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 require_once( 'ImageGallery.php' );
  13
  14 /**
  15  * Update this version number when the ParserOutput format
  16  * changes in an incompatible way, so the parser cache
  17  * can automatically discard old data.
  18  */
  19 define( 'MW_PARSER_VERSION', '1.6.1' );
  20
  21 /**
  22  * Variable substitution O(N^2) attack
  23  *
  24  * Without countermeasures, it would be possible to attack the parser by saving
  25  * a page filled with a large number of inclusions of large pages. The size of
  26  * the generated page would be proportional to the square of the input size.
  27  * Hence, we limit the number of inclusions of any given page, thus bringing any
  28  * attack back to O(N).
  29  */
  30
  31 define( 'MAX_INCLUDE_REPEAT', 100 );
  32 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  33
  34 define( 'RLH_FOR_UPDATE', 1 );
  35
  36 # Allowed values for $mOutputType
  37 define( 'OT_HTML', 1 );
  38 define( 'OT_WIKI', 2 );
  39 define( 'OT_MSG' , 3 );
  40
  41 # string parameter for extractTags which will cause it
  42 # to strip HTML comments in addition to regular
  43 # <XML>-style tags. This should not be anything we
  44 # may want to use in wikisyntax
  45 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  46
  47 # Constants needed for external link processing
  48 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  49 # Everything except bracket, space, or control characters
  50 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  51 # Including space
  52 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  53 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  54 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  55 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  56 define( 'EXT_IMAGE_REGEX',
  57         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  58         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  59         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  60 );
  61
  62 /**
  63  * PHP Parser
  64  *
  65  * Processes wiki markup
  66  *
  67  * <pre>
  68  * There are three main entry points into the Parser class:
  69  * parse()
  70  *   produces HTML output
  71  * preSaveTransform().
  72  *   produces altered wiki markup.
  73  * transformMsg()
  74  *   performs brace substitution on MediaWiki messages
  75  *
  76  * Globals used:
  77  *    objects:   $wgLang, $wgContLang
  78  *
  79  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  80  *
  81  * settings:
  82  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  83  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  84  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  85  *
  86  *  * only within ParserOptions
  87  * </pre>
  88  *
  89  * @package MediaWiki
  90  */
  91 class Parser
  92 {
  93         /**#@+
  94          * @private
  95          */
  96         # Persistent:
  97         var $mTagHooks, $mFunctionHooks;
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 101         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 103         var $mTemplates,        // cache of already loaded templates, avoids
 104                                 // multiple SQL queries for the same string
 105             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 106                                 // in this path. Used for loop detection.
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 114
 115         /**#@-*/
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function Parser() {
 123                 $this->mTagHooks = array();
 124                 $this->mFunctionHooks = array();
 125                 $this->clearState();
 126         }
 127
 128         /**
 129          * Clear Parser state
 130          *
 131          * @private
 132          */
 133         function clearState() {
 134                 $this->mOutput = new ParserOutput;
 135                 $this->mAutonumber = 0;
 136                 $this->mLastSection = '';
 137                 $this->mDTopen = false;
 138                 $this->mVariables = false;
 139                 $this->mIncludeCount = array();
 140                 $this->mStripState = array();
 141                 $this->mArgStack = array();
 142                 $this->mInPre = false;
 143                 $this->mInterwikiLinkHolders = array(
 144                         'texts' => array(),
 145                         'titles' => array()
 146                 );
 147                 $this->mLinkHolders = array(
 148                         'namespaces' => array(),
 149                         'dbkeys' => array(),
 150                         'queries' => array(),
 151                         'texts' => array(),
 152                         'titles' => array()
 153                 );
 154                 $this->mRevisionId = null;
 155
 156                 /**
 157                  * Prefix for temporary replacement strings for the multipass parser.
 158                  * \x07 should never appear in input as it's disallowed in XML.
 159                  * Using it at the front also gives us a little extra robustness
 160                  * since it shouldn't match when butted up against identifier-like
 161                  * string constructs.
 162                  */
 163                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 164
 165                 # Clear these on every parse, bug 4549
 166                 $this->mTemplates = array();
 167                 $this->mTemplatePath = array();
 168
 169                 $this->mShowToc = true;
 170                 $this->mForceTocPosition = false;
 171
 172                 wfRunHooks( 'ParserClearState', array( &$this ) );
 173         }
 174
 175         /**
 176          * Accessor for mUniqPrefix.
 177          *
 178          * @public
 179          */
 180         function UniqPrefix() {
 181                 return $this->mUniqPrefix;
 182         }
 183
 184         /**
 185          * Convert wikitext to HTML
 186          * Do not call this function recursively.
 187          *
 188          * @private
 189          * @param string $text Text we want to parse
 190          * @param Title &$title A title object
 191          * @param array $options
 192          * @param boolean $linestart
 193          * @param boolean $clearState
 194          * @param int $revid number to pass in {{REVISIONID}}
 195          * @return ParserOutput a ParserOutput
 196          */
 197         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 198                 /**
 199                  * First pass--just handle <nowiki> sections, pass the rest off
 200                  * to internalParse() which does all the real work.
 201                  */
 202
 203                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 204                 $fname = 'Parser::parse';
 205                 wfProfileIn( $fname );
 206
 207                 if ( $clearState ) {
 208                         $this->clearState();
 209                 }
 210
 211                 $this->mOptions = $options;
 212                 $this->mTitle =& $title;
 213                 $this->mRevisionId = $revid;
 214                 $this->mOutputType = OT_HTML;
 215
 216                 //$text = $this->strip( $text, $this->mStripState );
 217                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 218                 $x =& $this->mStripState;
 219
 220                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 221                 $text = $this->strip( $text, $x );
 222                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 223
 224                 # Hook to suspend the parser in this state
 225                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 226                         wfProfileOut( $fname );
 227                         return $text ;
 228                 }
 229
 230                 $text = $this->internalParse( $text );
 231
 232                 $text = $this->unstrip( $text, $this->mStripState );
 233
 234                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 235                 $fixtags = array(
 236                         # french spaces, last one Guillemet-left
 237                         # only if there is something before the space
 238                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 239                         # french spaces, Guillemet-right
 240                         '/(\\302\\253) /' => '\\1&nbsp;',
 241                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 242                 );
 243                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 244
 245                 # only once and last
 246                 $text = $this->doBlockLevels( $text, $linestart );
 247
 248                 $this->replaceLinkHolders( $text );
 249
 250                 # the position of the parserConvert() call should not be changed. it
 251                 # assumes that the links are all replaced and the only thing left
 252                 # is the <nowiki> mark.
 253                 # Side-effects: this calls $this->mOutput->setTitleText()
 254                 $text = $wgContLang->parserConvert( $text, $this );
 255
 256                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 257
 258                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 259
 260                 $text = Sanitizer::normalizeCharReferences( $text );
 261
 262                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 263                         $text = Parser::tidy($text);
 264                 } else {
 265                         # attempt to sanitize at least some nesting problems
 266                         # (bug #2702 and quite a few others)
 267                         $tidyregs = array(
 268                                 # ''Something [http://www.cool.com cool''] -->
 269                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 270                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 271                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 272                                 # fix up an anchor inside another anchor, only
 273                                 # at least for a single single nested link (bug 3695)
 274                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 275                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 276                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 277                                 # contains a div, so fix it up here; replace
 278                                 # div with escaped text
 279                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 280                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 281                                 # remove empty italic or bold tag pairs, some
 282                                 # introduced by rules above
 283                                 '/<([bi])><\/\\1>/' => ''
 284                         );
 285
 286                         $text = preg_replace(
 287                                 array_keys( $tidyregs ),
 288                                 array_values( $tidyregs ),
 289                                 $text );
 290                 }
 291
 292                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 293
 294                 $this->mOutput->setText( $text );
 295                 wfProfileOut( $fname );
 296
 297                 return $this->mOutput;
 298         }
 299
 300         /**
 301          * Get a random string
 302          *
 303          * @private
 304          * @static
 305          */
 306         function getRandomString() {
 307                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 308         }
 309
 310         function &getTitle() { return $this->mTitle; }
 311         function getOptions() { return $this->mOptions; }
 312
 313         /**
 314          * Replaces all occurrences of <$tag>content</$tag> in the text
 315          * with a random marker and returns the new text. the output parameter
 316          * $content will be an associative array filled with data on the form
 317          * $unique_marker => content.
 318          *
 319          * If $content is already set, the additional entries will be appended
 320          * If $tag is set to STRIP_COMMENTS, the function will extract
 321          * <!-- HTML comments -->
 322          *
 323          * $output: array( 'UNIQ-xxxxx' => array(
 324          *    'element',
 325          *    'tag content',
 326          *    array( 'param' => 'x' ),
 327          *    '<element param="x">' ) )
 328          * @private
 329          * @static
 330          */
 331         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 332                 $rand = Parser::getRandomString();
 333                 $n = 1;
 334                 $stripped = '';
 335                 $matches = array();
 336
 337                 if( $elements == STRIP_COMMENTS ) {
 338                         $start = '/<!--()()/';
 339                 } else {
 340                         $taglist = implode( '|', $elements );
 341                         $start = "/<($taglist)(\\s+[^>]*|\\s*\/?)>/i";
 342                 }
 343
 344                 while ( '' != $text ) {
 345                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 346                         $stripped .= $p[0];
 347                         if( count( $p ) < 4 ) {
 348                                 break;
 349                         }
 350                         $element    = $p[1];
 351                         $attributes = $p[2];
 352                         $inside     = $p[3];
 353
 354                         // If $attributes ends with '/', we have an empty element tag, <tag />
 355                         if( $element != '' && substr( $attributes, -1 ) == '/' ) {
 356                                 $attributes = substr( $attributes, 0, -1);
 357                                 $empty = '/';
 358                         } else {
 359                                 $empty = '';
 360                         }
 361
 362                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++);
 363                         $stripped .= $marker;
 364
 365                         if ( $empty === '/' ) {
 366                                 // Empty element tag, <tag />
 367                                 $content = null;
 368                                 $text = $inside;
 369                         } else {
 370                                 if( $element ) {
 371                                         $end = "/<\\/$element\\s*>/i";
 372                                 } else {
 373                                         $end = '/-->/';
 374                                 }
 375                                 $q = preg_split( $end, $inside, 2 );
 376                                 $content = $q[0];
 377                                 if( count( $q ) < 2 ) {
 378                                         # No end tag -- let it run out to the end of the text.
 379                                         $text = '';
 380                                 } else {
 381                                         $text = $q[1];
 382                                 }
 383                         }
 384
 385                         $matches[$marker] = array( $element,
 386                                 $content,
 387                                 Sanitizer::decodeTagAttributes( $attributes ),
 388                                 "<$element$attributes$empty>" );
 389                 }
 390                 return $stripped;
 391         }
 392
 393         /**
 394          * Strips and renders nowiki, pre, math, hiero
 395          * If $render is set, performs necessary rendering operations on plugins
 396          * Returns the text, and fills an array with data needed in unstrip()
 397          * If the $state is already a valid strip state, it adds to the state
 398          *
 399          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 400          *  will be stripped in addition to other tags. This is important
 401          *  for section editing, where these comments cause confusion when
 402          *  counting the sections in the wikisource
 403          *
 404          * @private
 405          */
 406         function strip( $text, &$state, $stripcomments = false ) {
 407                 $render = ($this->mOutputType == OT_HTML);
 408
 409                 # Replace any instances of the placeholders
 410                 $uniq_prefix = $this->mUniqPrefix;
 411                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 412
 413                 $elements = array_merge(
 414                         array( 'nowiki', 'pre', 'gallery' ),
 415                         array_keys( $this->mTagHooks ) );
 416                 global $wgRawHtml;
 417                 if( $wgRawHtml ) {
 418                         $elements[] = 'html';
 419                 }
 420                 if( $this->mOptions->getUseTeX() ) {
 421                         $elements[] = 'math';
 422                 }
 423
 424
 425                 // Strip comments in a first pass.
 426                 // This saves us from needlessly rendering extensions in comment text
 427                 $text = Parser::extractTagsAndParams(STRIP_COMMENTS, $text, $comment_matches, $uniq_prefix);
 428                 $commentState = array();
 429                 foreach( $comment_matches as $marker => $data ){
 430                         list( $element, $content, $params, $tag ) = $data;
 431                         $commentState[$marker] = '<!--' . $content . '-->';
 432                 }
 433
 434                 $matches = array();
 435                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 436
 437                 foreach( $matches as $marker => $data ) {
 438                         list( $element, $content, $params, $tag ) = $data;
 439                         // Restore any comments; the extension can deal with them.
 440                         if( $content !== null) {
 441                                 $content = strtr( $content, $commentState );
 442                         }
 443                         if( $render ) {
 444                                 switch( $element ) {
 445                                 case 'html':
 446                                         if( $wgRawHtml ) {
 447                                                 $output = $content;
 448                                                 break;
 449                                         }
 450                                         // Shouldn't happen otherwise. :)
 451                                 case 'nowiki':
 452                                         $output = wfEscapeHTMLTagsOnly( $content );
 453                                         break;
 454                                 case 'math':
 455                                         $output = renderMath( $content );
 456                                         break;
 457                                 case 'pre':
 458                                         // Backwards-compatibility hack
 459                                         $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $content );
 460                                         $output = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 461                                         break;
 462                                 case 'gallery':
 463                                         $output = $this->renderImageGallery( $content );
 464                                         break;
 465                                 default:
 466                                         $tagName = strtolower( $element );
 467                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 468                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 469                                                         array( $content, $params, $this ) );
 470                                         } else {
 471                                                 wfDebugDieBacktrace( "Invalid call hook $element" );
 472                                         }
 473                                 }
 474                         } else {
 475                                 // Just stripping tags; keep the source
 476                                 if( $content === null ) {
 477                                         $output = $tag;
 478                                 } else {
 479                                         $output = "$tag$content</$element>";
 480                                 }
 481                         }
 482                         $state[$element][$marker] = $output;
 483                 }
 484
 485                 # Unstrip comments unless explicitly told otherwise.
 486                 # (The comments are always stripped prior to this point, so as to
 487                 # not invoke any extension tags / parser hooks contained within
 488                 # a comment.)
 489                 if ( $stripcomments ) {
 490                         // Add remaining comments to the state array
 491                         foreach( $commentState as $marker => $content ) {
 492                                 $state['comment'][$marker] = $content;
 493                         }
 494                 } else {
 495                         // Put them all back and forget them
 496                         $text = strtr( $text, $commentState );
 497                 }
 498
 499                 return $text;
 500         }
 501
 502         /**
 503          * Restores pre, math, and other extensions removed by strip()
 504          *
 505          * always call unstripNoWiki() after this one
 506          * @private
 507          */
 508         function unstrip( $text, &$state ) {
 509                 if ( !is_array( $state ) ) {
 510                         return $text;
 511                 }
 512
 513                 $replacements = array();
 514                 foreach( $state as $tag => $contentDict ) {
 515                         if( $tag != 'nowiki' && $tag != 'html' ) {
 516                                 foreach( $contentDict as $uniq => $content ) {
 517                                         $replacements[$uniq] = $content;
 518                                 }
 519                         }
 520                 }
 521                 $text = strtr( $text, $replacements );
 522
 523                 return $text;
 524         }
 525
 526         /**
 527          * Always call this after unstrip() to preserve the order
 528          *
 529          * @private
 530          */
 531         function unstripNoWiki( $text, &$state ) {
 532                 if ( !is_array( $state ) ) {
 533                         return $text;
 534                 }
 535
 536                 $replacements = array();
 537                 foreach( $state as $tag => $contentDict ) {
 538                         if( $tag == 'nowiki' || $tag == 'html' ) {
 539                                 foreach( $contentDict as $uniq => $content ) {
 540                                         $replacements[$uniq] = $content;
 541                                 }
 542                         }
 543                 }
 544                 $text = strtr( $text, $replacements );
 545
 546                 return $text;
 547         }
 548
 549         /**
 550          * Add an item to the strip state
 551          * Returns the unique tag which must be inserted into the stripped text
 552          * The tag will be replaced with the original text in unstrip()
 553          *
 554          * @private
 555          */
 556         function insertStripItem( $text, &$state ) {
 557                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 558                 if ( !$state ) {
 559                         $state = array();
 560                 }
 561                 $state['item'][$rnd] = $text;
 562                 return $rnd;
 563         }
 564
 565         /**
 566          * Interface with html tidy, used if $wgUseTidy = true.
 567          * If tidy isn't able to correct the markup, the original will be
 568          * returned in all its glory with a warning comment appended.
 569          *
 570          * Either the external tidy program or the in-process tidy extension
 571          * will be used depending on availability. Override the default
 572          * $wgTidyInternal setting to disable the internal if it's not working.
 573          *
 574          * @param string $text Hideous HTML input
 575          * @return string Corrected HTML output
 576          * @public
 577          * @static
 578          */
 579         function tidy( $text ) {
 580                 global $wgTidyInternal;
 581                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 582 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 583 '<head><title>test</title></head><body>'.$text.'</body></html>';
 584                 if( $wgTidyInternal ) {
 585                         $correctedtext = Parser::internalTidy( $wrappedtext );
 586                 } else {
 587                         $correctedtext = Parser::externalTidy( $wrappedtext );
 588                 }
 589                 if( is_null( $correctedtext ) ) {
 590                         wfDebug( "Tidy error detected!\n" );
 591                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 592                 }
 593                 return $correctedtext;
 594         }
 595
 596         /**
 597          * Spawn an external HTML tidy process and get corrected markup back from it.
