3 * (X)HTML 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 * 59 Temple Place - Suite 330, Boston, MA 02111-1307, 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'
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
357 $htmlpairs = array();
358 $htmlsingle = array();
360 $tabletags = array();
363 $htmlsingle = array_merge( $tabletags, $htmlsingle );
364 $htmlelements = array_merge( $htmlsingle, $htmlpairs );
366 # Remove HTML comments
367 $text = Sanitizer::removeHTMLcomments( $text );
369 $bits = explode( '<', $text );
370 $text = array_shift( $bits );
372 $tagstack = array(); $tablestack = array();
373 foreach ( $bits as $x ) {
374 $prev = error_reporting( E_ALL & ~( E_NOTICE | E_WARNING ) );
375 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
377 list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
378 error_reporting( $prev );
381 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
385 if( in_array( $t, $htmlsingleonly ) ) {
387 } elseif( !in_array( $t, $htmlsingle ) &&
388 ( $ot = @array_pop( $tagstack ) ) != $t ) {
389 @array_push( $tagstack, $ot );
392 if ( $t == 'table' ) {
393 $tagstack = array_pop( $tablestack );
398 # Keep track for later
399 if ( in_array( $t, $tabletags ) &&
400 ! in_array( 'table', $tagstack ) ) {
402 } else if ( in_array( $t, $tagstack ) &&
403 ! in_array ( $t , $htmlnest ) ) {
405 } elseif( in_array( $t, $htmlsingleonly ) ) {
406 # Hack to force empty tag for uncloseable elements
408 } else if ( ! in_array( $t, $htmlsingle ) ) {
409 if ( $t == 'table' ) {
410 array_push( $tablestack, $tagstack );
413 array_push( $tagstack, $t );
416 # Replace any variables or template parameters with
418 if( is_callable( $processCallback ) ) {
419 call_user_func_array( $processCallback, array( &$params, $args ) );
422 # Strip non-approved attributes from the tag
423 $newparams = Sanitizer::fixTagAttributes( $params, $t );
426 $rest = str_replace( '>', '>', $rest );
427 $close = ( $brace == '/>' ) ? ' /' : '';
428 $text .= "<$slash$t$newparams$close>$rest";
432 $text .= '<' . str_replace( '>', '>', $x);
434 # Close off any remaining tags
435 while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
437 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
440 # this might be possible using tidy itself
441 foreach ( $bits as $x ) {
442 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
444 @list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
445 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
446 if( is_callable( $processCallback ) ) {
447 call_user_func_array( $processCallback, array( &$params, $args ) );
449 $newparams = Sanitizer::fixTagAttributes( $params, $t );
450 $rest = str_replace( '>', '>', $rest );
451 $text .= "<$slash$t$newparams$brace$rest";
453 $text .= '<' . str_replace( '>', '>', $x);
457 wfProfileOut( $fname );
462 * Remove '<!--', '-->', and everything between.
463 * To avoid leaving blank lines, when a comment is both preceded
464 * and followed by a newline (ignoring spaces), trim leading and
465 * trailing spaces and one of the newlines.
468 * @param string $text
471 function removeHTMLcomments( $text ) {
472 $fname='Parser::removeHTMLcomments';
473 wfProfileIn( $fname );
474 while (($start = strpos($text, '<!--')) !== false) {
475 $end = strpos($text, '-->', $start + 4);
476 if ($end === false) {
477 # Unterminated comment; bail out
483 # Trim space and newline if the comment is both
484 # preceded and followed by a newline
485 $spaceStart = max($start - 1, 0);
486 $spaceLen = $end - $spaceStart;
487 while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
491 while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
493 if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
494 # Remove the comment, leading and trailing
495 # spaces, and leave only one newline.
496 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
499 # Remove just the comment.
500 $text = substr_replace($text, '', $start, $end - $start);
503 wfProfileOut( $fname );
508 * Take a tag soup fragment listing an HTML element's attributes
509 * and normalize it to well-formed XML, discarding unwanted attributes.
511 * - Normalizes attribute names to lowercase
512 * - Discards attributes not on a whitelist for the given element
513 * - Turns broken or invalid entities into plaintext
514 * - Double-quotes all attribute values
515 * - Attributes without values are given the name as attribute
516 * - Double attributes are discarded
517 * - Unsafe style attributes are discarded
518 * - Prepends space if there are attributes.
520 * @param string $text
521 * @param string $element
524 * @todo Check for legal values where the DTD limits things.
525 * @todo Check for unique id attribute :P
527 function fixTagAttributes( $text, $element ) {
528 global $wgUrlProtocols;
529 if( trim( $text ) == '' ) {
534 # Since we quote this later, this can be anything distinguishable
535 # from the end of the attribute
544 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
546 foreach( $pairs as $set ) {
547 $attribute = strtolower( $set[1] );
548 if( !isset( $whitelist[$attribute] ) ) {
552 $raw = Sanitizer::getTagAttributeCallback( $set );
553 $value = Sanitizer::normalizeAttributeValue( $raw );
555 # Strip javascript "expression" from stylesheets.
