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 // State constants for the definition list colon extraction
  63 define( 'MW_COLON_STATE_TEXT', 0 );
  64 define( 'MW_COLON_STATE_TAG', 1 );
  65 define( 'MW_COLON_STATE_TAGSTART', 2 );
  66 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  67 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  68 define( 'MW_COLON_STATE_COMMENT', 5 );
  69 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  70 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  71
  72 /**
  73  * PHP Parser
  74  *
  75  * Processes wiki markup
  76  *
  77  * <pre>
  78  * There are three main entry points into the Parser class:
  79  * parse()
  80  *   produces HTML output
  81  * preSaveTransform().
  82  *   produces altered wiki markup.
  83  * transformMsg()
  84  *   performs brace substitution on MediaWiki messages
  85  *
  86  * Globals used:
  87  *    objects:   $wgLang, $wgContLang
  88  *
  89  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  90  *
  91  * settings:
  92  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  93  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  94  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  95  *
  96  *  * only within ParserOptions
  97  * </pre>
  98  *
  99  * @package MediaWiki
 100  */
 101 class Parser
 102 {
 103         /**#@+
 104          * @private
 105          */
 106         # Persistent:
 107         var $mTagHooks, $mFunctionHooks;
 108
 109         # Cleared with clearState():
 110         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 111         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 112         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 113         var $mTemplates,        // cache of already loaded templates, avoids
 114                                 // multiple SQL queries for the same string
 115             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 116                                 // in this path. Used for loop detection.
 117
 118         # Temporary
 119         # These are variables reset at least once per parse regardless of $clearState
 120         var $mOptions,      // ParserOptions object
 121                 $mTitle,        // Title context, used for self-link rendering and similar things
 122                 $mOutputType,   // Output type, one of the OT_xxx constants
 123                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 124
 125         /**#@-*/
 126
 127         /**
 128          * Constructor
 129          *
 130          * @public
 131          */
 132         function Parser() {
 133                 $this->mTagHooks = array();
 134                 $this->mFunctionHooks = array();
 135                 $this->clearState();
 136         }
 137
 138         /**
 139          * Clear Parser state
 140          *
 141          * @private
 142          */
 143         function clearState() {
 144                 $this->mOutput = new ParserOutput;
 145                 $this->mAutonumber = 0;
 146                 $this->mLastSection = '';
 147                 $this->mDTopen = false;
 148                 $this->mVariables = false;
 149                 $this->mIncludeCount = array();
 150                 $this->mStripState = array();
 151                 $this->mArgStack = array();
 152                 $this->mInPre = false;
 153                 $this->mInterwikiLinkHolders = array(
 154                         'texts' => array(),
 155                         'titles' => array()
 156                 );
 157                 $this->mLinkHolders = array(
 158                         'namespaces' => array(),
 159                         'dbkeys' => array(),
 160                         'queries' => array(),
 161                         'texts' => array(),
 162                         'titles' => array()
 163                 );
 164                 $this->mRevisionId = null;
 165
 166                 /**
 167                  * Prefix for temporary replacement strings for the multipass parser.
 168                  * \x07 should never appear in input as it's disallowed in XML.
 169                  * Using it at the front also gives us a little extra robustness
 170                  * since it shouldn't match when butted up against identifier-like
 171                  * string constructs.
 172                  */
 173                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 174
 175                 # Clear these on every parse, bug 4549
 176                 $this->mTemplates = array();
 177                 $this->mTemplatePath = array();
 178
 179                 $this->mShowToc = true;
 180                 $this->mForceTocPosition = false;
 181
 182                 wfRunHooks( 'ParserClearState', array( &$this ) );
 183         }
 184
 185         /**
 186          * Accessor for mUniqPrefix.
 187          *
 188          * @public
 189          */
 190         function UniqPrefix() {
 191                 return $this->mUniqPrefix;
 192         }
 193
 194         /**
 195          * Convert wikitext to HTML
 196          * Do not call this function recursively.
 197          *
 198          * @private
 199          * @param string $text Text we want to parse
 200          * @param Title &$title A title object
 201          * @param array $options
 202          * @param boolean $linestart
 203          * @param boolean $clearState
 204          * @param int $revid number to pass in {{REVISIONID}}
 205          * @return ParserOutput a ParserOutput
 206          */
 207         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 208                 /**
 209                  * First pass--just handle <nowiki> sections, pass the rest off
 210                  * to internalParse() which does all the real work.
 211                  */
 212
 213                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 214                 $fname = 'Parser::parse';
 215                 wfProfileIn( $fname );
 216
 217                 if ( $clearState ) {
 218                         $this->clearState();
 219                 }
 220
 221                 $this->mOptions = $options;
 222                 $this->mTitle =& $title;
 223                 $this->mRevisionId = $revid;
 224                 $this->mOutputType = OT_HTML;
 225
 226                 //$text = $this->strip( $text, $this->mStripState );
 227                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 228                 $x =& $this->mStripState;
 229
 230                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 231                 $text = $this->strip( $text, $x );
 232                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 233
 234                 # Hook to suspend the parser in this state
 235                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 236                         wfProfileOut( $fname );
 237                         return $text ;
 238                 }
 239
 240                 $text = $this->internalParse( $text );
 241
 242                 $text = $this->unstrip( $text, $this->mStripState );
 243
 244                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 245                 $fixtags = array(
 246                         # french spaces, last one Guillemet-left
 247                         # only if there is something before the space
 248                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 249                         # french spaces, Guillemet-right
 250                         '/(\\302\\253) /' => '\\1&nbsp;',
 251                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 252                 );
 253                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 254
 255                 # only once and last
 256                 $text = $this->doBlockLevels( $text, $linestart );
 257
 258                 $this->replaceLinkHolders( $text );
 259
 260                 # the position of the parserConvert() call should not be changed. it
 261                 # assumes that the links are all replaced and the only thing left
 262                 # is the <nowiki> mark.
 263                 # Side-effects: this calls $this->mOutput->setTitleText()
 264                 $text = $wgContLang->parserConvert( $text, $this );
 265
 266                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 267
 268                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 269
 270                 $text = Sanitizer::normalizeCharReferences( $text );
 271
 272                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 273                         $text = Parser::tidy($text);
 274                 } else {
 275                         # attempt to sanitize at least some nesting problems
 276                         # (bug #2702 and quite a few others)
 277                         $tidyregs = array(
 278                                 # ''Something [http://www.cool.com cool''] -->
 279                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 280                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 281                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 282                                 # fix up an anchor inside another anchor, only
 283                                 # at least for a single single nested link (bug 3695)
 284                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 285                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 286                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 287                                 # contains a div, so fix it up here; replace
 288                                 # div with escaped text
 289                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 290                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 291                                 # remove empty italic or bold tag pairs, some
 292                                 # introduced by rules above
 293                                 '/<([bi])><\/\\1>/' => ''
 294                         );
 295
 296                         $text = preg_replace(
 297                                 array_keys( $tidyregs ),
 298                                 array_values( $tidyregs ),
 299                                 $text );
 300                 }
 301
 302                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 303
 304                 $this->mOutput->setText( $text );
 305                 wfProfileOut( $fname );
 306
 307                 return $this->mOutput;
 308         }
 309
 310         /**
 311          * Get a random string
 312          *
 313          * @private
 314          * @static
 315          */
 316         function getRandomString() {
 317                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 318         }
 319
 320         function &getTitle() { return $this->mTitle; }
 321         function getOptions() { return $this->mOptions; }
 322
 323         /**
 324          * Replaces all occurrences of HTML-style comments and the given tags
 325          * in the text with a random marker and returns teh next text. The output
 326          * parameter $matches will be an associative array filled with data in
 327          * the form:
 328          *   'UNIQ-xxxxx' => array(
 329          *     'element',
 330          *     'tag content',
 331          *     array( 'param' => 'x' ),
 332          *     '<element param="x">tag content</element>' ) )
 333          *
 334          * @param $elements list of element names. Comments are always extracted.
 335          * @param $text Source text string.
 336          * @param $uniq_prefix
 337          *
 338          * @private
 339          * @static
 340          */
 341         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 342                 $rand = Parser::getRandomString();
 343                 $n = 1;
 344                 $stripped = '';
 345                 $matches = array();
 346
 347                 $taglist = implode( '|', $elements );
 348                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 349
 350                 while ( '' != $text ) {
 351                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 352                         $stripped .= $p[0];
 353                         if( count( $p ) < 5 ) {
 354                                 break;
 355                         }
 356                         if( count( $p ) > 5 ) {
 357                                 // comment
 358                                 $element    = $p[4];
 359                                 $attributes = '';
 360                                 $close      = '';
 361                                 $inside     = $p[5];
 362                         } else {
 363                                 // tag
 364                                 $element    = $p[1];
 365                                 $attributes = $p[2];
 366                                 $close      = $p[3];
 367                                 $inside     = $p[4];
 368                         }
 369
 370                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++);
 371                         $stripped .= $marker;
 372
 373                         if ( $close === '/>' ) {
 374                                 // Empty element tag, <tag />
 375                                 $content = null;
 376                                 $text = $inside;
 377                                 $tail = null;
 378                         } else {
 379                                 if( $element == '!--' ) {
 380                                         $end = '/(-->)/';
 381                                 } else {
 382                                         $end = "/(<\\/$element\\s*>)/i";
 383                                 }
 384                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 385                                 $content = $q[0];
 386                                 if( count( $q ) < 3 ) {
 387                                         # No end tag -- let it run out to the end of the text.
 388                                         $tail = '';
 389                                         $text = '';
 390                                 } else {
 391                                         $tail = $q[1];
 392                                         $text = $q[2];
 393                                 }
 394                         }
 395
 396                         $matches[$marker] = array( $element,
 397                                 $content,
 398                                 Sanitizer::decodeTagAttributes( $attributes ),
 399                                 "<$element$attributes$close$content$tail" );
 400                 }
 401                 return $stripped;
 402         }
 403
 404         /**
 405          * Strips and renders nowiki, pre, math, hiero
 406          * If $render is set, performs necessary rendering operations on plugins
 407          * Returns the text, and fills an array with data needed in unstrip()
 408          * If the $state is already a valid strip state, it adds to the state
 409          *
 410          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 411          *  will be stripped in addition to other tags. This is important
 412          *  for section editing, where these comments cause confusion when
 413          *  counting the sections in the wikisource
 414          *
 415          * @private
 416          */
 417         function strip( $text, &$state, $stripcomments = false ) {
 418                 $render = ($this->mOutputType == OT_HTML);
 419
 420                 # Replace any instances of the placeholders
 421                 $uniq_prefix = $this->mUniqPrefix;
 422                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 423                 $commentState = array();
 424
 425                 $elements = array_merge(
 426                         array( 'nowiki', 'pre', 'gallery' ),
 427                         array_keys( $this->mTagHooks ) );
 428                 global $wgRawHtml;
 429                 if( $wgRawHtml ) {
 430                         $elements[] = 'html';
 431                 }
 432                 if( $this->mOptions->getUseTeX() ) {
 433                         $elements[] = 'math';
 434                 }
 435
 436
 437                 $matches = array();
 438                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 439
 440                 foreach( $matches as $marker => $data ) {
 441                         list( $element, $content, $params, $tag ) = $data;
 442                         if( $render ) {
 443                                 $tagName = strtolower( $element );
 444                                 switch( $tagName ) {
 445                                 case '!--':
 446                                         // Comment
 447                                         if( substr( $tag, -3 ) == '-->' ) {
 448                                                 $output = $tag;
 449                                         } else {
 450                                                 // Unclosed comment in input.
 451                                                 // Close it so later stripping can remove it
 452                                                 $output = "$tag-->";
 453                                         }
 454                                         break;
 455                                 case 'html':
 456                                         if( $wgRawHtml ) {
 457                                                 $output = $content;
 458                                                 break;
 459                                         }
 460                                         // Shouldn't happen otherwise. :)
 461                                 case 'nowiki':
 462                                         $output = wfEscapeHTMLTagsOnly( $content );
 463                                         break;
 464                                 case 'math':
 465                                         $output = renderMath( $content );
 466                                         break;
 467                                 case 'pre':
 468                                         // Backwards-compatibility hack
 469                                         $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $content );
 470                                         $output = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 471                                         break;
 472                                 case 'gallery':
 473                                         $output = $this->renderImageGallery( $content );
 474                                         break;
 475                                 default:
 476                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 477                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 478                                                         array( $content, $params, $this ) );
 479                                         } else {
 480                                                 wfDebugDieBacktrace( "Invalid call hook $element" );
 481                                         }
 482                                 }
 483                         } else {
 484                                 // Just stripping tags; keep the source
 485                                 $output = $tag;
 486                         }
 487                         if( !$stripcomments && $element == '!--' ) {
 488                                 $commentState[$marker] = $output;
 489                         } else {
 490                                 $state[$element][$marker] = $output;
 491                         }
 492                 }
 493
 494                 # Unstrip comments unless explicitly told otherwise.
 495                 # (The comments are always stripped prior to this point, so as to
 496                 # not invoke any extension tags / parser hooks contained within
 497                 # a comment.)
 498                 if ( !$stripcomments ) {
 499                         // Put them all back and forget them
 500                         $text = strtr( $text, $commentState );
 501                 }
 502
 503                 return $text;
 504         }
 505
 506         /**
 507          * Restores pre, math, and other extensions removed by strip()
 508          *
 509          * always call unstripNoWiki() after this one
 510          * @private
 511          */
 512         function unstrip( $text, &$state ) {
 513                 if ( !is_array( $state ) ) {
 514                         return $text;
 515                 }
 516
 517                 $replacements = array();
 518                 foreach( $state as $tag => $contentDict ) {
 519                         if( $tag != 'nowiki' && $tag != 'html' ) {
 520                                 foreach( $contentDict as $uniq => $content ) {
 521                                         $replacements[$uniq] = $content;
 522                                 }
 523                         }
 524                 }
 525                 $text = strtr( $text, $replacements );
 526
 527                 return $text;
 528         }
 529
 530         /**
 531          * Always call this after unstrip() to preserve the order
 532          *
 533          * @private
 534          */
 535         function unstripNoWiki( $text, &$state ) {
 536                 if ( !is_array( $state ) ) {
 537                         return $text;
 538                 }
 539
 540                 $replacements = array();
 541                 foreach( $state as $tag => $contentDict ) {
 542                         if( $tag == 'nowiki' || $tag == 'html' ) {
 543                                 foreach( $contentDict as $uniq => $content ) {
 544                                         $replacements[$uniq] = $content;
 545                                 }
 546                         }
 547                 }
 548                 $text = strtr( $text, $replacements );
 549
 550                 return $text;
 551         }
 552
 553         /**
 554          * Add an item to the strip state
 555          * Returns the unique tag which must be inserted into the stripped text
 556          * The tag will be replaced with the original text in unstrip()
 557          *
 558          * @private
 559          */
 560         function insertStripItem( $text, &$state ) {
 561                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 562                 if ( !$state ) {
 563                         $state = array();
 564                 }
 565                 $state['item'][$rnd] = $text;
 566                 return $rnd;
 567         }
 568
 569         /**
 570          * Interface with html tidy, used if $wgUseTidy = true.
 571          * If tidy isn't able to correct the markup, the original will be
 572          * returned in all its glory with a warning comment appended.
 573          *
 574          * Either the external tidy program or the in-process tidy extension
 575          * will be used depending on availability. Override the default
 576          * $wgTidyInternal setting to disable the internal if it's not working.
 577          *
 578          * @param string $text Hideous HTML input
 579          * @return string Corrected HTML output
 580          * @public
 581          * @static
 582          */
 583         function tidy( $text ) {
 584                 global $wgTidyInternal;
 585                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 586 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 587 '<head><title>test</title></head><body>'.$text.'</body></html>';
 588                 if( $wgTidyInternal ) {
 589                         $correctedtext = Parser::internalTidy( $wrappedtext );
 590                 } else {
 591                         $correctedtext = Parser::externalTidy( $wrappedtext );
 592                 }
 593                 if( is_null( $correctedtext ) ) {
 594                         wfDebug( "Tidy error detected!\n" );
 595                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 596                 }
 597                 return $correctedtext;
 598         }
 599
 600         /**
 601          * Spawn an external HTML tidy process and get corrected markup back from it.
 602          *
 603          * @private
 604          * @static
 605          */
 606         function externalTidy( $text ) {
 607                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 608                 $fname = 'Parser::externalTidy';
 609                 wfProfileIn( $fname );
 610
 611                 $cleansource = '';
 612                 $opts = ' -utf8';
 613
 614                 $descriptorspec = array(
 615                         0 => array('pipe', 'r'),
 616                         1 => array('pipe', 'w'),
 617                         2 => array('file', '/dev/null', 'a')
 618                 );
 619                 $pipes = array();
 620                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 621                 if (is_resource($process)) {
 622                         // Theoretically, this style of communication could cause a deadlock
 623                         // here. If the stdout buffer fills up, then writes to stdin could
 624                         // block. This doesn't appear to happen with tidy, because tidy only
 625                         // writes to stdout after it's finished reading from stdin. Search
 626                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 627                         fwrite($pipes[0], $text);
 628                         fclose($pipes[0]);
 629                         while (!feof($pipes[1])) {
 630                                 $cleansource .= fgets($pipes[1], 1024);
 631                         }
 632                         fclose($pipes[1]);
 633                         proc_close($process);
 634                 }
 635
 636                 wfProfileOut( $fname );
 637
 638                 if( $cleansource == '' && $text != '') {
 639                         // Some kind of error happened, so we couldn't get the corrected text.
