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 a tag soup fragment listing an HTML element's attributes
542 * and normalize it to well-formed XML, discarding unwanted attributes.
544 * - Normalizes attribute names to lowercase
545 * - Discards attributes not on a whitelist for the given element
546 * - Turns broken or invalid entities into plaintext
547 * - Double-quotes all attribute values
548 * - Attributes without values are given the name as attribute
549 * - Double attributes are discarded
550 * - Unsafe style attributes are discarded
551 * - Prepends space if there are attributes.
553 * @param string $text
554 * @param string $element
557 * @todo Check for legal values where the DTD limits things.
558 * @todo Check for unique id attribute :P
560 function fixTagAttributes( $text, $element ) {
561 if( trim( $text ) == '' ) {
566 # Since we quote this later, this can be anything distinguishable
567 # from the end of the attribute
577 $whitelist = array_flip( Sanitizer
::attributeWhitelist( $element ) );
579 foreach( $pairs as $set ) {
580 $attribute = strtolower( $set[1] );
581 if( !isset( $whitelist[$attribute] ) ) {
585 $raw = Sanitizer
::getTagAttributeCallback( $set );
586 $value = Sanitizer
::normalizeAttributeValue( $raw );
588 # Strip javascript "expression" from stylesheets.
589 # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
590 if( $attribute == 'style' ) {
591 $stripped = Sanitizer
::decodeCharReferences( $value );
593 // Remove any comments; IE gets token splitting wrong
594 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
595 $value = htmlspecialchars( $stripped );
597 // ... and continue checks
598 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
599 'codepointToUtf8(hexdec("$1"))', $stripped );
600 $stripped = str_replace( '\\', '', $stripped );
601 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
608 if ( $attribute === 'id' )
609 $value = Sanitizer
::escapeId( $value );
611 # Templates and links may be expanded in later parsing,
612 # creating invalid or dangerous output. Suppress this.
613 $value = strtr( $value, array(
614 '<' => '<', // This should never happen,
615 '>' => '>', // we've received invalid input
616 '"' => '"', // which should have been escaped.
619 "''" => '''',
620 'ISBN' => 'ISBN',
622 'PMID' => 'PMID',
628 $value = preg_replace_callback(
629 '/(' . wfUrlProtocols() . ')/',
630 array( 'Sanitizer', 'armorLinksCallback' ),
633 // If this attribute was previously set, override it.
634 // Output should only have one attribute of each name.
635 $attribs[$attribute] = "$attribute=\"$value\"";
638 return count( $attribs ) ?
' ' . implode( ' ', $attribs ) : '';
642 * Given a value escape it so that it can be used in an id attribute and
643 * return it, this does not validate the value however (see first link)
645 * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
648 * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
657 function escapeId( $id ) {
658 static $replace = array(
663 $id = urlencode( Sanitizer
::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
665 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
669 * Regex replace callback for armoring links against further processing.
670 * @param array $matches
674 function armorLinksCallback( $matches ) {
675 return str_replace( ':', ':', $matches[1] );
679 * Return an associative array of attribute names and values from
680 * a partial tag string. Attribute names are forces to lowercase,
681 * character references are decoded to UTF-8 text.
686 function decodeTagAttributes( $text ) {
689 if( trim( $text ) == '' ) {
702 foreach( $pairs as $set ) {
703 $attribute = strtolower( $set[1] );
704 $value = Sanitizer
::getTagAttributeCallback( $set );
705 $attribs[$attribute] = Sanitizer
::decodeCharReferences( $value );
711 * Pick the appropriate attribute value from a match set from the
712 * MW_ATTRIBS_REGEX matches.
718 function getTagAttributeCallback( $set ) {
719 if( isset( $set[6] ) ) {
720 # Illegal #XXXXXX color with no quotes.
722 } elseif( isset( $set[5] ) ) {
725 } elseif( isset( $set[4] ) ) {
728 } elseif( isset( $set[3] ) ) {
731 } elseif( !isset( $set[2] ) ) {
732 # In XHTML, attributes must have a value.
