUBST
2300                         $mwSubst =& MagicWord::get( MAG_SUBST );
2301                         if (!$mwSubst->matchStartAndRemove( $varname ))
2302                                 $skip = true;
2303                         # Note that if we don't substitute the variable below,
2304                         # we don't remove the {{subst:}} magic word, in case
2305                         # it is a template rather than a magic variable.
2306                 }
2307                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2308                         $id = $this->mVariables[$varname];
2309                         $text = $this->getVariableValue( $id );
2310                         $this->mOutput->mContainsOldMagic = true;
2311                 } else {
2312                         $text = $matches[0];
2313                 }
2314                 wfProfileOut( $fname );
2315                 return $text;
2316         }
2317
2318         # Split template arguments
2319         function getTemplateArgs( $argsString ) {
2320                 if ( $argsString === '' ) {
2321                         return array();
2322                 }
2323
2324                 $args = explode( '|', substr( $argsString, 1 ) );
2325
2326                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2327                 # merged with the next arg because the '|' character between belongs
2328                 # to the link syntax and not the template parameter syntax.
2329                 $argc = count($args);
2330
2331                 for ( $i = 0; $i < $argc-1; $i++ ) {
2332                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2333                                 $args[$i] .= '|'.$args[$i+1];
2334                                 array_splice($args, $i+1, 1);
2335                                 $i--;
2336                                 $argc--;
2337                         }
2338                 }
2339
2340                 return $args;
2341         }
2342
2343         /**
2344          * Return the text of a template, after recursively
2345          * replacing any variables or templates within the template.
2346          *
2347          * @param array $piece The parts of the template
2348          *  $piece['text']: matched text
2349          *  $piece['title']: the title, i.e. the part before the |
2350          *  $piece['parts']: the parameter array
2351          * @return string the text of the template
2352          * @access private
2353          */
2354         function braceSubstitution( $piece ) {
2355                 global $wgContLang;
2356                 $fname = 'Parser::braceSubstitution';
2357                 wfProfileIn( $fname );
2358
2359                 $found = false;
2360                 $nowiki = false;
2361                 $noparse = false;
2362                 $replaceHeadings = false;
2363                 $isHTML = false;
2364
2365                 $title = NULL;
2366
2367                 $linestart = '';
2368
2369                 # $part1 is the bit before the first |, and must contain only title characters
2370                 # $args is a list of arguments, starting from index 0, not including $part1
2371
2372                 $part1 = $piece['title'];
2373                 # If the third subpattern matched anything, it will start with |
2374
2375                 if (null == $piece['parts']) {
2376                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2377                         if ($replaceWith != $piece['text']) {
2378                                 $text = $replaceWith;
2379                                 $found = true;
2380                                 $noparse = true;
2381                         }
2382                 }
2383
2384                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2385                 $argc = count( $args );
2386
2387                 # SUBST
2388                 if ( !$found ) {
2389                         $mwSubst =& MagicWord::get( MAG_SUBST );
2390                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2391                                 # One of two possibilities is true:
2392                                 # 1) Found SUBST but not in the PST phase
2393                                 # 2) Didn't find SUBST and in the PST phase
2394                                 # In either case, return without further processing
2395                                 $text = $piece['text'];
2396                                 $found = true;
2397                                 $noparse = true;
2398                         }
2399                 }
2400
2401                 # MSG, MSGNW and INT
2402                 if ( !$found ) {
2403                         # Check for MSGNW:
2404                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2405                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2406                                 $nowiki = true;
2407                         } else {
2408                                 # Remove obsolete MSG:
2409                                 $mwMsg =& MagicWord::get( MAG_MSG );
2410                                 $mwMsg->matchStartAndRemove( $part1 );
2411                         }
2412
2413                         # Check if it is an internal message
2414                         $mwInt =& MagicWord::get( MAG_INT );
2415                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2416                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2417                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2418                                         $found = true;
2419                                 }
2420                         }
2421                 }
2422
2423                 # NS
2424                 if ( !$found ) {
2425                         # Check for NS: (namespace expansion)
2426                         $mwNs = MagicWord::get( MAG_NS );
2427                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2428                                 if ( intval( $part1 ) || $part1 == "0" ) {
2429                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2430                                         $found = true;
2431                                 } else {
2432                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2433                                         if ( !is_null( $index ) ) {
2434                                                 $text = $linestart . $wgContLang->getNsText( $index );
2435                                                 $found = true;
2436                                         }
2437                                 }
2438                         }
2439                 }
2440
2441                 # LCFIRST, UCFIRST, LC and UC
2442                 if ( !$found ) {
2443                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2444                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2445                         $lc =& MagicWord::get( MAG_LC );
2446                         $uc =& MagicWord::get( MAG_UC );
2447                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2448                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2449                                 $found = true;
2450                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2451                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2452                                 $found = true;
2453                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2454                                 $text = $linestart . $wgContLang->lc( $part1 );
2455                                 $found = true;
2456                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2457                                  $text = $linestart . $wgContLang->uc( $part1 );
2458                                  $found = true;
2459                         }
2460                 }
2461
2462                 # LOCALURL and FULLURL
2463                 if ( !$found ) {
2464                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2465                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2466                         $mwFull =& MagicWord::get( MAG_FULLURL );
2467                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2468
2469
2470                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2471                                 $func = 'getLocalURL';
2472                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2473                                 $func = 'escapeLocalURL';
2474                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2475                                 $func = 'getFullURL';
2476                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2477                                 $func = 'escapeFullURL';
2478                         } else {
2479                                 $func = false;
2480                         }
2481
2482                         if ( $func !== false ) {
2483                                 $title = Title::newFromText( $part1 );
2484                                 if ( !is_null( $title ) ) {
2485                                         if ( $argc > 0 ) {
2486                                                 $text = $linestart . $title->$func( $args[0] );
2487                                         } else {
2488                                                 $text = $linestart . $title->$func();
2489                                         }
2490                                         $found = true;
2491                                 }
2492                         }
2493                 }
2494
2495                 # GRAMMAR
2496                 if ( !$found && $argc == 1 ) {
2497                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2498                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2499                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2500                                 $found = true;
2501                         }
2502                 }
2503
2504                 # PLURAL
2505                 if ( !$found && $argc >= 2 ) {
2506                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2507                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2508                                 if ($argc==2) {$args[2]=$args[1];}
2509                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2510                                 $found = true;
2511                         }
2512                 }
2513
2514                 # Template table test
2515
2516                 # Did we encounter this template already? If yes, it is in the cache
2517                 # and we need to check for loops.
2518                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2519                         $found = true;
2520
2521                         # Infinite loop test
2522                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2523                                 $noparse = true;
2524                                 $found = true;
2525                                 $text = $linestart .
2526                                         '{{' . $part1 . '}}' .
2527                                         '<!-- WARNING: template loop detected -->';
2528                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2529                         } else {
2530                                 # set $text to cached message.