 640                         // Just give up; we'll use the source text and append a warning.
 641                         return null;
 642                 } else {
 643                         return $cleansource;
 644                 }
 645         }
 646
 647         /**
 648          * Use the HTML tidy PECL extension to use the tidy library in-process,
 649          * saving the overhead of spawning a new process. Currently written to
 650          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 651          *
 652          * 'pear install tidy' should be able to compile the extension module.
 653          *
 654          * @private
 655          * @static
 656          */
 657         function internalTidy( $text ) {
 658                 global $wgTidyConf;
 659                 $fname = 'Parser::internalTidy';
 660                 wfProfileIn( $fname );
 661
 662                 tidy_load_config( $wgTidyConf );
 663                 tidy_set_encoding( 'utf8' );
 664                 tidy_parse_string( $text );
 665                 tidy_clean_repair();
 666                 if( tidy_get_status() == 2 ) {
 667                         // 2 is magic number for fatal error
 668                         // http://www.php.net/manual/en/function.tidy-get-status.php
 669                         $cleansource = null;
 670                 } else {
 671                         $cleansource = tidy_get_output();
 672                 }
 673                 wfProfileOut( $fname );
 674                 return $cleansource;
 675         }
 676
 677         /**
 678          * parse the wiki syntax used to render tables
 679          *
 680          * @private
 681          */
 682         function doTableStuff ( $t ) {
 683                 $fname = 'Parser::doTableStuff';
 684                 wfProfileIn( $fname );
 685
 686                 $t = explode ( "\n" , $t ) ;
 687                 $td = array () ; # Is currently a td tag open?
 688                 $ltd = array () ; # Was it TD or TH?
 689                 $tr = array () ; # Is currently a tr tag open?
 690                 $ltr = array () ; # tr attributes
 691                 $has_opened_tr = array(); # Did this table open a <tr> element?
 692                 $indent_level = 0; # indent level of the table
 693                 foreach ( $t AS $k => $x )
 694                 {
 695                         $x = trim ( $x ) ;
 696                         $fc = substr ( $x , 0 , 1 ) ;
 697                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 698                                 $indent_level = strlen( $matches[1] );
 699
 700                                 $attributes = $this->unstripForHTML( $matches[2] );
 701
 702                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 703                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 704                                 array_push ( $td , false ) ;
 705                                 array_push ( $ltd , '' ) ;
 706                                 array_push ( $tr , false ) ;
 707                                 array_push ( $ltr , '' ) ;
 708                                 array_push ( $has_opened_tr, false );
 709                         }
 710                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 711                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 712                                 $z = "</table>" . substr ( $x , 2);
 713                                 $l = array_pop ( $ltd ) ;
 714                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 715                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 716                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 717                                 array_pop ( $ltr ) ;
 718                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 719                         }
 720                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 721                                 $x = substr ( $x , 1 ) ;
 722                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 723                                 $z = '' ;
 724                                 $l = array_pop ( $ltd ) ;
 725                                 array_pop ( $has_opened_tr );
 726                                 array_push ( $has_opened_tr , true ) ;
 727                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 728                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 729                                 array_pop ( $ltr ) ;
 730                                 $t[$k] = $z ;
 731                                 array_push ( $tr , false ) ;
 732                                 array_push ( $td , false ) ;
 733                                 array_push ( $ltd , '' ) ;
 734                                 $attributes = $this->unstripForHTML( $x );
 735                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 736                         }
 737                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 738                                 # $x is a table row
 739                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 740                                         $fc = '+' ;
 741                                         $x = substr ( $x , 1 ) ;
 742                                 }
 743                                 $after = substr ( $x , 1 ) ;
 744                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 745
 746                                 // Split up multiple cells on the same line.
 747                                 // FIXME: This can result in improper nesting of tags processed
 748                                 // by earlier parser steps, but should avoid splitting up eg
 749                                 // attribute values containing literal "||".
 750                                 $after = wfExplodeMarkup( '||', $after );
 751
 752                                 $t[$k] = '' ;
 753
 754                                 # Loop through each table cell
 755                                 foreach ( $after AS $theline )
 756                                 {
 757                                         $z = '' ;
 758                                         if ( $fc != '+' )
 759                                         {
 760                                                 $tra = array_pop ( $ltr ) ;
 761                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 762                                                 array_push ( $tr , true ) ;
 763                                                 array_push ( $ltr , '' ) ;
 764                                                 array_pop ( $has_opened_tr );
 765                                                 array_push ( $has_opened_tr , true ) ;
 766                                         }
 767
 768                                         $l = array_pop ( $ltd ) ;
 769                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 770                                         if ( $fc == '|' ) $l = 'td' ;
 771                                         else if ( $fc == '!' ) $l = 'th' ;
 772                                         else if ( $fc == '+' ) $l = 'caption' ;
 773                                         else $l = '' ;
 774                                         array_push ( $ltd , $l ) ;
 775
 776                                         # Cell parameters
 777                                         $y = explode ( '|' , $theline , 2 ) ;
 778                                         # Note that a '|' inside an invalid link should not
 779                                         # be mistaken as delimiting cell parameters
 780                                         if ( strpos( $y[0], '[[' ) !== false ) {
 781                                                 $y = array ($theline);
 782                                         }
 783                                         if ( count ( $y ) == 1 )
 784                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 785                                         else {
 786                                                 $attributes = $this->unstripForHTML( $y[0] );
 787                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 788                                         }
 789                                         $t[$k] .= $y ;
 790                                         array_push ( $td , true ) ;
 791                                 }
 792                         }
 793                 }
 794
 795                 # Closing open td, tr && table
 796                 while ( count ( $td ) > 0 )
 797                 {
 798                         $l = array_pop ( $ltd ) ;
 799                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 800                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 801                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 802                         $t[] = '</table>' ;
 803                 }
 804
 805                 $t = implode ( "\n" , $t ) ;
 806                 # special case: don't return empty table
 807                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 808                         $t = '';
 809                 wfProfileOut( $fname );
 810                 return $t ;
 811         }
 812
 813         /**
 814          * Helper function for parse() that transforms wiki markup into
 815          * HTML. Only called for $mOutputType == OT_HTML.
 816          *
 817          * @private
 818          */
 819         function internalParse( $text ) {
 820                 $args = array();
 821                 $isMain = true;
 822                 $fname = 'Parser::internalParse';
 823                 wfProfileIn( $fname );
 824
 825                 # Remove <noinclude> tags and <includeonly> sections
 826                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 827                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 828                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 829
 830                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 831
 832                 $text = $this->replaceVariables( $text, $args );
 833
 834                 // Tables need to come after variable replacement for things to work
 835                 // properly; putting them before other transformations should keep
 836                 // exciting things like link expansions from showing up in surprising
 837                 // places.
 838                 $text = $this->doTableStuff( $text );
 839
 840                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 841
 842                 $text = $this->stripToc( $text );
 843                 $text = $this->doHeadings( $text );
 844                 if($this->mOptions->getUseDynamicDates()) {
 845                         $df =& DateFormatter::getInstance();
 846                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 847                 }
 848                 $text = $this->doAllQuotes( $text );
 849                 $text = $this->replaceInternalLinks( $text );
 850                 $text = $this->replaceExternalLinks( $text );
 851
 852                 # replaceInternalLinks may sometimes leave behind
 853                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 854                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 855
 856                 $text = $this->doMagicLinks( $text );
 857                 $text = $this->formatHeadings( $text, $isMain );
 858
 859                 wfProfileOut( $fname );
 860                 return $text;
 861         }
 862
 863         /**
 864          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 865          * magic external links.
 866          *
 867          * @private
 868          */
 869         function &doMagicLinks( &$text ) {
 870                 $text = $this->magicISBN( $text );
 871                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 872                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 873                 return $text;
 874         }
 875
 876         /**
 877          * Parse headers and return html
 878          *
 879          * @private
 880          */
 881         function doHeadings( $text ) {
 882                 $fname = 'Parser::doHeadings';
 883                 wfProfileIn( $fname );
 884                 for ( $i = 6; $i >= 1; --$i ) {
 885                         $h = str_repeat( '=', $i );
 886                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 887                           "<h{$i}>\\1</h{$i}>\\2", $text );
 888                 }
 889                 wfProfileOut( $fname );
 890                 return $text;
 891         }
 892
 893         /**
 894          * Replace single quotes with HTML markup
 895          * @private
 896          * @return string the altered text
 897          */
 898         function doAllQuotes( $text ) {
 899                 $fname = 'Parser::doAllQuotes';
 900                 wfProfileIn( $fname );
 901                 $outtext = '';
 902                 $lines = explode( "\n", $text );
 903                 foreach ( $lines as $line ) {
 904                         $outtext .= $this->doQuotes ( $line ) . "\n";
 905                 }
 906                 $outtext = substr($outtext, 0,-1);
 907                 wfProfileOut( $fname );
 908                 return $outtext;
 909         }
 910
 911         /**
 912          * Helper function for doAllQuotes()
 913          * @private
 914          */
 915         function doQuotes( $text ) {
 916                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 917                 if ( count( $arr ) == 1 )
 918                         return $text;
 919                 else
 920                 {
 921                         # First, do some preliminary work. This may shift some apostrophes from
 922                         # being mark-up to being text. It also counts the number of occurrences
 923                         # of bold and italics mark-ups.
 924                         $i = 0;
 925                         $numbold = 0;
 926                         $numitalics = 0;
 927                         foreach ( $arr as $r )
 928                         {
 929                                 if ( ( $i % 2 ) == 1 )
 930                                 {
 931                                         # If there are ever four apostrophes, assume the first is supposed to
 932                                         # be text, and the remaining three constitute mark-up for bold text.
 933                                         if ( strlen( $arr[$i] ) == 4 )
 934                                         {
 935                                                 $arr[$i-1] .= "'";
 936                                                 $arr[$i] = "'''";
 937                                         }
 938                                         # If there are more than 5 apostrophes in a row, assume they're all
 939                                         # text except for the last 5.
 940                                         else if ( strlen( $arr[$i] ) > 5 )
 941                                         {
 942                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 943                                                 $arr[$i] = "'''''";
 944                                         }
 945                                         # Count the number of occurrences of bold and italics mark-ups.
 946                                         # We are not counting sequences of five apostrophes.
 947                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 948                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 949                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 950                                 }
 951                                 $i++;
 952                         }
 953
 954                         # If there is an odd number of both bold and italics, it is likely
 955                         # that one of the bold ones was meant to be an apostrophe followed
 956                         # by italics. Which one we cannot know for certain, but it is more
 957                         # likely to be one that has a single-letter word before it.
 958                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 959                         {
 960                                 $i = 0;
 961                                 $firstsingleletterword = -1;
 962                                 $firstmultiletterword = -1;
 963                                 $firstspace = -1;
 964                                 foreach ( $arr as $r )
 965                                 {
 966                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 967                                         {
 968                                                 $x1 = substr ($arr[$i-1], -1);
 969                                                 $x2 = substr ($arr[$i-1], -2, 1);
 970                                                 if ($x1 == ' ') {
 971                                                         if ($firstspace == -1) $firstspace = $i;
 972                                                 } else if ($x2 == ' ') {
 973                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 974                                                 } else {
 975                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 976                                                 }
 977                                         }
 978                                         $i++;
 979                                 }
 980
 981                                 # If there is a single-letter word, use it!
 982                                 if ($firstsingleletterword > -1)
 983                                 {
 984                                         $arr [ $firstsingleletterword ] = "''";
 985                                         $arr [ $firstsingleletterword-1 ] .= "'";
 986                                 }
 987                                 # If not, but there's a multi-letter word, use that one.
 988                                 else if ($firstmultiletterword > -1)
 989                                 {
 990                                         $arr [ $firstmultiletterword ] = "''";
 991                                         $arr [ $firstmultiletterword-1 ] .= "'";
 992                                 }
 993                                 # ... otherwise use the first one that has neither.
 994                                 # (notice that it is possible for all three to be -1 if, for example,
 995                                 # there is only one pentuple-apostrophe in the line)
 996                                 else if ($firstspace > -1)
 997                                 {
 998                                         $arr [ $firstspace ] = "''";
 999                                         $arr [ $firstspace-1 ] .= "'";
1000                                 }
1001                         }
1002
1003                         # Now let's actually convert our apostrophic mush to HTML!
1004                         $output = '';
1005                         $buffer = '';
1006                         $state = '';
1007                         $i = 0;
1008                         foreach ($arr as $r)
1009                         {
1010                                 if (($i % 2) == 0)
1011                                 {
1012                                         if ($state == 'both')
1013                                                 $buffer .= $r;
1014                                         else
1015                                                 $output .= $r;
1016                                 }
1017                                 else
1018                                 {
1019                                         if (strlen ($r) == 2)
1020                                         {
1021                                                 if ($state == 'i')
1022                                                 { $output .= '</i>'; $state = ''; }
1023                                                 else if ($state == 'bi')
1024                                                 { $output .= '</i>'; $state = 'b'; }
1025                                                 else if ($state == 'ib')
1026                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1027                                                 else if ($state == 'both')
1028                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1029                                                 else # $state can be 'b' or ''
1030                                                 { $output .= '<i>'; $state .= 'i'; }
1031                                         }
1032                                         else if (strlen ($r) == 3)
1033                                         {
1034                                                 if ($state == 'b')
1035                                                 { $output .= '</b>'; $state = ''; }
1036                                                 else if ($state == 'bi')
1037                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1038                                                 else if ($state == 'ib')
1039                                                 { $output .= '</b>'; $state = 'i'; }
1040                                                 else if ($state == 'both')
1041                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1042                                                 else # $state can be 'i' or ''
1043                                                 { $output .= '<b>'; $state .= 'b'; }
1044                                         }
1045                                         else if (strlen ($r) == 5)
1046                                         {
1047                                                 if ($state == 'b')
1048                                                 { $output .= '</b><i>'; $state = 'i'; }
1049                                                 else if ($state == 'i')
1050                                                 { $output .= '</i><b>'; $state = 'b'; }
1051                                                 else if ($state == 'bi')
1052                                                 { $output .= '</i></b>'; $state = ''; }
1053                                                 else if ($state == 'ib')
1054                                                 { $output .= '</b></i>'; $state = ''; }
1055                                                 else if ($state == 'both')
1056                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1057                                                 else # ($state == '')
1058                                                 { $buffer = ''; $state = 'both'; }
1059                                         }
1060                                 }
1061                                 $i++;
1062                         }
1063                         # Now close all remaining tags.  Notice that the order is important.
1064                         if ($state == 'b' || $state == 'ib')
1065                                 $output .= '</b>';
1066                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1067                                 $output .= '</i>';
1068                         if ($state == 'bi')
1069                                 $output .= '</b>';
1070                         if ($state == 'both')
1071                                 $output .= '<b><i>'.$buffer.'</i></b>';
1072                         return $output;
1073                 }
1074         }
1075
1076         /**
1077          * Replace external links
1078          *
1079          * Note: this is all very hackish and the order of execution matters a lot.
1080          * Make sure to run maintenance/parserTests.php if you change this code.
1081          *
1082          * @private
1083          */
1084         function replaceExternalLinks( $text ) {
1085                 global $wgContLang;
1086                 $fname = 'Parser::replaceExternalLinks';
1087                 wfProfileIn( $fname );
1088
1089                 $sk =& $this->mOptions->getSkin();
1090
1091                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1092
1093                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1094
1095                 $i = 0;
1096                 while ( $i<count( $bits ) ) {
1097                         $url = $bits[$i++];
1098                         $protocol = $bits[$i++];
1099                         $text = $bits[$i++];
1100                         $trail = $bits[$i++];
1101
1102                         # The characters '<' and '>' (which were escaped by
1103                         # removeHTMLtags()) should not be included in
1104                         # URLs, per RFC 2396.
1105                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1106                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1107                                 $url = substr($url, 0, $m2[0][1]);
1108                         }
1109
1110                         # If the link text is an image URL, replace it with an <img> tag
1111                         # This happened by accident in the original parser, but some people used it extensively
1112                         $img = $this->maybeMakeExternalImage( $text );
1113                         if ( $img !== false ) {
1114                                 $text = $img;
1115                         }
1116
1117                         $dtrail = '';
1118
1119                         # Set linktype for CSS - if URL==text, link is essentially free
1120                         $linktype = ($text == $url) ? 'free' : 'text';
1121
1122                         # No link text, e.g. [http://domain.tld/some.link]
1123                         if ( $text == '' ) {
1124                                 # Autonumber if allowed. See bug #5918
1125                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1126                                         $text = '[' . ++$this->mAutonumber . ']';
1127                                         $linktype = 'autonumber';
1128                                 } else {
1129                                         # Otherwise just use the URL
1130                                         $text = htmlspecialchars( $url );
1131                                         $linktype = 'free';
1132                                 }
1133                         } else {
1134                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1135                                 # Check for trail
1136                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1137                         }
1138
1139                         $text = $wgContLang->markNoConversion($text);
1140
1141                         # Normalize any HTML entities in input. They will be
1142                         # re-escaped by makeExternalLink().
1143                         $url = Sanitizer::decodeCharReferences( $url );
1144
1145                         # Process the trail (i.e. everything after this link up until start of the next link),
1146                         # replacing any non-bracketed links
1147                         $trail = $this->replaceFreeExternalLinks( $trail );
1148
1149                         # Use the encoded URL
1150                         # This means that users can paste URLs directly into the text
1151                         # Funny characters like &ouml; aren't valid in URLs anyway
1152                         # This was changed in August 2004
1153                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1154
1155                         # Register link in the output object.
