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',
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(
319 /** @package MediaWiki */
322 * Cleans up HTML, removes dangerous tags and attributes, and
323 * removes HTML comments
325 * @param string $text
326 * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
327 * @param array $args for the processing callback
330 function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
331 global $wgUseTidy, $wgUserHtml;
332 $fname = 'Parser::removeHTMLtags';
333 wfProfileIn( $fname );
336 $htmlpairs = array( # Tags that must be closed
337 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
338 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
339 'strike', 'strong', 'tt', 'var', 'div', 'center',
340 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
341 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
344 'br', 'hr', 'li', 'dt', 'dd'
346 $htmlsingleonly = array( # Elements that cannot have close tags
349 $htmlnest = array( # Tags that can be nested--??
350 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
351 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
353 $tabletags = array( # Can only appear inside table
356 $htmllist = array( # Tags used by list
359 $listtags = array( # Tags that can appear in a list
364 $htmlpairs = array();
365 $htmlsingle = array();
367 $tabletags = array();
370 $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
371 $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
373 # Remove HTML comments
374 $text = Sanitizer
::removeHTMLcomments( $text );
375 $bits = explode( '<', $text );
376 $text = array_shift( $bits );
378 $tagstack = array(); $tablestack = array();
379 foreach ( $bits as $x ) {
380 $prev = error_reporting( E_ALL
& ~
( E_NOTICE | E_WARNING
) );
381 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
383 list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
384 error_reporting( $prev );
387 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
391 if( in_array( $t, $htmlsingleonly ) ) {
393 } elseif ( ( $ot = @array_pop
( $tagstack ) ) != $t ) {
394 if ( in_array($ot, $htmlsingleallowed) ) {
395 # Pop all elements with an optional close tag
396 # and see if we find a match below them
398 array_push ($optstack, $ot);
399 while ( ( ( $ot = @array_pop
( $tagstack ) ) != $t ) &&
400 in_array($ot, $htmlsingleallowed) ) {
401 array_push ($optstack, $ot);
404 # No match. Push the optinal elements back again
406 while ( $ot = @array_pop
( $optstack ) ) {
407 array_push( $tagstack, $ot );
411 @array_push
( $tagstack, $ot );
412 # <li> can be nested in <ul> or <ol>, skip those cases:
413 if(!(in_array($ot, $htmllist) && in_array($t, $listtags) )) {
418 if ( $t == 'table' ) {
419 $tagstack = array_pop( $tablestack );
424 # Keep track for later
425 if ( in_array( $t, $tabletags ) &&
426 ! in_array( 'table', $tagstack ) ) {
428 } else if ( in_array( $t, $tagstack ) &&
429 ! in_array ( $t , $htmlnest ) ) {
431 #Â Is it a self closed htmlpair ? (bug 5487)
432 } else if( $brace == '/>' &&
433 in_array($t, $htmlpairs) ) {
435 } elseif( in_array( $t, $htmlsingleonly ) ) {
436 # Hack to force empty tag for uncloseable elements
438 } else if( in_array( $t, $htmlsingle ) ) {
439 # Hack to not close $htmlsingle tags
442 if ( $t == 'table' ) {
443 array_push( $tablestack, $tagstack );
446 array_push( $tagstack, $t );
449 # Replace any variables or template parameters with
451 if( is_callable( $processCallback ) ) {
452 call_user_func_array( $processCallback, array( &$params, $args ) );
455 # Strip non-approved attributes from the tag
456 $newparams = Sanitizer
::fixTagAttributes( $params, $t );
459 $rest = str_replace( '>', '>', $rest );
460 $close = ( $brace == '/>' ) ?
' /' : '';
461 $text .= "<$slash$t$newparams$close>$rest";
465 $text .= '<' . str_replace( '>', '>', $x);
467 # Close off any remaining tags
468 while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
470 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
473 # this might be possible using tidy itself
474 foreach ( $bits as $x ) {
475 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
477 @list
( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
478 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
479 if( is_callable( $processCallback ) ) {
480 call_user_func_array( $processCallback, array( &$params, $args ) );
482 $newparams = Sanitizer
::fixTagAttributes( $params, $t );
483 $rest = str_replace( '>', '>', $rest );
484 $text .= "<$slash$t$newparams$brace$rest";
486 $text .= '<' . str_replace( '>', '>', $x);
490 wfProfileOut( $fname );
495 * Remove '<!--', '-->', and everything between.
