3 * XHTML sanitizer for MediaWiki
5 * Copyright (C) 2002-2005 Brion Vibber <brion@pobox.com> et al
6 * http://www.mediawiki.org/
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
18 * You should have received a copy of the GNU General Public License along
19 * with this program; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
21 * http://www.gnu.org/copyleft/gpl.html
28 * Regular expression to match various types of character references in
29 * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
31 define( 'MW_CHAR_REFS_REGEX',
32 '/&([A-Za-z0-9\x80-\xff]+);
39 * Regular expression to match HTML/XML attribute pairs within a tag.
40 * Allows some... latitude.
41 * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
43 $attrib = '[A-Za-z0-9]';
44 $space = '[\x09\x0a\x0d\x20]';
45 define( 'MW_ATTRIBS_REGEX',
46 "/(?:^|$space)($attrib+)
49 # The attribute value: quoted or alone
52 | ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
53 | (\#[0-9a-fA-F]+) # Technically wrong, but lots of
54 # colors are specified like this.
55 # We'll be normalizing it.
57 )?(?=$space|\$)/sx" );
60 * List of all named character entities defined in HTML 4.01
61 * http://www.w3.org/TR/html4/sgml/entities.html
64 global $wgHtmlEntities;
65 $wgHtmlEntities = array(
320 * Character entity aliases accepted by MediaWiki
322 global $wgHtmlEntityAliases;
323 $wgHtmlEntityAliases = array(
330 * XHTML sanitizer for MediaWiki
335 * Cleans up HTML, removes dangerous tags and attributes, and
336 * removes HTML comments
338 * @param $text String
339 * @param $processCallback Callback to do any variable or parameter replacements in HTML attribute values
340 * @param $args Array for the processing callback
341 * @param $extratags Array for any extra tags to include
342 * @param $removetags Array for any tags (default or extra) to exclude
345 static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array() ) {
348 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
349 $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
351 wfProfileIn( __METHOD__
);
353 if ( !$staticInitialised ) {
355 $htmlpairsStatic = array( # Tags that must be closed
356 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
357 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
358 'strike', 'strong', 'tt', 'var', 'div', 'center',
359 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
360 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
363 'br', 'hr', 'li', 'dt', 'dd'
365 $htmlsingleonly = array( # Elements that cannot have close tags
368 $htmlnest = array( # Tags that can be nested--??
369 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
370 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
372 $tabletags = array( # Can only appear inside table, we will close them
375 $htmllist = array( # Tags used by list
378 $listtags = array( # Tags that can appear in a list
382 $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
383 $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
385 # Convert them all to hashtables for faster lookup
386 $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
387 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
388 foreach ( $vars as $var ) {
389 $
$var = array_flip( $
$var );
391 $staticInitialised = true;
393 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
394 $extratags = array_flip( $extratags );
395 $removetags = array_flip( $removetags );
396 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
397 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ) , $removetags );
399 # Remove HTML comments
400 $text = Sanitizer
::removeHTMLcomments( $text );
401 $bits = explode( '<', $text );
402 $text = str_replace( '>', '>', array_shift( $bits ) );
404 $tagstack = $tablestack = array();
405 foreach ( $bits as $x ) {
407 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
408 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
410 $slash = $t = $params = $brace = $rest = null;
414 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
418 if( isset( $htmlsingleonly[$t] ) ) {
420 } elseif ( ( $ot = @array_pop
( $tagstack ) ) != $t ) {
421 if ( isset( $htmlsingleallowed[$ot] ) ) {
422 # Pop all elements with an optional close tag
423 # and see if we find a match below them
425 array_push ($optstack, $ot);
426 while ( ( ( $ot = @array_pop
( $tagstack ) ) != $t ) &&
427 isset( $htmlsingleallowed[$ot] ) )
429 array_push ($optstack, $ot);
432 # No match. Push the optinal elements back again
434 while ( $ot = @array_pop
( $optstack ) ) {
435 array_push( $tagstack, $ot );
439 @array_push
( $tagstack, $ot );
440 # <li> can be nested in <ul> or <ol>, skip those cases:
441 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
446 if ( $t == 'table' ) {
447 $tagstack = array_pop( $tablestack );
452 # Keep track for later
453 if ( isset( $tabletags[$t] ) &&
454 ! in_array( 'table', $tagstack ) ) {
456 } else if ( in_array( $t, $tagstack ) &&
457 ! isset( $htmlnest [$t ] ) ) {
459 #Â Is it a self closed htmlpair ? (bug 5487)
460 } else if( $brace == '/>' &&
461 isset( $htmlpairs[$t] ) ) {
463 } elseif( isset( $htmlsingleonly[$t] ) ) {
464 # Hack to force empty tag for uncloseable elements
466 } else if( isset( $htmlsingle[$t] ) ) {
467 # Hack to not close $htmlsingle tags
469 } else if( isset( $tabletags[$t] )
470 && in_array($t ,$tagstack) ) {
471 // New table tag but forgot to close the previous one
474 if ( $t == 'table' ) {
475 array_push( $tablestack, $tagstack );
478 array_push( $tagstack, $t );
481 # Replace any variables or template parameters with
483 if( is_callable( $processCallback ) ) {
484 call_user_func_array( $processCallback, array( &$params, $args ) );
487 # Strip non-approved attributes from the tag
488 $newparams = Sanitizer
::fixTagAttributes( $params, $t );
491 $rest = str_replace( '>', '>', $rest );
492 $close = ( $brace == '/>' && !$slash ) ?