1156                         # Replace unnecessary URL escape codes with the referenced character
1157                         # This prevents spammers from hiding links from the filters
1158                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1159                         $this->mOutput->addExternalLink( $pasteurized );
1160                 }
1161
1162                 wfProfileOut( $fname );
1163                 return $s;
1164         }
1165
1166         /**
1167          * Replace anything that looks like a URL with a link
1168          * @private
1169          */
1170         function replaceFreeExternalLinks( $text ) {
1171                 global $wgContLang;
1172                 $fname = 'Parser::replaceFreeExternalLinks';
1173                 wfProfileIn( $fname );
1174
1175                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1176                 $s = array_shift( $bits );
1177                 $i = 0;
1178
1179                 $sk =& $this->mOptions->getSkin();
1180
1181                 while ( $i < count( $bits ) ){
1182                         $protocol = $bits[$i++];
1183                         $remainder = $bits[$i++];
1184
1185                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1186                                 # Found some characters after the protocol that look promising
1187                                 $url = $protocol . $m[1];
1188                                 $trail = $m[2];
1189
1190                                 # special case: handle urls as url args:
1191                                 # http://www.example.com/foo?=http://www.example.com/bar
1192                                 if(strlen($trail) == 0 &&
1193                                         isset($bits[$i]) &&
1194                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1195                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1196                                 {
1197                                         # add protocol, arg
1198                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1199                                         $i += 2;
1200                                         $trail = $m[2];
1201                                 }
1202
1203                                 # The characters '<' and '>' (which were escaped by
1204                                 # removeHTMLtags()) should not be included in
1205                                 # URLs, per RFC 2396.
1206                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1207                                         $trail = substr($url, $m2[0][1]) . $trail;
1208                                         $url = substr($url, 0, $m2[0][1]);
1209                                 }
1210
1211                                 # Move trailing punctuation to $trail
1212                                 $sep = ',;\.:!?';
1213                                 # If there is no left bracket, then consider right brackets fair game too
1214                                 if ( strpos( $url, '(' ) === false ) {
1215                                         $sep .= ')';
1216                                 }
1217
1218                                 $numSepChars = strspn( strrev( $url ), $sep );
1219                                 if ( $numSepChars ) {
1220                                         $trail = substr( $url, -$numSepChars ) . $trail;
1221                                         $url = substr( $url, 0, -$numSepChars );
1222                                 }
1223
1224                                 # Normalize any HTML entities in input. They will be
1225                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1226                                 $url = Sanitizer::decodeCharReferences( $url );
1227
1228                                 # Is this an external image?
1229                                 $text = $this->maybeMakeExternalImage( $url );
1230                                 if ( $text === false ) {
1231                                         # Not an image, make a link
1232                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1233                                         # Register it in the output object...
1234                                         # Replace unnecessary URL escape codes with their equivalent characters
1235                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1236                                         $this->mOutput->addExternalLink( $pasteurized );
1237                                 }
1238                                 $s .= $text . $trail;
1239                         } else {
1240                                 $s .= $protocol . $remainder;
1241                         }
1242                 }
1243                 wfProfileOut( $fname );
1244                 return $s;
1245         }
1246
1247         /**
1248          * Replace unusual URL escape codes with their equivalent characters
1249          * @param string
1250          * @return string
1251          * @static
1252          * @fixme This can merge genuinely required bits in the path or query string,
1253          *        breaking legit URLs. A proper fix would treat the various parts of
1254          *        the URL differently; as a workaround, just use the output for
1255          *        statistical records, not for actual linking/output.
1256          */
1257         function replaceUnusualEscapes( $url ) {
1258                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1259                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1260         }
1261
1262         /**
1263          * Callback function used in replaceUnusualEscapes().
1264          * Replaces unusual URL escape codes with their equivalent character
1265          * @static
1266          * @private
1267          */
1268         function replaceUnusualEscapesCallback( $matches ) {
1269                 $char = urldecode( $matches[0] );
1270                 $ord = ord( $char );
1271                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1272                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1273                         // No, shouldn't be escaped
1274                         return $char;
1275                 } else {
1276                         // Yes, leave it escaped
1277                         return $matches[0];
1278                 }
1279         }
1280
1281         /**
1282          * make an image if it's allowed, either through the global
1283          * option or through the exception
1284          * @private
1285          */
1286         function maybeMakeExternalImage( $url ) {
1287                 $sk =& $this->mOptions->getSkin();
1288                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1289                 $imagesexception = !empty($imagesfrom);
1290                 $text = false;
1291                 if ( $this->mOptions->getAllowExternalImages()
1292                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1293                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1294                                 # Image found
1295                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1296                         }
1297                 }
1298                 return $text;
1299         }
1300
1301         /**
1302          * Process [[ ]] wikilinks
1303          *
1304          * @private
1305          */
1306         function replaceInternalLinks( $s ) {
1307                 global $wgContLang;
1308                 static $fname = 'Parser::replaceInternalLinks' ;
1309
1310                 wfProfileIn( $fname );
1311
1312                 wfProfileIn( $fname.'-setup' );
1313                 static $tc = FALSE;
1314                 # the % is needed to support urlencoded titles as well
1315                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1316
1317                 $sk =& $this->mOptions->getSkin();
1318
1319                 #split the entire text string on occurences of [[
1320                 $a = explode( '[[', ' ' . $s );
1321                 #get the first element (all text up to first [[), and remove the space we added
1322                 $s = array_shift( $a );
1323                 $s = substr( $s, 1 );
1324
1325                 # Match a link having the form [[namespace:link|alternate]]trail
1326                 static $e1 = FALSE;
1327                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1328                 # Match cases where there is no "]]", which might still be images
1329                 static $e1_img = FALSE;
1330                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1331                 # Match the end of a line for a word that's not followed by whitespace,
1332                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1333                 $e2 = wfMsgForContent( 'linkprefix' );
1334
1335                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1336
1337                 if( is_null( $this->mTitle ) ) {
1338                         wfDebugDieBacktrace( 'nooo' );
1339                 }
1340                 $nottalk = !$this->mTitle->isTalkPage();
1341
1342                 if ( $useLinkPrefixExtension ) {
1343                         if ( preg_match( $e2, $s, $m ) ) {
1344                                 $first_prefix = $m[2];
1345                         } else {
1346                                 $first_prefix = false;
1347                         }
1348                 } else {
1349                         $prefix = '';
1350                 }
1351
1352                 $selflink = $this->mTitle->getPrefixedText();
1353                 wfProfileOut( $fname.'-setup' );
1354
1355                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1356                 $useSubpages = $this->areSubpagesAllowed();
1357
1358                 # Loop for each link
1359                 for ($k = 0; isset( $a[$k] ); $k++) {
1360                         $line = $a[$k];
1361                         if ( $useLinkPrefixExtension ) {
1362                                 wfProfileIn( $fname.'-prefixhandling' );
1363                                 if ( preg_match( $e2, $s, $m ) ) {
1364                                         $prefix = $m[2];
1365                                         $s = $m[1];
1366                                 } else {
1367                                         $prefix='';
1368                                 }
1369                                 # first link
1370                                 if($first_prefix) {
1371                                         $prefix = $first_prefix;
1372                                         $first_prefix = false;
1373                                 }
1374                                 wfProfileOut( $fname.'-prefixhandling' );
1375                         }
1376
1377                         $might_be_img = false;
1378
1379                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1380                                 $text = $m[2];
1381                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1382                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1383                                 # the real problem is with the $e1 regex
1384                                 # See bug 1300.
1385                                 #
1386                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1387                                 # and no image is in sight. See bug 2095.
1388                                 #
1389                                 if( $text !== '' &&
1390                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1391                                         strpos($text, '[') !== false
1392                                 )
1393                                 {
1394                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1395                                         $m[3] = $n[1];
1396                                 }
1397                                 # fix up urlencoded title texts
1398                                 if(preg_match('/%/', $m[1] ))
1399                                         # Should anchors '#' also be rejected?
1400                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1401                                 $trail = $m[3];
1402                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1403                                 $might_be_img = true;
1404                                 $text = $m[2];
1405                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1406                                 $trail = "";
1407                         } else { # Invalid form; output directly
1408                                 $s .= $prefix . '[[' . $line ;
1409                                 continue;
1410                         }
1411
1412                         # Don't allow internal links to pages containing
1413                         # PROTO: where PROTO is a valid URL protocol; these
1414                         # should be external links.
1415                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1416                                 $s .= $prefix . '[[' . $line ;
1417                                 continue;
1418                         }
1419
1420                         # Make subpage if necessary
1421                         if( $useSubpages ) {
1422                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1423                         } else {
1424                                 $link = $m[1];
1425                         }
1426
1427                         $noforce = (substr($m[1], 0, 1) != ':');
1428                         if (!$noforce) {
1429                                 # Strip off leading ':'
1430                                 $link = substr($link, 1);
1431                         }
1432
1433                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1434                         if( !$nt ) {
1435                                 $s .= $prefix . '[[' . $line;
1436                                 continue;
1437                         }
1438
1439                         #check other language variants of the link
1440                         #if the article does not exist
1441                         if( $checkVariantLink
1442                             && $nt->getArticleID() == 0 ) {
1443                                 $wgContLang->findVariantLink($link, $nt);
1444                         }
1445
1446                         $ns = $nt->getNamespace();
1447                         $iw = $nt->getInterWiki();
1448
1449                         if ($might_be_img) { # if this is actually an invalid link
1450                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1451                                         $found = false;
1452                                         while (isset ($a[$k+1]) ) {
1453                                                 #look at the next 'line' to see if we can close it there
1454                                                 $spliced = array_splice( $a, $k + 1, 1 );
1455                                                 $next_line = array_shift( $spliced );
1456                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1457                                                 # the first ]] closes the inner link, the second the image
1458                                                         $found = true;
1459                                                         $text .= '[[' . $m[1];
1460                                                         $trail = $m[2];
1461                                                         break;
1462                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1463                                                         #if there's exactly one ]] that's fine, we'll keep looking
1464                                                         $text .= '[[' . $m[0];
1465                                                 } else {
1466                                                         #if $next_line is invalid too, we need look no further
1467                                                         $text .= '[[' . $next_line;
1468                                                         break;
1469                                                 }
1470                                         }
1471                                         if ( !$found ) {
1472                                                 # we couldn't find the end of this imageLink, so output it raw
1473                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1474                                                 $text = $this->replaceInternalLinks($text);
1475                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1476                                                 # note: no $trail, because without an end, there *is* no trail
1477                                                 continue;
1478                                         }
1479                                 } else { #it's not an image, so output it raw
1480                                         $s .= $prefix . '[[' . $link . '|' . $text;
1481                                         # note: no $trail, because without an end, there *is* no trail
1482                                         continue;
1483                                 }
1484                         }
1485
1486                         $wasblank = ( '' == $text );
1487                         if( $wasblank ) $text = $link;
1488
1489
1490                         # Link not escaped by : , create the various objects
1491                         if( $noforce ) {
1492
1493                                 # Interwikis
1494                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1495                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1496                                         $s = rtrim($s . "\n");
1497                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1498                                         continue;
1499                                 }
1500
1501                                 if ( $ns == NS_IMAGE ) {
1502                                         wfProfileIn( "$fname-image" );
1503                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1504                                                 # recursively parse links inside the image caption
1505                                                 # actually, this will parse them in any other parameters, too,
1506                                                 # but it might be hard to fix that, and it doesn't matter ATM
1507                                                 $text = $this->replaceExternalLinks($text);
1508                                                 $text = $this->replaceInternalLinks($text);
1509
1510                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1511                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1512                                                 $this->mOutput->addImage( $nt->getDBkey() );
1513
1514                                                 wfProfileOut( "$fname-image" );
1515                                                 continue;
1516                                         }
1517                                         wfProfileOut( "$fname-image" );
1518
1519                                 }
1520
1521                                 if ( $ns == NS_CATEGORY ) {
1522                                         wfProfileIn( "$fname-category" );
1523                                         $s = rtrim($s . "\n"); # bug 87
1524
1525                                         if ( $wasblank ) {
1526                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1527                                                         $sortkey = $this->mTitle->getText();
1528                                                 } else {
1529                                                         $sortkey = $this->mTitle->getPrefixedText();
1530                                                 }
1531                                         } else {
1532                                                 $sortkey = $text;
1533                                         }
1534                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1535                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1536                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1537
1538                                         /**
1539                                          * Strip the whitespace Category links produce, see bug 87
1540                                          * @todo We might want to use trim($tmp, "\n") here.
1541                                          */
1542                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1543
1544                                         wfProfileOut( "$fname-category" );
1545                                         continue;
1546                                 }
1547                         }
1548
1549                         if( ( $nt->getPrefixedText() === $selflink ) &&
1550                             ( $nt->getFragment() === '' ) ) {
1551                                 # Self-links are handled specially; generally de-link and change to bold.
1552                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1553                                 continue;
1554                         }
1555
1556                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1557                         if( $ns == NS_MEDIA ) {
1558                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1559                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1560                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1561                                 $this->mOutput->addImage( $nt->getDBkey() );
1562                                 continue;
1563                         } elseif( $ns == NS_SPECIAL ) {
1564                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1565                                 continue;
1566                         } elseif( $ns == NS_IMAGE ) {
1567                                 $img = Image::newFromTitle( $nt );
1568                                 if( $img->exists() ) {
1569                                         // Force a blue link if the file exists; may be a remote
1570                                         // upload on the shared repository, and we want to see its
1571                                         // auto-generated page.
1572                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1573                                         continue;
1574                                 }
1575                         }
1576                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1577                 }
1578                 wfProfileOut( $fname );
1579                 return $s;
1580         }
1581
1582         /**
1583          * Make a link placeholder. The text returned can be later resolved to a real link with
1584          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1585          * parsing of interwiki links, and secondly to allow all extistence checks and
1586          * article length checks (for stub links) to be bundled into a single query.
1587          *
1588          */
1589         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1590                 if ( ! is_object($nt) ) {
1591                         # Fail gracefully
1592                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1593                 } else {
1594                         # Separate the link trail from the rest of the link
1595                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1596
1597                         if ( $nt->isExternal() ) {
1598                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1599                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1600                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1601                         } else {
1602                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1603                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1604                                 $this->mLinkHolders['queries'][] = $query;
1605                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1606                                 $this->mLinkHolders['titles'][] = $nt;
1607
1608                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1609                         }
1610                 }
1611                 return $retVal;
1612         }
1613
1614         /**
1615          * Render a forced-blue link inline; protect against double expansion of
1616          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1617          * Since this little disaster has to split off the trail text to avoid
1618          * breaking URLs in the following text without breaking trails on the
1619          * wiki links, it's been made into a horrible function.
1620          *
1621          * @param Title $nt
1622          * @param string $text
1623          * @param string $query
1624          * @param string $trail
1625          * @param string $prefix
1626          * @return string HTML-wikitext mix oh yuck
1627          */
1628         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1629                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1630                 $sk =& $this->mOptions->getSkin();
1631                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1632                 return $this->armorLinks( $link ) . $trail;
1633         }
1634
1635         /**
1636          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1637          * going to go through further parsing steps before inline URL expansion.
1638          *
1639          * In particular this is important when using action=render, which causes
1640          * full URLs to be included.
1641          *
1642          * Oh man I hate our multi-layer parser!
1643          *
1644          * @param string more-or-less HTML
1645          * @return string less-or-more HTML with NOPARSE bits
1646          */
1647         function armorLinks( $text ) {
1648                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1649                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1650         }
1651
1652         /**
1653          * Return true if subpage links should be expanded on this page.
1654          * @return bool
1655          */
1656         function areSubpagesAllowed() {
1657                 # Some namespaces don't allow subpages
1658                 global $wgNamespacesWithSubpages;
1659                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1660         }
1661
1662         /**
1663          * Handle link to subpage if necessary
1664          * @param string $target the source of the link
1665          * @param string &$text the link text, modified as necessary
1666          * @return string the full name of the link
1667          * @private
1668          */
1669         function maybeDoSubpageLink($target, &$text) {
1670                 # Valid link forms:
1671                 # Foobar -- normal
1672                 # :Foobar -- override special treatment of prefix (images, language links)
1673                 # /Foobar -- convert to CurrentPage/Foobar
1674                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1675                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1676                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1677
1678                 $fname = 'Parser::maybeDoSubpageLink';
1679                 wfProfileIn( $fname );
1680                 $ret = $target; # default return value is no change
1681
1682                 # Some namespaces don't allow subpages,
1683                 # so only perform processing if subpages are allowed
1684                 if( $this->areSubpagesAllowed() ) {
1685                         # Look at the first character
1686                         if( $target != '' && $target{0} == '/' ) {
1687                                 # / at end means we don't want the slash to be shown
1688                                 if( substr( $target, -1, 1 ) == '/' ) {
1689                                         $target = substr( $target, 1, -1 );
1690                                         $noslash = $target;
1691                                 } else {
1692                                         $noslash = substr( $target, 1 );
1693                                 }
1694
1695                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1696                                 if( '' === $text ) {
1697                                         $text = $target;
1698                                 } # this might be changed for ugliness reasons
1699                         } else {
1700                                 # check for .. subpage backlinks
1701                                 $dotdotcount = 0;
1702                                 $nodotdot = $target;
1703                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1704                                         ++$dotdotcount;
1705                                         $nodotdot = substr( $nodotdot, 3 );
1706                                 }
1707                                 if($dotdotcount > 0) {
1708                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1709                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1710                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1711                                                 # / at the end means don't show full path
1712                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1713                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1714                                                         if( '' === $text ) {
1715                                                                 $text = $nodotdot;
1716                                                         }
1717                                                 }
1718                                                 $nodotdot = trim( $nodotdot );
1719                                                 if( $nodotdot != '' ) {
1720                                                         $ret .= '/' . $nodotdot;
1721                                                 }
1722                                         }
1723                                 }
1724                         }
1725                 }
1726
1727                 wfProfileOut( $fname );
1728                 return $ret;
1729         }
1730
1731         /**#@+
1732          * Used by doBlockLevels()
1733          * @private
1734          */
1735         /* private */ function closeParagraph() {
1736                 $result = '';
1737                 if ( '' != $this->mLastSection ) {
1738                         $result = '</' . $this->mLastSection  . ">\n";
1739                 }
1740                 $this->mInPre = false;
1741                 $this->mLastSection = '';
1742                 return $result;
1743         }
1744         # getCommon() returns the length of the longest common substring
1745         # of both arguments, starting at the beginning of both.