2531                                 $text = $linestart . $this->mTemplates[$part1];
2532                         }
2533                 }
2534
2535                 # Load from database
2536                 $lastPathLevel = $this->mTemplatePath;
2537                 if ( !$found ) {
2538                         $ns = NS_TEMPLATE;
2539                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2540                         if ($subpage !== '') {
2541                                 $ns = $this->mTitle->getNamespace();
2542                         }
2543                         $title = Title::newFromText( $part1, $ns );
2544
2545                         if ( !is_null( $title ) ) {
2546                                 if ( !$title->isExternal() ) {
2547                                         # Check for excessive inclusion
2548                                         $dbk = $title->getPrefixedDBkey();
2549                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2550                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2551                                                         # Capture special page output
2552                                                         $text = SpecialPage::capturePath( $title );
2553                                                         if ( is_string( $text ) ) {
2554                                                                 $found = true;
2555                                                                 $noparse = true;
2556                                                                 $isHTML = true;
2557                                                                 $this->disableCache();
2558                                                         }
2559                                                 } else {
2560                                                         $articleContent = $this->fetchTemplate( $title );
2561                                                         if ( $articleContent !== false ) {
2562                                                                 $found = true;
2563                                                                 $text = $articleContent;
2564                                                                 $replaceHeadings = true;
2565                                                         }
2566                                                 }
2567                                         }
2568
2569                                         # If the title is valid but undisplayable, make a link to it
2570                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2571                                                 $text = '[['.$title->getPrefixedText().']]';
2572                                                 $found = true;
2573                                         }
2574
2575                                         # Template cache array insertion
2576                                         if( $found ) {
2577                                                 $this->mTemplates[$part1] = $text;
2578                                                 $text = $linestart . $text;
2579                                         }
2580                                 } elseif ( $title->isTrans() ) {
2581                                         // Interwiki transclusion
2582                                         if ( $this->mOutputType == OT_HTML ) {
2583                                                 $text = $this->interwikiTransclude( $title, 'render' );
2584                                                 $isHTML = true;
2585                                                 $noparse = true;
2586                                         } else {
2587                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2588                                         }
2589                                         $found = true;
2590                                 }
2591                         }
2592                 }
2593
2594                 # Recursive parsing, escaping and link table handling
2595                 # Only for HTML output
2596                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2597                         $text = wfEscapeWikiText( $text );
2598                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2599                         # Clean up argument array
2600                         $assocArgs = array();
2601                         $index = 1;
2602                         foreach( $args as $arg ) {
2603                                 $eqpos = strpos( $arg, '=' );
2604                                 if ( $eqpos === false ) {
2605                                         $assocArgs[$index++] = $arg;
2606                                 } else {
2607                                         $name = trim( substr( $arg, 0, $eqpos ) );
2608                                         $value = trim( substr( $arg, $eqpos+1 ) );
2609                                         if ( $value === false ) {
2610                                                 $value = '';
2611                                         }
2612                                         if ( $name !== false ) {
2613                                                 $assocArgs[$name] = $value;
2614                                         }
2615                                 }
2616                         }
2617
2618                         # Add a new element to the templace recursion path
2619                         $this->mTemplatePath[$part1] = 1;
2620
2621                         # If there are any <onlyinclude> tags, only include them
2622                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2623                                 preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2624                                 $text = '';
2625                                 foreach ($m[1] as $piece)
2626                                         $text .= $piece;
2627                         }
2628                         # Remove <noinclude> sections and <includeonly> tags
2629                         $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2630                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2631
2632                         if( $this->mOutputType == OT_HTML ) {
2633                                 # Strip <nowiki>, <pre>, etc.
2634                                 $text = $this->strip( $text, $this->mStripState );
2635                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2636                         }
2637                         $text = $this->replaceVariables( $text, $assocArgs );
2638
2639                         # If the template begins with a table or block-level
2640                         # element, it should be treated as beginning a new line.
2641                         if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2642                                 $text = "\n" . $text;
2643                         }
2644                 }
2645                 # Prune lower levels off the recursion check path
2646                 $this->mTemplatePath = $lastPathLevel;
2647
2648                 if ( !$found ) {
2649                         wfProfileOut( $fname );
2650                         return $piece['text'];
2651                 } else {
2652                         if ( $isHTML ) {
2653                                 # Replace raw HTML by a placeholder
2654                                 # Add a blank line preceding, to prevent it from mucking up
2655                                 # immediately preceding headings
2656                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2657                         } else {
2658                                 # replace ==section headers==
2659                                 # XXX this needs to go away once we have a better parser.
2660                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2661                                         if( !is_null( $title ) )
2662                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2663                                         else
2664                                                 $encodedname = base64_encode("");
2665                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2666                                                 PREG_SPLIT_DELIM_CAPTURE);
2667                                         $text = '';
2668                                         $nsec = 0;
2669                                         for( $i = 0; $i < count($m); $i += 2 ) {
2670                                                 $text .= $m[$i];
2671                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2672                                                 $hl = $m[$i + 1];
2673                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2674                                                         $text .= $hl;
2675                                                         continue;
2676                                                 }
2677                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2678                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2679                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2680
2681                                                 $nsec++;
2682                                         }
2683                                 }
2684                         }
2685                 }
2686
2687                 # Prune lower levels off the recursion check path
2688                 $this->mTemplatePath = $lastPathLevel;
2689
2690                 if ( !$found ) {
2691                         wfProfileOut( $fname );
2692                         return $piece['text'];
2693                 } else {
2694                         wfProfileOut( $fname );
2695                         return $text;
2696                 }
2697         }
2698
2699         /**
2700          * Fetch the unparsed text of a template and register a reference to it.
2701          */
2702         function fetchTemplate( $title ) {
2703                 $text = false;
2704                 // Loop to fetch the article, with up to 1 redirect
2705                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2706                         $rev = Revision::newFromTitle( $title );
2707                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2708                         if ( !$rev ) {
2709                                 break;
2710                         }
2711                         $text = $rev->getText();
2712                         if ( $text === false ) {
2713                                 break;
2714                         }
2715                         // Redirect?
2716                         $title = Title::newFromRedirect( $text );
2717                 }
2718                 return $text;
2719         }
2720
2721         /**
2722          * Transclude an interwiki link.