496 * To avoid leaving blank lines, when a comment is both preceded
497 * and followed by a newline (ignoring spaces), trim leading and
498 * trailing spaces and one of the newlines.
501 * @param string $text
504 function removeHTMLcomments( $text ) {
505 $fname='Parser::removeHTMLcomments';
506 wfProfileIn( $fname );
507 while (($start = strpos($text, '<!--')) !== false) {
508 $end = strpos($text, '-->', $start +
4);
509 if ($end === false) {
510 # Unterminated comment; bail out
516 # Trim space and newline if the comment is both
517 # preceded and followed by a newline
518 $spaceStart = max($start - 1, 0);
519 $spaceLen = $end - $spaceStart;
520 while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
524 while (substr($text, $spaceStart +
$spaceLen, 1) === ' ')
526 if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart +
$spaceLen, 1) === "\n") {
527 # Remove the comment, leading and trailing
528 # spaces, and leave only one newline.
529 $text = substr_replace($text, "\n", $spaceStart, $spaceLen +
1);
532 # Remove just the comment.
533 $text = substr_replace($text, '', $start, $end - $start);
536 wfProfileOut( $fname );
541 * Take an array of attribute names and values and normalize or discard
542 * illegal values for the given element type.
544 * - Discards attributes not on a whitelist for the given element
545 * - Unsafe style attributes are discarded
547 * @param array $attribs
548 * @param string $element
551 * @todo Check for legal values where the DTD limits things.
552 * @todo Check for unique id attribute :P
554 function validateTagAttributes( $attribs, $element ) {
555 $whitelist = array_flip( Sanitizer
::attributeWhitelist( $element ) );
557 foreach( $attribs as $attribute => $value ) {
558 if( !isset( $whitelist[$attribute] ) ) {
561 # Strip javascript "expression" from stylesheets.
562 # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
563 if( $attribute == 'style' ) {
564 $stripped = Sanitizer
::decodeCharReferences( $value );
566 // Remove any comments; IE gets token splitting wrong
567 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
570 // ... and continue checks
571 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
572 'codepointToUtf8(hexdec("$1"))', $stripped );
573 $stripped = str_replace( '\\', '', $stripped );
574 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
581 if ( $attribute === 'id' )
582 $value = Sanitizer
::escapeId( $value );
584 // If this attribute was previously set, override it.
585 // Output should only have one attribute of each name.
586 $out[$attribute] = $value;
592 * Take a tag soup fragment listing an HTML element's attributes
593 * and normalize it to well-formed XML, discarding unwanted attributes.
594 * Output is safe for further wikitext processing, with escaping of
595 * values that could trigger problems.
597 * - Normalizes attribute names to lowercase
598 * - Discards attributes not on a whitelist for the given element
599 * - Turns broken or invalid entities into plaintext
600 * - Double-quotes all attribute values
601 * - Attributes without values are given the name as attribute
602 * - Double attributes are discarded
603 * - Unsafe style attributes are discarded
604 * - Prepends space if there are attributes.
606 * @param string $text
607 * @param string $element
610 function fixTagAttributes( $text, $element ) {
611 if( trim( $text ) == '' ) {
615 $stripped = Sanitizer
::validateTagAttributes(
616 Sanitizer
::decodeTagAttributes( $text ), $element );
619 foreach( $stripped as $attribute => $value ) {
620 $encAttribute = htmlspecialchars( $attribute );
622 $encValue = htmlspecialchars( $value );
623 # Templates and links may be expanded in later parsing,
624 # creating invalid or dangerous output. Suppress this.
625 $encValue = strtr( $encValue, array(
626 '<' => '<', // This should never happen,
627 '>' => '>', // we've received invalid input
628 '"' => '"', // which should have been escaped.
631 "''" => '''',
632 'ISBN' => 'ISBN',
634 'PMID' => 'PMID',
640 $encValue = preg_replace_callback(
641 '/(' . wfUrlProtocols() . ')/',
642 array( 'Sanitizer', 'armorLinksCallback' ),
645 $attribs[] = "$encAttribute=\"$encValue\"";
647 return count( $attribs ) ?
' ' . implode( ' ', $attribs ) : '';
651 * Given a value escape it so that it can be used in an id attribute and
652 * return it, this does not validate the value however (see first link)
654 * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
657 * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
666 function escapeId( $id ) {
667 static $replace = array(
672 $id = urlencode( Sanitizer
::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
674 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
678 * Regex replace callback for armoring links against further processing.