733 # For 'reduced' form, return explicitly the attribute name here.
736 wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
741 * Normalize whitespace and character references in an XML source-
742 * encoded text for an attribute value.
744 * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
745 * but note that we're not returning the value, but are returning
746 * XML source fragments that will be slapped into output.
748 * @param string $text
752 function normalizeAttributeValue( $text ) {
753 return str_replace( '"', '"',
755 '/\r\n|[\x20\x0d\x0a\x09]/',
757 Sanitizer
::normalizeCharReferences( $text ) ) );
761 * Ensure that any entities and character references are legal
762 * for XML and XHTML specifically. Any stray bits will be
763 * &-escaped to result in a valid text fragment.
765 * a. any named char refs must be known in XHTML
766 * b. any numeric char refs must be legal chars, not invalid or forbidden
767 * c. use &#x, not &#X
768 * d. fix or reject non-valid attributes
770 * @param string $text
774 function normalizeCharReferences( $text ) {
775 return preg_replace_callback(
777 array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
781 * @param string $matches
784 function normalizeCharReferencesCallback( $matches ) {
786 if( $matches[1] != '' ) {
787 $ret = Sanitizer
::normalizeEntity( $matches[1] );
788 } elseif( $matches[2] != '' ) {
789 $ret = Sanitizer
::decCharReference( $matches[2] );
790 } elseif( $matches[3] != '' ) {
791 $ret = Sanitizer
::hexCharReference( $matches[3] );
792 } elseif( $matches[4] != '' ) {
793 $ret = Sanitizer
::hexCharReference( $matches[4] );
795 if( is_null( $ret ) ) {
796 return htmlspecialchars( $matches[0] );
803 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
804 * return the named entity reference as is. Otherwise, returns
805 * HTML-escaped text of pseudo-entity source (eg &foo;)
807 * @param string $name
810 function normalizeEntity( $name ) {
811 global $wgHtmlEntities;
812 if( isset( $wgHtmlEntities[$name] ) ) {
815 return "&$name;";
819 function decCharReference( $codepoint ) {
820 $point = intval( $codepoint );
821 if( Sanitizer
::validateCodepoint( $point ) ) {
822 return sprintf( '&#%d;', $point );
828 function hexCharReference( $codepoint ) {
829 $point = hexdec( $codepoint );
830 if( Sanitizer
::validateCodepoint( $point ) ) {
831 return sprintf( '&#x%x;', $point );
838 * Returns true if a given Unicode codepoint is a valid character in XML.
839 * @param int $codepoint
842 function validateCodepoint( $codepoint ) {
843 return ($codepoint == 0x09)
844 ||
($codepoint == 0x0a)
845 ||
($codepoint == 0x0d)
846 ||
($codepoint >= 0x20 && $codepoint <= 0xd7ff)
847 ||
($codepoint >= 0xe000 && $codepoint <= 0xfffd)
848 ||
($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
852 * Decode any character references, numeric or named entities,
853 * in the text and return a UTF-8 string.