2723          */
2724         function interwikiTransclude( $title, $action ) {
2725                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2726
2727                 if (!$wgEnableScaryTranscluding)
2728                         return wfMsg('scarytranscludedisabled');
2729
2730                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2731                 // But we'll handle it generally anyway
2732                 if ( $title->getNamespace() ) {
2733                         // Use the canonical namespace, which should work anywhere
2734                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2735                 } else {
2736                         $articleName = $title->getDBkey();
2737                 }
2738
2739                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2740                 $url .= "?action=$action";
2741                 if (strlen($url) > 255)
2742                         return wfMsg('scarytranscludetoolong');
2743                 return $this->fetchScaryTemplateMaybeFromCache($url);
2744         }
2745
2746         function fetchScaryTemplateMaybeFromCache($url) {
2747                 global $wgTranscludeCacheExpiry;
2748                 $dbr =& wfGetDB(DB_SLAVE);
2749                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2750                                 array('tc_url' => $url));
2751                 if ($obj) {
2752                         $time = $obj->tc_time;
2753                         $text = $obj->tc_contents;
2754                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2755                                 return $text;
2756                         }
2757                 }
2758
2759                 $text = wfGetHTTP($url);
2760                 if (!$text)
2761                         return wfMsg('scarytranscludefailed', $url);
2762
2763                 $dbw =& wfGetDB(DB_MASTER);
2764                 $dbw->replace('transcache', array('tc_url'), array(
2765                         'tc_url' => $url,
2766                         'tc_time' => time(),
2767                         'tc_contents' => $text));
2768                 return $text;
2769         }
2770
2771
2772         /**
2773          * Triple brace replacement -- used for template arguments
2774          * @access private
2775          */
2776         function argSubstitution( $matches ) {
2777                 $arg = trim( $matches['title'] );
2778                 $text = $matches['text'];
2779                 $inputArgs = end( $this->mArgStack );
2780
2781                 if ( array_key_exists( $arg, $inputArgs ) ) {
2782                         $text = $inputArgs[$arg];
2783                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2784                         $text = $matches['parts'][0];
2785                 }
2786
2787                 return $text;
2788         }
2789
2790         /**
2791          * Returns true if the function is allowed to include this entity
2792          * @access private
2793          */
2794         function incrementIncludeCount( $dbk ) {
2795                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2796                         $this->mIncludeCount[$dbk] = 0;
2797                 }
2798                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2799                         return true;
2800                 } else {
2801                         return false;
2802                 }
2803         }
2804
2805         /**
2806          * This function accomplishes several tasks:
2807          * 1) Auto-number headings if that option is enabled
2808          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2809          * 3) Add a Table of contents on the top for users who have enabled the option
2810          * 4) Auto-anchor headings
2811          *
2812          * It loops through all headlines, collects the necessary data, then splits up the
2813          * string and re-inserts the newly formatted headlines.
2814          *
2815          * @param string $text
2816          * @param boolean $isMain
2817          * @access private
2818          */
2819         function formatHeadings( $text, $isMain=true ) {
2820                 global $wgMaxTocLevel, $wgContLang;
2821
2822                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2823                 $doShowToc = true;
2824                 $forceTocHere = false;
2825                 if( !$this->mTitle->userCanEdit() ) {
2826                         $showEditLink = 0;
2827                 } else {
2828                         $showEditLink = $this->mOptions->getEditSection();
2829                 }
2830
2831                 # Inhibit editsection links if requested in the page
2832                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2833                 if( $esw->matchAndRemove( $text ) ) {
2834                         $showEditLink = 0;
2835                 }
2836                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2837                 # do not add TOC
2838                 $mw =& MagicWord::get( MAG_NOTOC );
2839                 if( $mw->matchAndRemove( $text ) ) {
2840                         $doShowToc = false;
2841                 }
2842
2843                 # Get all headlines for numbering them and adding funky stuff like [edit]
2844                 # links - this is for later, but we need the number of headlines right now
2845                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2846
2847                 # if there are fewer than 4 headlines in the article, do not show TOC
2848                 if( $numMatches < 4 ) {
2849                         $doShowToc = false;
2850                 }
2851
2852                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2853                 # override above conditions and always show TOC at that place
2854
2855                 $mw =& MagicWord::get( MAG_TOC );
2856                 if($mw->match( $text ) ) {
2857                         $doShowToc = true;
2858                         $forceTocHere = true;
2859                 } else {
2860                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2861                         # override above conditions and always show TOC above first header
2862                         $mw =& MagicWord::get( MAG_FORCETOC );
2863                         if ($mw->matchAndRemove( $text ) ) {
2864                                 $doShowToc = true;
2865                         }
2866                 }
2867
2868                 # Never ever show TOC if no headers
2869                 if( $numMatches < 1 ) {
2870                         $doShowToc = false;
2871                 }
2872
2873                 # We need this to perform operations on the HTML
2874                 $sk =& $this->mOptions->getSkin();
2875
2876                 # headline counter
2877                 $headlineCount = 0;
2878                 $sectionCount = 0; # headlineCount excluding template sections
2879
2880                 # Ugh .. the TOC should have neat indentation levels which can be
2881                 # passed to the skin functions. These are determined here
2882                 $toc = '';
2883                 $full = '';
2884                 $head = array();
2885                 $sublevelCount = array();
2886                 $levelCount = array();
2887                 $toclevel = 0;
2888                 $level = 0;
2889                 $prevlevel = 0;
2890                 $toclevel = 0;
2891                 $prevtoclevel = 0;
2892
2893                 foreach( $matches[3] as $headline ) {
2894                         $istemplate = 0;
2895                         $templatetitle = '';
2896                         $templatesection = 0;
2897                         $numbering = '';
2898
2899                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2900                                 $istemplate = 1;
2901                                 $templatetitle = base64_decode($mat[1]);
2902                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2903                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2904                         }
2905
2906                         if( $toclevel ) {
2907                                 $prevlevel = $level;
2908                                 $prevtoclevel = $toclevel;
2909                         }
2910                         $level = $matches[1][$headlineCount];
2911
2912                         if( $doNumberHeadings || $doShowToc ) {
2913
2914                                 if ( $level > $prevlevel ) {
2915                                         # Increase TOC level
2916                                         $toclevel++;
2917                                         $sublevelCount[$toclevel] = 0;
2918                                         $toc .= $sk->tocIndent();
2919                                 }
2920                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2921                                         # Decrease TOC level, find level to jump to
2922
2923                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2924                                                 # Can only go down to level 1
2925                                                 $toclevel = 1;
2926                                         } else {
2927                                                 for ($i = $toclevel; $i > 0; $i--) {
2928                                                         if ( $levelCount[$i] == $level ) {
2929                                                                 # Found last matching level
2930                                                                 $toclevel = $i;
2931                                                                 break;
2932                                                         }
2933                                                         elseif ( $levelCount[$i] < $level ) {
2934                                                                 # Found first matching level below current level
2935                                                                 $toclevel = $i + 1;
2936                                                                 break;
2937                                                         }
2938                                                 }
2939                                         }
2940
2941                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2942                                 }
2943                                 else {
2944                                         # No change in level, end TOC line
2945                                         $toc .= $sk->tocLineEnd();
2946                                 }
2947
2948                                 $levelCount[$toclevel] = $level;
2949
2950                                 # count number of headlines for each level
2951                                 @$sublevelCount[$toclevel]++;
2952                                 $dot = 0;
2953                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2954                                         if( !empty( $sublevelCount[$i] ) ) {
2955                                                 if( $dot ) {
2956                                                         $numbering .= '.';
2957                                                 }
2958                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2959                                                 $dot = 1;
2960                                         }
2961                                 }
2962                         }
2963
2964                         # The canonized header is a version of the header text safe to use for links
2965                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2966                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2967                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2968
2969                         # Remove link placeholders by the link text.
