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 string $text
339 * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
340 * @param array $args for the processing callback
343 static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array() ) {
346 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
347 $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
349 wfProfileIn( __METHOD__ );
351 if ( !$staticInitialised ) {
353 $htmlpairs = array_merge( $extratags, array( # Tags that must be closed
354 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
355 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
356 'strike', 'strong', 'tt', 'var', 'div', 'center',
357 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
358 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
361 'br', 'hr', 'li', 'dt', 'dd'
363 $htmlsingleonly = array( # Elements that cannot have close tags
366 $htmlnest = array( # Tags that can be nested--??
367 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
368 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
370 $tabletags = array( # Can only appear inside table, we will close them
373 $htmllist = array( # Tags used by list
376 $listtags = array( # Tags that can appear in a list
380 $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
381 $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
383 # Convert them all to hashtables for faster lookup
384 $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
385 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
386 foreach ( $vars as $var ) {
387 $$var = array_flip( $$var );
389 $staticInitialised = true;
392 # Remove HTML comments
393 $text = Sanitizer::removeHTMLcomments( $text );
394 $bits = explode( '<', $text );
395 $text = str_replace( '>', '>', array_shift( $bits ) );
397 $tagstack = $tablestack = array();
398 foreach ( $bits as $x ) {
400 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
401 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
403 $slash = $t = $params = $brace = $rest = null;
407 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
411 if( isset( $htmlsingleonly[$t] ) ) {
413 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
414 if ( isset( $htmlsingleallowed[$ot] ) ) {
415 # Pop all elements with an optional close tag
416 # and see if we find a match below them
418 array_push ($optstack, $ot);
419 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
420 isset( $htmlsingleallowed[$ot] ) )
422 array_push ($optstack, $ot);
425 # No match. Push the optinal elements back again
427 while ( $ot = @array_pop( $optstack ) ) {
428 array_push( $tagstack, $ot );
432 @array_push( $tagstack, $ot );
433 # <li> can be nested in <ul> or <ol>, skip those cases:
434 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
439 if ( $t == 'table' ) {
440 $tagstack = array_pop( $tablestack );
445 # Keep track for later
446 if ( isset( $tabletags[$t] ) &&
447 ! in_array( 'table', $tagstack ) ) {
449 } else if ( in_array( $t, $tagstack ) &&
450 ! isset( $htmlnest [$t ] ) ) {
452 #Â Is it a self closed htmlpair ? (bug 5487)
453 } else if( $brace == '/>' &&
454 isset( $htmlpairs[$t] ) ) {
456 } elseif( isset( $htmlsingleonly[$t] ) ) {
457 # Hack to force empty tag for uncloseable elements
459 } else if( isset( $htmlsingle[$t] ) ) {
460 # Hack to not close $htmlsingle tags
462 } else if( isset( $tabletags[$t] )
463 && in_array($t ,$tagstack) ) {
464 // New table tag but forgot to close the previous one
467 if ( $t == 'table' ) {
468 array_push( $tablestack, $tagstack );
471 array_push( $tagstack, $t );
474 # Replace any variables or template parameters with
476 if( is_callable( $processCallback ) ) {
477 call_user_func_array( $processCallback, array( &$params, $args ) );
480 # Strip non-approved attributes from the tag
481 $newparams = Sanitizer::fixTagAttributes( $params, $t );
484 $rest = str_replace( '>', '>', $rest );
485 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
486 $text .= "<$slash$t$newparams$close>$rest";
490 $text .= '<' . str_replace( '>', '>', $x);
492 # Close off any remaining tags
493 while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
495 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
498 # this might be possible using tidy itself
499 foreach ( $bits as $x ) {
500 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
502 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
503 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
504 if( is_callable( $processCallback ) ) {
505 call_user_func_array( $processCallback, array( &$params, $args ) );
507 $newparams = Sanitizer::fixTagAttributes( $params, $t );
508 $rest = str_replace( '>', '>', $rest );
509 $text .= "<$slash$t$newparams$brace$rest";
511 $text .= '<' . str_replace( '>', '>', $x);
515 wfProfileOut( __METHOD__ );
520 * Remove '<!--', '-->', and everything between.