 598          *
 599          * @private
 600          * @static
 601          */
 602         function externalTidy( $text ) {
 603                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 604                 $fname = 'Parser::externalTidy';
 605                 wfProfileIn( $fname );
 606
 607                 $cleansource = '';
 608                 $opts = ' -utf8';
 609
 610                 $descriptorspec = array(
 611                         0 => array('pipe', 'r'),
 612                         1 => array('pipe', 'w'),
 613                         2 => array('file', '/dev/null', 'a')
 614                 );
 615                 $pipes = array();
 616                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 617                 if (is_resource($process)) {
 618                         // Theoretically, this style of communication could cause a deadlock
 619                         // here. If the stdout buffer fills up, then writes to stdin could
 620                         // block. This doesn't appear to happen with tidy, because tidy only
 621                         // writes to stdout after it's finished reading from stdin. Search
 622                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 623                         fwrite($pipes[0], $text);
 624                         fclose($pipes[0]);
 625                         while (!feof($pipes[1])) {
 626                                 $cleansource .= fgets($pipes[1], 1024);
 627                         }
 628                         fclose($pipes[1]);
 629                         proc_close($process);
 630                 }
 631
 632                 wfProfileOut( $fname );
 633
 634                 if( $cleansource == '' && $text != '') {
 635                         // Some kind of error happened, so we couldn't get the corrected text.
 636                         // Just give up; we'll use the source text and append a warning.
 637                         return null;
 638                 } else {
 639                         return $cleansource;
 640                 }
 641         }
 642
 643         /**
 644          * Use the HTML tidy PECL extension to use the tidy library in-process,
 645          * saving the overhead of spawning a new process. Currently written to
 646          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 647          *
 648          * 'pear install tidy' should be able to compile the extension module.
 649          *
 650          * @private
 651          * @static
 652          */
 653         function internalTidy( $text ) {
 654                 global $wgTidyConf;
 655                 $fname = 'Parser::internalTidy';
 656                 wfProfileIn( $fname );
 657
 658                 tidy_load_config( $wgTidyConf );
 659                 tidy_set_encoding( 'utf8' );
 660                 tidy_parse_string( $text );
 661                 tidy_clean_repair();
 662                 if( tidy_get_status() == 2 ) {
 663                         // 2 is magic number for fatal error
 664                         // http://www.php.net/manual/en/function.tidy-get-status.php
 665                         $cleansource = null;
 666                 } else {
 667                         $cleansource = tidy_get_output();
 668                 }
 669                 wfProfileOut( $fname );
 670                 return $cleansource;
 671         }
 672
 673         /**
 674          * parse the wiki syntax used to render tables
 675          *
 676          * @private
 677          */
 678         function doTableStuff ( $t ) {
 679                 $fname = 'Parser::doTableStuff';
 680                 wfProfileIn( $fname );
 681
 682                 $t = explode ( "\n" , $t ) ;
 683                 $td = array () ; # Is currently a td tag open?
 684                 $ltd = array () ; # Was it TD or TH?
 685                 $tr = array () ; # Is currently a tr tag open?
 686                 $ltr = array () ; # tr attributes
 687                 $has_opened_tr = array(); # Did this table open a <tr> element?
 688                 $indent_level = 0; # indent level of the table
 689                 foreach ( $t AS $k => $x )
 690                 {
 691                         $x = trim ( $x ) ;
 692                         $fc = substr ( $x , 0 , 1 ) ;
 693                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 694                                 $indent_level = strlen( $matches[1] );
 695
 696                                 $attributes = $this->unstripForHTML( $matches[2] );
 697
 698                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 699                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 700                                 array_push ( $td , false ) ;
 701                                 array_push ( $ltd , '' ) ;
 702                                 array_push ( $tr , false ) ;
 703                                 array_push ( $ltr , '' ) ;
 704                                 array_push ( $has_opened_tr, false );
 705                         }
 706                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 707                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 708                                 $z = "</table>" . substr ( $x , 2);
 709                                 $l = array_pop ( $ltd ) ;
 710                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 711                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 712                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 713                                 array_pop ( $ltr ) ;
 714                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 715                         }
 716                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 717                                 $x = substr ( $x , 1 ) ;
 718                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 719                                 $z = '' ;
 720                                 $l = array_pop ( $ltd ) ;
 721                                 array_pop ( $has_opened_tr );
 722                                 array_push ( $has_opened_tr , true ) ;
 723                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 724                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 725                                 array_pop ( $ltr ) ;
 726                                 $t[$k] = $z ;
 727                                 array_push ( $tr , false ) ;
 728                                 array_push ( $td , false ) ;
 729                                 array_push ( $ltd , '' ) ;
 730                                 $attributes = $this->unstripForHTML( $x );
 731                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 732                         }
 733                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 734                                 # $x is a table row
 735                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 736                                         $fc = '+' ;
 737                                         $x = substr ( $x , 1 ) ;
 738                                 }
 739                                 $after = substr ( $x , 1 ) ;
 740                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 741
 742                                 // Split up multiple cells on the same line.
 743                                 // FIXME: This can result in improper nesting of tags processed
 744                                 // by earlier parser steps, but should avoid splitting up eg
 745                                 // attribute values containing literal "||".
 746                                 $after = wfExplodeMarkup( '||', $after );
 747
 748                                 $t[$k] = '' ;
 749
 750                                 # Loop through each table cell
 751                                 foreach ( $after AS $theline )
 752                                 {
 753                                         $z = '' ;
 754                                         if ( $fc != '+' )
 755                                         {
 756                                                 $tra = array_pop ( $ltr ) ;
 757                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 758                                                 array_push ( $tr , true ) ;
 759                                                 array_push ( $ltr , '' ) ;
 760                                                 array_pop ( $has_opened_tr );
 761                                                 array_push ( $has_opened_tr , true ) ;
 762                                         }
 763
 764                                         $l = array_pop ( $ltd ) ;
 765                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 766                                         if ( $fc == '|' ) $l = 'td' ;
 767                                         else if ( $fc == '!' ) $l = 'th' ;
 768                                         else if ( $fc == '+' ) $l = 'caption' ;
 769                                         else $l = '' ;
 770                                         array_push ( $ltd , $l ) ;
 771
 772                                         # Cell parameters
 773                                         $y = explode ( '|' , $theline , 2 ) ;
 774                                         # Note that a '|' inside an invalid link should not
 775                                         # be mistaken as delimiting cell parameters
 776                                         if ( strpos( $y[0], '[[' ) !== false ) {
 777                                                 $y = array ($theline);
 778                                         }
 779                                         if ( count ( $y ) == 1 )
 780                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 781                                         else {
 782                                                 $attributes = $this->unstripForHTML( $y[0] );
 783                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 784                                         }
 785                                         $t[$k] .= $y ;
 786                                         array_push ( $td , true ) ;
 787                                 }
 788                         }
 789                 }
 790
 791                 # Closing open td, tr && table
 792                 while ( count ( $td ) > 0 )
 793                 {
 794                         $l = array_pop ( $ltd ) ;
 795                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 796                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 797                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 798                         $t[] = '</table>' ;
 799                 }
 800
 801                 $t = implode ( "\n" , $t ) ;
 802                 # special case: don't return empty table
 803                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 804                         $t = '';
 805                 wfProfileOut( $fname );
 806                 return $t ;
 807         }
 808
 809         /**
 810          * Helper function for parse() that transforms wiki markup into
 811          * HTML. Only called for $mOutputType == OT_HTML.
 812          *
 813          * @private
 814          */
 815         function internalParse( $text ) {
 816                 $args = array();
 817                 $isMain = true;
 818                 $fname = 'Parser::internalParse';
 819                 wfProfileIn( $fname );
 820
 821                 # Remove <noinclude> tags and <includeonly> sections
 822                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 823                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 824                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 825
 826                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 827
 828                 $text = $this->replaceVariables( $text, $args );
 829
 830                 // Tables need to come after variable replacement for things to work
 831                 // properly; putting them before other transformations should keep
 832                 // exciting things like link expansions from showing up in surprising
 833                 // places.
 834                 $text = $this->doTableStuff( $text );
 835
 836                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 837
 838                 $text = $this->stripToc( $text );
 839                 $text = $this->doHeadings( $text );
 840                 if($this->mOptions->getUseDynamicDates()) {
 841                         $df =& DateFormatter::getInstance();
 842                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 843                 }
 844                 $text = $this->doAllQuotes( $text );
 845                 $text = $this->replaceInternalLinks( $text );
 846                 $text = $this->replaceExternalLinks( $text );
 847
 848                 # replaceInternalLinks may sometimes leave behind
 849                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 850                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 851
 852                 $text = $this->doMagicLinks( $text );
 853                 $text = $this->formatHeadings( $text, $isMain );
 854
 855                 wfProfileOut( $fname );
 856                 return $text;
 857         }
 858
 859         /**
 860          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 861          * magic external links.
 862          *
 863          * @private
 864          */
 865         function &doMagicLinks( &$text ) {
 866                 $text = $this->magicISBN( $text );
 867                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 868                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 869                 return $text;
 870         }
 871
 872         /**
 873          * Parse headers and return html
 874          *
 875          * @private
 876          */
 877         function doHeadings( $text ) {
 878                 $fname = 'Parser::doHeadings';
 879                 wfProfileIn( $fname );
 880                 for ( $i = 6; $i >= 1; --$i ) {
 881                         $h = str_repeat( '=', $i );
 882                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 883                           "<h{$i}>\\1</h{$i}>\\2", $text );
 884                 }
 885                 wfProfileOut( $fname );
 886                 return $text;
 887         }
 888
 889         /**
 890          * Replace single quotes with HTML markup
 891          * @private
 892          * @return string the altered text
 893          */
 894         function doAllQuotes( $text ) {
 895                 $fname = 'Parser::doAllQuotes';
 896                 wfProfileIn( $fname );
 897                 $outtext = '';
 898                 $lines = explode( "\n", $text );
 899                 foreach ( $lines as $line ) {
 900                         $outtext .= $this->doQuotes ( $line ) . "\n";
 901                 }
 902                 $outtext = substr($outtext, 0,-1);
 903                 wfProfileOut( $fname );
 904                 return $outtext;
 905         }
 906
 907         /**
 908          * Helper function for doAllQuotes()
 909          * @private
 910          */
 911         function doQuotes( $text ) {
 912                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 913                 if ( count( $arr ) == 1 )
 914                         return $text;
 915                 else
 916                 {
 917                         # First, do some preliminary work. This may shift some apostrophes from
 918                         # being mark-up to being text. It also counts the number of occurrences
 919                         # of bold and italics mark-ups.
 920                         $i = 0;
 921                         $numbold = 0;
 922                         $numitalics = 0;
 923                         foreach ( $arr as $r )
 924                         {
 925                                 if ( ( $i % 2 ) == 1 )
 926                                 {
 927                                         # If there are ever four apostrophes, assume the first is supposed to
 928                                         # be text, and the remaining three constitute mark-up for bold text.
 929                                         if ( strlen( $arr[$i] ) == 4 )
 930                                         {
 931                                                 $arr[$i-1] .= "'";
 932                                                 $arr[$i] = "'''";
 933                                         }
 934                                         # If there are more than 5 apostrophes in a row, assume they're all
 935                                         # text except for the last 5.
 936                                         else if ( strlen( $arr[$i] ) > 5 )
 937                                         {
 938                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 939                                                 $arr[$i] = "'''''";
 940                                         }
 941                                         # Count the number of occurrences of bold and italics mark-ups.
 942                                         # We are not counting sequences of five apostrophes.
 943                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 944                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 945                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 946                                 }
 947                                 $i++;
 948                         }
 949
 950                         # If there is an odd number of both bold and italics, it is likely
 951                         # that one of the bold ones was meant to be an apostrophe followed
 952                         # by italics. Which one we cannot know for certain, but it is more
 953                         # likely to be one that has a single-letter word before it.
 954                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 955                         {
 956                                 $i = 0;
 957                                 $firstsingleletterword = -1;
 958                                 $firstmultiletterword = -1;
 959                                 $firstspace = -1;
 960                                 foreach ( $arr as $r )
 961                                 {
 962                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 963                                         {
 964                                                 $x1 = substr ($arr[$i-1], -1);
 965                                                 $x2 = substr ($arr[$i-1], -2, 1);
 966                                                 if ($x1 == ' ') {
 967                                                         if ($firstspace == -1) $firstspace = $i;
 968                                                 } else if ($x2 == ' ') {
 969                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 970                                                 } else {
 971                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 972                                                 }
 973                                         }
 974                                         $i++;
 975                                 }
 976
 977                                 # If there is a single-letter word, use it!
 978                                 if ($firstsingleletterword > -1)
 979                                 {
 980                                         $arr [ $firstsingleletterword ] = "''";
 981                                         $arr [ $firstsingleletterword-1 ] .= "'";
 982                                 }
 983                                 # If not, but there's a multi-letter word, use that one.
 984                                 else if ($firstmultiletterword > -1)
 985                                 {
 986                                         $arr [ $firstmultiletterword ] = "''";
 987                                         $arr [ $firstmultiletterword-1 ] .= "'";
 988                                 }
 989                                 # ... otherwise use the first one that has neither.
 990                                 # (notice that it is possible for all three to be -1 if, for example,
 991                                 # there is only one pentuple-apostrophe in the line)
 992                                 else if ($firstspace > -1)
 993                                 {
 994                                         $arr [ $firstspace ] = "''";
 995                                         $arr [ $firstspace-1 ] .= "'";
 996                                 }
 997                         }
 998
 999                         # Now let's actually convert our apostrophic mush to HTML!
1000                         $output = '';
1001                         $buffer = '';
1002                         $state = '';
1003                         $i = 0;
1004                         foreach ($arr as $r)
1005                         {
1006                                 if (($i % 2) == 0)
1007                                 {
1008                                         if ($state == 'both')
1009                                                 $buffer .= $r;
1010                                         else
1011                                                 $output .= $r;
1012                                 }
1013                                 else
1014                                 {
1015                                         if (strlen ($r) == 2)
1016                                         {
1017                                                 if ($state == 'i')
1018                                                 { $output .= '</i>'; $state = ''; }
1019                                                 else if ($state == 'bi')
1020                                                 { $output .= '</i>'; $state = 'b'; }
1021                                                 else if ($state == 'ib')
1022                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1023                                                 else if ($state == 'both')
1024                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1025                                                 else # $state can be 'b' or ''
1026                                                 { $output .= '<i>'; $state .= 'i'; }
1027                                         }
1028                                         else if (strlen ($r) == 3)
1029                                         {
1030                                                 if ($state == 'b')
1031                                                 { $output .= '</b>'; $state = ''; }
1032                                                 else if ($state == 'bi')
1033                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1034                                                 else if ($state == 'ib')
1035                                                 { $output .= '</b>'; $state = 'i'; }
1036                                                 else if ($state == 'both')
1037                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1038                                                 else # $state can be 'i' or ''
1039                                                 { $output .= '<b>'; $state .= 'b'; }
1040                                         }
1041                                         else if (strlen ($r) == 5)
1042                                         {
1043                                                 if ($state == 'b')
1044                                                 { $output .= '</b><i>'; $state = 'i'; }
1045                                                 else if ($state == 'i')
1046                                                 { $output .= '</i><b>'; $state = 'b'; }
1047                                                 else if ($state == 'bi')
1048                                                 { $output .= '</i></b>'; $state = ''; }
1049                                                 else if ($state == 'ib')
1050                                                 { $output .= '</b></i>'; $state = ''; }
1051                                                 else if ($state == 'both')
1052                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1053                                                 else # ($state == '')
1054                                                 { $buffer = ''; $state = 'both'; }
1055                                         }
1056                                 }
1057                                 $i++;
1058                         }
1059                         # Now close all remaining tags.  Notice that the order is important.
1060                         if ($state == 'b' || $state == 'ib')
1061                                 $output .= '</b>';
1062                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1063                                 $output .= '</i>';
1064                         if ($state == 'bi')
1065                                 $output .= '</b>';
1066                         if ($state == 'both')
1067                                 $output .= '<b><i>'.$buffer.'</i></b>';
1068                         return $output;
1069                 }
1070         }
1071
1072         /**
1073          * Replace external links
1074          *
1075          * Note: this is all very hackish and the order of execution matters a lot.
1076          * Make sure to run maintenance/parserTests.php if you change this code.
1077          *
1078          * @private
1079          */
1080         function replaceExternalLinks( $text ) {
1081                 global $wgContLang;
1082                 $fname = 'Parser::replaceExternalLinks';
1083                 wfProfileIn( $fname );
1084
1085                 $sk =& $this->mOptions->getSkin();
1086
1087                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1088
1089                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1090
1091                 $i = 0;
1092                 while ( $i<count( $bits ) ) {
1093                         $url = $bits[$i++];
1094                         $protocol = $bits[$i++];
1095                         $text = $bits[$i++];
1096                         $trail = $bits[$i++];
1097
1098                         # The characters '<' and '>' (which were escaped by
1099                         # removeHTMLtags()) should not be included in
1100                         # URLs, per RFC 2396.
