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 ) < 3 ) {
 344                                 break;
 345                         }
 346                         $attributes = $p[1];
 347                         $inside     = $p[2];
 348
 349                         // If $attributes ends with '/', we have an empty element tag, <tag />
 350                         if( $tag != STRIP_COMMENTS && substr( $attributes, -1 ) == '/' ) {
 351                                 $attributes = substr( $attributes, 0, -1);
 352                                 $empty = '/';
 353                         } else {
 354                                 $empty = '';
 355                         }
 356
 357                         $marker = $rnd . sprintf('%08X', $n++);
 358                         $stripped .= $marker;
 359
 360                         $tags[$marker] = "<$tag$attributes$empty>";
 361                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 362
 363                         if ( $empty === '/' ) {
 364                                 // Empty element tag, <tag />
 365                                 $content[$marker] = null;
 366                                 $text = $inside;
 367                         } else {
 368                                 $q = preg_split( $end, $inside, 2 );
 369                                 $content[$marker] = $q[0];
 370                                 if( count( $q ) < 2 ) {
 371                                         # No end tag -- let it run out to the end of the text.
 372                                         break;
 373                                 } else {
 374                                         $text = $q[1];
 375                                 }
 376                         }
 377                 }
 378                 return $stripped;
 379         }
 380
 381         /**
 382          * Wrapper function for extractTagsAndParams
 383          * for cases where $tags and $params isn't needed
 384          * i.e. where tags will never have params, like <nowiki>
 385          *
 386          * @access private
 387          * @static
 388          */
 389         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 390                 $dummy_tags = array();
 391                 $dummy_params = array();
 392
 393                 return Parser::extractTagsAndParams( $tag, $text, $content,
 394                         $dummy_tags, $dummy_params, $uniq_prefix );
 395         }
 396
 397         /**
 398          * Strips and renders nowiki, pre, math, hiero
 399          * If $render is set, performs necessary rendering operations on plugins
 400          * Returns the text, and fills an array with data needed in unstrip()
 401          * If the $state is already a valid strip state, it adds to the state
 402          *
 403          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 404          *  will be stripped in addition to other tags. This is important
 405          *  for section editing, where these comments cause confusion when
 406          *  counting the sections in the wikisource
 407          *
 408          * @access private
 409          */
 410         function strip( $text, &$state, $stripcomments = false ) {
 411                 $render = ($this->mOutputType == OT_HTML);
 412                 $html_content = array();
 413                 $nowiki_content = array();
 414                 $math_content = array();
 415                 $pre_content = array();
 416                 $comment_content = array();
 417                 $ext_content = array();
 418                 $ext_tags = array();
 419                 $ext_params = array();
 420                 $gallery_content = array();
 421
 422                 # Replace any instances of the placeholders
 423                 $uniq_prefix = $this->mUniqPrefix;
 424                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 425
 426                 # html
 427                 global $wgRawHtml;
 428                 if( $wgRawHtml ) {
 429                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 430                         foreach( $html_content as $marker => $content ) {
 431                                 if ($render ) {
 432                                         # Raw and unchecked for validity.
 433                                         $html_content[$marker] = $content;
 434                                 } else {
 435                                         $html_content[$marker] = '<html>'.$content.'</html>';
 436                                 }
 437                         }
 438                 }
 439
 440                 # nowiki
 441                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 442                 foreach( $nowiki_content as $marker => $content ) {
 443                         if( $render ){
 444                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 445                         } else {
 446                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 447                         }
 448                 }
 449
 450                 # math
 451                 if( $this->mOptions->getUseTeX() ) {
 452                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 453                         foreach( $math_content as $marker => $content ){
 454                                 if( $render ) {
 455                                         $math_content[$marker] = renderMath( $content );
 456                                 } else {
 457                                         $math_content[$marker] = '<math>'.$content.'</math>';
 458                                 }
 459                         }
 460                 }
 461
 462                 # pre
 463                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 464                 foreach( $pre_content as $marker => $content ){
 465                         if( $render ){
 466                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 467                         } else {
 468                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 469                         }
 470                 }
 471
 472                 # gallery
 473                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 474                 foreach( $gallery_content as $marker => $content ) {
 475                         require_once( 'ImageGallery.php' );
 476                         if ( $render ) {
 477                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 478                         } else {
 479                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 480                         }
 481                 }
 482
 483                 # Comments
 484                 $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 485                 foreach( $comment_content as $marker => $content ){
 486                         $comment_content[$marker] = '<!--'.$content.'-->';
 487                 }
 488
 489                 # Extensions
 490                 foreach ( $this->mTagHooks as $tag => $callback ) {
 491                         $ext_content[$tag] = array();
 492                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 493                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 494                         foreach( $ext_content[$tag] as $marker => $content ) {
 495                                 $full_tag = $ext_tags[$tag][$marker];
 496                                 $params = $ext_params[$tag][$marker];
 497                                 if ( $render )
 498                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 499                                 else {
 500                                         if ( is_null( $content ) ) {
 501                                                 // Empty element tag
 502                                                 $ext_content[$tag][$marker] = $full_tag;
 503                                         } else {
 504                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 505                                         }
 506                                 }
 507                         }
 508                 }
 509
 510                 # Unstrip comments unless explicitly told otherwise.
 511                 # (The comments are always stripped prior to this point, so as to
 512                 # not invoke any extension tags / parser hooks contained within
 513                 # a comment.)
 514                 if ( !$stripcomments ) {
 515                         $tempstate = array( 'comment' => $comment_content );
 516                         $text = $this->unstrip( $text, $tempstate );
 517                         $comment_content = array();
 518                 }
 519
 520                 # Merge state with the pre-existing state, if there is one
 521                 if ( $state ) {
 522                         $state['html'] = $state['html'] + $html_content;
 523                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 524                         $state['math'] = $state['math'] + $math_content;
 525                         $state['pre'] = $state['pre'] + $pre_content;
 526                         $state['gallery'] = $state['gallery'] + $gallery_content;
 527                         $state['comment'] = $state['comment'] + $comment_content;
 528
 529                         foreach( $ext_content as $tag => $array ) {
 530                                 if ( array_key_exists( $tag, $state ) ) {
 531                                         $state[$tag] = $state[$tag] + $array;
 532                                 }
 533                         }
 534                 } else {
 535                         $state = array(
 536                           'html' => $html_content,
 537                           'nowiki' => $nowiki_content,
 538                           'math' => $math_content,
 539                           'pre' => $pre_content,
 540                           'gallery' => $gallery_content,
 541                           'comment' => $comment_content,
 542                         ) + $ext_content;
 543                 }
 544                 return $text;
 545         }
 546
 547         /**
 548          * restores pre, math, and hiero removed by strip()
 549          *
 550          * always call unstripNoWiki() after this one
 551          * @access private
 552          */
 553         function unstrip( $text, &$state ) {
 554                 if ( !is_array( $state ) ) {
 555                         return $text;
 556                 }
 557
 558                 # Must expand in reverse order, otherwise nested tags will be corrupted
 559                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 560                         if( $tag != 'nowiki' && $tag != 'html' ) {
 561                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 562                                         $text = str_replace( $uniq, $content, $text );
 563                                 }
 564                         }
 565                 }
 566
 567                 return $text;
 568         }
 569
 570         /**
 571          * always call this after unstrip() to preserve the order
 572          *
 573          * @access private
 574          */
 575         function unstripNoWiki( $text, &$state ) {
 576                 if ( !is_array( $state ) ) {
 577                         return $text;
 578                 }
 579
 580                 # Must expand in reverse order, otherwise nested tags will be corrupted
 581                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 582                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 583                 }
 584
 585                 global $wgRawHtml;
 586                 if ($wgRawHtml) {
 587                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 588                                 $text = str_replace( key( $state['html'] ), $content, $text );
 589                         }
 590                 }
 591
 592                 return $text;
 593         }
 594
 595         /**
 596          * Add an item to the strip state
 597          * Returns the unique tag which must be inserted into the stripped text
 598          * The tag will be replaced with the original text in unstrip()
 599          *
 600          * @access private
 601          */
 602         function insertStripItem( $text, &$state ) {
 603                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 604                 if ( !$state ) {
 605                         $state = array(
 606                           'html' => array(),
 607                           'nowiki' => array(),
 608                           'math' => array(),
 609                           'pre' => array(),
 610                           'comment' => array(),
 611                           'gallery' => array(),
 612                         );
 613                 }
 614                 $state['item'][$rnd] = $text;
 615                 return $rnd;
 616         }
 617
 618         /**
 619          * Interface with html tidy, used if $wgUseTidy = true.
 620          * If tidy isn't able to correct the markup, the original will be
 621          * returned in all its glory with a warning comment appended.
 622          *
 623          * Either the external tidy program or the in-process tidy extension
 624          * will be used depending on availability. Override the default
 625          * $wgTidyInternal setting to disable the internal if it's not working.
 626          *
 627          * @param string $text Hideous HTML input
 628          * @return string Corrected HTML output
 629          * @access public
 630          * @static
 631          */
 632         function tidy( $text ) {
 633                 global $wgTidyInternal;
 634                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 635 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 636 '<head><title>test</title></head><body>'.$text.'</body></html>';
 637                 if( $wgTidyInternal ) {
 638                         $correctedtext = Parser::internalTidy( $wrappedtext );
 639                 } else {
 640                         $correctedtext = Parser::externalTidy( $wrappedtext );
 641                 }
 642                 if( is_null( $correctedtext ) ) {
 643                         wfDebug( "Tidy error detected!\n" );
 644                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 645                 }
 646                 return $correctedtext;
 647         }
 648
 649         /**
 650          * Spawn an external HTML tidy process and get corrected markup back from it.
 651          *
 652          * @access private
 653          * @static
 654          */
 655         function externalTidy( $text ) {
 656                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 657                 $fname = 'Parser::externalTidy';
 658                 wfProfileIn( $fname );
 659
 660                 $cleansource = '';
 661                 $opts = ' -utf8';
 662
 663                 $descriptorspec = array(
 664                         0 => array('pipe', 'r'),
 665                         1 => array('pipe', 'w'),
 666                         2 => array('file', '/dev/null', 'a')
 667                 );
 668                 $pipes = array();
 669                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 670                 if (is_resource($process)) {
 671                         // Theoretically, this style of communication could cause a deadlock
 672                         // here. If the stdout buffer fills up, then writes to stdin could
 673                         // block. This doesn't appear to happen with tidy, because tidy only
 674                         // writes to stdout after it's finished reading from stdin. Search
 675                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 676                         fwrite($pipes[0], $text);
 677                         fclose($pipes[0]);
 678                         while (!feof($pipes[1])) {
 679                                 $cleansource .= fgets($pipes[1], 1024);
 680                         }
 681                         fclose($pipes[1]);
 682                         proc_close($process);
 683                 }
 684
 685                 wfProfileOut( $fname );
 686
 687                 if( $cleansource == '' && $text != '') {
 688                         // Some kind of error happened, so we couldn't get the corrected text.
 689                         // Just give up; we'll use the source text and append a warning.
 690                         return null;
 691                 } else {
 692                         return $cleansource;
 693                 }
 694         }
 695
 696         /**
 697          * Use the HTML tidy PECL extension to use the tidy library in-process,
 698          * saving the overhead of spawning a new process. Currently written to
 699          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 700          *
 701          * 'pear install tidy' should be able to compile the extension module.
 702          *
 703          * @access private
 704          * @static
 705          */
 706         function internalTidy( $text ) {
 707                 global $wgTidyConf;
 708                 $fname = 'Parser::internalTidy';
 709                 wfProfileIn( $fname );
 710
 711                 tidy_load_config( $wgTidyConf );
 712                 tidy_set_encoding( 'utf8' );
 713                 tidy_parse_string( $text );
 714                 tidy_clean_repair();
 715                 if( tidy_get_status() == 2 ) {
 716                         // 2 is magic number for fatal error
 717                         // http://www.php.net/manual/en/function.tidy-get-status.php
 718                         $cleansource = null;
 719                 } else {
 720                         $cleansource = tidy_get_output();
 721                 }
 722                 wfProfileOut( $fname );
 723                 return $cleansource;
 724         }
 725
 726         /**
 727          * parse the wiki syntax used to render tables
 728          *
 729          * @access private
 730          */
 731         function doTableStuff ( $t ) {
 732                 $fname = 'Parser::doTableStuff';
 733                 wfProfileIn( $fname );
 734
 735                 $t = explode ( "\n" , $t ) ;
 736                 $td = array () ; # Is currently a td tag open?
 737                 $ltd = array () ; # Was it TD or TH?
 738                 $tr = array () ; # Is currently a tr tag open?
 739                 $ltr = array () ; # tr attributes
 740                 $has_opened_tr = array(); # Did this table open a <tr> element?
 741                 $indent_level = 0; # indent level of the table
 742                 foreach ( $t AS $k => $x )
 743                 {
 744                         $x = trim ( $x ) ;
 745                         $fc = substr ( $x , 0 , 1 ) ;
 746                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 747                                 $indent_level = strlen( $matches[1] );
 748
 749                                 $attributes = $this->unstripForHTML( $matches[2] );
 750
 751                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 752                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 753                                 array_push ( $td , false ) ;
 754                                 array_push ( $ltd , '' ) ;
 755                                 array_push ( $tr , false ) ;
 756                                 array_push ( $ltr , '' ) ;
 757                                 array_push ( $has_opened_tr, false );
 758                         }
 759                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 760                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 761                                 $z = "</table>" . substr ( $x , 2);
 762                                 $l = array_pop ( $ltd ) ;
 763                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 764                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 765                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 766                                 array_pop ( $ltr ) ;
 767                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 768                         }
 769                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 770                                 $x = substr ( $x , 1 ) ;
 771                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 772                                 $z = '' ;
 773                                 $l = array_pop ( $ltd ) ;
 774                                 array_pop ( $has_opened_tr );
 775                                 array_push ( $has_opened_tr , true ) ;
 776                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 777                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 778                                 array_pop ( $ltr ) ;
 779                                 $t[$k] = $z ;
 780                                 array_push ( $tr , false ) ;
 781                                 array_push ( $td , false ) ;
 782                                 array_push ( $ltd , '' ) ;
 783                                 $attributes = $this->unstripForHTML( $x );
 784                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 785                         }
 786                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 787                                 # $x is a table row
 788                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 789                                         $fc = '+' ;
 790                                         $x = substr ( $x , 1 ) ;
 791                                 }
 792                                 $after = substr ( $x , 1 ) ;
 793                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 794
 795                                 // Split up multiple cells on the same line.
 796                                 // FIXME: This can result in improper nesting of tags processed
 797                                 // by earlier parser steps, but should avoid splitting up eg
 798                                 // attribute values containing literal "||".