1746         #
1747         /* private */ function getCommon( $st1, $st2 ) {
1748                 $fl = strlen( $st1 );
1749                 $shorter = strlen( $st2 );
1750                 if ( $fl < $shorter ) { $shorter = $fl; }
1751
1752                 for ( $i = 0; $i < $shorter; ++$i ) {
1753                         if ( $st1{$i} != $st2{$i} ) { break; }
1754                 }
1755                 return $i;
1756         }
1757         # These next three functions open, continue, and close the list
1758         # element appropriate to the prefix character passed into them.
1759         #
1760         /* private */ function openList( $char ) {
1761                 $result = $this->closeParagraph();
1762
1763                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1764                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1765                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1766                 else if ( ';' == $char ) {
1767                         $result .= '<dl><dt>';
1768                         $this->mDTopen = true;
1769                 }
1770                 else { $result = '<!-- ERR 1 -->'; }
1771
1772                 return $result;
1773         }
1774
1775         /* private */ function nextItem( $char ) {
1776                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1777                 else if ( ':' == $char || ';' == $char ) {
1778                         $close = '</dd>';
1779                         if ( $this->mDTopen ) { $close = '</dt>'; }
1780                         if ( ';' == $char ) {
1781                                 $this->mDTopen = true;
1782                                 return $close . '<dt>';
1783                         } else {
1784                                 $this->mDTopen = false;
1785                                 return $close . '<dd>';
1786                         }
1787                 }
1788                 return '<!-- ERR 2 -->';
1789         }
1790
1791         /* private */ function closeList( $char ) {
1792                 if ( '*' == $char ) { $text = '</li></ul>'; }
1793                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1794                 else if ( ':' == $char ) {
1795                         if ( $this->mDTopen ) {
1796                                 $this->mDTopen = false;
1797                                 $text = '</dt></dl>';
1798                         } else {
1799                                 $text = '</dd></dl>';
1800                         }
1801                 }
1802                 else {  return '<!-- ERR 3 -->'; }
1803                 return $text."\n";
1804         }
1805         /**#@-*/
1806
1807         /**
1808          * Make lists from lines starting with ':', '*', '#', etc.
1809          *
1810          * @private
1811          * @return string the lists rendered as HTML
1812          */
1813         function doBlockLevels( $text, $linestart ) {
1814                 $fname = 'Parser::doBlockLevels';
1815                 wfProfileIn( $fname );
1816
1817                 # Parsing through the text line by line.  The main thing
1818                 # happening here is handling of block-level elements p, pre,
1819                 # and making lists from lines starting with * # : etc.
1820                 #
1821                 $textLines = explode( "\n", $text );
1822
1823                 $lastPrefix = $output = '';
1824                 $this->mDTopen = $inBlockElem = false;
1825                 $prefixLength = 0;
1826                 $paragraphStack = false;
1827
1828                 if ( !$linestart ) {
1829                         $output .= array_shift( $textLines );
1830                 }
1831                 foreach ( $textLines as $oLine ) {
1832                         $lastPrefixLength = strlen( $lastPrefix );
1833                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1834                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1835                         if ( !$this->mInPre ) {
1836                                 # Multiple prefixes may abut each other for nested lists.
1837                                 $prefixLength = strspn( $oLine, '*#:;' );
1838                                 $pref = substr( $oLine, 0, $prefixLength );
1839
1840                                 # eh?
1841                                 $pref2 = str_replace( ';', ':', $pref );
1842                                 $t = substr( $oLine, $prefixLength );
1843                                 $this->mInPre = !empty($preOpenMatch);
1844                         } else {
1845                                 # Don't interpret any other prefixes in preformatted text
1846                                 $prefixLength = 0;
1847                                 $pref = $pref2 = '';
1848                                 $t = $oLine;
1849                         }
1850
1851                         # List generation
1852                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1853                                 # Same as the last item, so no need to deal with nesting or opening stuff
1854                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1855                                 $paragraphStack = false;
1856
1857                                 if ( substr( $pref, -1 ) == ';') {
1858                                         # The one nasty exception: definition lists work like this:
1859                                         # ; title : definition text
1860                                         # So we check for : in the remainder text to split up the
1861                                         # title and definition, without b0rking links.
1862                                         $term = $t2 = '';
1863                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1864                                                 $t = $t2;
1865                                                 $output .= $term . $this->nextItem( ':' );
1866                                         }
1867                                 }
1868                         } elseif( $prefixLength || $lastPrefixLength ) {
1869                                 # Either open or close a level...
1870                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1871                                 $paragraphStack = false;
1872
1873                                 while( $commonPrefixLength < $lastPrefixLength ) {
1874                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1875                                         --$lastPrefixLength;
1876                                 }
1877                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1878                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1879                                 }
1880                                 while ( $prefixLength > $commonPrefixLength ) {
1881                                         $char = substr( $pref, $commonPrefixLength, 1 );
1882                                         $output .= $this->openList( $char );
1883
1884                                         if ( ';' == $char ) {
1885                                                 # FIXME: This is dupe of code above
1886                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1887                                                         $t = $t2;
1888                                                         $output .= $term . $this->nextItem( ':' );
1889                                                 }
1890                                         }
1891                                         ++$commonPrefixLength;
1892                                 }
1893                                 $lastPrefix = $pref2;
1894                         }
1895                         if( 0 == $prefixLength ) {
1896                                 wfProfileIn( "$fname-paragraph" );
1897                                 # No prefix (not in list)--go to paragraph mode
1898                                 // XXX: use a stack for nestable elements like span, table and div
1899                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1900                                 $closematch = preg_match(
1901                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1902                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1903                                 if ( $openmatch or $closematch ) {
1904                                         $paragraphStack = false;
1905                                         # TODO bug 5718: paragraph closed
1906                                         $output .= $this->closeParagraph();
1907                                         if ( $preOpenMatch and !$preCloseMatch ) {
1908                                                 $this->mInPre = true;
1909                                         }
1910                                         if ( $closematch ) {
1911                                                 $inBlockElem = false;
1912                                         } else {
1913                                                 $inBlockElem = true;
1914                                         }
1915                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1916                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1917                                                 // pre
1918                                                 if ($this->mLastSection != 'pre') {
1919                                                         $paragraphStack = false;
1920                                                         $output .= $this->closeParagraph().'<pre>';
1921                                                         $this->mLastSection = 'pre';
1922                                                 }
1923                                                 $t = substr( $t, 1 );
1924                                         } else {
1925                                                 // paragraph
1926                                                 if ( '' == trim($t) ) {
1927                                                         if ( $paragraphStack ) {
1928                                                                 $output .= $paragraphStack.'<br />';
1929                                                                 $paragraphStack = false;
1930                                                                 $this->mLastSection = 'p';
1931                                                         } else {
1932                                                                 if ($this->mLastSection != 'p' ) {
1933                                                                         $output .= $this->closeParagraph();
1934                                                                         $this->mLastSection = '';
1935                                                                         $paragraphStack = '<p>';
1936                                                                 } else {
1937                                                                         $paragraphStack = '</p><p>';
1938                                                                 }
1939                                                         }
1940                                                 } else {
1941                                                         if ( $paragraphStack ) {
1942                                                                 $output .= $paragraphStack;
1943                                                                 $paragraphStack = false;
1944                                                                 $this->mLastSection = 'p';
1945                                                         } else if ($this->mLastSection != 'p') {
1946                                                                 $output .= $this->closeParagraph().'<p>';
1947                                                                 $this->mLastSection = 'p';
1948                                                         }
1949                                                 }
1950                                         }
1951                                 }
1952                                 wfProfileOut( "$fname-paragraph" );
1953                         }
1954                         // somewhere above we forget to get out of pre block (bug 785)
1955                         if($preCloseMatch && $this->mInPre) {
1956                                 $this->mInPre = false;
1957                         }
1958                         if ($paragraphStack === false) {
1959                                 $output .= $t."\n";
1960                         }
1961                 }
1962                 while ( $prefixLength ) {
1963                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1964                         --$prefixLength;
1965                 }
1966                 if ( '' != $this->mLastSection ) {
1967                         $output .= '</' . $this->mLastSection . '>';
1968                         $this->mLastSection = '';
1969                 }
1970
1971                 wfProfileOut( $fname );
1972                 return $output;
1973         }
1974
1975         /**
1976          * Split up a string on ':', ignoring any occurences inside tags
1977          * to prevent illegal overlapping.
1978          * @param string $str the string to split
1979          * @param string &$before set to everything before the ':'
1980          * @param string &$after set to everything after the ':'
1981          * return string the position of the ':', or false if none found
1982          */
1983         function findColonNoLinks($str, &$before, &$after) {
1984                 $fname = 'Parser::findColonNoLinks';
1985                 wfProfileIn( $fname );
1986
1987                 $pos = strpos( $str, ':' );
1988                 if( $pos === false ) {
1989                         // Nothing to find!
1990                         wfProfileOut( $fname );
1991                         return false;
1992                 }
1993
1994                 if( strpos( $str, '<' ) === false ) {
1995                         // Easy; no tag nesting to worry about
1996                         $before = substr( $str, 0, $pos );
1997                         $after = substr( $str, $pos+1 );
1998                         wfProfileOut( $fname );
1999                         return $pos;
2000                 }
2001
2002                 // Ugly state machine to walk through avoiding tags.
2003                 $state = MW_COLON_STATE_TEXT;
2004                 $stack = 0;
2005                 $len = strlen( $str );
2006                 for( $i = 0; $i < $len; $i++ ) {
2007                         $c = $str{$i};
2008
2009                         switch( $state ) {
2010                         // (Using the number is a performance hack for common cases)
2011                         case 0: // MW_COLON_STATE_TEXT:
2012                                 switch( $c ) {
2013                                 case "<":
2014                                         // Could be either a <start> tag or an </end> tag
2015                                         $state = MW_COLON_STATE_TAGSTART;
2016                                         break;
2017                                 case ":":
2018                                         if( $stack == 0 ) {
2019                                                 // We found it!
2020                                                 $before = substr( $str, 0, $i );
2021                                                 $after = substr( $str, $i + 1 );
2022                                                 wfProfileOut( $fname );
2023                                                 return $i;
2024                                         }
2025                                         // Embedded in a tag; don't break it.
2026                                         break;
2027                                 default:
2028                                         // ignore
2029                                 }
2030                                 break;
2031                         case 1: // MW_COLON_STATE_TAG:
2032                                 // In a <tag>
2033                                 switch( $c ) {
2034                                 case ">":
2035                                         $stack++;
2036                                         $state = MW_COLON_STATE_TEXT;
2037                                         break;
2038                                 case "/":
2039                                         // Slash may be followed by >?
2040                                         $state = MW_COLON_STATE_TAGSLASH;
2041                                         break;
2042                                 default:
2043                                         // ignore
2044                                 }
2045                                 break;
2046                         case 2: // MW_COLON_STATE_TAGSTART:
2047                                 switch( $c ) {
2048                                 case "/":
2049                                         $state = MW_COLON_STATE_CLOSETAG;
2050                                         break;
2051                                 case "!":
2052                                         $state = MW_COLON_STATE_COMMENT;
2053                                         break;
2054                                 case ">":
2055                                         // Illegal early close? This shouldn't happen D:
2056                                         $state = MW_COLON_STATE_TEXT;
2057                                         break;
2058                                 default:
2059                                         $state = MW_COLON_STATE_TAG;
2060                                 }
2061                                 break;
2062                         case 3: // MW_COLON_STATE_CLOSETAG:
2063                                 // In a </tag>
2064                                 if( $c == ">" ) {
2065                                         $stack--;
2066                                         if( $stack < 0 ) {
2067                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2068                                                 wfProfileOut( $fname );
2069                                                 return false;
2070                                         }
2071                                         $state = MW_COLON_STATE_TEXT;
2072                                 }
2073                                 break;
2074                         case MW_COLON_STATE_TAGSLASH:
2075                                 if( $c == ">" ) {
2076                                         // Yes, a self-closed tag <blah/>
2077                                         $state = MW_COLON_STATE_TEXT;
2078                                 } else {
2079                                         // Probably we're jumping the gun, and this is an attribute
2080                                         $state = MW_COLON_STATE_TAG;
2081                                 }
2082                                 break;
2083                         case 5: // MW_COLON_STATE_COMMENT:
2084                                 if( $c == "-" ) {
2085                                         $state = MW_COLON_STATE_COMMENTDASH;
2086                                 }
2087                                 break;
2088                         case MW_COLON_STATE_COMMENTDASH:
2089                                 if( $c == "-" ) {
2090                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2091                                 } else {
2092                                         $state = MW_COLON_STATE_COMMENT;
2093                                 }
2094                                 break;
2095                         case MW_COLON_STATE_COMMENTDASHDASH:
2096                                 if( $c == ">" ) {
2097                                         $state = MW_COLON_STATE_TEXT;
2098                                 } else {
2099                                         $state = MW_COLON_STATE_COMMENT;
2100                                 }
2101                                 break;
2102                         default:
2103                                 wfDebugDieBacktrace( "State machine error in $fname" );
2104                         }
2105                 }
2106                 if( $stack > 0 ) {
2107                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2108                         return false;
2109                 }
2110                 wfProfileOut( $fname );
2111                 return false;
2112         }
2113
2114         /**
2115          * Return value of a magic variable (like PAGENAME)
2116          *
2117          * @private
2118          */
2119         function getVariableValue( $index ) {
2120                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2121
2122                 /**
2123                  * Some of these require message or data lookups and can be
2124                  * expensive to check many times.