521 * To avoid leaving blank lines, when a comment is both preceded
522 * and followed by a newline (ignoring spaces), trim leading and
523 * trailing spaces and one of the newlines.
526 * @param string $text
529 static function removeHTMLcomments( $text ) {
530 wfProfileIn( __METHOD__ );
531 while (($start = strpos($text, '<!--')) !== false) {
532 $end = strpos($text, '-->', $start + 4);
533 if ($end === false) {
534 # Unterminated comment; bail out
540 # Trim space and newline if the comment is both
541 # preceded and followed by a newline
542 $spaceStart = max($start - 1, 0);
543 $spaceLen = $end - $spaceStart;
544 while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
548 while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
550 if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
551 # Remove the comment, leading and trailing
552 # spaces, and leave only one newline.
553 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
556 # Remove just the comment.
557 $text = substr_replace($text, '', $start, $end - $start);
560 wfProfileOut( __METHOD__ );
565 * Take an array of attribute names and values and normalize or discard
566 * illegal values for the given element type.
568 * - Discards attributes not on a whitelist for the given element
569 * - Unsafe style attributes are discarded
570 * - Invalid id attributes are reencoded
572 * @param array $attribs
573 * @param string $element
576 * @todo Check for legal values where the DTD limits things.
577 * @todo Check for unique id attribute :P
579 static function validateTagAttributes( $attribs, $element ) {
580 return Sanitizer::validateAttributes( $attribs,
581 Sanitizer::attributeWhitelist( $element ) );
585 * Take an array of attribute names and values and normalize or discard
586 * illegal values for the given whitelist.
588 * - Discards attributes not the given whitelist
589 * - Unsafe style attributes are discarded
590 * - Invalid id attributes are reencoded
592 * @param array $attribs
593 * @param array $whitelist list of allowed attribute names
596 * @todo Check for legal values where the DTD limits things.
597 * @todo Check for unique id attribute :P
599 static function validateAttributes( $attribs, $whitelist ) {
600 $whitelist = array_flip( $whitelist );
602 foreach( $attribs as $attribute => $value ) {
603 if( !isset( $whitelist[$attribute] ) ) {
606 # Strip javascript "expression" from stylesheets.
607 # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
608 if( $attribute == 'style' ) {
609 $value = Sanitizer::checkCss( $value );
610 if( $value === false ) {
616 if ( $attribute === 'id' ) {
617 global $wgEnforceHtmlIds;
618 $value = Sanitizer::escapeId( $value,
619 $wgEnforceHtmlIds ? 'noninitial' : 'xml' );
622 // If this attribute was previously set, override it.
623 // Output should only have one attribute of each name.
624 $out[$attribute] = $value;
630 * Merge two sets of HTML attributes. Conflicting items in the second set
631 * will override those in the first, except for 'class' attributes which
632 * will be combined (if they're both strings).
634 * @todo implement merging for other attributes such as style
639 static function mergeAttributes( $a, $b ) {
640 $out = array_merge( $a, $b );
641 if( isset( $a['class'] ) && isset( $b['class'] )
642 && is_string( $a['class'] ) && is_string( $b['class'] )
643 && $a['class'] !== $b['class'] ) {
644 $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
645 -1, PREG_SPLIT_NO_EMPTY );
646 $out['class'] = implode( ' ', array_unique( $classes ) );
652 * Pick apart some CSS and check it for forbidden or unsafe structures.
653 * Returns a sanitized string, or false if it was just too evil.
655 * Currently URL references, 'expression', 'tps' are forbidden.
657 * @param string $value
660 static function checkCss( $value ) {
661 $value = Sanitizer::decodeCharReferences( $value );
663 // Remove any comments; IE gets token splitting wrong
664 $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
666 // Decode escape sequences and line continuation
667 // See the grammar in the CSS 2 spec, appendix D, Mozilla implements it accurately.
668 // IE 8 doesn't implement it at all, but there's no way to introduce url() into
669 // IE that doesn't hit Mozilla also.