 799                                 $after = wfExplodeMarkup( '||', $after );
 800
 801                                 $t[$k] = '' ;
 802
 803                                 # Loop through each table cell
 804                                 foreach ( $after AS $theline )
 805                                 {
 806                                         $z = '' ;
 807                                         if ( $fc != '+' )
 808                                         {
 809                                                 $tra = array_pop ( $ltr ) ;
 810                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 811                                                 array_push ( $tr , true ) ;
 812                                                 array_push ( $ltr , '' ) ;
 813                                                 array_pop ( $has_opened_tr );
 814                                                 array_push ( $has_opened_tr , true ) ;
 815                                         }
 816
 817                                         $l = array_pop ( $ltd ) ;
 818                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 819                                         if ( $fc == '|' ) $l = 'td' ;
 820                                         else if ( $fc == '!' ) $l = 'th' ;
 821                                         else if ( $fc == '+' ) $l = 'caption' ;
 822                                         else $l = '' ;
 823                                         array_push ( $ltd , $l ) ;
 824
 825                                         # Cell parameters
 826                                         $y = explode ( '|' , $theline , 2 ) ;
 827                                         # Note that a '|' inside an invalid link should not
 828                                         # be mistaken as delimiting cell parameters
 829                                         if ( strpos( $y[0], '[[' ) !== false ) {
 830                                                 $y = array ($theline);
 831                                         }
 832                                         if ( count ( $y ) == 1 )
 833                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 834                                         else {
 835                                                 $attributes = $this->unstripForHTML( $y[0] );
 836                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 837                                         }
 838                                         $t[$k] .= $y ;
 839                                         array_push ( $td , true ) ;
 840                                 }
 841                         }
 842                 }
 843
 844                 # Closing open td, tr && table
 845                 while ( count ( $td ) > 0 )
 846                 {
 847                         $l = array_pop ( $ltd ) ;
 848                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 849                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 850                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 851                         $t[] = '</table>' ;
 852                 }
 853
 854                 $t = implode ( "\n" , $t ) ;
 855                 # special case: don't return empty table
 856                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 857                         $t = '';
 858                 wfProfileOut( $fname );
 859                 return $t ;
 860         }
 861
 862         /**
 863          * Helper function for parse() that transforms wiki markup into
 864          * HTML. Only called for $mOutputType == OT_HTML.
 865          *
 866          * @access private
 867          */
 868         function internalParse( $text ) {
 869                 $args = array();
 870                 $isMain = true;
 871                 $fname = 'Parser::internalParse';
 872                 wfProfileIn( $fname );
 873
 874                 # Remove <noinclude> tags and <includeonly> sections
 875                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 876                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 877                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 878
 879                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 880                 $text = $this->replaceVariables( $text, $args );
 881
 882                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 883
 884                 $text = $this->doHeadings( $text );
 885                 if($this->mOptions->getUseDynamicDates()) {
 886                         $df =& DateFormatter::getInstance();
 887                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 888                 }
 889                 $text = $this->doAllQuotes( $text );
 890                 $text = $this->replaceInternalLinks( $text );
 891                 $text = $this->replaceExternalLinks( $text );
 892
 893                 # replaceInternalLinks may sometimes leave behind
 894                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 895                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 896
 897                 $text = $this->doMagicLinks( $text );
 898                 $text = $this->doTableStuff( $text );
 899                 $text = $this->formatHeadings( $text, $isMain );
 900
 901                 wfProfileOut( $fname );
 902                 return $text;
 903         }
 904
 905         /**
 906          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 907          * magic external links.
 908          *
 909          * @access private
 910          */
 911         function &doMagicLinks( &$text ) {
 912                 $text = $this->magicISBN( $text );
 913                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 914                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 915                 return $text;
 916         }
 917
 918         /**
 919          * Parse headers and return html
 920          *
 921          * @access private
 922          */
 923         function doHeadings( $text ) {
 924                 $fname = 'Parser::doHeadings';
 925                 wfProfileIn( $fname );
 926                 for ( $i = 6; $i >= 1; --$i ) {
 927                         $h = str_repeat( '=', $i );
 928                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 929                           "<h{$i}>\\1</h{$i}>\\2", $text );
 930                 }
 931                 wfProfileOut( $fname );
 932                 return $text;
 933         }
 934
 935         /**
 936          * Replace single quotes with HTML markup
 937          * @access private
 938          * @return string the altered text
 939          */
 940         function doAllQuotes( $text ) {
 941                 $fname = 'Parser::doAllQuotes';
 942                 wfProfileIn( $fname );
 943                 $outtext = '';
 944                 $lines = explode( "\n", $text );
 945                 foreach ( $lines as $line ) {
 946                         $outtext .= $this->doQuotes ( $line ) . "\n";
 947                 }
 948                 $outtext = substr($outtext, 0,-1);
 949                 wfProfileOut( $fname );
 950                 return $outtext;
 951         }
 952
 953         /**
 954          * Helper function for doAllQuotes()
 955          * @access private
 956          */
 957         function doQuotes( $text ) {
 958                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 959                 if ( count( $arr ) == 1 )
 960                         return $text;
 961                 else
 962                 {
 963                         # First, do some preliminary work. This may shift some apostrophes from
 964                         # being mark-up to being text. It also counts the number of occurrences
 965                         # of bold and italics mark-ups.
 966                         $i = 0;
 967                         $numbold = 0;
 968                         $numitalics = 0;
 969                         foreach ( $arr as $r )
 970                         {
 971                                 if ( ( $i % 2 ) == 1 )
 972                                 {
 973                                         # If there are ever four apostrophes, assume the first is supposed to
 974                                         # be text, and the remaining three constitute mark-up for bold text.
 975                                         if ( strlen( $arr[$i] ) == 4 )
 976                                         {
 977                                                 $arr[$i-1] .= "'";
 978                                                 $arr[$i] = "'''";
 979                                         }
 980                                         # If there are more than 5 apostrophes in a row, assume they're all
 981                                         # text except for the last 5.
 982                                         else if ( strlen( $arr[$i] ) > 5 )
 983                                         {
 984                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 985                                                 $arr[$i] = "'''''";
 986                                         }
 987                                         # Count the number of occurrences of bold and italics mark-ups.
 988                                         # We are not counting sequences of five apostrophes.
 989                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 990                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 991                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 992                                 }
 993                                 $i++;
 994                         }
 995
 996                         # If there is an odd number of both bold and italics, it is likely
 997                         # that one of the bold ones was meant to be an apostrophe followed
 998                         # by italics. Which one we cannot know for certain, but it is more
 999                         # likely to be one that has a single-letter word before it.
1000                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1001                         {
1002                                 $i = 0;
1003                                 $firstsingleletterword = -1;
1004                                 $firstmultiletterword = -1;
1005                                 $firstspace = -1;
1006                                 foreach ( $arr as $r )
1007                                 {
1008                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1009                                         {
1010                                                 $x1 = substr ($arr[$i-1], -1);
1011                                                 $x2 = substr ($arr[$i-1], -2, 1);
1012                                                 if ($x1 == ' ') {
1013                                                         if ($firstspace == -1) $firstspace = $i;
1014                                                 } else if ($x2 == ' ') {
1015                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1016                                                 } else {
1017                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1018                                                 }
1019                                         }
1020                                         $i++;
1021                                 }
1022
1023                                 # If there is a single-letter word, use it!
1024                                 if ($firstsingleletterword > -1)
1025                                 {
1026                                         $arr [ $firstsingleletterword ] = "''";
1027                                         $arr [ $firstsingleletterword-1 ] .= "'";
1028                                 }
1029                                 # If not, but there's a multi-letter word, use that one.
1030                                 else if ($firstmultiletterword > -1)
1031                                 {
1032                                         $arr [ $firstmultiletterword ] = "''";
1033                                         $arr [ $firstmultiletterword-1 ] .= "'";
1034                                 }
1035                                 # ... otherwise use the first one that has neither.
1036                                 # (notice that it is possible for all three to be -1 if, for example,
1037                                 # there is only one pentuple-apostrophe in the line)
1038                                 else if ($firstspace > -1)
1039                                 {
1040                                         $arr [ $firstspace ] = "''";
1041                                         $arr [ $firstspace-1 ] .= "'";
1042                                 }
1043                         }
1044
1045                         # Now let's actually convert our apostrophic mush to HTML!
1046                         $output = '';
1047                         $buffer = '';
1048                         $state = '';
1049                         $i = 0;
1050                         foreach ($arr as $r)
1051                         {
1052                                 if (($i % 2) == 0)
1053                                 {
1054                                         if ($state == 'both')
1055                                                 $buffer .= $r;
1056                                         else
1057                                                 $output .= $r;
1058                                 }
1059                                 else
1060                                 {
1061                                         if (strlen ($r) == 2)
1062                                         {
1063                                                 if ($state == 'i')
1064                                                 { $output .= '</i>'; $state = ''; }
1065                                                 else if ($state == 'bi')
1066                                                 { $output .= '</i>'; $state = 'b'; }
1067                                                 else if ($state == 'ib')
1068                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1069                                                 else if ($state == 'both')
1070                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1071                                                 else # $state can be 'b' or ''
1072                                                 { $output .= '<i>'; $state .= 'i'; }
1073                                         }
1074                                         else if (strlen ($r) == 3)
1075                                         {
1076                                                 if ($state == 'b')
1077                                                 { $output .= '</b>'; $state = ''; }
1078                                                 else if ($state == 'bi')
1079                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1080                                                 else if ($state == 'ib')
1081                                                 { $output .= '</b>'; $state = 'i'; }
1082                                                 else if ($state == 'both')
1083                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1084                                                 else # $state can be 'i' or ''
1085                                                 { $output .= '<b>'; $state .= 'b'; }
1086                                         }
1087                                         else if (strlen ($r) == 5)
1088                                         {
1089                                                 if ($state == 'b')
1090                                                 { $output .= '</b><i>'; $state = 'i'; }
1091                                                 else if ($state == 'i')
1092                                                 { $output .= '</i><b>'; $state = 'b'; }
1093                                                 else if ($state == 'bi')
1094                                                 { $output .= '</i></b>'; $state = ''; }
1095                                                 else if ($state == 'ib')
1096                                                 { $output .= '</b></i>'; $state = ''; }
1097                                                 else if ($state == 'both')
1098                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1099                                                 else # ($state == '')
1100                                                 { $buffer = ''; $state = 'both'; }
1101                                         }
1102                                 }
1103                                 $i++;
1104                         }
1105                         # Now close all remaining tags.  Notice that the order is important.
1106                         if ($state == 'b' || $state == 'ib')
1107                                 $output .= '</b>';
1108                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1109                                 $output .= '</i>';
1110                         if ($state == 'bi')
1111                                 $output .= '</b>';
1112                         if ($state == 'both')
1113                                 $output .= '<b><i>'.$buffer.'</i></b>';
1114                         return $output;
1115                 }
1116         }
1117
1118         /**
1119          * Replace external links
1120          *
1121          * Note: this is all very hackish and the order of execution matters a lot.
1122          * Make sure to run maintenance/parserTests.php if you change this code.
1123          *
1124          * @access private
1125          */
1126         function replaceExternalLinks( $text ) {
1127                 global $wgContLang;
1128                 $fname = 'Parser::replaceExternalLinks';
1129                 wfProfileIn( $fname );
1130
1131                 $sk =& $this->mOptions->getSkin();
1132
1133                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1134
1135                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1136
1137                 $i = 0;
1138                 while ( $i<count( $bits ) ) {
1139                         $url = $bits[$i++];
1140                         $protocol = $bits[$i++];
1141                         $text = $bits[$i++];
1142                         $trail = $bits[$i++];
1143
1144                         # The characters '<' and '>' (which were escaped by
1145                         # removeHTMLtags()) should not be included in
1146                         # URLs, per RFC 2396.
1147                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1148                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1149                                 $url = substr($url, 0, $m2[0][1]);
1150                         }
1151
1152                         # If the link text is an image URL, replace it with an <img> tag
1153                         # This happened by accident in the original parser, but some people used it extensively
1154                         $img = $this->maybeMakeExternalImage( $text );
1155                         if ( $img !== false ) {
1156                                 $text = $img;
1157                         }
1158
1159                         $dtrail = '';
1160
1161                         # Set linktype for CSS - if URL==text, link is essentially free
1162                         $linktype = ($text == $url) ? 'free' : 'text';
1163
1164                         # No link text, e.g. [http://domain.tld/some.link]
1165                         if ( $text == '' ) {
1166                                 # Autonumber if allowed
1167                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1168                                         $text = '[' . ++$this->mAutonumber . ']';
1169                                         $linktype = 'autonumber';
1170                                 } else {
1171                                         # Otherwise just use the URL
1172                                         $text = htmlspecialchars( $url );
1173                                         $linktype = 'free';
1174                                 }
1175                         } else {
1176                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1177                                 # Check for trail
1178                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1179                         }
1180
1181                         $text = $wgContLang->markNoConversion($text);
1182
1183                         # Replace &amp; from obsolete syntax with &.
1184                         # All HTML entities will be escaped by makeExternalLink()
1185                         $url = str_replace( '&amp;', '&', $url );
1186
1187                         # Process the trail (i.e. everything after this link up until start of the next link),
1188                         # replacing any non-bracketed links
1189                         $trail = $this->replaceFreeExternalLinks( $trail );
1190
1191                         # Use the encoded URL
1192                         # This means that users can paste URLs directly into the text
1193                         # Funny characters like &ouml; aren't valid in URLs anyway
1194                         # This was changed in August 2004
1195                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1196
1197                         # Register link in the output object.
1198                         # Replace unnecessary URL escape codes with the referenced character
1199                         # This prevents spammers from hiding links from the filters
1200                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1201                         $this->mOutput->addExternalLink( $pasteurized );
1202                 }
1203
1204                 wfProfileOut( $fname );
1205                 return $s;
1206         }
1207
1208         /**
1209          * Replace anything that looks like a URL with a link
1210          * @access private
1211          */
1212         function replaceFreeExternalLinks( $text ) {
1213                 global $wgContLang;
1214                 $fname = 'Parser::replaceFreeExternalLinks';
1215                 wfProfileIn( $fname );
1216
1217                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1218                 $s = array_shift( $bits );
1219                 $i = 0;
1220
1221                 $sk =& $this->mOptions->getSkin();
1222
1223                 while ( $i < count( $bits ) ){
1224                         $protocol = $bits[$i++];
1225                         $remainder = $bits[$i++];
1226
1227                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1228                                 # Found some characters after the protocol that look promising
1229                                 $url = $protocol . $m[1];
1230                                 $trail = $m[2];
1231
1232                                 # special case: handle urls as url args:
1233                                 # http://www.example.com/foo?=http://www.example.com/bar
1234                                 if(strlen($trail) == 0 &&
1235                                         isset($bits[$i]) &&
1236                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1237                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1238                                 {
1239                                         # add protocol, arg
1240                                         $url .= $bits[$i] . $bits[$i + 1]; # protocol, url as arg to previous link
1241                                         $i += 2;
1242                                         $trail = $m[2];
1243                                 }
1244
1245                                 # The characters '<' and '>' (which were escaped by
1246                                 # removeHTMLtags()) should not be included in
1247                                 # URLs, per RFC 2396.
1248                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1249                                         $trail = substr($url, $m2[0][1]) . $trail;
1250                                         $url = substr($url, 0, $m2[0][1]);
1251                                 }
1252
1253                                 # Move trailing punctuation to $trail
1254                                 $sep = ',;\.:!?';
1255                                 # If there is no left bracket, then consider right brackets fair game too
1256                                 if ( strpos( $url, '(' ) === false ) {
1257                                         $sep .= ')';
1258                                 }
1259
1260                                 $numSepChars = strspn( strrev( $url ), $sep );
1261                                 if ( $numSepChars ) {
1262                                         $trail = substr( $url, -$numSepChars ) . $trail;
1263                                         $url = substr( $url, 0, -$numSepChars );
1264                                 }
1265
1266                                 # Replace &amp; from obsolete syntax with &.
1267                                 # All HTML entities will be escaped by makeExternalLink()
1268                                 # or maybeMakeExternalImage()
1269                                 $url = str_replace( '&amp;', '&', $url );
1270
1271                                 # Is this an external image?