2970                         #     <!--LINK number-->
2971                         # turns into
2972                         #     link text with suffix
2973                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2974                                                             "\$this->mLinkHolders['texts'][\$1]",
2975                                                             $canonized_headline );
2976                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2977                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2978                                                             $canonized_headline );
2979
2980                         # strip out HTML
2981                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2982                         $tocline = trim( $canonized_headline );
2983                         $canonized_headline = Sanitizer::escapeId( $tocline );
2984                         $refers[$headlineCount] = $canonized_headline;
2985
2986                         # count how many in assoc. array so we can track dupes in anchors
2987                         @$refers[$canonized_headline]++;
2988                         $refcount[$headlineCount]=$refers[$canonized_headline];
2989
2990                         # Don't number the heading if it is the only one (looks silly)
2991                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2992                                 # the two are different if the line contains a link
2993                                 $headline=$numbering . ' ' . $headline;
2994                         }
2995
2996                         # Create the anchor for linking from the TOC to the section
2997                         $anchor = $canonized_headline;
2998                         if($refcount[$headlineCount] > 1 ) {
2999                                 $anchor .= '_' . $refcount[$headlineCount];
3000                         }
3001                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3002                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3003                         }
3004                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3005                                 if ( empty( $head[$headlineCount] ) ) {
3006                                         $head[$headlineCount] = '';
3007                                 }
3008                                 if( $istemplate )
3009                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3010                                 else
3011                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
3012                         }
3013
3014                         # give headline the correct <h#> tag
3015                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3016
3017                         $headlineCount++;
3018                         if( !$istemplate )
3019                                 $sectionCount++;
3020                 }
3021
3022                 if( $doShowToc ) {
3023                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3024                         $toc = $sk->tocList( $toc );
3025                 }
3026
3027                 # split up and insert constructed headlines
3028
3029                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3030                 $i = 0;
3031
3032                 foreach( $blocks as $block ) {
3033                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3034                                 # This is the [edit] link that appears for the top block of text when
3035                                 # section editing is enabled
3036
3037                                 # Disabled because it broke block formatting
3038                                 # For example, a bullet point in the top line
3039                                 # $full .= $sk->editSectionLink(0);
3040                         }
3041                         $full .= $block;
3042                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3043                         # Top anchor now in skin
3044                                 $full = $full.$toc;
3045                         }
3046
3047                         if( !empty( $head[$i] ) ) {
3048                                 $full .= $head[$i];
3049                         }
3050                         $i++;
3051                 }
3052                 if($forceTocHere) {
3053                         $mw =& MagicWord::get( MAG_TOC );
3054                         return $mw->replace( $toc, $full );
3055                 } else {
3056                         return $full;
3057                 }
3058         }
3059
3060         /**
3061          * Return an HTML link for the "ISBN 123456" text
3062          * @access private
3063          */
3064         function magicISBN( $text ) {
3065                 $fname = 'Parser::magicISBN';
3066                 wfProfileIn( $fname );
3067
3068                 $a = split( 'ISBN ', ' '.$text );
3069                 if ( count ( $a ) < 2 ) {
3070                         wfProfileOut( $fname );
3071                         return $text;
3072                 }
3073                 $text = substr( array_shift( $a ), 1);
3074                 $valid = '0123456789-Xx';
3075
3076                 foreach ( $a as $x ) {
3077                         $isbn = $blank = '' ;
3078                         while ( ' ' == $x{0} ) {
3079                                 $blank .= ' ';
3080                                 $x = substr( $x, 1 );
3081                         }
3082                         if ( $x == '' ) { # blank isbn
3083                                 $text .= "ISBN $blank";
3084                                 continue;
3085                         }
3086                         while ( strstr( $valid, $x{0} ) != false ) {
3087                                 $isbn .= $x{0};
3088                                 $x = substr( $x, 1 );
3089                         }
3090                         $num = str_replace( '-', '', $isbn );
3091                         $num = str_replace( ' ', '', $num );
3092                         $num = str_replace( 'x', 'X', $num );
3093
3094                         if ( '' == $num ) {
3095                                 $text .= "ISBN $blank$x";
3096                         } else {
3097                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3098                                 $text .= '<a href="' .
3099                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3100                                         "\" class=\"internal\">ISBN $isbn</a>";
3101                                 $text .= $x;
3102                         }
3103                 }
3104                 wfProfileOut( $fname );
3105                 return $text;
3106         }
3107
3108         /**
3109          * Return an HTML link for the "RFC 1234" text
3110          *
3111          * @access private
3112          * @param string $text     Text to be processed
3113          * @param string $keyword  Magic keyword to use (default RFC)
3114          * @param string $urlmsg   Interface message to use (default rfcurl)
3115          * @return string
3116          */
3117         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3118
3119                 $valid = '0123456789';
3120                 $internal = false;
3121
3122                 $a = split( $keyword, ' '.$text );
3123                 if ( count ( $a ) < 2 ) {
3124                         return $text;
3125                 }
3126                 $text = substr( array_shift( $a ), 1);
3127
3128                 /* Check if keyword is preceed by [[.
3129                  * This test is made here cause of the array_shift above
3130                  * that prevent the test to be done in the foreach.
3131                  */
3132                 if ( substr( $text, -2 ) == '[[' ) {
3133                         $internal = true;
3134                 }
3135
3136                 foreach ( $a as $x ) {
3137                         /* token might be empty if we have RFC RFC 1234 */
3138                         if ( $x=='' ) {
3139                                 $text.=$keyword;
3140                                 continue;
3141                                 }
3142
3143                         $id = $blank = '' ;
3144
3145                         /** remove and save whitespaces in $blank */
3146                         while ( $x{0} == ' ' ) {
3147                                 $blank .= ' ';
3148                                 $x = substr( $x, 1 );
3149                         }
3150
3151                         /** remove and save the rfc number in $id */
3152                         while ( strstr( $valid, $x{0} ) != false ) {
3153                                 $id .= $x{0};
3154                                 $x = substr( $x, 1 );
3155                         }
3156
3157                         if ( $id == '' ) {
3158                                 /* call back stripped spaces*/
3159                                 $text .= $keyword.$blank.$x;
3160                         } elseif( $internal ) {
3161                                 /* normal link */
3162                                 $text .= $keyword.$id.$x;
3163                         } else {
3164                                 /* build the external link*/
3165                                 $url = wfMsg( $urlmsg, $id);
3166                                 $sk =& $this->mOptions->getSkin();
3167                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3168                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3169                         }
3170
3171                         /* Check if the next RFC keyword is preceed by [[ */
3172                         $internal = ( substr($x,-2) == '[[' );
3173                 }
3174                 return $text;
3175         }
3176
3177         /**
3178          * Transform wiki markup when saving a page by doing \r\n -> \n
3179          * conversion, substitting signatures, {{subst:}} templates, etc.