1101                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1102                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1103                                 $url = substr($url, 0, $m2[0][1]);
1104                         }
1105
1106                         # If the link text is an image URL, replace it with an <img> tag
1107                         # This happened by accident in the original parser, but some people used it extensively
1108                         $img = $this->maybeMakeExternalImage( $text );
1109                         if ( $img !== false ) {
1110                                 $text = $img;
1111                         }
1112
1113                         $dtrail = '';
1114
1115                         # Set linktype for CSS - if URL==text, link is essentially free
1116                         $linktype = ($text == $url) ? 'free' : 'text';
1117
1118                         # No link text, e.g. [http://domain.tld/some.link]
1119                         if ( $text == '' ) {
1120                                 # Autonumber if allowed. See bug #5918
1121                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1122                                         $text = '[' . ++$this->mAutonumber . ']';
1123                                         $linktype = 'autonumber';
1124                                 } else {
1125                                         # Otherwise just use the URL
1126                                         $text = htmlspecialchars( $url );
1127                                         $linktype = 'free';
1128                                 }
1129                         } else {
1130                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1131                                 # Check for trail
1132                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1133                         }
1134
1135                         $text = $wgContLang->markNoConversion($text);
1136
1137                         # Normalize any HTML entities in input. They will be
1138                         # re-escaped by makeExternalLink().
1139                         $url = Sanitizer::decodeCharReferences( $url );
1140
1141                         # Process the trail (i.e. everything after this link up until start of the next link),
1142                         # replacing any non-bracketed links
1143                         $trail = $this->replaceFreeExternalLinks( $trail );
1144
1145                         # Use the encoded URL
1146                         # This means that users can paste URLs directly into the text
1147                         # Funny characters like &ouml; aren't valid in URLs anyway
1148                         # This was changed in August 2004
1149                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1150
1151                         # Register link in the output object.
1152                         # Replace unnecessary URL escape codes with the referenced character
1153                         # This prevents spammers from hiding links from the filters
1154                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1155                         $this->mOutput->addExternalLink( $pasteurized );
1156                 }
1157
1158                 wfProfileOut( $fname );
1159                 return $s;
1160         }
1161
1162         /**
1163          * Replace anything that looks like a URL with a link
1164          * @private
1165          */
1166         function replaceFreeExternalLinks( $text ) {
1167                 global $wgContLang;
1168                 $fname = 'Parser::replaceFreeExternalLinks';
1169                 wfProfileIn( $fname );
1170
1171                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1172                 $s = array_shift( $bits );
1173                 $i = 0;
1174
1175                 $sk =& $this->mOptions->getSkin();
1176
1177                 while ( $i < count( $bits ) ){
1178                         $protocol = $bits[$i++];
1179                         $remainder = $bits[$i++];
1180
1181                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1182                                 # Found some characters after the protocol that look promising
1183                                 $url = $protocol . $m[1];
1184                                 $trail = $m[2];
1185
1186                                 # special case: handle urls as url args:
1187                                 # http://www.example.com/foo?=http://www.example.com/bar
1188                                 if(strlen($trail) == 0 &&
1189                                         isset($bits[$i]) &&
1190                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1191                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1192                                 {
1193                                         # add protocol, arg
1194                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1195                                         $i += 2;
1196                                         $trail = $m[2];
1197                                 }
1198
1199                                 # The characters '<' and '>' (which were escaped by
1200                                 # removeHTMLtags()) should not be included in
1201                                 # URLs, per RFC 2396.
1202                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1203                                         $trail = substr($url, $m2[0][1]) . $trail;
1204                                         $url = substr($url, 0, $m2[0][1]);
1205                                 }
1206
1207                                 # Move trailing punctuation to $trail
1208                                 $sep = ',;\.:!?';
1209                                 # If there is no left bracket, then consider right brackets fair game too
1210                                 if ( strpos( $url, '(' ) === false ) {
1211                                         $sep .= ')';
1212                                 }
1213
1214                                 $numSepChars = strspn( strrev( $url ), $sep );
1215                                 if ( $numSepChars ) {
1216                                         $trail = substr( $url, -$numSepChars ) . $trail;
1217                                         $url = substr( $url, 0, -$numSepChars );
1218                                 }
1219
1220                                 # Normalize any HTML entities in input. They will be
1221                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1222                                 $url = Sanitizer::decodeCharReferences( $url );
1223
1224                                 # Is this an external image?
1225                                 $text = $this->maybeMakeExternalImage( $url );
1226                                 if ( $text === false ) {
1227                                         # Not an image, make a link
1228                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1229                                         # Register it in the output object...
1230                                         # Replace unnecessary URL escape codes with their equivalent characters
1231                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1232                                         $this->mOutput->addExternalLink( $pasteurized );
1233                                 }
1234                                 $s .= $text . $trail;
1235                         } else {
1236                                 $s .= $protocol . $remainder;
1237                         }
1238                 }
1239                 wfProfileOut( $fname );
1240                 return $s;
1241         }
1242
1243         /**
1244          * Replace unusual URL escape codes with their equivalent characters
1245          * @param string
1246          * @return string
1247          * @static
1248          * @fixme This can merge genuinely required bits in the path or query string,
1249          *        breaking legit URLs. A proper fix would treat the various parts of
1250          *        the URL differently; as a workaround, just use the output for
1251          *        statistical records, not for actual linking/output.
1252          */
1253         function replaceUnusualEscapes( $url ) {
1254                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1255                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1256         }
1257
1258         /**
1259          * Callback function used in replaceUnusualEscapes().
1260          * Replaces unusual URL escape codes with their equivalent character
1261          * @static
1262          * @private
1263          */
1264         function replaceUnusualEscapesCallback( $matches ) {
1265                 $char = urldecode( $matches[0] );
1266                 $ord = ord( $char );
1267                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1268                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1269                         // No, shouldn't be escaped
1270                         return $char;
1271                 } else {
1272                         // Yes, leave it escaped
1273                         return $matches[0];
1274                 }
1275         }
1276
1277         /**
1278          * make an image if it's allowed, either through the global
1279          * option or through the exception
1280          * @private
1281          */
1282         function maybeMakeExternalImage( $url ) {
1283                 $sk =& $this->mOptions->getSkin();
1284                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1285                 $imagesexception = !empty($imagesfrom);
1286                 $text = false;
1287                 if ( $this->mOptions->getAllowExternalImages()
1288                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1289                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1290                                 # Image found
1291                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1292                         }
1293                 }
1294                 return $text;
1295         }
1296
1297         /**
1298          * Process [[ ]] wikilinks
1299          *
1300          * @private
1301          */
1302         function replaceInternalLinks( $s ) {
1303                 global $wgContLang;
1304                 static $fname = 'Parser::replaceInternalLinks' ;
1305
1306                 wfProfileIn( $fname );
1307
1308                 wfProfileIn( $fname.'-setup' );
1309                 static $tc = FALSE;
1310                 # the % is needed to support urlencoded titles as well
1311                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1312
1313                 $sk =& $this->mOptions->getSkin();
1314
1315                 #split the entire text string on occurences of [[
1316                 $a = explode( '[[', ' ' . $s );
1317                 #get the first element (all text up to first [[), and remove the space we added
1318                 $s = array_shift( $a );
1319                 $s = substr( $s, 1 );
1320
1321                 # Match a link having the form [[namespace:link|alternate]]trail
1322                 static $e1 = FALSE;
1323                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1324                 # Match cases where there is no "]]", which might still be images
1325                 static $e1_img = FALSE;
1326                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1327                 # Match the end of a line for a word that's not followed by whitespace,
1328                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1329                 $e2 = wfMsgForContent( 'linkprefix' );
1330
1331                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1332
1333                 if( is_null( $this->mTitle ) ) {
1334                         wfDebugDieBacktrace( 'nooo' );
1335                 }
1336                 $nottalk = !$this->mTitle->isTalkPage();
1337
1338                 if ( $useLinkPrefixExtension ) {
1339                         if ( preg_match( $e2, $s, $m ) ) {
1340                                 $first_prefix = $m[2];
1341                         } else {
1342                                 $first_prefix = false;
1343                         }
1344                 } else {
1345                         $prefix = '';
1346                 }
1347
1348                 $selflink = $this->mTitle->getPrefixedText();
1349                 wfProfileOut( $fname.'-setup' );
1350
1351                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1352                 $useSubpages = $this->areSubpagesAllowed();
1353
1354                 # Loop for each link
1355                 for ($k = 0; isset( $a[$k] ); $k++) {
1356                         $line = $a[$k];
1357                         if ( $useLinkPrefixExtension ) {
1358                                 wfProfileIn( $fname.'-prefixhandling' );
1359                                 if ( preg_match( $e2, $s, $m ) ) {
1360                                         $prefix = $m[2];
1361                                         $s = $m[1];
1362                                 } else {
1363                                         $prefix='';
1364                                 }
1365                                 # first link
1366                                 if($first_prefix) {
1367                                         $prefix = $first_prefix;
1368                                         $first_prefix = false;
1369                                 }
1370                                 wfProfileOut( $fname.'-prefixhandling' );
1371                         }
1372
1373                         $might_be_img = false;
1374
1375                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1376                                 $text = $m[2];
1377                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1378                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1379                                 # the real problem is with the $e1 regex
1380                                 # See bug 1300.
1381                                 #
1382                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1383                                 # and no image is in sight. See bug 2095.
1384                                 #
1385                                 if( $text !== '' &&
1386                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1387                                         strpos($text, '[') !== false
1388                                 )
1389                                 {
1390                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1391                                         $m[3] = $n[1];
1392                                 }
1393                                 # fix up urlencoded title texts
1394                                 if(preg_match('/%/', $m[1] ))
1395                                         # Should anchors '#' also be rejected?
1396                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1397                                 $trail = $m[3];
1398                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1399                                 $might_be_img = true;
1400                                 $text = $m[2];
1401                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1402                                 $trail = "";
1403                         } else { # Invalid form; output directly
1404                                 $s .= $prefix . '[[' . $line ;
1405                                 continue;
1406                         }
1407
1408                         # Don't allow internal links to pages containing
1409                         # PROTO: where PROTO is a valid URL protocol; these
1410                         # should be external links.
1411                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1412                                 $s .= $prefix . '[[' . $line ;
1413                                 continue;
1414                         }
1415
1416                         # Make subpage if necessary
1417                         if( $useSubpages ) {
1418                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1419                         } else {
1420                                 $link = $m[1];
1421                         }
1422
1423                         $noforce = (substr($m[1], 0, 1) != ':');
1424                         if (!$noforce) {
1425                                 # Strip off leading ':'
1426                                 $link = substr($link, 1);
1427                         }
1428
1429                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1430                         if( !$nt ) {
1431                                 $s .= $prefix . '[[' . $line;
1432                                 continue;
1433                         }
1434
1435                         #check other language variants of the link
1436                         #if the article does not exist
1437                         if( $checkVariantLink
1438                             && $nt->getArticleID() == 0 ) {
1439                                 $wgContLang->findVariantLink($link, $nt);
1440                         }
1441
1442                         $ns = $nt->getNamespace();
1443                         $iw = $nt->getInterWiki();
1444
1445                         if ($might_be_img) { # if this is actually an invalid link
1446                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1447                                         $found = false;
1448                                         while (isset ($a[$k+1]) ) {
1449                                                 #look at the next 'line' to see if we can close it there
1450                                                 $spliced = array_splice( $a, $k + 1, 1 );
1451                                                 $next_line = array_shift( $spliced );
1452                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1453                                                 # the first ]] closes the inner link, the second the image
1454                                                         $found = true;
1455                                                         $text .= '[[' . $m[1];
1456                                                         $trail = $m[2];
1457                                                         break;
1458                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1459                                                         #if there's exactly one ]] that's fine, we'll keep looking
1460                                                         $text .= '[[' . $m[0];
1461                                                 } else {
1462                                                         #if $next_line is invalid too, we need look no further
1463                                                         $text .= '[[' . $next_line;
1464                                                         break;
1465                                                 }
1466                                         }
1467                                         if ( !$found ) {
1468                                                 # we couldn't find the end of this imageLink, so output it raw
1469                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1470                                                 $text = $this->replaceInternalLinks($text);
1471                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1472                                                 # note: no $trail, because without an end, there *is* no trail
1473                                                 continue;
1474                                         }
1475                                 } else { #it's not an image, so output it raw
1476                                         $s .= $prefix . '[[' . $link . '|' . $text;
1477                                         # note: no $trail, because without an end, there *is* no trail
1478                                         continue;
1479                                 }
1480                         }
1481
1482                         $wasblank = ( '' == $text );
1483                         if( $wasblank ) $text = $link;
1484
1485
1486                         # Link not escaped by : , create the various objects
1487                         if( $noforce ) {
1488
1489                                 # Interwikis
1490                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1491                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1492                                         $s = rtrim($s . "\n");
1493                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1494                                         continue;
1495                                 }
1496
1497                                 if ( $ns == NS_IMAGE ) {
1498                                         wfProfileIn( "$fname-image" );
1499                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1500                                                 # recursively parse links inside the image caption
1501                                                 # actually, this will parse them in any other parameters, too,
1502                                                 # but it might be hard to fix that, and it doesn't matter ATM
1503                                                 $text = $this->replaceExternalLinks($text);
1504                                                 $text = $this->replaceInternalLinks($text);
1505
1506                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1507                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1508                                                 $this->mOutput->addImage( $nt->getDBkey() );
1509
1510                                                 wfProfileOut( "$fname-image" );
1511                                                 continue;
1512                                         }
1513                                         wfProfileOut( "$fname-image" );
1514
1515                                 }
1516
1517                                 if ( $ns == NS_CATEGORY ) {
1518                                         wfProfileIn( "$fname-category" );
1519                                         $s = rtrim($s . "\n"); # bug 87
1520
1521                                         if ( $wasblank ) {
1522                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1523                                                         $sortkey = $this->mTitle->getText();
1524                                                 } else {
1525                                                         $sortkey = $this->mTitle->getPrefixedText();
1526                                                 }
1527                                         } else {
1528                                                 $sortkey = $text;
1529                                         }
1530                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1531                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1532                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1533
1534                                         /**
1535                                          * Strip the whitespace Category links produce, see bug 87
1536                                          * @todo We might want to use trim($tmp, "\n") here.
1537                                          */
1538                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1539
1540                                         wfProfileOut( "$fname-category" );
1541                                         continue;
1542                                 }
1543                         }
1544
1545                         if( ( $nt->getPrefixedText() === $selflink ) &&
1546                             ( $nt->getFragment() === '' ) ) {
1547                                 # Self-links are handled specially; generally de-link and change to bold.
1548                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1549                                 continue;
1550                         }
1551
1552                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1553                         if( $ns == NS_MEDIA ) {
1554                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1555                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1556                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1557                                 $this->mOutput->addImage( $nt->getDBkey() );
1558                                 continue;
1559                         } elseif( $ns == NS_SPECIAL ) {
1560                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1561                                 continue;
1562                         } elseif( $ns == NS_IMAGE ) {
1563                                 $img = Image::newFromTitle( $nt );
1564                                 if( $img->exists() ) {
1565                                         // Force a blue link if the file exists; may be a remote
1566                                         // upload on the shared repository, and we want to see its
1567                                         // auto-generated page.
1568                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1569                                         continue;
1570                                 }
1571                         }
1572                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1573                 }
1574                 wfProfileOut( $fname );
1575                 return $s;
1576         }
1577
1578         /**
1579          * Make a link placeholder. The text returned can be later resolved to a real link with
1580          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1581          * parsing of interwiki links, and secondly to allow all extistence checks and
1582          * article length checks (for stub links) to be bundled into a single query.
1583          *
1584          */
1585         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1586                 if ( ! is_object($nt) ) {
1587                         # Fail gracefully
1588                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1589                 } else {
1590                         # Separate the link trail from the rest of the link
1591                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1592
1593                         if ( $nt->isExternal() ) {
1594                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1595                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1596                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1597                         } else {
1598                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1599                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1600                                 $this->mLinkHolders['queries'][] = $query;
1601                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1602                                 $this->mLinkHolders['titles'][] = $nt;
1603
1604                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1605                         }
1606                 }
1607                 return $retVal;
1608         }
1609
1610         /**
1611          * Render a forced-blue link inline; protect against double expansion of
1612          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1613          * Since this little disaster has to split off the trail text to avoid
1614          * breaking URLs in the following text without breaking trails on the
1615          * wiki links, it's been made into a horrible function.
1616          *
1617          * @param Title $nt
1618          * @param string $text
1619          * @param string $query
1620          * @param string $trail
1621          * @param string $prefix
1622          * @return string HTML-wikitext mix oh yuck
1623          */
1624         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1625                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1626                 $sk =& $this->mOptions->getSkin();
1627                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1628                 return $this->armorLinks( $link ) . $trail;
1629         }
1630
1631         /**
1632          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1633          * going to go through further parsing steps before inline URL expansion.
1634          *
1635          * In particular this is important when using action=render, which causes
1636          * full URLs to be included.
1637          *
1638          * Oh man I hate our multi-layer parser!
1639          *
1640          * @param string more-or-less HTML
1641          * @return string less-or-more HTML with NOPARSE bits
1642          */
1643         function armorLinks( $text ) {
1644                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1645                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1646         }
1647
1648         /**
1649          * Return true if subpage links should be expanded on this page.