1272                                 $text = $this->maybeMakeExternalImage( $url );
1273                                 if ( $text === false ) {
1274                                         # Not an image, make a link
1275                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1276                                         # Register it in the output object...
1277                                         # Replace unnecessary URL escape codes with their equivalent characters
1278                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1279                                         $this->mOutput->addExternalLink( $pasteurized );
1280                                 }
1281                                 $s .= $text . $trail;
1282                         } else {
1283                                 $s .= $protocol . $remainder;
1284                         }
1285                 }
1286                 wfProfileOut( $fname );
1287                 return $s;
1288         }
1289
1290         /**
1291          * Replace unusual URL escape codes with their equivalent characters
1292          * @param string
1293          * @return string
1294          * @static
1295          * @fixme This can merge genuinely required bits in the path or query string,
1296          *        breaking legit URLs. A proper fix would treat the various parts of
1297          *        the URL differently; as a workaround, just use the output for
1298          *        statistical records, not for actual linking/output.
1299          */
1300         function replaceUnusualEscapes( $url ) {
1301                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1302                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1303         }
1304
1305         /**
1306          * Callback function used in replaceUnusualEscapes().
1307          * Replaces unusual URL escape codes with their equivalent character
1308          * @static
1309          * @access private
1310          */
1311         function replaceUnusualEscapesCallback( $matches ) {
1312                 $char = urldecode( $matches[0] );
1313                 $ord = ord( $char );
1314                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1315                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1316                         // No, shouldn't be escaped
1317                         return $char;
1318                 } else {
1319                         // Yes, leave it escaped
1320                         return $matches[0];
1321                 }
1322         }
1323
1324         /**
1325          * make an image if it's allowed, either through the global
1326          * option or through the exception
1327          * @access private
1328          */
1329         function maybeMakeExternalImage( $url ) {
1330                 $sk =& $this->mOptions->getSkin();
1331                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1332                 $imagesexception = !empty($imagesfrom);
1333                 $text = false;
1334                 if ( $this->mOptions->getAllowExternalImages()
1335                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1336                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1337                                 # Image found
1338                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1339                         }
1340                 }
1341                 return $text;
1342         }
1343
1344         /**
1345          * Process [[ ]] wikilinks
1346          *
1347          * @access private
1348          */
1349         function replaceInternalLinks( $s ) {
1350                 global $wgContLang;
1351                 static $fname = 'Parser::replaceInternalLinks' ;
1352
1353                 wfProfileIn( $fname );
1354
1355                 wfProfileIn( $fname.'-setup' );
1356                 static $tc = FALSE;
1357                 # the % is needed to support urlencoded titles as well
1358                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1359
1360                 $sk =& $this->mOptions->getSkin();
1361
1362                 #split the entire text string on occurences of [[
1363                 $a = explode( '[[', ' ' . $s );
1364                 #get the first element (all text up to first [[), and remove the space we added
1365                 $s = array_shift( $a );
1366                 $s = substr( $s, 1 );
1367
1368                 # Match a link having the form [[namespace:link|alternate]]trail
1369                 static $e1 = FALSE;
1370                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1371                 # Match cases where there is no "]]", which might still be images
1372                 static $e1_img = FALSE;
1373                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1374                 # Match the end of a line for a word that's not followed by whitespace,
1375                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1376                 $e2 = wfMsgForContent( 'linkprefix' );
1377
1378                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1379
1380                 if( is_null( $this->mTitle ) ) {
1381                         wfDebugDieBacktrace( 'nooo' );
1382                 }
1383                 $nottalk = !$this->mTitle->isTalkPage();
1384
1385                 if ( $useLinkPrefixExtension ) {
1386                         if ( preg_match( $e2, $s, $m ) ) {
1387                                 $first_prefix = $m[2];
1388                         } else {
1389                                 $first_prefix = false;
1390                         }
1391                 } else {
1392                         $prefix = '';
1393                 }
1394
1395                 $selflink = $this->mTitle->getPrefixedText();
1396                 wfProfileOut( $fname.'-setup' );
1397
1398                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1399                 $useSubpages = $this->areSubpagesAllowed();
1400
1401                 # Loop for each link
1402                 for ($k = 0; isset( $a[$k] ); $k++) {
1403                         $line = $a[$k];
1404                         if ( $useLinkPrefixExtension ) {
1405                                 wfProfileIn( $fname.'-prefixhandling' );
1406                                 if ( preg_match( $e2, $s, $m ) ) {
1407                                         $prefix = $m[2];
1408                                         $s = $m[1];
1409                                 } else {
1410                                         $prefix='';
1411                                 }
1412                                 # first link
1413                                 if($first_prefix) {
1414                                         $prefix = $first_prefix;
1415                                         $first_prefix = false;
1416                                 }
1417                                 wfProfileOut( $fname.'-prefixhandling' );
1418                         }
1419
1420                         $might_be_img = false;
1421
1422                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1423                                 $text = $m[2];
1424                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1425                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1426                                 # the real problem is with the $e1 regex
1427                                 # See bug 1300.
1428                                 #
1429                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1430                                 # and no image is in sight. See bug 2095.
1431                                 #
1432                                 if( $text !== '' &&
1433                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1434                                         strpos($text, '[') !== false
1435                                 )
1436                                 {
1437                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1438                                         $m[3] = $n[1];
1439                                 }
1440                                 # fix up urlencoded title texts
1441                                 if(preg_match('/%/', $m[1] ))
1442                                         # Should anchors '#' also be rejected?
1443                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1444                                 $trail = $m[3];
1445                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1446                                 $might_be_img = true;
1447                                 $text = $m[2];
1448                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1449                                 $trail = "";
1450                         } else { # Invalid form; output directly
1451                                 $s .= $prefix . '[[' . $line ;
1452                                 continue;
1453                         }
1454
1455                         # Don't allow internal links to pages containing
1456                         # PROTO: where PROTO is a valid URL protocol; these
1457                         # should be external links.
1458                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1459                                 $s .= $prefix . '[[' . $line ;
1460                                 continue;
1461                         }
1462
1463                         # Make subpage if necessary
1464                         if( $useSubpages ) {
1465                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1466                         } else {
1467                                 $link = $m[1];
1468                         }
1469
1470                         $noforce = (substr($m[1], 0, 1) != ':');
1471                         if (!$noforce) {
1472                                 # Strip off leading ':'
1473                                 $link = substr($link, 1);
1474                         }
1475
1476                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1477                         if( !$nt ) {
1478                                 $s .= $prefix . '[[' . $line;
1479                                 continue;
1480                         }
1481
1482                         #check other language variants of the link
1483                         #if the article does not exist
1484                         if( $checkVariantLink
1485                             && $nt->getArticleID() == 0 ) {
1486                                 $wgContLang->findVariantLink($link, $nt);
1487                         }
1488
1489                         $ns = $nt->getNamespace();
1490                         $iw = $nt->getInterWiki();
1491
1492                         if ($might_be_img) { # if this is actually an invalid link
1493                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1494                                         $found = false;
1495                                         while (isset ($a[$k+1]) ) {
1496                                                 #look at the next 'line' to see if we can close it there
1497                                                 $spliced = array_splice( $a, $k + 1, 1 );
1498                                                 $next_line = array_shift( $spliced );
1499                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1500                                                 # the first ]] closes the inner link, the second the image
1501                                                         $found = true;
1502                                                         $text .= '[[' . $m[1];
1503                                                         $trail = $m[2];
1504                                                         break;
1505                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1506                                                         #if there's exactly one ]] that's fine, we'll keep looking
1507                                                         $text .= '[[' . $m[0];
1508                                                 } else {
1509                                                         #if $next_line is invalid too, we need look no further
1510                                                         $text .= '[[' . $next_line;
1511                                                         break;
1512                                                 }
1513                                         }
1514                                         if ( !$found ) {
1515                                                 # we couldn't find the end of this imageLink, so output it raw
1516                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1517                                                 $text = $this->replaceInternalLinks($text);
1518                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1519                                                 # note: no $trail, because without an end, there *is* no trail
1520                                                 continue;
1521                                         }
1522                                 } else { #it's not an image, so output it raw
1523                                         $s .= $prefix . '[[' . $link . '|' . $text;
1524                                         # note: no $trail, because without an end, there *is* no trail
1525                                         continue;
1526                                 }
1527                         }
1528
1529                         $wasblank = ( '' == $text );
1530                         if( $wasblank ) $text = $link;
1531
1532
1533                         # Link not escaped by : , create the various objects
1534                         if( $noforce ) {
1535
1536                                 # Interwikis
1537                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1538                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1539                                         $s = rtrim($s . "\n");
1540                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1541                                         continue;
1542                                 }
1543
1544                                 if ( $ns == NS_IMAGE ) {
1545                                         wfProfileIn( "$fname-image" );
1546                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1547                                                 # recursively parse links inside the image caption
1548                                                 # actually, this will parse them in any other parameters, too,
1549                                                 # but it might be hard to fix that, and it doesn't matter ATM
1550                                                 $text = $this->replaceExternalLinks($text);
1551                                                 $text = $this->replaceInternalLinks($text);
1552
1553                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1554                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1555                                                 $this->mOutput->addImage( $nt->getDBkey() );
1556
1557                                                 wfProfileOut( "$fname-image" );
1558                                                 continue;
1559                                         }
1560                                         wfProfileOut( "$fname-image" );
1561
1562                                 }
1563
1564                                 if ( $ns == NS_CATEGORY ) {
1565                                         wfProfileIn( "$fname-category" );
1566                                         $s = rtrim($s . "\n"); # bug 87
1567
1568                                         if ( $wasblank ) {
1569                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1570                                                         $sortkey = $this->mTitle->getText();
1571                                                 } else {
1572                                                         $sortkey = $this->mTitle->getPrefixedText();
1573                                                 }
1574                                         } else {
1575                                                 $sortkey = $text;
1576                                         }
1577                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1578                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1579                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1580
1581                                         /**
1582                                          * Strip the whitespace Category links produce, see bug 87
1583                                          * @todo We might want to use trim($tmp, "\n") here.
1584                                          */
1585                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1586
1587                                         wfProfileOut( "$fname-category" );
1588                                         continue;
1589                                 }
1590                         }
1591
1592                         if( ( $nt->getPrefixedText() === $selflink ) &&
1593                             ( $nt->getFragment() === '' ) ) {
1594                                 # Self-links are handled specially; generally de-link and change to bold.
1595                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1596                                 continue;
1597                         }
1598
1599                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1600                         if( $ns == NS_MEDIA ) {
1601                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1602                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1603                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1604                                 $this->mOutput->addImage( $nt->getDBkey() );
1605                                 continue;
1606                         } elseif( $ns == NS_SPECIAL ) {
1607                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1608                                 continue;
1609                         } elseif( $ns == NS_IMAGE ) {
1610                                 $img = Image::newFromTitle( $nt );
1611                                 if( $img->exists() ) {
1612                                         // Force a blue link if the file exists; may be a remote
1613                                         // upload on the shared repository, and we want to see its
1614                                         // auto-generated page.
1615                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1616                                         continue;
1617                                 }
1618                         }
1619                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1620                 }
1621                 wfProfileOut( $fname );
1622                 return $s;
1623         }
1624
1625         /**
1626          * Make a link placeholder. The text returned can be later resolved to a real link with
1627          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1628          * parsing of interwiki links, and secondly to allow all extistence checks and
1629          * article length checks (for stub links) to be bundled into a single query.
1630          *
1631          */
1632         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1633                 if ( ! is_object($nt) ) {
1634                         # Fail gracefully
1635                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1636                 } else {
1637                         # Separate the link trail from the rest of the link
1638                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1639
1640                         if ( $nt->isExternal() ) {
1641                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1642                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1643                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1644                         } else {
1645                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1646                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1647                                 $this->mLinkHolders['queries'][] = $query;
1648                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1649                                 $this->mLinkHolders['titles'][] = $nt;
1650
1651                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1652                         }
1653                 }
1654                 return $retVal;
1655         }
1656
1657         /**
1658          * Render a forced-blue link inline; protect against double expansion of
1659          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1660          * Since this little disaster has to split off the trail text to avoid
1661          * breaking URLs in the following text without breaking trails on the
1662          * wiki links, it's been made into a horrible function.
1663          *
1664          * @param Title $nt
1665          * @param string $text
1666          * @param string $query
1667          * @param string $trail
1668          * @param string $prefix
1669          * @return string HTML-wikitext mix oh yuck
1670          */
1671         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1672                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1673                 $sk =& $this->mOptions->getSkin();
1674                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1675                 return $this->armorLinks( $link ) . $trail;
1676         }
1677
1678         /**
1679          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1680          * going to go through further parsing steps before inline URL expansion.
1681          *
1682          * In particular this is important when using action=render, which causes
1683          * full URLs to be included.
1684          *
1685          * Oh man I hate our multi-layer parser!
1686          *
1687          * @param string more-or-less HTML
1688          * @return string less-or-more HTML with NOPARSE bits
1689          */
1690         function armorLinks( $text ) {
1691                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1692                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1693         }
1694
1695         /**
1696          * Return true if subpage links should be expanded on this page.
1697          * @return bool
1698          */
1699         function areSubpagesAllowed() {
1700                 # Some namespaces don't allow subpages
1701                 global $wgNamespacesWithSubpages;
1702                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1703         }
1704
1705         /**
1706          * Handle link to subpage if necessary
1707          * @param string $target the source of the link
1708          * @param string &$text the link text, modified as necessary
1709          * @return string the full name of the link
1710          * @access private
1711          */
1712         function maybeDoSubpageLink($target, &$text) {
1713                 # Valid link forms:
1714                 # Foobar -- normal
1715                 # :Foobar -- override special treatment of prefix (images, language links)
1716                 # /Foobar -- convert to CurrentPage/Foobar
1717                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1718                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1719                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1720
1721                 $fname = 'Parser::maybeDoSubpageLink';
1722                 wfProfileIn( $fname );
1723                 $ret = $target; # default return value is no change
1724
1725                 # Some namespaces don't allow subpages,
1726                 # so only perform processing if subpages are allowed
1727                 if( $this->areSubpagesAllowed() ) {
1728                         # Look at the first character
1729                         if( $target != '' && $target{0} == '/' ) {
1730                                 # / at end means we don't want the slash to be shown
1731                                 if( substr( $target, -1, 1 ) == '/' ) {
1732                                         $target = substr( $target, 1, -1 );
1733                                         $noslash = $target;
1734                                 } else {
1735                                         $noslash = substr( $target, 1 );
1736                                 }
1737
1738                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1739                                 if( '' === $text ) {
1740                                         $text = $target;
1741                                 } # this might be changed for ugliness reasons
1742                         } else {
1743                                 # check for .. subpage backlinks
1744                                 $dotdotcount = 0;
1745                                 $nodotdot = $target;
1746                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1747                                         ++$dotdotcount;
1748                                         $nodotdot = substr( $nodotdot, 3 );
1749                                 }
1750                                 if($dotdotcount > 0) {
1751                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1752                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1753                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1754                                                 # / at the end means don't show full path
1755                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1756                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1757                                                         if( '' === $text ) {
1758                                                                 $text = $nodotdot;
1759                                                         }
1760                                                 }
1761                                                 $nodotdot = trim( $nodotdot );
1762                                                 if( $nodotdot != '' ) {
1763                                                         $ret .= '/' . $nodotdot;
1764                                                 }
1765                                         }
1766                                 }
1767                         }
1768                 }
1769
1770                 wfProfileOut( $fname );
1771                 return $ret;
1772         }
1773
1774         /**#@+
1775          * Used by doBlockLevels()
1776          * @access private
1777          */
1778         /* private */ function closeParagraph() {
1779                 $result = '';
1780                 if ( '' != $this->mLastSection ) {
1781                         $result = '</' . $this->mLastSection  . ">\n";
1782                 }
1783                 $this->mInPre = false;
1784                 $this->mLastSection = '';
1785                 return $result;
1786         }
1787         # getCommon() returns the length of the longest common substring
1788         # of both arguments, starting at the beginning of both.