671 if ( !$decodeRegex ) {
672 $space = '[\\x20\\t\\r\\n\\f]';
673 $nl = '(?:\\n|\\r\\n|\\r|\\f)';
675 $decodeRegex = "/ $backslash
677 ($nl) | # 1. Line continuation
678 ([0-9A-Fa-f]{1,6})$space? | # 2. character number
679 (.) # 3. backslash cancelling special meaning
682 $decoded = preg_replace_callback( $decodeRegex,
683 array( __CLASS__, 'cssDecodeCallback' ), $value );
684 if ( preg_match( '!expression|https?://|url\s*\(!i', $decoded ) ) {
688 // Allowed, return CSS with comments stripped
693 static function cssDecodeCallback( $matches ) {
694 if ( $matches[1] !== '' ) {
696 } elseif ( $matches[2] !== '' ) {
697 return codepointToUtf8( hexdec( $matches[2] ) );
698 } elseif ( $matches[3] !== '' ) {
701 throw new MWException( __METHOD__.': invalid match' );
706 * Take a tag soup fragment listing an HTML element's attributes
707 * and normalize it to well-formed XML, discarding unwanted attributes.
708 * Output is safe for further wikitext processing, with escaping of
709 * values that could trigger problems.
711 * - Normalizes attribute names to lowercase
712 * - Discards attributes not on a whitelist for the given element
713 * - Turns broken or invalid entities into plaintext
714 * - Double-quotes all attribute values
715 * - Attributes without values are given the name as attribute
716 * - Double attributes are discarded
717 * - Unsafe style attributes are discarded
718 * - Prepends space if there are attributes.
720 * @param string $text
721 * @param string $element
724 static function fixTagAttributes( $text, $element ) {
725 if( trim( $text ) == '' ) {
729 $stripped = Sanitizer::validateTagAttributes(
730 Sanitizer::decodeTagAttributes( $text ), $element );
733 foreach( $stripped as $attribute => $value ) {
734 $encAttribute = htmlspecialchars( $attribute );
735 $encValue = Sanitizer::safeEncodeAttribute( $value );
737 $attribs[] = "$encAttribute=\"$encValue\"";
739 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
743 * Encode an attribute value for HTML output.
745 * @return HTML-encoded text fragment
747 static function encodeAttribute( $text ) {
748 $encValue = htmlspecialchars( $text, ENT_QUOTES );
750 // Whitespace is normalized during attribute decoding,
751 // so if we've been passed non-spaces we must encode them
752 // ahead of time or they won't be preserved.
753 $encValue = strtr( $encValue, array(
763 * Encode an attribute value for HTML tags, with extra armoring
764 * against further wiki processing.
766 * @return HTML-encoded text fragment
768 static function safeEncodeAttribute( $text ) {
769 $encValue = Sanitizer::encodeAttribute( $text );
771 # Templates and links may be expanded in later parsing,
772 # creating invalid or dangerous output. Suppress this.
773 $encValue = strtr( $encValue, array(
774 '<' => '<', // This should never happen,
775 '>' => '>', // we've received invalid input
776 '"' => '"', // which should have been escaped.
779 "''" => '''',
780 'ISBN' => 'ISBN',
782 'PMID' => 'PMID',
788 $encValue = preg_replace_callback(
789 '/(' . wfUrlProtocols() . ')/',
790 array( 'Sanitizer', 'armorLinksCallback' ),
796 * Given a value escape it so that it can be used in an id attribute and
797 * return it, this does not validate the value however (see first link)
799 * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
802 * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
804 * @param string $id Id to validate
805 * @param mixed $options String or array of strings (default is array()):
806 * 'noninitial': This is a non-initial fragment of an id, not a full id,
807 * so don't pay attention if the first character isn't valid at the
808 * beginning of an id.
809 * 'xml': Don't restrict the id to be HTML4-compatible. This option
810 * allows any alphabetic character to be used, per the XML standard.
811 * Therefore, it also completely changes the type of escaping: instead
812 * of weird dot-encoding, runs of invalid characters (mostly
813 * whitespace) are just compressed into a single underscore.