1650          * @return bool
1651          */
1652         function areSubpagesAllowed() {
1653                 # Some namespaces don't allow subpages
1654                 global $wgNamespacesWithSubpages;
1655                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1656         }
1657
1658         /**
1659          * Handle link to subpage if necessary
1660          * @param string $target the source of the link
1661          * @param string &$text the link text, modified as necessary
1662          * @return string the full name of the link
1663          * @private
1664          */
1665         function maybeDoSubpageLink($target, &$text) {
1666                 # Valid link forms:
1667                 # Foobar -- normal
1668                 # :Foobar -- override special treatment of prefix (images, language links)
1669                 # /Foobar -- convert to CurrentPage/Foobar
1670                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1671                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1672                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1673
1674                 $fname = 'Parser::maybeDoSubpageLink';
1675                 wfProfileIn( $fname );
1676                 $ret = $target; # default return value is no change
1677
1678                 # Some namespaces don't allow subpages,
1679                 # so only perform processing if subpages are allowed
1680                 if( $this->areSubpagesAllowed() ) {
1681                         # Look at the first character
1682                         if( $target != '' && $target{0} == '/' ) {
1683                                 # / at end means we don't want the slash to be shown
1684                                 if( substr( $target, -1, 1 ) == '/' ) {
1685                                         $target = substr( $target, 1, -1 );
1686                                         $noslash = $target;
1687                                 } else {
1688                                         $noslash = substr( $target, 1 );
1689                                 }
1690
1691                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1692                                 if( '' === $text ) {
1693                                         $text = $target;
1694                                 } # this might be changed for ugliness reasons
1695                         } else {
1696                                 # check for .. subpage backlinks
1697                                 $dotdotcount = 0;
1698                                 $nodotdot = $target;
1699                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1700                                         ++$dotdotcount;
1701                                         $nodotdot = substr( $nodotdot, 3 );
1702                                 }
1703                                 if($dotdotcount > 0) {
1704                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1705                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1706                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1707                                                 # / at the end means don't show full path
1708                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1709                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1710                                                         if( '' === $text ) {
1711                                                                 $text = $nodotdot;
1712                                                         }
1713                                                 }
1714                                                 $nodotdot = trim( $nodotdot );
1715                                                 if( $nodotdot != '' ) {
1716                                                         $ret .= '/' . $nodotdot;
1717                                                 }
1718                                         }
1719                                 }
1720                         }
1721                 }
1722
1723                 wfProfileOut( $fname );
1724                 return $ret;
1725         }
1726
1727         /**#@+
1728          * Used by doBlockLevels()
1729          * @private
1730          */
1731         /* private */ function closeParagraph() {
1732                 $result = '';
1733                 if ( '' != $this->mLastSection ) {
1734                         $result = '</' . $this->mLastSection  . ">\n";
1735                 }
1736                 $this->mInPre = false;
1737                 $this->mLastSection = '';
1738                 return $result;
1739         }
1740         # getCommon() returns the length of the longest common substring
1741         # of both arguments, starting at the beginning of both.
1742         #
1743         /* private */ function getCommon( $st1, $st2 ) {
1744                 $fl = strlen( $st1 );
1745                 $shorter = strlen( $st2 );
1746                 if ( $fl < $shorter ) { $shorter = $fl; }
1747
1748                 for ( $i = 0; $i < $shorter; ++$i ) {
1749                         if ( $st1{$i} != $st2{$i} ) { break; }
1750                 }
1751                 return $i;
1752         }
1753         # These next three functions open, continue, and close the list
1754         # element appropriate to the prefix character passed into them.
1755         #
1756         /* private */ function openList( $char ) {
1757                 $result = $this->closeParagraph();
1758
1759                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1760                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1761                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1762                 else if ( ';' == $char ) {
1763                         $result .= '<dl><dt>';
1764                         $this->mDTopen = true;
1765                 }
1766                 else { $result = '<!-- ERR 1 -->'; }
1767
1768                 return $result;
1769         }
1770
1771         /* private */ function nextItem( $char ) {
1772                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1773                 else if ( ':' == $char || ';' == $char ) {
1774                         $close = '</dd>';
1775                         if ( $this->mDTopen ) { $close = '</dt>'; }
1776                         if ( ';' == $char ) {
1777                                 $this->mDTopen = true;
1778                                 return $close . '<dt>';
1779                         } else {
1780                                 $this->mDTopen = false;
1781                                 return $close . '<dd>';
1782                         }
1783                 }
1784                 return '<!-- ERR 2 -->';
1785         }
1786
1787         /* private */ function closeList( $char ) {
1788                 if ( '*' == $char ) { $text = '</li></ul>'; }
1789                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1790                 else if ( ':' == $char ) {
1791                         if ( $this->mDTopen ) {
1792                                 $this->mDTopen = false;
1793                                 $text = '</dt></dl>';
1794                         } else {
1795                                 $text = '</dd></dl>';
1796                         }
1797                 }
1798                 else {  return '<!-- ERR 3 -->'; }
1799                 return $text."\n";
1800         }
1801         /**#@-*/
1802
1803         /**
1804          * Make lists from lines starting with ':', '*', '#', etc.
1805          *
1806          * @private
1807          * @return string the lists rendered as HTML
1808          */
1809         function doBlockLevels( $text, $linestart ) {
1810                 $fname = 'Parser::doBlockLevels';
1811                 wfProfileIn( $fname );
1812
1813                 # Parsing through the text line by line.  The main thing
1814                 # happening here is handling of block-level elements p, pre,
1815                 # and making lists from lines starting with * # : etc.
1816                 #
1817                 $textLines = explode( "\n", $text );
1818
1819                 $lastPrefix = $output = '';
1820                 $this->mDTopen = $inBlockElem = false;
1821                 $prefixLength = 0;
1822                 $paragraphStack = false;
1823
1824                 if ( !$linestart ) {
1825                         $output .= array_shift( $textLines );
1826                 }
1827                 foreach ( $textLines as $oLine ) {
1828                         $lastPrefixLength = strlen( $lastPrefix );
1829                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1830                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1831                         if ( !$this->mInPre ) {
1832                                 # Multiple prefixes may abut each other for nested lists.
1833                                 $prefixLength = strspn( $oLine, '*#:;' );
1834                                 $pref = substr( $oLine, 0, $prefixLength );
1835
1836                                 # eh?
1837                                 $pref2 = str_replace( ';', ':', $pref );
1838                                 $t = substr( $oLine, $prefixLength );
1839                                 $this->mInPre = !empty($preOpenMatch);
1840                         } else {
1841                                 # Don't interpret any other prefixes in preformatted text
1842                                 $prefixLength = 0;
1843                                 $pref = $pref2 = '';
1844                                 $t = $oLine;
1845                         }
1846
1847                         # List generation
1848                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1849                                 # Same as the last item, so no need to deal with nesting or opening stuff
1850                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1851                                 $paragraphStack = false;
1852
1853                                 if ( substr( $pref, -1 ) == ';') {
1854                                         # The one nasty exception: definition lists work like this:
1855                                         # ; title : definition text
1856                                         # So we check for : in the remainder text to split up the
1857                                         # title and definition, without b0rking links.
1858                                         $term = $t2 = '';
1859                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1860                                                 $t = $t2;
1861                                                 $output .= $term . $this->nextItem( ':' );
1862                                         }
1863                                 }
1864                         } elseif( $prefixLength || $lastPrefixLength ) {
1865                                 # Either open or close a level...
1866                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1867                                 $paragraphStack = false;
1868
1869                                 while( $commonPrefixLength < $lastPrefixLength ) {
1870                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1871                                         --$lastPrefixLength;
1872                                 }
1873                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1874                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1875                                 }
1876                                 while ( $prefixLength > $commonPrefixLength ) {
1877                                         $char = substr( $pref, $commonPrefixLength, 1 );
1878                                         $output .= $this->openList( $char );
1879
1880                                         if ( ';' == $char ) {
1881                                                 # FIXME: This is dupe of code above
1882                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1883                                                         $t = $t2;
1884                                                         $output .= $term . $this->nextItem( ':' );
1885                                                 }
1886                                         }
1887                                         ++$commonPrefixLength;
1888                                 }
1889                                 $lastPrefix = $pref2;
1890                         }
1891                         if( 0 == $prefixLength ) {
1892                                 wfProfileIn( "$fname-paragraph" );
1893                                 # No prefix (not in list)--go to paragraph mode
1894                                 // XXX: use a stack for nestable elements like span, table and div
1895                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1896                                 $closematch = preg_match(
1897                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1898                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1899                                 if ( $openmatch or $closematch ) {
1900                                         $paragraphStack = false;
1901                                         # TODO bug 5718: paragraph closed
1902                                         $output .= $this->closeParagraph();
1903                                         if ( $preOpenMatch and !$preCloseMatch ) {
1904                                                 $this->mInPre = true;
1905                                         }
1906                                         if ( $closematch ) {
1907                                                 $inBlockElem = false;
1908                                         } else {
1909                                                 $inBlockElem = true;
1910                                         }
1911                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1912                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1913                                                 // pre
1914                                                 if ($this->mLastSection != 'pre') {
1915                                                         $paragraphStack = false;
1916                                                         $output .= $this->closeParagraph().'<pre>';
1917                                                         $this->mLastSection = 'pre';
1918                                                 }
1919                                                 $t = substr( $t, 1 );
1920                                         } else {
1921                                                 // paragraph
1922                                                 if ( '' == trim($t) ) {
1923                                                         if ( $paragraphStack ) {
1924                                                                 $output .= $paragraphStack.'<br />';
1925                                                                 $paragraphStack = false;
1926                                                                 $this->mLastSection = 'p';
1927                                                         } else {
1928                                                                 if ($this->mLastSection != 'p' ) {
1929                                                                         $output .= $this->closeParagraph();
1930                                                                         $this->mLastSection = '';
1931                                                                         $paragraphStack = '<p>';
1932                                                                 } else {
1933                                                                         $paragraphStack = '</p><p>';
1934                                                                 }
1935                                                         }
1936                                                 } else {
1937                                                         if ( $paragraphStack ) {
1938                                                                 $output .= $paragraphStack;
1939                                                                 $paragraphStack = false;
1940                                                                 $this->mLastSection = 'p';
1941                                                         } else if ($this->mLastSection != 'p') {
1942                                                                 $output .= $this->closeParagraph().'<p>';
1943                                                                 $this->mLastSection = 'p';
1944                                                         }
1945                                                 }
1946                                         }
1947                                 }
1948                                 wfProfileOut( "$fname-paragraph" );
1949                         }
1950                         // somewhere above we forget to get out of pre block (bug 785)
1951                         if($preCloseMatch && $this->mInPre) {
1952                                 $this->mInPre = false;
1953                         }
1954                         if ($paragraphStack === false) {
1955                                 $output .= $t."\n";
1956                         }
1957                 }
1958                 while ( $prefixLength ) {
1959                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1960                         --$prefixLength;
1961                 }
1962                 if ( '' != $this->mLastSection ) {
1963                         $output .= '</' . $this->mLastSection . '>';
1964                         $this->mLastSection = '';
1965                 }
1966
1967                 wfProfileOut( $fname );
1968                 return $output;
1969         }
1970
1971         /**
1972          * Split up a string on ':', ignoring any occurences inside
1973          * <a>..</a> or <span>...</span>
1974          * @param string $str the string to split
1975          * @param string &$before set to everything before the ':'
1976          * @param string &$after set to everything after the ':'
1977          * return string the position of the ':', or false if none found
1978          */
1979         function findColonNoLinks($str, &$before, &$after) {
1980                 # I wonder if we should make this count all tags, not just <a>
1981                 # and <span>. That would prevent us from matching a ':' that
1982                 # comes in the middle of italics other such formatting....
1983                 # -- Wil
1984                 $fname = 'Parser::findColonNoLinks';
1985                 wfProfileIn( $fname );
1986                 $pos = 0;
1987                 do {
1988                         $colon = strpos($str, ':', $pos);
1989
1990                         if ($colon !== false) {
1991                                 $before = substr($str, 0, $colon);
1992                                 $after = substr($str, $colon + 1);
1993
1994                                 # Skip any ':' within <a> or <span> pairs
1995                                 $a = substr_count($before, '<a');
1996                                 $s = substr_count($before, '<span');
1997                                 $ca = substr_count($before, '</a>');
1998                                 $cs = substr_count($before, '</span>');
1999
2000                                 if ($a <= $ca and $s <= $cs) {
2001                                         # Tags are balanced before ':'; ok
2002                                         break;
2003                                 }
2004                                 $pos = $colon + 1;
2005                         }
2006                 } while ($colon !== false);
2007                 wfProfileOut( $fname );
2008                 return $colon;
2009         }
2010
2011         /**
2012          * Return value of a magic variable (like PAGENAME)
2013          *
2014          * @private
2015          */
2016         function getVariableValue( $index ) {
2017                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2018
2019                 /**
2020                  * Some of these require message or data lookups and can be
2021                  * expensive to check many times.