1789         #
1790         /* private */ function getCommon( $st1, $st2 ) {
1791                 $fl = strlen( $st1 );
1792                 $shorter = strlen( $st2 );
1793                 if ( $fl < $shorter ) { $shorter = $fl; }
1794
1795                 for ( $i = 0; $i < $shorter; ++$i ) {
1796                         if ( $st1{$i} != $st2{$i} ) { break; }
1797                 }
1798                 return $i;
1799         }
1800         # These next three functions open, continue, and close the list
1801         # element appropriate to the prefix character passed into them.
1802         #
1803         /* private */ function openList( $char ) {
1804                 $result = $this->closeParagraph();
1805
1806                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1807                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1808                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1809                 else if ( ';' == $char ) {
1810                         $result .= '<dl><dt>';
1811                         $this->mDTopen = true;
1812                 }
1813                 else { $result = '<!-- ERR 1 -->'; }
1814
1815                 return $result;
1816         }
1817
1818         /* private */ function nextItem( $char ) {
1819                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1820                 else if ( ':' == $char || ';' == $char ) {
1821                         $close = '</dd>';
1822                         if ( $this->mDTopen ) { $close = '</dt>'; }
1823                         if ( ';' == $char ) {
1824                                 $this->mDTopen = true;
1825                                 return $close . '<dt>';
1826                         } else {
1827                                 $this->mDTopen = false;
1828                                 return $close . '<dd>';
1829                         }
1830                 }
1831                 return '<!-- ERR 2 -->';
1832         }
1833
1834         /* private */ function closeList( $char ) {
1835                 if ( '*' == $char ) { $text = '</li></ul>'; }
1836                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1837                 else if ( ':' == $char ) {
1838                         if ( $this->mDTopen ) {
1839                                 $this->mDTopen = false;
1840                                 $text = '</dt></dl>';
1841                         } else {
1842                                 $text = '</dd></dl>';
1843                         }
1844                 }
1845                 else {  return '<!-- ERR 3 -->'; }
1846                 return $text."\n";
1847         }
1848         /**#@-*/
1849
1850         /**
1851          * Make lists from lines starting with ':', '*', '#', etc.
1852          *
1853          * @access private
1854          * @return string the lists rendered as HTML
1855          */
1856         function doBlockLevels( $text, $linestart ) {
1857                 $fname = 'Parser::doBlockLevels';
1858                 wfProfileIn( $fname );
1859
1860                 # Parsing through the text line by line.  The main thing
1861                 # happening here is handling of block-level elements p, pre,
1862                 # and making lists from lines starting with * # : etc.
1863                 #
1864                 $textLines = explode( "\n", $text );
1865
1866                 $lastPrefix = $output = '';
1867                 $this->mDTopen = $inBlockElem = false;
1868                 $prefixLength = 0;
1869                 $paragraphStack = false;
1870
1871                 if ( !$linestart ) {
1872                         $output .= array_shift( $textLines );
1873                 }
1874                 foreach ( $textLines as $oLine ) {
1875                         $lastPrefixLength = strlen( $lastPrefix );
1876                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1877                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1878                         if ( !$this->mInPre ) {
1879                                 # Multiple prefixes may abut each other for nested lists.
1880                                 $prefixLength = strspn( $oLine, '*#:;' );
1881                                 $pref = substr( $oLine, 0, $prefixLength );
1882
1883                                 # eh?
1884                                 $pref2 = str_replace( ';', ':', $pref );
1885                                 $t = substr( $oLine, $prefixLength );
1886                                 $this->mInPre = !empty($preOpenMatch);
1887                         } else {
1888                                 # Don't interpret any other prefixes in preformatted text
1889                                 $prefixLength = 0;
1890                                 $pref = $pref2 = '';
1891                                 $t = $oLine;
1892                         }
1893
1894                         # List generation
1895                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1896                                 # Same as the last item, so no need to deal with nesting or opening stuff
1897                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1898                                 $paragraphStack = false;
1899
1900                                 if ( substr( $pref, -1 ) == ';') {
1901                                         # The one nasty exception: definition lists work like this:
1902                                         # ; title : definition text
1903                                         # So we check for : in the remainder text to split up the
1904                                         # title and definition, without b0rking links.
1905                                         $term = $t2 = '';
1906                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1907                                                 $t = $t2;
1908                                                 $output .= $term . $this->nextItem( ':' );
1909                                         }
1910                                 }
1911                         } elseif( $prefixLength || $lastPrefixLength ) {
1912                                 # Either open or close a level...
1913                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1914                                 $paragraphStack = false;
1915
1916                                 while( $commonPrefixLength < $lastPrefixLength ) {
1917                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1918                                         --$lastPrefixLength;
1919                                 }
1920                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1921                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1922                                 }
1923                                 while ( $prefixLength > $commonPrefixLength ) {
1924                                         $char = substr( $pref, $commonPrefixLength, 1 );
1925                                         $output .= $this->openList( $char );
1926
1927                                         if ( ';' == $char ) {
1928                                                 # FIXME: This is dupe of code above
1929                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1930                                                         $t = $t2;
1931                                                         $output .= $term . $this->nextItem( ':' );
1932                                                 }
1933                                         }
1934                                         ++$commonPrefixLength;
1935                                 }
1936                                 $lastPrefix = $pref2;
1937                         }
1938                         if( 0 == $prefixLength ) {
1939                                 wfProfileIn( "$fname-paragraph" );
1940                                 # No prefix (not in list)--go to paragraph mode
1941                                 // XXX: use a stack for nestable elements like span, table and div
1942                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1943                                 $closematch = preg_match(
1944                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1945                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1946                                 if ( $openmatch or $closematch ) {
1947                                         $paragraphStack = false;
1948                                         $output .= $this->closeParagraph();
1949                                         if ( $preOpenMatch and !$preCloseMatch ) {
1950                                                 $this->mInPre = true;
1951                                         }
1952                                         if ( $closematch ) {
1953                                                 $inBlockElem = false;
1954                                         } else {
1955                                                 $inBlockElem = true;
1956                                         }
1957                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1958                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1959                                                 // pre
1960                                                 if ($this->mLastSection != 'pre') {
1961                                                         $paragraphStack = false;
1962                                                         $output .= $this->closeParagraph().'<pre>';
1963                                                         $this->mLastSection = 'pre';
1964                                                 }
1965                                                 $t = substr( $t, 1 );
1966                                         } else {
1967                                                 // paragraph
1968                                                 if ( '' == trim($t) ) {
1969                                                         if ( $paragraphStack ) {
1970                                                                 $output .= $paragraphStack.'<br />';
1971                                                                 $paragraphStack = false;
1972                                                                 $this->mLastSection = 'p';
1973                                                         } else {
1974                                                                 if ($this->mLastSection != 'p' ) {
1975                                                                         $output .= $this->closeParagraph();
1976                                                                         $this->mLastSection = '';
1977                                                                         $paragraphStack = '<p>';
1978                                                                 } else {
1979                                                                         $paragraphStack = '</p><p>';
1980                                                                 }
1981                                                         }
1982                                                 } else {
1983                                                         if ( $paragraphStack ) {
1984                                                                 $output .= $paragraphStack;
1985                                                                 $paragraphStack = false;
1986                                                                 $this->mLastSection = 'p';
1987                                                         } else if ($this->mLastSection != 'p') {
1988                                                                 $output .= $this->closeParagraph().'<p>';
1989                                                                 $this->mLastSection = 'p';
1990                                                         }
1991                                                 }
1992                                         }
1993                                 }
1994                                 wfProfileOut( "$fname-paragraph" );
1995                         }
1996                         // somewhere above we forget to get out of pre block (bug 785)
1997                         if($preCloseMatch && $this->mInPre) {
1998                                 $this->mInPre = false;
1999                         }
2000                         if ($paragraphStack === false) {
2001                                 $output .= $t."\n";
2002                         }
2003                 }
2004                 while ( $prefixLength ) {
2005                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2006                         --$prefixLength;
2007                 }
2008                 if ( '' != $this->mLastSection ) {
2009                         $output .= '</' . $this->mLastSection . '>';
2010                         $this->mLastSection = '';
2011                 }
2012
2013                 wfProfileOut( $fname );
2014                 return $output;
2015         }
2016
2017         /**
2018          * Split up a string on ':', ignoring any occurences inside
2019          * <a>..</a> or <span>...</span>
2020          * @param string $str the string to split
2021          * @param string &$before set to everything before the ':'
2022          * @param string &$after set to everything after the ':'
2023          * return string the position of the ':', or false if none found
2024          */
2025         function findColonNoLinks($str, &$before, &$after) {
2026                 # I wonder if we should make this count all tags, not just <a>
2027                 # and <span>. That would prevent us from matching a ':' that
2028                 # comes in the middle of italics other such formatting....
2029                 # -- Wil
2030                 $fname = 'Parser::findColonNoLinks';
2031                 wfProfileIn( $fname );
2032                 $pos = 0;
2033                 do {
2034                         $colon = strpos($str, ':', $pos);
2035
2036                         if ($colon !== false) {
2037                                 $before = substr($str, 0, $colon);
2038                                 $after = substr($str, $colon + 1);
2039
2040                                 # Skip any ':' within <a> or <span> pairs
2041                                 $a = substr_count($before, '<a');
2042                                 $s = substr_count($before, '<span');
2043                                 $ca = substr_count($before, '</a>');
2044                                 $cs = substr_count($before, '</span>');
2045
2046                                 if ($a <= $ca and $s <= $cs) {
2047                                         # Tags are balanced before ':'; ok
2048                                         break;
2049                                 }
2050                                 $pos = $colon + 1;
2051                         }
2052                 } while ($colon !== false);
2053                 wfProfileOut( $fname );
2054                 return $colon;
2055         }
2056
2057         /**
2058          * Return value of a magic variable (like PAGENAME)
2059          *
2060          * @access private
2061          */
2062         function getVariableValue( $index ) {
2063                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2064
2065                 /**
2066                  * Some of these require message or data lookups and can be
2067                  * expensive to check many times.
2068                  */
2069                 static $varCache = array();
2070                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2071                         if ( isset( $varCache[$index] ) )
2072                                 return $varCache[$index];
2073
2074                 $ts = time();
2075                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2076
2077                 switch ( $index ) {
2078                         case MAG_CURRENTMONTH:
2079                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2080                         case MAG_CURRENTMONTHNAME:
2081                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2082                         case MAG_CURRENTMONTHNAMEGEN:
2083                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2084                         case MAG_CURRENTMONTHABBREV:
2085                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2086                         case MAG_CURRENTDAY:
2087                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2088                         case MAG_CURRENTDAY2:
2089                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2090                         case MAG_PAGENAME:
2091                                 return $this->mTitle->getText();
2092                         case MAG_PAGENAMEE:
2093                                 return $this->mTitle->getPartialURL();
2094                         case MAG_FULLPAGENAME:
2095                                 return $this->mTitle->getPrefixedText();
2096                         case MAG_FULLPAGENAMEE:
2097                                 return $this->mTitle->getPrefixedURL();
2098                         case MAG_SUBPAGENAME:
2099                                 return $this->mTitle->getSubpageText();
2100                         case MAG_SUBPAGENAMEE:
2101                                 return $this->mTitle->getSubpageUrlForm();
2102                         case MAG_REVISIONID:
2103                                 return $this->mRevisionId;
2104                         case MAG_NAMESPACE:
2105                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2106                         case MAG_NAMESPACEE:
2107                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2108                         case MAG_CURRENTDAYNAME:
2109                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2110                         case MAG_CURRENTYEAR:
2111                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2112                         case MAG_CURRENTTIME:
2113                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2114                         case MAG_CURRENTWEEK:
2115                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2116                                 // int to remove the padding
2117                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2118                         case MAG_CURRENTDOW:
2119                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2120                         case MAG_NUMBEROFARTICLES:
2121                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2122                         case MAG_NUMBEROFFILES:
2123                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2124                         case MAG_SITENAME:
2125                                 return $wgSitename;
2126                         case MAG_SERVER:
2127                                 return $wgServer;
2128                         case MAG_SERVERNAME:
2129                                 return $wgServerName;
2130                         case MAG_SCRIPTPATH:
2131                                 return $wgScriptPath;
2132                         default:
2133                                 $ret = null;
2134                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2135                                         return $ret;
2136                                 else
2137                                         return null;
2138                 }
2139         }
2140
2141         /**
2142          * initialise the magic variables (like CURRENTMONTHNAME)
2143          *
2144          * @access private
2145          */
2146         function initialiseVariables() {
2147                 $fname = 'Parser::initialiseVariables';
2148                 wfProfileIn( $fname );
2149                 global $wgVariableIDs;
2150                 $this->mVariables = array();
2151                 foreach ( $wgVariableIDs as $id ) {
2152                         $mw =& MagicWord::get( $id );
2153                         $mw->addToArray( $this->mVariables, $id );
2154                 }
2155                 wfProfileOut( $fname );
2156         }
2157
2158         /**
2159          * parse any parentheses in format ((title|part|part))
2160          * and call callbacks to get a replacement text for any found piece
2161          *
2162          * @param string $text The text to parse
2163          * @param array $callbacks rules in form:
2164          *     '{' => array(                            # opening parentheses
2165          *                                      'end' => '}',   # closing parentheses
2166          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2167          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2168          *                                                                )
2169          *                                      )
2170          * @access private
2171          */
2172         function replace_callback ($text, $callbacks) {
2173                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2174                 $lastOpeningBrace = -1;                 # last not closed parentheses
2175
2176                 for ($i = 0; $i < strlen($text); $i++) {
2177                         # check for any opening brace
2178                         $rule = null;
2179                         $nextPos = -1;
2180                         foreach ($callbacks as $key => $value) {
2181                                 $pos = strpos ($text, $key, $i);
2182                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2183                                         $rule = $value;
2184                                         $nextPos = $pos;
2185                                 }
2186                         }
2187
2188                         if ($lastOpeningBrace >= 0) {
2189                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2190
2191                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2192                                         $rule = null;
2193                                         $nextPos = $pos;
2194                                 }
2195
2196                                 $pos = strpos ($text, '|', $i);
2197
2198                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2199                                         $rule = null;
2200                                         $nextPos = $pos;
2201                                 }
2202                         }
2203
2204                         if ($nextPos == -1)
2205                                 break;
2206
2207                         $i = $nextPos;
2208
2209                         # found openning brace, lets add it to parentheses stack
2210                         if (null != $rule) {
2211                                 $piece = array('brace' => $text[$i],
2212                                                            'braceEnd' => $rule['end'],
2213                                                            'count' => 1,
2214                                                            'title' => '',
2215                                                            'parts' => null);
2216
2217                                 # count openning brace characters
2218                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2219                                         $piece['count']++;
2220                                         $i++;
2221                                 }
2222
2223                                 $piece['startAt'] = $i+1;
2224                                 $piece['partStart'] = $i+1;
2225
2226                                 # we need to add to stack only if openning brace count is enough for any given rule
2227                                 foreach ($rule['cb'] as $cnt => $fn) {
2228                                         if ($piece['count'] >= $cnt) {
2229                                                 $lastOpeningBrace ++;
2230                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2231                                                 break;
2232                                         }
2233                                 }
2234
2235                                 continue;
2236                         }
2237                         else if ($lastOpeningBrace >= 0) {
2238                                 # first check if it is a closing brace
2239                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2240                                         # lets check if it is enough characters for closing brace
2241                                         $count = 1;
2242                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2243                                                 $count++;
2244
2245                                         # if there are more closing parentheses than opening ones, we parse less
2246                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2247                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2248
2249                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2250                                         $matchingCount = 0;
2251                                         $matchingCallback = null;
2252                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2253                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2254                                                         $matchingCount = $cnt;
2255                                                         $matchingCallback = $fn;
2256                                                 }
2257                                         }
2258
2259                                         if ($matchingCount == 0) {
2260                                                 $i += $count - 1;
2261                                                 continue;
2262                                         }
2263
2264                                         # lets set a title or last part (if '|' was found)
2265                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2266                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2267                                         else
2268                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2269
2270                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2271                                         $pieceEnd = $i + $matchingCount;
2272
2273                                         if( is_callable( $matchingCallback ) ) {
2274                                                 $cbArgs = array (
2275                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2276                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2277                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2278                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2279                                                                                  );
2280                                                 # finally we can call a user callback and replace piece of text
2281                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2282                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2283                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2284                                         }
2285                                         else {
2286                                                 # null value for callback means that parentheses should be parsed, but not replaced
2287                                                 $i += $matchingCount - 1;
2288                                         }
2289
2290                                         # reset last openning parentheses, but keep it in case there are unused characters
2291                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2292                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2293                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2294                                                                    'title' => '',
2295                                                                    'parts' => null,
2296                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2297                                         $openingBraceStack[$lastOpeningBrace--] = null;
2298
2299                                         if ($matchingCount < $piece['count']) {
2300                                                 $piece['count'] -= $matchingCount;
2301                                                 $piece['startAt'] -= $matchingCount;
2302                                                 $piece['partStart'] = $piece['startAt'];
2303                                                 # do we still qualify for any callback with remaining count?