3180          *
3181          * @param string $text the text to transform
3182          * @param Title &$title the Title object for the current article
3183          * @param User &$user the User object describing the current user
3184          * @param ParserOptions $options parsing options
3185          * @param bool $clearState whether to clear the parser state first
3186          * @return string the altered wiki markup
3187          * @access public
3188          */
3189         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3190                 $this->mOptions = $options;
3191                 $this->mTitle =& $title;
3192                 $this->mOutputType = OT_WIKI;
3193
3194                 if ( $clearState ) {
3195                         $this->clearState();
3196                 }
3197
3198                 $stripState = false;
3199                 $pairs = array(
3200                         "\r\n" => "\n",
3201                 );
3202                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3203                 $text = $this->strip( $text, $stripState, true );
3204                 $text = $this->pstPass2( $text, $user );
3205                 $text = $this->unstrip( $text, $stripState );
3206                 $text = $this->unstripNoWiki( $text, $stripState );
3207                 return $text;
3208         }
3209
3210         /**
3211          * Pre-save transform helper function
3212          * @access private
3213          */
3214         function pstPass2( $text, &$user ) {
3215                 global $wgContLang, $wgLocaltimezone;
3216
3217                 /* Note: This is the timestamp saved as hardcoded wikitext to
3218                  * the database, we use $wgContLang here in order to give
3219                  * everyone the same signiture and use the default one rather
3220                  * than the one selected in each users preferences.
3221                  */
3222                 if ( isset( $wgLocaltimezone ) ) {
3223                         $oldtz = getenv( 'TZ' );
3224                         putenv( 'TZ='.$wgLocaltimezone );
3225                 }
3226                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3227                   ' (' . date( 'T' ) . ')';
3228                 if ( isset( $wgLocaltimezone ) ) {
3229                         putenv( 'TZ='.$oldtz );
3230                 }
3231
3232                 # Variable replacement
3233                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3234                 $text = $this->replaceVariables( $text );
3235
3236                 # Signatures
3237                 $sigText = $this->getUserSig( $user );
3238                 $text = strtr( $text, array(
3239                         '~~~~~' => $d,
3240                         '~~~~' => "$sigText $d",
3241                         '~~~' => $sigText
3242                 ) );
3243
3244                 # Context links: [[|name]] and [[name (context)|]]
3245                 #
3246                 global $wgLegalTitleChars;
3247                 $tc = "[$wgLegalTitleChars]";
3248                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3249
3250                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3251                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3252
3253                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3254                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3255                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3256                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3257                 $context = '';
3258                 $t = $this->mTitle->getText();
3259                 if ( preg_match( $conpat, $t, $m ) ) {
3260                         $context = $m[2];
3261                 }
3262                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3263                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3264                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3265
3266                 if ( '' == $context ) {
3267                         $text = preg_replace( $p2, '[[\\1]]', $text );
3268                 } else {
3269                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3270                 }
3271
3272                 # Trim trailing whitespace
3273                 # MAG_END (__END__) tag allows for trailing
3274                 # whitespace to be deliberately included
3275                 $text = rtrim( $text );
3276                 $mw =& MagicWord::get( MAG_END );
3277                 $mw->matchAndRemove( $text );
3278
3279                 return $text;
3280         }
3281
3282         /**
3283          * Fetch the user's signature text, if any, and normalize to
3284          * validated, ready-to-insert wikitext.
3285          *
3286          * @param User $user
3287          * @return string
3288          * @access private
3289          */
3290         function getUserSig( &$user ) {
3291                 global $wgContLang;
3292
3293                 $username = $user->getName();
3294                 $nickname = $user->getOption( 'nickname' );
3295                 $nickname = $nickname === '' ? $username : $nickname;
3296
3297                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3298                         # Sig. might contain markup; validate this
3299                         if( $this->validateSig( $nickname ) !== false ) {
3300                                 # Validated; clean up (if needed) and return it
3301                                 return( $this->cleanSig( $nickname ) );
3302                         } else {
3303                                 # Failed to validate; fall back to the default
3304                                 $nickname = $username;
3305                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3306                         }
3307                 }
3308
3309                 # If we're still here, make it a link to the user page
3310                 $userpage = $user->getUserPage();
3311                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3312         }
3313
3314         /**
3315          * Check that the user's signature contains no bad XML
3316          *
3317          * @param string $text
3318          * @return mixed An expanded string, or false if invalid.
3319          */
3320         function validateSig( $text ) {
3321                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3322         }
3323
3324         /**
3325          * Clean up signature text
3326          *
3327          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3328          * 2) Substitute all transclusions
3329          *
3330          * @param string $text
3331          * @return string Signature text
3332          */
3333         function cleanSig( $text ) {
3334                 $substWord = MagicWord::get( MAG_SUBST );
3335                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3336                 $substText = '{{' . $substWord->getSynonym( 0 );
3337
3338                 $text = preg_replace( $substRegex, $substText, $text );
3339                 $text = preg_replace( '/~{3,5}/', '', $text );
3340                 $text = $this->replaceVariables( $text );
3341
3342                 return $text;
3343         }
3344
3345         /**
3346          * Set up some variables which are usually set up in parse()
3347          * so that an external function can call some class members with confidence
3348          * @access public
3349          */
3350         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3351                 $this->mTitle =& $title;
3352                 $this->mOptions = $options;
3353                 $this->mOutputType = $outputType;
3354                 if ( $clearState ) {
3355                         $this->clearState();
3356                 }
3357         }
3358
3359         /**
3360          * Transform a MediaWiki message by replacing magic variables.