' /' : '';
493 $text .= "<$slash$t$newparams$close>$rest";
497 $text .= '<' . str_replace( '>', '>', $x);
499 # Close off any remaining tags
500 while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
502 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
505 # this might be possible using tidy itself
506 foreach ( $bits as $x ) {
507 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
509 @list
( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
510 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
511 if( is_callable( $processCallback ) ) {
512 call_user_func_array( $processCallback, array( &$params, $args ) );
514 $newparams = Sanitizer
::fixTagAttributes( $params, $t );
515 $rest = str_replace( '>', '>', $rest );
516 $text .= "<$slash$t$newparams$brace$rest";
518 $text .= '<' . str_replace( '>', '>', $x);
522 wfProfileOut( __METHOD__
);
527 * Remove '<!--', '-->', and everything between.
528 * To avoid leaving blank lines, when a comment is both preceded
529 * and followed by a newline (ignoring spaces), trim leading and
530 * trailing spaces and one of the newlines.
533 * @param $text String
536 static function removeHTMLcomments( $text ) {
537 wfProfileIn( __METHOD__
);
538 while (($start = strpos($text, '<!--')) !== false) {
539 $end = strpos($text, '-->', $start +
4);
540 if ($end === false) {
541 # Unterminated comment; bail out
547 # Trim space and newline if the comment is both
548 # preceded and followed by a newline
549 $spaceStart = max($start - 1, 0);
550 $spaceLen = $end - $spaceStart;
551 while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
555 while (substr($text, $spaceStart +
$spaceLen, 1) === ' ')
557 if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart +
$spaceLen, 1) === "\n") {
558 # Remove the comment, leading and trailing
559 # spaces, and leave only one newline.
560 $text = substr_replace($text, "\n", $spaceStart, $spaceLen +
1);
563 # Remove just the comment.
564 $text = substr_replace($text, '', $start, $end - $start);
567 wfProfileOut( __METHOD__
);
572 * Take an array of attribute names and values and normalize or discard
573 * illegal values for the given element type.
575 * - Discards attributes not on a whitelist for the given element
576 * - Unsafe style attributes are discarded
577 * - Invalid id attributes are reencoded
579 * @param $attribs Array
580 * @param $element String
583 * @todo Check for legal values where the DTD limits things.
584 * @todo Check for unique id attribute :P
586 static function validateTagAttributes( $attribs, $element ) {
587 return Sanitizer
::validateAttributes( $attribs,
588 Sanitizer
::attributeWhitelist( $element ) );
592 * Take an array of attribute names and values and normalize or discard
593 * illegal values for the given whitelist.
595 * - Discards attributes not the given whitelist
596 * - Unsafe style attributes are discarded
597 * - Invalid id attributes are reencoded
599 * @param $attribs Array
600 * @param $whitelist Array: list of allowed attribute names
603 * @todo Check for legal values where the DTD limits things.
604 * @todo Check for unique id attribute :P
606 static function validateAttributes( $attribs, $whitelist ) {
607 $whitelist = array_flip( $whitelist );
609 foreach( $attribs as $attribute => $value ) {
610 if( !isset( $whitelist[$attribute] ) ) {
613 # Strip javascript "expression" from stylesheets.