2304                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2305                                                         if ($piece['count'] >= $cnt) {
2306                                                                 $lastOpeningBrace ++;
2307                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2308                                                                 break;
2309                                                         }
2310                                                 }
2311                                         }
2312                                         continue;
2313                                 }
2314
2315                                 # lets set a title if it is a first separator, or next part otherwise
2316                                 if ($text[$i] == '|') {
2317                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2318                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2319                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2320                                         }
2321                                         else
2322                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2323
2324                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2325                                 }
2326                         }
2327                 }
2328
2329                 return $text;
2330         }
2331
2332         /**
2333          * Replace magic variables, templates, and template arguments
2334          * with the appropriate text. Templates are substituted recursively,
2335          * taking care to avoid infinite loops.
2336          *
2337          * Note that the substitution depends on value of $mOutputType:
2338          *  OT_WIKI: only {{subst:}} templates
2339          *  OT_MSG: only magic variables
2340          *  OT_HTML: all templates and magic variables
2341          *
2342          * @param string $tex The text to transform
2343          * @param array $args Key-value pairs representing template parameters to substitute
2344          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2345          * @access private
2346          */
2347         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2348                 # Prevent too big inclusions
2349                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2350                         return $text;
2351                 }
2352
2353                 $fname = 'Parser::replaceVariables';
2354                 wfProfileIn( $fname );
2355
2356                 # This function is called recursively. To keep track of arguments we need a stack:
2357                 array_push( $this->mArgStack, $args );
2358
2359                 $braceCallbacks = array();
2360                 if ( !$argsOnly ) {
2361                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2362                 }
2363                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2364                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2365                 }
2366                 $callbacks = array();
2367                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2368                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2369                 $text = $this->replace_callback ($text, $callbacks);
2370
2371                 array_pop( $this->mArgStack );
2372
2373                 wfProfileOut( $fname );
2374                 return $text;
2375         }
2376
2377         /**
2378          * Replace magic variables
2379          * @access private
2380          */
2381         function variableSubstitution( $matches ) {
2382                 $fname = 'Parser::variableSubstitution';
2383                 $varname = $matches[1];
2384                 wfProfileIn( $fname );
2385                 if ( !$this->mVariables ) {
2386                         $this->initialiseVariables();
2387                 }
2388                 $skip = false;
2389                 if ( $this->mOutputType == OT_WIKI ) {
2390                         # Do only magic variables prefixed by SUBST
2391                         $mwSubst =& MagicWord::get( MAG_SUBST );
2392                         if (!$mwSubst->matchStartAndRemove( $varname ))
2393                                 $skip = true;
2394                         # Note that if we don't substitute the variable below,
2395                         # we don't remove the {{subst:}} magic word, in case
2396                         # it is a template rather than a magic variable.
2397                 }
2398                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2399                         $id = $this->mVariables[$varname];
2400                         $text = $this->getVariableValue( $id );
2401                         $this->mOutput->mContainsOldMagic = true;
2402                 } else {
2403                         $text = $matches[0];
2404                 }
2405                 wfProfileOut( $fname );
2406                 return $text;
2407         }
2408
2409         # Split template arguments
2410         function getTemplateArgs( $argsString ) {
2411                 if ( $argsString === '' ) {
2412                         return array();
2413                 }
2414
2415                 $args = explode( '|', substr( $argsString, 1 ) );
2416
2417                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2418                 # merged with the next arg because the '|' character between belongs
2419                 # to the link syntax and not the template parameter syntax.
2420                 $argc = count($args);
2421
2422                 for ( $i = 0; $i < $argc-1; $i++ ) {
2423                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2424                                 $args[$i] .= '|'.$args[$i+1];
2425                                 array_splice($args, $i+1, 1);
2426                                 $i--;
2427                                 $argc--;
2428                         }
2429                 }
2430
2431                 return $args;
2432         }
2433
2434         /**
2435          * Return the text of a template, after recursively
2436          * replacing any variables or templates within the template.
2437          *
2438          * @param array $piece The parts of the template
2439          *  $piece['text']: matched text
2440          *  $piece['title']: the title, i.e. the part before the |
2441          *  $piece['parts']: the parameter array
2442          * @return string the text of the template
2443          * @access private
2444          */
2445         function braceSubstitution( $piece ) {
2446                 global $wgContLang;
2447                 $fname = 'Parser::braceSubstitution';
2448                 wfProfileIn( $fname );
2449
2450                 # Flags
2451                 $found = false;             # $text has been filled
2452                 $nowiki = false;            # wiki markup in $text should be escaped
2453                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2454                 $noargs = false;            # Don't replace triple-brace arguments in $text
2455                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2456                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2457                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2458
2459                 # Title object, where $text came from
2460                 $title = NULL;
2461
2462                 $linestart = '';
2463
2464                 # $part1 is the bit before the first |, and must contain only title characters
2465                 # $args is a list of arguments, starting from index 0, not including $part1
2466
2467                 $part1 = $piece['title'];
2468                 # If the third subpattern matched anything, it will start with |
2469
2470                 if (null == $piece['parts']) {
2471                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2472                         if ($replaceWith != $piece['text']) {
2473                                 $text = $replaceWith;
2474                                 $found = true;
2475                                 $noparse = true;
2476                                 $noargs = true;
2477                         }
2478                 }
2479
2480                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2481                 $argc = count( $args );
2482
2483                 # SUBST
2484                 if ( !$found ) {
2485                         $mwSubst =& MagicWord::get( MAG_SUBST );
2486                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2487                                 # One of two possibilities is true:
2488                                 # 1) Found SUBST but not in the PST phase
2489                                 # 2) Didn't find SUBST and in the PST phase
2490                                 # In either case, return without further processing
2491                                 $text = $piece['text'];
2492                                 $found = true;
2493                                 $noparse = true;
2494                                 $noargs = true;
2495                         }
2496                 }
2497
2498                 # MSG, MSGNW, INT and RAW
2499                 if ( !$found ) {
2500                         # Check for MSGNW:
2501                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2502                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2503                                 $nowiki = true;
2504                         } else {
2505                                 # Remove obsolete MSG:
2506                                 $mwMsg =& MagicWord::get( MAG_MSG );
2507                                 $mwMsg->matchStartAndRemove( $part1 );
2508                         }
2509
2510                         # Check for RAW:
2511                         $mwRaw =& MagicWord::get( MAG_RAW );
2512                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2513                                 $forceRawInterwiki = true;
2514                         }
2515
2516                         # Check if it is an internal message
2517                         $mwInt =& MagicWord::get( MAG_INT );
2518                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2519                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2520                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2521                                         $found = true;
2522                                 }
2523                         }
2524                 }
2525
2526                 # NS
2527                 if ( !$found ) {
2528                         # Check for NS: (namespace expansion)
2529                         $mwNs = MagicWord::get( MAG_NS );
2530                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2531                                 if ( intval( $part1 ) || $part1 == "0" ) {
2532                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2533                                         $found = true;
2534                                 } else {
2535                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2536                                         if ( !is_null( $index ) ) {
2537                                                 $text = $linestart . $wgContLang->getNsText( $index );
2538                                                 $found = true;
2539                                         }
2540                                 }
2541                         }
2542                 }
2543
2544                 # LCFIRST, UCFIRST, LC and UC
2545                 if ( !$found ) {
2546                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2547                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2548                         $lc =& MagicWord::get( MAG_LC );
2549                         $uc =& MagicWord::get( MAG_UC );
2550                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2551                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2552                                 $found = true;
2553                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2554                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2555                                 $found = true;
2556                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2557                                 $text = $linestart . $wgContLang->lc( $part1 );
2558                                 $found = true;
2559                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2560                                  $text = $linestart . $wgContLang->uc( $part1 );
2561                                  $found = true;
2562                         }
2563                 }
2564
2565                 # LOCALURL and FULLURL
2566                 if ( !$found ) {
2567                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2568                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2569                         $mwFull =& MagicWord::get( MAG_FULLURL );
2570                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2571
2572
2573                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2574                                 $func = 'getLocalURL';
2575                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2576                                 $func = 'escapeLocalURL';
2577                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2578                                 $func = 'getFullURL';
2579                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2580                                 $func = 'escapeFullURL';
2581                         } else {
2582                                 $func = false;
2583                         }
2584
2585                         if ( $func !== false ) {
2586                                 $title = Title::newFromText( $part1 );
2587                                 if ( !is_null( $title ) ) {
2588                                         if ( $argc > 0 ) {
2589                                                 $text = $linestart . $title->$func( $args[0] );
2590                                         } else {
2591                                                 $text = $linestart . $title->$func();
2592                                         }
2593                                         $found = true;
2594                                 }
2595                         }
2596                 }
2597
2598                 # GRAMMAR
2599                 if ( !$found && $argc == 1 ) {
2600                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2601                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2602                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2603                                 $found = true;
2604                         }
2605                 }
2606
2607                 # PLURAL
2608                 if ( !$found && $argc >= 2 ) {
2609                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2610                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2611                                 if ($argc==2) {$args[2]=$args[1];}
2612                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2613                                 $found = true;
2614                         }
2615                 }
2616
2617                 # Extensions
2618                 if ( !$found ) {
2619                         $colonPos = strpos( $part1, ':' );
2620                         if ( $colonPos !== false ) {
2621                                 $function = strtolower( substr( $part1, 0, $colonPos ) );
2622                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2623                                         $funcArgs = array_merge( array( &$this, substr( $part1, $colonPos + 1 ) ), $args );
2624                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2625                                         $found = true;
2626                                         if ( is_array( $result ) ) {
2627                                                 $text = $linestart . $result[0];
2628                                                 unset( $result[0] );
2629
2630                                                 // Extract flags into the local scope
2631                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2632                                                 extract( $result );
2633                                         } else {
2634                                                 $text = $linestart . $result;
2635                                         }
2636                                 }
2637                         }
2638                 }
2639
2640                 # Template table test
2641
2642                 # Did we encounter this template already? If yes, it is in the cache
2643                 # and we need to check for loops.
2644                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2645                         $found = true;
2646
2647                         # Infinite loop test
2648                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2649                                 $noparse = true;
2650                                 $noargs = true;
2651                                 $found = true;
2652                                 $text = $linestart .
2653                                         '{{' . $part1 . '}}' .
2654                                         '<!-- WARNING: template loop detected -->';
2655                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2656                         } else {
2657                                 # set $text to cached message.
2658                                 $text = $linestart . $this->mTemplates[$piece['title']];
2659                         }
2660                 }
2661
2662                 # Load from database
2663                 $lastPathLevel = $this->mTemplatePath;
2664                 if ( !$found ) {
2665                         $ns = NS_TEMPLATE;
2666                         # declaring $subpage directly in the function call
2667                         # does not work correctly with references and breaks
2668                         # {{/subpage}}-style inclusions
2669                         $subpage = '';
2670                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2671                         if ($subpage !== '') {
2672                                 $ns = $this->mTitle->getNamespace();
2673                         }
2674                         $title = Title::newFromText( $part1, $ns );
2675
2676                         if ( !is_null( $title ) ) {
2677                                 if ( !$title->isExternal() ) {
2678                                         # Check for excessive inclusion
2679                                         $dbk = $title->getPrefixedDBkey();
2680                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2681                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2682                                                         # Capture special page output
2683                                                         $text = SpecialPage::capturePath( $title );
2684                                                         if ( is_string( $text ) ) {
2685                                                                 $found = true;
2686                                                                 $noparse = true;
2687                                                                 $noargs = true;
2688                                                                 $isHTML = true;
2689                                                                 $this->disableCache();
2690                                                         }
2691                                                 } else {
2692                                                         $articleContent = $this->fetchTemplate( $title );
2693                                                         if ( $articleContent !== false ) {
2694                                                                 $found = true;
2695                                                                 $text = $articleContent;
2696                                                                 $replaceHeadings = true;
2697                                                         }
2698                                                 }
2699                                         }
2700
2701                                         # If the title is valid but undisplayable, make a link to it
2702                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2703                                                 $text = '[['.$title->getPrefixedText().']]';
2704                                                 $found = true;
2705                                         }
2706                                 } elseif ( $title->isTrans() ) {
2707                                         // Interwiki transclusion
2708                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2709                                                 $text = $this->interwikiTransclude( $title, 'render' );
2710                                                 $isHTML = true;
2711                                                 $noparse = true;
2712                                         } else {
2713                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2714                                                 $replaceHeadings = true;
2715                                         }
2716                                         $found = true;
2717                                 }
2718
2719                                 # Template cache array insertion
2720                                 # Use the original $piece['title'] not the mangled $part1, so that
2721                                 # modifiers such as RAW: produce separate cache entries
2722                                 if( $found ) {
2723                                         $this->mTemplates[$piece['title']] = $text;
2724                                         $text = $linestart . $text;
2725                                 }
2726                         }
2727                 }
2728
2729                 # Recursive parsing, escaping and link table handling
2730                 # Only for HTML output
2731                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2732                         $text = wfEscapeWikiText( $text );
2733                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2734                         if ( !$noargs ) {
2735                                 # Clean up argument array
2736                                 $assocArgs = array();
2737                                 $index = 1;
2738                                 foreach( $args as $arg ) {
2739                                         $eqpos = strpos( $arg, '=' );
2740                                         if ( $eqpos === false ) {
2741                                                 $assocArgs[$index++] = $arg;
2742                                         } else {
2743                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2744                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2745                                                 if ( $value === false ) {
2746                                                         $value = '';
2747                                                 }
2748                                                 if ( $name !== false ) {
2749                                                         $assocArgs[$name] = $value;
2750                                                 }
2751                                         }
2752                                 }
2753
2754                                 # Add a new element to the templace recursion path
2755                                 $this->mTemplatePath[$part1] = 1;
2756                         }
2757
2758                         if ( !$noparse ) {
2759                                 # If there are any <onlyinclude> tags, only include them
2760                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2761                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2762                                         $text = '';
2763                                         foreach ($m[1] as $piece)
2764                                                 $text .= $piece;
2765                                 }
2766                                 # Remove <noinclude> sections and <includeonly> tags
2767                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2768                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2769
2770                                 if( $this->mOutputType == OT_HTML ) {
2771                                         # Strip <nowiki>, <pre>, etc.
