3 * HTML sanitizer for %MediaWiki.
5 * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
6 * https://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 * HTML sanitizer for MediaWiki
33 * Regular expression to match various types of character references in
34 * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
36 const CHAR_REFS_REGEX
=
37 '/&([A-Za-z0-9\x80-\xff]+);
39 |&\#[xX]([0-9A-Fa-f]+);
43 * Acceptable tag name charset from HTML5 parsing spec
44 * https://www.w3.org/TR/html5/syntax.html#tag-open-state
46 const ELEMENT_BITS_REGEX
= '!^(/?)([A-Za-z][^\t\n\v />\0]*+)([^>]*?)(/?>)([^<]*)$!';
49 * Blacklist for evil uris like javascript:
50 * WARNING: DO NOT use this in any place that actually requires blacklisting
51 * for security reasons. There are NUMEROUS[1] ways to bypass blacklisting, the
52 * only way to be secure from javascript: uri based xss vectors is to whitelist
53 * things that you know are safe and deny everything else.
54 * [1]: http://ha.ckers.org/xss.html
56 const EVIL_URI_PATTERN
= '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i';
57 const XMLNS_ATTRIBUTE_PATTERN
= "/^xmlns:[:A-Z_a-z-.0-9]+$/";
60 * Tells escapeUrlForHtml() to encode the ID using the wiki's primary encoding.
67 * Tells escapeUrlForHtml() to encode the ID using the fallback encoding, or return false
68 * if no fallback is configured.
72 const ID_FALLBACK
= 1;
75 * List of all named character entities defined in HTML 4.01
76 * https://www.w3.org/TR/html4/sgml/entities.html
77 * As well as ' which is only defined starting in XHTML1.
79 private static $htmlEntities = [
95 'apos' => 39, // New in XHTML & HTML 5; avoid in output for compatibility with IE.
336 * Character entity aliases accepted by MediaWiki
338 private static $htmlEntityAliases = [
344 * Lazy-initialised attributes regex, see getAttribsRegex()
346 private static $attribsRegex;
349 * Regular expression to match HTML/XML attribute pairs within a tag.
350 * Allows some... latitude. Based on,
351 * https://www.w3.org/TR/html5/syntax.html#before-attribute-value-state
352 * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
355 static function getAttribsRegex() {
356 if ( self
::$attribsRegex === null ) {
357 $attribFirst = "[:_\p{L}\p{N}]";
358 $attrib = "[:_\.\-\p{L}\p{N}]";
359 $space = '[\x09\x0a\x0c\x0d\x20]';
360 self
::$attribsRegex =
361 "/(?:^|$space)({$attribFirst}{$attrib}*)
364 # The attribute value: quoted or alone
369 )?(?=$space|\$)/sxu";
371 return self
::$attribsRegex;
375 * Return the various lists of recognized tags
376 * @param array $extratags For any extra tags to include
377 * @param array $removetags For any tags (default or extra) to exclude
380 public static function getRecognizedTagData( $extratags = [], $removetags = [] ) {
381 global $wgAllowImageTag;
383 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
384 $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
386 // Base our staticInitialised variable off of the global config state so that if the globals
387 // are changed (like in the screwed up test system) we will re-initialise the settings.
388 $globalContext = $wgAllowImageTag;
389 if ( !$staticInitialised ||
$staticInitialised != $globalContext ) {
390 $htmlpairsStatic = [ # Tags that must be closed
391 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
392 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
393 'strike', 'strong', 'tt', 'var', 'div', 'center',
394 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
395 'ruby', 'rb', 'rp', 'rt', 'rtc', 'p', 'span', 'abbr', 'dfn',
396 'kbd', 'samp', 'data', 'time', 'mark'
399 'br', 'wbr', 'hr', 'li', 'dt', 'dd', 'meta', 'link'
402 # Elements that cannot have close tags. This is (not coincidentally)
403 # also the list of tags for which the HTML 5 parsing algorithm
404 # requires you to "acknowledge the token's self-closing flag", i.e.
405 # a self-closing tag like <br/> is not an HTML 5 parse error only
408 'br', 'wbr', 'hr', 'meta', 'link'
411 $htmlnest = [ # Tags that can be nested--??
412 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
413 'li', 'dl', 'dt', 'dd', 'font', 'big', 'small', 'sub', 'sup', 'span',
414 'var', 'kbd', 'samp', 'em', 'strong', 'q', 'ruby', 'bdo'
416 $tabletags = [ # Can only appear inside table, we will close them
419 $htmllist = [ # Tags used by list
422 $listtags = [ # Tags that can appear in a list
426 if ( $wgAllowImageTag ) {
427 $htmlsingle[] = 'img';
428 $htmlsingleonly[] = 'img';
431 $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
432 $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
434 # Convert them all to hashtables for faster lookup
435 $vars = [ 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
436 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' ];
437 foreach ( $vars as $var ) {
438 $
$var = array_flip( $
$var );
440 $staticInitialised = $globalContext;
443 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
444 $extratags = array_flip( $extratags );
445 $removetags = array_flip( $removetags );
446 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
447 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ), $removetags );
450 'htmlpairs' => $htmlpairs,
451 'htmlsingle' => $htmlsingle,
452 'htmlsingleonly' => $htmlsingleonly,
453 'htmlnest' => $htmlnest,
454 'tabletags' => $tabletags,
455 'htmllist' => $htmllist,
456 'listtags' => $listtags,
457 'htmlsingleallowed' => $htmlsingleallowed,
458 'htmlelements' => $htmlelements,
463 * Cleans up HTML, removes dangerous tags and attributes, and
464 * removes HTML comments
465 * @param string $text
466 * @param callable $processCallback Callback to do any variable or parameter
467 * replacements in HTML attribute values
468 * @param array|bool $args Arguments for the processing callback
469 * @param array $extratags For any extra tags to include
470 * @param array $removetags For any tags (default or extra) to exclude
471 * @param callable $warnCallback (Deprecated) Callback allowing the
472 * addition of a tracking category when bad input is encountered.