614 # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
615 if( $attribute == 'style' ) {
616 $value = Sanitizer
::checkCss( $value );
617 if( $value === false ) {
623 if ( $attribute === 'id' ) {
624 global $wgEnforceHtmlIds;
625 $value = Sanitizer
::escapeId( $value,
626 $wgEnforceHtmlIds ?
'noninitial' : 'xml' );
629 // If this attribute was previously set, override it.
630 // Output should only have one attribute of each name.
631 $out[$attribute] = $value;
637 * Merge two sets of HTML attributes. Conflicting items in the second set
638 * will override those in the first, except for 'class' attributes which
639 * will be combined (if they're both strings).
641 * @todo implement merging for other attributes such as style
646 static function mergeAttributes( $a, $b ) {
647 $out = array_merge( $a, $b );
648 if( isset( $a['class'] ) && isset( $b['class'] )
649 && is_string( $a['class'] ) && is_string( $b['class'] )
650 && $a['class'] !== $b['class'] ) {
651 $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
652 -1, PREG_SPLIT_NO_EMPTY
);
653 $out['class'] = implode( ' ', array_unique( $classes ) );
659 * Pick apart some CSS and check it for forbidden or unsafe structures.
660 * Returns a sanitized string, or false if it was just too evil.
662 * Currently URL references, 'expression', 'tps' are forbidden.
664 * @param $value String
667 static function checkCss( $value ) {
668 $stripped = Sanitizer
::decodeCharReferences( $value );
670 // Remove any comments; IE gets token splitting wrong
671 $stripped = StringUtils
::delimiterReplace( '/*', '*/', ' ', $stripped );
675 // ... and continue checks
676 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
677 'codepointToUtf8(hexdec("$1"))', $stripped );
678 $stripped = str_replace( '\\', '', $stripped );
679 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
689 * Take a tag soup fragment listing an HTML element's attributes
690 * and normalize it to well-formed XML, discarding unwanted attributes.
691 * Output is safe for further wikitext processing, with escaping of
692 * values that could trigger problems.
694 * - Normalizes attribute names to lowercase
695 * - Discards attributes not on a whitelist for the given element
696 * - Turns broken or invalid entities into plaintext
697 * - Double-quotes all attribute values
698 * - Attributes without values are given the name as attribute
699 * - Double attributes are discarded
700 * - Unsafe style attributes are discarded
701 * - Prepends space if there are attributes.
703 * @param $text String
704 * @param $element String
707 static function fixTagAttributes( $text, $element ) {
708 if( trim( $text ) == '' ) {
712 $stripped = Sanitizer
::validateTagAttributes(
713 Sanitizer
::decodeTagAttributes( $text ), $element );
716 foreach( $stripped as $attribute => $value ) {
717 $encAttribute = htmlspecialchars( $attribute );
718 $encValue = Sanitizer
::safeEncodeAttribute( $value );
720 $attribs[] = "$encAttribute=\"$encValue\"";
722 return count( $attribs ) ?
' ' . implode( ' ', $attribs ) : '';
726 * Encode an attribute value for HTML output.
727 * @param $text String
728 * @return HTML-encoded text fragment
730 static function encodeAttribute( $text ) {
731 $encValue = htmlspecialchars( $text, ENT_QUOTES
);
733 // Whitespace is normalized during attribute decoding,
734 // so if we've been passed non-spaces we must encode them
735 // ahead of time or they won't be preserved.
736 $encValue = strtr( $encValue, array(
746 * Encode an attribute value for HTML tags, with extra armoring
747 * against further wiki processing.
748 * @param $text String
749 * @return HTML-encoded text fragment
751 static function safeEncodeAttribute( $text ) {
752 $encValue = Sanitizer
::encodeAttribute( $text );
754 # Templates and links may be expanded in later parsing,
755 # creating invalid or dangerous output. Suppress this.
756 $encValue = strtr( $encValue, array(
757 '<' => '<', // This should never happen,
758 '>' => '>', // we've received invalid input
759 '"' => '"', // which should have been escaped.
762 "''" => '''',
763 'ISBN' => 'ISBN',
765 'PMID' => 'PMID',
771 $encValue = preg_replace_callback(
772 '/(' . wfUrlProtocols() . ')/',
773 array( 'Sanitizer', 'armorLinksCallback' ),
779 * Given a value escape it so that it can be used in an id attribute and
780 * return it, this does not validate the value however (see first link)
782 * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
785 * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
787 * @param $id String: id to validate
788 * @param $options Mixed: string or array of strings (default is array()):
789 * 'noninitial': This is a non-initial fragment of an id, not a full id,
790 * so don't pay attention if the first character isn't valid at the
791 * beginning of an id.