2772                                         $text = $this->strip( $text, $this->mStripState );
2773                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2774                                 }
2775                                 $text = $this->replaceVariables( $text, $assocArgs );
2776
2777                                 # If the template begins with a table or block-level
2778                                 # element, it should be treated as beginning a new line.
2779                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2780                                         $text = "\n" . $text;
2781                                 }
2782                         } elseif ( !$noargs ) {
2783                                 # $noparse and !$noargs
2784                                 # Just replace the arguments, not any double-brace items
2785                                 # This is used for rendered interwiki transclusion
2786                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2787                         }
2788                 }
2789                 # Prune lower levels off the recursion check path
2790                 $this->mTemplatePath = $lastPathLevel;
2791
2792                 if ( !$found ) {
2793                         wfProfileOut( $fname );
2794                         return $piece['text'];
2795                 } else {
2796                         if ( $isHTML ) {
2797                                 # Replace raw HTML by a placeholder
2798                                 # Add a blank line preceding, to prevent it from mucking up
2799                                 # immediately preceding headings
2800                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2801                         } else {
2802                                 # replace ==section headers==
2803                                 # XXX this needs to go away once we have a better parser.
2804                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2805                                         if( !is_null( $title ) )
2806                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2807                                         else
2808                                                 $encodedname = base64_encode("");
2809                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2810                                                 PREG_SPLIT_DELIM_CAPTURE);
2811                                         $text = '';
2812                                         $nsec = 0;
2813                                         for( $i = 0; $i < count($m); $i += 2 ) {
2814                                                 $text .= $m[$i];
2815                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2816                                                 $hl = $m[$i + 1];
2817                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2818                                                         $text .= $hl;
2819                                                         continue;
2820                                                 }
2821                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2822                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2823                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2824
2825                                                 $nsec++;
2826                                         }
2827                                 }
2828                         }
2829                 }
2830
2831                 # Prune lower levels off the recursion check path
2832                 $this->mTemplatePath = $lastPathLevel;
2833
2834                 if ( !$found ) {
2835                         wfProfileOut( $fname );
2836                         return $piece['text'];
2837                 } else {
2838                         wfProfileOut( $fname );
2839                         return $text;
2840                 }
2841         }
2842
2843         /**
2844          * Fetch the unparsed text of a template and register a reference to it.
2845          */
2846         function fetchTemplate( $title ) {
2847                 $text = false;
2848                 // Loop to fetch the article, with up to 1 redirect
2849                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2850                         $rev = Revision::newFromTitle( $title );
2851                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2852                         if ( !$rev ) {
2853                                 break;
2854                         }
2855                         $text = $rev->getText();
2856                         if ( $text === false ) {
2857                                 break;
2858                         }
2859                         // Redirect?
2860                         $title = Title::newFromRedirect( $text );
2861                 }
2862                 return $text;
2863         }
2864
2865         /**
2866          * Transclude an interwiki link.
2867          */
2868         function interwikiTransclude( $title, $action ) {
2869                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2870
2871                 if (!$wgEnableScaryTranscluding)
2872                         return wfMsg('scarytranscludedisabled');
2873
2874                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2875                 // But we'll handle it generally anyway
2876                 if ( $title->getNamespace() ) {
2877                         // Use the canonical namespace, which should work anywhere
2878                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2879                 } else {
2880                         $articleName = $title->getDBkey();
2881                 }
2882
2883                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2884                 $url .= "?action=$action";
2885                 if (strlen($url) > 255)
2886                         return wfMsg('scarytranscludetoolong');
2887                 return $this->fetchScaryTemplateMaybeFromCache($url);
2888         }
2889
2890         function fetchScaryTemplateMaybeFromCache($url) {
2891                 global $wgTranscludeCacheExpiry;
2892                 $dbr =& wfGetDB(DB_SLAVE);
2893                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2894                                 array('tc_url' => $url));
2895                 if ($obj) {
2896                         $time = $obj->tc_time;
2897                         $text = $obj->tc_contents;
2898                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2899                                 return $text;
2900                         }
2901                 }
2902
2903                 $text = wfGetHTTP($url);
2904                 if (!$text)
2905                         return wfMsg('scarytranscludefailed', $url);
2906
2907                 $dbw =& wfGetDB(DB_MASTER);
2908                 $dbw->replace('transcache', array('tc_url'), array(
2909                         'tc_url' => $url,
2910                         'tc_time' => time(),
2911                         'tc_contents' => $text));
2912                 return $text;
2913         }
2914
2915
2916         /**
2917          * Triple brace replacement -- used for template arguments
2918          * @access private
2919          */
2920         function argSubstitution( $matches ) {
2921                 $arg = trim( $matches['title'] );
2922                 $text = $matches['text'];
2923                 $inputArgs = end( $this->mArgStack );
2924
2925                 if ( array_key_exists( $arg, $inputArgs ) ) {
2926                         $text = $inputArgs[$arg];
2927                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2928                         $text = $matches['parts'][0];
2929                 }
2930
2931                 return $text;
2932         }
2933
2934         /**
2935          * Returns true if the function is allowed to include this entity
2936          * @access private
2937          */
2938         function incrementIncludeCount( $dbk ) {
2939                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2940                         $this->mIncludeCount[$dbk] = 0;
2941                 }
2942                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2943                         return true;
2944                 } else {
2945                         return false;
2946                 }
2947         }
2948
2949         /**
2950          * This function accomplishes several tasks:
2951          * 1) Auto-number headings if that option is enabled
2952          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2953          * 3) Add a Table of contents on the top for users who have enabled the option
2954          * 4) Auto-anchor headings
2955          *
2956          * It loops through all headlines, collects the necessary data, then splits up the
2957          * string and re-inserts the newly formatted headlines.
2958          *
2959          * @param string $text
2960          * @param boolean $isMain
2961          * @access private
2962          */
2963         function formatHeadings( $text, $isMain=true ) {
2964                 global $wgMaxTocLevel, $wgContLang;
2965
2966                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2967                 $doShowToc = true;
2968                 $forceTocHere = false;
2969                 if( !$this->mTitle->userCanEdit() ) {
2970                         $showEditLink = 0;
2971                 } else {
2972                         $showEditLink = $this->mOptions->getEditSection();
2973                 }
2974
2975                 # Inhibit editsection links if requested in the page
2976                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2977                 if( $esw->matchAndRemove( $text ) ) {
2978                         $showEditLink = 0;
2979                 }
2980                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2981                 # do not add TOC
2982                 $mw =& MagicWord::get( MAG_NOTOC );
2983                 if( $mw->matchAndRemove( $text ) ) {
2984                         $doShowToc = false;
2985                 }
2986
2987                 # Get all headlines for numbering them and adding funky stuff like [edit]
2988                 # links - this is for later, but we need the number of headlines right now
2989                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2990
2991                 # if there are fewer than 4 headlines in the article, do not show TOC
2992                 if( $numMatches < 4 ) {
2993                         $doShowToc = false;
2994                 }
2995
2996                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2997                 # override above conditions and always show TOC at that place
2998
2999                 $mw =& MagicWord::get( MAG_TOC );
3000                 if($mw->match( $text ) ) {
3001                         $doShowToc = true;
3002                         $forceTocHere = true;
3003                 } else {
3004                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3005                         # override above conditions and always show TOC above first header
3006                         $mw =& MagicWord::get( MAG_FORCETOC );
3007                         if ($mw->matchAndRemove( $text ) ) {
3008                                 $doShowToc = true;
3009                         }
3010                 }
3011
3012                 # Never ever show TOC if no headers
3013                 if( $numMatches < 1 ) {
3014                         $doShowToc = false;
3015                 }
3016
3017                 # We need this to perform operations on the HTML
3018                 $sk =& $this->mOptions->getSkin();
3019
3020                 # headline counter
3021                 $headlineCount = 0;
3022                 $sectionCount = 0; # headlineCount excluding template sections
3023
3024                 # Ugh .. the TOC should have neat indentation levels which can be
3025                 # passed to the skin functions. These are determined here
3026                 $toc = '';
3027                 $full = '';
3028                 $head = array();
3029                 $sublevelCount = array();
3030                 $levelCount = array();
3031                 $toclevel = 0;
3032                 $level = 0;
3033                 $prevlevel = 0;
3034                 $toclevel = 0;
3035                 $prevtoclevel = 0;
3036
3037                 foreach( $matches[3] as $headline ) {
3038                         $istemplate = 0;
3039                         $templatetitle = '';
3040                         $templatesection = 0;
3041                         $numbering = '';
3042
3043                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3044                                 $istemplate = 1;
3045                                 $templatetitle = base64_decode($mat[1]);
3046                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3047                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3048                         }
3049
3050                         if( $toclevel ) {
3051                                 $prevlevel = $level;
3052                                 $prevtoclevel = $toclevel;
3053                         }
3054                         $level = $matches[1][$headlineCount];
3055
3056                         if( $doNumberHeadings || $doShowToc ) {
3057
3058                                 if ( $level > $prevlevel ) {
3059                                         # Increase TOC level
3060                                         $toclevel++;
3061                                         $sublevelCount[$toclevel] = 0;
3062                                         $toc .= $sk->tocIndent();
3063                                 }
3064                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3065                                         # Decrease TOC level, find level to jump to
3066
3067                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3068                                                 # Can only go down to level 1
3069                                                 $toclevel = 1;
3070                                         } else {
3071                                                 for ($i = $toclevel; $i > 0; $i--) {
3072                                                         if ( $levelCount[$i] == $level ) {
3073                                                                 # Found last matching level
3074                                                                 $toclevel = $i;
3075                                                                 break;
3076                                                         }
3077                                                         elseif ( $levelCount[$i] < $level ) {
3078                                                                 # Found first matching level below current level
3079                                                                 $toclevel = $i + 1;
3080                                                                 break;
3081                                                         }
3082                                                 }
3083                                         }
3084
3085                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3086                                 }
3087                                 else {
3088                                         # No change in level, end TOC line
3089                                         $toc .= $sk->tocLineEnd();
3090                                 }
3091
3092                                 $levelCount[$toclevel] = $level;
3093
3094                                 # count number of headlines for each level
3095                                 @$sublevelCount[$toclevel]++;
3096                                 $dot = 0;
3097                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3098                                         if( !empty( $sublevelCount[$i] ) ) {
3099                                                 if( $dot ) {
3100                                                         $numbering .= '.';
3101                                                 }
3102                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3103                                                 $dot = 1;
3104                                         }
3105                                 }
3106                         }
3107
3108                         # The canonized header is a version of the header text safe to use for links
3109                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3110                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3111                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3112
3113                         # Remove link placeholders by the link text.
3114                         #     <!--LINK number-->
3115                         # turns into
3116                         #     link text with suffix
3117                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3118                                                             "\$this->mLinkHolders['texts'][\$1]",
3119                                                             $canonized_headline );
3120                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3121                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3122                                                             $canonized_headline );
3123
3124                         # strip out HTML
3125                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3126                         $tocline = trim( $canonized_headline );
3127                         # Save headline for section edit hint before it's escaped
3128                         $headline_hint = trim( $canonized_headline );
3129                         $canonized_headline = Sanitizer::escapeId( $tocline );
3130                         $refers[$headlineCount] = $canonized_headline;
3131
3132                         # count how many in assoc. array so we can track dupes in anchors
3133                         @$refers[$canonized_headline]++;
3134                         $refcount[$headlineCount]=$refers[$canonized_headline];
3135
3136                         # Don't number the heading if it is the only one (looks silly)
3137                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3138                                 # the two are different if the line contains a link
3139                                 $headline=$numbering . ' ' . $headline;
3140                         }
3141
3142                         # Create the anchor for linking from the TOC to the section
3143                         $anchor = $canonized_headline;
3144                         if($refcount[$headlineCount] > 1 ) {
3145                                 $anchor .= '_' . $refcount[$headlineCount];
3146                         }
3147                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3148                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3149                         }
3150                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3151                                 if ( empty( $head[$headlineCount] ) ) {
3152                                         $head[$headlineCount] = '';
3153                                 }
3154                                 if( $istemplate )
3155                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3156                                 else
3157                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3158                         }
3159
3160                         # give headline the correct <h#> tag
3161                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3162
3163                         $headlineCount++;
3164                         if( !$istemplate )
3165                                 $sectionCount++;
3166                 }
3167
3168                 if( $doShowToc ) {
3169                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3170                         $toc = $sk->tocList( $toc );
3171                 }
3172
3173                 # split up and insert constructed headlines
3174
3175                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3176                 $i = 0;
3177
3178                 foreach( $blocks as $block ) {
3179                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3180                                 # This is the [edit] link that appears for the top block of text when
3181                                 # section editing is enabled
3182
3183                                 # Disabled because it broke block formatting
3184                                 # For example, a bullet point in the top line
3185                                 # $full .= $sk->editSectionLink(0);
3186                         }
3187                         $full .= $block;
3188                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3189                         # Top anchor now in skin
3190                                 $full = $full.$toc;
3191                         }
3192
3193                         if( !empty( $head[$i] ) ) {
3194                                 $full .= $head[$i];
3195                         }
3196                         $i++;
3197                 }
3198                 if($forceTocHere) {
3199                         $mw =& MagicWord::get( MAG_TOC );
3200                         return $mw->replace( $toc, $full );
3201                 } else {
3202                         return $full;
3203                 }
3204         }
3205
3206         /**
3207          * Return an HTML link for the "ISBN 123456" text
3208          * @access private
3209          */
3210         function magicISBN( $text ) {
3211                 $fname = 'Parser::magicISBN';
3212                 wfProfileIn( $fname );
3213
3214                 $a = split( 'ISBN ', ' '.$text );
3215                 if ( count ( $a ) < 2 ) {
3216                         wfProfileOut( $fname );
3217                         return $text;
3218                 }
3219                 $text = substr( array_shift( $a ), 1);
3220                 $valid = '0123456789-Xx';
3221
3222                 foreach ( $a as $x ) {
3223                         # hack: don't replace inside thumbnail title/alt
3224                         # attributes
3225                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3226                                 $text .= "ISBN $x";
3227                                 continue;
3228                         }
3229
3230                         $isbn = $blank = '' ;
3231                         while ( ' ' == $x{0} ) {
3232                                 $blank .= ' ';
3233                                 $x = substr( $x, 1 );
3234                         }
3235                         if ( $x == '' ) { # blank isbn
3236                                 $text .= "ISBN $blank";
3237                                 continue;
3238                         }
3239                         while ( strstr( $valid, $x{0} ) != false ) {
3240                                 $isbn .= $x{0};
3241                                 $x = substr( $x, 1 );
3242                         }
3243                         $num = str_replace( '-', '', $isbn );
3244                         $num = str_replace( ' ', '', $num );
3245                         $num = str_replace( 'x', 'X', $num );
3246
3247                         if ( '' == $num ) {
3248                                 $text .= "ISBN $blank$x";
3249                         } else {
3250                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3251                                 $text .= '<a href="' .
