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, $wgContLang
  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, $mFunctionHooks;
  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         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 122                 $this->mTagHooks = array();
 123                 $this->mFunctionHooks = array();
 124                 $this->clearState();
 125         }
 126
 127         /**
 128          * Clear Parser state
 129          *
 130          * @access private
 131          */
 132         function clearState() {
 133                 $this->mOutput = new ParserOutput;
 134                 $this->mAutonumber = 0;
 135                 $this->mLastSection = '';
 136                 $this->mDTopen = false;
 137                 $this->mVariables = false;
 138                 $this->mIncludeCount = array();
 139                 $this->mStripState = array();
 140                 $this->mArgStack = array();
 141                 $this->mInPre = false;
 142                 $this->mInterwikiLinkHolders = array(
 143                         'texts' => array(),
 144                         'titles' => array()
 145                 );
 146                 $this->mLinkHolders = array(
 147                         'namespaces' => array(),
 148                         'dbkeys' => array(),
 149                         'queries' => array(),
 150                         'texts' => array(),
 151                         'titles' => array()
 152                 );
 153                 $this->mRevisionId = null;
 154                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 155
 156                 # Clear these on every parse, bug 4549
 157                 $this->mTemplates = array();
 158                 $this->mTemplatePath = array();
 159
 160                 wfRunHooks( 'ParserClearState', array( &$this ) );
 161         }
 162
 163         /**
 164          * Accessor for mUniqPrefix.
 165          *
 166          * @access public
 167          */
 168         function UniqPrefix() {
 169                 return $this->mUniqPrefix;
 170         }
 171
 172         /**
 173          * Convert wikitext to HTML
 174          * Do not call this function recursively.
 175          *
 176          * @access private
 177          * @param string $text Text we want to parse
 178          * @param Title &$title A title object
 179          * @param array $options
 180          * @param boolean $linestart
 181          * @param boolean $clearState
 182          * @param int $revid number to pass in {{REVISIONID}}
 183          * @return ParserOutput a ParserOutput
 184          */
 185         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 186                 /**
 187                  * First pass--just handle <nowiki> sections, pass the rest off
 188                  * to internalParse() which does all the real work.
 189                  */
 190
 191                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 192                 $fname = 'Parser::parse';
 193                 wfProfileIn( $fname );
 194
 195                 if ( $clearState ) {
 196                         $this->clearState();
 197                 }
 198
 199                 $this->mOptions = $options;
 200                 $this->mTitle =& $title;
 201                 $this->mRevisionId = $revid;
 202                 $this->mOutputType = OT_HTML;
 203
 204                 $this->mStripState = NULL;
 205
 206                 //$text = $this->strip( $text, $this->mStripState );
 207                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 208                 $x =& $this->mStripState;
 209
 210                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 211                 $text = $this->strip( $text, $x );
 212                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 213
 214                 # Hook to suspend the parser in this state
 215                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 216                         wfProfileOut( $fname );
 217                         return $text ;
 218                 }
 219
 220                 $text = $this->internalParse( $text );
 221
 222                 $text = $this->unstrip( $text, $this->mStripState );
 223
 224                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 225                 $fixtags = array(
 226                         # french spaces, last one Guillemet-left
 227                         # only if there is something before the space
 228                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 229                         # french spaces, Guillemet-right
 230                         '/(\\302\\253) /' => '\\1&nbsp;',
 231                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 232                 );
 233                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 234
 235                 # only once and last
 236                 $text = $this->doBlockLevels( $text, $linestart );
 237
 238                 $this->replaceLinkHolders( $text );
 239
 240                 # the position of the parserConvert() call should not be changed. it
 241                 # assumes that the links are all replaced and the only thing left
 242                 # is the <nowiki> mark.
 243                 # Side-effects: this calls $this->mOutput->setTitleText()
 244                 $text = $wgContLang->parserConvert( $text, $this );
 245
 246                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 247
 248                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 249
 250                 $text = Sanitizer::normalizeCharReferences( $text );
 251
 252                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 253                         $text = Parser::tidy($text);
 254                 } else {
 255                         # attempt to sanitize at least some nesting problems
 256                         # (bug #2702 and quite a few others)
 257                         $tidyregs = array(
 258                                 # ''Something [http://www.cool.com cool''] -->
 259                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 260                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 261                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 262                                 # fix up an anchor inside another anchor, only
 263                                 # at least for a single single nested link (bug 3695)
 264                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 265                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 266                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 267                                 # contains a div, so fix it up here; replace
 268                                 # div with escaped text
 269                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 270                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 271                                 # remove empty italic or bold tag pairs, some
 272                                 # introduced by rules above
 273                                 '/<([bi])><\/\\1>/' => ''
 274                         );
 275
 276                         $text = preg_replace(
 277                                 array_keys( $tidyregs ),
 278                                 array_values( $tidyregs ),
 279                                 $text );
 280                 }
 281
 282                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 283
 284                 $this->mOutput->setText( $text );
 285                 wfProfileOut( $fname );
 286
 287                 return $this->mOutput;
 288         }
 289
 290         /**
 291          * Get a random string
 292          *
 293          * @access private
 294          * @static
 295          */
 296         function getRandomString() {
 297                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 298         }
 299
 300         function &getTitle() { return $this->mTitle; }
 301         function getOptions() { return $this->mOptions; }
 302
 303         /**
 304          * Replaces all occurrences of <$tag>content</$tag> in the text
 305          * with a random marker and returns the new text. the output parameter
 306          * $content will be an associative array filled with data on the form
 307          * $unique_marker => content.
 308          *
 309          * If $content is already set, the additional entries will be appended
 310          * If $tag is set to STRIP_COMMENTS, the function will extract
 311          * <!-- HTML comments -->
 312          *
 313          * @access private
 314          * @static
 315          */
 316         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 317                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 318                 if ( !$content ) {
 319                         $content = array( );
 320                 }
 321                 $n = 1;
 322                 $stripped = '';
 323
 324                 if ( !$tags ) {
 325                         $tags = array( );
 326                 }
 327
 328                 if ( !$params ) {
 329                         $params = array( );
 330                 }
 331
 332                 if( $tag == STRIP_COMMENTS ) {
 333                         $start = '/<!--()()/';
 334                         $end   = '/-->/';
 335                 } else {
 336                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 337                         $end   = "/<\\/$tag\\s*>/i";
 338                 }
 339
 340                 while ( '' != $text ) {
 341                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 342                         $stripped .= $p[0];
 343                         if( count( $p ) < 4 ) {
 344                                 break;
 345                         }
 346                         $attributes = $p[1];
 347                         $empty      = $p[2];
 348                         $inside     = $p[3];
 349
 350                         $marker = $rnd . sprintf('%08X', $n++);
 351                         $stripped .= $marker;
 352
 353                         $tags[$marker] = "<$tag$attributes$empty>";
 354                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 355
 356                         if ( $empty === '/' ) {
 357                                 // Empty element tag, <tag />
 358                                 $content[$marker] = null;
 359                                 $text = $inside;
 360                         } else {
 361                                 $q = preg_split( $end, $inside, 2 );
 362                                 $content[$marker] = $q[0];
 363                                 if( count( $q ) < 2 ) {
 364                                         # No end tag -- let it run out to the end of the text.
 365                                         break;
 366                                 } else {
 367                                         $text = $q[1];
 368                                 }
 369                         }
 370                 }
 371                 return $stripped;
 372         }
 373
 374         /**
 375          * Wrapper function for extractTagsAndParams
 376          * for cases where $tags and $params isn't needed
 377          * i.e. where tags will never have params, like <nowiki>
 378          *
 379          * @access private
 380          * @static
 381          */
 382         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 383                 $dummy_tags = array();
 384                 $dummy_params = array();
 385
 386                 return Parser::extractTagsAndParams( $tag, $text, $content,
 387                         $dummy_tags, $dummy_params, $uniq_prefix );
 388         }
 389
 390         /**
 391          * Strips and renders nowiki, pre, math, hiero
 392          * If $render is set, performs necessary rendering operations on plugins
 393          * Returns the text, and fills an array with data needed in unstrip()
 394          * If the $state is already a valid strip state, it adds to the state
 395          *
 396          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 397          *  will be stripped in addition to other tags. This is important
 398          *  for section editing, where these comments cause confusion when
 399          *  counting the sections in the wikisource
 400          *
 401          * @access private
 402          */
 403         function strip( $text, &$state, $stripcomments = false ) {
 404                 $render = ($this->mOutputType == OT_HTML);
 405                 $html_content = array();
 406                 $nowiki_content = array();
 407                 $math_content = array();
 408                 $pre_content = array();
 409                 $comment_content = array();
 410                 $ext_content = array();
 411                 $ext_tags = array();
 412                 $ext_params = array();
 413                 $gallery_content = array();
 414
 415                 # Replace any instances of the placeholders
 416                 $uniq_prefix = $this->mUniqPrefix;
 417                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 418
 419                 # html
 420                 global $wgRawHtml;
 421                 if( $wgRawHtml ) {
 422                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 423                         foreach( $html_content as $marker => $content ) {
 424                                 if ($render ) {
 425                                         # Raw and unchecked for validity.
 426                                         $html_content[$marker] = $content;
 427                                 } else {
 428                                         $html_content[$marker] = '<html>'.$content.'</html>';
 429                                 }
 430                         }
 431                 }
 432
 433                 # nowiki
 434                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 435                 foreach( $nowiki_content as $marker => $content ) {
 436                         if( $render ){
 437                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 438                         } else {
 439                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 440                         }
 441                 }
 442
 443                 # math
 444                 if( $this->mOptions->getUseTeX() ) {
 445                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 446                         foreach( $math_content as $marker => $content ){
 447                                 if( $render ) {
 448                                         $math_content[$marker] = renderMath( $content );
 449                                 } else {
 450                                         $math_content[$marker] = '<math>'.$content.'</math>';
 451                                 }
 452                         }
 453                 }
 454
 455                 # pre
 456                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 457                 foreach( $pre_content as $marker => $content ){
 458                         if( $render ){
 459                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 460                         } else {
 461                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 462                         }
 463                 }
 464
 465                 # gallery
 466                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 467                 foreach( $gallery_content as $marker => $content ) {
 468                         require_once( 'ImageGallery.php' );
 469                         if ( $render ) {
 470                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 471                         } else {
 472                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 473                         }
 474                 }
 475
 476                 # Comments
 477                 $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 478                 foreach( $comment_content as $marker => $content ){
 479                         $comment_content[$marker] = '<!--'.$content.'-->';
 480                 }
 481
 482                 # Extensions
 483                 foreach ( $this->mTagHooks as $tag => $callback ) {
 484                         $ext_content[$tag] = array();
 485                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 486                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 487                         foreach( $ext_content[$tag] as $marker => $content ) {
 488                                 $full_tag = $ext_tags[$tag][$marker];
 489                                 $params = $ext_params[$tag][$marker];
 490                                 if ( $render )
 491                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 492                                 else {
 493                                         if ( is_null( $content ) ) {
 494                                                 // Empty element tag
 495                                                 $ext_content[$tag][$marker] = $full_tag;
 496                                         } else {
 497                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 498                                         }
 499                                 }
 500                         }
 501                 }
 502
 503                 # Unstrip comments unless explicitly told otherwise.
 504                 # (The comments are always stripped prior to this point, so as to
 505                 # not invoke any extension tags / parser hooks contained within
 506                 # a comment.)
 507                 if ( !$stripcomments ) {
 508                         $tempstate = array( 'comment' => $comment_content );
 509                         $text = $this->unstrip( $text, $tempstate );
 510                         $comment_content = array();
 511                 }
 512
 513                 # Merge state with the pre-existing state, if there is one
 514                 if ( $state ) {
 515                         $state['html'] = $state['html'] + $html_content;
 516                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 517                         $state['math'] = $state['math'] + $math_content;
 518                         $state['pre'] = $state['pre'] + $pre_content;
 519                         $state['gallery'] = $state['gallery'] + $gallery_content;
 520                         $state['comment'] = $state['comment'] + $comment_content;
 521
 522                         foreach( $ext_content as $tag => $array ) {
 523                                 if ( array_key_exists( $tag, $state ) ) {
 524                                         $state[$tag] = $state[$tag] + $array;
 525                                 }
 526                         }
 527                 } else {
 528                         $state = array(
 529                           'html' => $html_content,
 530                           'nowiki' => $nowiki_content,
 531                           'math' => $math_content,
 532                           'pre' => $pre_content,
 533                           'gallery' => $gallery_content,
 534                           'comment' => $comment_content,
 535                         ) + $ext_content;
 536                 }
 537                 return $text;
 538         }
 539
 540         /**
 541          * restores pre, math, and hiero removed by strip()
 542          *
 543          * always call unstripNoWiki() after this one
 544          * @access private
 545          */
 546         function unstrip( $text, &$state ) {
 547                 if ( !is_array( $state ) ) {
 548                         return $text;
 549                 }
 550
 551                 # Must expand in reverse order, otherwise nested tags will be corrupted
 552                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 553                         if( $tag != 'nowiki' && $tag != 'html' ) {
 554                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 555                                         $text = str_replace( $uniq, $content, $text );
 556                                 }
 557                         }
 558                 }
 559
 560                 return $text;
 561         }
 562
 563         /**
 564          * always call this after unstrip() to preserve the order
 565          *
 566          * @access private
 567          */
 568         function unstripNoWiki( $text, &$state ) {
 569                 if ( !is_array( $state ) ) {
 570                         return $text;
 571                 }
 572
 573                 # Must expand in reverse order, otherwise nested tags will be corrupted
 574                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 575                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 576                 }
 577
 578                 global $wgRawHtml;
 579                 if ($wgRawHtml) {
 580                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 581                                 $text = str_replace( key( $state['html'] ), $content, $text );
 582                         }
 583                 }
 584
 585                 return $text;
 586         }
 587
 588         /**
 589          * Add an item to the strip state
 590          * Returns the unique tag which must be inserted into the stripped text
 591          * The tag will be replaced with the original text in unstrip()
 592          *
 593          * @access private
 594          */
 595         function insertStripItem( $text, &$state ) {
 596                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 597                 if ( !$state ) {
 598                         $state = array(
 599                           'html' => array(),
 600                           'nowiki' => array(),
 601                           'math' => array(),
 602                           'pre' => array(),
 603                           'comment' => array(),
 604                           'gallery' => array(),
 605                         );
 606                 }
 607                 $state['item'][$rnd] = $text;
 608                 return $rnd;
 609         }
 610
 611         /**
 612          * Interface with html tidy, used if $wgUseTidy = true.
 613          * If tidy isn't able to correct the markup, the original will be
 614          * returned in all its glory with a warning comment appended.
 615          *
 616          * Either the external tidy program or the in-process tidy extension
 617          * will be used depending on availability. Override the default
 618          * $wgTidyInternal setting to disable the internal if it's not working.
 619          *
 620          * @param string $text Hideous HTML input
 621          * @return string Corrected HTML output
 622          * @access public
 623          * @static
 624          */
 625         function tidy( $text ) {
 626                 global $wgTidyInternal;
 627                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 628 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 629 '<head><title>test</title></head><body>'.$text.'</body></html>';
 630                 if( $wgTidyInternal ) {
 631                         $correctedtext = Parser::internalTidy( $wrappedtext );
 632                 } else {
 633                         $correctedtext = Parser::externalTidy( $wrappedtext );
 634                 }
 635                 if( is_null( $correctedtext ) ) {
 636                         wfDebug( "Tidy error detected!\n" );
 637                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 638                 }
 639                 return $correctedtext;
 640         }
 641
 642         /**
 643          * Spawn an external HTML tidy process and get corrected markup back from it.