2022                  */
2023                 static $varCache = array();
2024                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2025                         if ( isset( $varCache[$index] ) )
2026                                 return $varCache[$index];
2027
2028                 $ts = time();
2029                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2030
2031                 switch ( $index ) {
2032                         case MAG_CURRENTMONTH:
2033                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2034                         case MAG_CURRENTMONTHNAME:
2035                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2036                         case MAG_CURRENTMONTHNAMEGEN:
2037                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2038                         case MAG_CURRENTMONTHABBREV:
2039                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2040                         case MAG_CURRENTDAY:
2041                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2042                         case MAG_CURRENTDAY2:
2043                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2044                         case MAG_PAGENAME:
2045                                 return $this->mTitle->getText();
2046                         case MAG_PAGENAMEE:
2047                                 return $this->mTitle->getPartialURL();
2048                         case MAG_FULLPAGENAME:
2049                                 return $this->mTitle->getPrefixedText();
2050                         case MAG_FULLPAGENAMEE:
2051                                 return $this->mTitle->getPrefixedURL();
2052                         case MAG_SUBPAGENAME:
2053                                 return $this->mTitle->getSubpageText();
2054                         case MAG_SUBPAGENAMEE:
2055                                 return $this->mTitle->getSubpageUrlForm();
2056                         case MAG_BASEPAGENAME:
2057                                 return $this->mTitle->getBaseText();
2058                         case MAG_BASEPAGENAMEE:
2059                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2060                         case MAG_TALKPAGENAME:
2061                                 if( $this->mTitle->canTalk() ) {
2062                                         $talkPage = $this->mTitle->getTalkPage();
2063                                         return $talkPage->getPrefixedText();
2064                                 } else {
2065                                         return '';
2066                                 }
2067                         case MAG_TALKPAGENAMEE:
2068                                 if( $this->mTitle->canTalk() ) {
2069                                         $talkPage = $this->mTitle->getTalkPage();
2070                                         return $talkPage->getPrefixedUrl();
2071                                 } else {
2072                                         return '';
2073                                 }
2074                         case MAG_SUBJECTPAGENAME:
2075                                 $subjPage = $this->mTitle->getSubjectPage();
2076                                 return $subjPage->getPrefixedText();
2077                         case MAG_SUBJECTPAGENAMEE:
2078                                 $subjPage = $this->mTitle->getSubjectPage();
2079                                 return $subjPage->getPrefixedUrl();
2080                         case MAG_REVISIONID:
2081                                 return $this->mRevisionId;
2082                         case MAG_NAMESPACE:
2083                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2084                         case MAG_NAMESPACEE:
2085                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2086                         case MAG_TALKSPACE:
2087                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2088                         case MAG_TALKSPACEE:
2089                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2090                         case MAG_SUBJECTSPACE:
2091                                 return $this->mTitle->getSubjectNsText();
2092                         case MAG_SUBJECTSPACEE:
2093                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2094                         case MAG_CURRENTDAYNAME:
2095                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2096                         case MAG_CURRENTYEAR:
2097                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2098                         case MAG_CURRENTTIME:
2099                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2100                         case MAG_CURRENTWEEK:
2101                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2102                                 // int to remove the padding
2103                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2104                         case MAG_CURRENTDOW:
2105                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2106                         case MAG_NUMBEROFARTICLES:
2107                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2108                         case MAG_NUMBEROFFILES:
2109                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2110                         case MAG_NUMBEROFUSERS:
2111                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2112                         case MAG_NUMBEROFPAGES:
2113                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2114                         case MAG_CURRENTTIMESTAMP:
2115                                 return $varCache[$index] = wfTimestampNow();
2116                         case MAG_CURRENTVERSION:
2117                                 global $wgVersion;
2118                                 return $wgVersion;
2119                         case MAG_SITENAME:
2120                                 return $wgSitename;
2121                         case MAG_SERVER:
2122                                 return $wgServer;
2123                         case MAG_SERVERNAME:
2124                                 return $wgServerName;
2125                         case MAG_SCRIPTPATH:
2126                                 return $wgScriptPath;
2127                         case MAG_DIRECTIONMARK:
2128                                 return $wgContLang->getDirMark();
2129                         default:
2130                                 $ret = null;
2131                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2132                                         return $ret;
2133                                 else
2134                                         return null;
2135                 }
2136         }
2137
2138         /**
2139          * initialise the magic variables (like CURRENTMONTHNAME)
2140          *
2141          * @private
2142          */
2143         function initialiseVariables() {
2144                 $fname = 'Parser::initialiseVariables';
2145                 wfProfileIn( $fname );
2146                 global $wgVariableIDs;
2147                 $this->mVariables = array();
2148                 foreach ( $wgVariableIDs as $id ) {
2149                         $mw =& MagicWord::get( $id );
2150                         $mw->addToArray( $this->mVariables, $id );
2151                 }
2152                 wfProfileOut( $fname );
2153         }
2154
2155         /**
2156          * parse any parentheses in format ((title|part|part))
2157          * and call callbacks to get a replacement text for any found piece
2158          *
2159          * @param string $text The text to parse
2160          * @param array $callbacks rules in form:
2161          *     '{' => array(                            # opening parentheses
2162          *                                      'end' => '}',   # closing parentheses
2163          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2164          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2165          *                                                                )
2166          *                                      )
2167          * @private
2168          */
2169         function replace_callback ($text, $callbacks) {
2170                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2171                 $lastOpeningBrace = -1;                 # last not closed parentheses
2172
2173                 for ($i = 0; $i < strlen($text); $i++) {
2174                         # check for any opening brace
2175                         $rule = null;
2176                         $nextPos = -1;
2177                         foreach ($callbacks as $key => $value) {
2178                                 $pos = strpos ($text, $key, $i);
2179                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2180                                         $rule = $value;
2181                                         $nextPos = $pos;
2182                                 }
2183                         }
2184
2185                         if ($lastOpeningBrace >= 0) {
2186                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2187
2188                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2189                                         $rule = null;
2190                                         $nextPos = $pos;
2191                                 }
2192
2193                                 $pos = strpos ($text, '|', $i);
2194
2195                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2196                                         $rule = null;
2197                                         $nextPos = $pos;
2198                                 }
2199                         }
2200
2201                         if ($nextPos == -1)
2202                                 break;
2203
2204                         $i = $nextPos;
2205
2206                         # found openning brace, lets add it to parentheses stack
2207                         if (null != $rule) {
2208                                 $piece = array('brace' => $text[$i],
2209                                                            'braceEnd' => $rule['end'],
2210                                                            'count' => 1,
2211                                                            'title' => '',
2212                                                            'parts' => null);
2213
2214                                 # count openning brace characters
2215                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2216                                         $piece['count']++;
2217                                         $i++;
2218                                 }
2219
2220                                 $piece['startAt'] = $i+1;
2221                                 $piece['partStart'] = $i+1;
2222
2223                                 # we need to add to stack only if openning brace count is enough for any given rule
2224                                 foreach ($rule['cb'] as $cnt => $fn) {
2225                                         if ($piece['count'] >= $cnt) {
2226                                                 $lastOpeningBrace ++;
2227                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2228                                                 break;
2229                                         }
2230                                 }
2231
2232                                 continue;
2233                         }
2234                         else if ($lastOpeningBrace >= 0) {
2235                                 # first check if it is a closing brace
2236                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2237                                         # lets check if it is enough characters for closing brace
2238                                         $count = 1;
2239                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2240                                                 $count++;
2241
2242                                         # if there are more closing parentheses than opening ones, we parse less
2243                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2244                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2245
2246                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2247                                         $matchingCount = 0;
2248                                         $matchingCallback = null;
2249                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2250                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2251                                                         $matchingCount = $cnt;
2252                                                         $matchingCallback = $fn;
2253                                                 }
2254                                         }
2255
2256                                         if ($matchingCount == 0) {
2257                                                 $i += $count - 1;
2258                                                 continue;
2259                                         }
2260
2261                                         # lets set a title or last part (if '|' was found)
2262                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2263                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2264                                         else
2265                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2266
2267                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2268                                         $pieceEnd = $i + $matchingCount;
2269
2270                                         if( is_callable( $matchingCallback ) ) {
2271                                                 $cbArgs = array (
2272                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2273                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2274                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2275                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2276                                                                                  );
2277                                                 # finally we can call a user callback and replace piece of text
2278                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2279                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2280                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2281                                         }
2282                                         else {
2283                                                 # null value for callback means that parentheses should be parsed, but not replaced
2284                                                 $i += $matchingCount - 1;
2285                                         }
2286
2287                                         # reset last openning parentheses, but keep it in case there are unused characters
2288                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2289                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2290                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2291                                                                    'title' => '',
2292                                                                    'parts' => null,
2293                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2294                                         $openingBraceStack[$lastOpeningBrace--] = null;
2295
2296                                         if ($matchingCount < $piece['count']) {
2297                                                 $piece['count'] -= $matchingCount;
2298                                                 $piece['startAt'] -= $matchingCount;
2299                                                 $piece['partStart'] = $piece['startAt'];
2300                                                 # do we still qualify for any callback with remaining count?
2301                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2302                                                         if ($piece['count'] >= $cnt) {
2303                                                                 $lastOpeningBrace ++;
2304                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2305                                                                 break;
2306                                                         }
2307                                                 }
2308                                         }
2309                                         continue;
2310                                 }
2311
2312                                 # lets set a title if it is a first separator, or next part otherwise
2313                                 if ($text[$i] == '|') {
2314                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2315                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2316                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2317                                         }
2318                                         else
2319                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2320
2321                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2322                                 }
2323                         }
2324                 }
2325
2326                 return $text;
2327         }
2328
2329         /**
2330          * Replace magic variables, templates, and template arguments
2331          * with the appropriate text. Templates are substituted recursively,
2332          * taking care to avoid infinite loops.
2333          *
2334          * Note that the substitution depends on value of $mOutputType:
2335          *  OT_WIKI: only {{subst:}} templates
2336          *  OT_MSG: only magic variables
2337          *  OT_HTML: all templates and magic variables
2338          *
2339          * @param string $tex The text to transform
2340          * @param array $args Key-value pairs representing template parameters to substitute
2341          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2342          * @private
2343          */
2344         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2345                 # Prevent too big inclusions
2346                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2347                         return $text;
2348                 }
2349
2350                 $fname = 'Parser::replaceVariables';
2351                 wfProfileIn( $fname );
2352
2353                 # This function is called recursively. To keep track of arguments we need a stack:
2354                 array_push( $this->mArgStack, $args );
2355
2356                 $braceCallbacks = array();
2357                 if ( !$argsOnly ) {
2358                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2359                 }
2360                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2361                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2362                 }
2363                 $callbacks = array();
2364                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2365                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2366                 $text = $this->replace_callback ($text, $callbacks);
2367
2368                 array_pop( $this->mArgStack );
2369
2370                 wfProfileOut( $fname );
2371                 return $text;
2372         }
2373
2374         /**
2375          * Replace magic variables
2376          * @private
2377          */
2378         function variableSubstitution( $matches ) {
2379                 $fname = 'Parser::variableSubstitution';
2380                 $varname = $matches[1];
2381                 wfProfileIn( $fname );
2382                 if ( !$this->mVariables ) {
2383                         $this->initialiseVariables();
2384                 }
2385                 $skip = false;
2386                 if ( $this->mOutputType == OT_WIKI ) {
2387                         # Do only magic variables prefixed by SUBST
2388                         $mwSubst =& MagicWord::get( MAG_SUBST );
2389                         if (!$mwSubst->matchStartAndRemove( $varname ))
2390                                 $skip = true;
2391                         # Note that if we don't substitute the variable below,
2392                         # we don't remove the {{subst:}} magic word, in case
2393                         # it is a template rather than a magic variable.
2394                 }
2395                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2396                         $id = $this->mVariables[$varname];
2397                         $text = $this->getVariableValue( $id );
2398                         $this->mOutput->mContainsOldMagic = true;
2399                 } else {
2400                         $text = $matches[0];
2401                 }
2402                 wfProfileOut( $fname );
2403                 return $text;
2404         }
2405
2406         # Split template arguments
2407         function getTemplateArgs( $argsString ) {
2408                 if ( $argsString === '' ) {
2409                         return array();
2410                 }
2411
2412                 $args = explode( '|', substr( $argsString, 1 ) );
2413
2414                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2415                 # merged with the next arg because the '|' character between belongs
2416                 # to the link syntax and not the template parameter syntax.
2417                 $argc = count($args);
2418
2419                 for ( $i = 0; $i < $argc-1; $i++ ) {
2420                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2421                                 $args[$i] .= '|'.$args[$i+1];
2422                                 array_splice($args, $i+1, 1);
2423                                 $i--;
2424                                 $argc--;
2425                         }
2426                 }
2427
2428                 return $args;
2429         }
2430
2431         /**
2432          * Return the text of a template, after recursively
2433          * replacing any variables or templates within the template.
2434          *
2435          * @param array $piece The parts of the template
2436          *  $piece['text']: matched text
2437          *  $piece['title']: the title, i.e. the part before the |
2438          *  $piece['parts']: the parameter array
2439          * @return string the text of the template
2440          * @private
2441          */
2442         function braceSubstitution( $piece ) {
2443                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2444                 $fname = 'Parser::braceSubstitution';
2445                 wfProfileIn( $fname );
2446
2447                 # Flags
2448                 $found = false;             # $text has been filled
2449                 $nowiki = false;            # wiki markup in $text should be escaped
2450                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2451                 $noargs = false;            # Don't replace triple-brace arguments in $text
2452                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2453                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2454                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2455
2456                 # Title object, where $text came from
2457                 $title = NULL;
2458
2459                 $linestart = '';
2460
2461                 # $part1 is the bit before the first |, and must contain only title characters
2462                 # $args is a list of arguments, starting from index 0, not including $part1
2463
2464                 $part1 = $piece['title'];
2465                 # If the third subpattern matched anything, it will start with |
2466
2467                 if (null == $piece['parts']) {
2468                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2469                         if ($replaceWith != $piece['text']) {
2470                                 $text = $replaceWith;
2471                                 $found = true;
2472                                 $noparse = true;
2473                                 $noargs = true;
2474                         }
2475                 }
2476
2477                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2478                 $argc = count( $args );
2479
2480                 # SUBST
2481                 if ( !$found ) {
2482                         $mwSubst =& MagicWord::get( MAG_SUBST );
2483                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2484                                 # One of two possibilities is true:
2485                                 # 1) Found SUBST but not in the PST phase
2486                                 # 2) Didn't find SUBST and in the PST phase
2487                                 # In either case, return without further processing
2488                                 $text = $piece['text'];
2489                                 $found = true;
2490                                 $noparse = true;
2491                                 $noargs = true;
2492                         }
2493                 }
2494
2495                 # MSG, MSGNW, INT and RAW
2496                 if ( !$found ) {
2497                         # Check for MSGNW:
2498                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2499                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2500                                 $nowiki = true;
2501                         } else {
2502                                 # Remove obsolete MSG:
2503                                 $mwMsg =& MagicWord::get( MAG_MSG );
2504                                 $mwMsg->matchStartAndRemove( $part1 );
2505                         }
2506
2507                         # Check for RAW:
2508                         $mwRaw =& MagicWord::get( MAG_RAW );
2509                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2510                                 $forceRawInterwiki = true;
2511                         }
2512
2513                         # Check if it is an internal message
2514                         $mwInt =& MagicWord::get( MAG_INT );
2515                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2516                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2517                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2518                                         $found = true;
2519                                 }
2520                         }
2521                 }
2522
2523                 # NS
2524                 if ( !$found ) {
2525                         # Check for NS: (namespace expansion)
2526                         $mwNs = MagicWord::get( MAG_NS );
2527                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2528                                 if ( intval( $part1 ) || $part1 == "0" ) {
2529                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2530                                         $found = true;
2531                                 } else {
2532                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2533                                         if ( !is_null( $index ) ) {
2534                                                 $text = $linestart . $wgContLang->getNsText( $index );
2535                                                 $found = true;
2536                                         }
2537                                 }
2538                         }
2539                 }
2540
2541                 # URLENCODE
2542                 if( !$found ) {
2543                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2544                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2545                                 $text = $linestart . urlencode( $part1 );
2546                                 $found = true;
2547                         }
2548                 }
2549
2550                 # LCFIRST, UCFIRST, LC and UC
2551                 if ( !$found ) {
2552                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2553                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2554                         $lc =& MagicWord::get( MAG_LC );
2555                         $uc =& MagicWord::get( MAG_UC );
2556                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2557                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2558                                 $found = true;
2559                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2560                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2561                                 $found = true;
2562                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2563                                 $text = $linestart . $wgContLang->lc( $part1 );
2564                                 $found = true;
2565                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2566                                  $text = $linestart . $wgContLang->uc( $part1 );
2567                                  $found = true;
2568                         }
2569                 }
2570
2571                 # LOCALURL and FULLURL
2572                 if ( !$found ) {
2573                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2574                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2575                         $mwFull =& MagicWord::get( MAG_FULLURL );
2576                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2577
2578
2579                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2580                                 $func = 'getLocalURL';
2581                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2582                                 $func = 'escapeLocalURL';
2583                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2584                                 $func = 'getFullURL';
2585                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2586                                 $func = 'escapeFullURL';
2587                         } else {
2588                                 $func = false;
2589                         }
2590
2591                         if ( $func !== false ) {
2592                                 $title = Title::newFromText( $part1 );
2593                                 # Due to order of execution of a lot of bits, the values might be encoded
2594                                 # before arriving here; if that's true, then the title can't be created
2595                                 # and the variable will fail. If we can't get a decent title from the first
2596                                 # attempt, url-decode and try for a second.
2597                                 if( is_null( $title ) )
2598                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2599                                 if ( !is_null( $title ) ) {
2600                                         if ( $argc > 0 ) {
2601                                                 $text = $linestart . $title->$func( $args[0] );
2602                                         } else {
2603                                                 $text = $linestart . $title->$func();
2604                                         }
2605                                         $found = true;
2606                                 }
2607                         }
2608                 }
2609
2610                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2611                 # GRAMMAR
2612                 if ( !$found && $argc == 1 ) {
2613                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2614                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2615                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2616                                 $found = true;
2617                         }
2618                 }
2619
2620                 # PLURAL
2621                 if ( !$found && $argc >= 2 ) {
2622                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2623                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2624                                 if ($argc==2) {$args[2]=$args[1];}
2625                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2626                                 $found = true;
2627                         }
2628                 }
2629
2630                 # DISPLAYTITLE
2631                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2632                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2633                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2634
2635                                 # Set title in parser output object
2636                                 $param = $args[0];
2637                                 $parserOptions = new ParserOptions;
2638                                 $local_parser = new Parser ();
2639                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2640                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2641
2642                                 # Add subtitle
2643                                 $t = $this->mTitle->getPrefixedText();
2644                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2645                                 $text = "" ;
2646                                 $found = true ;
2647                         }
2648                 }
2649
2650                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2651                 if( !$found ) {
2652                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2653                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2654                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2655                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2656                         foreach( $mwWordsToCheck as $word => $func ) {
2657                                 $mwCurrentWord =& MagicWord::get( $word );
2658                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2659                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2660                                         if( $mwRawSuffix->match( $args[0] ) ) {
2661                                                 # Raw and unformatted
2662                                                 $text = $linestart . call_user_func( $func );
2663                                         } else {
2664                                                 # Formatted according to the content default
2665                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2666                                         }
2667                                         $found = true;
2668                                 }
2669                         }
2670                 }
2671
2672                 # #LANGUAGE:
2673                 if( !$found ) {
2674                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2675                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2676                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2677                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2678                                 $found = true;
2679                         }
2680                 }
2681
2682                 # Extensions
2683                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2684                         $colonPos = strpos( $part1, ':' );
2685                         if ( $colonPos !== false ) {
2686                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2687                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2688                                         $funcArgs = array_map( 'trim', $args );
2689                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2690                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2691                                         $found = true;
2692
2693                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2694                                         //$noargs = true;
2695                                         //$noparse = true;
2696
2697                                         if ( is_array( $result ) ) {
2698                                                 $text = $linestart . $result[0];
2699                                                 unset( $result[0] );
2700
2701                                                 // Extract flags into the local scope
2702                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2703                                                 extract( $result );
2704                                         } else {
2705                                                 $text = $linestart . $result;
2706                                         }
2707                                 }
2708                         }
2709                 }
2710
2711                 # Template table test
2712
2713                 # Did we encounter this template already? If yes, it is in the cache
2714                 # and we need to check for loops.
2715                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2716                         $found = true;
2717
2718                         # Infinite loop test
2719                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2720                                 $noparse = true;
2721                                 $noargs = true;
2722                                 $found = true;
2723                                 $text = $linestart .
2724                                         '{{' . $part1 . '}}' .
2725                                         '<!-- WARNING: template loop detected -->';
2726                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2727                         } else {
2728                                 # set $text to cached message.