3361          *
3362          * @param string $text the text to transform
3363          * @param ParserOptions $options  options
3364          * @return string the text with variables substituted
3365          * @access public
3366          */
3367         function transformMsg( $text, $options ) {
3368                 global $wgTitle;
3369                 static $executing = false;
3370
3371                 $fname = "Parser::transformMsg";
3372
3373                 # Guard against infinite recursion
3374                 if ( $executing ) {
3375                         return $text;
3376                 }
3377                 $executing = true;
3378
3379                 wfProfileIn($fname);
3380
3381                 $this->mTitle = $wgTitle;
3382                 $this->mOptions = $options;
3383                 $this->mOutputType = OT_MSG;
3384                 $this->clearState();
3385                 $text = $this->replaceVariables( $text );
3386
3387                 $executing = false;
3388                 wfProfileOut($fname);
3389                 return $text;
3390         }
3391
3392         /**
3393          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3394          * Callback will be called with the text within
3395          * Transform and return the text within
3396          *
3397          * @access public
3398          *
3399          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3400          * @param mixed $callback The callback function (and object) to use for the tag
3401          *
3402          * @return The old value of the mTagHooks array associated with the hook
3403          */
3404         function setHook( $tag, $callback ) {
3405                 $oldVal = @$this->mTagHooks[$tag];
3406                 $this->mTagHooks[$tag] = $callback;
3407
3408                 return $oldVal;
3409         }
3410
3411         /**
3412          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3413          * Placeholders created in Skin::makeLinkObj()
3414          * Returns an array of links found, indexed by PDBK:
3415          *  0 - broken
3416          *  1 - normal link
3417          *  2 - stub
3418          * $options is a bit field, RLH_FOR_UPDATE to select for update
3419          */
3420         function replaceLinkHolders( &$text, $options = 0 ) {
3421                 global $wgUser;
3422                 global $wgOutputReplace;
3423
3424                 $fname = 'Parser::replaceLinkHolders';
3425                 wfProfileIn( $fname );
3426
3427                 $pdbks = array();
3428                 $colours = array();
3429                 $sk =& $this->mOptions->getSkin();
3430                 $linkCache =& LinkCache::singleton();
3431
3432                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3433                         wfProfileIn( $fname.'-check' );
3434                         $dbr =& wfGetDB( DB_SLAVE );
3435                         $page = $dbr->tableName( 'page' );
3436                         $threshold = $wgUser->getOption('stubthreshold');
3437
3438                         # Sort by namespace
3439                         asort( $this->mLinkHolders['namespaces'] );
3440
3441                         # Generate query
3442                         $query = false;
3443                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3444                                 # Make title object
3445                                 $title = $this->mLinkHolders['titles'][$key];
3446
3447                                 # Skip invalid entries.
3448                                 # Result will be ugly, but prevents crash.
3449                                 if ( is_null( $title ) ) {
3450                                         continue;
3451                                 }
3452                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3453
3454                                 # Check if it's a static known link, e.g. interwiki
3455                                 if ( $title->isAlwaysKnown() ) {
3456                                         $colours[$pdbk] = 1;
3457                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3458                                         $colours[$pdbk] = 1;
3459                                         $this->mOutput->addLink( $title, $id );
3460                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3461                                         $colours[$pdbk] = 0;
3462                                 } else {
3463                                         # Not in the link cache, add it to the query
3464                                         if ( !isset( $current ) ) {
3465                                                 $current = $ns;
3466                                                 $query =  "SELECT page_id, page_namespace, page_title";
3467                                                 if ( $threshold > 0 ) {
3468                                                         $query .= ', page_len, page_is_redirect';
3469                                                 }
3470                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3471                                         } elseif ( $current != $ns ) {
3472                                                 $current = $ns;
3473                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3474                                         } else {
3475                                                 $query .= ', ';
3476                                         }
3477
3478                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3479                                 }
3480                         }
3481                         if ( $query ) {
3482                                 $query .= '))';
3483                                 if ( $options & RLH_FOR_UPDATE ) {
3484                                         $query .= ' FOR UPDATE';
3485                                 }
3486
3487                                 $res = $dbr->query( $query, $fname );
3488
3489                                 # Fetch data and form into an associative array
3490                                 # non-existent = broken
3491                                 # 1 = known
3492                                 # 2 = stub
3493                                 while ( $s = $dbr->fetchObject($res) ) {
3494                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3495                                         $pdbk = $title->getPrefixedDBkey();
3496                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3497                                         $this->mOutput->addLink( $title, $s->page_id );
3498
3499                                         if ( $threshold >  0 ) {
3500                                                 $size = $s->page_len;
3501                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3502                                                         $colours[$pdbk] = 1;
3503                                                 } else {
3504                                                         $colours[$pdbk] = 2;
3505                                                 }
3506                                         } else {
3507                                                 $colours[$pdbk] = 1;
3508                                         }
3509                                 }
3510                         }
3511                         wfProfileOut( $fname.'-check' );
3512
3513                         # Construct search and replace arrays
3514                         wfProfileIn( $fname.'-construct' );
3515                         $wgOutputReplace = array();
3516                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3517                                 $pdbk = $pdbks[$key];
3518                                 $searchkey = "<!--LINK $key-->";
3519                                 $title = $this->mLinkHolders['titles'][$key];
3520                                 if ( empty( $colours[$pdbk] ) ) {
3521                                         $linkCache->addBadLinkObj( $title );
3522                                         $colours[$pdbk] = 0;
3523                                         $this->mOutput->addLink( $title, 0 );
3524                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3525                                                                         $this->mLinkHolders['texts'][$key],
3526                                                                         $this->mLinkHolders['queries'][$key] );
3527                                 } elseif ( $colours[$pdbk] == 1 ) {
3528                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3529                                                                         $this->mLinkHolders['texts'][$key],
3530                                                                         $this->mLinkHolders['queries'][$key] );
3531                                 } elseif ( $colours[$pdbk] == 2 ) {
3532                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3533                                                                         $this->mLinkHolders['texts'][$key],
3534                                                                         $this->mLinkHolders['queries'][$key] );
3535                                 }
3536                         }
3537                         wfProfileOut( $fname.'-construct' );
3538
3539                         # Do the thing
3540                         wfProfileIn( $fname.'-replace' );
3541
3542                         $text = preg_replace_callback(
3543                                 '/(<!--LINK .*?-->)/',
3544                                 "wfOutputReplaceMatches",
3545                                 $text);
3546
3547                         wfProfileOut( $fname.'-replace' );
3548                 }
3549
3550                 # Now process interwiki link holders
3551                 # This is quite a bit simpler than internal links
3552                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3553                         wfProfileIn( $fname.'-interwiki' );
3554                         # Make interwiki link HTML
3555                         $wgOutputReplace = array();
3556                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3557                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3558                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3559                         }
3560
3561                         $text = preg_replace_callback(
3562                                 '/<!--IWLINK (.*?)-->/',
3563                                 "wfOutputReplaceMatches",
3564                                 $text );
3565                         wfProfileOut( $fname.'-interwiki' );
3566                 }
3567
3568                 wfProfileOut( $fname );
3569                 return $colours;
3570         }
3571
3572         /**
3573          * Replace <!--LINK--> link placeholders with plain text of links
3574          * (not HTML-formatted).
3575          * @param string $text
3576          * @return string
3577          */
3578         function replaceLinkHoldersText( $text ) {
3579                 global $wgUser;
3580                 global $wgOutputReplace;
3581
3582                 $fname = 'Parser::replaceLinkHoldersText';
3583                 wfProfileIn( $fname );
3584
3585                 $text = preg_replace_callback(
3586                         '/<!--(LINK|IWLINK) (.*?)-->/',
3587                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3588                         $text );
3589
3590                 wfProfileOut( $fname );
3591                 return $text;
3592         }
3593
3594         /**
3595          * @param array $matches
3596          * @return string
3597          * @access private
3598          */
3599         function replaceLinkHoldersTextCallback( $matches ) {
3600                 $type = $matches[1];
3601                 $key  = $matches[2];
3602                 if( $type == 'LINK' ) {
3603                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3604                                 return $this->mLinkHolders['texts'][$key];
3605                         }
3606                 } elseif( $type == 'IWLINK' ) {
3607                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3608                                 return $this->mInterwikiLinkHolders['texts'][$key];
3609                         }
3610                 }
3611                 return $matches[0];
3612         }
3613
3614         /**
3615          * Renders an image gallery from a text with one line per image.