2125                  */
2126                 static $varCache = array();
2127                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2128                         if ( isset( $varCache[$index] ) )
2129                                 return $varCache[$index];
2130
2131                 $ts = time();
2132                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2133
2134                 switch ( $index ) {
2135                         case MAG_CURRENTMONTH:
2136                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2137                         case MAG_CURRENTMONTHNAME:
2138                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2139                         case MAG_CURRENTMONTHNAMEGEN:
2140                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2141                         case MAG_CURRENTMONTHABBREV:
2142                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2143                         case MAG_CURRENTDAY:
2144                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2145                         case MAG_CURRENTDAY2:
2146                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2147                         case MAG_PAGENAME:
2148                                 return $this->mTitle->getText();
2149                         case MAG_PAGENAMEE:
2150                                 return $this->mTitle->getPartialURL();
2151                         case MAG_FULLPAGENAME:
2152                                 return $this->mTitle->getPrefixedText();
2153                         case MAG_FULLPAGENAMEE:
2154                                 return $this->mTitle->getPrefixedURL();
2155                         case MAG_SUBPAGENAME:
2156                                 return $this->mTitle->getSubpageText();
2157                         case MAG_SUBPAGENAMEE:
2158                                 return $this->mTitle->getSubpageUrlForm();
2159                         case MAG_BASEPAGENAME:
2160                                 return $this->mTitle->getBaseText();
2161                         case MAG_BASEPAGENAMEE:
2162                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2163                         case MAG_TALKPAGENAME:
2164                                 if( $this->mTitle->canTalk() ) {
2165                                         $talkPage = $this->mTitle->getTalkPage();
2166                                         return $talkPage->getPrefixedText();
2167                                 } else {
2168                                         return '';
2169                                 }
2170                         case MAG_TALKPAGENAMEE:
2171                                 if( $this->mTitle->canTalk() ) {
2172                                         $talkPage = $this->mTitle->getTalkPage();
2173                                         return $talkPage->getPrefixedUrl();
2174                                 } else {
2175                                         return '';
2176                                 }
2177                         case MAG_SUBJECTPAGENAME:
2178                                 $subjPage = $this->mTitle->getSubjectPage();
2179                                 return $subjPage->getPrefixedText();
2180                         case MAG_SUBJECTPAGENAMEE:
2181                                 $subjPage = $this->mTitle->getSubjectPage();
2182                                 return $subjPage->getPrefixedUrl();
2183                         case MAG_REVISIONID:
2184                                 return $this->mRevisionId;
2185                         case MAG_NAMESPACE:
2186                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2187                         case MAG_NAMESPACEE:
2188                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2189                         case MAG_TALKSPACE:
2190                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2191                         case MAG_TALKSPACEE:
2192                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2193                         case MAG_SUBJECTSPACE:
2194                                 return $this->mTitle->getSubjectNsText();
2195                         case MAG_SUBJECTSPACEE:
2196                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2197                         case MAG_CURRENTDAYNAME:
2198                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2199                         case MAG_CURRENTYEAR:
2200                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2201                         case MAG_CURRENTTIME:
2202                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2203                         case MAG_CURRENTWEEK:
2204                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2205                                 // int to remove the padding
2206                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2207                         case MAG_CURRENTDOW:
2208                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2209                         case MAG_NUMBEROFARTICLES:
2210                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2211                         case MAG_NUMBEROFFILES:
2212                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2213                         case MAG_NUMBEROFUSERS:
2214                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2215                         case MAG_NUMBEROFPAGES:
2216                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2217                         case MAG_CURRENTTIMESTAMP:
2218                                 return $varCache[$index] = wfTimestampNow();
2219                         case MAG_CURRENTVERSION:
2220                                 global $wgVersion;
2221                                 return $wgVersion;
2222                         case MAG_SITENAME:
2223                                 return $wgSitename;
2224                         case MAG_SERVER:
2225                                 return $wgServer;
2226                         case MAG_SERVERNAME:
2227                                 return $wgServerName;
2228                         case MAG_SCRIPTPATH:
2229                                 return $wgScriptPath;
2230                         case MAG_DIRECTIONMARK:
2231                                 return $wgContLang->getDirMark();
2232                         default:
2233                                 $ret = null;
2234                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2235                                         return $ret;
2236                                 else
2237                                         return null;
2238                 }
2239         }
2240
2241         /**
2242          * initialise the magic variables (like CURRENTMONTHNAME)
2243          *
2244          * @private
2245          */
2246         function initialiseVariables() {
2247                 $fname = 'Parser::initialiseVariables';
2248                 wfProfileIn( $fname );
2249                 global $wgVariableIDs;
2250                 $this->mVariables = array();
2251                 foreach ( $wgVariableIDs as $id ) {
2252                         $mw =& MagicWord::get( $id );
2253                         $mw->addToArray( $this->mVariables, $id );
2254                 }
2255                 wfProfileOut( $fname );
2256         }
2257
2258         /**
2259          * parse any parentheses in format ((title|part|part))
2260          * and call callbacks to get a replacement text for any found piece
2261          *
2262          * @param string $text The text to parse
2263          * @param array $callbacks rules in form:
2264          *     '{' => array(                            # opening parentheses
2265          *                                      'end' => '}',   # closing parentheses
2266          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2267          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2268          *                                                                )
2269          *                                      )
2270          * @private
2271          */
2272         function replace_callback ($text, $callbacks) {
2273                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2274                 $lastOpeningBrace = -1;                 # last not closed parentheses
2275
2276                 for ($i = 0; $i < strlen($text); $i++) {
2277                         # check for any opening brace
2278                         $rule = null;
2279                         $nextPos = -1;
2280                         foreach ($callbacks as $key => $value) {
2281                                 $pos = strpos ($text, $key, $i);
2282                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2283                                         $rule = $value;
2284                                         $nextPos = $pos;
2285                                 }
2286                         }
2287
2288                         if ($lastOpeningBrace >= 0) {
2289                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2290
2291                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2292                                         $rule = null;
2293                                         $nextPos = $pos;
2294                                 }
2295
2296                                 $pos = strpos ($text, '|', $i);
2297
2298                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2299                                         $rule = null;
2300                                         $nextPos = $pos;
2301                                 }
2302                         }
2303
2304                         if ($nextPos == -1)
2305                                 break;
2306
2307                         $i = $nextPos;
2308
2309                         # found openning brace, lets add it to parentheses stack
2310                         if (null != $rule) {
2311                                 $piece = array('brace' => $text[$i],
2312                                                            'braceEnd' => $rule['end'],
2313                                                            'count' => 1,
2314                                                            'title' => '',
2315                                                            'parts' => null);
2316
2317                                 # count openning brace characters
2318                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2319                                         $piece['count']++;
2320                                         $i++;
2321                                 }
2322
2323                                 $piece['startAt'] = $i+1;
2324                                 $piece['partStart'] = $i+1;
2325
2326                                 # we need to add to stack only if openning brace count is enough for any given rule
2327                                 foreach ($rule['cb'] as $cnt => $fn) {
2328                                         if ($piece['count'] >= $cnt) {
2329                                                 $lastOpeningBrace ++;
2330                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2331                                                 break;
2332                                         }
2333                                 }
2334
2335                                 continue;
2336                         }
2337                         else if ($lastOpeningBrace >= 0) {
2338                                 # first check if it is a closing brace
2339                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2340                                         # lets check if it is enough characters for closing brace
2341                                         $count = 1;
2342                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2343                                                 $count++;
2344
2345                                         # if there are more closing parentheses than opening ones, we parse less
2346                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2347                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2348
2349                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2350                                         $matchingCount = 0;
2351                                         $matchingCallback = null;
2352                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2353                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2354                                                         $matchingCount = $cnt;
2355                                                         $matchingCallback = $fn;
2356                                                 }
2357                                         }
2358
2359                                         if ($matchingCount == 0) {
2360                                                 $i += $count - 1;
2361                                                 continue;
2362                                         }
2363
2364                                         # lets set a title or last part (if '|' was found)
2365                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2366                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2367                                         else
2368                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2369
2370                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2371                                         $pieceEnd = $i + $matchingCount;
2372
2373                                         if( is_callable( $matchingCallback ) ) {
2374                                                 $cbArgs = array (
2375                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2376                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2377                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2378                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2379                                                                                  );
2380                                                 # finally we can call a user callback and replace piece of text
2381                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2382                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2383                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2384                                         }
2385                                         else {
2386                                                 # null value for callback means that parentheses should be parsed, but not replaced
2387                                                 $i += $matchingCount - 1;
2388                                         }
2389
2390                                         # reset last openning parentheses, but keep it in case there are unused characters
2391                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2392                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2393                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2394                                                                    'title' => '',
2395                                                                    'parts' => null,
2396                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2397                                         $openingBraceStack[$lastOpeningBrace--] = null;
2398
2399                                         if ($matchingCount < $piece['count']) {
2400                                                 $piece['count'] -= $matchingCount;
2401                                                 $piece['startAt'] -= $matchingCount;
2402                                                 $piece['partStart'] = $piece['startAt'];
2403                                                 # do we still qualify for any callback with remaining count?
2404                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2405                                                         if ($piece['count'] >= $cnt) {
2406                                                                 $lastOpeningBrace ++;
2407                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2408                                                                 break;
2409                                                         }
2410                                                 }
2411                                         }
2412                                         continue;
2413                                 }
2414
2415                                 # lets set a title if it is a first separator, or next part otherwise
2416                                 if ($text[$i] == '|') {
2417                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2418                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2419                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2420                                         }
2421                                         else
2422                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2423
2424                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2425                                 }
2426                         }
2427                 }
2428
2429                 return $text;
2430         }
2431
2432         /**
2433          * Replace magic variables, templates, and template arguments
2434          * with the appropriate text. Templates are substituted recursively,
2435          * taking care to avoid infinite loops.
2436          *
2437          * Note that the substitution depends on value of $mOutputType:
2438          *  OT_WIKI: only {{subst:}} templates
2439          *  OT_MSG: only magic variables
2440          *  OT_HTML: all templates and magic variables
2441          *
2442          * @param string $tex The text to transform
2443          * @param array $args Key-value pairs representing template parameters to substitute
2444          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2445          * @private
2446          */
2447         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2448                 # Prevent too big inclusions
2449                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2450                         return $text;
2451                 }
2452
2453                 $fname = 'Parser::replaceVariables';
2454                 wfProfileIn( $fname );
2455
2456                 # This function is called recursively. To keep track of arguments we need a stack:
2457                 array_push( $this->mArgStack, $args );
2458
2459                 $braceCallbacks = array();
2460                 if ( !$argsOnly ) {
2461                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2462                 }
2463                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2464                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2465                 }
2466                 $callbacks = array();
2467                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2468                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2469                 $text = $this->replace_callback ($text, $callbacks);
2470
2471                 array_pop( $this->mArgStack );
2472
2473                 wfProfileOut( $fname );
2474                 return $text;
2475         }
2476
2477         /**
2478          * Replace magic variables
2479          * @private
2480          */
2481         function variableSubstitution( $matches ) {
2482                 $fname = 'Parser::variableSubstitution';
2483                 $varname = $matches[1];
2484                 wfProfileIn( $fname );
2485                 if ( !$this->mVariables ) {
2486                         $this->initialiseVariables();
2487                 }
2488                 $skip = false;
2489                 if ( $this->mOutputType == OT_WIKI ) {
2490                         # Do only magic variables prefixed by SUBST
2491                         $mwSubst =& MagicWord::get( MAG_SUBST );
2492                         if (!$mwSubst->matchStartAndRemove( $varname ))
2493                                 $skip = true;
2494                         # Note that if we don't substitute the variable below,
2495                         # we don't remove the {{subst:}} magic word, in case
2496                         # it is a template rather than a magic variable.
2497                 }
2498                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2499                         $id = $this->mVariables[$varname];
2500                         $text = $this->getVariableValue( $id );
2501                         $this->mOutput->mContainsOldMagic = true;
2502                 } else {
2503                         $text = $matches[0];
2504                 }
2505                 wfProfileOut( $fname );
2506                 return $text;
2507         }
2508
2509         # Split template arguments
2510         function getTemplateArgs( $argsString ) {
2511                 if ( $argsString === '' ) {
2512                         return array();
2513                 }
2514
2515                 $args = explode( '|', substr( $argsString, 1 ) );
2516
2517                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2518                 # merged with the next arg because the '|' character between belongs
2519                 # to the link syntax and not the template parameter syntax.
2520                 $argc = count($args);
2521
2522                 for ( $i = 0; $i < $argc-1; $i++ ) {
2523                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2524                                 $args[$i] .= '|'.$args[$i+1];
2525                                 array_splice($args, $i+1, 1);
2526                                 $i--;
2527                                 $argc--;
2528                         }
2529                 }
2530
2531                 return $args;
2532         }
2533
2534         /**
2535          * Return the text of a template, after recursively
2536          * replacing any variables or templates within the template.
2537          *
2538          * @param array $piece The parts of the template
2539          *  $piece['text']: matched text
2540          *  $piece['title']: the title, i.e. the part before the |
2541          *  $piece['parts']: the parameter array
2542          * @return string the text of the template
2543          * @private
2544          */
2545         function braceSubstitution( $piece ) {
2546                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2547                 $fname = 'Parser::braceSubstitution';
2548                 wfProfileIn( $fname );
2549
2550                 # Flags
2551                 $found = false;             # $text has been filled
2552                 $nowiki = false;            # wiki markup in $text should be escaped
2553                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2554                 $noargs = false;            # Don't replace triple-brace arguments in $text
2555                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2556                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2557                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2558
2559                 # Title object, where $text came from
2560                 $title = NULL;
2561
2562                 $linestart = '';
2563
2564                 # $part1 is the bit before the first |, and must contain only title characters
2565                 # $args is a list of arguments, starting from index 0, not including $part1
2566
2567                 $part1 = $piece['title'];
2568                 # If the third subpattern matched anything, it will start with |
2569
2570                 if (null == $piece['parts']) {
2571                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2572                         if ($replaceWith != $piece['text']) {
2573                                 $text = $replaceWith;
2574                                 $found = true;
2575                                 $noparse = true;
2576                                 $noargs = true;
2577                         }
2578                 }
2579
2580                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2581                 $argc = count( $args );
2582
2583                 # SUBST
2584                 if ( !$found ) {
2585                         $mwSubst =& MagicWord::get( MAG_SUBST );
2586                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2587                                 # One of two possibilities is true:
2588                                 # 1) Found SUBST but not in the PST phase
2589                                 # 2) Didn't find SUBST and in the PST phase
2590                                 # In either case, return without further processing
2591                                 $text = $piece['text'];
2592                                 $found = true;
2593                                 $noparse = true;
2594                                 $noargs = true;
2595                         }
2596                 }
2597
2598                 # MSG, MSGNW, INT and RAW
2599                 if ( !$found ) {
2600                         # Check for MSGNW:
2601                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2602                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2603                                 $nowiki = true;
2604                         } else {
2605                                 # Remove obsolete MSG:
2606                                 $mwMsg =& MagicWord::get( MAG_MSG );
2607                                 $mwMsg->matchStartAndRemove( $part1 );
2608                         }
2609
2610                         # Check for RAW:
2611                         $mwRaw =& MagicWord::get( MAG_RAW );
2612                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2613                                 $forceRawInterwiki = true;
2614                         }
2615
2616                         # Check if it is an internal message
2617                         $mwInt =& MagicWord::get( MAG_INT );
2618                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2619                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2620                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2621                                         $found = true;
2622                                 }
2623                         }
2624                 }
2625
2626                 # NS
2627                 if ( !$found ) {
2628                         # Check for NS: (namespace expansion)
2629                         $mwNs = MagicWord::get( MAG_NS );
2630                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2631                                 if ( intval( $part1 ) || $part1 == "0" ) {
2632                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2633                                         $found = true;
2634                                 } else {
2635                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2636                                         if ( !is_null( $index ) ) {
2637                                                 $text = $linestart . $wgContLang->getNsText( $index );
2638                                                 $found = true;
2639                                         }
2640                                 }
2641                         }
2642                 }
2643
2644                 # URLENCODE
2645                 if( !$found ) {
2646                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2647                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2648                                 $text = $linestart . urlencode( $part1 );
2649                                 $found = true;
2650                         }
2651                 }
2652
2653                 # LCFIRST, UCFIRST, LC and UC
2654                 if ( !$found ) {
2655                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2656                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2657                         $lc =& MagicWord::get( MAG_LC );
2658                         $uc =& MagicWord::get( MAG_UC );
2659                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2660                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2661                                 $found = true;
2662                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2663                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2664                                 $found = true;
2665                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2666                                 $text = $linestart . $wgContLang->lc( $part1 );
2667                                 $found = true;
2668                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2669                                  $text = $linestart . $wgContLang->uc( $part1 );
2670                                  $found = true;
2671                         }
2672                 }
2673
2674                 # LOCALURL and FULLURL
2675                 if ( !$found ) {
2676                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2677                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2678                         $mwFull =& MagicWord::get( MAG_FULLURL );
2679                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2680
2681
2682                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2683                                 $func = 'getLocalURL';
2684                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2685                                 $func = 'escapeLocalURL';
2686                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2687                                 $func = 'getFullURL';
2688                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2689                                 $func = 'escapeFullURL';
2690                         } else {
2691                                 $func = false;
2692                         }
2693
2694                         if ( $func !== false ) {
2695                                 $title = Title::newFromText( $part1 );
2696                                 # Due to order of execution of a lot of bits, the values might be encoded
2697                                 # before arriving here; if that's true, then the title can't be created
2698                                 # and the variable will fail. If we can't get a decent title from the first
2699                                 # attempt, url-decode and try for a second.
2700                                 if( is_null( $title ) )
2701                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2702                                 if ( !is_null( $title ) ) {
2703                                         if ( $argc > 0 ) {
2704                                                 $text = $linestart . $title->$func( $args[0] );
2705                                         } else {
2706                                                 $text = $linestart . $title->$func();
2707                                         }
2708                                         $found = true;
2709                                 }
2710                         }
2711                 }
2712
2713                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2714                 # GRAMMAR
2715                 if ( !$found && $argc == 1 ) {
2716                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2717                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2718                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2719                                 $found = true;
2720                         }
2721                 }
2722
2723                 # PLURAL
2724                 if ( !$found && $argc >= 2 ) {
2725                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2726                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2727                                 if ($argc==2) {$args[2]=$args[1];}
2728                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2729                                 $found = true;
2730                         }
2731                 }
2732
2733                 # DISPLAYTITLE
2734                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2735                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2736                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2737
2738                                 # Set title in parser output object
2739                                 $param = $args[0];
2740                                 $parserOptions = new ParserOptions;
2741                                 $local_parser = new Parser ();
2742                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2743                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2744
2745                                 # Add subtitle
2746                                 $t = $this->mTitle->getPrefixedText();
2747                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2748                                 $text = "" ;
2749                                 $found = true ;
2750                         }
2751                 }
2752
2753                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2754                 if( !$found ) {
2755                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2756                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2757                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2758                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2759                         foreach( $mwWordsToCheck as $word => $func ) {
2760                                 $mwCurrentWord =& MagicWord::get( $word );
2761                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2762                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2763                                         if( $mwRawSuffix->match( $args[0] ) ) {
2764                                                 # Raw and unformatted
2765                                                 $text = $linestart . call_user_func( $func );
2766                                         } else {
2767                                                 # Formatted according to the content default
2768                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2769                                         }
2770                                         $found = true;
2771                                 }
2772                         }
2773                 }
2774
2775                 # #LANGUAGE:
2776                 if( !$found ) {
2777                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2778                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2779                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2780                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2781                                 $found = true;
2782                         }
2783                 }
2784
2785                 # Extensions
2786                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2787                         $colonPos = strpos( $part1, ':' );
2788                         if ( $colonPos !== false ) {
2789                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2790                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2791                                         $funcArgs = array_map( 'trim', $args );
2792                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2793                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2794                                         $found = true;
2795
2796                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2797                                         //$noargs = true;
2798                                         //$noparse = true;
2799
2800                                         if ( is_array( $result ) ) {
2801                                                 $text = $linestart . $result[0];
2802                                                 unset( $result[0] );
2803
2804                                                 // Extract flags into the local scope
2805                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2806                                                 extract( $result );
2807                                         } else {
2808                                                 $text = $linestart . $result;
2809                                         }
2810                                 }
2811                         }
2812                 }
2813
2814                 # Template table test
2815
2816                 # Did we encounter this template already? If yes, it is in the cache
2817                 # and we need to check for loops.