792 * 'xml': Don't restrict the id to be HTML4-compatible. This option
793 * allows any alphabetic character to be used, per the XML standard.
794 * Therefore, it also completely changes the type of escaping: instead
795 * of weird dot-encoding, runs of invalid characters (mostly
796 * whitespace) are just compressed into a single underscore.
799 static function escapeId( $id, $options = array() ) {
800 $options = (array)$options;
802 if ( !in_array( 'xml', $options ) ) {
803 # HTML4-style escaping
804 static $replace = array(
809 $id = urlencode( Sanitizer
::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
810 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
812 if ( !preg_match( '/^[a-zA-Z]/', $id )
813 && !in_array( 'noninitial', $options ) ) {
814 // Initial character must be a letter!
820 # XML-style escaping. For the patterns used, see the XML 1.0 standard,
821 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
822 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
823 . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
824 . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
825 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
826 . '\x{203F}-\x{2040}';
827 # Replace _ as well so we don't get multiple consecutive underscores
828 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
829 $id = trim( $id, '_' );
831 if ( !preg_match( "/^[$nameStartChar]/u", $id )
832 && !in_array( 'noninitial', $options ) ) {
840 * Given a value, escape it so that it can be used as a CSS class and
843 * @todo For extra validity, input should be validated UTF-8.
845 * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
847 * @param $class String
850 static function escapeClass( $class ) {
851 // Convert ugly stuff to underscores and kill underscores in ugly places
852 return rtrim(preg_replace(
853 array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
859 * Given HTML input, escape with htmlspecialchars but un-escape entites.
860 * This allows (generally harmless) entities like to survive.
862 * @param $html String to escape
863 * @return String: escaped input
865 static function escapeHtmlAllowEntities( $html ) {
866 # It seems wise to escape ' as well as ", as a matter of course. Can't
868 $html = htmlspecialchars( $html, ENT_QUOTES
);
869 $html = str_replace( '&', '&', $html );
870 $html = Sanitizer
::normalizeCharReferences( $html );
875 * Regex replace callback for armoring links against further processing.
876 * @param $matches Array
879 private static function armorLinksCallback( $matches ) {
880 return str_replace( ':', ':', $matches[1] );
884 * Return an associative array of attribute names and values from
885 * a partial tag string. Attribute names are forces to lowercase,
886 * character references are decoded to UTF-8 text.
888 * @param $text String
891 public static function decodeTagAttributes( $text ) {
894 if( trim( $text ) == '' ) {
907 foreach( $pairs as $set ) {
908 $attribute = strtolower( $set[1] );
909 $value = Sanitizer
::getTagAttributeCallback( $set );
911 // Normalize whitespace
912 $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
913 $value = trim( $value );
915 // Decode character references
916 $attribs[$attribute] = Sanitizer
::decodeCharReferences( $value );
922 * Pick the appropriate attribute value from a match set from the
923 * MW_ATTRIBS_REGEX matches.
928 private static function getTagAttributeCallback( $set ) {
929 if( isset( $set[6] ) ) {
930 # Illegal #XXXXXX color with no quotes.
932 } elseif( isset( $set[5] ) ) {
935 } elseif( isset( $set[4] ) ) {
938 } elseif( isset( $set[3] ) ) {
941 } elseif( !isset( $set[2] ) ) {
942 # In XHTML, attributes must have a value.
943 # For 'reduced' form, return explicitly the attribute name here.
946 throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
951 * Normalize whitespace and character references in an XML source-
952 * encoded text for an attribute value.
954 * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
955 * but note that we're not returning the value, but are returning
956 * XML source fragments that will be slapped into output.
958 * @param $text String
961 private static function normalizeAttributeValue( $text ) {
962 return str_replace( '"', '"',
963 self
::normalizeWhitespace(
964 Sanitizer
::normalizeCharReferences( $text ) ) );
967 private static function normalizeWhitespace( $text ) {
969 '/\r\n|[\x20\x0d\x0a\x09]/',
975 * Ensure that any entities and character references are legal
976 * for XML and XHTML specifically. Any stray bits will be
977 * &-escaped to result in a valid text fragment.