3616          * text labels may be given by using |-style alternative text. E.g.
3617          *   Image:one.jpg|The number "1"
3618          *   Image:tree.jpg|A tree
3619          * given as text will return the HTML of a gallery with two images,
3620          * labeled 'The number "1"' and
3621          * 'A tree'.
3622          */
3623         function renderImageGallery( $text ) {
3624                 # Setup the parser
3625                 $parserOptions = new ParserOptions;
3626                 $localParser = new Parser();
3627
3628                 $ig = new ImageGallery();
3629                 $ig->setShowBytes( false );
3630                 $ig->setShowFilename( false );
3631                 $lines = explode( "\n", $text );
3632
3633                 foreach ( $lines as $line ) {
3634                         # match lines like these:
3635                         # Image:someimage.jpg|This is some image
3636                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3637                         # Skip empty lines
3638                         if ( count( $matches ) == 0 ) {
3639                                 continue;
3640                         }
3641                         $nt =& Title::newFromText( $matches[1] );
3642                         if( is_null( $nt ) ) {
3643                                 # Bogus title. Ignore these so we don't bomb out later.
3644                                 continue;
3645                         }
3646                         if ( isset( $matches[3] ) ) {
3647                                 $label = $matches[3];
3648                         } else {
3649                                 $label = '';
3650                         }
3651
3652                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3653                         $html = $pout->getText();
3654
3655                         $ig->add( new Image( $nt ), $html );
3656                         $this->mOutput->addImage( $nt->getDBkey() );
3657                 }
3658                 return $ig->toHTML();
3659         }
3660
3661         /**
3662          * Parse image options text and use it to make an image
3663          */
3664         function makeImage( &$nt, $options ) {
3665                 global $wgContLang, $wgUseImageResize, $wgUser;
3666
3667                 $align = '';
3668
3669                 # Check if the options text is of the form "options|alt text"
3670                 # Options are:
3671                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3672                 #  * left               no resizing, just left align. label is used for alt= only
3673                 #  * right              same, but right aligned
3674                 #  * none               same, but not aligned
3675                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3676                 #  * center             center the image
3677                 #  * framed             Keep original image size, no magnify-button.
3678
3679                 $part = explode( '|', $options);
3680
3681                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3682                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3683                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3684                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3685                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3686                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3687                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3688                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3689                 $caption = '';
3690
3691                 $width = $height = $framed = $thumb = false;
3692                 $manual_thumb = '' ;
3693
3694                 foreach( $part as $key => $val ) {
3695                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3696                                 $thumb=true;
3697                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3698                                 # use manually specified thumbnail
3699                                 $thumb=true;
3700                                 $manual_thumb = $match;
3701                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3702                                 # remember to set an alignment, don't render immediately
3703                                 $align = 'right';
3704                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3705                                 # remember to set an alignment, don't render immediately
3706                                 $align = 'left';
3707                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3708                                 # remember to set an alignment, don't render immediately
3709                                 $align = 'center';
3710                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3711                                 # remember to set an alignment, don't render immediately
3712                                 $align = 'none';
3713                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3714                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3715                                 # $match is the image width in pixels
3716                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3717                                         $width = intval( $m[1] );
3718                                         $height = intval( $m[2] );
3719                                 } else {
3720                                         $width = intval($match);
3721                                 }
3722                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3723                                 $framed=true;
3724                         } else {
3725                                 $caption = $val;
3726                         }
3727                 }
3728                 # Strip bad stuff out of the alt text
3729                 $alt = $this->replaceLinkHoldersText( $caption );
3730                 $alt = Sanitizer::stripAllTags( $alt );
3731
3732                 # Linker does the rest
3733                 $sk =& $this->mOptions->getSkin();
3734                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3735         }
3736
3737         /**
3738          * Set a flag in the output object indicating that the content is dynamic and
3739          * shouldn't be cached.
3740          */
3741         function disableCache() {
3742                 $this->mOutput->mCacheTime = -1;
3743         }
3744
3745         /**#@+
3746          * Callback from the Sanitizer for expanding items found in HTML attribute
3747          * values, so they can be safely tested and escaped.
3748          * @param string $text
3749          * @param array $args
3750          * @return string
3751          * @access private
3752          */
3753         function attributeStripCallback( &$text, $args ) {
3754                 $text = $this->replaceVariables( $text, $args );
3755                 $text = $this->unstripForHTML( $text );
3756                 return $text;
3757         }
3758
3759         function unstripForHTML( $text ) {
3760                 $text = $this->unstrip( $text, $this->mStripState );
3761                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3762                 return $text;
3763         }
3764         /**#@-*/
3765
3766         /**#@+
3767          * Accessor/mutator
3768          */
3769         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3770         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3771         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3772         /**#@-*/
3773
3774         /**#@+
3775          * Accessor
3776          */
3777         function getTags() { return array_keys( $this->mTagHooks ); }
3778         /**#@-*/
3779 }
3780
3781 /**
3782  * @todo document
3783  * @package MediaWiki
3784  */
3785 class ParserOutput
3786 {
3787         var $mText,             # The output text
3788                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3789                 $mCategories,       # Map of category names to sort keys
3790                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3791                 $mCacheTime,        # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3792                 $mVersion,          # Compatibility check
3793                 $mTitleText,        # title text of the chosen language variant
3794                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3795                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3796                 $mImages,           # DB keys of the images used, in the array key only
3797                 $mExternalLinks;    # External link URLs, in the key only
3798
3799         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3800                 $containsOldMagic = false, $titletext = '' )
3801         {
3802                 $this->mText = $text;
3803                 $this->mLanguageLinks = $languageLinks;
3804                 $this->mCategories = $categoryLinks;
3805                 $this->mContainsOldMagic = $containsOldMagic;
3806                 $this->mCacheTime = '';
3807                 $this->mVersion = MW_PARSER_VERSION;
3808                 $this->mTitleText = $titletext;
3809                 $this->mLinks = array();
3810                 $this->mTemplates = array();
3811                 $this->mImages = array();
3812                 $this->mExternalLinks = array();
3813         }
3814
3815         function getText()                   { return $this->mText; }
3816         function getLanguageLinks()          { return $this->mLanguageLinks; }
3817         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3818         function &getCategories()            { return $this->mCategories; }
3819         function getCacheTime()              { return $this->mCacheTime; }
3820         function getTitleText()              { return $this->mTitleText; }
3821         function &getLinks()                 { return $this->mLinks; }
3822         function &getTemplates()             { return $this->mTemplates; }
3823         function &getImages()                { return $this->mImages; }
3824         function &getExternalLinks()         { return $this->mExternalLinks; }
3825
3826         function containsOldMagic()          { return $this->mContainsOldMagic; }
3827         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3828         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3829         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3830         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3831         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3832         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3833
3834         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3835         function addImage( $name )           { $this->mImages[$name] = 1; }
3836         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3837         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3838
3839         function addLink( $title, $id ) {
3840                 $ns = $title->getNamespace();
3841                 $dbk = $title->getDBkey();
3842                 if ( !isset( $this->mLinks[$ns] ) ) {
3843                         $this->mLinks[$ns] = array();
3844                 }
3845                 $this->mLinks[$ns][$dbk] = $id;
3846         }
3847
3848         function addTemplate( $title, $id ) {
3849                 $ns = $title->getNamespace();
3850                 $dbk = $title->getDBkey();
3851                 if ( !isset( $this->mTemplates[$ns] ) ) {
3852                         $this->mTemplates[$ns] = array();
3853                 }
3854                 $this->mTemplates[$ns][$dbk] = $id;
3855         }
3856
3857         /**
3858          * @deprecated
3859          */
3860         /*
3861         function merge( $other ) {
3862                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3863                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3864                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3865         }*/
3866
3867         /**
3868          * Return true if this cached output object predates the global or
3869          * per-article cache invalidation timestamps, or if it comes from
3870          * an incompatible older version.