2729                                 $text = $linestart . $this->mTemplates[$piece['title']];
2730                         }
2731                 }
2732
2733                 # Load from database
2734                 $lastPathLevel = $this->mTemplatePath;
2735                 if ( !$found ) {
2736                         $ns = NS_TEMPLATE;
2737                         # declaring $subpage directly in the function call
2738                         # does not work correctly with references and breaks
2739                         # {{/subpage}}-style inclusions
2740                         $subpage = '';
2741                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2742                         if ($subpage !== '') {
2743                                 $ns = $this->mTitle->getNamespace();
2744                         }
2745                         $title = Title::newFromText( $part1, $ns );
2746
2747                         if ( !is_null( $title ) ) {
2748                                 if ( !$title->isExternal() ) {
2749                                         # Check for excessive inclusion
2750                                         $dbk = $title->getPrefixedDBkey();
2751                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2752                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2753                                                         $text = SpecialPage::capturePath( $title );
2754                                                         if ( is_string( $text ) ) {
2755                                                                 $found = true;
2756                                                                 $noparse = true;
2757                                                                 $noargs = true;
2758                                                                 $isHTML = true;
2759                                                                 $this->disableCache();
2760                                                         }
2761                                                 } else {
2762                                                         $articleContent = $this->fetchTemplate( $title );
2763                                                         if ( $articleContent !== false ) {
2764                                                                 $found = true;
2765                                                                 $text = $articleContent;
2766                                                                 $replaceHeadings = true;
2767                                                         }
2768                                                 }
2769                                         }
2770
2771                                         # If the title is valid but undisplayable, make a link to it
2772                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2773                                                 $text = '[['.$title->getPrefixedText().']]';
2774                                                 $found = true;
2775                                         }
2776                                 } elseif ( $title->isTrans() ) {
2777                                         // Interwiki transclusion
2778                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2779                                                 $text = $this->interwikiTransclude( $title, 'render' );
2780                                                 $isHTML = true;
2781                                                 $noparse = true;
2782                                         } else {
2783                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2784                                                 $replaceHeadings = true;
2785                                         }
2786                                         $found = true;
2787                                 }
2788
2789                                 # Template cache array insertion
2790                                 # Use the original $piece['title'] not the mangled $part1, so that
2791                                 # modifiers such as RAW: produce separate cache entries
2792                                 if( $found ) {
2793                                         $this->mTemplates[$piece['title']] = $text;
2794                                         $text = $linestart . $text;
2795                                 }
2796                         }
2797                 }
2798
2799                 # Recursive parsing, escaping and link table handling
2800                 # Only for HTML output
2801                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2802                         $text = wfEscapeWikiText( $text );
2803                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2804                         if ( $noargs ) {
2805                                 $assocArgs = array();
2806                         } else {
2807                                 # Clean up argument array
2808                                 $assocArgs = array();
2809                                 $index = 1;
2810                                 foreach( $args as $arg ) {
2811                                         $eqpos = strpos( $arg, '=' );
2812                                         if ( $eqpos === false ) {
2813                                                 $assocArgs[$index++] = $arg;
2814                                         } else {
2815                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2816                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2817                                                 if ( $value === false ) {
2818                                                         $value = '';
2819                                                 }
2820                                                 if ( $name !== false ) {
2821                                                         $assocArgs[$name] = $value;
2822                                                 }
2823                                         }
2824                                 }
2825
2826                                 # Add a new element to the templace recursion path
2827                                 $this->mTemplatePath[$part1] = 1;
2828                         }
2829
2830                         if ( !$noparse ) {
2831                                 # If there are any <onlyinclude> tags, only include them
2832                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2833                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2834                                         $text = '';
2835                                         foreach ($m[1] as $piece)
2836                                                 $text .= $piece;
2837                                 }
2838                                 # Remove <noinclude> sections and <includeonly> tags
2839                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2840                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2841
2842                                 if( $this->mOutputType == OT_HTML ) {
2843                                         # Strip <nowiki>, <pre>, etc.
2844                                         $text = $this->strip( $text, $this->mStripState );
2845                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2846                                 }
2847                                 $text = $this->replaceVariables( $text, $assocArgs );
2848
2849                                 # If the template begins with a table or block-level
2850                                 # element, it should be treated as beginning a new line.
2851                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2852                                         $text = "\n" . $text;
2853                                 }
2854                         } elseif ( !$noargs ) {
2855                                 # $noparse and !$noargs
2856                                 # Just replace the arguments, not any double-brace items
2857                                 # This is used for rendered interwiki transclusion
2858                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2859                         }
2860                 }
2861                 # Prune lower levels off the recursion check path
2862                 $this->mTemplatePath = $lastPathLevel;
2863
2864                 if ( !$found ) {
2865                         wfProfileOut( $fname );
2866                         return $piece['text'];
2867                 } else {
2868                         if ( $isHTML ) {
2869                                 # Replace raw HTML by a placeholder
2870                                 # Add a blank line preceding, to prevent it from mucking up
2871                                 # immediately preceding headings
2872                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2873                         } else {
2874                                 # replace ==section headers==
2875                                 # XXX this needs to go away once we have a better parser.
2876                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2877                                         if( !is_null( $title ) )
2878                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2879                                         else
2880                                                 $encodedname = base64_encode("");
2881                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2882                                                 PREG_SPLIT_DELIM_CAPTURE);
2883                                         $text = '';
2884                                         $nsec = 0;
2885                                         for( $i = 0; $i < count($m); $i += 2 ) {
2886                                                 $text .= $m[$i];
2887                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2888                                                 $hl = $m[$i + 1];
2889                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2890                                                         $text .= $hl;
2891                                                         continue;
2892                                                 }
2893                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2894                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2895                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2896
2897                                                 $nsec++;
2898                                         }
2899                                 }
2900                         }
2901                 }
2902
2903                 # Prune lower levels off the recursion check path
2904                 $this->mTemplatePath = $lastPathLevel;
2905
2906                 if ( !$found ) {
2907                         wfProfileOut( $fname );
2908                         return $piece['text'];
2909                 } else {
2910                         wfProfileOut( $fname );
2911                         return $text;
2912                 }
2913         }
2914
2915         /**
2916          * Fetch the unparsed text of a template and register a reference to it.
2917          */
2918         function fetchTemplate( $title ) {
2919                 $text = false;
2920                 // Loop to fetch the article, with up to 1 redirect
2921                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2922                         $rev = Revision::newFromTitle( $title );
2923                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2924                         if ( !$rev ) {
2925                                 break;
2926                         }
2927                         $text = $rev->getText();
2928                         if ( $text === false ) {
2929                                 break;
2930                         }
2931                         // Redirect?
2932                         $title = Title::newFromRedirect( $text );
2933                 }
2934                 return $text;
2935         }
2936
2937         /**
2938          * Transclude an interwiki link.
2939          */
2940         function interwikiTransclude( $title, $action ) {
2941                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2942
2943                 if (!$wgEnableScaryTranscluding)
2944                         return wfMsg('scarytranscludedisabled');
2945
2946                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2947                 // But we'll handle it generally anyway
2948                 if ( $title->getNamespace() ) {
2949                         // Use the canonical namespace, which should work anywhere
2950                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2951                 } else {
2952                         $articleName = $title->getDBkey();
2953                 }
2954
2955                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2956                 $url .= "?action=$action";
2957                 if (strlen($url) > 255)
2958                         return wfMsg('scarytranscludetoolong');
2959                 return $this->fetchScaryTemplateMaybeFromCache($url);
2960         }
2961
2962         function fetchScaryTemplateMaybeFromCache($url) {
2963                 global $wgTranscludeCacheExpiry;
2964                 $dbr =& wfGetDB(DB_SLAVE);
2965                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2966                                 array('tc_url' => $url));
2967                 if ($obj) {
2968                         $time = $obj->tc_time;
2969                         $text = $obj->tc_contents;
2970                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2971                                 return $text;
2972                         }
2973                 }
2974
2975                 $text = wfGetHTTP($url);
2976                 if (!$text)
2977                         return wfMsg('scarytranscludefailed', $url);
2978
2979                 $dbw =& wfGetDB(DB_MASTER);
2980                 $dbw->replace('transcache', array('tc_url'), array(
2981                         'tc_url' => $url,
2982                         'tc_time' => time(),
2983                         'tc_contents' => $text));
2984                 return $text;
2985         }
2986
2987
2988         /**
2989          * Triple brace replacement -- used for template arguments
2990          * @private
2991          */
2992         function argSubstitution( $matches ) {
2993                 $arg = trim( $matches['title'] );
2994                 $text = $matches['text'];
2995                 $inputArgs = end( $this->mArgStack );
2996
2997                 if ( array_key_exists( $arg, $inputArgs ) ) {
2998                         $text = $inputArgs[$arg];
2999                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3000                         $text = $matches['parts'][0];
3001                 }
3002
3003                 return $text;
3004         }
3005
3006         /**
3007          * Returns true if the function is allowed to include this entity
3008          * @private
3009          */
3010         function incrementIncludeCount( $dbk ) {
3011                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3012                         $this->mIncludeCount[$dbk] = 0;
3013                 }
3014                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3015                         return true;
3016                 } else {
3017                         return false;
3018                 }
3019         }
3020
3021         /**
3022          * Detect __TOC__ magic word and set a placeholder
3023          */
3024         function stripToc( $text ) {
3025                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3026                 # do not add TOC
3027                 $mw = MagicWord::get( MAG_NOTOC );
3028                 if( $mw->matchAndRemove( $text ) ) {
3029                         $this->mShowToc = false;
3030                 }
3031
3032                 $mw = MagicWord::get( MAG_TOC );
3033                 if( $mw->match( $text ) ) {
3034                         $this->mShowToc = true;
3035                         $this->mForceTocPosition = true;
3036
3037                         // Set a placeholder. At the end we'll fill it in with the TOC.
3038                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3039
3040                         // Only keep the first one.
3041                         $text = $mw->replace( '', $text );
3042                 }
3043                 return $text;
3044         }
3045
3046         /**
3047          * This function accomplishes several tasks:
3048          * 1) Auto-number headings if that option is enabled
3049          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3050          * 3) Add a Table of contents on the top for users who have enabled the option
3051          * 4) Auto-anchor headings
3052          *
3053          * It loops through all headlines, collects the necessary data, then splits up the
3054          * string and re-inserts the newly formatted headlines.
3055          *
3056          * @param string $text
3057          * @param boolean $isMain
3058          * @private
3059          */
3060         function formatHeadings( $text, $isMain=true ) {
3061                 global $wgMaxTocLevel, $wgContLang;
3062
3063                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3064                 if( !$this->mTitle->userCanEdit() ) {
3065                         $showEditLink = 0;
3066                 } else {
3067                         $showEditLink = $this->mOptions->getEditSection();
3068                 }
3069
3070                 # Inhibit editsection links if requested in the page
3071                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3072                 if( $esw->matchAndRemove( $text ) ) {
3073                         $showEditLink = 0;
3074                 }
3075
3076                 # Get all headlines for numbering them and adding funky stuff like [edit]
3077                 # links - this is for later, but we need the number of headlines right now
3078                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3079
3080                 # if there are fewer than 4 headlines in the article, do not show TOC
3081                 # unless it's been explicitly enabled.
3082                 $enoughToc = $this->mShowToc &&
3083                         (($numMatches >= 4) || $this->mForceTocPosition);
3084
3085                 # Allow user to stipulate that a page should have a "new section"
3086                 # link added via __NEWSECTIONLINK__
3087                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3088                 if( $mw->matchAndRemove( $text ) )
3089                         $this->mOutput->setNewSection( true );
3090
3091                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3092                 # override above conditions and always show TOC above first header
3093                 $mw =& MagicWord::get( MAG_FORCETOC );
3094                 if ($mw->matchAndRemove( $text ) ) {
3095                         $this->mShowToc = true;
3096                         $enoughToc = true;
3097                 }
3098
3099                 # Never ever show TOC if no headers
3100                 if( $numMatches < 1 ) {
3101                         $enoughToc = false;
3102                 }
3103
3104                 # We need this to perform operations on the HTML
3105                 $sk =& $this->mOptions->getSkin();
3106
3107                 # headline counter
3108                 $headlineCount = 0;
3109                 $sectionCount = 0; # headlineCount excluding template sections
3110
3111                 # Ugh .. the TOC should have neat indentation levels which can be
3112                 # passed to the skin functions. These are determined here
3113                 $toc = '';
3114                 $full = '';
3115                 $head = array();
3116                 $sublevelCount = array();
3117                 $levelCount = array();
3118                 $toclevel = 0;
3119                 $level = 0;
3120                 $prevlevel = 0;
3121                 $toclevel = 0;
3122                 $prevtoclevel = 0;
3123
3124                 foreach( $matches[3] as $headline ) {
3125                         $istemplate = 0;
3126                         $templatetitle = '';
3127                         $templatesection = 0;
3128                         $numbering = '';
3129
3130                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3131                                 $istemplate = 1;
3132                                 $templatetitle = base64_decode($mat[1]);
3133                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3134                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3135                         }
3136
3137                         if( $toclevel ) {
3138                                 $prevlevel = $level;
3139                                 $prevtoclevel = $toclevel;
3140                         }
3141                         $level = $matches[1][$headlineCount];
3142
3143                         if( $doNumberHeadings || $enoughToc ) {
3144
3145                                 if ( $level > $prevlevel ) {
3146                                         # Increase TOC level
3147                                         $toclevel++;
3148                                         $sublevelCount[$toclevel] = 0;
3149                                         if( $toclevel<$wgMaxTocLevel ) {
3150                                                 $toc .= $sk->tocIndent();
3151                                         }
3152                                 }
3153                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3154                                         # Decrease TOC level, find level to jump to
3155
3156                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3157                                                 # Can only go down to level 1
3158                                                 $toclevel = 1;
3159                                         } else {
3160                                                 for ($i = $toclevel; $i > 0; $i--) {
3161                                                         if ( $levelCount[$i] == $level ) {
3162                                                                 # Found last matching level
3163                                                                 $toclevel = $i;
3164                                                                 break;
3165                                                         }
3166                                                         elseif ( $levelCount[$i] < $level ) {
3167                                                                 # Found first matching level below current level
3168                                                                 $toclevel = $i + 1;
3169                                                                 break;
3170                                                         }
3171                                                 }
3172                                         }
3173                                         if( $toclevel<$wgMaxTocLevel ) {
3174                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3175                                         }
3176                                 }
3177                                 else {
3178                                         # No change in level, end TOC line
3179                                         if( $toclevel<$wgMaxTocLevel ) {
3180                                                 $toc .= $sk->tocLineEnd();
3181                                         }
3182                                 }
3183
3184                                 $levelCount[$toclevel] = $level;
3185
3186                                 # count number of headlines for each level
3187                                 @$sublevelCount[$toclevel]++;
3188                                 $dot = 0;
3189                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3190                                         if( !empty( $sublevelCount[$i] ) ) {
3191                                                 if( $dot ) {
3192                                                         $numbering .= '.';
3193                                                 }
3194                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3195                                                 $dot = 1;
3196                                         }
3197                                 }
3198                         }
3199
3200                         # The canonized header is a version of the header text safe to use for links
3201                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3202                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3203                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3204
3205                         # Remove link placeholders by the link text.
3206                         #     <!--LINK number-->
3207                         # turns into
3208                         #     link text with suffix
3209                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3210                                                             "\$this->mLinkHolders['texts'][\$1]",
3211                                                             $canonized_headline );
3212                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3213                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3214                                                             $canonized_headline );
3215
3216                         # strip out HTML
3217                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3218                         $tocline = trim( $canonized_headline );
3219                         # Save headline for section edit hint before it's escaped
3220                         $headline_hint = trim( $canonized_headline );
3221                         $canonized_headline = Sanitizer::escapeId( $tocline );
3222                         $refers[$headlineCount] = $canonized_headline;
3223
3224                         # count how many in assoc. array so we can track dupes in anchors
3225                         @$refers[$canonized_headline]++;
3226                         $refcount[$headlineCount]=$refers[$canonized_headline];
3227
3228                         # Don't number the heading if it is the only one (looks silly)
3229                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3230                                 # the two are different if the line contains a link
3231                                 $headline=$numbering . ' ' . $headline;
3232                         }
3233
3234                         # Create the anchor for linking from the TOC to the section
3235                         $anchor = $canonized_headline;
3236                         if($refcount[$headlineCount] > 1 ) {
3237                                 $anchor .= '_' . $refcount[$headlineCount];
3238                         }
3239                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3240                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3241                         }
3242                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3243                                 if ( empty( $head[$headlineCount] ) ) {
3244                                         $head[$headlineCount] = '';
3245                                 }
3246                                 if( $istemplate )
3247                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3248                                 else
3249                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3250                         }
3251
3252                         # give headline the correct <h#> tag
3253                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3254
3255                         $headlineCount++;
3256                         if( !$istemplate )
3257                                 $sectionCount++;
3258                 }
3259
3260                 if( $enoughToc ) {
3261                         if( $toclevel<$wgMaxTocLevel ) {
3262                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3263                         }
3264                         $toc = $sk->tocList( $toc );
3265                 }
3266
3267                 # split up and insert constructed headlines
3268
3269                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3270                 $i = 0;
3271
3272                 foreach( $blocks as $block ) {
3273                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3274                                 # This is the [edit] link that appears for the top block of text when
3275                                 # section editing is enabled
3276
3277                                 # Disabled because it broke block formatting
3278                                 # For example, a bullet point in the top line
3279                                 # $full .= $sk->editSectionLink(0);
3280                         }
3281                         $full .= $block;
3282                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3283                                 # Top anchor now in skin
3284                                 $full = $full.$toc;
3285                         }
3286
3287                         if( !empty( $head[$i] ) ) {
3288                                 $full .= $head[$i];
3289                         }
3290                         $i++;
3291                 }
3292                 if( $this->mForceTocPosition ) {
3293                         return str_replace( '<!--MWTOC-->', $toc, $full );
3294                 } else {
3295                         return $full;
3296                 }
3297         }
3298
3299         /**
3300          * Return an HTML link for the "ISBN 123456" text
3301          * @private
3302          */
3303         function magicISBN( $text ) {
3304                 $fname = 'Parser::magicISBN';
3305                 wfProfileIn( $fname );
3306
3307                 $a = split( 'ISBN ', ' '.$text );
3308                 if ( count ( $a ) < 2 ) {
3309                         wfProfileOut( $fname );
3310                         return $text;
3311                 }
3312                 $text = substr( array_shift( $a ), 1);
3313                 $valid = '0123456789-Xx';
3314
3315                 foreach ( $a as $x ) {
3316                         # hack: don't replace inside thumbnail title/alt
3317                         # attributes
3318                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3319                                 $text .= "ISBN $x";
3320                                 continue;
3321                         }
3322
3323                         $isbn = $blank = '' ;
3324                         while ( ' ' == $x{0} ) {
3325                                 $blank .= ' ';
3326                                 $x = substr( $x, 1 );
3327                         }
3328                         if ( $x == '' ) { # blank isbn
3329                                 $text .= "ISBN $blank";
3330                                 continue;
3331                         }
3332                         while ( strstr( $valid, $x{0} ) != false ) {
3333                                 $isbn .= $x{0};
3334                                 $x = substr( $x, 1 );
3335                         }
3336                         $num = str_replace( '-', '', $isbn );
3337                         $num = str_replace( ' ', '', $num );
3338                         $num = str_replace( 'x', 'X', $num );
3339
3340                         if ( '' == $num ) {
3341                                 $text .= "ISBN $blank$x";
3342                         } else {
3343                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3344                                 $text .= '<a href="' .