816 static function escapeId( $id, $options = array() ) {
817 $options = (array)$options;
819 if ( !in_array( 'xml', $options ) ) {
820 # HTML4-style escaping
821 static $replace = array(
826 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
827 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
829 if ( !preg_match( '/^[a-zA-Z]/', $id )
830 && !in_array( 'noninitial', $options ) ) {
831 // Initial character must be a letter!
837 # XML-style escaping. For the patterns used, see the XML 1.0 standard,
838 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
839 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
840 . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
841 . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
842 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
843 . '\x{203F}-\x{2040}';
844 # Replace _ as well so we don't get multiple consecutive underscores
845 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
846 $id = trim( $id, '_' );
848 if ( !preg_match( "/^[$nameStartChar]/u", $id )
849 && !in_array( 'noninitial', $options ) ) {
857 * Given a value, escape it so that it can be used as a CSS class and
860 * @todo For extra validity, input should be validated UTF-8.
862 * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
864 * @param string $class
867 static function escapeClass( $class ) {
868 // Convert ugly stuff to underscores and kill underscores in ugly places
869 return rtrim(preg_replace(
870 array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
876 * Given HTML input, escape with htmlspecialchars but un-escape entites.
877 * This allows (generally harmless) entities like to survive.
879 * @param string $html String to escape
880 * @return string Escaped input
882 static function escapeHtmlAllowEntities( $html ) {
883 # It seems wise to escape ' as well as ", as a matter of course. Can't
885 $html = htmlspecialchars( $html, ENT_QUOTES );
886 $html = str_replace( '&', '&', $html );
887 $html = Sanitizer::normalizeCharReferences( $html );
892 * Regex replace callback for armoring links against further processing.
893 * @param array $matches
897 private static function armorLinksCallback( $matches ) {
898 return str_replace( ':', ':', $matches[1] );
902 * Return an associative array of attribute names and values from
903 * a partial tag string. Attribute names are forces to lowercase,
904 * character references are decoded to UTF-8 text.
909 public static function decodeTagAttributes( $text ) {
912 if( trim( $text ) == '' ) {
925 foreach( $pairs as $set ) {
926 $attribute = strtolower( $set[1] );
927 $value = Sanitizer::getTagAttributeCallback( $set );
929 // Normalize whitespace
930 $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
931 $value = trim( $value );
933 // Decode character references
934 $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
940 * Pick the appropriate attribute value from a match set from the
941 * MW_ATTRIBS_REGEX matches.
947 private static function getTagAttributeCallback( $set ) {
948 if( isset( $set[6] ) ) {
949 # Illegal #XXXXXX color with no quotes.
951 } elseif( isset( $set[5] ) ) {
954 } elseif( isset( $set[4] ) ) {
957 } elseif( isset( $set[3] ) ) {
960 } elseif( !isset( $set[2] ) ) {
961 # In XHTML, attributes must have a value.
962 # For 'reduced' form, return explicitly the attribute name here.
965 throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
970 * Normalize whitespace and character references in an XML source-
971 * encoded text for an attribute value.
973 * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
974 * but note that we're not returning the value, but are returning
975 * XML source fragments that will be slapped into output.
977 * @param string $text
981 private static function normalizeAttributeValue( $text ) {
982 return str_replace( '"', '"',
983 self::normalizeWhitespace(
984 Sanitizer::normalizeCharReferences( $text ) ) );
987 private static function normalizeWhitespace( $text ) {
989 '/\r\n|[\x20\x0d\x0a\x09]/',
995 * Ensure that any entities and character references are legal
996 * for XML and XHTML specifically. Any stray bits will be
997 * &-escaped to result in a valid text fragment.