3871          *
3872          * @param string $touched the affected article's last touched timestamp
3873          * @return bool
3874          * @access public
3875          */
3876         function expired( $touched ) {
3877                 global $wgCacheEpoch;
3878                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3879                        $this->getCacheTime() < $touched ||
3880                        $this->getCacheTime() <= $wgCacheEpoch ||
3881                        !isset( $this->mVersion ) ||
3882                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3883         }
3884 }
3885
3886 /**
3887  * Set options of the Parser
3888  * @todo document
3889  * @package MediaWiki
3890  */
3891 class ParserOptions
3892 {
3893         # All variables are private
3894         var $mUseTeX;                    # Use texvc to expand <math> tags
3895         var $mUseDynamicDates;           # Use DateFormatter to format dates
3896         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3897         var $mAllowExternalImages;       # Allow external images inline
3898         var $mAllowExternalImagesFrom;   # If not, any exception?
3899         var $mSkin;                      # Reference to the preferred skin
3900         var $mDateFormat;                # Date format index
3901         var $mEditSection;               # Create "edit section" links
3902         var $mNumberHeadings;            # Automatically number headings
3903         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3904         var $mTidy;                      # Ask for tidy cleanup
3905
3906         function getUseTeX()                        { return $this->mUseTeX; }
3907         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3908         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3909         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3910         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3911         function &getSkin()                         { return $this->mSkin; }
3912         function getDateFormat()                    { return $this->mDateFormat; }
3913         function getEditSection()                   { return $this->mEditSection; }
3914         function getNumberHeadings()                { return $this->mNumberHeadings; }
3915         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3916         function getTidy()                          { return $this->mTidy; }
3917
3918         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3919         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3920         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3921         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3922         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3923         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3924         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3925         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3926         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3927         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
3928         function setSkin( &$x ) { $this->mSkin =& $x; }
3929
3930         function ParserOptions() {
3931                 global $wgUser;
3932                 $this->initialiseFromUser( $wgUser );
3933         }
3934
3935         /**
3936          * Get parser options
3937          * @static
3938          */
3939         function newFromUser( &$user ) {
3940                 $popts = new ParserOptions;
3941                 $popts->initialiseFromUser( $user );
3942                 return $popts;
3943         }
3944
3945         /** Get user options */
3946         function initialiseFromUser( &$userInput ) {
3947                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3948                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3949                 $fname = 'ParserOptions::initialiseFromUser';
3950                 wfProfileIn( $fname );
3951                 if ( !$userInput ) {
3952                         $user = new User;
3953                         $user->setLoaded( true );
3954                 } else {
3955                         $user =& $userInput;
3956                 }
3957
3958                 $this->mUseTeX = $wgUseTeX;
3959                 $this->mUseDynamicDates = $wgUseDynamicDates;
3960                 $this->mInterwikiMagic = $wgInterwikiMagic;
3961                 $this->mAllowExternalImages = $wgAllowExternalImages;
3962                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
3963                 wfProfileIn( $fname.'-skin' );
3964                 $this->mSkin =& $user->getSkin();
3965                 wfProfileOut( $fname.'-skin' );
3966                 $this->mDateFormat = $user->getOption( 'date' );
3967                 $this->mEditSection = true;
3968                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3969                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3970                 $this->mTidy = false;
3971                 wfProfileOut( $fname );
3972         }
3973 }
3974
3975 /**
3976  * Callback function used by Parser::replaceLinkHolders()
3977  * to substitute link placeholders.
3978  */
3979 function &wfOutputReplaceMatches( $matches ) {
3980         global $wgOutputReplace;
3981         return $wgOutputReplace[$matches[1]];
3982 }
3983
3984 /**
3985  * Return the total number of articles
3986  */
3987 function wfNumberOfArticles() {
3988         global $wgNumberOfArticles;
3989
3990         wfLoadSiteStats();
3991         return $wgNumberOfArticles;
3992 }
3993
3994 /**
3995  * Return the number of files
3996  */
3997 function wfNumberOfFiles() {
3998         $fname = 'Parser::wfNumberOfFiles';
3999
4000         wfProfileIn( $fname );
4001         $dbr =& wfGetDB( DB_SLAVE );
4002         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4003         wfProfileOut( $fname );
4004
4005         return $res;
4006 }
4007
4008 /**
4009  * Get various statistics from the database
4010  * @access private
4011  */
4012 function wfLoadSiteStats() {
4013         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4014         $fname = 'wfLoadSiteStats';
4015
4016         if ( -1 != $wgNumberOfArticles ) return;
4017         $dbr =& wfGetDB( DB_SLAVE );
4018         $s = $dbr->selectRow( 'site_stats',
4019                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4020                 array( 'ss_row_id' => 1 ), $fname
4021         );
4022
4023         if ( $s === false ) {
4024                 return;
4025         } else {
4026                 $wgTotalViews = $s->ss_total_views;
4027                 $wgTotalEdits = $s->ss_total_edits;
4028                 $wgNumberOfArticles = $s->ss_good_articles;
4029         }
4030 }
4031
4032 /**
4033  * Escape html tags
4034  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4035  *
4036  * @param string $in Text that might contain HTML tags
4037  * @return string Escaped string
4038  */
4039 function wfEscapeHTMLTagsOnly( $in ) {
4040         return str_replace(
4041                 array( '"', '>', '<' ),
4042                 array( '&quot;', '&gt;', '&lt;' ),
4043                 $in );
4044 }
4045
4046 ?>