473 * DO NOT ADD NEW PARAMETERS AFTER $warnCallback, since it will be
477 public static function removeHTMLtags( $text, $processCallback = null,
478 $args = [], $extratags = [], $removetags = [], $warnCallback = null
480 $tagData = self
::getRecognizedTagData( $extratags, $removetags );
481 $htmlpairs = $tagData['htmlpairs'];
482 $htmlsingle = $tagData['htmlsingle'];
483 $htmlsingleonly = $tagData['htmlsingleonly'];
484 $htmlnest = $tagData['htmlnest'];
485 $tabletags = $tagData['tabletags'];
486 $htmllist = $tagData['htmllist'];
487 $listtags = $tagData['listtags'];
488 $htmlsingleallowed = $tagData['htmlsingleallowed'];
489 $htmlelements = $tagData['htmlelements'];
491 # Remove HTML comments
492 $text = self
::removeHTMLcomments( $text );
493 $bits = explode( '<', $text );
494 $text = str_replace( '>', '>', array_shift( $bits ) );
495 if ( !MWTidy
::isEnabled() ) {
496 $tagstack = $tablestack = [];
497 foreach ( $bits as $x ) {
499 # $slash: Does the current element start with a '/'?
500 # $t: Current element name
501 # $params: String between element name and >
502 # $brace: Ending '>' or '/>'
503 # $rest: Everything until the next element of $bits
504 if ( preg_match( self
::ELEMENT_BITS_REGEX
, $x, $regs ) ) {
505 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
507 $slash = $t = $params = $brace = $rest = null;
511 $t = strtolower( $t );
512 if ( isset( $htmlelements[$t] ) ) {
514 if ( $slash && isset( $htmlsingleonly[$t] ) ) {
516 } elseif ( $slash ) {
517 # Closing a tag... is it the one we just opened?
518 MediaWiki\
suppressWarnings();
519 $ot = array_pop( $tagstack );
520 MediaWiki\restoreWarnings
();
523 if ( isset( $htmlsingleallowed[$ot] ) ) {
524 # Pop all elements with an optional close tag
525 # and see if we find a match below them
527 array_push( $optstack, $ot );
528 MediaWiki\
suppressWarnings();
529 $ot = array_pop( $tagstack );
530 MediaWiki\restoreWarnings
();
531 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
532 array_push( $optstack, $ot );
533 MediaWiki\
suppressWarnings();
534 $ot = array_pop( $tagstack );
535 MediaWiki\restoreWarnings
();
538 # No match. Push the optional elements back again
540 MediaWiki\
suppressWarnings();
541 $ot = array_pop( $optstack );
542 MediaWiki\restoreWarnings
();
544 array_push( $tagstack, $ot );
545 MediaWiki\
suppressWarnings();
546 $ot = array_pop( $optstack );
547 MediaWiki\restoreWarnings
();
551 MediaWiki\
suppressWarnings();
552 array_push( $tagstack, $ot );
553 MediaWiki\restoreWarnings
();
555 # <li> can be nested in <ul> or <ol>, skip those cases:
556 if ( !isset( $htmllist[$ot] ) ||
!isset( $listtags[$t] ) ) {
561 if ( $t == 'table' ) {
562 $tagstack = array_pop( $tablestack );
567 # Keep track for later
568 if ( isset( $tabletags[$t] ) && !in_array( 'table', $tagstack ) ) {
570 } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
572 # Is it a self closed htmlpair ? (T7487)
573 } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
574 // Eventually we'll just remove the self-closing
575 // slash, in order to be consistent with HTML5
578 // For now, let's just warn authors to clean up.
579 if ( is_callable( $warnCallback ) ) {
580 call_user_func_array( $warnCallback, [ 'deprecated-self-close-category' ] );
583 } elseif ( isset( $htmlsingleonly[$t] ) ) {
584 # Hack to force empty tag for unclosable elements
586 } elseif ( isset( $htmlsingle[$t] ) ) {
587 # Hack to not close $htmlsingle tags
589 # Still need to push this optionally-closed tag to
590 # the tag stack so that we can match end tags
591 # instead of marking them as bad.
592 array_push( $tagstack, $t );
593 } elseif ( isset( $tabletags[$t] ) && in_array( $t, $tagstack ) ) {
594 // New table tag but forgot to close the previous one
597 if ( $t == 'table' ) {
598 array_push( $tablestack, $tagstack );
601 array_push( $tagstack, $t );
604 # Replace any variables or template parameters with
606 if ( is_callable( $processCallback ) ) {
607 call_user_func_array( $processCallback, [ &$params, $args ] );
610 if ( !self
::validateTag( $params, $t ) ) {
614 # Strip non-approved attributes from the tag
615 $newparams = self
::fixTagAttributes( $params, $t );
618 $rest = str_replace( '>', '>', $rest );
619 $close = ( $brace == '/>' && !$slash ) ?
' /' : '';
620 $text .= "<$slash$t$newparams$close>$rest";
624 $text .= '<' . str_replace( '>', '>', $x );
626 # Close off any remaining tags
627 while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
629 if ( $t == 'table' ) {
630 $tagstack = array_pop( $tablestack );
634 # this might be possible using tidy itself
635 foreach ( $bits as $x ) {
636 if ( preg_match( self
::ELEMENT_BITS_REGEX
, $x, $regs ) ) {
637 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
640 $t = strtolower( $t );
641 if ( isset( $htmlelements[$t] ) ) {
642 if ( is_callable( $processCallback ) ) {
643 call_user_func_array( $processCallback, [ &$params, $args ] );
646 if ( $brace == '/>' && !( isset( $htmlsingle[$t] ) ||
isset( $htmlsingleonly[$t] ) ) ) {
647 // Eventually we'll just remove the self-closing
648 // slash, in order to be consistent with HTML5
651 // For now, let's just warn authors to clean up.
652 if ( is_callable( $warnCallback ) ) {
653 call_user_func_array( $warnCallback, [ 'deprecated-self-close-category' ] );
656 if ( !self
::validateTag( $params, $t ) ) {
660 $newparams = self
::fixTagAttributes( $params, $t );
662 if ( $brace === '/>' && !isset( $htmlsingleonly[$t] ) ) {
663 # Interpret self-closing tags as empty tags even when
664 # HTML 5 would interpret them as start tags. Such input
665 # is commonly seen on Wikimedia wikis with this intention.
669 $rest = str_replace( '>', '>', $rest );
670 $text .= "<$slash$t$newparams$brace$rest";
675 $text .= '<' . str_replace( '>', '>', $x );
682 * Remove '<!--', '-->', and everything between.
683 * To avoid leaving blank lines, when a comment is both preceded
684 * and followed by a newline (ignoring spaces), trim leading and
685 * trailing spaces and one of the newlines.