 644          *
 645          * @access private
 646          * @static
 647          */
 648         function externalTidy( $text ) {
 649                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 650                 $fname = 'Parser::externalTidy';
 651                 wfProfileIn( $fname );
 652
 653                 $cleansource = '';
 654                 $opts = ' -utf8';
 655
 656                 $descriptorspec = array(
 657                         0 => array('pipe', 'r'),
 658                         1 => array('pipe', 'w'),
 659                         2 => array('file', '/dev/null', 'a')
 660                 );
 661                 $pipes = array();
 662                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 663                 if (is_resource($process)) {
 664                         // Theoretically, this style of communication could cause a deadlock
 665                         // here. If the stdout buffer fills up, then writes to stdin could
 666                         // block. This doesn't appear to happen with tidy, because tidy only
 667                         // writes to stdout after it's finished reading from stdin. Search
 668                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 669                         fwrite($pipes[0], $text);
 670                         fclose($pipes[0]);
 671                         while (!feof($pipes[1])) {
 672                                 $cleansource .= fgets($pipes[1], 1024);
 673                         }
 674                         fclose($pipes[1]);
 675                         proc_close($process);
 676                 }
 677
 678                 wfProfileOut( $fname );
 679
 680                 if( $cleansource == '' && $text != '') {
 681                         // Some kind of error happened, so we couldn't get the corrected text.
 682                         // Just give up; we'll use the source text and append a warning.
 683                         return null;
 684                 } else {
 685                         return $cleansource;
 686                 }
 687         }
 688
 689         /**
 690          * Use the HTML tidy PECL extension to use the tidy library in-process,
 691          * saving the overhead of spawning a new process. Currently written to
 692          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 693          *
 694          * 'pear install tidy' should be able to compile the extension module.
 695          *
 696          * @access private
 697          * @static
 698          */
 699         function internalTidy( $text ) {
 700                 global $wgTidyConf;
 701                 $fname = 'Parser::internalTidy';
 702                 wfProfileIn( $fname );
 703
 704                 tidy_load_config( $wgTidyConf );
 705                 tidy_set_encoding( 'utf8' );
 706                 tidy_parse_string( $text );
 707                 tidy_clean_repair();
 708                 if( tidy_get_status() == 2 ) {
 709                         // 2 is magic number for fatal error
 710                         // http://www.php.net/manual/en/function.tidy-get-status.php
 711                         $cleansource = null;
 712                 } else {
 713                         $cleansource = tidy_get_output();
 714                 }
 715                 wfProfileOut( $fname );
 716                 return $cleansource;
 717         }
 718
 719         /**
 720          * parse the wiki syntax used to render tables
 721          *
 722          * @access private
 723          */
 724         function doTableStuff ( $t ) {
 725                 $fname = 'Parser::doTableStuff';
 726                 wfProfileIn( $fname );
 727
 728                 $t = explode ( "\n" , $t ) ;
 729                 $td = array () ; # Is currently a td tag open?
 730                 $ltd = array () ; # Was it TD or TH?
 731                 $tr = array () ; # Is currently a tr tag open?
 732                 $ltr = array () ; # tr attributes
 733                 $has_opened_tr = array(); # Did this table open a <tr> element?
 734                 $indent_level = 0; # indent level of the table
 735                 foreach ( $t AS $k => $x )
 736                 {
 737                         $x = trim ( $x ) ;
 738                         $fc = substr ( $x , 0 , 1 ) ;
 739                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 740                                 $indent_level = strlen( $matches[1] );
 741
 742                                 $attributes = $this->unstripForHTML( $matches[2] );
 743
 744                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 745                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 746                                 array_push ( $td , false ) ;
 747                                 array_push ( $ltd , '' ) ;
 748                                 array_push ( $tr , false ) ;
 749                                 array_push ( $ltr , '' ) ;
 750                                 array_push ( $has_opened_tr, false );
 751                         }
 752                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 753                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 754                                 $z = "</table>" . substr ( $x , 2);
 755                                 $l = array_pop ( $ltd ) ;
 756                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 757                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 758                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 759                                 array_pop ( $ltr ) ;
 760                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 761                         }
 762                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 763                                 $x = substr ( $x , 1 ) ;
 764                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 765                                 $z = '' ;
 766                                 $l = array_pop ( $ltd ) ;
 767                                 array_pop ( $has_opened_tr );
 768                                 array_push ( $has_opened_tr , true ) ;
 769                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 770                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 771                                 array_pop ( $ltr ) ;
 772                                 $t[$k] = $z ;
 773                                 array_push ( $tr , false ) ;
 774                                 array_push ( $td , false ) ;
 775                                 array_push ( $ltd , '' ) ;
 776                                 $attributes = $this->unstripForHTML( $x );
 777                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 778                         }
 779                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 780                                 # $x is a table row
 781                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 782                                         $fc = '+' ;
 783                                         $x = substr ( $x , 1 ) ;
 784                                 }
 785                                 $after = substr ( $x , 1 ) ;
 786                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 787
 788                                 // Split up multiple cells on the same line.
 789                                 // FIXME: This can result in improper nesting of tags processed
 790                                 // by earlier parser steps, but should avoid splitting up eg
 791                                 // attribute values containing literal "||".
 792                                 $after = wfExplodeMarkup( '||', $after );
 793
 794                                 $t[$k] = '' ;
 795
 796                                 # Loop through each table cell
 797                                 foreach ( $after AS $theline )
 798                                 {
 799                                         $z = '' ;
 800                                         if ( $fc != '+' )
 801                                         {
 802                                                 $tra = array_pop ( $ltr ) ;
 803                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 804                                                 array_push ( $tr , true ) ;
 805                                                 array_push ( $ltr , '' ) ;
 806                                                 array_pop ( $has_opened_tr );
 807                                                 array_push ( $has_opened_tr , true ) ;
 808                                         }
 809
 810                                         $l = array_pop ( $ltd ) ;
 811                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 812                                         if ( $fc == '|' ) $l = 'td' ;
 813                                         else if ( $fc == '!' ) $l = 'th' ;
 814                                         else if ( $fc == '+' ) $l = 'caption' ;
 815                                         else $l = '' ;
 816                                         array_push ( $ltd , $l ) ;
 817
 818                                         # Cell parameters
 819                                         $y = explode ( '|' , $theline , 2 ) ;
 820                                         # Note that a '|' inside an invalid link should not
 821                                         # be mistaken as delimiting cell parameters
 822                                         if ( strpos( $y[0], '[[' ) !== false ) {
 823                                                 $y = array ($theline);
 824                                         }
 825                                         if ( count ( $y ) == 1 )
 826                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 827                                         else {
 828                                                 $attributes = $this->unstripForHTML( $y[0] );
 829                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 830                                         }
 831                                         $t[$k] .= $y ;
 832                                         array_push ( $td , true ) ;
 833                                 }
 834                         }
 835                 }
 836
 837                 # Closing open td, tr && table
 838                 while ( count ( $td ) > 0 )
 839                 {
 840                         $l = array_pop ( $ltd ) ;
 841                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 842                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 843                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 844                         $t[] = '</table>' ;
 845                 }
 846
 847                 $t = implode ( "\n" , $t ) ;
 848                 # special case: don't return empty table
 849                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 850                         $t = '';
 851                 wfProfileOut( $fname );
 852                 return $t ;
 853         }
 854
 855         /**
 856          * Helper function for parse() that transforms wiki markup into
 857          * HTML. Only called for $mOutputType == OT_HTML.
 858          *
 859          * @access private
 860          */
 861         function internalParse( $text ) {
 862                 $args = array();
 863                 $isMain = true;
 864                 $fname = 'Parser::internalParse';
 865                 wfProfileIn( $fname );
 866
 867                 # Remove <noinclude> tags and <includeonly> sections
 868                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 869                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 870                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 871
 872                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 873                 $text = $this->replaceVariables( $text, $args );
 874
 875                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 876
 877                 $text = $this->doHeadings( $text );
 878                 if($this->mOptions->getUseDynamicDates()) {
 879                         $df =& DateFormatter::getInstance();
 880                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 881                 }
 882                 $text = $this->doAllQuotes( $text );
 883                 $text = $this->replaceInternalLinks( $text );
 884                 $text = $this->replaceExternalLinks( $text );
 885
 886                 # replaceInternalLinks may sometimes leave behind
 887                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 888                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 889
 890                 $text = $this->doMagicLinks( $text );
 891                 $text = $this->doTableStuff( $text );
 892                 $text = $this->formatHeadings( $text, $isMain );
 893
 894                 wfProfileOut( $fname );
 895                 return $text;
 896         }
 897
 898         /**
 899          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 900          * magic external links.
 901          *
 902          * @access private
 903          */
 904         function &doMagicLinks( &$text ) {
 905                 $text = $this->magicISBN( $text );
 906                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 907                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 908                 return $text;
 909         }
 910
 911         /**
 912          * Parse headers and return html
 913          *
 914          * @access private
 915          */
 916         function doHeadings( $text ) {
 917                 $fname = 'Parser::doHeadings';
 918                 wfProfileIn( $fname );
 919                 for ( $i = 6; $i >= 1; --$i ) {
 920                         $h = str_repeat( '=', $i );
 921                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 922                           "<h{$i}>\\1</h{$i}>\\2", $text );
 923                 }
 924                 wfProfileOut( $fname );
 925                 return $text;
 926         }
 927
 928         /**
 929          * Replace single quotes with HTML markup
 930          * @access private
 931          * @return string the altered text
 932          */
 933         function doAllQuotes( $text ) {
 934                 $fname = 'Parser::doAllQuotes';
 935                 wfProfileIn( $fname );
 936                 $outtext = '';
 937                 $lines = explode( "\n", $text );
 938                 foreach ( $lines as $line ) {
 939                         $outtext .= $this->doQuotes ( $line ) . "\n";
 940                 }
 941                 $outtext = substr($outtext, 0,-1);
 942                 wfProfileOut( $fname );
 943                 return $outtext;
 944         }
 945
 946         /**
 947          * Helper function for doAllQuotes()
 948          * @access private
 949          */
 950         function doQuotes( $text ) {
 951                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 952                 if ( count( $arr ) == 1 )
 953                         return $text;
 954                 else
 955                 {
 956                         # First, do some preliminary work. This may shift some apostrophes from
 957                         # being mark-up to being text. It also counts the number of occurrences
 958                         # of bold and italics mark-ups.
 959                         $i = 0;
 960                         $numbold = 0;
 961                         $numitalics = 0;
 962                         foreach ( $arr as $r )
 963                         {
 964                                 if ( ( $i % 2 ) == 1 )
 965                                 {
 966                                         # If there are ever four apostrophes, assume the first is supposed to
 967                                         # be text, and the remaining three constitute mark-up for bold text.
 968                                         if ( strlen( $arr[$i] ) == 4 )
 969                                         {
 970                                                 $arr[$i-1] .= "'";
 971                                                 $arr[$i] = "'''";
 972                                         }
 973                                         # If there are more than 5 apostrophes in a row, assume they're all
 974                                         # text except for the last 5.
 975                                         else if ( strlen( $arr[$i] ) > 5 )
 976                                         {
 977                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 978                                                 $arr[$i] = "'''''";
 979                                         }
 980                                         # Count the number of occurrences of bold and italics mark-ups.
 981                                         # We are not counting sequences of five apostrophes.
 982                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 983                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 984                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 985                                 }
 986                                 $i++;
 987                         }
 988
 989                         # If there is an odd number of both bold and italics, it is likely
 990                         # that one of the bold ones was meant to be an apostrophe followed
 991                         # by italics. Which one we cannot know for certain, but it is more
 992                         # likely to be one that has a single-letter word before it.
 993                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 994                         {
 995                                 $i = 0;
 996                                 $firstsingleletterword = -1;
 997                                 $firstmultiletterword = -1;
 998                                 $firstspace = -1;
 999                                 foreach ( $arr as $r )
1000                                 {
1001                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1002                                         {
1003                                                 $x1 = substr ($arr[$i-1], -1);
1004                                                 $x2 = substr ($arr[$i-1], -2, 1);
1005                                                 if ($x1 == ' ') {
1006                                                         if ($firstspace == -1) $firstspace = $i;
1007                                                 } else if ($x2 == ' ') {
1008                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1009                                                 } else {
1010                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1011                                                 }
1012                                         }
1013                                         $i++;
1014                                 }
1015
1016                                 # If there is a single-letter word, use it!
1017                                 if ($firstsingleletterword > -1)
1018                                 {
1019                                         $arr [ $firstsingleletterword ] = "''";
1020                                         $arr [ $firstsingleletterword-1 ] .= "'";
1021                                 }
1022                                 # If not, but there's a multi-letter word, use that one.
1023                                 else if ($firstmultiletterword > -1)
1024                                 {
1025                                         $arr [ $firstmultiletterword ] = "''";
1026                                         $arr [ $firstmultiletterword-1 ] .= "'";
1027                                 }
1028                                 # ... otherwise use the first one that has neither.
1029                                 # (notice that it is possible for all three to be -1 if, for example,
1030                                 # there is only one pentuple-apostrophe in the line)
1031                                 else if ($firstspace > -1)
1032                                 {
1033                                         $arr [ $firstspace ] = "''";
1034                                         $arr [ $firstspace-1 ] .= "'";
1035                                 }
1036                         }
1037
1038                         # Now let's actually convert our apostrophic mush to HTML!
1039                         $output = '';
1040                         $buffer = '';
1041                         $state = '';
1042                         $i = 0;
1043                         foreach ($arr as $r)
1044                         {
1045                                 if (($i % 2) == 0)
1046                                 {
1047                                         if ($state == 'both')
1048                                                 $buffer .= $r;
1049                                         else
1050                                                 $output .= $r;
1051                                 }
1052                                 else
1053                                 {
1054                                         if (strlen ($r) == 2)
1055                                         {
1056                                                 if ($state == 'i')
1057                                                 { $output .= '</i>'; $state = ''; }
1058                                                 else if ($state == 'bi')
1059                                                 { $output .= '</i>'; $state = 'b'; }
1060                                                 else if ($state == 'ib')
1061                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1062                                                 else if ($state == 'both')
1063                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1064                                                 else # $state can be 'b' or ''
1065                                                 { $output .= '<i>'; $state .= 'i'; }
1066                                         }
1067                                         else if (strlen ($r) == 3)
1068                                         {
1069                                                 if ($state == 'b')
1070                                                 { $output .= '</b>'; $state = ''; }
1071                                                 else if ($state == 'bi')
1072                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1073                                                 else if ($state == 'ib')
1074                                                 { $output .= '</b>'; $state = 'i'; }
1075                                                 else if ($state == 'both')
1076                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1077                                                 else # $state can be 'i' or ''
1078                                                 { $output .= '<b>'; $state .= 'b'; }
1079                                         }
1080                                         else if (strlen ($r) == 5)
1081                                         {
1082                                                 if ($state == 'b')
1083                                                 { $output .= '</b><i>'; $state = 'i'; }
1084                                                 else if ($state == 'i')
1085                                                 { $output .= '</i><b>'; $state = 'b'; }
1086                                                 else if ($state == 'bi')
1087                                                 { $output .= '</i></b>'; $state = ''; }
1088                                                 else if ($state == 'ib')
1089                                                 { $output .= '</b></i>'; $state = ''; }
1090                                                 else if ($state == 'both')
1091                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1092                                                 else # ($state == '')
1093                                                 { $buffer = ''; $state = 'both'; }
1094                                         }
1095                                 }
1096                                 $i++;
1097                         }
1098                         # Now close all remaining tags.  Notice that the order is important.
1099                         if ($state == 'b' || $state == 'ib')
1100                                 $output .= '</b>';
1101                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1102                                 $output .= '</i>';
1103                         if ($state == 'bi')
1104                                 $output .= '</b>';
1105                         if ($state == 'both')
1106                                 $output .= '<b><i>'.$buffer.'</i></b>';
1107                         return $output;
1108                 }
1109         }
1110
1111         /**
1112          * Replace external links
1113          *
1114          * Note: this is all very hackish and the order of execution matters a lot.
1115          * Make sure to run maintenance/parserTests.php if you change this code.
1116          *
1117          * @access private
1118          */
1119         function replaceExternalLinks( $text ) {
1120                 global $wgContLang;
1121                 $fname = 'Parser::replaceExternalLinks';
1122                 wfProfileIn( $fname );
1123
1124                 $sk =& $this->mOptions->getSkin();
1125
1126                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1127
1128                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1129
1130                 $i = 0;
1131                 while ( $i<count( $bits ) ) {
1132                         $url = $bits[$i++];
1133                         $protocol = $bits[$i++];
1134                         $text = $bits[$i++];
1135                         $trail = $bits[$i++];
1136
1137                         # The characters '<' and '>' (which were escaped by
1138                         # removeHTMLtags()) should not be included in
1139                         # URLs, per RFC 2396.