556 # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
557 if( $attribute == 'style' ) {
558 $stripped = Sanitizer::decodeCharReferences( $value );
560 // Remove any comments; IE gets token splitting wrong
561 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
562 $value = htmlspecialchars( $stripped );
564 // ... and continue checks
565 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
566 'codepointToUtf8(hexdec("$1"))', $stripped );
567 $stripped = str_replace( '\\', '', $stripped );
568 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
575 # Templates and links may be expanded in later parsing,
576 # creating invalid or dangerous output. Suppress this.
577 $value = strtr( $value, array(
580 "''" => '''',
581 'ISBN' => 'ISBN',
583 'PMID' => 'PMID',
587 $value = preg_replace_callback(
588 '/(' . $wgUrlProtocols . ')/',
589 array( 'Sanitizer', 'armorLinksCallback' ),
592 // If this attribute was previously set, override it.
593 // Output should only have one attribute of each name.
594 $attribs[$attribute] = "$attribute=\"$value\"";
596 if( empty( $attribs ) ) {
599 return ' ' . implode( ' ', $attribs );
604 * Regex replace callback for armoring links against further processing.
605 * @param array $matches
609 function armorLinksCallback( $matches ) {
610 return str_replace( ':', ':', $matches[1] );
614 * Return an associative array of attribute names and values from
615 * a partial tag string. Attribute names are forces to lowercase,
616 * character references are decoded to UTF-8 text.
621 function decodeTagAttributes( $text ) {
624 if( trim( $text ) == '' ) {
636 foreach( $pairs as $set ) {
637 $attribute = strtolower( $set[1] );
638 $value = Sanitizer::getTagAttributeCallback( $set );
639 $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
645 * Pick the appropriate attribute value from a match set from the
646 * MW_ATTRIBS_REGEX matches.
652 function getTagAttributeCallback( $set ) {
653 if( isset( $set[6] ) ) {
654 # Illegal #XXXXXX color with no quotes.
656 } elseif( isset( $set[5] ) ) {
659 } elseif( isset( $set[4] ) ) {
662 } elseif( isset( $set[3] ) ) {
665 } elseif( !isset( $set[2] ) ) {
666 # In XHTML, attributes must have a value.
667 # For 'reduced' form, return explicitly the attribute name here.
670 wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
675 * Normalize whitespace and character references in an XML source-
676 * encoded text for an attribute value.
678 * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
679 * but note that we're not returning the value, but are returning
680 * XML source fragments that will be slapped into output.
682 * @param string $text
686 function normalizeAttributeValue( $text ) {
687 return str_replace( '"', '"',
689 '/\r\n|[\x20\x0d\x0a\x09]/',
691 Sanitizer::normalizeCharReferences( $text ) ) );
695 * Ensure that any entities and character references are legal
696 * for XML and XHTML specifically. Any stray bits will be
697 * &-escaped to result in a valid text fragment.
699 * a. any named char refs must be known in XHTML
700 * b. any numeric char refs must be legal chars, not invalid or forbidden
701 * c. use &#x, not &#X
702 * d. fix or reject non-valid attributes
704 * @param string $text
708 function normalizeCharReferences( $text ) {
709 return preg_replace_callback(
711 array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
715 * @param string $matches
718 function normalizeCharReferencesCallback( $matches ) {
720 if( $matches[1] != '' ) {
721 $ret = Sanitizer::normalizeEntity( $matches[1] );
722 } elseif( $matches[2] != '' ) {
723 $ret = Sanitizer::decCharReference( $matches[2] );
724 } elseif( $matches[3] != '' ) {
725 $ret = Sanitizer::hexCharReference( $matches[3] );
726 } elseif( $matches[4] != '' ) {
727 $ret = Sanitizer::hexCharReference( $matches[4] );
729 if( is_null( $ret ) ) {
730 return htmlspecialchars( $matches[0] );
737 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
738 * return the named entity reference as is. Otherwise, returns
739 * HTML-escaped text of pseudo-entity source (eg &foo;)
741 * @param string $name
744 function normalizeEntity( $name ) {
745 global $wgHtmlEntities;
746 if( isset( $wgHtmlEntities[$name] ) ) {
749 return "&$name;";
753 function decCharReference( $codepoint ) {
754 $point = IntVal( $codepoint );
755 if( Sanitizer::validateCodepoint( $point ) ) {
756 return sprintf( '&#%d;', $point );
762 function hexCharReference( $codepoint ) {
763 $point = hexdec( $codepoint );
764 if( Sanitizer::validateCodepoint( $point ) ) {
765 return sprintf( '&#x%x;', $point );
772 * Returns true if a given Unicode codepoint is a valid character in XML.