687 * @param string $text
690 public static function removeHTMLcomments( $text ) {
691 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
692 $end = strpos( $text, '-->', $start +
4 );
693 if ( $end === false ) {
694 # Unterminated comment; bail out
700 # Trim space and newline if the comment is both
701 # preceded and followed by a newline
702 $spaceStart = max( $start - 1, 0 );
703 $spaceLen = $end - $spaceStart;
704 while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
708 while ( substr( $text, $spaceStart +
$spaceLen, 1 ) === ' ' ) {
711 if ( substr( $text, $spaceStart, 1 ) === "\n"
712 && substr( $text, $spaceStart +
$spaceLen, 1 ) === "\n" ) {
713 # Remove the comment, leading and trailing
714 # spaces, and leave only one newline.
715 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen +
1 );
717 # Remove just the comment.
718 $text = substr_replace( $text, '', $start, $end - $start );
725 * Takes attribute names and values for a tag and the tag name and
726 * validates that the tag is allowed to be present.
727 * This DOES NOT validate the attributes, nor does it validate the
728 * tags themselves. This method only handles the special circumstances
729 * where we may want to allow a tag within content but ONLY when it has
730 * specific attributes set.
732 * @param string $params
733 * @param string $element
736 static function validateTag( $params, $element ) {
737 $params = self
::decodeTagAttributes( $params );
739 if ( $element == 'meta' ||
$element == 'link' ) {
740 if ( !isset( $params['itemprop'] ) ) {
741 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
744 if ( $element == 'meta' && !isset( $params['content'] ) ) {
745 // <meta> must have a content="" for the itemprop
748 if ( $element == 'link' && !isset( $params['href'] ) ) {
749 // <link> must have an associated href=""
758 * Take an array of attribute names and values and normalize or discard
759 * illegal values for the given element type.
761 * - Discards attributes not on a whitelist for the given element
762 * - Unsafe style attributes are discarded
763 * - Invalid id attributes are re-encoded
765 * @param array $attribs
766 * @param string $element
769 * @todo Check for legal values where the DTD limits things.
770 * @todo Check for unique id attribute :P
772 static function validateTagAttributes( $attribs, $element ) {
773 return self
::validateAttributes( $attribs,
774 self
::attributeWhitelist( $element ) );
778 * Take an array of attribute names and values and normalize or discard
779 * illegal values for the given whitelist.
781 * - Discards attributes not on the given whitelist
782 * - Unsafe style attributes are discarded
783 * - Invalid id attributes are re-encoded
785 * @param array $attribs
786 * @param array $whitelist List of allowed attribute names
789 * @todo Check for legal values where the DTD limits things.
790 * @todo Check for unique id attribute :P
792 static function validateAttributes( $attribs, $whitelist ) {
793 $whitelist = array_flip( $whitelist );
794 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
797 foreach ( $attribs as $attribute => $value ) {
798 # Allow XML namespace declaration to allow RDFa
799 if ( preg_match( self
::XMLNS_ATTRIBUTE_PATTERN
, $attribute ) ) {
800 if ( !preg_match( self
::EVIL_URI_PATTERN
, $value ) ) {
801 $out[$attribute] = $value;
807 # Allow any attribute beginning with "data-"
809 # * Disallow data attributes used by MediaWiki code
810 # * Ensure that the attribute is not namespaced by banning
812 if ( !preg_match( '/^data-[^:]*$/i', $attribute )
813 && !isset( $whitelist[$attribute] )
814 || self
::isReservedDataAttribute( $attribute )
819 # Strip javascript "expression" from stylesheets.
820 # https://msdn.microsoft.com/en-us/library/ms537634.aspx
821 if ( $attribute == 'style' ) {
822 $value = self
::checkCss( $value );
825 # Escape HTML id attributes
826 if ( $attribute === 'id' ) {
827 $value = self
::escapeIdForAttribute( $value, self
::ID_PRIMARY
);
830 # Escape HTML id reference lists
831 if ( $attribute === 'aria-describedby'
832 ||
$attribute === 'aria-flowto'
833 ||
$attribute === 'aria-labelledby'
834 ||
$attribute === 'aria-owns'
836 $value = self
::escapeIdReferenceList( $value );
839 // RDFa and microdata properties allow URLs, URIs and/or CURIs.
840 // Check them for sanity.
841 if ( $attribute === 'rel' ||
$attribute === 'rev'
843 ||
$attribute === 'about' ||
$attribute === 'property'
844 ||
$attribute === 'resource' ||
$attribute === 'datatype'
845 ||
$attribute === 'typeof'
847 ||
$attribute === 'itemid' ||
$attribute === 'itemprop'
848 ||
$attribute === 'itemref' ||
$attribute === 'itemscope'
849 ||
$attribute === 'itemtype'
851 // Paranoia. Allow "simple" values but suppress javascript
852 if ( preg_match( self
::EVIL_URI_PATTERN
, $value ) ) {
857 # NOTE: even though elements using href/src are not allowed directly, supply
858 # validation code that can be used by tag hook handlers, etc
859 if ( $attribute === 'href' ||
$attribute === 'src' ||
$attribute === 'poster' ) {
860 if ( !preg_match( $hrefExp, $value ) ) {
861 continue; // drop any href or src attributes not using an allowed protocol.
862 // NOTE: this also drops all relative URLs
866 // If this attribute was previously set, override it.
867 // Output should only have one attribute of each name.
868 $out[$attribute] = $value;
871 # itemtype, itemid, itemref don't make sense without itemscope
872 if ( !array_key_exists( 'itemscope', $out ) ) {
873 unset( $out['itemtype'] );
874 unset( $out['itemid'] );
875 unset( $out['itemref'] );
877 # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
883 * Given an attribute name, checks whether it is a reserved data attribute
884 * (such as data-mw-foo) which is unavailable to user-generated HTML so MediaWiki
885 * core and extension code can safely use it to communicate with frontend code.
886 * @param string $attr Attribute name.
889 public static function isReservedDataAttribute( $attr ) {
890 // data-ooui is reserved for ooui.
891 // data-mw and data-parsoid are reserved for parsoid.
892 // data-mw-<name here> is reserved for extensions (or core) if
893 // they need to communicate some data to the client and want to be
894 // sure that it isn't coming from an untrusted user.
895 // We ignore the possibility of namespaces since user-generated HTML
896 // can't use them anymore.
897 return (bool)preg_match( '/^data-(ooui|mw|parsoid)/i', $attr );
901 * Merge two sets of HTML attributes. Conflicting items in the second set
902 * will override those in the first, except for 'class' attributes which
903 * will be combined (if they're both strings).