1140                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1141                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1142                                 $url = substr($url, 0, $m2[0][1]);
1143                         }
1144
1145                         # If the link text is an image URL, replace it with an <img> tag
1146                         # This happened by accident in the original parser, but some people used it extensively
1147                         $img = $this->maybeMakeExternalImage( $text );
1148                         if ( $img !== false ) {
1149                                 $text = $img;
1150                         }
1151
1152                         $dtrail = '';
1153
1154                         # Set linktype for CSS - if URL==text, link is essentially free
1155                         $linktype = ($text == $url) ? 'free' : 'text';
1156
1157                         # No link text, e.g. [http://domain.tld/some.link]
1158                         if ( $text == '' ) {
1159                                 # Autonumber if allowed
1160                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1161                                         $text = '[' . ++$this->mAutonumber . ']';
1162                                         $linktype = 'autonumber';
1163                                 } else {
1164                                         # Otherwise just use the URL
1165                                         $text = htmlspecialchars( $url );
1166                                         $linktype = 'free';
1167                                 }
1168                         } else {
1169                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1170                                 # Check for trail
1171                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1172                         }
1173
1174                         $text = $wgContLang->markNoConversion($text);
1175
1176                         # Replace &amp; from obsolete syntax with &.
1177                         # All HTML entities will be escaped by makeExternalLink()
1178                         $url = str_replace( '&amp;', '&', $url );
1179
1180                         # Process the trail (i.e. everything after this link up until start of the next link),
1181                         # replacing any non-bracketed links
1182                         $trail = $this->replaceFreeExternalLinks( $trail );
1183
1184                         # Use the encoded URL
1185                         # This means that users can paste URLs directly into the text
1186                         # Funny characters like &ouml; aren't valid in URLs anyway
1187                         # This was changed in August 2004
1188                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1189
1190                         # Register link in the output object.
1191                         # Replace unnecessary URL escape codes with the referenced character
1192                         # This prevents spammers from hiding links from the filters
1193                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1194                         $this->mOutput->addExternalLink( $pasteurized );
1195                 }
1196
1197                 wfProfileOut( $fname );
1198                 return $s;
1199         }
1200
1201         /**
1202          * Replace anything that looks like a URL with a link
1203          * @access private
1204          */
1205         function replaceFreeExternalLinks( $text ) {
1206                 global $wgContLang;
1207                 $fname = 'Parser::replaceFreeExternalLinks';
1208                 wfProfileIn( $fname );
1209
1210                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1211                 $s = array_shift( $bits );
1212                 $i = 0;
1213
1214                 $sk =& $this->mOptions->getSkin();
1215
1216                 while ( $i < count( $bits ) ){
1217                         $protocol = $bits[$i++];
1218                         $remainder = $bits[$i++];
1219
1220                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1221                                 # Found some characters after the protocol that look promising
1222                                 $url = $protocol . $m[1];
1223                                 $trail = $m[2];
1224
1225                                 # special case: handle urls as url args:
1226                                 # http://www.example.com/foo?=http://www.example.com/bar
1227                                 if(strlen($trail) == 0 &&
1228                                         isset($bits[$i]) &&
1229                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1230                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1231                                 {
1232                                         # add protocol, arg
1233                                         $url .= $bits[$i] . $bits[$i + 1]; # protocol, url as arg to previous link
1234                                         $i += 2;
1235                                         $trail = $m[2];
1236                                 }
1237
1238                                 # The characters '<' and '>' (which were escaped by
1239                                 # removeHTMLtags()) should not be included in
1240                                 # URLs, per RFC 2396.
1241                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1242                                         $trail = substr($url, $m2[0][1]) . $trail;
1243                                         $url = substr($url, 0, $m2[0][1]);
1244                                 }
1245
1246                                 # Move trailing punctuation to $trail
1247                                 $sep = ',;\.:!?';
1248                                 # If there is no left bracket, then consider right brackets fair game too
1249                                 if ( strpos( $url, '(' ) === false ) {
1250                                         $sep .= ')';
1251                                 }
1252
1253                                 $numSepChars = strspn( strrev( $url ), $sep );
1254                                 if ( $numSepChars ) {
1255                                         $trail = substr( $url, -$numSepChars ) . $trail;
1256                                         $url = substr( $url, 0, -$numSepChars );
1257                                 }
1258
1259                                 # Replace &amp; from obsolete syntax with &.
1260                                 # All HTML entities will be escaped by makeExternalLink()
1261                                 # or maybeMakeExternalImage()
1262                                 $url = str_replace( '&amp;', '&', $url );
1263
1264                                 # Is this an external image?
1265                                 $text = $this->maybeMakeExternalImage( $url );
1266                                 if ( $text === false ) {
1267                                         # Not an image, make a link
1268                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1269                                         # Register it in the output object...
1270                                         # Replace unnecessary URL escape codes with their equivalent characters
1271                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1272                                         $this->mOutput->addExternalLink( $pasteurized );
1273                                 }
1274                                 $s .= $text . $trail;
1275                         } else {
1276                                 $s .= $protocol . $remainder;
1277                         }
1278                 }
1279                 wfProfileOut( $fname );
1280                 return $s;
1281         }
1282
1283         /**
1284          * Replace unusual URL escape codes with their equivalent characters
1285          * @param string
1286          * @return string
1287          * @static
1288          * @fixme This can merge genuinely required bits in the path or query string,
1289          *        breaking legit URLs. A proper fix would treat the various parts of
1290          *        the URL differently; as a workaround, just use the output for
1291          *        statistical records, not for actual linking/output.
1292          */
1293         function replaceUnusualEscapes( $url ) {
1294                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1295                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1296         }
1297
1298         /**
1299          * Callback function used in replaceUnusualEscapes().
1300          * Replaces unusual URL escape codes with their equivalent character
1301          * @static
1302          * @access private
1303          */
1304         function replaceUnusualEscapesCallback( $matches ) {
1305                 $char = urldecode( $matches[0] );
1306                 $ord = ord( $char );
1307                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1308                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1309                         // No, shouldn't be escaped
1310                         return $char;
1311                 } else {
1312                         // Yes, leave it escaped
1313                         return $matches[0];
1314                 }
1315         }
1316
1317         /**
1318          * make an image if it's allowed, either through the global
1319          * option or through the exception
1320          * @access private
1321          */
1322         function maybeMakeExternalImage( $url ) {
1323                 $sk =& $this->mOptions->getSkin();
1324                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1325                 $imagesexception = !empty($imagesfrom);
1326                 $text = false;
1327                 if ( $this->mOptions->getAllowExternalImages()
1328                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1329                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1330                                 # Image found
1331                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1332                         }
1333                 }
1334                 return $text;
1335         }
1336
1337         /**
1338          * Process [[ ]] wikilinks
1339          *
1340          * @access private
1341          */
1342         function replaceInternalLinks( $s ) {
1343                 global $wgContLang;
1344                 static $fname = 'Parser::replaceInternalLinks' ;
1345
1346                 wfProfileIn( $fname );
1347
1348                 wfProfileIn( $fname.'-setup' );
1349                 static $tc = FALSE;
1350                 # the % is needed to support urlencoded titles as well
1351                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1352
1353                 $sk =& $this->mOptions->getSkin();
1354
1355                 #split the entire text string on occurences of [[
1356                 $a = explode( '[[', ' ' . $s );
1357                 #get the first element (all text up to first [[), and remove the space we added
1358                 $s = array_shift( $a );
1359                 $s = substr( $s, 1 );
1360
1361                 # Match a link having the form [[namespace:link|alternate]]trail
1362                 static $e1 = FALSE;
1363                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1364                 # Match cases where there is no "]]", which might still be images
1365                 static $e1_img = FALSE;
1366                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1367                 # Match the end of a line for a word that's not followed by whitespace,
1368                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1369                 $e2 = wfMsgForContent( 'linkprefix' );
1370
1371                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1372
1373                 if( is_null( $this->mTitle ) ) {
1374                         wfDebugDieBacktrace( 'nooo' );
1375                 }
1376                 $nottalk = !$this->mTitle->isTalkPage();
1377
1378                 if ( $useLinkPrefixExtension ) {
1379                         if ( preg_match( $e2, $s, $m ) ) {
1380                                 $first_prefix = $m[2];
1381                         } else {
1382                                 $first_prefix = false;
1383                         }
1384                 } else {
1385                         $prefix = '';
1386                 }
1387
1388                 $selflink = $this->mTitle->getPrefixedText();
1389                 wfProfileOut( $fname.'-setup' );
1390
1391                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1392                 $useSubpages = $this->areSubpagesAllowed();
1393
1394                 # Loop for each link
1395                 for ($k = 0; isset( $a[$k] ); $k++) {
1396                         $line = $a[$k];
1397                         if ( $useLinkPrefixExtension ) {
1398                                 wfProfileIn( $fname.'-prefixhandling' );
1399                                 if ( preg_match( $e2, $s, $m ) ) {
1400                                         $prefix = $m[2];
1401                                         $s = $m[1];
1402                                 } else {
1403                                         $prefix='';
1404                                 }
1405                                 # first link
1406                                 if($first_prefix) {
1407                                         $prefix = $first_prefix;
1408                                         $first_prefix = false;
1409                                 }
1410                                 wfProfileOut( $fname.'-prefixhandling' );
1411                         }
1412
1413                         $might_be_img = false;
1414
1415                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1416                                 $text = $m[2];
1417                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1418                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1419                                 # the real problem is with the $e1 regex
1420                                 # See bug 1300.
1421                                 #
1422                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1423                                 # and no image is in sight. See bug 2095.
1424                                 #
1425                                 if( $text !== '' &&
1426                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1427                                         strpos($text, '[') !== false
1428                                 )
1429                                 {
1430                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1431                                         $m[3] = $n[1];
1432                                 }
1433                                 # fix up urlencoded title texts
1434                                 if(preg_match('/%/', $m[1] ))
1435                                         # Should anchors '#' also be rejected?
1436                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1437                                 $trail = $m[3];
1438                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1439                                 $might_be_img = true;
1440                                 $text = $m[2];
1441                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1442                                 $trail = "";
1443                         } else { # Invalid form; output directly
1444                                 $s .= $prefix . '[[' . $line ;
1445                                 continue;
1446                         }
1447
1448                         # Don't allow internal links to pages containing
1449                         # PROTO: where PROTO is a valid URL protocol; these
1450                         # should be external links.
1451                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1452                                 $s .= $prefix . '[[' . $line ;
1453                                 continue;
1454                         }
1455
1456                         # Make subpage if necessary
1457                         if( $useSubpages ) {
1458                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1459                         } else {
1460                                 $link = $m[1];
1461                         }
1462
1463                         $noforce = (substr($m[1], 0, 1) != ':');
1464                         if (!$noforce) {
1465                                 # Strip off leading ':'
1466                                 $link = substr($link, 1);
1467                         }
1468
1469                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1470                         if( !$nt ) {
1471                                 $s .= $prefix . '[[' . $line;
1472                                 continue;
1473                         }
1474
1475                         #check other language variants of the link
1476                         #if the article does not exist
1477                         if( $checkVariantLink
1478                             && $nt->getArticleID() == 0 ) {
1479                                 $wgContLang->findVariantLink($link, $nt);
1480                         }
1481
1482                         $ns = $nt->getNamespace();
1483                         $iw = $nt->getInterWiki();
1484
1485                         if ($might_be_img) { # if this is actually an invalid link
1486                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1487                                         $found = false;
1488                                         while (isset ($a[$k+1]) ) {
1489                                                 #look at the next 'line' to see if we can close it there
1490                                                 $spliced = array_splice( $a, $k + 1, 1 );
1491                                                 $next_line = array_shift( $spliced );
1492                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1493                                                 # the first ]] closes the inner link, the second the image
1494                                                         $found = true;
1495                                                         $text .= '[[' . $m[1];
1496                                                         $trail = $m[2];
1497                                                         break;
1498                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1499                                                         #if there's exactly one ]] that's fine, we'll keep looking
1500                                                         $text .= '[[' . $m[0];
1501                                                 } else {
1502                                                         #if $next_line is invalid too, we need look no further
1503                                                         $text .= '[[' . $next_line;
1504                                                         break;
1505                                                 }
1506                                         }
1507                                         if ( !$found ) {
1508                                                 # we couldn't find the end of this imageLink, so output it raw
1509                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1510                                                 $text = $this->replaceInternalLinks($text);
1511                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1512                                                 # note: no $trail, because without an end, there *is* no trail
1513                                                 continue;
1514                                         }
1515                                 } else { #it's not an image, so output it raw
1516                                         $s .= $prefix . '[[' . $link . '|' . $text;
1517                                         # note: no $trail, because without an end, there *is* no trail
1518                                         continue;
1519                                 }
1520                         }
1521
1522                         $wasblank = ( '' == $text );
1523                         if( $wasblank ) $text = $link;
1524
1525
1526                         # Link not escaped by : , create the various objects
1527                         if( $noforce ) {
1528
1529                                 # Interwikis
1530                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1531                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1532                                         $s = rtrim($s . "\n");
1533                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1534                                         continue;
1535                                 }
1536
1537                                 if ( $ns == NS_IMAGE ) {
1538                                         wfProfileIn( "$fname-image" );
1539                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1540                                                 # recursively parse links inside the image caption
1541                                                 # actually, this will parse them in any other parameters, too,
1542                                                 # but it might be hard to fix that, and it doesn't matter ATM
1543                                                 $text = $this->replaceExternalLinks($text);
1544                                                 $text = $this->replaceInternalLinks($text);
1545
1546                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1547                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1548                                                 $this->mOutput->addImage( $nt->getDBkey() );
1549
1550                                                 wfProfileOut( "$fname-image" );
1551                                                 continue;
1552                                         }
1553                                         wfProfileOut( "$fname-image" );
1554
1555                                 }
1556
1557                                 if ( $ns == NS_CATEGORY ) {
1558                                         wfProfileIn( "$fname-category" );
1559                                         $s = rtrim($s . "\n"); # bug 87
1560
1561                                         if ( $wasblank ) {
1562                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1563                                                         $sortkey = $this->mTitle->getText();
1564                                                 } else {
1565                                                         $sortkey = $this->mTitle->getPrefixedText();
1566                                                 }
1567                                         } else {
1568                                                 $sortkey = $text;
1569                                         }
1570                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1571                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1572                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1573
1574                                         /**
1575                                          * Strip the whitespace Category links produce, see bug 87
1576                                          * @todo We might want to use trim($tmp, "\n") here.
1577                                          */
1578                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1579
1580                                         wfProfileOut( "$fname-category" );
1581                                         continue;
1582                                 }
1583                         }
1584
1585                         if( ( $nt->getPrefixedText() === $selflink ) &&
1586                             ( $nt->getFragment() === '' ) ) {
1587                                 # Self-links are handled specially; generally de-link and change to bold.
1588                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1589                                 continue;
1590                         }
1591
1592                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1593                         if( $ns == NS_MEDIA ) {
1594                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1595                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1596                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1597                                 $this->mOutput->addImage( $nt->getDBkey() );
1598                                 continue;
1599                         } elseif( $ns == NS_SPECIAL ) {
1600                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1601                                 continue;
1602                         } elseif( $ns == NS_IMAGE ) {
1603                                 $img = Image::newFromTitle( $nt );
1604                                 if( $img->exists() ) {
1605                                         // Force a blue link if the file exists; may be a remote
1606                                         // upload on the shared repository, and we want to see its
1607                                         // auto-generated page.
1608                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1609                                         continue;
1610                                 }
1611                         }
1612                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1613                 }
1614                 wfProfileOut( $fname );
1615                 return $s;
1616         }
1617
1618         /**
1619          * Make a link placeholder. The text returned can be later resolved to a real link with
1620          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1621          * parsing of interwiki links, and secondly to allow all extistence checks and
1622          * article length checks (for stub links) to be bundled into a single query.
1623          *
1624          */
1625         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1626                 if ( ! is_object($nt) ) {
1627                         # Fail gracefully
1628                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1629                 } else {
1630                         # Separate the link trail from the rest of the link
1631                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1632
1633                         if ( $nt->isExternal() ) {
1634                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1635                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1636                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1637                         } else {
1638                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1639                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1640                                 $this->mLinkHolders['queries'][] = $query;
1641                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1642                                 $this->mLinkHolders['titles'][] = $nt;
1643
1644                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1645                         }
1646                 }
1647                 return $retVal;
1648         }
1649
1650         /**
1651          * Render a forced-blue link inline; protect against double expansion of
1652          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1653          * Since this little disaster has to split off the trail text to avoid
1654          * breaking URLs in the following text without breaking trails on the
1655          * wiki links, it's been made into a horrible function.