3252                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3253                                         "\" class=\"internal\">ISBN $isbn</a>";
3254                                 $text .= $x;
3255                         }
3256                 }
3257                 wfProfileOut( $fname );
3258                 return $text;
3259         }
3260
3261         /**
3262          * Return an HTML link for the "RFC 1234" text
3263          *
3264          * @access private
3265          * @param string $text     Text to be processed
3266          * @param string $keyword  Magic keyword to use (default RFC)
3267          * @param string $urlmsg   Interface message to use (default rfcurl)
3268          * @return string
3269          */
3270         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3271
3272                 $valid = '0123456789';
3273                 $internal = false;
3274
3275                 $a = split( $keyword, ' '.$text );
3276                 if ( count ( $a ) < 2 ) {
3277                         return $text;
3278                 }
3279                 $text = substr( array_shift( $a ), 1);
3280
3281                 /* Check if keyword is preceed by [[.
3282                  * This test is made here cause of the array_shift above
3283                  * that prevent the test to be done in the foreach.
3284                  */
3285                 if ( substr( $text, -2 ) == '[[' ) {
3286                         $internal = true;
3287                 }
3288
3289                 foreach ( $a as $x ) {
3290                         /* token might be empty if we have RFC RFC 1234 */
3291                         if ( $x=='' ) {
3292                                 $text.=$keyword;
3293                                 continue;
3294                                 }
3295
3296                         # hack: don't replace inside thumbnail title/alt
3297                         # attributes
3298                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3299                                 $text .= $keyword . $x;
3300                                 continue;
3301                         }
3302
3303                         $id = $blank = '' ;
3304
3305                         /** remove and save whitespaces in $blank */
3306                         while ( $x{0} == ' ' ) {
3307                                 $blank .= ' ';
3308                                 $x = substr( $x, 1 );
3309                         }
3310
3311                         /** remove and save the rfc number in $id */
3312                         while ( strstr( $valid, $x{0} ) != false ) {
3313                                 $id .= $x{0};
3314                                 $x = substr( $x, 1 );
3315                         }
3316
3317                         if ( $id == '' ) {
3318                                 /* call back stripped spaces*/
3319                                 $text .= $keyword.$blank.$x;
3320                         } elseif( $internal ) {
3321                                 /* normal link */
3322                                 $text .= $keyword.$id.$x;
3323                         } else {
3324                                 /* build the external link*/
3325                                 $url = wfMsg( $urlmsg, $id);
3326                                 $sk =& $this->mOptions->getSkin();
3327                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3328                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3329                         }
3330
3331                         /* Check if the next RFC keyword is preceed by [[ */
3332                         $internal = ( substr($x,-2) == '[[' );
3333                 }
3334                 return $text;
3335         }
3336
3337         /**
3338          * Transform wiki markup when saving a page by doing \r\n -> \n
3339          * conversion, substitting signatures, {{subst:}} templates, etc.
3340          *
3341          * @param string $text the text to transform
3342          * @param Title &$title the Title object for the current article
3343          * @param User &$user the User object describing the current user
3344          * @param ParserOptions $options parsing options
3345          * @param bool $clearState whether to clear the parser state first
3346          * @return string the altered wiki markup
3347          * @access public
3348          */
3349         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3350                 $this->mOptions = $options;
3351                 $this->mTitle =& $title;
3352                 $this->mOutputType = OT_WIKI;
3353
3354                 if ( $clearState ) {
3355                         $this->clearState();
3356                 }
3357
3358                 $stripState = false;
3359                 $pairs = array(
3360                         "\r\n" => "\n",
3361                 );
3362                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3363                 $text = $this->strip( $text, $stripState, true );
3364                 $text = $this->pstPass2( $text, $user );
3365                 $text = $this->unstrip( $text, $stripState );
3366                 $text = $this->unstripNoWiki( $text, $stripState );
3367                 return $text;
3368         }
3369
3370         /**
3371          * Pre-save transform helper function
3372          * @access private
3373          */
3374         function pstPass2( $text, &$user ) {
3375                 global $wgContLang, $wgLocaltimezone;
3376
3377                 /* Note: This is the timestamp saved as hardcoded wikitext to
3378                  * the database, we use $wgContLang here in order to give
3379                  * everyone the same signiture and use the default one rather
3380                  * than the one selected in each users preferences.
3381                  */
3382                 if ( isset( $wgLocaltimezone ) ) {
3383                         $oldtz = getenv( 'TZ' );
3384                         putenv( 'TZ='.$wgLocaltimezone );
3385                 }
3386                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3387                   ' (' . date( 'T' ) . ')';
3388                 if ( isset( $wgLocaltimezone ) ) {
3389                         putenv( 'TZ='.$oldtz );
3390                 }
3391
3392                 # Variable replacement
3393                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3394                 $text = $this->replaceVariables( $text );
3395
3396                 # Signatures
3397                 $sigText = $this->getUserSig( $user );
3398                 $text = strtr( $text, array(
3399                         '~~~~~' => $d,
3400                         '~~~~' => "$sigText $d",
3401                         '~~~' => $sigText
3402                 ) );
3403
3404                 # Context links: [[|name]] and [[name (context)|]]
3405                 #
3406                 global $wgLegalTitleChars;
3407                 $tc = "[$wgLegalTitleChars]";
3408                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3409
3410                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3411                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3412
3413                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3414                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3415                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3416                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3417                 $context = '';
3418                 $t = $this->mTitle->getText();
3419                 if ( preg_match( $conpat, $t, $m ) ) {
3420                         $context = $m[2];
3421                 }
3422                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3423                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3424                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3425
3426                 if ( '' == $context ) {
3427                         $text = preg_replace( $p2, '[[\\1]]', $text );
3428                 } else {
3429                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3430                 }
3431
3432                 # Trim trailing whitespace
3433                 # MAG_END (__END__) tag allows for trailing
3434                 # whitespace to be deliberately included
3435                 $text = rtrim( $text );
3436                 $mw =& MagicWord::get( MAG_END );
3437                 $mw->matchAndRemove( $text );
3438
3439                 return $text;
3440         }
3441
3442         /**
3443          * Fetch the user's signature text, if any, and normalize to
3444          * validated, ready-to-insert wikitext.
3445          *
3446          * @param User $user
3447          * @return string
3448          * @access private
3449          */
3450         function getUserSig( &$user ) {
3451                 $username = $user->getName();
3452                 $nickname = $user->getOption( 'nickname' );
3453                 $nickname = $nickname === '' ? $username : $nickname;
3454
3455                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3456                         # Sig. might contain markup; validate this
3457                         if( $this->validateSig( $nickname ) !== false ) {
3458                                 # Validated; clean up (if needed) and return it
3459                                 return( $this->cleanSig( $nickname ) );
3460                         } else {
3461                                 # Failed to validate; fall back to the default
3462                                 $nickname = $username;
3463                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3464                         }
3465                 }
3466
3467                 # If we're still here, make it a link to the user page
3468                 $userpage = $user->getUserPage();
3469                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3470         }
3471
3472         /**
3473          * Check that the user's signature contains no bad XML
3474          *
3475          * @param string $text
3476          * @return mixed An expanded string, or false if invalid.
3477          */
3478         function validateSig( $text ) {
3479                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3480         }
3481
3482         /**
3483          * Clean up signature text
3484          *
3485          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3486          * 2) Substitute all transclusions
3487          *
3488          * @param string $text
3489          * @return string Signature text
3490          */
3491         function cleanSig( $text ) {
3492                 $substWord = MagicWord::get( MAG_SUBST );
3493                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3494                 $substText = '{{' . $substWord->getSynonym( 0 );
3495
3496                 $text = preg_replace( $substRegex, $substText, $text );
3497                 $text = preg_replace( '/~{3,5}/', '', $text );
3498                 $text = $this->replaceVariables( $text );
3499
3500                 return $text;
3501         }
3502
3503         /**
3504          * Set up some variables which are usually set up in parse()
3505          * so that an external function can call some class members with confidence
3506          * @access public
3507          */
3508         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3509                 $this->mTitle =& $title;
3510                 $this->mOptions = $options;
3511                 $this->mOutputType = $outputType;
3512                 if ( $clearState ) {
3513                         $this->clearState();
3514                 }
3515         }
3516
3517         /**
3518          * Transform a MediaWiki message by replacing magic variables.
3519          *
3520          * @param string $text the text to transform
3521          * @param ParserOptions $options  options
3522          * @return string the text with variables substituted
3523          * @access public
3524          */
3525         function transformMsg( $text, $options ) {
3526                 global $wgTitle;
3527                 static $executing = false;
3528
3529                 $fname = "Parser::transformMsg";
3530
3531                 # Guard against infinite recursion
3532                 if ( $executing ) {
3533                         return $text;
3534                 }
3535                 $executing = true;
3536
3537                 wfProfileIn($fname);
3538
3539                 $this->mTitle = $wgTitle;
3540                 $this->mOptions = $options;
3541                 $this->mOutputType = OT_MSG;
3542                 $this->clearState();
3543                 $text = $this->replaceVariables( $text );
3544
3545                 $executing = false;
3546                 wfProfileOut($fname);
3547                 return $text;
3548         }
3549
3550         /**
3551          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3552          * The callback should have the following form:
3553          *    function myParserHook( $text, $params, &$parser ) { ... }
3554          *
3555          * Transform and return $text. Use $parser for any required context, e.g. use
3556          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3557          *
3558          * @access public
3559          *
3560          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3561          * @param mixed $callback The callback function (and object) to use for the tag
3562          *
3563          * @return The old value of the mTagHooks array associated with the hook
3564          */
3565         function setHook( $tag, $callback ) {
3566                 $oldVal = @$this->mTagHooks[$tag];
3567                 $this->mTagHooks[$tag] = $callback;
3568
3569                 return $oldVal;
3570         }
3571
3572         /**
3573          * Create a function, e.g. {{sum:1|2|3}}
3574          * The callback function should have the form:
3575          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3576          *
3577          * The callback may either return the text result of the function, or an array with the text
3578          * in element 0, and a number of flags in the other elements. The names of the flags are
3579          * specified in the keys. Valid flags are:
3580          *   found                     The text returned is valid, stop processing the template. This
3581          *                             is on by default.
3582          *   nowiki                    Wiki markup in the return value should be escaped
3583          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3584          *   noargs                    Don't replace triple-brace arguments in the return value
3585          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3586          *
3587          * @access public
3588          *
3589          * @param string $name The function name. Function names are case-insensitive.
3590          * @param mixed $callback The callback function (and object) to use
3591          *
3592          * @return The old callback function for this name, if any
3593          */
3594         function setFunctionHook( $name, $callback ) {
3595                 $name = strtolower( $name );
3596                 $oldVal = @$this->mFunctionHooks[$name];
3597                 $this->mFunctionHooks[$name] = $callback;
3598                 return $oldVal;
3599         }
3600
3601         /**
3602          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3603          * Placeholders created in Skin::makeLinkObj()
3604          * Returns an array of links found, indexed by PDBK:
3605          *  0 - broken
3606          *  1 - normal link
3607          *  2 - stub
3608          * $options is a bit field, RLH_FOR_UPDATE to select for update
3609          */
3610         function replaceLinkHolders( &$text, $options = 0 ) {
3611                 global $wgUser;
3612                 global $wgOutputReplace;
3613
3614                 $fname = 'Parser::replaceLinkHolders';
3615                 wfProfileIn( $fname );
3616
3617                 $pdbks = array();
3618                 $colours = array();
3619                 $sk =& $this->mOptions->getSkin();
3620                 $linkCache =& LinkCache::singleton();
3621
3622                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3623                         wfProfileIn( $fname.'-check' );
3624                         $dbr =& wfGetDB( DB_SLAVE );
3625                         $page = $dbr->tableName( 'page' );
3626                         $threshold = $wgUser->getOption('stubthreshold');
3627
3628                         # Sort by namespace
3629                         asort( $this->mLinkHolders['namespaces'] );
3630
3631                         # Generate query
3632                         $query = false;
3633                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3634                                 # Make title object
3635                                 $title = $this->mLinkHolders['titles'][$key];
3636
3637                                 # Skip invalid entries.
3638                                 # Result will be ugly, but prevents crash.
3639                                 if ( is_null( $title ) ) {
3640                                         continue;
3641                                 }
3642                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3643
3644                                 # Check if it's a static known link, e.g. interwiki
3645                                 if ( $title->isAlwaysKnown() ) {
3646                                         $colours[$pdbk] = 1;
3647                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3648                                         $colours[$pdbk] = 1;
3649                                         $this->mOutput->addLink( $title, $id );
3650                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3651                                         $colours[$pdbk] = 0;
3652                                 } else {
3653                                         # Not in the link cache, add it to the query
3654                                         if ( !isset( $current ) ) {
3655                                                 $current = $ns;
3656                                                 $query =  "SELECT page_id, page_namespace, page_title";
3657                                                 if ( $threshold > 0 ) {
3658                                                         $query .= ', page_len, page_is_redirect';
3659                                                 }
3660                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3661                                         } elseif ( $current != $ns ) {
3662                                                 $current = $ns;
3663                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3664                                         } else {
3665                                                 $query .= ', ';
3666                                         }
3667
3668                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3669                                 }
3670                         }
3671                         if ( $query ) {
3672                                 $query .= '))';
3673                                 if ( $options & RLH_FOR_UPDATE ) {
3674                                         $query .= ' FOR UPDATE';
3675                                 }
3676
3677                                 $res = $dbr->query( $query, $fname );
3678
3679                                 # Fetch data and form into an associative array
3680                                 # non-existent = broken
3681                                 # 1 = known
3682                                 # 2 = stub
3683                                 while ( $s = $dbr->fetchObject($res) ) {
3684                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3685                                         $pdbk = $title->getPrefixedDBkey();
3686                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3687                                         $this->mOutput->addLink( $title, $s->page_id );
3688
3689                                         if ( $threshold >  0 ) {
3690                                                 $size = $s->page_len;
3691                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3692                                                         $colours[$pdbk] = 1;
3693                                                 } else {
3694                                                         $colours[$pdbk] = 2;
3695                                                 }
3696                                         } else {
3697                                                 $colours[$pdbk] = 1;
3698                                         }
3699                                 }
3700                         }
3701                         wfProfileOut( $fname.'-check' );
3702
3703                         # Construct search and replace arrays
3704                         wfProfileIn( $fname.'-construct' );
3705                         $wgOutputReplace = array();
3706                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3707                                 $pdbk = $pdbks[$key];
3708                                 $searchkey = "<!--LINK $key-->";
3709                                 $title = $this->mLinkHolders['titles'][$key];
3710                                 if ( empty( $colours[$pdbk] ) ) {
3711                                         $linkCache->addBadLinkObj( $title );
3712                                         $colours[$pdbk] = 0;
3713                                         $this->mOutput->addLink( $title, 0 );
3714                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3715                                                                         $this->mLinkHolders['texts'][$key],
3716                                                                         $this->mLinkHolders['queries'][$key] );
3717                                 } elseif ( $colours[$pdbk] == 1 ) {
3718                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3719                                                                         $this->mLinkHolders['texts'][$key],
3720                                                                         $this->mLinkHolders['queries'][$key] );
3721                                 } elseif ( $colours[$pdbk] == 2 ) {
3722                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3723                                                                         $this->mLinkHolders['texts'][$key],
3724                                                                         $this->mLinkHolders['queries'][$key] );
3725                                 }
3726                         }
3727                         wfProfileOut( $fname.'-construct' );
3728
3729                         # Do the thing
3730                         wfProfileIn( $fname.'-replace' );
3731
3732                         $text = preg_replace_callback(
3733                                 '/(<!--LINK .*?-->)/',
3734                                 "wfOutputReplaceMatches",
3735                                 $text);
3736
3737                         wfProfileOut( $fname.'-replace' );
3738                 }
3739
3740                 # Now process interwiki link holders
3741                 # This is quite a bit simpler than internal links
3742                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3743                         wfProfileIn( $fname.'-interwiki' );
3744                         # Make interwiki link HTML
3745                         $wgOutputReplace = array();
3746                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3747                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3748                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3749                         }
3750
3751                         $text = preg_replace_callback(
3752                                 '/<!--IWLINK (.*?)-->/',
3753                                 "wfOutputReplaceMatches",
3754                                 $text );
3755                         wfProfileOut( $fname.'-interwiki' );
3756                 }
3757
3758                 wfProfileOut( $fname );
3759                 return $colours;
3760         }
3761
3762         /**
3763          * Replace <!--LINK--> link placeholders with plain text of links
3764          * (not HTML-formatted).