905 * @todo implement merging for other attributes such as style
910 static function mergeAttributes( $a, $b ) {
911 $out = array_merge( $a, $b );
912 if ( isset( $a['class'] ) && isset( $b['class'] )
913 && is_string( $a['class'] ) && is_string( $b['class'] )
914 && $a['class'] !== $b['class']
916 $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
917 -1, PREG_SPLIT_NO_EMPTY
);
918 $out['class'] = implode( ' ', array_unique( $classes ) );
924 * Normalize CSS into a format we can easily search for hostile input
925 * - decode character references
926 * - decode escape sequences
927 * - convert characters that IE6 interprets into ascii
928 * - remove comments, unless the entire value is one single comment
929 * @param string $value the css string
930 * @return string normalized css
932 public static function normalizeCss( $value ) {
933 // Decode character references like {
934 $value = self
::decodeCharReferences( $value );
936 // Decode escape sequences and line continuation
937 // See the grammar in the CSS 2 spec, appendix D.
938 // This has to be done AFTER decoding character references.
939 // This means it isn't possible for this function to return
940 // unsanitized escape sequences. It is possible to manufacture
941 // input that contains character references that decode to
942 // escape sequences that decode to character references, but
943 // it's OK for the return value to contain character references
944 // because the caller is supposed to escape those anyway.
946 if ( !$decodeRegex ) {
947 $space = '[\\x20\\t\\r\\n\\f]';
948 $nl = '(?:\\n|\\r\\n|\\r|\\f)';
950 $decodeRegex = "/ $backslash
952 ($nl) | # 1. Line continuation
953 ([0-9A-Fa-f]{1,6})$space? | # 2. character number
954 (.) | # 3. backslash cancelling special meaning
955 () | # 4. backslash at end of string
958 $value = preg_replace_callback( $decodeRegex,
959 [ __CLASS__
, 'cssDecodeCallback' ], $value );
961 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
962 $value = preg_replace_callback(
963 '/[!-[]-z]/u', // U+FF01 to U+FF5A, excluding U+FF3C (T60088)
964 function ( $matches ) {
965 $cp = UtfNormal\Utils
::utf8ToCodepoint( $matches[0] );
966 if ( $cp === false ) {
969 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
974 // Convert more characters IE6 might treat as ascii
975 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
976 $value = str_replace(
977 [ 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ],
978 [ 'r', 'n', 'n', 'l', 'i', '(', '(' ],
982 // Let the value through if it's nothing but a single comment, to
983 // allow other functions which may reject it to pass some error
985 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
986 // Remove any comments; IE gets token splitting wrong
987 // This must be done AFTER decoding character references and
988 // escape sequences, because those steps can introduce comments
989 // This step cannot introduce character references or escape
990 // sequences, because it replaces comments with spaces rather
991 // than removing them completely.
992 $value = StringUtils
::delimiterReplace( '/*', '*/', ' ', $value );
994 // Remove anything after a comment-start token, to guard against
995 // incorrect client implementations.
996 $commentPos = strpos( $value, '/*' );
997 if ( $commentPos !== false ) {
998 $value = substr( $value, 0, $commentPos );
1002 // S followed by repeat, iteration, or prolonged sound marks,
1003 // which IE will treat as "ss"
1004 $value = preg_replace(
1006 \xE3\x80\xB1 | # U+3031
1007 \xE3\x82\x9D | # U+309D
1008 \xE3\x83\xBC | # U+30FC
1009 \xE3\x83\xBD | # U+30FD
1010 \xEF\xB9\xBC | # U+FE7C
1011 \xEF\xB9\xBD | # U+FE7D
1012 \xEF\xBD\xB0 # U+FF70
1022 * Pick apart some CSS and check it for forbidden or unsafe structures.
1023 * Returns a sanitized string. This sanitized string will have
1024 * character references and escape sequences decoded and comments
1025 * stripped (unless it is itself one valid comment, in which case the value
1026 * will be passed through). If the input is just too evil, only a comment
1027 * complaining about evilness will be returned.
1029 * Currently URL references, 'expression', 'tps' are forbidden.
1031 * NOTE: Despite the fact that character references are decoded, the
1032 * returned string may contain character references given certain
1033 * clever input strings. These character references must
1034 * be escaped before the return value is embedded in HTML.
1036 * @param string $value
1039 static function checkCss( $value ) {
1040 $value = self
::normalizeCss( $value );
1042 // Reject problematic keywords and control characters
1043 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ||
1044 strpos( $value, UtfNormal\Constants
::UTF8_REPLACEMENT
) !== false ) {
1045 return '/* invalid control char */';
1046 } elseif ( preg_match(
1051 | -o-link-source\s*:
1056 | attr\s*\([^)]+[\s,]+url
1058 return '/* insecure input */';
1064 * @param array $matches
1067 static function cssDecodeCallback( $matches ) {
1068 if ( $matches[1] !== '' ) {
1069 // Line continuation
1071 } elseif ( $matches[2] !== '' ) {
1072 $char = UtfNormal\Utils
::codepointToUtf8( hexdec( $matches[2] ) );
1073 } elseif ( $matches[3] !== '' ) {
1074 $char = $matches[3];
1078 if ( $char == "\n" ||
$char == '"' ||
$char == "'" ||
$char == '\\' ) {
1079 // These characters need to be escaped in strings
1080 // Clean up the escape sequence to avoid parsing errors by clients
1081 return '\\' . dechex( ord( $char ) ) . ' ';
1083 // Decode unnecessary escape
1089 * Take a tag soup fragment listing an HTML element's attributes
1090 * and normalize it to well-formed XML, discarding unwanted attributes.
1091 * Output is safe for further wikitext processing, with escaping of
1092 * values that could trigger problems.
1094 * - Normalizes attribute names to lowercase
1095 * - Discards attributes not on a whitelist for the given element
1096 * - Turns broken or invalid entities into plaintext
1097 * - Double-quotes all attribute values
1098 * - Attributes without values are given the name as attribute
1099 * - Double attributes are discarded
1100 * - Unsafe style attributes are discarded
1101 * - Prepends space if there are attributes.
1102 * - (Optionally) Sorts attributes by name.