1656          *
1657          * @param Title $nt
1658          * @param string $text
1659          * @param string $query
1660          * @param string $trail
1661          * @param string $prefix
1662          * @return string HTML-wikitext mix oh yuck
1663          */
1664         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1665                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1666                 $sk =& $this->mOptions->getSkin();
1667                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1668                 return $this->armorLinks( $link ) . $trail;
1669         }
1670
1671         /**
1672          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1673          * going to go through further parsing steps before inline URL expansion.
1674          *
1675          * In particular this is important when using action=render, which causes
1676          * full URLs to be included.
1677          *
1678          * Oh man I hate our multi-layer parser!
1679          *
1680          * @param string more-or-less HTML
1681          * @return string less-or-more HTML with NOPARSE bits
1682          */
1683         function armorLinks( $text ) {
1684                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1685                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1686         }
1687
1688         /**
1689          * Return true if subpage links should be expanded on this page.
1690          * @return bool
1691          */
1692         function areSubpagesAllowed() {
1693                 # Some namespaces don't allow subpages
1694                 global $wgNamespacesWithSubpages;
1695                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1696         }
1697
1698         /**
1699          * Handle link to subpage if necessary
1700          * @param string $target the source of the link
1701          * @param string &$text the link text, modified as necessary
1702          * @return string the full name of the link
1703          * @access private
1704          */
1705         function maybeDoSubpageLink($target, &$text) {
1706                 # Valid link forms:
1707                 # Foobar -- normal
1708                 # :Foobar -- override special treatment of prefix (images, language links)
1709                 # /Foobar -- convert to CurrentPage/Foobar
1710                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1711                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1712                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1713
1714                 $fname = 'Parser::maybeDoSubpageLink';
1715                 wfProfileIn( $fname );
1716                 $ret = $target; # default return value is no change
1717
1718                 # Some namespaces don't allow subpages,
1719                 # so only perform processing if subpages are allowed
1720                 if( $this->areSubpagesAllowed() ) {
1721                         # Look at the first character
1722                         if( $target != '' && $target{0} == '/' ) {
1723                                 # / at end means we don't want the slash to be shown
1724                                 if( substr( $target, -1, 1 ) == '/' ) {
1725                                         $target = substr( $target, 1, -1 );
1726                                         $noslash = $target;
1727                                 } else {
1728                                         $noslash = substr( $target, 1 );
1729                                 }
1730
1731                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1732                                 if( '' === $text ) {
1733                                         $text = $target;
1734                                 } # this might be changed for ugliness reasons
1735                         } else {
1736                                 # check for .. subpage backlinks
1737                                 $dotdotcount = 0;
1738                                 $nodotdot = $target;
1739                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1740                                         ++$dotdotcount;
1741                                         $nodotdot = substr( $nodotdot, 3 );
1742                                 }
1743                                 if($dotdotcount > 0) {
1744                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1745                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1746                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1747                                                 # / at the end means don't show full path
1748                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1749                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1750                                                         if( '' === $text ) {
1751                                                                 $text = $nodotdot;
1752                                                         }
1753                                                 }
1754                                                 $nodotdot = trim( $nodotdot );
1755                                                 if( $nodotdot != '' ) {
1756                                                         $ret .= '/' . $nodotdot;
1757                                                 }
1758                                         }
1759                                 }
1760                         }
1761                 }
1762
1763                 wfProfileOut( $fname );
1764                 return $ret;
1765         }
1766
1767         /**#@+
1768          * Used by doBlockLevels()
1769          * @access private
1770          */
1771         /* private */ function closeParagraph() {
1772                 $result = '';
1773                 if ( '' != $this->mLastSection ) {
1774                         $result = '</' . $this->mLastSection  . ">\n";
1775                 }
1776                 $this->mInPre = false;
1777                 $this->mLastSection = '';
1778                 return $result;
1779         }
1780         # getCommon() returns the length of the longest common substring
1781         # of both arguments, starting at the beginning of both.
1782         #
1783         /* private */ function getCommon( $st1, $st2 ) {
1784                 $fl = strlen( $st1 );
1785                 $shorter = strlen( $st2 );
1786                 if ( $fl < $shorter ) { $shorter = $fl; }
1787
1788                 for ( $i = 0; $i < $shorter; ++$i ) {
1789                         if ( $st1{$i} != $st2{$i} ) { break; }
1790                 }
1791                 return $i;
1792         }
1793         # These next three functions open, continue, and close the list
1794         # element appropriate to the prefix character passed into them.
1795         #
1796         /* private */ function openList( $char ) {
1797                 $result = $this->closeParagraph();
1798
1799                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1800                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1801                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1802                 else if ( ';' == $char ) {
1803                         $result .= '<dl><dt>';
1804                         $this->mDTopen = true;
1805                 }
1806                 else { $result = '<!-- ERR 1 -->'; }
1807
1808                 return $result;
1809         }
1810
1811         /* private */ function nextItem( $char ) {
1812                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1813                 else if ( ':' == $char || ';' == $char ) {
1814                         $close = '</dd>';
1815                         if ( $this->mDTopen ) { $close = '</dt>'; }
1816                         if ( ';' == $char ) {
1817                                 $this->mDTopen = true;
1818                                 return $close . '<dt>';
1819                         } else {
1820                                 $this->mDTopen = false;
1821                                 return $close . '<dd>';
1822                         }
1823                 }
1824                 return '<!-- ERR 2 -->';
1825         }
1826
1827         /* private */ function closeList( $char ) {
1828                 if ( '*' == $char ) { $text = '</li></ul>'; }
1829                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1830                 else if ( ':' == $char ) {
1831                         if ( $this->mDTopen ) {
1832                                 $this->mDTopen = false;
1833                                 $text = '</dt></dl>';
1834                         } else {
1835                                 $text = '</dd></dl>';
1836                         }
1837                 }
1838                 else {  return '<!-- ERR 3 -->'; }
1839                 return $text."\n";
1840         }
1841         /**#@-*/
1842
1843         /**
1844          * Make lists from lines starting with ':', '*', '#', etc.
1845          *
1846          * @access private
1847          * @return string the lists rendered as HTML
1848          */
1849         function doBlockLevels( $text, $linestart ) {
1850                 $fname = 'Parser::doBlockLevels';
1851                 wfProfileIn( $fname );
1852
1853                 # Parsing through the text line by line.  The main thing
1854                 # happening here is handling of block-level elements p, pre,
1855                 # and making lists from lines starting with * # : etc.
1856                 #
1857                 $textLines = explode( "\n", $text );
1858
1859                 $lastPrefix = $output = '';
1860                 $this->mDTopen = $inBlockElem = false;
1861                 $prefixLength = 0;
1862                 $paragraphStack = false;
1863
1864                 if ( !$linestart ) {
1865                         $output .= array_shift( $textLines );
1866                 }
1867                 foreach ( $textLines as $oLine ) {
1868                         $lastPrefixLength = strlen( $lastPrefix );
1869                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1870                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1871                         if ( !$this->mInPre ) {
1872                                 # Multiple prefixes may abut each other for nested lists.
1873                                 $prefixLength = strspn( $oLine, '*#:;' );
1874                                 $pref = substr( $oLine, 0, $prefixLength );
1875
1876                                 # eh?
1877                                 $pref2 = str_replace( ';', ':', $pref );
1878                                 $t = substr( $oLine, $prefixLength );
1879                                 $this->mInPre = !empty($preOpenMatch);
1880                         } else {
1881                                 # Don't interpret any other prefixes in preformatted text
1882                                 $prefixLength = 0;
1883                                 $pref = $pref2 = '';
1884                                 $t = $oLine;
1885                         }
1886
1887                         # List generation
1888                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1889                                 # Same as the last item, so no need to deal with nesting or opening stuff
1890                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1891                                 $paragraphStack = false;
1892
1893                                 if ( substr( $pref, -1 ) == ';') {
1894                                         # The one nasty exception: definition lists work like this:
1895                                         # ; title : definition text
1896                                         # So we check for : in the remainder text to split up the
1897                                         # title and definition, without b0rking links.
1898                                         $term = $t2 = '';
1899                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1900                                                 $t = $t2;
1901                                                 $output .= $term . $this->nextItem( ':' );
1902                                         }
1903                                 }
1904                         } elseif( $prefixLength || $lastPrefixLength ) {
1905                                 # Either open or close a level...
1906                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1907                                 $paragraphStack = false;
1908
1909                                 while( $commonPrefixLength < $lastPrefixLength ) {
1910                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1911                                         --$lastPrefixLength;
1912                                 }
1913                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1914                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1915                                 }
1916                                 while ( $prefixLength > $commonPrefixLength ) {
1917                                         $char = substr( $pref, $commonPrefixLength, 1 );
1918                                         $output .= $this->openList( $char );
1919
1920                                         if ( ';' == $char ) {
1921                                                 # FIXME: This is dupe of code above
1922                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1923                                                         $t = $t2;
1924                                                         $output .= $term . $this->nextItem( ':' );
1925                                                 }
1926                                         }
1927                                         ++$commonPrefixLength;
1928                                 }
1929                                 $lastPrefix = $pref2;
1930                         }
1931                         if( 0 == $prefixLength ) {
1932                                 wfProfileIn( "$fname-paragraph" );
1933                                 # No prefix (not in list)--go to paragraph mode
1934                                 // XXX: use a stack for nestable elements like span, table and div
1935                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1936                                 $closematch = preg_match(
1937                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1938                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1939                                 if ( $openmatch or $closematch ) {
1940                                         $paragraphStack = false;
1941                                         $output .= $this->closeParagraph();
1942                                         if ( $preOpenMatch and !$preCloseMatch ) {
1943                                                 $this->mInPre = true;
1944                                         }
1945                                         if ( $closematch ) {
1946                                                 $inBlockElem = false;
1947                                         } else {
1948                                                 $inBlockElem = true;
1949                                         }
1950                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1951                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1952                                                 // pre
1953                                                 if ($this->mLastSection != 'pre') {
1954                                                         $paragraphStack = false;
1955                                                         $output .= $this->closeParagraph().'<pre>';
1956                                                         $this->mLastSection = 'pre';
1957                                                 }
1958                                                 $t = substr( $t, 1 );
1959                                         } else {
1960                                                 // paragraph
1961                                                 if ( '' == trim($t) ) {
1962                                                         if ( $paragraphStack ) {
1963                                                                 $output .= $paragraphStack.'<br />';
1964                                                                 $paragraphStack = false;
1965                                                                 $this->mLastSection = 'p';
1966                                                         } else {
1967                                                                 if ($this->mLastSection != 'p' ) {
1968                                                                         $output .= $this->closeParagraph();
1969                                                                         $this->mLastSection = '';
1970                                                                         $paragraphStack = '<p>';
1971                                                                 } else {
1972                                                                         $paragraphStack = '</p><p>';
1973                                                                 }
1974                                                         }
1975                                                 } else {
1976                                                         if ( $paragraphStack ) {
1977                                                                 $output .= $paragraphStack;
1978                                                                 $paragraphStack = false;
1979                                                                 $this->mLastSection = 'p';
1980                                                         } else if ($this->mLastSection != 'p') {
1981                                                                 $output .= $this->closeParagraph().'<p>';
1982                                                                 $this->mLastSection = 'p';
1983                                                         }
1984                                                 }
1985                                         }
1986                                 }
1987                                 wfProfileOut( "$fname-paragraph" );
1988                         }
1989                         // somewhere above we forget to get out of pre block (bug 785)
1990                         if($preCloseMatch && $this->mInPre) {
1991                                 $this->mInPre = false;
1992                         }
1993                         if ($paragraphStack === false) {
1994                                 $output .= $t."\n";
1995                         }
1996                 }
1997                 while ( $prefixLength ) {
1998                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1999                         --$prefixLength;
2000                 }
2001                 if ( '' != $this->mLastSection ) {
2002                         $output .= '</' . $this->mLastSection . '>';
2003                         $this->mLastSection = '';
2004                 }
2005
2006                 wfProfileOut( $fname );
2007                 return $output;
2008         }
2009
2010         /**
2011          * Split up a string on ':', ignoring any occurences inside
2012          * <a>..</a> or <span>...</span>
2013          * @param string $str the string to split
2014          * @param string &$before set to everything before the ':'
2015          * @param string &$after set to everything after the ':'
2016          * return string the position of the ':', or false if none found
2017          */
2018         function findColonNoLinks($str, &$before, &$after) {
2019                 # I wonder if we should make this count all tags, not just <a>
2020                 # and <span>. That would prevent us from matching a ':' that
2021                 # comes in the middle of italics other such formatting....
2022                 # -- Wil
2023                 $fname = 'Parser::findColonNoLinks';
2024                 wfProfileIn( $fname );
2025                 $pos = 0;
2026                 do {
2027                         $colon = strpos($str, ':', $pos);
2028
2029                         if ($colon !== false) {
2030                                 $before = substr($str, 0, $colon);
2031                                 $after = substr($str, $colon + 1);
2032
2033                                 # Skip any ':' within <a> or <span> pairs
2034                                 $a = substr_count($before, '<a');
2035                                 $s = substr_count($before, '<span');
2036                                 $ca = substr_count($before, '</a>');
2037                                 $cs = substr_count($before, '</span>');
2038
2039                                 if ($a <= $ca and $s <= $cs) {
2040                                         # Tags are balanced before ':'; ok
2041                                         break;
2042                                 }
2043                                 $pos = $colon + 1;
2044                         }
2045                 } while ($colon !== false);
2046                 wfProfileOut( $fname );
2047                 return $colon;
2048         }
2049
2050         /**
2051          * Return value of a magic variable (like PAGENAME)
2052          *
2053          * @access private
2054          */
2055         function getVariableValue( $index ) {
2056                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2057
2058                 /**
2059                  * Some of these require message or data lookups and can be
2060                  * expensive to check many times.