2818                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2819                         $found = true;
2820
2821                         # Infinite loop test
2822                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2823                                 $noparse = true;
2824                                 $noargs = true;
2825                                 $found = true;
2826                                 $text = $linestart .
2827                                         '{{' . $part1 . '}}' .
2828                                         '<!-- WARNING: template loop detected -->';
2829                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2830                         } else {
2831                                 # set $text to cached message.
2832                                 $text = $linestart . $this->mTemplates[$piece['title']];
2833                         }
2834                 }
2835
2836                 # Load from database
2837                 $lastPathLevel = $this->mTemplatePath;
2838                 if ( !$found ) {
2839                         $ns = NS_TEMPLATE;
2840                         # declaring $subpage directly in the function call
2841                         # does not work correctly with references and breaks
2842                         # {{/subpage}}-style inclusions
2843                         $subpage = '';
2844                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2845                         if ($subpage !== '') {
2846                                 $ns = $this->mTitle->getNamespace();
2847                         }
2848                         $title = Title::newFromText( $part1, $ns );
2849
2850                         if ( !is_null( $title ) ) {
2851                                 if ( !$title->isExternal() ) {
2852                                         # Check for excessive inclusion
2853                                         $dbk = $title->getPrefixedDBkey();
2854                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2855                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2856                                                         $text = SpecialPage::capturePath( $title );
2857                                                         if ( is_string( $text ) ) {
2858                                                                 $found = true;
2859                                                                 $noparse = true;
2860                                                                 $noargs = true;
2861                                                                 $isHTML = true;
2862                                                                 $this->disableCache();
2863                                                         }
2864                                                 } else {
2865                                                         $articleContent = $this->fetchTemplate( $title );
2866                                                         if ( $articleContent !== false ) {
2867                                                                 $found = true;
2868                                                                 $text = $articleContent;
2869                                                                 $replaceHeadings = true;
2870                                                         }
2871                                                 }
2872                                         }
2873
2874                                         # If the title is valid but undisplayable, make a link to it
2875                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2876                                                 $text = '[['.$title->getPrefixedText().']]';
2877                                                 $found = true;
2878                                         }
2879                                 } elseif ( $title->isTrans() ) {
2880                                         // Interwiki transclusion
2881                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2882                                                 $text = $this->interwikiTransclude( $title, 'render' );
2883                                                 $isHTML = true;
2884                                                 $noparse = true;
2885                                         } else {
2886                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2887                                                 $replaceHeadings = true;
2888                                         }
2889                                         $found = true;
2890                                 }
2891
2892                                 # Template cache array insertion
2893                                 # Use the original $piece['title'] not the mangled $part1, so that
2894                                 # modifiers such as RAW: produce separate cache entries
2895                                 if( $found ) {
2896                                         $this->mTemplates[$piece['title']] = $text;
2897                                         $text = $linestart . $text;
2898                                 }
2899                         }
2900                 }
2901
2902                 # Recursive parsing, escaping and link table handling
2903                 # Only for HTML output
2904                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2905                         $text = wfEscapeWikiText( $text );
2906                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2907                         if ( $noargs ) {
2908                                 $assocArgs = array();
2909                         } else {
2910                                 # Clean up argument array
2911                                 $assocArgs = array();
2912                                 $index = 1;
2913                                 foreach( $args as $arg ) {
2914                                         $eqpos = strpos( $arg, '=' );
2915                                         if ( $eqpos === false ) {
2916                                                 $assocArgs[$index++] = $arg;
2917                                         } else {
2918                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2919                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2920                                                 if ( $value === false ) {
2921                                                         $value = '';
2922                                                 }
2923                                                 if ( $name !== false ) {
2924                                                         $assocArgs[$name] = $value;
2925                                                 }
2926                                         }
2927                                 }
2928
2929                                 # Add a new element to the templace recursion path
2930                                 $this->mTemplatePath[$part1] = 1;
2931                         }
2932
2933                         if ( !$noparse ) {
2934                                 # If there are any <onlyinclude> tags, only include them
2935                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2936                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2937                                         $text = '';
2938                                         foreach ($m[1] as $piece)
2939                                                 $text .= $piece;
2940                                 }
2941                                 # Remove <noinclude> sections and <includeonly> tags
2942                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2943                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2944
2945                                 if( $this->mOutputType == OT_HTML ) {
2946                                         # Strip <nowiki>, <pre>, etc.
2947                                         $text = $this->strip( $text, $this->mStripState );
2948                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2949                                 }
2950                                 $text = $this->replaceVariables( $text, $assocArgs );
2951
2952                                 # If the template begins with a table or block-level
2953                                 # element, it should be treated as beginning a new line.
2954                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2955                                         $text = "\n" . $text;
2956                                 }
2957                         } elseif ( !$noargs ) {
2958                                 # $noparse and !$noargs
2959                                 # Just replace the arguments, not any double-brace items
2960                                 # This is used for rendered interwiki transclusion
2961                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2962                         }
2963                 }
2964                 # Prune lower levels off the recursion check path
2965                 $this->mTemplatePath = $lastPathLevel;
2966
2967                 if ( !$found ) {
2968                         wfProfileOut( $fname );
2969                         return $piece['text'];
2970                 } else {
2971                         if ( $isHTML ) {
2972                                 # Replace raw HTML by a placeholder
2973                                 # Add a blank line preceding, to prevent it from mucking up
2974                                 # immediately preceding headings
2975                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2976                         } else {
2977                                 # replace ==section headers==
2978                                 # XXX this needs to go away once we have a better parser.
2979                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2980                                         if( !is_null( $title ) )
2981                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2982                                         else
2983                                                 $encodedname = base64_encode("");
2984                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2985                                                 PREG_SPLIT_DELIM_CAPTURE);
2986                                         $text = '';
2987                                         $nsec = 0;
2988                                         for( $i = 0; $i < count($m); $i += 2 ) {
2989                                                 $text .= $m[$i];
2990                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2991                                                 $hl = $m[$i + 1];
2992                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2993                                                         $text .= $hl;
2994                                                         continue;
2995                                                 }
2996                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2997                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2998                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2999
3000                                                 $nsec++;
3001                                         }
3002                                 }
3003                         }
3004                 }
3005
3006                 # Prune lower levels off the recursion check path
3007                 $this->mTemplatePath = $lastPathLevel;
3008
3009                 if ( !$found ) {
3010                         wfProfileOut( $fname );
3011                         return $piece['text'];
3012                 } else {
3013                         wfProfileOut( $fname );
3014                         return $text;
3015                 }
3016         }
3017
3018         /**
3019          * Fetch the unparsed text of a template and register a reference to it.
3020          */
3021         function fetchTemplate( $title ) {
3022                 $text = false;
3023                 // Loop to fetch the article, with up to 1 redirect
3024                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3025                         $rev = Revision::newFromTitle( $title );
3026                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3027                         if ( !$rev ) {
3028                                 break;
3029                         }
3030                         $text = $rev->getText();
3031                         if ( $text === false ) {
3032                                 break;
3033                         }
3034                         // Redirect?
3035                         $title = Title::newFromRedirect( $text );
3036                 }
3037                 return $text;
3038         }
3039
3040         /**
3041          * Transclude an interwiki link.
3042          */
3043         function interwikiTransclude( $title, $action ) {
3044                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3045
3046                 if (!$wgEnableScaryTranscluding)
3047                         return wfMsg('scarytranscludedisabled');
3048
3049                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3050                 // But we'll handle it generally anyway
3051                 if ( $title->getNamespace() ) {
3052                         // Use the canonical namespace, which should work anywhere
3053                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3054                 } else {
3055                         $articleName = $title->getDBkey();
3056                 }
3057
3058                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3059                 $url .= "?action=$action";
3060                 if (strlen($url) > 255)
3061                         return wfMsg('scarytranscludetoolong');
3062                 return $this->fetchScaryTemplateMaybeFromCache($url);
3063         }
3064
3065         function fetchScaryTemplateMaybeFromCache($url) {
3066                 global $wgTranscludeCacheExpiry;
3067                 $dbr =& wfGetDB(DB_SLAVE);
3068                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3069                                 array('tc_url' => $url));
3070                 if ($obj) {
3071                         $time = $obj->tc_time;
3072                         $text = $obj->tc_contents;
3073                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3074                                 return $text;
3075                         }
3076                 }
3077
3078                 $text = wfGetHTTP($url);
3079                 if (!$text)
3080                         return wfMsg('scarytranscludefailed', $url);
3081
3082                 $dbw =& wfGetDB(DB_MASTER);
3083                 $dbw->replace('transcache', array('tc_url'), array(
3084                         'tc_url' => $url,
3085                         'tc_time' => time(),
3086                         'tc_contents' => $text));
3087                 return $text;
3088         }
3089
3090
3091         /**
3092          * Triple brace replacement -- used for template arguments
3093          * @private
3094          */
3095         function argSubstitution( $matches ) {
3096                 $arg = trim( $matches['title'] );
3097                 $text = $matches['text'];
3098                 $inputArgs = end( $this->mArgStack );
3099
3100                 if ( array_key_exists( $arg, $inputArgs ) ) {
3101                         $text = $inputArgs[$arg];
3102                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3103                         $text = $matches['parts'][0];
3104                 }
3105
3106                 return $text;
3107         }
3108
3109         /**
3110          * Returns true if the function is allowed to include this entity
3111          * @private
3112          */
3113         function incrementIncludeCount( $dbk ) {
3114                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3115                         $this->mIncludeCount[$dbk] = 0;
3116                 }
3117                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3118                         return true;
3119                 } else {
3120                         return false;
3121                 }
3122         }
3123
3124         /**
3125          * Detect __TOC__ magic word and set a placeholder
3126          */
3127         function stripToc( $text ) {
3128                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3129                 # do not add TOC
3130                 $mw = MagicWord::get( MAG_NOTOC );
3131                 if( $mw->matchAndRemove( $text ) ) {
3132                         $this->mShowToc = false;
3133                 }
3134
3135                 $mw = MagicWord::get( MAG_TOC );
3136                 if( $mw->match( $text ) ) {
3137                         $this->mShowToc = true;
3138                         $this->mForceTocPosition = true;
3139
3140                         // Set a placeholder. At the end we'll fill it in with the TOC.
3141                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3142
3143                         // Only keep the first one.
3144                         $text = $mw->replace( '', $text );
3145                 }
3146                 return $text;
3147         }
3148
3149         /**
3150          * This function accomplishes several tasks:
3151          * 1) Auto-number headings if that option is enabled
3152          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3153          * 3) Add a Table of contents on the top for users who have enabled the option
3154          * 4) Auto-anchor headings
3155          *
3156          * It loops through all headlines, collects the necessary data, then splits up the
3157          * string and re-inserts the newly formatted headlines.
3158          *
3159          * @param string $text
3160          * @param boolean $isMain
3161          * @private
3162          */
3163         function formatHeadings( $text, $isMain=true ) {
3164                 global $wgMaxTocLevel, $wgContLang;
3165
3166                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3167                 if( !$this->mTitle->userCanEdit() ) {
3168                         $showEditLink = 0;
3169                 } else {
3170                         $showEditLink = $this->mOptions->getEditSection();
3171                 }
3172
3173                 # Inhibit editsection links if requested in the page
3174                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3175                 if( $esw->matchAndRemove( $text ) ) {
3176                         $showEditLink = 0;
3177                 }
3178
3179                 # Get all headlines for numbering them and adding funky stuff like [edit]
3180                 # links - this is for later, but we need the number of headlines right now
3181                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3182
3183                 # if there are fewer than 4 headlines in the article, do not show TOC
3184                 # unless it's been explicitly enabled.
3185                 $enoughToc = $this->mShowToc &&
3186                         (($numMatches >= 4) || $this->mForceTocPosition);
3187
3188                 # Allow user to stipulate that a page should have a "new section"
3189                 # link added via __NEWSECTIONLINK__
3190                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3191                 if( $mw->matchAndRemove( $text ) )
3192                         $this->mOutput->setNewSection( true );
3193
3194                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3195                 # override above conditions and always show TOC above first header
3196                 $mw =& MagicWord::get( MAG_FORCETOC );
3197                 if ($mw->matchAndRemove( $text ) ) {
3198                         $this->mShowToc = true;
3199                         $enoughToc = true;
3200                 }
3201
3202                 # Never ever show TOC if no headers
3203                 if( $numMatches < 1 ) {
3204                         $enoughToc = false;
3205                 }
3206
3207                 # We need this to perform operations on the HTML
3208                 $sk =& $this->mOptions->getSkin();
3209
3210                 # headline counter
3211                 $headlineCount = 0;
3212                 $sectionCount = 0; # headlineCount excluding template sections
3213
3214                 # Ugh .. the TOC should have neat indentation levels which can be
3215                 # passed to the skin functions. These are determined here
3216                 $toc = '';
3217                 $full = '';
3218                 $head = array();
3219                 $sublevelCount = array();
3220                 $levelCount = array();
3221                 $toclevel = 0;
3222                 $level = 0;
3223                 $prevlevel = 0;
3224                 $toclevel = 0;
3225                 $prevtoclevel = 0;
3226
3227                 foreach( $matches[3] as $headline ) {
3228                         $istemplate = 0;
3229                         $templatetitle = '';
3230                         $templatesection = 0;
3231                         $numbering = '';
3232
3233                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3234                                 $istemplate = 1;
3235                                 $templatetitle = base64_decode($mat[1]);
3236                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3237                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3238                         }
3239
3240                         if( $toclevel ) {
3241                                 $prevlevel = $level;
3242                                 $prevtoclevel = $toclevel;
3243                         }
3244                         $level = $matches[1][$headlineCount];
3245
3246                         if( $doNumberHeadings || $enoughToc ) {
3247
3248                                 if ( $level > $prevlevel ) {
3249                                         # Increase TOC level
3250                                         $toclevel++;
3251                                         $sublevelCount[$toclevel] = 0;
3252                                         if( $toclevel<$wgMaxTocLevel ) {
3253                                                 $toc .= $sk->tocIndent();
3254                                         }
3255                                 }
3256                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3257                                         # Decrease TOC level, find level to jump to
3258
3259                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3260                                                 # Can only go down to level 1
3261                                                 $toclevel = 1;
3262                                         } else {
3263                                                 for ($i = $toclevel; $i > 0; $i--) {
3264                                                         if ( $levelCount[$i] == $level ) {
3265                                                                 # Found last matching level
3266                                                                 $toclevel = $i;
3267                                                                 break;
3268                                                         }
3269                                                         elseif ( $levelCount[$i] < $level ) {
3270                                                                 # Found first matching level below current level
3271                                                                 $toclevel = $i + 1;
3272                                                                 break;
3273                                                         }
3274                                                 }
3275                                         }
3276                                         if( $toclevel<$wgMaxTocLevel ) {
3277                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3278                                         }
3279                                 }
3280                                 else {
3281                                         # No change in level, end TOC line
3282                                         if( $toclevel<$wgMaxTocLevel ) {
3283                                                 $toc .= $sk->tocLineEnd();
3284                                         }
3285                                 }
3286
3287                                 $levelCount[$toclevel] = $level;
3288
3289                                 # count number of headlines for each level
3290                                 @$sublevelCount[$toclevel]++;
3291                                 $dot = 0;
3292                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3293                                         if( !empty( $sublevelCount[$i] ) ) {
3294                                                 if( $dot ) {
3295                                                         $numbering .= '.';
3296                                                 }
3297                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3298                                                 $dot = 1;
3299                                         }
3300                                 }
3301                         }
3302
3303                         # The canonized header is a version of the header text safe to use for links
3304                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3305                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3306                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3307
3308                         # Remove link placeholders by the link text.
