/**
* Take an array of attribute names and values and normalize or discard
* illegal values for the given whitelist.
*
* - Discards attributes not on the given whitelist
* - Unsafe style attributes are discarded
* - Invalid id attributes are re-encoded
*
* @param array $attribs
* @param array $whitelist List of allowed attribute names
* @return array
*
* @todo Check for legal values where the DTD limits things.
* @todo Check for unique id attribute :P
*/
static function validateAttributes($attribs, $whitelist)
{
$whitelist = array_flip($whitelist);
$hrefExp = '/^(' . wfUrlProtocols() . ')[^\\s]+$/';
$out = [];
foreach ($attribs as $attribute => $value) {
# Allow XML namespace declaration to allow RDFa
if (preg_match(self::XMLNS_ATTRIBUTE_PATTERN, $attribute)) {
if (!preg_match(self::EVIL_URI_PATTERN, $value)) {
$out[$attribute] = $value;
}
continue;
}
# Allow any attribute beginning with "data-"
# However:
# * data-ooui is reserved for ooui
# * data-mw and data-parsoid are reserved for parsoid
# * data-mw-<name here> is reserved for extensions (or core) if
# they need to communicate some data to the client and want to be
# sure that it isn't coming from an untrusted user.
# * Ensure that the attribute is not namespaced by banning
# colons.
if (!preg_match('/^data-(?!ooui|mw|parsoid)[^:]*$/i', $attribute) && !isset($whitelist[$attribute])) {
continue;
}
# Strip javascript "expression" from stylesheets.
# http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
if ($attribute == 'style') {
$value = Sanitizer::checkCss($value);
}
# Escape HTML id attributes
if ($attribute === 'id') {
$value = Sanitizer::escapeId($value, 'noninitial');
}
# Escape HTML id reference lists
if ($attribute === 'aria-describedby' || $attribute === 'aria-flowto' || $attribute === 'aria-labelledby' || $attribute === 'aria-owns') {
$value = Sanitizer::escapeIdReferenceList($value, 'noninitial');
}
// RDFa and microdata properties allow URLs, URIs and/or CURIs.
// Check them for sanity.
if ($attribute === 'rel' || $attribute === 'rev' || $attribute === 'about' || $attribute === 'property' || $attribute === 'resource' || $attribute === 'datatype' || $attribute === 'typeof' || $attribute === 'itemid' || $attribute === 'itemprop' || $attribute === 'itemref' || $attribute === 'itemscope' || $attribute === 'itemtype') {
// Paranoia. Allow "simple" values but suppress javascript
if (preg_match(self::EVIL_URI_PATTERN, $value)) {
continue;
}
}
# NOTE: even though elements using href/src are not allowed directly, supply
# validation code that can be used by tag hook handlers, etc
if ($attribute === 'href' || $attribute === 'src') {
if (!preg_match($hrefExp, $value)) {
continue;
// drop any href or src attributes not using an allowed protocol.
// NOTE: this also drops all relative URLs
}
}
// If this attribute was previously set, override it.
// Output should only have one attribute of each name.
$out[$attribute] = $value;
}
# itemtype, itemid, itemref don't make sense without itemscope
if (!array_key_exists('itemscope', $out)) {
unset($out['itemtype']);
unset($out['itemid']);
unset($out['itemref']);
}
# TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
return $out;
}
/**
* Test escapeIdReferenceList for consistency with escapeId
*
* @dataProvider provideEscapeIdReferenceList
* @covers Sanitizer::escapeIdReferenceList
*/
public function testEscapeIdReferenceList($referenceList, $id1, $id2)
{
$this->assertEquals(Sanitizer::escapeIdReferenceList($referenceList, 'noninitial'), Sanitizer::escapeId($id1, 'noninitial') . ' ' . Sanitizer::escapeId($id2, 'noninitial'));
}