3345                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3346                                         "\" class=\"internal\">ISBN $isbn</a>";
3347                                 $text .= $x;
3348                         }
3349                 }
3350                 wfProfileOut( $fname );
3351                 return $text;
3352         }
3353
3354         /**
3355          * Return an HTML link for the "RFC 1234" text
3356          *
3357          * @private
3358          * @param string $text     Text to be processed
3359          * @param string $keyword  Magic keyword to use (default RFC)
3360          * @param string $urlmsg   Interface message to use (default rfcurl)
3361          * @return string
3362          */
3363         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3364
3365                 $valid = '0123456789';
3366                 $internal = false;
3367
3368                 $a = split( $keyword, ' '.$text );
3369                 if ( count ( $a ) < 2 ) {
3370                         return $text;
3371                 }
3372                 $text = substr( array_shift( $a ), 1);
3373
3374                 /* Check if keyword is preceed by [[.
3375                  * This test is made here cause of the array_shift above
3376                  * that prevent the test to be done in the foreach.
3377                  */
3378                 if ( substr( $text, -2 ) == '[[' ) {
3379                         $internal = true;
3380                 }
3381
3382                 foreach ( $a as $x ) {
3383                         /* token might be empty if we have RFC RFC 1234 */
3384                         if ( $x=='' ) {
3385                                 $text.=$keyword;
3386                                 continue;
3387                                 }
3388
3389                         # hack: don't replace inside thumbnail title/alt
3390                         # attributes
3391                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3392                                 $text .= $keyword . $x;
3393                                 continue;
3394                         }
3395
3396                         $id = $blank = '' ;
3397
3398                         /** remove and save whitespaces in $blank */
3399                         while ( $x{0} == ' ' ) {
3400                                 $blank .= ' ';
3401                                 $x = substr( $x, 1 );
3402                         }
3403
3404                         /** remove and save the rfc number in $id */
3405                         while ( strstr( $valid, $x{0} ) != false ) {
3406                                 $id .= $x{0};
3407                                 $x = substr( $x, 1 );
3408                         }
3409
3410                         if ( $id == '' ) {
3411                                 /* call back stripped spaces*/
3412                                 $text .= $keyword.$blank.$x;
3413                         } elseif( $internal ) {
3414                                 /* normal link */
3415                                 $text .= $keyword.$id.$x;
3416                         } else {
3417                                 /* build the external link*/
3418                                 $url = wfMsg( $urlmsg, $id);
3419                                 $sk =& $this->mOptions->getSkin();
3420                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3421                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3422                         }
3423
3424                         /* Check if the next RFC keyword is preceed by [[ */
3425                         $internal = ( substr($x,-2) == '[[' );
3426                 }
3427                 return $text;
3428         }
3429
3430         /**
3431          * Transform wiki markup when saving a page by doing \r\n -> \n
3432          * conversion, substitting signatures, {{subst:}} templates, etc.
3433          *
3434          * @param string $text the text to transform
3435          * @param Title &$title the Title object for the current article
3436          * @param User &$user the User object describing the current user
3437          * @param ParserOptions $options parsing options
3438          * @param bool $clearState whether to clear the parser state first
3439          * @return string the altered wiki markup
3440          * @public
3441          */
3442         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3443                 $this->mOptions = $options;
3444                 $this->mTitle =& $title;
3445                 $this->mOutputType = OT_WIKI;
3446
3447                 if ( $clearState ) {
3448                         $this->clearState();
3449                 }
3450
3451                 $stripState = false;
3452                 $pairs = array(
3453                         "\r\n" => "\n",
3454                 );
3455                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3456                 $text = $this->strip( $text, $stripState, true );
3457                 $text = $this->pstPass2( $text, $stripState, $user );
3458                 $text = $this->unstrip( $text, $stripState );
3459                 $text = $this->unstripNoWiki( $text, $stripState );
3460                 return $text;
3461         }
3462
3463         /**
3464          * Pre-save transform helper function
3465          * @private
3466          */
3467         function pstPass2( $text, &$stripState, &$user ) {
3468                 global $wgContLang, $wgLocaltimezone;
3469
3470                 /* Note: This is the timestamp saved as hardcoded wikitext to
3471                  * the database, we use $wgContLang here in order to give
3472                  * everyone the same signature and use the default one rather
3473                  * than the one selected in each user's preferences.
3474                  */
3475                 if ( isset( $wgLocaltimezone ) ) {
3476                         $oldtz = getenv( 'TZ' );
3477                         putenv( 'TZ='.$wgLocaltimezone );
3478                 }
3479                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3480                   ' (' . date( 'T' ) . ')';
3481                 if ( isset( $wgLocaltimezone ) ) {
3482                         putenv( 'TZ='.$oldtz );
3483                 }
3484
3485                 # Variable replacement
3486                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3487                 $text = $this->replaceVariables( $text );
3488
3489                 # Strip out <nowiki> etc. added via replaceVariables
3490                 $text = $this->strip( $text, $stripState );
3491
3492                 # Signatures
3493                 $sigText = $this->getUserSig( $user );
3494                 $text = strtr( $text, array(
3495                         '~~~~~' => $d,
3496                         '~~~~' => "$sigText $d",
3497                         '~~~' => $sigText
3498                 ) );
3499
3500                 # Context links: [[|name]] and [[name (context)|]]
3501                 #
3502                 global $wgLegalTitleChars;
3503                 $tc = "[$wgLegalTitleChars]";
3504                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3505
3506                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3507                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3508
3509                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3510                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3511                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3512                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3513                 $context = '';
3514                 $t = $this->mTitle->getText();
3515                 if ( preg_match( $conpat, $t, $m ) ) {
3516                         $context = $m[2];
3517                 }
3518                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3519                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3520                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3521
3522                 if ( '' == $context ) {
3523                         $text = preg_replace( $p2, '[[\\1]]', $text );
3524                 } else {
3525                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3526                 }
3527
3528                 # Trim trailing whitespace
3529                 # MAG_END (__END__) tag allows for trailing
3530                 # whitespace to be deliberately included
3531                 $text = rtrim( $text );
3532                 $mw =& MagicWord::get( MAG_END );
3533                 $mw->matchAndRemove( $text );
3534
3535                 return $text;
3536         }
3537
3538         /**
3539          * Fetch the user's signature text, if any, and normalize to
3540          * validated, ready-to-insert wikitext.
3541          *
3542          * @param User $user
3543          * @return string
3544          * @private
3545          */
3546         function getUserSig( &$user ) {
3547                 $username = $user->getName();
3548                 $nickname = $user->getOption( 'nickname' );
3549                 $nickname = $nickname === '' ? $username : $nickname;
3550
3551                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3552                         # Sig. might contain markup; validate this
3553                         if( $this->validateSig( $nickname ) !== false ) {
3554                                 # Validated; clean up (if needed) and return it
3555                                 return $this->cleanSig( $nickname, true );
3556                         } else {
3557                                 # Failed to validate; fall back to the default
3558                                 $nickname = $username;
3559                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3560                         }
3561                 }
3562
3563                 # If we're still here, make it a link to the user page
3564                 $userpage = $user->getUserPage();
3565                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3566         }
3567
3568         /**
3569          * Check that the user's signature contains no bad XML
3570          *
3571          * @param string $text
3572          * @return mixed An expanded string, or false if invalid.
3573          */
3574         function validateSig( $text ) {
3575                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3576         }
3577
3578         /**
3579          * Clean up signature text
3580          *
3581          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3582          * 2) Substitute all transclusions
3583          *
3584          * @param string $text
3585          * @param $parsing Whether we're cleaning (preferences save) or parsing
3586          * @return string Signature text
3587          */
3588         function cleanSig( $text, $parsing = false ) {
3589                 global $wgTitle;
3590                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3591
3592                 $substWord = MagicWord::get( MAG_SUBST );
3593                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3594                 $substText = '{{' . $substWord->getSynonym( 0 );
3595
3596                 $text = preg_replace( $substRegex, $substText, $text );
3597                 $text = preg_replace( '/~{3,5}/', '', $text );
3598                 $text = $this->replaceVariables( $text );
3599
3600                 $this->clearState();
3601                 return $text;
3602         }
3603
3604         /**
3605          * Set up some variables which are usually set up in parse()
3606          * so that an external function can call some class members with confidence
3607          * @public
3608          */
3609         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3610                 $this->mTitle =& $title;
3611                 $this->mOptions = $options;
3612                 $this->mOutputType = $outputType;
3613                 if ( $clearState ) {
3614                         $this->clearState();
3615                 }
3616         }
3617
3618         /**
3619          * Transform a MediaWiki message by replacing magic variables.
3620          *
3621          * @param string $text the text to transform
3622          * @param ParserOptions $options  options
3623          * @return string the text with variables substituted
3624          * @public
3625          */
3626         function transformMsg( $text, $options ) {
3627                 global $wgTitle;
3628                 static $executing = false;
3629
3630                 $fname = "Parser::transformMsg";
3631
3632                 # Guard against infinite recursion
3633                 if ( $executing ) {
3634                         return $text;
3635                 }
3636                 $executing = true;
3637
3638                 wfProfileIn($fname);
3639
3640                 $this->mTitle = $wgTitle;
3641                 $this->mOptions = $options;
3642                 $this->mOutputType = OT_MSG;
3643                 $this->clearState();
3644                 $text = $this->replaceVariables( $text );
3645
3646                 $executing = false;
3647                 wfProfileOut($fname);
3648                 return $text;
3649         }
3650
3651         /**
3652          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3653          * The callback should have the following form:
3654          *    function myParserHook( $text, $params, &$parser ) { ... }
3655          *
3656          * Transform and return $text. Use $parser for any required context, e.g. use
3657          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3658          *
3659          * @public
3660          *
3661          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3662          * @param mixed $callback The callback function (and object) to use for the tag
3663          *
3664          * @return The old value of the mTagHooks array associated with the hook
3665          */
3666         function setHook( $tag, $callback ) {
3667                 $tag = strtolower( $tag );
3668                 $oldVal = @$this->mTagHooks[$tag];
3669                 $this->mTagHooks[$tag] = $callback;
3670
3671                 return $oldVal;
3672         }
3673
3674         /**
3675          * Create a function, e.g. {{sum:1|2|3}}
3676          * The callback function should have the form:
3677          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3678          *
3679          * The callback may either return the text result of the function, or an array with the text
3680          * in element 0, and a number of flags in the other elements. The names of the flags are
3681          * specified in the keys. Valid flags are:
3682          *   found                     The text returned is valid, stop processing the template. This
3683          *                             is on by default.
3684          *   nowiki                    Wiki markup in the return value should be escaped
3685          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3686          *   noargs                    Don't replace triple-brace arguments in the return value
3687          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3688          *
3689          * @public
3690          *
3691          * @param string $name The function name. Function names are case-insensitive.
3692          * @param mixed $callback The callback function (and object) to use
3693          *
3694          * @return The old callback function for this name, if any
3695          */
3696         function setFunctionHook( $name, $callback ) {
3697                 $name = strtolower( $name );
3698                 $oldVal = @$this->mFunctionHooks[$name];
3699                 $this->mFunctionHooks[$name] = $callback;
3700                 return $oldVal;
3701         }
3702
3703         /**
3704          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3705          * Placeholders created in Skin::makeLinkObj()
3706          * Returns an array of links found, indexed by PDBK:
3707          *  0 - broken
3708          *  1 - normal link
3709          *  2 - stub
3710          * $options is a bit field, RLH_FOR_UPDATE to select for update
3711          */
3712         function replaceLinkHolders( &$text, $options = 0 ) {
3713                 global $wgUser;
3714                 global $wgOutputReplace;
3715
3716                 $fname = 'Parser::replaceLinkHolders';
3717                 wfProfileIn( $fname );
3718
3719                 $pdbks = array();
3720                 $colours = array();
3721                 $sk =& $this->mOptions->getSkin();
3722                 $linkCache =& LinkCache::singleton();
3723
3724                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3725                         wfProfileIn( $fname.'-check' );
3726                         $dbr =& wfGetDB( DB_SLAVE );
3727                         $page = $dbr->tableName( 'page' );
3728                         $threshold = $wgUser->getOption('stubthreshold');
3729
3730                         # Sort by namespace
3731                         asort( $this->mLinkHolders['namespaces'] );
3732
3733                         # Generate query
3734                         $query = false;
3735                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3736                                 # Make title object
3737                                 $title = $this->mLinkHolders['titles'][$key];
3738
3739                                 # Skip invalid entries.
3740                                 # Result will be ugly, but prevents crash.
3741                                 if ( is_null( $title ) ) {
3742                                         continue;
3743                                 }
3744                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3745
3746                                 # Check if it's a static known link, e.g. interwiki
3747                                 if ( $title->isAlwaysKnown() ) {
3748                                         $colours[$pdbk] = 1;
3749                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3750                                         $colours[$pdbk] = 1;
3751                                         $this->mOutput->addLink( $title, $id );
3752                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3753                                         $colours[$pdbk] = 0;
3754                                 } else {
3755                                         # Not in the link cache, add it to the query
3756                                         if ( !isset( $current ) ) {
3757                                                 $current = $ns;
3758                                                 $query =  "SELECT page_id, page_namespace, page_title";
3759                                                 if ( $threshold > 0 ) {
3760                                                         $query .= ', page_len, page_is_redirect';
3761                                                 }
3762                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3763                                         } elseif ( $current != $ns ) {
3764                                                 $current = $ns;
3765                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3766                                         } else {
3767                                                 $query .= ', ';
3768                                         }
3769
3770                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3771                                 }
3772                         }
3773                         if ( $query ) {
3774                                 $query .= '))';
3775                                 if ( $options & RLH_FOR_UPDATE ) {
3776                                         $query .= ' FOR UPDATE';
3777                                 }
3778
3779                                 $res = $dbr->query( $query, $fname );
3780
3781                                 # Fetch data and form into an associative array
3782                                 # non-existent = broken
3783                                 # 1 = known
3784                                 # 2 = stub
3785                                 while ( $s = $dbr->fetchObject($res) ) {
3786                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3787                                         $pdbk = $title->getPrefixedDBkey();
3788                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3789                                         $this->mOutput->addLink( $title, $s->page_id );
3790
3791                                         if ( $threshold >  0 ) {
3792                                                 $size = $s->page_len;
3793                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3794                                                         $colours[$pdbk] = 1;
3795                                                 } else {
3796                                                         $colours[$pdbk] = 2;
3797                                                 }
3798                                         } else {
3799                                                 $colours[$pdbk] = 1;
3800                                         }
3801                                 }
3802                         }
3803                         wfProfileOut( $fname.'-check' );
3804
3805                         # Construct search and replace arrays
3806                         wfProfileIn( $fname.'-construct' );
3807                         $wgOutputReplace = array();
3808                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3809                                 $pdbk = $pdbks[$key];
3810                                 $searchkey = "<!--LINK $key-->";
3811                                 $title = $this->mLinkHolders['titles'][$key];
3812                                 if ( empty( $colours[$pdbk] ) ) {
3813                                         $linkCache->addBadLinkObj( $title );
3814                                         $colours[$pdbk] = 0;
3815                                         $this->mOutput->addLink( $title, 0 );
3816                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3817                                                                         $this->mLinkHolders['texts'][$key],
3818                                                                         $this->mLinkHolders['queries'][$key] );
3819                                 } elseif ( $colours[$pdbk] == 1 ) {
3820                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3821                                                                         $this->mLinkHolders['texts'][$key],
3822                                                                         $this->mLinkHolders['queries'][$key] );
3823                                 } elseif ( $colours[$pdbk] == 2 ) {
3824                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3825                                                                         $this->mLinkHolders['texts'][$key],
3826                                                                         $this->mLinkHolders['queries'][$key] );
3827                                 }
3828                         }
3829                         wfProfileOut( $fname.'-construct' );
3830
3831                         # Do the thing
3832                         wfProfileIn( $fname.'-replace' );
3833
3834                         $text = preg_replace_callback(
3835                                 '/(<!--LINK .*?-->)/',
3836                                 "wfOutputReplaceMatches",
3837                                 $text);
3838
3839                         wfProfileOut( $fname.'-replace' );
3840                 }
3841
3842                 # Now process interwiki link holders
3843                 # This is quite a bit simpler than internal links
3844                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3845                         wfProfileIn( $fname.'-interwiki' );
3846                         # Make interwiki link HTML
3847                         $wgOutputReplace = array();
3848                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3849                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3850                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3851                         }
3852
3853                         $text = preg_replace_callback(
3854                                 '/<!--IWLINK (.*?)-->/',
3855                                 "wfOutputReplaceMatches",
3856                                 $text );
3857                         wfProfileOut( $fname.'-interwiki' );
3858                 }
3859
3860                 wfProfileOut( $fname );
3861                 return $colours;
3862         }
3863
3864         /**
3865          * Replace <!--LINK--> link placeholders with plain text of links
3866          * (not HTML-formatted).