979 * a. any named char refs must be known in XHTML
980 * b. any numeric char refs must be legal chars, not invalid or forbidden
981 * c. use &#x, not &#X
982 * d. fix or reject non-valid attributes
984 * @param $text String
988 static function normalizeCharReferences( $text ) {
989 return preg_replace_callback(
991 array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
995 * @param $matches String
998 static function normalizeCharReferencesCallback( $matches ) {
1000 if( $matches[1] != '' ) {
1001 $ret = Sanitizer
::normalizeEntity( $matches[1] );
1002 } elseif( $matches[2] != '' ) {
1003 $ret = Sanitizer
::decCharReference( $matches[2] );
1004 } elseif( $matches[3] != '' ) {
1005 $ret = Sanitizer
::hexCharReference( $matches[3] );
1006 } elseif( $matches[4] != '' ) {
1007 $ret = Sanitizer
::hexCharReference( $matches[4] );
1009 if( is_null( $ret ) ) {
1010 return htmlspecialchars( $matches[0] );
1017 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1018 * return the named entity reference as is. If the entity is a
1019 * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1020 * returns HTML-escaped text of pseudo-entity source (eg &foo;)
1022 * @param $name String
1025 static function normalizeEntity( $name ) {
1026 global $wgHtmlEntities, $wgHtmlEntityAliases;
1027 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1028 return "&{$wgHtmlEntityAliases[$name]};";
1029 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1032 return "&$name;";
1036 static function decCharReference( $codepoint ) {
1037 $point = intval( $codepoint );
1038 if( Sanitizer
::validateCodepoint( $point ) ) {
1039 return sprintf( '&#%d;', $point );
1045 static function hexCharReference( $codepoint ) {
1046 $point = hexdec( $codepoint );
1047 if( Sanitizer
::validateCodepoint( $point ) ) {
1048 return sprintf( '&#x%x;', $point );
1055 * Returns true if a given Unicode codepoint is a valid character in XML.
1056 * @param $codepoint Integer
1059 private static function validateCodepoint( $codepoint ) {
1060 return ($codepoint == 0x09)
1061 ||
($codepoint == 0x0a)
1062 ||
($codepoint == 0x0d)
1063 ||
($codepoint >= 0x20 && $codepoint <= 0xd7ff)
1064 ||
($codepoint >= 0xe000 && $codepoint <= 0xfffd)
1065 ||
($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1069 * Decode any character references, numeric or named entities,
1070 * in the text and return a UTF-8 string.
1072 * @param $text String
1075 public static function decodeCharReferences( $text ) {
1076 return preg_replace_callback(
1078 array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1083 * @param $matches String
1086 static function decodeCharReferencesCallback( $matches ) {
1087 if( $matches[1] != '' ) {
1088 return Sanitizer
::decodeEntity( $matches[1] );
1089 } elseif( $matches[2] != '' ) {
1090 return Sanitizer
::decodeChar( intval( $matches[2] ) );
1091 } elseif( $matches[3] != '' ) {
1092 return Sanitizer
::decodeChar( hexdec( $matches[3] ) );
1093 } elseif( $matches[4] != '' ) {
1094 return Sanitizer
::decodeChar( hexdec( $matches[4] ) );
1096 # Last case should be an ampersand by itself
1101 * Return UTF-8 string for a codepoint if that is a valid
1102 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1103 * @param $codepoint Integer
1107 static function decodeChar( $codepoint ) {
1108 if( Sanitizer
::validateCodepoint( $codepoint ) ) {
1109 return codepointToUtf8( $codepoint );
1111 return UTF8_REPLACEMENT
;
1116 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1117 * return the UTF-8 encoding of that character. Otherwise, returns
1118 * pseudo-entity source (eg &foo;)
1120 * @param $name Strings
1123 static function decodeEntity( $name ) {
1124 global $wgHtmlEntities, $wgHtmlEntityAliases;
1125 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1126 $name = $wgHtmlEntityAliases[$name];
1128 if( isset( $wgHtmlEntities[$name] ) ) {
1129 return codepointToUtf8( $wgHtmlEntities[$name] );
1136 * Fetch the whitelist of acceptable attributes for a given element name.
1138 * @param $element String
1141 static function attributeWhitelist( $element ) {
1143 if( !isset( $list ) ) {
1144 $list = Sanitizer
::setupAttributeWhitelist();
1146 return isset( $list[$element] )
1152 * Foreach array key (an allowed HTML element), return an array
1153 * of allowed attributes
1156 static function setupAttributeWhitelist() {
1157 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1158 $block = array_merge( $common, array( 'align' ) );
1159 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1160 $tablecell = array( 'abbr',
1166 'nowrap', # deprecated
1167 'width', # deprecated
1168 'height', # deprecated
1169 'bgcolor' # deprecated
1172 # Numbers refer to sections in HTML 4.01 standard describing the element.