1104 * @param string $text
1105 * @param string $element
1106 * @param bool $sorted Whether to sort the attributes (default: false)
1109 static function fixTagAttributes( $text, $element, $sorted = false ) {
1110 if ( trim( $text ) == '' ) {
1114 $decoded = self
::decodeTagAttributes( $text );
1115 $stripped = self
::validateTagAttributes( $decoded, $element );
1121 return self
::safeEncodeTagAttributes( $stripped );
1125 * Encode an attribute value for HTML output.
1126 * @param string $text
1127 * @return string HTML-encoded text fragment
1129 static function encodeAttribute( $text ) {
1130 $encValue = htmlspecialchars( $text, ENT_QUOTES
);
1132 // Whitespace is normalized during attribute decoding,
1133 // so if we've been passed non-spaces we must encode them
1134 // ahead of time or they won't be preserved.
1135 $encValue = strtr( $encValue, [
1145 * Encode an attribute value for HTML tags, with extra armoring
1146 * against further wiki processing.
1147 * @param string $text
1148 * @return string HTML-encoded text fragment
1150 static function safeEncodeAttribute( $text ) {
1151 $encValue = self
::encodeAttribute( $text );
1153 # Templates and links may be expanded in later parsing,
1154 # creating invalid or dangerous output. Suppress this.
1155 $encValue = strtr( $encValue, [
1156 '<' => '<', // This should never happen,
1157 '>' => '>', // we've received invalid input
1158 '"' => '"', // which should have been escaped.
1160 '}' => '}', // prevent unpaired language conversion syntax
1163 "''" => '''',
1164 'ISBN' => 'ISBN',
1166 'PMID' => 'PMID',
1172 $encValue = preg_replace_callback(
1173 '/((?i)' . wfUrlProtocols() . ')/',
1174 [ 'Sanitizer', 'armorLinksCallback' ],
1180 * Given a value, escape it so that it can be used in an id attribute and
1181 * return it. This will use HTML5 validation if $wgExperimentalHtmlIds is
1182 * true, allowing anything but ASCII whitespace. Otherwise it will use
1183 * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1184 * escaped with lots of dots.
1186 * To ensure we don't have to bother escaping anything, we also strip ', ",
1187 * & even if $wgExperimentalIds is true. TODO: Is this the best tactic?
1188 * We also strip # because it upsets IE, and % because it could be
1189 * ambiguous if it's part of something that looks like a percent escape
1190 * (which don't work reliably in fragments cross-browser).
1192 * @deprecated since 1.30, use one of this class' escapeIdFor*() functions
1194 * @see https://www.w3.org/TR/html401/types.html#type-name Valid characters
1195 * in the id and name attributes
1196 * @see https://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1198 * @see https://www.w3.org/TR/html5/dom.html#the-id-attribute
1199 * HTML5 definition of id attribute
1201 * @param string $id Id to escape
1202 * @param string|array $options String or array of strings (default is array()):
1203 * 'noninitial': This is a non-initial fragment of an id, not a full id,
1204 * so don't pay attention if the first character isn't valid at the
1205 * beginning of an id. Only matters if $wgExperimentalHtmlIds is
1207 * 'legacy': Behave the way the old HTML 4-based ID escaping worked even
1208 * if $wgExperimentalHtmlIds is used, so we can generate extra
1209 * anchors and links won't break.
1212 static function escapeId( $id, $options = [] ) {
1213 global $wgExperimentalHtmlIds;
1214 $options = (array)$options;
1216 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1217 $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1218 $id = trim( $id, '_' );
1220 // Must have been all whitespace to start with.
1227 // HTML4-style escaping
1233 $id = urlencode( strtr( $id, ' ', '_' ) );
1234 $id = strtr( $id, $replace );
1236 if ( !preg_match( '/^[a-zA-Z]/', $id ) && !in_array( 'noninitial', $options ) ) {
1237 // Initial character must be a letter!
1244 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1245 * a valid HTML id attribute.
1247 * WARNING: unlike escapeId(), the output of this function is not guaranteed to be HTML safe,
1248 * be sure to use proper escaping.
1250 * @param string $id String to escape
1251 * @param int $mode One of ID_* constants, specifying whether the primary or fallback encoding
1253 * @return string|bool Escaped ID or false if fallback encoding is requested but it's not
1258 public static function escapeIdForAttribute( $id, $mode = self
::ID_PRIMARY
) {
1259 global $wgFragmentMode;
1261 if ( !isset( $wgFragmentMode[$mode] ) ) {
1262 if ( $mode === self
::ID_PRIMARY
) {
1263 throw new UnexpectedValueException( '$wgFragmentMode is configured with no primary mode' );
1268 $internalMode = $wgFragmentMode[$mode];
1270 return self
::escapeIdInternal( $id, $internalMode );
1274 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1275 * a valid URL fragment.
1277 * WARNING: unlike escapeId(), the output of this function is not guaranteed to be HTML safe,
1278 * be sure to use proper escaping.
1280 * @param string $id String to escape
1281 * @return string Escaped ID
1285 public static function escapeIdForLink( $id ) {
1286 global $wgFragmentMode;
1288 if ( !isset( $wgFragmentMode[self
::ID_PRIMARY
] ) ) {
1289 throw new UnexpectedValueException( '$wgFragmentMode is configured with no primary mode' );
1292 $mode = $wgFragmentMode[self
::ID_PRIMARY
];
1294 $id = self
::escapeIdInternal( $id, $mode );
1300 * Given a section name or other user-generated or otherwise unsafe string, escapes it to be
1301 * a valid URL fragment for external interwikis.
1303 * @param string $id String to escape
1304 * @return string Escaped ID
1308 public static function escapeIdForExternalInterwiki( $id ) {
1309 global $wgExternalInterwikiFragmentMode;
1311 $id = self
::escapeIdInternal( $id, $wgExternalInterwikiFragmentMode );
1317 * Helper for escapeIdFor*() functions. Performs most of the actual escaping.
1319 * @param string $id String to escape
1320 * @param string $mode One of modes from $wgFragmentMode
1323 private static function escapeIdInternal( $id, $mode ) {
1326 $id = str_replace( ' ', '_', $id );
1329 // This corresponds to 'noninitial' mode of the old escapeId()
1335 $id = urlencode( str_replace( ' ', '_', $id ) );
1336 $id = strtr( $id, $replace );
1338 case 'html5-legacy':
1339 $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1340 $id = trim( $id, '_' );
1342 // Must have been all whitespace to start with.