999 * a. any named char refs must be known in XHTML
1000 * b. any numeric char refs must be legal chars, not invalid or forbidden
1001 * c. use &#x, not &#X
1002 * d. fix or reject non-valid attributes
1004 * @param string $text
1008 static function normalizeCharReferences( $text ) {
1009 return preg_replace_callback(
1011 array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1015 * @param string $matches
1018 static function normalizeCharReferencesCallback( $matches ) {
1020 if( $matches[1] != '' ) {
1021 $ret = Sanitizer::normalizeEntity( $matches[1] );
1022 } elseif( $matches[2] != '' ) {
1023 $ret = Sanitizer::decCharReference( $matches[2] );
1024 } elseif( $matches[3] != '' ) {
1025 $ret = Sanitizer::hexCharReference( $matches[3] );
1026 } elseif( $matches[4] != '' ) {
1027 $ret = Sanitizer::hexCharReference( $matches[4] );
1029 if( is_null( $ret ) ) {
1030 return htmlspecialchars( $matches[0] );
1037 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1038 * return the named entity reference as is. If the entity is a
1039 * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1040 * returns HTML-escaped text of pseudo-entity source (eg &foo;)
1042 * @param string $name
1046 static function normalizeEntity( $name ) {
1047 global $wgHtmlEntities, $wgHtmlEntityAliases;
1048 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1049 return "&{$wgHtmlEntityAliases[$name]};";
1050 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1053 return "&$name;";
1057 static function decCharReference( $codepoint ) {
1058 $point = intval( $codepoint );
1059 if( Sanitizer::validateCodepoint( $point ) ) {
1060 return sprintf( '&#%d;', $point );
1066 static function hexCharReference( $codepoint ) {
1067 $point = hexdec( $codepoint );
1068 if( Sanitizer::validateCodepoint( $point ) ) {
1069 return sprintf( '&#x%x;', $point );
1076 * Returns true if a given Unicode codepoint is a valid character in XML.
1077 * @param int $codepoint
1080 private static function validateCodepoint( $codepoint ) {
1081 return ($codepoint == 0x09)
1082 || ($codepoint == 0x0a)
1083 || ($codepoint == 0x0d)
1084 || ($codepoint >= 0x20 && $codepoint <= 0xd7ff)
1085 || ($codepoint >= 0xe000 && $codepoint <= 0xfffd)
1086 || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1090 * Decode any character references, numeric or named entities,
1091 * in the text and return a UTF-8 string.
1093 * @param string $text
1098 public static function decodeCharReferences( $text ) {
1099 return preg_replace_callback(
1101 array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1106 * @param string $matches
1109 static function decodeCharReferencesCallback( $matches ) {
1110 if( $matches[1] != '' ) {
1111 return Sanitizer::decodeEntity( $matches[1] );
1112 } elseif( $matches[2] != '' ) {
1113 return Sanitizer::decodeChar( intval( $matches[2] ) );
1114 } elseif( $matches[3] != '' ) {
1115 return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1116 } elseif( $matches[4] != '' ) {
1117 return Sanitizer::decodeChar( hexdec( $matches[4] ) );
1119 # Last case should be an ampersand by itself
1124 * Return UTF-8 string for a codepoint if that is a valid
1125 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1126 * @param int $codepoint
1130 static function decodeChar( $codepoint ) {
1131 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1132 return codepointToUtf8( $codepoint );
1134 return UTF8_REPLACEMENT;
1139 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1140 * return the UTF-8 encoding of that character. Otherwise, returns
1141 * pseudo-entity source (eg &foo;)
1143 * @param string $name
1146 static function decodeEntity( $name ) {
1147 global $wgHtmlEntities, $wgHtmlEntityAliases;
1148 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1149 $name = $wgHtmlEntityAliases[$name];
1151 if( isset( $wgHtmlEntities[$name] ) ) {
1152 return codepointToUtf8( $wgHtmlEntities[$name] );
1159 * Fetch the whitelist of acceptable attributes for a given
1162 * @param string $element
1165 static function attributeWhitelist( $element ) {
1167 if( !isset( $list ) ) {
1168 $list = Sanitizer::setupAttributeWhitelist();
1170 return isset( $list[$element] )
1176 * Foreach array key (an allowed HTML element), return an array
1177 * of allowed attributes
1180 static function setupAttributeWhitelist() {
1181 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1182 $block = array_merge( $common, array( 'align' ) );
1183 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1184 $tablecell = array( 'abbr',
1190 'nowrap', # deprecated
1191 'width', # deprecated
1192 'height', # deprecated
1193 'bgcolor' # deprecated
1196 # Numbers refer to sections in HTML 4.01 standard describing the element.