3309                         #     <!--LINK number-->
3310                         # turns into
3311                         #     link text with suffix
3312                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3313                                                             "\$this->mLinkHolders['texts'][\$1]",
3314                                                             $canonized_headline );
3315                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3316                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3317                                                             $canonized_headline );
3318
3319                         # strip out HTML
3320                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3321                         $tocline = trim( $canonized_headline );
3322                         # Save headline for section edit hint before it's escaped
3323                         $headline_hint = trim( $canonized_headline );
3324                         $canonized_headline = Sanitizer::escapeId( $tocline );
3325                         $refers[$headlineCount] = $canonized_headline;
3326
3327                         # count how many in assoc. array so we can track dupes in anchors
3328                         @$refers[$canonized_headline]++;
3329                         $refcount[$headlineCount]=$refers[$canonized_headline];
3330
3331                         # Don't number the heading if it is the only one (looks silly)
3332                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3333                                 # the two are different if the line contains a link
3334                                 $headline=$numbering . ' ' . $headline;
3335                         }
3336
3337                         # Create the anchor for linking from the TOC to the section
3338                         $anchor = $canonized_headline;
3339                         if($refcount[$headlineCount] > 1 ) {
3340                                 $anchor .= '_' . $refcount[$headlineCount];
3341                         }
3342                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3343                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3344                         }
3345                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3346                                 if ( empty( $head[$headlineCount] ) ) {
3347                                         $head[$headlineCount] = '';
3348                                 }
3349                                 if( $istemplate )
3350                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3351                                 else
3352                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3353                         }
3354
3355                         # give headline the correct <h#> tag
3356                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3357
3358                         $headlineCount++;
3359                         if( !$istemplate )
3360                                 $sectionCount++;
3361                 }
3362
3363                 if( $enoughToc ) {
3364                         if( $toclevel<$wgMaxTocLevel ) {
3365                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3366                         }
3367                         $toc = $sk->tocList( $toc );
3368                 }
3369
3370                 # split up and insert constructed headlines
3371
3372                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3373                 $i = 0;
3374
3375                 foreach( $blocks as $block ) {
3376                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3377                                 # This is the [edit] link that appears for the top block of text when
3378                                 # section editing is enabled
3379
3380                                 # Disabled because it broke block formatting
3381                                 # For example, a bullet point in the top line
3382                                 # $full .= $sk->editSectionLink(0);
3383                         }
3384                         $full .= $block;
3385                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3386                                 # Top anchor now in skin
3387                                 $full = $full.$toc;
3388                         }
3389
3390                         if( !empty( $head[$i] ) ) {
3391                                 $full .= $head[$i];
3392                         }
3393                         $i++;
3394                 }
3395                 if( $this->mForceTocPosition ) {
3396                         return str_replace( '<!--MWTOC-->', $toc, $full );
3397                 } else {
3398                         return $full;
3399                 }
3400         }
3401
3402         /**
3403          * Return an HTML link for the "ISBN 123456" text
3404          * @private
3405          */
3406         function magicISBN( $text ) {
3407                 $fname = 'Parser::magicISBN';
3408                 wfProfileIn( $fname );
3409
3410                 $a = split( 'ISBN ', ' '.$text );
3411                 if ( count ( $a ) < 2 ) {
3412                         wfProfileOut( $fname );
3413                         return $text;
3414                 }
3415                 $text = substr( array_shift( $a ), 1);
3416                 $valid = '0123456789-Xx';
3417
3418                 foreach ( $a as $x ) {
3419                         # hack: don't replace inside thumbnail title/alt
3420                         # attributes
3421                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3422                                 $text .= "ISBN $x";
3423                                 continue;
3424                         }
3425
3426                         $isbn = $blank = '' ;
3427                         while ( ' ' == $x{0} ) {
3428                                 $blank .= ' ';
3429                                 $x = substr( $x, 1 );
3430                         }
3431                         if ( $x == '' ) { # blank isbn
3432                                 $text .= "ISBN $blank";
3433                                 continue;
3434                         }
3435                         while ( strstr( $valid, $x{0} ) != false ) {
3436                                 $isbn .= $x{0};
3437                                 $x = substr( $x, 1 );
3438                         }
3439                         $num = str_replace( '-', '', $isbn );
3440                         $num = str_replace( ' ', '', $num );
3441                         $num = str_replace( 'x', 'X', $num );
3442
3443                         if ( '' == $num ) {
3444                                 $text .= "ISBN $blank$x";
3445                         } else {
3446                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3447                                 $text .= '<a href="' .
3448                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3449                                         "\" class=\"internal\">ISBN $isbn</a>";
3450                                 $text .= $x;
3451                         }
3452                 }
3453                 wfProfileOut( $fname );
3454                 return $text;
3455         }
3456
3457         /**
3458          * Return an HTML link for the "RFC 1234" text
3459          *
3460          * @private
3461          * @param string $text     Text to be processed
3462          * @param string $keyword  Magic keyword to use (default RFC)
3463          * @param string $urlmsg   Interface message to use (default rfcurl)
3464          * @return string
3465          */
3466         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3467
3468                 $valid = '0123456789';
3469                 $internal = false;
3470
3471                 $a = split( $keyword, ' '.$text );
3472                 if ( count ( $a ) < 2 ) {
3473                         return $text;
3474                 }
3475                 $text = substr( array_shift( $a ), 1);
3476
3477                 /* Check if keyword is preceed by [[.
3478                  * This test is made here cause of the array_shift above
3479                  * that prevent the test to be done in the foreach.
3480                  */
3481                 if ( substr( $text, -2 ) == '[[' ) {
3482                         $internal = true;
3483                 }
3484
3485                 foreach ( $a as $x ) {
3486                         /* token might be empty if we have RFC RFC 1234 */
3487                         if ( $x=='' ) {
3488                                 $text.=$keyword;
3489                                 continue;
3490                                 }
3491
3492                         # hack: don't replace inside thumbnail title/alt
3493                         # attributes
3494                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3495                                 $text .= $keyword . $x;
3496                                 continue;
3497                         }
3498
3499                         $id = $blank = '' ;
3500
3501                         /** remove and save whitespaces in $blank */
3502                         while ( $x{0} == ' ' ) {
3503                                 $blank .= ' ';
3504                                 $x = substr( $x, 1 );
3505                         }
3506
3507                         /** remove and save the rfc number in $id */
3508                         while ( strstr( $valid, $x{0} ) != false ) {
3509                                 $id .= $x{0};
3510                                 $x = substr( $x, 1 );
3511                         }
3512
3513                         if ( $id == '' ) {
3514                                 /* call back stripped spaces*/
3515                                 $text .= $keyword.$blank.$x;
3516                         } elseif( $internal ) {
3517                                 /* normal link */
3518                                 $text .= $keyword.$id.$x;
3519                         } else {
3520                                 /* build the external link*/
3521                                 $url = wfMsg( $urlmsg, $id);
3522                                 $sk =& $this->mOptions->getSkin();
3523                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3524                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3525                         }
3526
3527                         /* Check if the next RFC keyword is preceed by [[ */
3528                         $internal = ( substr($x,-2) == '[[' );
3529                 }
3530                 return $text;
3531         }
3532
3533         /**
3534          * Transform wiki markup when saving a page by doing \r\n -> \n
3535          * conversion, substitting signatures, {{subst:}} templates, etc.
3536          *
3537          * @param string $text the text to transform
3538          * @param Title &$title the Title object for the current article
3539          * @param User &$user the User object describing the current user
3540          * @param ParserOptions $options parsing options
3541          * @param bool $clearState whether to clear the parser state first
3542          * @return string the altered wiki markup
3543          * @public
3544          */
3545         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3546                 $this->mOptions = $options;
3547                 $this->mTitle =& $title;
3548                 $this->mOutputType = OT_WIKI;
3549
3550                 if ( $clearState ) {
3551                         $this->clearState();
3552                 }
3553
3554                 $stripState = false;
3555                 $pairs = array(
3556                         "\r\n" => "\n",
3557                 );
3558                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3559                 $text = $this->strip( $text, $stripState, true );
3560                 $text = $this->pstPass2( $text, $stripState, $user );
3561                 $text = $this->unstrip( $text, $stripState );
3562                 $text = $this->unstripNoWiki( $text, $stripState );
3563                 return $text;
3564         }
3565
3566         /**
3567          * Pre-save transform helper function
3568          * @private
3569          */
3570         function pstPass2( $text, &$stripState, &$user ) {
3571                 global $wgContLang, $wgLocaltimezone;
3572
3573                 /* Note: This is the timestamp saved as hardcoded wikitext to
3574                  * the database, we use $wgContLang here in order to give
3575                  * everyone the same signature and use the default one rather
3576                  * than the one selected in each user's preferences.
3577                  */
3578                 if ( isset( $wgLocaltimezone ) ) {
3579                         $oldtz = getenv( 'TZ' );
3580                         putenv( 'TZ='.$wgLocaltimezone );
3581                 }
3582                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3583                   ' (' . date( 'T' ) . ')';
3584                 if ( isset( $wgLocaltimezone ) ) {
3585                         putenv( 'TZ='.$oldtz );
3586                 }
3587
3588                 # Variable replacement
3589                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3590                 $text = $this->replaceVariables( $text );
3591
3592                 # Strip out <nowiki> etc. added via replaceVariables
3593                 $text = $this->strip( $text, $stripState );
3594
3595                 # Signatures
3596                 $sigText = $this->getUserSig( $user );
3597                 $text = strtr( $text, array(
3598                         '~~~~~' => $d,
3599                         '~~~~' => "$sigText $d",
3600                         '~~~' => $sigText
3601                 ) );
3602
3603                 # Context links: [[|name]] and [[name (context)|]]
3604                 #
3605                 global $wgLegalTitleChars;
3606                 $tc = "[$wgLegalTitleChars]";
3607                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3608
3609                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3610                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3611
3612                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3613                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3614                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3615                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3616                 $context = '';
3617                 $t = $this->mTitle->getText();
3618                 if ( preg_match( $conpat, $t, $m ) ) {
3619                         $context = $m[2];
3620                 }
3621                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3622                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3623                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3624
3625                 if ( '' == $context ) {
3626                         $text = preg_replace( $p2, '[[\\1]]', $text );
3627                 } else {
3628                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3629                 }
3630
3631                 # Trim trailing whitespace
3632                 # MAG_END (__END__) tag allows for trailing
3633                 # whitespace to be deliberately included
3634                 $text = rtrim( $text );
3635                 $mw =& MagicWord::get( MAG_END );
3636                 $mw->matchAndRemove( $text );
3637
3638                 return $text;
3639         }
3640
3641         /**
3642          * Fetch the user's signature text, if any, and normalize to
3643          * validated, ready-to-insert wikitext.
3644          *
3645          * @param User $user
3646          * @return string
3647          * @private
3648          */
3649         function getUserSig( &$user ) {
3650                 $username = $user->getName();
3651                 $nickname = $user->getOption( 'nickname' );
3652                 $nickname = $nickname === '' ? $username : $nickname;
3653
3654                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3655                         # Sig. might contain markup; validate this
3656                         if( $this->validateSig( $nickname ) !== false ) {
3657                                 # Validated; clean up (if needed) and return it
3658                                 return $this->cleanSig( $nickname, true );
3659                         } else {
3660                                 # Failed to validate; fall back to the default
3661                                 $nickname = $username;
3662                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3663                         }
3664                 }
3665
3666                 # If we're still here, make it a link to the user page
3667                 $userpage = $user->getUserPage();
3668                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3669         }
3670
3671         /**
3672          * Check that the user's signature contains no bad XML
3673          *
3674          * @param string $text
3675          * @return mixed An expanded string, or false if invalid.
3676          */
3677         function validateSig( $text ) {
3678                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3679         }
3680
3681         /**
3682          * Clean up signature text
3683          *
3684          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3685          * 2) Substitute all transclusions
3686          *
3687          * @param string $text
3688          * @param $parsing Whether we're cleaning (preferences save) or parsing
3689          * @return string Signature text
3690          */
3691         function cleanSig( $text, $parsing = false ) {
3692                 global $wgTitle;
3693                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3694
3695                 $substWord = MagicWord::get( MAG_SUBST );
3696                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3697                 $substText = '{{' . $substWord->getSynonym( 0 );
3698
3699                 $text = preg_replace( $substRegex, $substText, $text );
3700                 $text = preg_replace( '/~{3,5}/', '', $text );
3701                 $text = $this->replaceVariables( $text );
3702
3703                 $this->clearState();
3704                 return $text;
3705         }
3706
3707         /**
3708          * Set up some variables which are usually set up in parse()
3709          * so that an external function can call some class members with confidence
3710          * @public
3711          */
3712         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3713                 $this->mTitle =& $title;
3714                 $this->mOptions = $options;
3715                 $this->mOutputType = $outputType;
3716                 if ( $clearState ) {
3717                         $this->clearState();
3718                 }
3719         }
3720
3721         /**
3722          * Transform a MediaWiki message by replacing magic variables.
3723          *
3724          * @param string $text the text to transform
3725          * @param ParserOptions $options  options
3726          * @return string the text with variables substituted
3727          * @public
3728          */
3729         function transformMsg( $text, $options ) {
3730                 global $wgTitle;
3731                 static $executing = false;
3732
3733                 $fname = "Parser::transformMsg";
3734
3735                 # Guard against infinite recursion
3736                 if ( $executing ) {
3737                         return $text;
3738                 }
3739                 $executing = true;
3740
3741                 wfProfileIn($fname);
3742
3743                 $this->mTitle = $wgTitle;
3744                 $this->mOptions = $options;
3745                 $this->mOutputType = OT_MSG;
3746                 $this->clearState();
3747                 $text = $this->replaceVariables( $text );
3748
3749                 $executing = false;
3750                 wfProfileOut($fname);
3751                 return $text;
3752         }
3753
3754         /**
3755          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3756          * The callback should have the following form:
3757          *    function myParserHook( $text, $params, &$parser ) { ... }
3758          *
3759          * Transform and return $text. Use $parser for any required context, e.g. use
3760          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3761          *
3762          * @public
3763          *
3764          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3765          * @param mixed $callback The callback function (and object) to use for the tag
3766          *
3767          * @return The old value of the mTagHooks array associated with the hook
3768          */
3769         function setHook( $tag, $callback ) {
3770                 $tag = strtolower( $tag );
3771                 $oldVal = @$this->mTagHooks[$tag];
3772                 $this->mTagHooks[$tag] = $callback;
3773
3774                 return $oldVal;
3775         }
3776
3777         /**
3778          * Create a function, e.g. {{sum:1|2|3}}
3779          * The callback function should have the form:
3780          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3781          *
3782          * The callback may either return the text result of the function, or an array with the text
3783          * in element 0, and a number of flags in the other elements. The names of the flags are
3784          * specified in the keys. Valid flags are:
3785          *   found                     The text returned is valid, stop processing the template. This
3786          *                             is on by default.
3787          *   nowiki                    Wiki markup in the return value should be escaped
3788          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3789          *   noargs                    Don't replace triple-brace arguments in the return value
3790          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3791          *
3792          * @public
3793          *
3794          * @param string $name The function name. Function names are case-insensitive.
3795          * @param mixed $callback The callback function (and object) to use
3796          *
3797          * @return The old callback function for this name, if any
3798          */
3799         function setFunctionHook( $name, $callback ) {
3800                 $name = strtolower( $name );
3801                 $oldVal = @$this->mFunctionHooks[$name];
3802                 $this->mFunctionHooks[$name] = $callback;
3803                 return $oldVal;
3804         }
3805
3806         /**
3807          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3808          * Placeholders created in Skin::makeLinkObj()
3809          * Returns an array of links found, indexed by PDBK:
3810          *  0 - broken
3811          *  1 - normal link
3812          *  2 - stub
3813          * $options is a bit field, RLH_FOR_UPDATE to select for update
3814          */
3815         function replaceLinkHolders( &$text, $options = 0 ) {
3816                 global $wgUser;
3817                 global $wgOutputReplace;
3818
3819                 $fname = 'Parser::replaceLinkHolders';
3820                 wfProfileIn( $fname );
3821
3822                 $pdbks = array();
3823                 $colours = array();
3824                 $sk =& $this->mOptions->getSkin();
3825                 $linkCache =& LinkCache::singleton();
3826
3827                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3828                         wfProfileIn( $fname.'-check' );
3829                         $dbr =& wfGetDB( DB_SLAVE );
3830                         $page = $dbr->tableName( 'page' );
3831                         $threshold = $wgUser->getOption('stubthreshold');
3832
3833                         # Sort by namespace
3834                         asort( $this->mLinkHolders['namespaces'] );
3835
3836                         # Generate query
3837                         $query = false;
3838                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3839                                 # Make title object
3840                                 $title = $this->mLinkHolders['titles'][$key];
3841
3842                                 # Skip invalid entries.
3843                                 # Result will be ugly, but prevents crash.
3844                                 if ( is_null( $title ) ) {
3845                                         continue;
3846                                 }
3847                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3848
3849                                 # Check if it's a static known link, e.g. interwiki
3850                                 if ( $title->isAlwaysKnown() ) {
3851                                         $colours[$pdbk] = 1;
3852                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3853                                         $colours[$pdbk] = 1;
3854                                         $this->mOutput->addLink( $title, $id );
3855                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3856                                         $colours[$pdbk] = 0;
3857                                 } else {
3858                                         # Not in the link cache, add it to the query
3859                                         if ( !isset( $current ) ) {
3860                                                 $current = $ns;
3861                                                 $query =  "SELECT page_id, page_namespace, page_title";
3862                                                 if ( $threshold > 0 ) {
3863                                                         $query .= ', page_len, page_is_redirect';
3864                                                 }
3865                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3866                                         } elseif ( $current != $ns ) {
3867                                                 $current = $ns;
3868                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3869                                         } else {
3870                                                 $query .= ', ';
3871                                         }
3872
3873                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3874                                 }
3875                         }
3876                         if ( $query ) {
3877                                 $query .= '))';
3878                                 if ( $options & RLH_FOR_UPDATE ) {
3879                                         $query .= ' FOR UPDATE';
3880                                 }
3881
3882                                 $res = $dbr->query( $query, $fname );
3883
3884                                 # Fetch data and form into an associative array
3885                                 # non-existent = broken
3886                                 # 1 = known
3887                                 # 2 = stub
3888                                 while ( $s = $dbr->fetchObject($res) ) {
3889                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3890                                         $pdbk = $title->getPrefixedDBkey();
3891                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3892                                         $this->mOutput->addLink( $title, $s->page_id );
3893
3894                                         if ( $threshold >  0 ) {
3895                                                 $size = $s->page_len;
3896                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3897                                                         $colours[$pdbk] = 1;
3898                                                 } else {
3899                                                         $colours[$pdbk] = 2;
3900                                                 }
3901                                         } else {
3902                                                 $colours[$pdbk] = 1;
3903                                         }
3904                                 }
3905                         }
3906                         wfProfileOut( $fname.'-check' );
3907
3908                         # Construct search and replace arrays
3909                         wfProfileIn( $fname.'-construct' );
3910                         $wgOutputReplace = array();
3911                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3912                                 $pdbk = $pdbks[$key];
3913                                 $searchkey = "<!--LINK $key-->";
3914                                 $title = $this->mLinkHolders['titles'][$key];
3915                                 if ( empty( $colours[$pdbk] ) ) {
3916                                         $linkCache->addBadLinkObj( $title );
3917                                         $colours[$pdbk] = 0;
3918                                         $this->mOutput->addLink( $title, 0 );
3919                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3920                                                                         $this->mLinkHolders['texts'][$key],
3921                                                                         $this->mLinkHolders['queries'][$key] );
3922                                 } elseif ( $colours[$pdbk] == 1 ) {
3923                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3924                                                                         $this->mLinkHolders['texts'][$key],
3925                                                                         $this->mLinkHolders['queries'][$key] );
3926                                 } elseif ( $colours[$pdbk] == 2 ) {
3927                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3928                                                                         $this->mLinkHolders['texts'][$key],
3929                                                                         $this->mLinkHolders['queries'][$key] );
3930                                 }
3931                         }
3932                         wfProfileOut( $fname.'-construct' );
3933
3934                         # Do the thing
3935                         wfProfileIn( $fname.'-replace' );
3936
3937                         $text = preg_replace_callback(
3938                                 '/(<!--LINK .*?-->)/',
3939                                 "wfOutputReplaceMatches",
3940                                 $text);
3941
3942                         wfProfileOut( $fname.'-replace' );
3943                 }
3944
3945                 # Now process interwiki link holders
3946                 # This is quite a bit simpler than internal links
3947                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3948                         wfProfileIn( $fname.'-interwiki' );
3949                         # Make interwiki link HTML
3950                         $wgOutputReplace = array();
3951                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3952                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3953                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3954                         }
3955
3956                         $text = preg_replace_callback(
3957                                 '/<!--IWLINK (.*?)-->/',
3958                                 "wfOutputReplaceMatches",
3959                                 $text );
3960                         wfProfileOut( $fname.'-interwiki' );
3961                 }
3962
3963                 wfProfileOut( $fname );
3964                 return $colours;
3965         }
3966
3967         /**
3968          * Replace <!--LINK--> link placeholders with plain text of links
3969          * (not HTML-formatted).