2061                  */
2062                 static $varCache = array();
2063                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2064                         if ( isset( $varCache[$index] ) )
2065                                 return $varCache[$index];
2066
2067                 $ts = time();
2068                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2069
2070                 switch ( $index ) {
2071                         case MAG_CURRENTMONTH:
2072                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2073                         case MAG_CURRENTMONTHNAME:
2074                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2075                         case MAG_CURRENTMONTHNAMEGEN:
2076                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2077                         case MAG_CURRENTMONTHABBREV:
2078                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2079                         case MAG_CURRENTDAY:
2080                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2081                         case MAG_CURRENTDAY2:
2082                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2083                         case MAG_PAGENAME:
2084                                 return $this->mTitle->getText();
2085                         case MAG_PAGENAMEE:
2086                                 return $this->mTitle->getPartialURL();
2087                         case MAG_FULLPAGENAME:
2088                                 return $this->mTitle->getPrefixedText();
2089                         case MAG_FULLPAGENAMEE:
2090                                 return $this->mTitle->getPrefixedURL();
2091                         case MAG_SUBPAGENAME:
2092                                 return $this->mTitle->getSubpageText();
2093                         case MAG_SUBPAGENAMEE:
2094                                 return $this->mTitle->getSubpageUrlForm();
2095                         case MAG_REVISIONID:
2096                                 return $this->mRevisionId;
2097                         case MAG_NAMESPACE:
2098                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2099                         case MAG_NAMESPACEE:
2100                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2101                         case MAG_CURRENTDAYNAME:
2102                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2103                         case MAG_CURRENTYEAR:
2104                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2105                         case MAG_CURRENTTIME:
2106                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2107                         case MAG_CURRENTWEEK:
2108                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2109                                 // int to remove the padding
2110                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2111                         case MAG_CURRENTDOW:
2112                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2113                         case MAG_NUMBEROFARTICLES:
2114                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2115                         case MAG_NUMBEROFFILES:
2116                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2117                         case MAG_SITENAME:
2118                                 return $wgSitename;
2119                         case MAG_SERVER:
2120                                 return $wgServer;
2121                         case MAG_SERVERNAME:
2122                                 return $wgServerName;
2123                         case MAG_SCRIPTPATH:
2124                                 return $wgScriptPath;
2125                         default:
2126                                 $ret = null;
2127                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2128                                         return $ret;
2129                                 else
2130                                         return null;
2131                 }
2132         }
2133
2134         /**
2135          * initialise the magic variables (like CURRENTMONTHNAME)
2136          *
2137          * @access private
2138          */
2139         function initialiseVariables() {
2140                 $fname = 'Parser::initialiseVariables';
2141                 wfProfileIn( $fname );
2142                 global $wgVariableIDs;
2143                 $this->mVariables = array();
2144                 foreach ( $wgVariableIDs as $id ) {
2145                         $mw =& MagicWord::get( $id );
2146                         $mw->addToArray( $this->mVariables, $id );
2147                 }
2148                 wfProfileOut( $fname );
2149         }
2150
2151         /**
2152          * parse any parentheses in format ((title|part|part))
2153          * and call callbacks to get a replacement text for any found piece
2154          *
2155          * @param string $text The text to parse
2156          * @param array $callbacks rules in form:
2157          *     '{' => array(                            # opening parentheses
2158          *                                      'end' => '}',   # closing parentheses
2159          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2160          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2161          *                                                                )
2162          *                                      )
2163          * @access private
2164          */
2165         function replace_callback ($text, $callbacks) {
2166                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2167                 $lastOpeningBrace = -1;                 # last not closed parentheses
2168
2169                 for ($i = 0; $i < strlen($text); $i++) {
2170                         # check for any opening brace
2171                         $rule = null;
2172                         $nextPos = -1;
2173                         foreach ($callbacks as $key => $value) {
2174                                 $pos = strpos ($text, $key, $i);
2175                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2176                                         $rule = $value;
2177                                         $nextPos = $pos;
2178                                 }
2179                         }
2180
2181                         if ($lastOpeningBrace >= 0) {
2182                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2183
2184                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2185                                         $rule = null;
2186                                         $nextPos = $pos;
2187                                 }
2188
2189                                 $pos = strpos ($text, '|', $i);
2190
2191                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2192                                         $rule = null;
2193                                         $nextPos = $pos;
2194                                 }
2195                         }
2196
2197                         if ($nextPos == -1)
2198                                 break;
2199
2200                         $i = $nextPos;
2201
2202                         # found openning brace, lets add it to parentheses stack
2203                         if (null != $rule) {
2204                                 $piece = array('brace' => $text[$i],
2205                                                            'braceEnd' => $rule['end'],
2206                                                            'count' => 1,
2207                                                            'title' => '',
2208                                                            'parts' => null);
2209
2210                                 # count openning brace characters
2211                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2212                                         $piece['count']++;
2213                                         $i++;
2214                                 }
2215
2216                                 $piece['startAt'] = $i+1;
2217                                 $piece['partStart'] = $i+1;
2218
2219                                 # we need to add to stack only if openning brace count is enough for any given rule
2220                                 foreach ($rule['cb'] as $cnt => $fn) {
2221                                         if ($piece['count'] >= $cnt) {
2222                                                 $lastOpeningBrace ++;
2223                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2224                                                 break;
2225                                         }
2226                                 }
2227
2228                                 continue;
2229                         }
2230                         else if ($lastOpeningBrace >= 0) {
2231                                 # first check if it is a closing brace
2232                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2233                                         # lets check if it is enough characters for closing brace
2234                                         $count = 1;
2235                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2236                                                 $count++;
2237
2238                                         # if there are more closing parentheses than opening ones, we parse less
2239                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2240                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2241
2242                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2243                                         $matchingCount = 0;
2244                                         $matchingCallback = null;
2245                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2246                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2247                                                         $matchingCount = $cnt;
2248                                                         $matchingCallback = $fn;
2249                                                 }
2250                                         }
2251
2252                                         if ($matchingCount == 0) {
2253                                                 $i += $count - 1;
2254                                                 continue;
2255                                         }
2256
2257                                         # lets set a title or last part (if '|' was found)
2258                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2259                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2260                                         else
2261                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2262
2263                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2264                                         $pieceEnd = $i + $matchingCount;
2265
2266                                         if( is_callable( $matchingCallback ) ) {
2267                                                 $cbArgs = array (
2268                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2269                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2270                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2271                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2272                                                                                  );
2273                                                 # finally we can call a user callback and replace piece of text
2274                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2275                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2276                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2277                                         }
2278                                         else {
2279                                                 # null value for callback means that parentheses should be parsed, but not replaced
2280                                                 $i += $matchingCount - 1;
2281                                         }
2282
2283                                         # reset last openning parentheses, but keep it in case there are unused characters
2284                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2285                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2286                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2287                                                                    'title' => '',
2288                                                                    'parts' => null,
2289                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2290                                         $openingBraceStack[$lastOpeningBrace--] = null;
2291
2292                                         if ($matchingCount < $piece['count']) {
2293                                                 $piece['count'] -= $matchingCount;
2294                                                 $piece['startAt'] -= $matchingCount;
2295                                                 $piece['partStart'] = $piece['startAt'];
2296                                                 # do we still qualify for any callback with remaining count?
2297                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2298                                                         if ($piece['count'] >= $cnt) {
2299                                                                 $lastOpeningBrace ++;
2300                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2301                                                                 break;
2302                                                         }
2303                                                 }
2304                                         }
2305                                         continue;
2306                                 }
2307
2308                                 # lets set a title if it is a first separator, or next part otherwise
2309                                 if ($text[$i] == '|') {
2310                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2311                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2312                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2313                                         }
2314                                         else
2315                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2316
2317                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2318                                 }
2319                         }
2320                 }
2321
2322                 return $text;
2323         }
2324
2325         /**
2326          * Replace magic variables, templates, and template arguments
2327          * with the appropriate text. Templates are substituted recursively,
2328          * taking care to avoid infinite loops.
2329          *
2330          * Note that the substitution depends on value of $mOutputType:
2331          *  OT_WIKI: only {{subst:}} templates
2332          *  OT_MSG: only magic variables
2333          *  OT_HTML: all templates and magic variables
2334          *
2335          * @param string $tex The text to transform
2336          * @param array $args Key-value pairs representing template parameters to substitute
2337          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2338          * @access private
2339          */
2340         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2341                 # Prevent too big inclusions
2342                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2343                         return $text;
2344                 }
2345
2346                 $fname = 'Parser::replaceVariables';
2347                 wfProfileIn( $fname );
2348
2349                 # This function is called recursively. To keep track of arguments we need a stack:
2350                 array_push( $this->mArgStack, $args );
2351
2352                 $braceCallbacks = array();
2353                 if ( !$argsOnly ) {
2354                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2355                 }
2356                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2357                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2358                 }
2359                 $callbacks = array();
2360                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2361                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2362                 $text = $this->replace_callback ($text, $callbacks);
2363
2364                 array_pop( $this->mArgStack );
2365
2366                 wfProfileOut( $fname );
2367                 return $text;
2368         }
2369
2370         /**
2371          * Replace magic variables
2372          * @access private
2373          */
2374         function variableSubstitution( $matches ) {
2375                 $fname = 'Parser::variableSubstitution';
2376                 $varname = $matches[1];
2377                 wfProfileIn( $fname );
2378                 if ( !$this->mVariables ) {
2379                         $this->initialiseVariables();
2380                 }
2381                 $skip = false;
2382                 if ( $this->mOutputType == OT_WIKI ) {
2383                         # Do only magic variables prefixed by SUBST
2384                         $mwSubst =& MagicWord::get( MAG_SUBST );
2385                         if (!$mwSubst->matchStartAndRemove( $varname ))
2386                                 $skip = true;
2387                         # Note that if we don't substitute the variable below,
2388                         # we don't remove the {{subst:}} magic word, in case
2389                         # it is a template rather than a magic variable.
2390                 }
2391                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2392                         $id = $this->mVariables[$varname];
2393                         $text = $this->getVariableValue( $id );
2394                         $this->mOutput->mContainsOldMagic = true;
2395                 } else {
2396                         $text = $matches[0];
2397                 }
2398                 wfProfileOut( $fname );
2399                 return $text;
2400         }
2401
2402         # Split template arguments
2403         function getTemplateArgs( $argsString ) {
2404                 if ( $argsString === '' ) {
2405                         return array();
2406                 }
2407
2408                 $args = explode( '|', substr( $argsString, 1 ) );
2409
2410                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2411                 # merged with the next arg because the '|' character between belongs
2412                 # to the link syntax and not the template parameter syntax.
2413                 $argc = count($args);
2414
2415                 for ( $i = 0; $i < $argc-1; $i++ ) {
2416                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2417                                 $args[$i] .= '|'.$args[$i+1];
2418                                 array_splice($args, $i+1, 1);
2419                                 $i--;
2420                                 $argc--;
2421                         }
2422                 }
2423
2424                 return $args;
2425         }
2426
2427         /**
2428          * Return the text of a template, after recursively
2429          * replacing any variables or templates within the template.
2430          *
2431          * @param array $piece The parts of the template
2432          *  $piece['text']: matched text
2433          *  $piece['title']: the title, i.e. the part before the |
2434          *  $piece['parts']: the parameter array
2435          * @return string the text of the template
2436          * @access private
2437          */
2438         function braceSubstitution( $piece ) {
2439                 global $wgContLang;
2440                 $fname = 'Parser::braceSubstitution';
2441                 wfProfileIn( $fname );
2442
2443                 # Flags
2444                 $found = false;             # $text has been filled
2445                 $nowiki = false;            # wiki markup in $text should be escaped
2446                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2447                 $noargs = false;            # Don't replace triple-brace arguments in $text
2448                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2449                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2450                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2451
2452                 # Title object, where $text came from
2453                 $title = NULL;
2454
2455                 $linestart = '';
2456
2457                 # $part1 is the bit before the first |, and must contain only title characters
2458                 # $args is a list of arguments, starting from index 0, not including $part1
2459
2460                 $part1 = $piece['title'];
2461                 # If the third subpattern matched anything, it will start with |
2462
2463                 if (null == $piece['parts']) {
2464                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2465                         if ($replaceWith != $piece['text']) {
2466                                 $text = $replaceWith;
2467                                 $found = true;
2468                                 $noparse = true;
2469                                 $noargs = true;
2470                         }
2471                 }
2472
2473                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2474                 $argc = count( $args );
2475
2476                 # SUBST
2477                 if ( !$found ) {
2478                         $mwSubst =& MagicWord::get( MAG_SUBST );
2479                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2480                                 # One of two possibilities is true:
2481                                 # 1) Found SUBST but not in the PST phase
2482                                 # 2) Didn't find SUBST and in the PST phase
2483                                 # In either case, return without further processing
2484                                 $text = $piece['text'];
2485                                 $found = true;
2486                                 $noparse = true;
2487                                 $noargs = true;
2488                         }
2489                 }
2490
2491                 # MSG, MSGNW, INT and RAW
2492                 if ( !$found ) {
2493                         # Check for MSGNW:
2494                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2495                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2496                                 $nowiki = true;
2497                         } else {
2498                                 # Remove obsolete MSG:
2499                                 $mwMsg =& MagicWord::get( MAG_MSG );
2500                                 $mwMsg->matchStartAndRemove( $part1 );
2501                         }
2502
2503                         # Check for RAW:
2504                         $mwRaw =& MagicWord::get( MAG_RAW );
2505                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2506                                 $forceRawInterwiki = true;
2507                         }
2508
2509                         # Check if it is an internal message
2510                         $mwInt =& MagicWord::get( MAG_INT );
2511                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2512                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2513                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2514                                         $found = true;
2515                                 }
2516                         }
2517                 }
2518
2519                 # NS
2520                 if ( !$found ) {
2521                         # Check for NS: (namespace expansion)
2522                         $mwNs = MagicWord::get( MAG_NS );
2523                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2524                                 if ( intval( $part1 ) || $part1 == "0" ) {
2525                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2526                                         $found = true;
2527                                 } else {
2528                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2529                                         if ( !is_null( $index ) ) {
2530                                                 $text = $linestart . $wgContLang->getNsText( $index );
2531                                                 $found = true;
2532                                         }
2533                                 }
2534                         }
2535                 }
2536
2537                 # LCFIRST, UCFIRST, LC and UC
2538                 if ( !$found ) {
2539                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2540                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2541                         $lc =& MagicWord::get( MAG_LC );
2542                         $uc =& MagicWord::get( MAG_UC );
2543                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2544                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2545                                 $found = true;
2546                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2547                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2548                                 $found = true;
2549                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2550                                 $text = $linestart . $wgContLang->lc( $part1 );
2551                                 $found = true;
2552                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2553                                  $text = $linestart . $wgContLang->uc( $part1 );
2554                                  $found = true;
2555                         }
2556                 }
2557
2558                 # LOCALURL and FULLURL
2559                 if ( !$found ) {
2560                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2561                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2562                         $mwFull =& MagicWord::get( MAG_FULLURL );
2563                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2564
2565
2566                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2567                                 $func = 'getLocalURL';
2568                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2569                                 $func = 'escapeLocalURL';
2570                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2571                                 $func = 'getFullURL';
2572                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2573                                 $func = 'escapeFullURL';
2574                         } else {
2575                                 $func = false;
2576                         }
2577
2578                         if ( $func !== false ) {
2579                                 $title = Title::newFromText( $part1 );
2580                                 if ( !is_null( $title ) ) {
2581                                         if ( $argc > 0 ) {
2582                                                 $text = $linestart . $title->$func( $args[0] );
2583                                         } else {
2584                                                 $text = $linestart . $title->$func();
2585                                         }
2586                                         $found = true;
2587                                 }
2588                         }
2589                 }
2590
2591                 # GRAMMAR
2592                 if ( !$found && $argc == 1 ) {
2593                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2594                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2595                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2596                                 $found = true;
2597                         }
2598                 }
2599
2600                 # PLURAL
2601                 if ( !$found && $argc >= 2 ) {
2602                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2603                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2604                                 if ($argc==2) {$args[2]=$args[1];}
2605                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2606                                 $found = true;
2607                         }
2608                 }
2609
2610                 # Extensions
2611                 if ( !$found ) {
2612                         $colonPos = strpos( $part1, ':' );
2613                         if ( $colonPos !== false ) {
2614                                 $function = strtolower( substr( $part1, 0, $colonPos ) );
2615                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2616                                         $funcArgs = $args;
2617                                         array_unshift( $funcArgs, &$this, substr( $part1, $colonPos + 1 ) );
2618                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2619                                         $found = true;
2620                                         if ( is_array( $result ) ) {
2621                                                 $text = $linestart . $result[0];
2622                                                 unset( $result[0] );
2623
2624                                                 // Extract flags into the local scope
2625                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2626                                                 extract( $result );
2627                                         } else {
2628                                                 $text = $linestart . $result;
2629                                         }
2630                                 }
2631                         }
2632                 }
2633
2634                 # Template table test
2635
2636                 # Did we encounter this template already? If yes, it is in the cache
2637                 # and we need to check for loops.
2638                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2639                         $found = true;
2640
2641                         # Infinite loop test
2642                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2643                                 $noparse = true;
2644                                 $noargs = true;
2645                                 $found = true;
2646                                 $text = $linestart .
2647                                         '{{' . $part1 . '}}' .
2648                                         '<!-- WARNING: template loop detected -->';
2649                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2650                         } else {
2651                                 # set $text to cached message.