3867          * @param string $text
3868          * @return string
3869          */
3870         function replaceLinkHoldersText( $text ) {
3871                 $fname = 'Parser::replaceLinkHoldersText';
3872                 wfProfileIn( $fname );
3873
3874                 $text = preg_replace_callback(
3875                         '/<!--(LINK|IWLINK) (.*?)-->/',
3876                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3877                         $text );
3878
3879                 wfProfileOut( $fname );
3880                 return $text;
3881         }
3882
3883         /**
3884          * @param array $matches
3885          * @return string
3886          * @private
3887          */
3888         function replaceLinkHoldersTextCallback( $matches ) {
3889                 $type = $matches[1];
3890                 $key  = $matches[2];
3891                 if( $type == 'LINK' ) {
3892                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3893                                 return $this->mLinkHolders['texts'][$key];
3894                         }
3895                 } elseif( $type == 'IWLINK' ) {
3896                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3897                                 return $this->mInterwikiLinkHolders['texts'][$key];
3898                         }
3899                 }
3900                 return $matches[0];
3901         }
3902
3903         /**
3904          * Renders an image gallery from a text with one line per image.
3905          * text labels may be given by using |-style alternative text. E.g.
3906          *   Image:one.jpg|The number "1"
3907          *   Image:tree.jpg|A tree
3908          * given as text will return the HTML of a gallery with two images,
3909          * labeled 'The number "1"' and
3910          * 'A tree'.
3911          */
3912         function renderImageGallery( $text ) {
3913                 # Setup the parser
3914                 $parserOptions = new ParserOptions;
3915                 $localParser = new Parser();
3916
3917                 $ig = new ImageGallery();
3918                 $ig->setShowBytes( false );
3919                 $ig->setShowFilename( false );
3920                 $ig->setParsing();
3921                 $lines = explode( "\n", $text );
3922
3923                 foreach ( $lines as $line ) {
3924                         # match lines like these:
3925                         # Image:someimage.jpg|This is some image
3926                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3927                         # Skip empty lines
3928                         if ( count( $matches ) == 0 ) {
3929                                 continue;
3930                         }
3931                         $nt =& Title::newFromText( $matches[1] );
3932                         if( is_null( $nt ) ) {
3933                                 # Bogus title. Ignore these so we don't bomb out later.
3934                                 continue;
3935                         }
3936                         if ( isset( $matches[3] ) ) {
3937                                 $label = $matches[3];
3938                         } else {
3939                                 $label = '';
3940                         }
3941
3942                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3943                         $html = $pout->getText();
3944
3945                         $ig->add( new Image( $nt ), $html );
3946
3947                         # Only add real images (bug #5586)
3948                         if ( $nt->getNamespace() == NS_IMAGE ) {
3949                                 $this->mOutput->addImage( $nt->getDBkey() );
3950                         }
3951
3952                         # Register links with the parent parser
3953                         foreach( $pout->getLinks() as $ns => $keys ) {
3954                                 foreach( $keys as $dbk => $id )
3955                                         $this->mOutput->addLink( Title::makeTitle( $ns, $dbk ), $id );
3956                         }
3957
3958                 }
3959                 return $ig->toHTML();
3960         }
3961
3962         /**
3963          * Parse image options text and use it to make an image
3964          */
3965         function makeImage( &$nt, $options ) {
3966                 global $wgUseImageResize;
3967
3968                 $align = '';
3969
3970                 # Check if the options text is of the form "options|alt text"
3971                 # Options are:
3972                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3973                 #  * left               no resizing, just left align. label is used for alt= only
3974                 #  * right              same, but right aligned
3975                 #  * none               same, but not aligned
3976                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3977                 #  * center             center the image
3978                 #  * framed             Keep original image size, no magnify-button.
3979
3980                 $part = explode( '|', $options);
3981
3982                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3983                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3984                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3985                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3986                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3987                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3988                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3989                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3990                 $caption = '';
3991
3992                 $width = $height = $framed = $thumb = false;
3993                 $manual_thumb = '' ;
3994
3995                 foreach( $part as $key => $val ) {
3996                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3997                                 $thumb=true;
3998                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3999                                 # use manually specified thumbnail
4000                                 $thumb=true;
4001                                 $manual_thumb = $match;
4002                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4003                                 # remember to set an alignment, don't render immediately
4004                                 $align = 'right';
4005                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4006                                 # remember to set an alignment, don't render immediately
4007                                 $align = 'left';
4008                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4009                                 # remember to set an alignment, don't render immediately
4010                                 $align = 'center';
4011                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4012                                 # remember to set an alignment, don't render immediately
4013                                 $align = 'none';
4014                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4015                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4016                                 # $match is the image width in pixels
4017                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4018                                         $width = intval( $m[1] );
4019                                         $height = intval( $m[2] );
4020                                 } else {
4021                                         $width = intval($match);
4022                                 }
4023                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4024                                 $framed=true;
4025                         } else {
4026                                 $caption = $val;
4027                         }
4028                 }
4029                 # Strip bad stuff out of the alt text
4030                 $alt = $this->replaceLinkHoldersText( $caption );
4031
4032                 # make sure there are no placeholders in thumbnail attributes
4033                 # that are later expanded to html- so expand them now and
4034                 # remove the tags
4035                 $alt = $this->unstrip($alt, $this->mStripState);
4036                 $alt = Sanitizer::stripAllTags( $alt );
4037
4038                 # Linker does the rest
4039                 $sk =& $this->mOptions->getSkin();
4040                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4041         }
4042
4043         /**
4044          * Set a flag in the output object indicating that the content is dynamic and
4045          * shouldn't be cached.
4046          */
4047         function disableCache() {
4048                 wfDebug( "Parser output marked as uncacheable.\n" );
4049                 $this->mOutput->mCacheTime = -1;
4050         }
4051
4052         /**#@+
4053          * Callback from the Sanitizer for expanding items found in HTML attribute
4054          * values, so they can be safely tested and escaped.
4055          * @param string $text
4056          * @param array $args
4057          * @return string
4058          * @private
4059          */
4060         function attributeStripCallback( &$text, $args ) {
4061                 $text = $this->replaceVariables( $text, $args );
4062                 $text = $this->unstripForHTML( $text );
4063                 return $text;
4064         }
4065
4066         function unstripForHTML( $text ) {
4067                 $text = $this->unstrip( $text, $this->mStripState );
4068                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4069                 return $text;
4070         }
4071         /**#@-*/
4072
4073         /**#@+
4074          * Accessor/mutator
4075          */
4076         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4077         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4078         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4079         /**#@-*/
4080
4081         /**#@+
4082          * Accessor
4083          */
4084         function getTags() { return array_keys( $this->mTagHooks ); }
4085         /**#@-*/
4086 }
4087
4088 /**
4089  * @todo document
4090  * @package MediaWiki
4091  */
4092 class ParserOutput
4093 {
4094         var $mText,             # The output text
4095                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4096                 $mCategories,       # Map of category names to sort keys
4097                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4098                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4099                 $mVersion,          # Compatibility check
4100                 $mTitleText,        # title text of the chosen language variant
4101                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4102                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4103                 $mImages,           # DB keys of the images used, in the array key only
4104                 $mExternalLinks,    # External link URLs, in the key only
4105                 $mHTMLtitle,            # Display HTML title
4106                 $mSubtitle,                     # Additional subtitle
4107                 $mNewSection;           # Show a new section link?
4108
4109         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4110                 $containsOldMagic = false, $titletext = '' )
4111         {
4112                 $this->mText = $text;
4113                 $this->mLanguageLinks = $languageLinks;
4114                 $this->mCategories = $categoryLinks;
4115                 $this->mContainsOldMagic = $containsOldMagic;
4116                 $this->mCacheTime = '';
4117                 $this->mVersion = MW_PARSER_VERSION;
4118                 $this->mTitleText = $titletext;
4119                 $this->mLinks = array();
4120                 $this->mTemplates = array();
4121                 $this->mImages = array();
4122                 $this->mExternalLinks = array();
4123                 $this->mHTMLtitle = "" ;
4124                 $this->mSubtitle = "" ;
4125                 $this->mNewSection = false;
4126         }
4127
4128         function getText()                   { return $this->mText; }
4129         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4130         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4131         function &getCategories()            { return $this->mCategories; }
4132         function getCacheTime()              { return $this->mCacheTime; }
4133         function getTitleText()              { return $this->mTitleText; }
4134         function &getLinks()                 { return $this->mLinks; }
4135         function &getTemplates()             { return $this->mTemplates; }
4136         function &getImages()                { return $this->mImages; }
4137         function &getExternalLinks()         { return $this->mExternalLinks; }
4138
4139         function containsOldMagic()          { return $this->mContainsOldMagic; }
4140         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4141         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4142         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4143         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4144         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4145         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4146
4147         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4148         function addImage( $name )           { $this->mImages[$name] = 1; }
4149         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4150         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4151
4152         function setNewSection( $value ) {
4153                 $this->mNewSection = (bool)$value;
4154         }
4155         function getNewSection() {
4156                 return (bool)$this->mNewSection;
4157         }
4158
4159         function addLink( $title, $id ) {
4160                 $ns = $title->getNamespace();
4161                 $dbk = $title->getDBkey();
4162                 if ( !isset( $this->mLinks[$ns] ) ) {
4163                         $this->mLinks[$ns] = array();
4164                 }
4165                 $this->mLinks[$ns][$dbk] = $id;
4166         }
4167
4168         function addTemplate( $title, $id ) {
4169                 $ns = $title->getNamespace();
4170                 $dbk = $title->getDBkey();
4171                 if ( !isset( $this->mTemplates[$ns] ) ) {
4172                         $this->mTemplates[$ns] = array();
4173                 }
4174                 $this->mTemplates[$ns][$dbk] = $id;
4175         }
4176
4177         /**
4178          * Return true if this cached output object predates the global or
4179          * per-article cache invalidation timestamps, or if it comes from
4180          * an incompatible older version.
4181          *
4182          * @param string $touched the affected article's last touched timestamp
4183          * @return bool
4184          * @public
4185          */
4186         function expired( $touched ) {
4187                 global $wgCacheEpoch;
4188                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4189                        $this->getCacheTime() < $touched ||
4190                        $this->getCacheTime() <= $wgCacheEpoch ||
4191                        !isset( $this->mVersion ) ||
4192                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4193         }
4194 }
4195
4196 /**
4197  * Set options of the Parser
4198  * @todo document
4199  * @package MediaWiki
4200  */
4201 class ParserOptions
4202 {
4203         # All variables are private
4204         var $mUseTeX;                    # Use texvc to expand <math> tags
4205         var $mUseDynamicDates;           # Use DateFormatter to format dates
4206         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4207         var $mAllowExternalImages;       # Allow external images inline
4208         var $mAllowExternalImagesFrom;   # If not, any exception?
4209         var $mSkin;                      # Reference to the preferred skin
4210         var $mDateFormat;                # Date format index
4211         var $mEditSection;               # Create "edit section" links
4212         var $mNumberHeadings;            # Automatically number headings
4213         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4214         var $mTidy;                      # Ask for tidy cleanup
4215         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4216
4217         function getUseTeX()                        { return $this->mUseTeX; }
4218         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4219         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4220         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4221         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4222         function &getSkin()                         { return $this->mSkin; }
4223         function getDateFormat()                    { return $this->mDateFormat; }
4224         function getEditSection()                   { return $this->mEditSection; }
4225         function getNumberHeadings()                { return $this->mNumberHeadings; }
4226         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4227         function getTidy()                          { return $this->mTidy; }
4228         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4229
4230         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4231         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4232         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4233         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4234         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4235         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4236         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4237         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4238         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4239         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4240         function setSkin( &$x ) { $this->mSkin =& $x; }
4241         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4242
4243         function ParserOptions() {
4244                 global $wgUser;
4245                 $this->initialiseFromUser( $wgUser );
4246         }
4247
4248         /**
4249          * Get parser options
4250          * @static
4251          */
4252         function newFromUser( &$user ) {
4253                 $popts = new ParserOptions;
4254                 $popts->initialiseFromUser( $user );
4255                 return $popts;
4256         }
4257
4258         /** Get user options */
4259         function initialiseFromUser( &$userInput ) {
4260                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4261                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4262                 $fname = 'ParserOptions::initialiseFromUser';
4263                 wfProfileIn( $fname );
4264                 if ( !$userInput ) {
4265                         $user = new User;
4266                         $user->setLoaded( true );
4267                 } else {
4268                         $user =& $userInput;
4269                 }
4270
4271                 $this->mUseTeX = $wgUseTeX;
4272                 $this->mUseDynamicDates = $wgUseDynamicDates;
4273                 $this->mInterwikiMagic = $wgInterwikiMagic;
4274                 $this->mAllowExternalImages = $wgAllowExternalImages;
4275                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4276                 wfProfileIn( $fname.'-skin' );
4277                 $this->mSkin =& $user->getSkin();
4278                 wfProfileOut( $fname.'-skin' );
4279                 $this->mDateFormat = $user->getOption( 'date' );
4280                 $this->mEditSection = true;
4281                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4282                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4283                 $this->mTidy = false;
4284                 $this->mInterfaceMessage = false;
4285                 wfProfileOut( $fname );
4286         }
4287 }
4288
4289 /**
4290  * Callback function used by Parser::replaceLinkHolders()
4291  * to substitute link placeholders.
4292  */
4293 function &wfOutputReplaceMatches( $matches ) {
4294         global $wgOutputReplace;
4295         return $wgOutputReplace[$matches[1]];
4296 }
4297
4298 /**
4299  * Return the total number of articles
4300  */
4301 function wfNumberOfArticles() {
4302         global $wgNumberOfArticles;
4303
4304         wfLoadSiteStats();
4305         return $wgNumberOfArticles;
4306 }
4307
4308 /**
4309  * Return the number of files
4310  */
4311 function wfNumberOfFiles() {
4312         $fname = 'wfNumberOfFiles';
4313
4314         wfProfileIn( $fname );
4315         $dbr =& wfGetDB( DB_SLAVE );
4316         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4317         wfProfileOut( $fname );
4318
4319         return $numImages;
4320 }
4321
4322 /**
4323  * Return the number of user accounts
4324  * @return integer
4325  */
4326 function wfNumberOfUsers() {
4327         wfProfileIn( 'wfNumberOfUsers' );
4328         $dbr =& wfGetDB( DB_SLAVE );
4329         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4330         wfProfileOut( 'wfNumberOfUsers' );
4331         return (int)$count;
4332 }
4333
4334 /**
4335  * Return the total number of pages
4336  * @return integer
4337  */
4338 function wfNumberOfPages() {
4339         wfProfileIn( 'wfNumberOfPages' );
4340         $dbr =& wfGetDB( DB_SLAVE );
4341         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4342         wfProfileOut( 'wfNumberOfPages' );
4343         return (int)$count;
4344 }
4345
4346 /**
4347  * Get various statistics from the database
4348  * @private
4349  */
4350 function wfLoadSiteStats() {
4351         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4352         $fname = 'wfLoadSiteStats';
4353
4354         if ( -1 != $wgNumberOfArticles ) return;
4355         $dbr =& wfGetDB( DB_SLAVE );
4356         $s = $dbr->selectRow( 'site_stats',
4357                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4358                 array( 'ss_row_id' => 1 ), $fname
4359         );
4360
4361         if ( $s === false ) {
4362                 return;
4363         } else {
4364                 $wgTotalViews = $s->ss_total_views;
4365                 $wgTotalEdits = $s->ss_total_edits;
4366                 $wgNumberOfArticles = $s->ss_good_articles;
4367         }
4368 }
4369
4370 /**
4371  * Escape html tags
4372  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4373  *
4374  * @param $in String: text that might contain HTML tags.
4375  * @return string Escaped string
4376  */
4377 function wfEscapeHTMLTagsOnly( $in ) {
4378         return str_replace(
4379                 array( '"', '>', '<' ),
4380                 array( '&quot;', '&gt;', '&lt;' ),
4381                 $in );
4382 }
4383
4384 ?>