679 * @param array $matches
683 function armorLinksCallback( $matches ) {
684 return str_replace( ':', ':', $matches[1] );
688 * Return an associative array of attribute names and values from
689 * a partial tag string. Attribute names are forces to lowercase,
690 * character references are decoded to UTF-8 text.
695 function decodeTagAttributes( $text ) {
698 if( trim( $text ) == '' ) {
711 foreach( $pairs as $set ) {
712 $attribute = strtolower( $set[1] );
713 $value = Sanitizer
::getTagAttributeCallback( $set );
714 $attribs[$attribute] = Sanitizer
::decodeCharReferences( $value );
720 * Pick the appropriate attribute value from a match set from the
721 * MW_ATTRIBS_REGEX matches.
727 function getTagAttributeCallback( $set ) {
728 if( isset( $set[6] ) ) {
729 # Illegal #XXXXXX color with no quotes.
731 } elseif( isset( $set[5] ) ) {
734 } elseif( isset( $set[4] ) ) {
737 } elseif( isset( $set[3] ) ) {
740 } elseif( !isset( $set[2] ) ) {
741 # In XHTML, attributes must have a value.
742 # For 'reduced' form, return explicitly the attribute name here.
745 throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
750 * Normalize whitespace and character references in an XML source-
751 * encoded text for an attribute value.
753 * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
754 * but note that we're not returning the value, but are returning
755 * XML source fragments that will be slapped into output.
757 * @param string $text
761 function normalizeAttributeValue( $text ) {
762 return str_replace( '"', '"',
764 '/\r\n|[\x20\x0d\x0a\x09]/',
766 Sanitizer
::normalizeCharReferences( $text ) ) );
770 * Ensure that any entities and character references are legal
771 * for XML and XHTML specifically. Any stray bits will be
772 * &-escaped to result in a valid text fragment.
774 * a. any named char refs must be known in XHTML
775 * b. any numeric char refs must be legal chars, not invalid or forbidden
776 * c. use &#x, not &#X
777 * d. fix or reject non-valid attributes
779 * @param string $text
783 function normalizeCharReferences( $text ) {
784 return preg_replace_callback(
786 array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
790 * @param string $matches
793 function normalizeCharReferencesCallback( $matches ) {
795 if( $matches[1] != '' ) {
796 $ret = Sanitizer
::normalizeEntity( $matches[1] );
797 } elseif( $matches[2] != '' ) {
798 $ret = Sanitizer
::decCharReference( $matches[2] );
799 } elseif( $matches[3] != '' ) {
800 $ret = Sanitizer
::hexCharReference( $matches[3] );
801 } elseif( $matches[4] != '' ) {
802 $ret = Sanitizer
::hexCharReference( $matches[4] );
804 if( is_null( $ret ) ) {
805 return htmlspecialchars( $matches[0] );
812 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
813 * return the named entity reference as is. Otherwise, returns
814 * HTML-escaped text of pseudo-entity source (eg &foo;)
816 * @param string $name
819 function normalizeEntity( $name ) {
820 global $wgHtmlEntities;
821 if( isset( $wgHtmlEntities[$name] ) ) {
824 return "&$name;";
828 function decCharReference( $codepoint ) {
829 $point = intval( $codepoint );
830 if( Sanitizer
::validateCodepoint( $point ) ) {
831 return sprintf( '&#%d;', $point );
837 function hexCharReference( $codepoint ) {
838 $point = hexdec( $codepoint );
839 if( Sanitizer
::validateCodepoint( $point ) ) {
840 return sprintf( '&#x%x;', $point );
847 * Returns true if a given Unicode codepoint is a valid character in XML.
848 * @param int $codepoint
851 function validateCodepoint( $codepoint ) {
852 return ($codepoint == 0x09)
853 ||
($codepoint == 0x0a)
854 ||
($codepoint == 0x0d)
855 ||
($codepoint >= 0x20 && $codepoint <= 0xd7ff)
856 ||
($codepoint >= 0xe000 && $codepoint <= 0xfffd)
857 ||
($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
861 * Decode any character references, numeric or named entities,
862 * in the text and return a UTF-8 string.