1347 throw new InvalidArgumentException( "Invalid mode '$mode' passed to '" . __METHOD__
);
1354 * Given a string containing a space delimited list of ids, escape each id
1355 * to match ids escaped by the escapeId() function.
1357 * @todo remove $options completely in 1.32
1361 * @param string $referenceString Space delimited list of ids
1362 * @param string|array $options Deprecated and does nothing.
1365 static function escapeIdReferenceList( $referenceString, $options = [] ) {
1367 wfDeprecated( __METHOD__
. ' with $options', '1.31' );
1369 # Explode the space delimited list string into an array of tokens
1370 $references = preg_split( '/\s+/', "{$referenceString}", -1, PREG_SPLIT_NO_EMPTY
);
1372 # Escape each token as an id
1373 foreach ( $references as &$ref ) {
1374 $ref = self
::escapeIdForAttribute( $ref );
1377 # Merge the array back to a space delimited list string
1378 # If the array is empty, the result will be an empty string ('')
1379 $referenceString = implode( ' ', $references );
1381 return $referenceString;
1385 * Given a value, escape it so that it can be used as a CSS class and
1388 * @todo For extra validity, input should be validated UTF-8.
1390 * @see https://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1392 * @param string $class
1395 static function escapeClass( $class ) {
1396 // Convert ugly stuff to underscores and kill underscores in ugly places
1397 return rtrim( preg_replace(
1398 [ '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ],
1404 * Given HTML input, escape with htmlspecialchars but un-escape entities.
1405 * This allows (generally harmless) entities like   to survive.
1407 * @param string $html HTML to escape
1408 * @return string Escaped input
1410 static function escapeHtmlAllowEntities( $html ) {
1411 $html = self
::decodeCharReferences( $html );
1412 # It seems wise to escape ' as well as ", as a matter of course. Can't
1413 # hurt. Use ENT_SUBSTITUTE so that incorrectly truncated multibyte characters
1414 # don't cause the entire string to disappear.
1415 $html = htmlspecialchars( $html, ENT_QUOTES | ENT_SUBSTITUTE
);
1420 * Regex replace callback for armoring links against further processing.
1421 * @param array $matches
1424 private static function armorLinksCallback( $matches ) {
1425 return str_replace( ':', ':', $matches[1] );
1429 * Return an associative array of attribute names and values from
1430 * a partial tag string. Attribute names are forced to lowercase,
1431 * character references are decoded to UTF-8 text.
1433 * @param string $text
1436 public static function decodeTagAttributes( $text ) {
1437 if ( trim( $text ) == '' ) {
1443 if ( !preg_match_all(
1444 self
::getAttribsRegex(),
1447 PREG_SET_ORDER
) ) {
1451 foreach ( $pairs as $set ) {
1452 $attribute = strtolower( $set[1] );
1453 $value = self
::getTagAttributeCallback( $set );
1455 // Normalize whitespace
1456 $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1457 $value = trim( $value );
1459 // Decode character references
1460 $attribs[$attribute] = self
::decodeCharReferences( $value );
1466 * Build a partial tag string from an associative array of attribute
1467 * names and values as returned by decodeTagAttributes.
1469 * @param array $assoc_array
1472 public static function safeEncodeTagAttributes( $assoc_array ) {
1474 foreach ( $assoc_array as $attribute => $value ) {
1475 $encAttribute = htmlspecialchars( $attribute );
1476 $encValue = self
::safeEncodeAttribute( $value );
1478 $attribs[] = "$encAttribute=\"$encValue\"";
1480 return count( $attribs ) ?
' ' . implode( ' ', $attribs ) : '';
1484 * Pick the appropriate attribute value from a match set from the
1485 * attribs regex matches.
1488 * @throws MWException When tag conditions are not met.
1491 private static function getTagAttributeCallback( $set ) {
1492 if ( isset( $set[5] ) ) {
1495 } elseif ( isset( $set[4] ) ) {
1498 } elseif ( isset( $set[3] ) ) {
1501 } elseif ( !isset( $set[2] ) ) {
1502 # In XHTML, attributes must have a value so return an empty string.
1503 # See "Empty attribute syntax",
1504 # https://www.w3.org/TR/html5/syntax.html#syntax-attribute-name
1507 throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1512 * @param string $text
1515 private static function normalizeWhitespace( $text ) {
1516 return preg_replace(
1517 '/\r\n|[\x20\x0d\x0a\x09]/',
1523 * Normalizes whitespace in a section name, such as might be returned
1524 * by Parser::stripSectionName(), for use in the id's that are used for
1527 * @param string $section
1530 static function normalizeSectionNameWhitespace( $section ) {
1531 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1535 * Ensure that any entities and character references are legal
1536 * for XML and XHTML specifically. Any stray bits will be
1537 * &-escaped to result in a valid text fragment.
1539 * a. named char refs can only be < > & ", others are
1540 * numericized (this way we're well-formed even without a DTD)
1541 * b. any numeric char refs must be legal chars, not invalid or forbidden
1542 * c. use lower cased "&#x", not "&#X"
1543 * d. fix or reject non-valid attributes
1545 * @param string $text
1549 static function normalizeCharReferences( $text ) {
1550 return preg_replace_callback(
1551 self
::CHAR_REFS_REGEX
,
1552 [ 'Sanitizer', 'normalizeCharReferencesCallback' ],
1557 * @param string $matches
1560 static function normalizeCharReferencesCallback( $matches ) {
1562 if ( $matches[1] != '' ) {
1563 $ret = self
::normalizeEntity( $matches[1] );
1564 } elseif ( $matches[2] != '' ) {
1565 $ret = self
::decCharReference( $matches[2] );
1566 } elseif ( $matches[3] != '' ) {
1567 $ret = self
::hexCharReference( $matches[3] );
1569 if ( is_null( $ret ) ) {
1570 return htmlspecialchars( $matches[0] );
1577 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1578 * return the equivalent numeric entity reference (except for the core <
1579 * > & "). If the entity is a MediaWiki-specific alias, returns
1580 * the HTML equivalent. Otherwise, returns HTML-escaped text of
1581 * pseudo-entity source (eg &foo;)
1583 * @param string $name
1586 static function normalizeEntity( $name ) {
1587 if ( isset( self
::$htmlEntityAliases[$name] ) ) {
1588 return '&' . self
::$htmlEntityAliases[$name] . ';';
1589 } elseif ( in_array( $name, [ 'lt', 'gt', 'amp', 'quot' ] ) ) {
1591 } elseif ( isset( self
::$htmlEntities[$name] ) ) {
1592 return '&#' . self
::$htmlEntities[$name] . ';';
1594 return "&$name;";
1599 * @param int $codepoint
1600 * @return null|string
1602 static function decCharReference( $codepoint ) {
1603 $point = intval( $codepoint );
1604 if ( self
::validateCodepoint( $point ) ) {
1605 return sprintf( '&#%d;', $point );
1612 * @param int $codepoint
1613 * @return null|string
1615 static function hexCharReference( $codepoint ) {
1616 $point = hexdec( $codepoint );
1617 if ( self
::validateCodepoint( $point ) ) {
1618 return sprintf( '&#x%x;', $point );
1625 * Returns true if a given Unicode codepoint is a valid character in
1626 * both HTML5 and XML.