855 * @param string $text
859 function decodeCharReferences( $text ) {
860 return preg_replace_callback(
862 array( 'Sanitizer', 'decodeCharReferencesCallback' ),
867 * @param string $matches
870 function decodeCharReferencesCallback( $matches ) {
871 if( $matches[1] != '' ) {
872 return Sanitizer
::decodeEntity( $matches[1] );
873 } elseif( $matches[2] != '' ) {
874 return Sanitizer
::decodeChar( intval( $matches[2] ) );
875 } elseif( $matches[3] != '' ) {
876 return Sanitizer
::decodeChar( hexdec( $matches[3] ) );
877 } elseif( $matches[4] != '' ) {
878 return Sanitizer
::decodeChar( hexdec( $matches[4] ) );
880 # Last case should be an ampersand by itself
885 * Return UTF-8 string for a codepoint if that is a valid
886 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
887 * @param int $codepoint
891 function decodeChar( $codepoint ) {
892 if( Sanitizer
::validateCodepoint( $codepoint ) ) {
893 return codepointToUtf8( $codepoint );
895 return UTF8_REPLACEMENT
;
900 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
901 * return the UTF-8 encoding of that character. Otherwise, returns
902 * pseudo-entity source (eg &foo;)
904 * @param string $name
907 function decodeEntity( $name ) {
908 global $wgHtmlEntities;
909 if( isset( $wgHtmlEntities[$name] ) ) {
910 return codepointToUtf8( $wgHtmlEntities[$name] );
917 * Fetch the whitelist of acceptable attributes for a given
920 * @param string $element
923 function attributeWhitelist( $element ) {
925 if( !isset( $list ) ) {
926 $list = Sanitizer
::setupAttributeWhitelist();
928 return isset( $list[$element] )
936 function setupAttributeWhitelist() {
937 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
938 $block = array_merge( $common, array( 'align' ) );
939 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
940 $tablecell = array( 'abbr',
946 'nowrap', # deprecated
947 'width', # deprecated
948 'height', # deprecated
949 'bgcolor' # deprecated
952 # Numbers refer to sections in HTML 4.01 standard describing the element.
953 # See: http://www.w3.org/TR/html4/
957 'center' => $common, # deprecated
958 'span' => $block, # ??
987 'blockquote' => array_merge( $common, array( 'cite' ) ),
998 'br' => array( 'id', 'class', 'title', 'style', 'clear' ),
1001 'pre' => array_merge( $common, array( 'width' ) ),
1004 'ins' => array_merge( $common, array( 'cite', 'datetime' ) ),
1005 'del' => array_merge( $common, array( 'cite', 'datetime' ) ),
1008 'ul' => array_merge( $common, array( 'type' ) ),
1009 'ol' => array_merge( $common, array( 'type', 'start' ) ),
1010 'li' => array_merge( $common, array( 'type', 'value' ) ),
1018 'table' => array_merge( $common,
1019 array( 'summary', 'width', 'border', 'frame',
1020 'rules', 'cellspacing', 'cellpadding',
1021 'align', 'bgcolor', 'frame', 'rules',
1025 'caption' => array_merge( $common, array( 'align' ) ),
1028 'thead' => array_merge( $common, $tablealign ),
1029 'tfoot' => array_merge( $common, $tablealign ),
1030 'tbody' => array_merge( $common, $tablealign ),
1033 'colgroup' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1034 'col' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1037 'tr' => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1040 'td' => array_merge( $common, $tablecell, $tablealign ),
1041 'th' => array_merge( $common, $tablecell, $tablealign ),
1049 'strike' => $common,
1054 'font' => array_merge( $common, array( 'size', 'color', 'face' ) ),
1058 'hr' => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1060 # XHTML Ruby annotation text module, simple ruby only.
1061 # http://www.w3c.org/TR/ruby/
1066 'rt' => $common, #array_merge( $common, array( 'rbspan' ) ),
1073 * Take a fragment of (potentially invalid) HTML and return
1074 * a version with any tags removed, encoded suitably for literal
1075 * inclusion in an attribute value.
1077 * @param string $text HTML fragment
1080 function stripAllTags( $text ) {
1082 $text = preg_replace( '/ < .*? > /x', '', $text );
1084 # Normalize &entities and whitespace
1085 $text = Sanitizer
::normalizeAttributeValue( $text );
1087 # Will be placed into "double-quoted" attributes,
1088 # make sure remaining bits are safe.
1089 $text = str_replace(
1090 array('<', '>', '"'),
1091 array('<', '>', '"'),
1098 * Hack up a private DOCTYPE with HTML's standard entity declarations.
1099 * PHP 4 seemed to know these if you gave it an HTML doctype, but
1102 * Use for passing XHTML fragments to PHP's XML parsing functions
1107 function hackDocType() {
1108 global $wgHtmlEntities;
1109 $out = "<!DOCTYPE html [\n";
1110 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1111 $out .= "<!ENTITY $entity \"&#$codepoint;\">";