2652                                 $text = $linestart . $this->mTemplates[$piece['title']];
2653                         }
2654                 }
2655
2656                 # Load from database
2657                 $lastPathLevel = $this->mTemplatePath;
2658                 if ( !$found ) {
2659                         $ns = NS_TEMPLATE;
2660                         # declaring $subpage directly in the function call
2661                         # does not work correctly with references and breaks
2662                         # {{/subpage}}-style inclusions
2663                         $subpage = '';
2664                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2665                         if ($subpage !== '') {
2666                                 $ns = $this->mTitle->getNamespace();
2667                         }
2668                         $title = Title::newFromText( $part1, $ns );
2669
2670                         if ( !is_null( $title ) ) {
2671                                 if ( !$title->isExternal() ) {
2672                                         # Check for excessive inclusion
2673                                         $dbk = $title->getPrefixedDBkey();
2674                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2675                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2676                                                         # Capture special page output
2677                                                         $text = SpecialPage::capturePath( $title );
2678                                                         if ( is_string( $text ) ) {
2679                                                                 $found = true;
2680                                                                 $noparse = true;
2681                                                                 $noargs = true;
2682                                                                 $isHTML = true;
2683                                                                 $this->disableCache();
2684                                                         }
2685                                                 } else {
2686                                                         $articleContent = $this->fetchTemplate( $title );
2687                                                         if ( $articleContent !== false ) {
2688                                                                 $found = true;
2689                                                                 $text = $articleContent;
2690                                                                 $replaceHeadings = true;
2691                                                         }
2692                                                 }
2693                                         }
2694
2695                                         # If the title is valid but undisplayable, make a link to it
2696                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2697                                                 $text = '[['.$title->getPrefixedText().']]';
2698                                                 $found = true;
2699                                         }
2700                                 } elseif ( $title->isTrans() ) {
2701                                         // Interwiki transclusion
2702                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2703                                                 $text = $this->interwikiTransclude( $title, 'render' );
2704                                                 $isHTML = true;
2705                                                 $noparse = true;
2706                                         } else {
2707                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2708                                                 $replaceHeadings = true;
2709                                         }
2710                                         $found = true;
2711                                 }
2712
2713                                 # Template cache array insertion
2714                                 # Use the original $piece['title'] not the mangled $part1, so that
2715                                 # modifiers such as RAW: produce separate cache entries
2716                                 if( $found ) {
2717                                         $this->mTemplates[$piece['title']] = $text;
2718                                         $text = $linestart . $text;
2719                                 }
2720                         }
2721                 }
2722
2723                 # Recursive parsing, escaping and link table handling
2724                 # Only for HTML output
2725                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2726                         $text = wfEscapeWikiText( $text );
2727                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2728                         if ( !$noargs ) {
2729                                 # Clean up argument array
2730                                 $assocArgs = array();
2731                                 $index = 1;
2732                                 foreach( $args as $arg ) {
2733                                         $eqpos = strpos( $arg, '=' );
2734                                         if ( $eqpos === false ) {
2735                                                 $assocArgs[$index++] = $arg;
2736                                         } else {
2737                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2738                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2739                                                 if ( $value === false ) {
2740                                                         $value = '';
2741                                                 }
2742                                                 if ( $name !== false ) {
2743                                                         $assocArgs[$name] = $value;
2744                                                 }
2745                                         }
2746                                 }
2747
2748                                 # Add a new element to the templace recursion path
2749                                 $this->mTemplatePath[$part1] = 1;
2750                         }
2751
2752                         if ( !$noparse ) {
2753                                 # If there are any <onlyinclude> tags, only include them
2754                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2755                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2756                                         $text = '';
2757                                         foreach ($m[1] as $piece)
2758                                                 $text .= $piece;
2759                                 }
2760                                 # Remove <noinclude> sections and <includeonly> tags
2761                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2762                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2763
2764                                 if( $this->mOutputType == OT_HTML ) {
2765                                         # Strip <nowiki>, <pre>, etc.
2766                                         $text = $this->strip( $text, $this->mStripState );
2767                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2768                                 }
2769                                 $text = $this->replaceVariables( $text, $assocArgs );
2770
2771                                 # If the template begins with a table or block-level
2772                                 # element, it should be treated as beginning a new line.
2773                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2774                                         $text = "\n" . $text;
2775                                 }
2776                         } elseif ( !$noargs ) {
2777                                 # $noparse and !$noargs
2778                                 # Just replace the arguments, not any double-brace items
2779                                 # This is used for rendered interwiki transclusion
2780                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2781                         }
2782                 }
2783                 # Prune lower levels off the recursion check path
2784                 $this->mTemplatePath = $lastPathLevel;
2785
2786                 if ( !$found ) {
2787                         wfProfileOut( $fname );
2788                         return $piece['text'];
2789                 } else {
2790                         if ( $isHTML ) {
2791                                 # Replace raw HTML by a placeholder
2792                                 # Add a blank line preceding, to prevent it from mucking up
2793                                 # immediately preceding headings
2794                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2795                         } else {
2796                                 # replace ==section headers==
2797                                 # XXX this needs to go away once we have a better parser.
2798                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2799                                         if( !is_null( $title ) )
2800                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2801                                         else
2802                                                 $encodedname = base64_encode("");
2803                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2804                                                 PREG_SPLIT_DELIM_CAPTURE);
2805                                         $text = '';
2806                                         $nsec = 0;
2807                                         for( $i = 0; $i < count($m); $i += 2 ) {
2808                                                 $text .= $m[$i];
2809                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2810                                                 $hl = $m[$i + 1];
2811                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2812                                                         $text .= $hl;
2813                                                         continue;
2814                                                 }
2815                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2816                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2817                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2818
2819                                                 $nsec++;
2820                                         }
2821                                 }
2822                         }
2823                 }
2824
2825                 # Prune lower levels off the recursion check path
2826                 $this->mTemplatePath = $lastPathLevel;
2827
2828                 if ( !$found ) {
2829                         wfProfileOut( $fname );
2830                         return $piece['text'];
2831                 } else {
2832                         wfProfileOut( $fname );
2833                         return $text;
2834                 }
2835         }
2836
2837         /**
2838          * Fetch the unparsed text of a template and register a reference to it.
2839          */
2840         function fetchTemplate( $title ) {
2841                 $text = false;
2842                 // Loop to fetch the article, with up to 1 redirect
2843                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2844                         $rev = Revision::newFromTitle( $title );
2845                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2846                         if ( !$rev ) {
2847                                 break;
2848                         }
2849                         $text = $rev->getText();
2850                         if ( $text === false ) {
2851                                 break;
2852                         }
2853                         // Redirect?
2854                         $title = Title::newFromRedirect( $text );
2855                 }
2856                 return $text;
2857         }
2858
2859         /**
2860          * Transclude an interwiki link.
2861          */
2862         function interwikiTransclude( $title, $action ) {
2863                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2864
2865                 if (!$wgEnableScaryTranscluding)
2866                         return wfMsg('scarytranscludedisabled');
2867
2868                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2869                 // But we'll handle it generally anyway
2870                 if ( $title->getNamespace() ) {
2871                         // Use the canonical namespace, which should work anywhere
2872                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2873                 } else {
2874                         $articleName = $title->getDBkey();
2875                 }
2876
2877                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2878                 $url .= "?action=$action";
2879                 if (strlen($url) > 255)
2880                         return wfMsg('scarytranscludetoolong');
2881                 return $this->fetchScaryTemplateMaybeFromCache($url);
2882         }
2883
2884         function fetchScaryTemplateMaybeFromCache($url) {
2885                 global $wgTranscludeCacheExpiry;
2886                 $dbr =& wfGetDB(DB_SLAVE);
2887                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2888                                 array('tc_url' => $url));
2889                 if ($obj) {
2890                         $time = $obj->tc_time;
2891                         $text = $obj->tc_contents;
2892                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2893                                 return $text;
2894                         }
2895                 }
2896
2897                 $text = wfGetHTTP($url);
2898                 if (!$text)
2899                         return wfMsg('scarytranscludefailed', $url);
2900
2901                 $dbw =& wfGetDB(DB_MASTER);
2902                 $dbw->replace('transcache', array('tc_url'), array(
2903                         'tc_url' => $url,
2904                         'tc_time' => time(),
2905                         'tc_contents' => $text));
2906                 return $text;
2907         }
2908
2909
2910         /**
2911          * Triple brace replacement -- used for template arguments
2912          * @access private
2913          */
2914         function argSubstitution( $matches ) {
2915                 $arg = trim( $matches['title'] );
2916                 $text = $matches['text'];
2917                 $inputArgs = end( $this->mArgStack );
2918
2919                 if ( array_key_exists( $arg, $inputArgs ) ) {
2920                         $text = $inputArgs[$arg];
2921                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2922                         $text = $matches['parts'][0];
2923                 }
2924
2925                 return $text;
2926         }
2927
2928         /**
2929          * Returns true if the function is allowed to include this entity
2930          * @access private
2931          */
2932         function incrementIncludeCount( $dbk ) {
2933                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2934                         $this->mIncludeCount[$dbk] = 0;
2935                 }
2936                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2937                         return true;
2938                 } else {
2939                         return false;
2940                 }
2941         }
2942
2943         /**
2944          * This function accomplishes several tasks:
2945          * 1) Auto-number headings if that option is enabled
2946          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2947          * 3) Add a Table of contents on the top for users who have enabled the option
2948          * 4) Auto-anchor headings
2949          *
2950          * It loops through all headlines, collects the necessary data, then splits up the
2951          * string and re-inserts the newly formatted headlines.
2952          *
2953          * @param string $text
2954          * @param boolean $isMain
2955          * @access private
2956          */
2957         function formatHeadings( $text, $isMain=true ) {
2958                 global $wgMaxTocLevel, $wgContLang;
2959
2960                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2961                 $doShowToc = true;
2962                 $forceTocHere = false;
2963                 if( !$this->mTitle->userCanEdit() ) {
2964                         $showEditLink = 0;
2965                 } else {
2966                         $showEditLink = $this->mOptions->getEditSection();
2967                 }
2968
2969                 # Inhibit editsection links if requested in the page
2970                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2971                 if( $esw->matchAndRemove( $text ) ) {
2972                         $showEditLink = 0;
2973                 }
2974                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2975                 # do not add TOC
2976                 $mw =& MagicWord::get( MAG_NOTOC );
2977                 if( $mw->matchAndRemove( $text ) ) {
2978                         $doShowToc = false;
2979                 }
2980
2981                 # Get all headlines for numbering them and adding funky stuff like [edit]
2982                 # links - this is for later, but we need the number of headlines right now
2983                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2984
2985                 # if there are fewer than 4 headlines in the article, do not show TOC
2986                 if( $numMatches < 4 ) {
2987                         $doShowToc = false;
2988                 }
2989
2990                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2991                 # override above conditions and always show TOC at that place
2992
2993                 $mw =& MagicWord::get( MAG_TOC );
2994                 if($mw->match( $text ) ) {
2995                         $doShowToc = true;
2996                         $forceTocHere = true;
2997                 } else {
2998                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2999                         # override above conditions and always show TOC above first header
3000                         $mw =& MagicWord::get( MAG_FORCETOC );
3001                         if ($mw->matchAndRemove( $text ) ) {
3002                                 $doShowToc = true;
3003                         }
3004                 }
3005
3006                 # Never ever show TOC if no headers
3007                 if( $numMatches < 1 ) {
3008                         $doShowToc = false;
3009                 }
3010
3011                 # We need this to perform operations on the HTML
3012                 $sk =& $this->mOptions->getSkin();
3013
3014                 # headline counter
3015                 $headlineCount = 0;
3016                 $sectionCount = 0; # headlineCount excluding template sections
3017
3018                 # Ugh .. the TOC should have neat indentation levels which can be
3019                 # passed to the skin functions. These are determined here
3020                 $toc = '';
3021                 $full = '';
3022                 $head = array();
3023                 $sublevelCount = array();
3024                 $levelCount = array();
3025                 $toclevel = 0;
3026                 $level = 0;
3027                 $prevlevel = 0;
3028                 $toclevel = 0;
3029                 $prevtoclevel = 0;
3030
3031                 foreach( $matches[3] as $headline ) {
3032                         $istemplate = 0;
3033                         $templatetitle = '';
3034                         $templatesection = 0;
3035                         $numbering = '';
3036
3037                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3038                                 $istemplate = 1;
3039                                 $templatetitle = base64_decode($mat[1]);
3040                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3041                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3042                         }
3043
3044                         if( $toclevel ) {
3045                                 $prevlevel = $level;
3046                                 $prevtoclevel = $toclevel;
3047                         }
3048                         $level = $matches[1][$headlineCount];
3049
3050                         if( $doNumberHeadings || $doShowToc ) {
3051
3052                                 if ( $level > $prevlevel ) {
3053                                         # Increase TOC level
3054                                         $toclevel++;
3055                                         $sublevelCount[$toclevel] = 0;
3056                                         $toc .= $sk->tocIndent();
3057                                 }
3058                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3059                                         # Decrease TOC level, find level to jump to
3060
3061                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3062                                                 # Can only go down to level 1
3063                                                 $toclevel = 1;
3064                                         } else {
3065                                                 for ($i = $toclevel; $i > 0; $i--) {
3066                                                         if ( $levelCount[$i] == $level ) {
3067                                                                 # Found last matching level
3068                                                                 $toclevel = $i;
3069                                                                 break;
3070                                                         }
3071                                                         elseif ( $levelCount[$i] < $level ) {
3072                                                                 # Found first matching level below current level
3073                                                                 $toclevel = $i + 1;
3074                                                                 break;
3075                                                         }
3076                                                 }
3077                                         }
3078
3079                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3080                                 }
3081                                 else {
3082                                         # No change in level, end TOC line
3083                                         $toc .= $sk->tocLineEnd();
3084                                 }
3085
3086                                 $levelCount[$toclevel] = $level;
3087
3088                                 # count number of headlines for each level
3089                                 @$sublevelCount[$toclevel]++;
3090                                 $dot = 0;
3091                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3092                                         if( !empty( $sublevelCount[$i] ) ) {
3093                                                 if( $dot ) {
3094                                                         $numbering .= '.';
3095                                                 }
3096                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3097                                                 $dot = 1;
3098                                         }
3099                                 }
3100                         }
3101
3102                         # The canonized header is a version of the header text safe to use for links
3103                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3104                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3105                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3106
3107                         # Remove link placeholders by the link text.
3108                         #     <!--LINK number-->
3109                         # turns into
3110                         #     link text with suffix
3111                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3112                                                             "\$this->mLinkHolders['texts'][\$1]",
3113                                                             $canonized_headline );
3114                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3115                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3116                                                             $canonized_headline );
3117
3118                         # strip out HTML
3119                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3120                         $tocline = trim( $canonized_headline );
3121                         # Save headline for section edit hint before it's escaped
3122                         $headline_hint = trim( $canonized_headline );
3123                         $canonized_headline = Sanitizer::escapeId( $tocline );
3124                         $refers[$headlineCount] = $canonized_headline;
3125
3126                         # count how many in assoc. array so we can track dupes in anchors
3127                         @$refers[$canonized_headline]++;
3128                         $refcount[$headlineCount]=$refers[$canonized_headline];
3129
3130                         # Don't number the heading if it is the only one (looks silly)
3131                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3132                                 # the two are different if the line contains a link
3133                                 $headline=$numbering . ' ' . $headline;
3134                         }
3135
3136                         # Create the anchor for linking from the TOC to the section
3137                         $anchor = $canonized_headline;
3138                         if($refcount[$headlineCount] > 1 ) {
3139                                 $anchor .= '_' . $refcount[$headlineCount];
3140                         }
3141                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3142                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3143                         }
3144                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3145                                 if ( empty( $head[$headlineCount] ) ) {
3146                                         $head[$headlineCount] = '';
3147                                 }
3148                                 if( $istemplate )
3149                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3150                                 else
3151                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3152                         }
3153
3154                         # give headline the correct <h#> tag
3155                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3156
3157                         $headlineCount++;
3158                         if( !$istemplate )
3159                                 $sectionCount++;
3160                 }
3161
3162                 if( $doShowToc ) {
3163                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3164                         $toc = $sk->tocList( $toc );
3165                 }
3166
3167                 # split up and insert constructed headlines
3168
3169                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3170                 $i = 0;
3171
3172                 foreach( $blocks as $block ) {
3173                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3174                                 # This is the [edit] link that appears for the top block of text when
3175                                 # section editing is enabled
3176
3177                                 # Disabled because it broke block formatting
3178                                 # For example, a bullet point in the top line
3179                                 # $full .= $sk->editSectionLink(0);
3180                         }
3181                         $full .= $block;
3182                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3183                         # Top anchor now in skin
3184                                 $full = $full.$toc;
3185                         }
3186
3187                         if( !empty( $head[$i] ) ) {
3188                                 $full .= $head[$i];
3189                         }
3190                         $i++;
3191                 }
3192                 if($forceTocHere) {
3193                         $mw =& MagicWord::get( MAG_TOC );
3194                         return $mw->replace( $toc, $full );
3195                 } else {
3196                         return $full;
3197                 }
3198         }
3199
3200         /**
3201          * Return an HTML link for the "ISBN 123456" text
3202          * @access private
3203          */
3204         function magicISBN( $text ) {
3205                 $fname = 'Parser::magicISBN';
3206                 wfProfileIn( $fname );
3207
3208                 $a = split( 'ISBN ', ' '.$text );
3209                 if ( count ( $a ) < 2 ) {
3210                         wfProfileOut( $fname );
3211                         return $text;
3212                 }
3213                 $text = substr( array_shift( $a ), 1);
3214                 $valid = '0123456789-Xx';
3215
3216                 foreach ( $a as $x ) {
3217                         # hack: don't replace inside thumbnail title/alt
3218                         # attributes
3219                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3220                                 $text .= "ISBN $x";
3221                                 continue;
3222                         }
3223
3224                         $isbn = $blank = '' ;
3225                         while ( ' ' == $x{0} ) {
3226                                 $blank .= ' ';
3227                                 $x = substr( $x, 1 );
3228                         }
3229                         if ( $x == '' ) { # blank isbn
3230                                 $text .= "ISBN $blank";
3231                                 continue;
3232                         }
3233                         while ( strstr( $valid, $x{0} ) != false ) {
3234                                 $isbn .= $x{0};
3235                                 $x = substr( $x, 1 );
3236                         }
3237                         $num = str_replace( '-', '', $isbn );
3238                         $num = str_replace( ' ', '', $num );
3239                         $num = str_replace( 'x', 'X', $num );
3240
3241                         if ( '' == $num ) {
3242                                 $text .= "ISBN $blank$x";
3243                         } else {
3244                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3245                                 $text .= '<a href="' .