1627 * @param int $codepoint
1630 private static function validateCodepoint( $codepoint ) {
1631 # U+000C is valid in HTML5 but not allowed in XML.
1632 # U+000D is valid in XML but not allowed in HTML5.
1633 # U+007F - U+009F are disallowed in HTML5 (control characters).
1634 return $codepoint == 0x09
1635 ||
$codepoint == 0x0a
1636 ||
( $codepoint >= 0x20 && $codepoint <= 0x7e )
1637 ||
( $codepoint >= 0xa0 && $codepoint <= 0xd7ff )
1638 ||
( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1639 ||
( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1643 * Decode any character references, numeric or named entities,
1644 * in the text and return a UTF-8 string.
1646 * @param string $text
1649 public static function decodeCharReferences( $text ) {
1650 return preg_replace_callback(
1651 self
::CHAR_REFS_REGEX
,
1652 [ 'Sanitizer', 'decodeCharReferencesCallback' ],
1657 * Decode any character references, numeric or named entities,
1658 * in the next and normalize the resulting string. (T16952)
1660 * This is useful for page titles, not for text to be displayed,
1661 * MediaWiki allows HTML entities to escape normalization as a feature.
1663 * @param string $text Already normalized, containing entities
1664 * @return string Still normalized, without entities
1666 public static function decodeCharReferencesAndNormalize( $text ) {
1668 $text = preg_replace_callback(
1669 self
::CHAR_REFS_REGEX
,
1670 [ 'Sanitizer', 'decodeCharReferencesCallback' ],
1677 return $wgContLang->normalize( $text );
1684 * @param string $matches
1687 static function decodeCharReferencesCallback( $matches ) {
1688 if ( $matches[1] != '' ) {
1689 return self
::decodeEntity( $matches[1] );
1690 } elseif ( $matches[2] != '' ) {
1691 return self
::decodeChar( intval( $matches[2] ) );
1692 } elseif ( $matches[3] != '' ) {
1693 return self
::decodeChar( hexdec( $matches[3] ) );
1695 # Last case should be an ampersand by itself
1700 * Return UTF-8 string for a codepoint if that is a valid
1701 * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1702 * @param int $codepoint
1706 static function decodeChar( $codepoint ) {
1707 if ( self
::validateCodepoint( $codepoint ) ) {
1708 return UtfNormal\Utils
::codepointToUtf8( $codepoint );
1710 return UtfNormal\Constants
::UTF8_REPLACEMENT
;
1715 * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1716 * return the UTF-8 encoding of that character. Otherwise, returns
1717 * pseudo-entity source (eg "&foo;")
1719 * @param string $name
1722 static function decodeEntity( $name ) {
1723 if ( isset( self
::$htmlEntityAliases[$name] ) ) {
1724 $name = self
::$htmlEntityAliases[$name];
1726 if ( isset( self
::$htmlEntities[$name] ) ) {
1727 return UtfNormal\Utils
::codepointToUtf8( self
::$htmlEntities[$name] );
1734 * Fetch the whitelist of acceptable attributes for a given element name.
1736 * @param string $element
1739 static function attributeWhitelist( $element ) {
1740 $list = self
::setupAttributeWhitelist();
1741 return isset( $list[$element] )
1747 * Foreach array key (an allowed HTML element), return an array
1748 * of allowed attributes
1751 static function setupAttributeWhitelist() {
1754 if ( $whitelist !== null ) {
1776 # These attributes are specified in section 9 of
1777 # https://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1784 # Microdata. These are specified by
1785 # https://html.spec.whatwg.org/multipage/microdata.html#the-microdata-model
1793 $block = array_merge( $common, [ 'align' ] );
1794 $tablealign = [ 'align', 'valign' ];
1802 'nowrap', # deprecated
1803 'width', # deprecated
1804 'height', # deprecated
1805 'bgcolor', # deprecated
1808 # Numbers refer to sections in HTML 4.01 standard describing the element.
1809 # See: https://www.w3.org/TR/html4/
1813 'center' => $common, # deprecated
1832 'strong' => $common,
1843 'blockquote' => array_merge( $common, [ 'cite' ] ),
1844 'q' => array_merge( $common, [ 'cite' ] ),
1854 'br' => array_merge( $common, [ 'clear' ] ),
1856 # https://www.w3.org/TR/html5/text-level-semantics.html#the-wbr-element
1860 'pre' => array_merge( $common, [ 'width' ] ),
1863 'ins' => array_merge( $common, [ 'cite', 'datetime' ] ),
1864 'del' => array_merge( $common, [ 'cite', 'datetime' ] ),
1867 'ul' => array_merge( $common, [ 'type' ] ),
1868 'ol' => array_merge( $common, [ 'type', 'start', 'reversed' ] ),
1869 'li' => array_merge( $common, [ 'type', 'value' ] ),
1877 'table' => array_merge( $common,
1878 [ 'summary', 'width', 'border', 'frame',
1879 'rules', 'cellspacing', 'cellpadding',
1884 'caption' => $block,
1892 'colgroup' => array_merge( $common, [ 'span' ] ),
1893 'col' => array_merge( $common, [ 'span' ] ),
1896 'tr' => array_merge( $common, [ 'bgcolor' ], $tablealign ),
1899 'td' => array_merge( $common, $tablecell, $tablealign ),
1900 'th' => array_merge( $common, $tablecell, $tablealign ),
1903 # NOTE: <a> is not allowed directly, but the attrib
1904 # whitelist is used from the Parser object
1905 'a' => array_merge( $common, [ 'href', 'rel', 'rev' ] ), # rel/rev esp. for RDFa
1908 # Not usually allowed, but may be used for extension-style hooks
1909 # such as <math> when it is rasterized, or if $wgAllowImageTag is
1911 'img' => array_merge( $common, [ 'alt', 'src', 'width', 'height', 'srcset' ] ),
1913 'video' => array_merge( $common, [ 'poster', 'controls', 'preload', 'width', 'height' ] ),
1914 'source' => array_merge( $common, [ 'type', 'src' ] ),
1915 'track' => array_merge( $common, [ 'type', 'src', 'srclang', 'kind', 'label' ] ),
1923 'strike' => $common,
1928 'font' => array_merge( $common, [ 'size', 'color', 'face' ] ),
1932 'hr' => array_merge( $common, [ 'width' ] ),
1934 # HTML Ruby annotation text module, simple ruby only.