773 * @param int $codepoint
776 function validateCodepoint( $codepoint ) {
777 return ($codepoint == 0x09)
778 || ($codepoint == 0x0a)
779 || ($codepoint == 0x0d)
780 || ($codepoint >= 0x20 && $codepoint <= 0xd7ff)
781 || ($codepoint >= 0xe000 && $codepoint <= 0xfffd)
782 || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
786 * Decode any character references, numeric or named entities,
787 * in the text and return a UTF-8 string.
789 * @param string $text
793 function decodeCharReferences( $text ) {
794 return preg_replace_callback(
796 array( 'Sanitizer', 'decodeCharReferencesCallback' ),
801 * @param string $matches
804 function decodeCharReferencesCallback( $matches ) {
805 if( $matches[1] != '' ) {
806 return Sanitizer::decodeEntity( $matches[1] );
807 } elseif( $matches[2] != '' ) {
808 return Sanitizer::decodeChar( intval( $matches[2] ) );
809 } elseif( $matches[3] != '' ) {
810 return Sanitizer::decodeChar( hexdec( $matches[3] ) );
811 } elseif( $matches[4] != '' ) {
812 return Sanitizer::decodeChar( hexdec( $matches[4] ) );
814 # Last case should be an ampersand by itself
819 * Return UTF-8 string for a codepoint if that is a valid
820 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
821 * @param int $codepoint
825 function decodeChar( $codepoint ) {
826 if( Sanitizer::validateCodepoint( $codepoint ) ) {
827 return codepointToUtf8( $codepoint );
829 return UTF8_REPLACEMENT;
834 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
835 * return the UTF-8 encoding of that character. Otherwise, returns
836 * pseudo-entity source (eg &foo;)
838 * @param string $name
841 function decodeEntity( $name ) {
842 global $wgHtmlEntities;
843 if( isset( $wgHtmlEntities[$name] ) ) {
844 return codepointToUtf8( $wgHtmlEntities[$name] );
851 * Fetch the whitelist of acceptable attributes for a given
854 * @param string $element
857 function attributeWhitelist( $element ) {
859 if( !isset( $list ) ) {
860 $list = Sanitizer::setupAttributeWhitelist();
862 return isset( $list[$element] )
870 function setupAttributeWhitelist() {
871 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
872 $block = array_merge( $common, array( 'align' ) );
873 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
874 $tablecell = array( 'abbr',
880 'nowrap', # deprecated
881 'width', # deprecated
882 'height', # deprecated
883 'bgcolor' # deprecated
886 # Numbers refer to sections in HTML 4.01 standard describing the element.
887 # See: http://www.w3.org/TR/html4/
891 'center' => $common, # deprecated
892 'span' => $block, # ??
921 'blockquote' => array_merge( $common, array( 'cite' ) ),
932 'br' => array( 'id', 'class', 'title', 'style', 'clear' ),
935 'pre' => array_merge( $common, array( 'width' ) ),
938 'ins' => array_merge( $common, array( 'cite', 'datetime' ) ),
939 'del' => array_merge( $common, array( 'cite', 'datetime' ) ),
942 'ul' => array_merge( $common, array( 'type' ) ),
943 'ol' => array_merge( $common, array( 'type', 'start' ) ),
944 'li' => array_merge( $common, array( 'type', 'value' ) ),
952 'table' => array_merge( $common,
953 array( 'summary', 'width', 'border', 'frame',
954 'rules', 'cellspacing', 'cellpadding',
955 'align', 'bgcolor', 'frame', 'rules',
959 'caption' => array_merge( $common, array( 'align' ) ),
962 'thead' => array_merge( $common, $tablealign ),
963 'tfoot' => array_merge( $common, $tablealign ),
964 'tbody' => array_merge( $common, $tablealign ),
967 'colgroup' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
968 'col' => array_merge( $common, array( 'span', 'width' ), $tablealign ),
971 'tr' => array_merge( $common, array( 'bgcolor' ), $tablealign ),
974 'td' => array_merge( $common, $tablecell, $tablealign ),
975 'th' => array_merge( $common, $tablecell, $tablealign ),
988 'font' => array_merge( $common, array( 'size', 'color', 'face' ) ),
992 'hr' => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
994 # XHTML Ruby annotation text module, simple ruby only.
995 # http://www.w3c.org/TR/ruby/
1000 'rt' => $common, #array_merge( $common, array( 'rbspan' ) ),
1007 * Take a fragment of (potentially invalid) HTML and return
1008 * a version with any tags removed, encoded suitably for literal
1009 * inclusion in an attribute value.
1011 * @param string $text HTML fragment
1014 function stripAllTags( $text ) {
1016 $text = preg_replace( '/<[^>]*>/', '', $text );
1018 # Normalize &entities and whitespace
1019 $text = Sanitizer::normalizeAttributeValue( $text );
1021 # Will be placed into "double-quoted" attributes,
1022 # make sure remaining bits are safe.
1023 $text = str_replace(
1024 array('<', '>', '"'),
1025 array('<', '>', '"'),