1173 # See: http://www.w3.org/TR/html4/
1174 $whitelist = array (
1177 'center' => $common, # deprecated
1178 'span' => $block, # ??
1196 'strong' => $common,
1207 'blockquote' => array_merge( $common, array( 'cite' ) ),
1218 'br' => array( 'id', 'class', 'title', 'style', 'clear' ),
1221 'pre' => array_merge( $common, array( 'width' ) ),
1224 'ins' => array_merge( $common, array( 'cite', 'datetime' ) ),
1225 'del' => array_merge( $common, array( 'cite', 'datetime' ) ),
1228 'ul' => array_merge( $common, array( 'type' ) ),
1229 'ol' => array_merge( $common, array( 'type', 'start' ) ),
1230 'li' => array_merge( $common, array( 'type', 'value' ) ),
1238 'table' => array_merge( $common,
1239 array( 'summary', 'width', 'border', 'frame',
1240 'rules', 'cellspacing', 'cellpadding',
1245 'caption' => array_merge( $common, array( 'align' ) ),
1248 'thead' => array_merge( $common, $tablealign ),
1249 'tfoot' => array_merge( $common, $tablealign ),
1250 'tbody' => array_merge( $common, $tablealign ),
1253 'colgroup' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1254 'col' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1257 'tr' => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1260 'td' => array_merge( $common, $tablecell, $tablealign ),
1261 'th' => array_merge( $common, $tablecell, $tablealign ),
1264 # Not usually allowed, but may be used for extension-style hooks
1265 # such as <math> when it is rasterized
1266 'img' => array_merge( $common, array( 'alt' ) ),
1274 'strike' => $common,
1279 'font' => array_merge( $common, array( 'size', 'color', 'face' ) ),
1283 'hr' => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1285 # XHTML Ruby annotation text module, simple ruby only.
1286 # http://www.w3c.org/TR/ruby/
1291 'rt' => $common, #array_merge( $common, array( 'rbspan' ) ),
1294 # MathML root element, where used for extensions
1295 # 'title' may not be 100% valid here; it's XHTML
1296 # http://www.w3.org/TR/REC-MathML/
1297 'math' => array( 'class', 'style', 'id', 'title' ),
1303 * Take a fragment of (potentially invalid) HTML and return
1304 * a version with any tags removed, encoded as plain text.
1306 * Warning: this return value must be further escaped for literal
1307 * inclusion in HTML output as of 1.10!
1309 * @param $text String: HTML fragment
1312 static function stripAllTags( $text ) {
1314 $text = StringUtils
::delimiterReplace( '<', '>', '', $text );
1316 # Normalize &entities and whitespace
1317 $text = self
::decodeCharReferences( $text );
1318 $text = self
::normalizeWhitespace( $text );
1324 * Hack up a private DOCTYPE with HTML's standard entity declarations.
1325 * PHP 4 seemed to know these if you gave it an HTML doctype, but
1328 * Use for passing XHTML fragments to PHP's XML parsing functions
1332 static function hackDocType() {
1333 global $wgHtmlEntities;
1334 $out = "<!DOCTYPE html [\n";
1335 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1336 $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1342 static function cleanUrl( $url ) {
1343 # Normalize any HTML entities in input. They will be
1344 # re-escaped by makeExternalLink().
1345 $url = Sanitizer
::decodeCharReferences( $url );
1347 # Escape any control characters introduced by the above step
1348 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1350 # Validate hostname portion
1352 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1353 list( /* $whole */, $protocol, $host, $rest ) = $matches;
1355 // Characters that will be ignored in IDNs.
1356 // http://tools.ietf.org/html/3454#section-3.1
1357 // Strip them before further processing so blacklists and such work.
1359 \\s| # general whitespace
1360 \xc2\xad| # 00ad SOFT HYPHEN
1361 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1362 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1363 \xe2\x81\xa0| # 2060 WORD JOINER
1364 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1365 \xcd\x8f| # 034f COMBINING GRAPHEME JOINER
1366 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1367 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1368 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1369 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1370 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1371 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1374 $host = preg_replace( $strip, '', $host );
1376 // @fixme: validate hostnames here
1378 return $protocol . $host . $rest;