3246                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3247                                         "\" class=\"internal\">ISBN $isbn</a>";
3248                                 $text .= $x;
3249                         }
3250                 }
3251                 wfProfileOut( $fname );
3252                 return $text;
3253         }
3254
3255         /**
3256          * Return an HTML link for the "RFC 1234" text
3257          *
3258          * @access private
3259          * @param string $text     Text to be processed
3260          * @param string $keyword  Magic keyword to use (default RFC)
3261          * @param string $urlmsg   Interface message to use (default rfcurl)
3262          * @return string
3263          */
3264         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3265
3266                 $valid = '0123456789';
3267                 $internal = false;
3268
3269                 $a = split( $keyword, ' '.$text );
3270                 if ( count ( $a ) < 2 ) {
3271                         return $text;
3272                 }
3273                 $text = substr( array_shift( $a ), 1);
3274
3275                 /* Check if keyword is preceed by [[.
3276                  * This test is made here cause of the array_shift above
3277                  * that prevent the test to be done in the foreach.
3278                  */
3279                 if ( substr( $text, -2 ) == '[[' ) {
3280                         $internal = true;
3281                 }
3282
3283                 foreach ( $a as $x ) {
3284                         /* token might be empty if we have RFC RFC 1234 */
3285                         if ( $x=='' ) {
3286                                 $text.=$keyword;
3287                                 continue;
3288                                 }
3289
3290                         # hack: don't replace inside thumbnail title/alt
3291                         # attributes
3292                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3293                                 $text .= $keyword . $x;
3294                                 continue;
3295                         }
3296
3297                         $id = $blank = '' ;
3298
3299                         /** remove and save whitespaces in $blank */
3300                         while ( $x{0} == ' ' ) {
3301                                 $blank .= ' ';
3302                                 $x = substr( $x, 1 );
3303                         }
3304
3305                         /** remove and save the rfc number in $id */
3306                         while ( strstr( $valid, $x{0} ) != false ) {
3307                                 $id .= $x{0};
3308                                 $x = substr( $x, 1 );
3309                         }
3310
3311                         if ( $id == '' ) {
3312                                 /* call back stripped spaces*/
3313                                 $text .= $keyword.$blank.$x;
3314                         } elseif( $internal ) {
3315                                 /* normal link */
3316                                 $text .= $keyword.$id.$x;
3317                         } else {
3318                                 /* build the external link*/
3319                                 $url = wfMsg( $urlmsg, $id);
3320                                 $sk =& $this->mOptions->getSkin();
3321                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3322                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3323                         }
3324
3325                         /* Check if the next RFC keyword is preceed by [[ */
3326                         $internal = ( substr($x,-2) == '[[' );
3327                 }
3328                 return $text;
3329         }
3330
3331         /**
3332          * Transform wiki markup when saving a page by doing \r\n -> \n
3333          * conversion, substitting signatures, {{subst:}} templates, etc.
3334          *
3335          * @param string $text the text to transform
3336          * @param Title &$title the Title object for the current article
3337          * @param User &$user the User object describing the current user
3338          * @param ParserOptions $options parsing options
3339          * @param bool $clearState whether to clear the parser state first
3340          * @return string the altered wiki markup
3341          * @access public
3342          */
3343         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3344                 $this->mOptions = $options;
3345                 $this->mTitle =& $title;
3346                 $this->mOutputType = OT_WIKI;
3347
3348                 if ( $clearState ) {
3349                         $this->clearState();
3350                 }
3351
3352                 $stripState = false;
3353                 $pairs = array(
3354                         "\r\n" => "\n",
3355                 );
3356                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3357                 $text = $this->strip( $text, $stripState, true );
3358                 $text = $this->pstPass2( $text, $user );
3359                 $text = $this->unstrip( $text, $stripState );
3360                 $text = $this->unstripNoWiki( $text, $stripState );
3361                 return $text;
3362         }
3363
3364         /**
3365          * Pre-save transform helper function
3366          * @access private
3367          */
3368         function pstPass2( $text, &$user ) {
3369                 global $wgContLang, $wgLocaltimezone;
3370
3371                 /* Note: This is the timestamp saved as hardcoded wikitext to
3372                  * the database, we use $wgContLang here in order to give
3373                  * everyone the same signiture and use the default one rather
3374                  * than the one selected in each users preferences.
3375                  */
3376                 if ( isset( $wgLocaltimezone ) ) {
3377                         $oldtz = getenv( 'TZ' );
3378                         putenv( 'TZ='.$wgLocaltimezone );
3379                 }
3380                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3381                   ' (' . date( 'T' ) . ')';
3382                 if ( isset( $wgLocaltimezone ) ) {
3383                         putenv( 'TZ='.$oldtz );
3384                 }
3385
3386                 # Variable replacement
3387                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3388                 $text = $this->replaceVariables( $text );
3389
3390                 # Signatures
3391                 $sigText = $this->getUserSig( $user );
3392                 $text = strtr( $text, array(
3393                         '~~~~~' => $d,
3394                         '~~~~' => "$sigText $d",
3395                         '~~~' => $sigText
3396                 ) );
3397
3398                 # Context links: [[|name]] and [[name (context)|]]
3399                 #
3400                 global $wgLegalTitleChars;
3401                 $tc = "[$wgLegalTitleChars]";
3402                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3403
3404                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3405                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3406
3407                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3408                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3409                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3410                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3411                 $context = '';
3412                 $t = $this->mTitle->getText();
3413                 if ( preg_match( $conpat, $t, $m ) ) {
3414                         $context = $m[2];
3415                 }
3416                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3417                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3418                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3419
3420                 if ( '' == $context ) {
3421                         $text = preg_replace( $p2, '[[\\1]]', $text );
3422                 } else {
3423                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3424                 }
3425
3426                 # Trim trailing whitespace
3427                 # MAG_END (__END__) tag allows for trailing
3428                 # whitespace to be deliberately included
3429                 $text = rtrim( $text );
3430                 $mw =& MagicWord::get( MAG_END );
3431                 $mw->matchAndRemove( $text );
3432
3433                 return $text;
3434         }
3435
3436         /**
3437          * Fetch the user's signature text, if any, and normalize to
3438          * validated, ready-to-insert wikitext.
3439          *
3440          * @param User $user
3441          * @return string
3442          * @access private
3443          */
3444         function getUserSig( &$user ) {
3445                 $username = $user->getName();
3446                 $nickname = $user->getOption( 'nickname' );
3447                 $nickname = $nickname === '' ? $username : $nickname;
3448
3449                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3450                         # Sig. might contain markup; validate this
3451                         if( $this->validateSig( $nickname ) !== false ) {
3452                                 # Validated; clean up (if needed) and return it
3453                                 return( $this->cleanSig( $nickname ) );
3454                         } else {
3455                                 # Failed to validate; fall back to the default
3456                                 $nickname = $username;
3457                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3458                         }
3459                 }
3460
3461                 # If we're still here, make it a link to the user page
3462                 $userpage = $user->getUserPage();
3463                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3464         }
3465
3466         /**
3467          * Check that the user's signature contains no bad XML
3468          *
3469          * @param string $text
3470          * @return mixed An expanded string, or false if invalid.
3471          */
3472         function validateSig( $text ) {
3473                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3474         }
3475
3476         /**
3477          * Clean up signature text
3478          *
3479          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3480          * 2) Substitute all transclusions
3481          *
3482          * @param string $text
3483          * @return string Signature text
3484          */
3485         function cleanSig( $text ) {
3486                 $substWord = MagicWord::get( MAG_SUBST );
3487                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3488                 $substText = '{{' . $substWord->getSynonym( 0 );
3489
3490                 $text = preg_replace( $substRegex, $substText, $text );
3491                 $text = preg_replace( '/~{3,5}/', '', $text );
3492                 $text = $this->replaceVariables( $text );
3493
3494                 return $text;
3495         }
3496
3497         /**
3498          * Set up some variables which are usually set up in parse()
3499          * so that an external function can call some class members with confidence
3500          * @access public
3501          */
3502         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3503                 $this->mTitle =& $title;
3504                 $this->mOptions = $options;
3505                 $this->mOutputType = $outputType;
3506                 if ( $clearState ) {
3507                         $this->clearState();
3508                 }
3509         }
3510
3511         /**
3512          * Transform a MediaWiki message by replacing magic variables.
3513          *
3514          * @param string $text the text to transform
3515          * @param ParserOptions $options  options
3516          * @return string the text with variables substituted
3517          * @access public
3518          */
3519         function transformMsg( $text, $options ) {
3520                 global $wgTitle;
3521                 static $executing = false;
3522
3523                 $fname = "Parser::transformMsg";
3524
3525                 # Guard against infinite recursion
3526                 if ( $executing ) {
3527                         return $text;
3528                 }
3529                 $executing = true;
3530
3531                 wfProfileIn($fname);
3532
3533                 $this->mTitle = $wgTitle;
3534                 $this->mOptions = $options;
3535                 $this->mOutputType = OT_MSG;
3536                 $this->clearState();
3537                 $text = $this->replaceVariables( $text );
3538
3539                 $executing = false;
3540                 wfProfileOut($fname);
3541                 return $text;
3542         }
3543
3544         /**
3545          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3546          * The callback should have the following form:
3547          *    function myParserHook( $text, $params, &$parser ) { ... }
3548          *
3549          * Transform and return $text. Use $parser for any required context, e.g. use
3550          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3551          *
3552          * @access public
3553          *
3554          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3555          * @param mixed $callback The callback function (and object) to use for the tag
3556          *
3557          * @return The old value of the mTagHooks array associated with the hook
3558          */
3559         function setHook( $tag, $callback ) {
3560                 $oldVal = @$this->mTagHooks[$tag];
3561                 $this->mTagHooks[$tag] = $callback;
3562
3563                 return $oldVal;
3564         }
3565
3566         /**
3567          * Create a function, e.g. {{sum:1|2|3}}
3568          * The callback function should have the form:
3569          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3570          *
3571          * The callback may either return the text result of the function, or an array with the text
3572          * in element 0, and a number of flags in the other elements. The names of the flags are
3573          * specified in the keys. Valid flags are:
3574          *   found                     The text returned is valid, stop processing the template. This
3575          *                             is on by default.
3576          *   nowiki                    Wiki markup in the return value should be escaped
3577          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3578          *   noargs                    Don't replace triple-brace arguments in the return value
3579          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3580          *
3581          * @access public
3582          *
3583          * @param string $name The function name. Function names are case-insensitive.
3584          * @param mixed $callback The callback function (and object) to use
3585          *
3586          * @return The old callback function for this name, if any
3587          */
3588         function setFunctionHook( $name, $callback ) {
3589                 $name = strtolower( $name );
3590                 $oldVal = @$this->mFunctionHooks[$name];
3591                 $this->mFunctionHooks[$name] = $callback;
3592                 return $oldVal;
3593         }
3594
3595         /**
3596          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3597          * Placeholders created in Skin::makeLinkObj()
3598          * Returns an array of links found, indexed by PDBK:
3599          *  0 - broken
3600          *  1 - normal link
3601          *  2 - stub
3602          * $options is a bit field, RLH_FOR_UPDATE to select for update
3603          */
3604         function replaceLinkHolders( &$text, $options = 0 ) {
3605                 global $wgUser;
3606                 global $wgOutputReplace;
3607
3608                 $fname = 'Parser::replaceLinkHolders';
3609                 wfProfileIn( $fname );
3610
3611                 $pdbks = array();
3612                 $colours = array();
3613                 $sk =& $this->mOptions->getSkin();
3614                 $linkCache =& LinkCache::singleton();
3615
3616                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3617                         wfProfileIn( $fname.'-check' );
3618                         $dbr =& wfGetDB( DB_SLAVE );
3619                         $page = $dbr->tableName( 'page' );
3620                         $threshold = $wgUser->getOption('stubthreshold');
3621
3622                         # Sort by namespace
3623                         asort( $this->mLinkHolders['namespaces'] );
3624
3625                         # Generate query
3626                         $query = false;
3627                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3628                                 # Make title object
3629                                 $title = $this->mLinkHolders['titles'][$key];
3630
3631                                 # Skip invalid entries.
3632                                 # Result will be ugly, but prevents crash.
3633                                 if ( is_null( $title ) ) {
3634                                         continue;
3635                                 }
3636                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3637
3638                                 # Check if it's a static known link, e.g. interwiki
3639                                 if ( $title->isAlwaysKnown() ) {
3640                                         $colours[$pdbk] = 1;
3641                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3642                                         $colours[$pdbk] = 1;
3643                                         $this->mOutput->addLink( $title, $id );
3644                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3645                                         $colours[$pdbk] = 0;
3646                                 } else {
3647                                         # Not in the link cache, add it to the query
3648                                         if ( !isset( $current ) ) {
3649                                                 $current = $ns;
3650                                                 $query =  "SELECT page_id, page_namespace, page_title";
3651                                                 if ( $threshold > 0 ) {
3652                                                         $query .= ', page_len, page_is_redirect';
3653                                                 }
3654                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3655                                         } elseif ( $current != $ns ) {
3656                                                 $current = $ns;
3657                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3658                                         } else {
3659                                                 $query .= ', ';
3660                                         }
3661
3662                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3663                                 }
3664                         }
3665                         if ( $query ) {
3666                                 $query .= '))';
3667                                 if ( $options & RLH_FOR_UPDATE ) {
3668                                         $query .= ' FOR UPDATE';
3669                                 }
3670
3671                                 $res = $dbr->query( $query, $fname );
3672
3673                                 # Fetch data and form into an associative array
3674                                 # non-existent = broken
3675                                 # 1 = known
3676                                 # 2 = stub
3677                                 while ( $s = $dbr->fetchObject($res) ) {
3678                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3679                                         $pdbk = $title->getPrefixedDBkey();
3680                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3681                                         $this->mOutput->addLink( $title, $s->page_id );
3682
3683                                         if ( $threshold >  0 ) {
3684                                                 $size = $s->page_len;
3685                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3686                                                         $colours[$pdbk] = 1;
3687                                                 } else {
3688                                                         $colours[$pdbk] = 2;
3689                                                 }
3690                                         } else {
3691                                                 $colours[$pdbk] = 1;
3692                                         }
3693                                 }
3694                         }
3695                         wfProfileOut( $fname.'-check' );
3696
3697                         # Construct search and replace arrays
3698                         wfProfileIn( $fname.'-construct' );
3699                         $wgOutputReplace = array();
3700                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3701                                 $pdbk = $pdbks[$key];
3702                                 $searchkey = "<!--LINK $key-->";
3703                                 $title = $this->mLinkHolders['titles'][$key];
3704                                 if ( empty( $colours[$pdbk] ) ) {
3705                                         $linkCache->addBadLinkObj( $title );
3706                                         $colours[$pdbk] = 0;
3707                                         $this->mOutput->addLink( $title, 0 );
3708                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3709                                                                         $this->mLinkHolders['texts'][$key],
3710                                                                         $this->mLinkHolders['queries'][$key] );
3711                                 } elseif ( $colours[$pdbk] == 1 ) {
3712                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3713                                                                         $this->mLinkHolders['texts'][$key],
3714                                                                         $this->mLinkHolders['queries'][$key] );
3715                                 } elseif ( $colours[$pdbk] == 2 ) {
3716                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3717                                                                         $this->mLinkHolders['texts'][$key],
3718                                                                         $this->mLinkHolders['queries'][$key] );
3719                                 }
3720                         }
3721                         wfProfileOut( $fname.'-construct' );
3722
3723                         # Do the thing
3724                         wfProfileIn( $fname.'-replace' );
3725
3726                         $text = preg_replace_callback(
3727                                 '/(<!--LINK .*?-->)/',
3728                                 "wfOutputReplaceMatches",
3729                                 $text);
3730
3731                         wfProfileOut( $fname.'-replace' );
3732                 }
3733
3734                 # Now process interwiki link holders
3735                 # This is quite a bit simpler than internal links
3736                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3737                         wfProfileIn( $fname.'-interwiki' );
3738                         # Make interwiki link HTML
3739                         $wgOutputReplace = array();
3740                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3741                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3742                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3743                         }
3744
3745                         $text = preg_replace_callback(
3746                                 '/<!--IWLINK (.*?)-->/',
3747                                 "wfOutputReplaceMatches",
3748                                 $text );
3749                         wfProfileOut( $fname.'-interwiki' );
3750                 }
3751
3752                 wfProfileOut( $fname );
3753                 return $colours;
3754         }
3755
3756         /**
3757          * Replace <!--LINK--> link placeholders with plain text of links
3758          * (not HTML-formatted).