1935 # https://www.w3.org/TR/html5/text-level-semantics.html#the-ruby-element
1940 'rt' => $common, # array_merge( $common, array( 'rbspan' ) ),
1943 # MathML root element, where used for extensions
1944 # 'title' may not be 100% valid here; it's XHTML
1945 # https://www.w3.org/TR/REC-MathML/
1946 'math' => [ 'class', 'style', 'id', 'title' ],
1948 // HTML 5 section 4.5
1949 'figure' => $common,
1950 'figcaption' => $common,
1952 # HTML 5 section 4.6
1955 # HTML5 elements, defined by:
1956 # https://html.spec.whatwg.org/multipage/semantics.html#the-data-element
1957 'data' => array_merge( $common, [ 'value' ] ),
1958 'time' => array_merge( $common, [ 'datetime' ] ),
1961 // meta and link are only permitted by removeHTMLtags when Microdata
1962 // is enabled so we don't bother adding a conditional to hide these
1963 // Also meta and link are only valid in WikiText as Microdata elements
1964 // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1965 // So we don't bother including $common attributes that have no purpose.
1966 'meta' => [ 'itemprop', 'content' ],
1967 'link' => [ 'itemprop', 'href', 'title' ],
1974 * Take a fragment of (potentially invalid) HTML and return
1975 * a version with any tags removed, encoded as plain text.
1977 * Warning: this return value must be further escaped for literal
1978 * inclusion in HTML output as of 1.10!
1980 * @param string $html HTML fragment
1983 static function stripAllTags( $html ) {
1984 // Use RemexHtml to tokenize $html and extract the text
1985 $handler = new RemexStripTagHandler
;
1986 $tokenizer = new RemexHtml\Tokenizer\
Tokenizer( $handler, $html, [
1987 'ignoreErrors' => true,
1988 // don't ignore char refs, we want them to be decoded
1989 'ignoreNulls' => true,
1990 'skipPreprocess' => true,
1992 $tokenizer->execute();
1993 $text = $handler->getResult();
1995 $text = self
::normalizeWhitespace( $text );
2000 * Hack up a private DOCTYPE with HTML's standard entity declarations.
2001 * PHP 4 seemed to know these if you gave it an HTML doctype, but
2004 * Use for passing XHTML fragments to PHP's XML parsing functions
2008 static function hackDocType() {
2009 $out = "<!DOCTYPE html [\n";
2010 foreach ( self
::$htmlEntities as $entity => $codepoint ) {
2011 $out .= "<!ENTITY $entity \"&#$codepoint;\">";
2018 * @param string $url
2019 * @return mixed|string
2021 static function cleanUrl( $url ) {
2022 # Normalize any HTML entities in input. They will be
2023 # re-escaped by makeExternalLink().
2024 $url = self
::decodeCharReferences( $url );
2026 # Escape any control characters introduced by the above step
2027 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
2028 [ __CLASS__
, 'cleanUrlCallback' ], $url );
2030 # Validate hostname portion
2032 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
2033 list( /* $whole */, $protocol, $host, $rest ) = $matches;
2035 // Characters that will be ignored in IDNs.
2036 // https://tools.ietf.org/html/rfc3454#section-3.1
2037 // Strip them before further processing so blacklists and such work.
2039 \\s| # general whitespace
2040 \xc2\xad| # 00ad SOFT HYPHEN
2041 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
2042 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
2043 \xe2\x81\xa0| # 2060 WORD JOINER
2044 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
2045 \xcd\x8f| # 034f COMBINING GRAPHEME JOINER
2046 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
2047 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
2048 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
2049 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
2050 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
2051 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
2054 $host = preg_replace( $strip, '', $host );
2056 // IPv6 host names are bracketed with []. Url-decode these.
2057 if ( substr_compare( "//%5B", $host, 0, 5 ) === 0 &&
2058 preg_match( '!^//%5B([0-9A-Fa-f:.]+)%5D((:\d+)?)$!', $host, $matches )
2060 $host = '//[' . $matches[1] . ']' . $matches[2];
2063 // @todo FIXME: Validate hostnames here
2065 return $protocol . $host . $rest;
2072 * @param array $matches
2075 static function cleanUrlCallback( $matches ) {
2076 return urlencode( $matches[0] );
2080 * Does a string look like an e-mail address?
2082 * This validates an email address using an HTML5 specification found at:
2083 * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
2084 * Which as of 2011-01-24 says:
2086 * A valid e-mail address is a string that matches the ABNF production
2087 * 1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
2088 * in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
2091 * This function is an implementation of the specification as requested in
2094 * Client-side forms will use the same standard validation rules via JS or
2095 * HTML 5 validation; additional restrictions can be enforced server-side
2096 * by extensions via the 'isValidEmailAddr' hook.
2098 * Note that this validation doesn't 100% match RFC 2822, but is believed
2099 * to be liberal enough for wide use. Some invalid addresses will still
2100 * pass validation here.
2104 * @param string $addr E-mail address
2107 public static function validateEmail( $addr ) {
2109 if ( !Hooks
::run( 'isValidEmailAddr', [ $addr, &$result ] ) ) {
2113 // Please note strings below are enclosed in brackets [], this make the
2114 // hyphen "-" a range indicator. Hence it is double backslashed below.
2116 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
2117 $rfc1034_ldh_str = "a-z0-9\\-";
2119 $html5_email_regexp = "/
2121 [$rfc5322_atext\\.]+ # user part which is liberal :p
2123 [$rfc1034_ldh_str]+ # First domain part
2124 (\\.[$rfc1034_ldh_str]+)* # Following part prefixed with a dot
2126 /ix"; // case Insensitive, eXtended
2128 return (bool)preg_match( $html5_email_regexp, $addr );