3970          * @param string $text
3971          * @return string
3972          */
3973         function replaceLinkHoldersText( $text ) {
3974                 $fname = 'Parser::replaceLinkHoldersText';
3975                 wfProfileIn( $fname );
3976
3977                 $text = preg_replace_callback(
3978                         '/<!--(LINK|IWLINK) (.*?)-->/',
3979                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3980                         $text );
3981
3982                 wfProfileOut( $fname );
3983                 return $text;
3984         }
3985
3986         /**
3987          * @param array $matches
3988          * @return string
3989          * @private
3990          */
3991         function replaceLinkHoldersTextCallback( $matches ) {
3992                 $type = $matches[1];
3993                 $key  = $matches[2];
3994                 if( $type == 'LINK' ) {
3995                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3996                                 return $this->mLinkHolders['texts'][$key];
3997                         }
3998                 } elseif( $type == 'IWLINK' ) {
3999                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4000                                 return $this->mInterwikiLinkHolders['texts'][$key];
4001                         }
4002                 }
4003                 return $matches[0];
4004         }
4005
4006         /**
4007          * Renders an image gallery from a text with one line per image.
4008          * text labels may be given by using |-style alternative text. E.g.
4009          *   Image:one.jpg|The number "1"
4010          *   Image:tree.jpg|A tree
4011          * given as text will return the HTML of a gallery with two images,
4012          * labeled 'The number "1"' and
4013          * 'A tree'.
4014          */
4015         function renderImageGallery( $text ) {
4016                 $ig = new ImageGallery();
4017                 $ig->setShowBytes( false );
4018                 $ig->setShowFilename( false );
4019                 $ig->setParsing();
4020                 $lines = explode( "\n", $text );
4021
4022                 foreach ( $lines as $line ) {
4023                         # match lines like these:
4024                         # Image:someimage.jpg|This is some image
4025                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4026                         # Skip empty lines
4027                         if ( count( $matches ) == 0 ) {
4028                                 continue;
4029                         }
4030                         $nt =& Title::newFromText( $matches[1] );
4031                         if( is_null( $nt ) ) {
4032                                 # Bogus title. Ignore these so we don't bomb out later.
4033                                 continue;
4034                         }
4035                         if ( isset( $matches[3] ) ) {
4036                                 $label = $matches[3];
4037                         } else {
4038                                 $label = '';
4039                         }
4040
4041                         $pout = $this->parse( $label,
4042                                 $this->mTitle,
4043                                 $this->mOptions,
4044                                 false, // Strip whitespace...?
4045                                 false  // Don't clear state!
4046                         );
4047                         $html = $pout->getText();
4048
4049                         $ig->add( new Image( $nt ), $html );
4050
4051                         # Only add real images (bug #5586)
4052                         if ( $nt->getNamespace() == NS_IMAGE ) {
4053                                 $this->mOutput->addImage( $nt->getDBkey() );
4054                         }
4055                 }
4056                 return $ig->toHTML();
4057         }
4058
4059         /**
4060          * Parse image options text and use it to make an image
4061          */
4062         function makeImage( &$nt, $options ) {
4063                 global $wgUseImageResize;
4064
4065                 $align = '';
4066
4067                 # Check if the options text is of the form "options|alt text"
4068                 # Options are:
4069                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4070                 #  * left               no resizing, just left align. label is used for alt= only
4071                 #  * right              same, but right aligned
4072                 #  * none               same, but not aligned
4073                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4074                 #  * center             center the image
4075                 #  * framed             Keep original image size, no magnify-button.
4076
4077                 $part = explode( '|', $options);
4078
4079                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4080                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4081                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4082                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4083                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4084                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4085                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4086                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4087                 $caption = '';
4088
4089                 $width = $height = $framed = $thumb = false;
4090                 $manual_thumb = '' ;
4091
4092                 foreach( $part as $key => $val ) {
4093                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4094                                 $thumb=true;
4095                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4096                                 # use manually specified thumbnail
4097                                 $thumb=true;
4098                                 $manual_thumb = $match;
4099                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4100                                 # remember to set an alignment, don't render immediately
4101                                 $align = 'right';
4102                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4103                                 # remember to set an alignment, don't render immediately
4104                                 $align = 'left';
4105                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4106                                 # remember to set an alignment, don't render immediately
4107                                 $align = 'center';
4108                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4109                                 # remember to set an alignment, don't render immediately
4110                                 $align = 'none';
4111                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4112                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4113                                 # $match is the image width in pixels
4114                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4115                                         $width = intval( $m[1] );
4116                                         $height = intval( $m[2] );
4117                                 } else {
4118                                         $width = intval($match);
4119                                 }
4120                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4121                                 $framed=true;
4122                         } else {
4123                                 $caption = $val;
4124                         }
4125                 }
4126                 # Strip bad stuff out of the alt text
4127                 $alt = $this->replaceLinkHoldersText( $caption );
4128
4129                 # make sure there are no placeholders in thumbnail attributes
4130                 # that are later expanded to html- so expand them now and
4131                 # remove the tags
4132                 $alt = $this->unstrip($alt, $this->mStripState);
4133                 $alt = Sanitizer::stripAllTags( $alt );
4134
4135                 # Linker does the rest
4136                 $sk =& $this->mOptions->getSkin();
4137                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4138         }
4139
4140         /**
4141          * Set a flag in the output object indicating that the content is dynamic and
4142          * shouldn't be cached.
4143          */
4144         function disableCache() {
4145                 wfDebug( "Parser output marked as uncacheable.\n" );
4146                 $this->mOutput->mCacheTime = -1;
4147         }
4148
4149         /**#@+
4150          * Callback from the Sanitizer for expanding items found in HTML attribute
4151          * values, so they can be safely tested and escaped.
4152          * @param string $text
4153          * @param array $args
4154          * @return string
4155          * @private
4156          */
4157         function attributeStripCallback( &$text, $args ) {
4158                 $text = $this->replaceVariables( $text, $args );
4159                 $text = $this->unstripForHTML( $text );
4160                 return $text;
4161         }
4162
4163         function unstripForHTML( $text ) {
4164                 $text = $this->unstrip( $text, $this->mStripState );
4165                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4166                 return $text;
4167         }
4168         /**#@-*/
4169
4170         /**#@+
4171          * Accessor/mutator
4172          */
4173         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4174         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4175         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4176         /**#@-*/
4177
4178         /**#@+
4179          * Accessor
4180          */
4181         function getTags() { return array_keys( $this->mTagHooks ); }
4182         /**#@-*/
4183 }
4184
4185 /**
4186  * @todo document
4187  * @package MediaWiki
4188  */
4189 class ParserOutput
4190 {
4191         var $mText,             # The output text
4192                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4193                 $mCategories,       # Map of category names to sort keys
4194                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4195                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4196                 $mVersion,          # Compatibility check
4197                 $mTitleText,        # title text of the chosen language variant
4198                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4199                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4200                 $mImages,           # DB keys of the images used, in the array key only
4201                 $mExternalLinks,    # External link URLs, in the key only
4202                 $mHTMLtitle,            # Display HTML title
4203                 $mSubtitle,                     # Additional subtitle
4204                 $mNewSection;           # Show a new section link?
4205
4206         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4207                 $containsOldMagic = false, $titletext = '' )
4208         {
4209                 $this->mText = $text;
4210                 $this->mLanguageLinks = $languageLinks;
4211                 $this->mCategories = $categoryLinks;
4212                 $this->mContainsOldMagic = $containsOldMagic;
4213                 $this->mCacheTime = '';
4214                 $this->mVersion = MW_PARSER_VERSION;
4215                 $this->mTitleText = $titletext;
4216                 $this->mLinks = array();
4217                 $this->mTemplates = array();
4218                 $this->mImages = array();
4219                 $this->mExternalLinks = array();
4220                 $this->mHTMLtitle = "" ;
4221                 $this->mSubtitle = "" ;
4222                 $this->mNewSection = false;
4223         }
4224
4225         function getText()                   { return $this->mText; }
4226         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4227         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4228         function &getCategories()            { return $this->mCategories; }
4229         function getCacheTime()              { return $this->mCacheTime; }
4230         function getTitleText()              { return $this->mTitleText; }
4231         function &getLinks()                 { return $this->mLinks; }
4232         function &getTemplates()             { return $this->mTemplates; }
4233         function &getImages()                { return $this->mImages; }
4234         function &getExternalLinks()         { return $this->mExternalLinks; }
4235
4236         function containsOldMagic()          { return $this->mContainsOldMagic; }
4237         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4238         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4239         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4240         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4241         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4242         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4243
4244         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4245         function addImage( $name )           { $this->mImages[$name] = 1; }
4246         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4247         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4248
4249         function setNewSection( $value ) {
4250                 $this->mNewSection = (bool)$value;
4251         }
4252         function getNewSection() {
4253                 return (bool)$this->mNewSection;
4254         }
4255
4256         function addLink( $title, $id ) {
4257                 $ns = $title->getNamespace();
4258                 $dbk = $title->getDBkey();
4259                 if ( !isset( $this->mLinks[$ns] ) ) {
4260                         $this->mLinks[$ns] = array();
4261                 }
4262                 $this->mLinks[$ns][$dbk] = $id;
4263         }
4264
4265         function addTemplate( $title, $id ) {
4266                 $ns = $title->getNamespace();
4267                 $dbk = $title->getDBkey();
4268                 if ( !isset( $this->mTemplates[$ns] ) ) {
4269                         $this->mTemplates[$ns] = array();
4270                 }
4271                 $this->mTemplates[$ns][$dbk] = $id;
4272         }
4273
4274         /**
4275          * Return true if this cached output object predates the global or
4276          * per-article cache invalidation timestamps, or if it comes from
4277          * an incompatible older version.
4278          *
4279          * @param string $touched the affected article's last touched timestamp
4280          * @return bool
4281          * @public
4282          */
4283         function expired( $touched ) {
4284                 global $wgCacheEpoch;
4285                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4286                        $this->getCacheTime() < $touched ||
4287                        $this->getCacheTime() <= $wgCacheEpoch ||
4288                        !isset( $this->mVersion ) ||
4289                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4290         }
4291 }
4292
4293 /**
4294  * Set options of the Parser
4295  * @todo document
4296  * @package MediaWiki
4297  */
4298 class ParserOptions
4299 {
4300         # All variables are private
4301         var $mUseTeX;                    # Use texvc to expand <math> tags
4302         var $mUseDynamicDates;           # Use DateFormatter to format dates
4303         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4304         var $mAllowExternalImages;       # Allow external images inline
4305         var $mAllowExternalImagesFrom;   # If not, any exception?
4306         var $mSkin;                      # Reference to the preferred skin
4307         var $mDateFormat;                # Date format index
4308         var $mEditSection;               # Create "edit section" links
4309         var $mNumberHeadings;            # Automatically number headings
4310         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4311         var $mTidy;                      # Ask for tidy cleanup
4312         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4313
4314         function getUseTeX()                        { return $this->mUseTeX; }
4315         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4316         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4317         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4318         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4319         function &getSkin()                         { return $this->mSkin; }
4320         function getDateFormat()                    { return $this->mDateFormat; }
4321         function getEditSection()                   { return $this->mEditSection; }
4322         function getNumberHeadings()                { return $this->mNumberHeadings; }
4323         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4324         function getTidy()                          { return $this->mTidy; }
4325         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4326
4327         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4328         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4329         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4330         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4331         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4332         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4333         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4334         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4335         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4336         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4337         function setSkin( &$x ) { $this->mSkin =& $x; }
4338         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4339
4340         function ParserOptions() {
4341                 global $wgUser;
4342                 $this->initialiseFromUser( $wgUser );
4343         }
4344
4345         /**
4346          * Get parser options
4347          * @static
4348          */
4349         function newFromUser( &$user ) {
4350                 $popts = new ParserOptions;
4351                 $popts->initialiseFromUser( $user );
4352                 return $popts;
4353         }
4354
4355         /** Get user options */
4356         function initialiseFromUser( &$userInput ) {
4357                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4358                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4359                 $fname = 'ParserOptions::initialiseFromUser';
4360                 wfProfileIn( $fname );
4361                 if ( !$userInput ) {
4362                         $user = new User;
4363                         $user->setLoaded( true );
4364                 } else {
4365                         $user =& $userInput;
4366                 }
4367
4368                 $this->mUseTeX = $wgUseTeX;
4369                 $this->mUseDynamicDates = $wgUseDynamicDates;
4370                 $this->mInterwikiMagic = $wgInterwikiMagic;
4371                 $this->mAllowExternalImages = $wgAllowExternalImages;
4372                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4373                 wfProfileIn( $fname.'-skin' );
4374                 $this->mSkin =& $user->getSkin();
4375                 wfProfileOut( $fname.'-skin' );
4376                 $this->mDateFormat = $user->getOption( 'date' );
4377                 $this->mEditSection = true;
4378                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4379                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4380                 $this->mTidy = false;
4381                 $this->mInterfaceMessage = false;
4382                 wfProfileOut( $fname );
4383         }
4384 }
4385
4386 /**
4387  * Callback function used by Parser::replaceLinkHolders()
4388  * to substitute link placeholders.
4389  */
4390 function &wfOutputReplaceMatches( $matches ) {
4391         global $wgOutputReplace;
4392         return $wgOutputReplace[$matches[1]];
4393 }
4394
4395 /**
4396  * Return the total number of articles
4397  */
4398 function wfNumberOfArticles() {
4399         global $wgNumberOfArticles;
4400
4401         wfLoadSiteStats();
4402         return $wgNumberOfArticles;
4403 }
4404
4405 /**
4406  * Return the number of files
4407  */
4408 function wfNumberOfFiles() {
4409         $fname = 'wfNumberOfFiles';
4410
4411         wfProfileIn( $fname );
4412         $dbr =& wfGetDB( DB_SLAVE );
4413         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4414         wfProfileOut( $fname );
4415
4416         return $numImages;
4417 }
4418
4419 /**
4420  * Return the number of user accounts
4421  * @return integer
4422  */
4423 function wfNumberOfUsers() {
4424         wfProfileIn( 'wfNumberOfUsers' );
4425         $dbr =& wfGetDB( DB_SLAVE );
4426         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4427         wfProfileOut( 'wfNumberOfUsers' );
4428         return (int)$count;
4429 }
4430
4431 /**
4432  * Return the total number of pages
4433  * @return integer
4434  */
4435 function wfNumberOfPages() {
4436         wfProfileIn( 'wfNumberOfPages' );
4437         $dbr =& wfGetDB( DB_SLAVE );
4438         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4439         wfProfileOut( 'wfNumberOfPages' );
4440         return (int)$count;
4441 }
4442
4443 /**
4444  * Get various statistics from the database
4445  * @private
4446  */
4447 function wfLoadSiteStats() {
4448         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4449         $fname = 'wfLoadSiteStats';
4450
4451         if ( -1 != $wgNumberOfArticles ) return;
4452         $dbr =& wfGetDB( DB_SLAVE );
4453         $s = $dbr->selectRow( 'site_stats',
4454                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4455                 array( 'ss_row_id' => 1 ), $fname
4456         );
4457
4458         if ( $s === false ) {
4459                 return;
4460         } else {
4461                 $wgTotalViews = $s->ss_total_views;
4462                 $wgTotalEdits = $s->ss_total_edits;
4463                 $wgNumberOfArticles = $s->ss_good_articles;
4464         }
4465 }
4466
4467 /**
4468  * Escape html tags
4469  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4470  *
4471  * @param $in String: text that might contain HTML tags.
4472  * @return string Escaped string
4473  */
4474 function wfEscapeHTMLTagsOnly( $in ) {
4475         return str_replace(
4476                 array( '"', '>', '<' ),
4477                 array( '&quot;', '&gt;', '&lt;' ),
4478                 $in );
4479 }
4480
4481 ?>