3759          * @param string $text
3760          * @return string
3761          */
3762         function replaceLinkHoldersText( $text ) {
3763                 $fname = 'Parser::replaceLinkHoldersText';
3764                 wfProfileIn( $fname );
3765
3766                 $text = preg_replace_callback(
3767                         '/<!--(LINK|IWLINK) (.*?)-->/',
3768                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3769                         $text );
3770
3771                 wfProfileOut( $fname );
3772                 return $text;
3773         }
3774
3775         /**
3776          * @param array $matches
3777          * @return string
3778          * @access private
3779          */
3780         function replaceLinkHoldersTextCallback( $matches ) {
3781                 $type = $matches[1];
3782                 $key  = $matches[2];
3783                 if( $type == 'LINK' ) {
3784                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3785                                 return $this->mLinkHolders['texts'][$key];
3786                         }
3787                 } elseif( $type == 'IWLINK' ) {
3788                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3789                                 return $this->mInterwikiLinkHolders['texts'][$key];
3790                         }
3791                 }
3792                 return $matches[0];
3793         }
3794
3795         /**
3796          * Renders an image gallery from a text with one line per image.
3797          * text labels may be given by using |-style alternative text. E.g.
3798          *   Image:one.jpg|The number "1"
3799          *   Image:tree.jpg|A tree
3800          * given as text will return the HTML of a gallery with two images,
3801          * labeled 'The number "1"' and
3802          * 'A tree'.
3803          */
3804         function renderImageGallery( $text ) {
3805                 # Setup the parser
3806                 $parserOptions = new ParserOptions;
3807                 $localParser = new Parser();
3808
3809                 $ig = new ImageGallery();
3810                 $ig->setShowBytes( false );
3811                 $ig->setShowFilename( false );
3812                 $lines = explode( "\n", $text );
3813
3814                 foreach ( $lines as $line ) {
3815                         # match lines like these:
3816                         # Image:someimage.jpg|This is some image
3817                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3818                         # Skip empty lines
3819                         if ( count( $matches ) == 0 ) {
3820                                 continue;
3821                         }
3822                         $nt =& Title::newFromText( $matches[1] );
3823                         if( is_null( $nt ) ) {
3824                                 # Bogus title. Ignore these so we don't bomb out later.
3825                                 continue;
3826                         }
3827                         if ( isset( $matches[3] ) ) {
3828                                 $label = $matches[3];
3829                         } else {
3830                                 $label = '';
3831                         }
3832
3833                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3834                         $html = $pout->getText();
3835
3836                         $ig->add( new Image( $nt ), $html );
3837                         $this->mOutput->addImage( $nt->getDBkey() );
3838                 }
3839                 return $ig->toHTML();
3840         }
3841
3842         /**
3843          * Parse image options text and use it to make an image
3844          */
3845         function makeImage( &$nt, $options ) {
3846                 global $wgUseImageResize;
3847
3848                 $align = '';
3849
3850                 # Check if the options text is of the form "options|alt text"
3851                 # Options are:
3852                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3853                 #  * left               no resizing, just left align. label is used for alt= only
3854                 #  * right              same, but right aligned
3855                 #  * none               same, but not aligned
3856                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3857                 #  * center             center the image
3858                 #  * framed             Keep original image size, no magnify-button.
3859
3860                 $part = explode( '|', $options);
3861
3862                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3863                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3864                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3865                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3866                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3867                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3868                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3869                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3870                 $caption = '';
3871
3872                 $width = $height = $framed = $thumb = false;
3873                 $manual_thumb = '' ;
3874
3875                 foreach( $part as $key => $val ) {
3876                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3877                                 $thumb=true;
3878                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3879                                 # use manually specified thumbnail
3880                                 $thumb=true;
3881                                 $manual_thumb = $match;
3882                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3883                                 # remember to set an alignment, don't render immediately
3884                                 $align = 'right';
3885                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3886                                 # remember to set an alignment, don't render immediately
3887                                 $align = 'left';
3888                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3889                                 # remember to set an alignment, don't render immediately
3890                                 $align = 'center';
3891                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3892                                 # remember to set an alignment, don't render immediately
3893                                 $align = 'none';
3894                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3895                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3896                                 # $match is the image width in pixels
3897                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3898                                         $width = intval( $m[1] );
3899                                         $height = intval( $m[2] );
3900                                 } else {
3901                                         $width = intval($match);
3902                                 }
3903                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3904                                 $framed=true;
3905                         } else {
3906                                 $caption = $val;
3907                         }
3908                 }
3909                 # Strip bad stuff out of the alt text
3910                 $alt = $this->replaceLinkHoldersText( $caption );
3911
3912                 # make sure there are no placeholders in thumbnail attributes
3913                 # that are later expanded to html- so expand them now and
3914                 # remove the tags
3915                 $alt = $this->unstrip($alt, $this->mStripState);
3916                 $alt = Sanitizer::stripAllTags( $alt );
3917
3918                 # Linker does the rest
3919                 $sk =& $this->mOptions->getSkin();
3920                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3921         }
3922
3923         /**
3924          * Set a flag in the output object indicating that the content is dynamic and
3925          * shouldn't be cached.
3926          */
3927         function disableCache() {
3928                 $this->mOutput->mCacheTime = -1;
3929         }
3930
3931         /**#@+
3932          * Callback from the Sanitizer for expanding items found in HTML attribute
3933          * values, so they can be safely tested and escaped.
3934          * @param string $text
3935          * @param array $args
3936          * @return string
3937          * @access private
3938          */
3939         function attributeStripCallback( &$text, $args ) {
3940                 $text = $this->replaceVariables( $text, $args );
3941                 $text = $this->unstripForHTML( $text );
3942                 return $text;
3943         }
3944
3945         function unstripForHTML( $text ) {
3946                 $text = $this->unstrip( $text, $this->mStripState );
3947                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3948                 return $text;
3949         }
3950         /**#@-*/
3951
3952         /**#@+
3953          * Accessor/mutator
3954          */
3955         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3956         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3957         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3958         /**#@-*/
3959
3960         /**#@+
3961          * Accessor
3962          */
3963         function getTags() { return array_keys( $this->mTagHooks ); }
3964         /**#@-*/
3965 }
3966
3967 /**
3968  * @todo document
3969  * @package MediaWiki
3970  */
3971 class ParserOutput
3972 {
3973         var $mText,             # The output text
3974                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3975                 $mCategories,       # Map of category names to sort keys
3976                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3977                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
3978                 $mVersion,          # Compatibility check
3979                 $mTitleText,        # title text of the chosen language variant
3980                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3981                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3982                 $mImages,           # DB keys of the images used, in the array key only
3983                 $mExternalLinks;    # External link URLs, in the key only
3984
3985         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3986                 $containsOldMagic = false, $titletext = '' )
3987         {
3988                 $this->mText = $text;
3989                 $this->mLanguageLinks = $languageLinks;
3990                 $this->mCategories = $categoryLinks;
3991                 $this->mContainsOldMagic = $containsOldMagic;
3992                 $this->mCacheTime = '';
3993                 $this->mVersion = MW_PARSER_VERSION;
3994                 $this->mTitleText = $titletext;
3995                 $this->mLinks = array();
3996                 $this->mTemplates = array();
3997                 $this->mImages = array();
3998                 $this->mExternalLinks = array();
3999         }
4000
4001         function getText()                   { return $this->mText; }
4002         function getLanguageLinks()          { return $this->mLanguageLinks; }
4003         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4004         function &getCategories()            { return $this->mCategories; }
4005         function getCacheTime()              { return $this->mCacheTime; }
4006         function getTitleText()              { return $this->mTitleText; }
4007         function &getLinks()                 { return $this->mLinks; }
4008         function &getTemplates()             { return $this->mTemplates; }
4009         function &getImages()                { return $this->mImages; }
4010         function &getExternalLinks()         { return $this->mExternalLinks; }
4011
4012         function containsOldMagic()          { return $this->mContainsOldMagic; }
4013         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4014         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4015         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4016         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4017         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4018         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4019
4020         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4021         function addImage( $name )           { $this->mImages[$name] = 1; }
4022         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4023         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4024
4025         function addLink( $title, $id ) {
4026                 $ns = $title->getNamespace();
4027                 $dbk = $title->getDBkey();
4028                 if ( !isset( $this->mLinks[$ns] ) ) {
4029                         $this->mLinks[$ns] = array();
4030                 }
4031                 $this->mLinks[$ns][$dbk] = $id;
4032         }
4033
4034         function addTemplate( $title, $id ) {
4035                 $ns = $title->getNamespace();
4036                 $dbk = $title->getDBkey();
4037                 if ( !isset( $this->mTemplates[$ns] ) ) {
4038                         $this->mTemplates[$ns] = array();
4039                 }
4040                 $this->mTemplates[$ns][$dbk] = $id;
4041         }
4042
4043         /**
4044          * @deprecated
4045          */
4046         /*
4047         function merge( $other ) {
4048                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
4049                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
4050                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
4051         }*/
4052
4053         /**
4054          * Return true if this cached output object predates the global or
4055          * per-article cache invalidation timestamps, or if it comes from
4056          * an incompatible older version.
4057          *
4058          * @param string $touched the affected article's last touched timestamp
4059          * @return bool
4060          * @access public
4061          */
4062         function expired( $touched ) {
4063                 global $wgCacheEpoch;
4064                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4065                        $this->getCacheTime() < $touched ||
4066                        $this->getCacheTime() <= $wgCacheEpoch ||
4067                        !isset( $this->mVersion ) ||
4068                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4069         }
4070 }
4071
4072 /**
4073  * Set options of the Parser
4074  * @todo document
4075  * @package MediaWiki
4076  */
4077 class ParserOptions
4078 {
4079         # All variables are private
4080         var $mUseTeX;                    # Use texvc to expand <math> tags
4081         var $mUseDynamicDates;           # Use DateFormatter to format dates
4082         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4083         var $mAllowExternalImages;       # Allow external images inline
4084         var $mAllowExternalImagesFrom;   # If not, any exception?
4085         var $mSkin;                      # Reference to the preferred skin
4086         var $mDateFormat;                # Date format index
4087         var $mEditSection;               # Create "edit section" links
4088         var $mNumberHeadings;            # Automatically number headings
4089         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4090         var $mTidy;                      # Ask for tidy cleanup
4091
4092         function getUseTeX()                        { return $this->mUseTeX; }
4093         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4094         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4095         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4096         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4097         function &getSkin()                         { return $this->mSkin; }
4098         function getDateFormat()                    { return $this->mDateFormat; }
4099         function getEditSection()                   { return $this->mEditSection; }
4100         function getNumberHeadings()                { return $this->mNumberHeadings; }
4101         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4102         function getTidy()                          { return $this->mTidy; }
4103
4104         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4105         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4106         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4107         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4108         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4109         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4110         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4111         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4112         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4113         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4114         function setSkin( &$x ) { $this->mSkin =& $x; }
4115
4116         function ParserOptions() {
4117                 global $wgUser;
4118                 $this->initialiseFromUser( $wgUser );
4119         }
4120
4121         /**
4122          * Get parser options
4123          * @static
4124          */
4125         function newFromUser( &$user ) {
4126                 $popts = new ParserOptions;
4127                 $popts->initialiseFromUser( $user );
4128                 return $popts;
4129         }
4130
4131         /** Get user options */
4132         function initialiseFromUser( &$userInput ) {
4133                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4134                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4135                 $fname = 'ParserOptions::initialiseFromUser';
4136                 wfProfileIn( $fname );
4137                 if ( !$userInput ) {
4138                         $user = new User;
4139                         $user->setLoaded( true );
4140                 } else {
4141                         $user =& $userInput;
4142                 }
4143
4144                 $this->mUseTeX = $wgUseTeX;
4145                 $this->mUseDynamicDates = $wgUseDynamicDates;
4146                 $this->mInterwikiMagic = $wgInterwikiMagic;
4147                 $this->mAllowExternalImages = $wgAllowExternalImages;
4148                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4149                 wfProfileIn( $fname.'-skin' );
4150                 $this->mSkin =& $user->getSkin();
4151                 wfProfileOut( $fname.'-skin' );
4152                 $this->mDateFormat = $user->getOption( 'date' );
4153                 $this->mEditSection = true;
4154                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4155                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4156                 $this->mTidy = false;
4157                 wfProfileOut( $fname );
4158         }
4159 }
4160
4161 /**
4162  * Callback function used by Parser::replaceLinkHolders()
4163  * to substitute link placeholders.
4164  */
4165 function &wfOutputReplaceMatches( $matches ) {
4166         global $wgOutputReplace;
4167         return $wgOutputReplace[$matches[1]];
4168 }
4169
4170 /**
4171  * Return the total number of articles
4172  */
4173 function wfNumberOfArticles() {
4174         global $wgNumberOfArticles;
4175
4176         wfLoadSiteStats();
4177         return $wgNumberOfArticles;
4178 }
4179
4180 /**
4181  * Return the number of files
4182  */
4183 function wfNumberOfFiles() {
4184         $fname = 'wfNumberOfFiles';
4185
4186         wfProfileIn( $fname );
4187         $dbr =& wfGetDB( DB_SLAVE );
4188         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4189         wfProfileOut( $fname );
4190
4191         return $numImages;
4192 }
4193
4194 /**
4195  * Get various statistics from the database
4196  * @access private
4197  */
4198 function wfLoadSiteStats() {
4199         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4200         $fname = 'wfLoadSiteStats';
4201
4202         if ( -1 != $wgNumberOfArticles ) return;
4203         $dbr =& wfGetDB( DB_SLAVE );
4204         $s = $dbr->selectRow( 'site_stats',
4205                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4206                 array( 'ss_row_id' => 1 ), $fname
4207         );
4208
4209         if ( $s === false ) {
4210                 return;
4211         } else {
4212                 $wgTotalViews = $s->ss_total_views;
4213                 $wgTotalEdits = $s->ss_total_edits;
4214                 $wgNumberOfArticles = $s->ss_good_articles;
4215         }
4216 }
4217
4218 /**
4219  * Escape html tags
4220  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4221  *
4222  * @param string $in Text that might contain HTML tags
4223  * @return string Escaped string
4224  */
4225 function wfEscapeHTMLTagsOnly( $in ) {
4226         return str_replace(
4227                 array( '"', '>', '<' ),
4228                 array( '&quot;', '&gt;', '&lt;' ),
4229                 $in );
4230 }
4231
4232 ?>