864 * @param string $text
868 function decodeCharReferences( $text ) {
869 return preg_replace_callback(
871 array( 'Sanitizer', 'decodeCharReferencesCallback' ),
876 * @param string $matches
879 function decodeCharReferencesCallback( $matches ) {
880 if( $matches[1] != '' ) {
881 return Sanitizer
::decodeEntity( $matches[1] );
882 } elseif( $matches[2] != '' ) {
883 return Sanitizer
::decodeChar( intval( $matches[2] ) );
884 } elseif( $matches[3] != '' ) {
885 return Sanitizer
::decodeChar( hexdec( $matches[3] ) );
886 } elseif( $matches[4] != '' ) {
887 return Sanitizer
::decodeChar( hexdec( $matches[4] ) );
889 # Last case should be an ampersand by itself
894 * Return UTF-8 string for a codepoint if that is a valid
895 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
896 * @param int $codepoint
900 function decodeChar( $codepoint ) {
901 if( Sanitizer
::validateCodepoint( $codepoint ) ) {
902 return codepointToUtf8( $codepoint );
904 return UTF8_REPLACEMENT
;
909 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
910 * return the UTF-8 encoding of that character. Otherwise, returns
911 * pseudo-entity source (eg &foo;)
913 * @param string $name
916 function decodeEntity( $name ) {
917 global $wgHtmlEntities;
918 if( isset( $wgHtmlEntities[$name] ) ) {
919 return codepointToUtf8( $wgHtmlEntities[$name] );
926 * Fetch the whitelist of acceptable attributes for a given
929 * @param string $element
932 function attributeWhitelist( $element ) {
934 if( !isset( $list ) ) {
935 $list = Sanitizer
::setupAttributeWhitelist();
937 return isset( $list[$element] )
945 function setupAttributeWhitelist() {
946 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
947 $block = array_merge( $common, array( 'align' ) );
948 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
949 $tablecell = array( 'abbr',
955 'nowrap', # deprecated
956 'width', # deprecated
957 'height', # deprecated
958 'bgcolor' # deprecated
961 # Numbers refer to sections in HTML 4.01 standard describing the element.
962 # See: http://www.w3.org/TR/html4/
966 'center' => $common, # deprecated
967 'span' => $block, # ??
996 'blockquote' => array_merge( $common, array( 'cite' ) ),
1007 'br' => array( 'id', 'class', 'title', 'style', 'clear' ),
1010 'pre' => array_merge( $common, array( 'width' ) ),
1013 'ins' => array_merge( $common, array( 'cite', 'datetime' ) ),
1014 'del' => array_merge( $common, array( 'cite', 'datetime' ) ),
1017 'ul' => array_merge( $common, array( 'type' ) ),
1018 'ol' => array_merge( $common, array( 'type', 'start' ) ),
1019 'li' => array_merge( $common, array( 'type', 'value' ) ),
1027 'table' => array_merge( $common,
1028 array( 'summary', 'width', 'border', 'frame',
1029 'rules', 'cellspacing', 'cellpadding',
1030 'align', 'bgcolor', 'frame', 'rules',
1034 'caption' => array_merge( $common, array( 'align' ) ),
1037 'thead' => array_merge( $common, $tablealign ),
1038 'tfoot' => array_merge( $common, $tablealign ),
1039 'tbody' => array_merge( $common, $tablealign ),
1042 'colgroup' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1043 'col' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1046 'tr' => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1049 'td' => array_merge( $common, $tablecell, $tablealign ),
1050 'th' => array_merge( $common, $tablecell, $tablealign ),
1058 'strike' => $common,
1063 'font' => array_merge( $common, array( 'size', 'color', 'face' ) ),
1067 'hr' => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1069 # XHTML Ruby annotation text module, simple ruby only.
1070 # http://www.w3c.org/TR/ruby/
1075 'rt' => $common, #array_merge( $common, array( 'rbspan' ) ),
1082 * Take a fragment of (potentially invalid) HTML and return
1083 * a version with any tags removed, encoded suitably for literal
1084 * inclusion in an attribute value.
1086 * @param string $text HTML fragment
1089 function stripAllTags( $text ) {
1091 $text = preg_replace( '/ < .*? > /x', '', $text );
1093 # Normalize &entities and whitespace
1094 $text = Sanitizer
::normalizeAttributeValue( $text );
1096 # Will be placed into "double-quoted" attributes,
1097 # make sure remaining bits are safe.
1098 $text = str_replace(
1099 array('<', '>', '"'),
1100 array('<', '>', '"'),
1107 * Hack up a private DOCTYPE with HTML's standard entity declarations.
1108 * PHP 4 seemed to know these if you gave it an HTML doctype, but
1111 * Use for passing XHTML fragments to PHP's XML parsing functions
1116 function hackDocType() {
1117 global $wgHtmlEntities;
1118 $out = "<!DOCTYPE html [\n";
1119 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1120 $out .= "<!ENTITY $entity \"&#$codepoint;\">";