1197 # See: http://www.w3.org/TR/html4/
1198 $whitelist = array (
1201 'center' => $common, # deprecated
1202 'span' => $block, # ??
1220 'strong' => $common,
1231 'blockquote' => array_merge( $common, array( 'cite' ) ),
1242 'br' => array( 'id', 'class', 'title', 'style', 'clear' ),
1245 'pre' => array_merge( $common, array( 'width' ) ),
1248 'ins' => array_merge( $common, array( 'cite', 'datetime' ) ),
1249 'del' => array_merge( $common, array( 'cite', 'datetime' ) ),
1252 'ul' => array_merge( $common, array( 'type' ) ),
1253 'ol' => array_merge( $common, array( 'type', 'start' ) ),
1254 'li' => array_merge( $common, array( 'type', 'value' ) ),
1262 'table' => array_merge( $common,
1263 array( 'summary', 'width', 'border', 'frame',
1264 'rules', 'cellspacing', 'cellpadding',
1269 'caption' => array_merge( $common, array( 'align' ) ),
1272 'thead' => array_merge( $common, $tablealign ),
1273 'tfoot' => array_merge( $common, $tablealign ),
1274 'tbody' => array_merge( $common, $tablealign ),
1277 'colgroup' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1278 'col' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1281 'tr' => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1284 'td' => array_merge( $common, $tablecell, $tablealign ),
1285 'th' => array_merge( $common, $tablecell, $tablealign ),
1288 # Not usually allowed, but may be used for extension-style hooks
1289 # such as <math> when it is rasterized
1290 'img' => array_merge( $common, array( 'alt' ) ),
1298 'strike' => $common,
1303 'font' => array_merge( $common, array( 'size', 'color', 'face' ) ),
1307 'hr' => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1309 # XHTML Ruby annotation text module, simple ruby only.
1310 # http://www.w3c.org/TR/ruby/
1315 'rt' => $common, #array_merge( $common, array( 'rbspan' ) ),
1318 # MathML root element, where used for extensions
1319 # 'title' may not be 100% valid here; it's XHTML
1320 # http://www.w3.org/TR/REC-MathML/
1321 'math' => array( 'class', 'style', 'id', 'title' ),
1327 * Take a fragment of (potentially invalid) HTML and return
1328 * a version with any tags removed, encoded as plain text.
1330 * Warning: this return value must be further escaped for literal
1331 * inclusion in HTML output as of 1.10!
1333 * @param string $text HTML fragment
1336 static function stripAllTags( $text ) {
1338 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1340 # Normalize &entities and whitespace
1341 $text = self::decodeCharReferences( $text );
1342 $text = self::normalizeWhitespace( $text );
1348 * Hack up a private DOCTYPE with HTML's standard entity declarations.
1349 * PHP 4 seemed to know these if you gave it an HTML doctype, but
1352 * Use for passing XHTML fragments to PHP's XML parsing functions
1357 static function hackDocType() {
1358 global $wgHtmlEntities;
1359 $out = "<!DOCTYPE html [\n";
1360 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1361 $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1367 static function cleanUrl( $url ) {
1368 # Normalize any HTML entities in input. They will be
1369 # re-escaped by makeExternalLink().
1370 $url = Sanitizer::decodeCharReferences( $url );
1372 # Escape any control characters introduced by the above step
1373 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1375 # Validate hostname portion
1377 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1378 list( /* $whole */, $protocol, $host, $rest ) = $matches;
1380 // Characters that will be ignored in IDNs.
1381 // http://tools.ietf.org/html/3454#section-3.1
1382 // Strip them before further processing so blacklists and such work.
1384 \\s| # general whitespace
1385 \xc2\xad| # 00ad SOFT HYPHEN
1386 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1387 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1388 \xe2\x81\xa0| # 2060 WORD JOINER
1389 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1390 \xcd\x8f| # 034f COMBINING GRAPHEME JOINER
1391 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1392 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1393 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1394 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1395 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1396 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1399 $host = preg_replace( $strip, '', $host );
1401 // @fixme: validate hostnames here
1403 return $protocol . $host . $rest;