3765          * @param string $text
3766          * @return string
3767          */
3768         function replaceLinkHoldersText( $text ) {
3769                 $fname = 'Parser::replaceLinkHoldersText';
3770                 wfProfileIn( $fname );
3771
3772                 $text = preg_replace_callback(
3773                         '/<!--(LINK|IWLINK) (.*?)-->/',
3774                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3775                         $text );
3776
3777                 wfProfileOut( $fname );
3778                 return $text;
3779         }
3780
3781         /**
3782          * @param array $matches
3783          * @return string
3784          * @access private
3785          */
3786         function replaceLinkHoldersTextCallback( $matches ) {
3787                 $type = $matches[1];
3788                 $key  = $matches[2];
3789                 if( $type == 'LINK' ) {
3790                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3791                                 return $this->mLinkHolders['texts'][$key];
3792                         }
3793                 } elseif( $type == 'IWLINK' ) {
3794                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3795                                 return $this->mInterwikiLinkHolders['texts'][$key];
3796                         }
3797                 }
3798                 return $matches[0];
3799         }
3800
3801         /**
3802          * Renders an image gallery from a text with one line per image.
3803          * text labels may be given by using |-style alternative text. E.g.
3804          *   Image:one.jpg|The number "1"
3805          *   Image:tree.jpg|A tree
3806          * given as text will return the HTML of a gallery with two images,
3807          * labeled 'The number "1"' and
3808          * 'A tree'.
3809          */
3810         function renderImageGallery( $text ) {
3811                 # Setup the parser
3812                 $parserOptions = new ParserOptions;
3813                 $localParser = new Parser();
3814
3815                 $ig = new ImageGallery();
3816                 $ig->setShowBytes( false );
3817                 $ig->setShowFilename( false );
3818                 $lines = explode( "\n", $text );
3819
3820                 foreach ( $lines as $line ) {
3821                         # match lines like these:
3822                         # Image:someimage.jpg|This is some image
3823                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3824                         # Skip empty lines
3825                         if ( count( $matches ) == 0 ) {
3826                                 continue;
3827                         }
3828                         $nt =& Title::newFromText( $matches[1] );
3829                         if( is_null( $nt ) ) {
3830                                 # Bogus title. Ignore these so we don't bomb out later.
3831                                 continue;
3832                         }
3833                         if ( isset( $matches[3] ) ) {
3834                                 $label = $matches[3];
3835                         } else {
3836                                 $label = '';
3837                         }
3838
3839                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3840                         $html = $pout->getText();
3841
3842                         $ig->add( new Image( $nt ), $html );
3843                         $this->mOutput->addImage( $nt->getDBkey() );
3844                 }
3845                 return $ig->toHTML();
3846         }
3847
3848         /**
3849          * Parse image options text and use it to make an image
3850          */
3851         function makeImage( &$nt, $options ) {
3852                 global $wgUseImageResize;
3853
3854                 $align = '';
3855
3856                 # Check if the options text is of the form "options|alt text"
3857                 # Options are:
3858                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3859                 #  * left               no resizing, just left align. label is used for alt= only
3860                 #  * right              same, but right aligned
3861                 #  * none               same, but not aligned
3862                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3863                 #  * center             center the image
3864                 #  * framed             Keep original image size, no magnify-button.
3865
3866                 $part = explode( '|', $options);
3867
3868                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3869                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3870                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3871                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3872                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3873                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3874                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3875                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3876                 $caption = '';
3877
3878                 $width = $height = $framed = $thumb = false;
3879                 $manual_thumb = '' ;
3880
3881                 foreach( $part as $key => $val ) {
3882                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3883                                 $thumb=true;
3884                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3885                                 # use manually specified thumbnail
3886                                 $thumb=true;
3887                                 $manual_thumb = $match;
3888                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3889                                 # remember to set an alignment, don't render immediately
3890                                 $align = 'right';
3891                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3892                                 # remember to set an alignment, don't render immediately
3893                                 $align = 'left';
3894                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3895                                 # remember to set an alignment, don't render immediately
3896                                 $align = 'center';
3897                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3898                                 # remember to set an alignment, don't render immediately
3899                                 $align = 'none';
3900                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3901                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3902                                 # $match is the image width in pixels
3903                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3904                                         $width = intval( $m[1] );
3905                                         $height = intval( $m[2] );
3906                                 } else {
3907                                         $width = intval($match);
3908                                 }
3909                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3910                                 $framed=true;
3911                         } else {
3912                                 $caption = $val;
3913                         }
3914                 }
3915                 # Strip bad stuff out of the alt text
3916                 $alt = $this->replaceLinkHoldersText( $caption );
3917
3918                 # make sure there are no placeholders in thumbnail attributes
3919                 # that are later expanded to html- so expand them now and
3920                 # remove the tags
3921                 $alt = $this->unstrip($alt, $this->mStripState);
3922                 $alt = Sanitizer::stripAllTags( $alt );
3923
3924                 # Linker does the rest
3925                 $sk =& $this->mOptions->getSkin();
3926                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3927         }
3928
3929         /**
3930          * Set a flag in the output object indicating that the content is dynamic and
3931          * shouldn't be cached.
3932          */
3933         function disableCache() {
3934                 $this->mOutput->mCacheTime = -1;
3935         }
3936
3937         /**#@+
3938          * Callback from the Sanitizer for expanding items found in HTML attribute
3939          * values, so they can be safely tested and escaped.
3940          * @param string $text
3941          * @param array $args
3942          * @return string
3943          * @access private
3944          */
3945         function attributeStripCallback( &$text, $args ) {
3946                 $text = $this->replaceVariables( $text, $args );
3947                 $text = $this->unstripForHTML( $text );
3948                 return $text;
3949         }
3950
3951         function unstripForHTML( $text ) {
3952                 $text = $this->unstrip( $text, $this->mStripState );
3953                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3954                 return $text;
3955         }
3956         /**#@-*/
3957
3958         /**#@+
3959          * Accessor/mutator
3960          */
3961         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3962         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3963         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3964         /**#@-*/
3965
3966         /**#@+
3967          * Accessor
3968          */
3969         function getTags() { return array_keys( $this->mTagHooks ); }
3970         /**#@-*/
3971 }
3972
3973 /**
3974  * @todo document
3975  * @package MediaWiki
3976  */
3977 class ParserOutput
3978 {
3979         var $mText,             # The output text
3980                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3981                 $mCategories,       # Map of category names to sort keys
3982                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3983                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
3984                 $mVersion,          # Compatibility check
3985                 $mTitleText,        # title text of the chosen language variant
3986                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3987                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3988                 $mImages,           # DB keys of the images used, in the array key only
3989                 $mExternalLinks;    # External link URLs, in the key only
3990
3991         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3992                 $containsOldMagic = false, $titletext = '' )
3993         {
3994                 $this->mText = $text;
3995                 $this->mLanguageLinks = $languageLinks;
3996                 $this->mCategories = $categoryLinks;
3997                 $this->mContainsOldMagic = $containsOldMagic;
3998                 $this->mCacheTime = '';
3999                 $this->mVersion = MW_PARSER_VERSION;
4000                 $this->mTitleText = $titletext;
4001                 $this->mLinks = array();
4002                 $this->mTemplates = array();
4003                 $this->mImages = array();
4004                 $this->mExternalLinks = array();
4005         }
4006
4007         function getText()                   { return $this->mText; }
4008         function getLanguageLinks()          { return $this->mLanguageLinks; }
4009         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4010         function &getCategories()            { return $this->mCategories; }
4011         function getCacheTime()              { return $this->mCacheTime; }
4012         function getTitleText()              { return $this->mTitleText; }
4013         function &getLinks()                 { return $this->mLinks; }
4014         function &getTemplates()             { return $this->mTemplates; }
4015         function &getImages()                { return $this->mImages; }
4016         function &getExternalLinks()         { return $this->mExternalLinks; }
4017
4018         function containsOldMagic()          { return $this->mContainsOldMagic; }
4019         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4020         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4021         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4022         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4023         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4024         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4025
4026         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4027         function addImage( $name )           { $this->mImages[$name] = 1; }
4028         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4029         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4030
4031         function addLink( $title, $id ) {
4032                 $ns = $title->getNamespace();
4033                 $dbk = $title->getDBkey();
4034                 if ( !isset( $this->mLinks[$ns] ) ) {
4035                         $this->mLinks[$ns] = array();
4036                 }
4037                 $this->mLinks[$ns][$dbk] = $id;
4038         }
4039
4040         function addTemplate( $title, $id ) {
4041                 $ns = $title->getNamespace();
4042                 $dbk = $title->getDBkey();
4043                 if ( !isset( $this->mTemplates[$ns] ) ) {
4044                         $this->mTemplates[$ns] = array();
4045                 }
4046                 $this->mTemplates[$ns][$dbk] = $id;
4047         }
4048
4049         /**
4050          * @deprecated
4051          */
4052         /*
4053         function merge( $other ) {
4054                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
4055                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
4056                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
4057         }*/
4058
4059         /**
4060          * Return true if this cached output object predates the global or
4061          * per-article cache invalidation timestamps, or if it comes from
4062          * an incompatible older version.
4063          *
4064          * @param string $touched the affected article's last touched timestamp
4065          * @return bool
4066          * @access public
4067          */
4068         function expired( $touched ) {
4069                 global $wgCacheEpoch;
4070                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4071                        $this->getCacheTime() < $touched ||
4072                        $this->getCacheTime() <= $wgCacheEpoch ||
4073                        !isset( $this->mVersion ) ||
4074                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4075         }
4076 }
4077
4078 /**
4079  * Set options of the Parser
4080  * @todo document
4081  * @package MediaWiki
4082  */
4083 class ParserOptions
4084 {
4085         # All variables are private
4086         var $mUseTeX;                    # Use texvc to expand <math> tags
4087         var $mUseDynamicDates;           # Use DateFormatter to format dates
4088         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4089         var $mAllowExternalImages;       # Allow external images inline
4090         var $mAllowExternalImagesFrom;   # If not, any exception?
4091         var $mSkin;                      # Reference to the preferred skin
4092         var $mDateFormat;                # Date format index
4093         var $mEditSection;               # Create "edit section" links
4094         var $mNumberHeadings;            # Automatically number headings
4095         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4096         var $mTidy;                      # Ask for tidy cleanup
4097
4098         function getUseTeX()                        { return $this->mUseTeX; }
4099         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4100         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4101         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4102         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4103         function &getSkin()                         { return $this->mSkin; }
4104         function getDateFormat()                    { return $this->mDateFormat; }
4105         function getEditSection()                   { return $this->mEditSection; }
4106         function getNumberHeadings()                { return $this->mNumberHeadings; }
4107         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4108         function getTidy()                          { return $this->mTidy; }
4109
4110         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4111         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4112         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4113         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4114         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4115         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4116         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4117         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4118         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4119         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4120         function setSkin( &$x ) { $this->mSkin =& $x; }
4121
4122         function ParserOptions() {
4123                 global $wgUser;
4124                 $this->initialiseFromUser( $wgUser );
4125         }
4126
4127         /**
4128          * Get parser options
4129          * @static
4130          */
4131         function newFromUser( &$user ) {
4132                 $popts = new ParserOptions;
4133                 $popts->initialiseFromUser( $user );
4134                 return $popts;
4135         }
4136
4137         /** Get user options */
4138         function initialiseFromUser( &$userInput ) {
4139                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4140                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4141                 $fname = 'ParserOptions::initialiseFromUser';
4142                 wfProfileIn( $fname );
4143                 if ( !$userInput ) {
4144                         $user = new User;
4145                         $user->setLoaded( true );
4146                 } else {
4147                         $user =& $userInput;
4148                 }
4149
4150                 $this->mUseTeX = $wgUseTeX;
4151                 $this->mUseDynamicDates = $wgUseDynamicDates;
4152                 $this->mInterwikiMagic = $wgInterwikiMagic;
4153                 $this->mAllowExternalImages = $wgAllowExternalImages;
4154                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4155                 wfProfileIn( $fname.'-skin' );
4156                 $this->mSkin =& $user->getSkin();
4157                 wfProfileOut( $fname.'-skin' );
4158                 $this->mDateFormat = $user->getOption( 'date' );
4159                 $this->mEditSection = true;
4160                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4161                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4162                 $this->mTidy = false;
4163                 wfProfileOut( $fname );
4164         }
4165 }
4166
4167 /**
4168  * Callback function used by Parser::replaceLinkHolders()
4169  * to substitute link placeholders.
4170  */
4171 function &wfOutputReplaceMatches( $matches ) {
4172         global $wgOutputReplace;
4173         return $wgOutputReplace[$matches[1]];
4174 }
4175
4176 /**
4177  * Return the total number of articles
4178  */
4179 function wfNumberOfArticles() {
4180         global $wgNumberOfArticles;
4181
4182         wfLoadSiteStats();
4183         return $wgNumberOfArticles;
4184 }
4185
4186 /**
4187  * Return the number of files
4188  */
4189 function wfNumberOfFiles() {
4190         $fname = 'wfNumberOfFiles';
4191
4192         wfProfileIn( $fname );
4193         $dbr =& wfGetDB( DB_SLAVE );
4194         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4195         wfProfileOut( $fname );
4196
4197         return $numImages;
4198 }
4199
4200 /**
4201  * Get various statistics from the database
4202  * @access private
4203  */
4204 function wfLoadSiteStats() {
4205         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4206         $fname = 'wfLoadSiteStats';
4207
4208         if ( -1 != $wgNumberOfArticles ) return;
4209         $dbr =& wfGetDB( DB_SLAVE );
4210         $s = $dbr->selectRow( 'site_stats',
4211                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4212                 array( 'ss_row_id' => 1 ), $fname
4213         );
4214
4215         if ( $s === false ) {
4216                 return;
4217         } else {
4218                 $wgTotalViews = $s->ss_total_views;
4219                 $wgTotalEdits = $s->ss_total_edits;
4220                 $wgNumberOfArticles = $s->ss_good_articles;
4221         }
4222 }
4223
4224 /**
4225  * Escape html tags
4226  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4227  *
4228  * @param string $in Text that might contain HTML tags
4229  * @return string Escaped string
4230  */
4231 function wfEscapeHTMLTagsOnly( $in ) {
4232         return str_replace(
4233                 array( '"', '>', '<' ),
4234                 array( '&quot;', '&gt;', '&lt;' ),
4235                 $in );
4236 }
4237
4238 ?>