function FilterUnitTestCase::testFilterXSS in Drupal 7
Tests limiting allowed tags and XSS prevention.
XSS tests assume that script is disallowed by default and src is allowed by default, but on* and style attributes are disallowed.
Script injection vectors mostly adopted from http://ha.ckers.org/xss.html.
Relevant CVEs:
- CVE-2002-1806, ~CVE-2005-0682, ~CVE-2005-2106, CVE-2005-3973, CVE-2006-1226 (= rev. 1.112?), CVE-2008-0273, CVE-2008-3740.
File
- modules/
filter/ filter.test, line 983 - Tests for filter.module.
Class
- FilterUnitTestCase
- Unit tests for core filters.
Code
function testFilterXSS() {
// Tag stripping, different ways to work around removal of HTML tags.
$f = filter_xss('<script>alert(0)</script>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping -- simple script without special characters.');
$f = filter_xss('<script src="http://www.example.com" />');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping -- empty script with source.');
$f = filter_xss('<ScRipt sRc=http://www.example.com/>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- varying case.');
$f = filter_xss("<script\nsrc\n=\nhttp://www.example.com/\n>");
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- multiline tag.');
$f = filter_xss('<script/a src=http://www.example.com/a.js></script>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- non whitespace character after tag name.');
$f = filter_xss('<script/src=http://www.example.com/a.js></script>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- no space between tag and attribute.');
// Null between < and tag name works at least with IE6.
$f = filter_xss("<\0scr\0ipt>alert(0)</script>");
$this
->assertNoNormalized($f, 'ipt', 'HTML tag stripping evasion -- breaking HTML with nulls.');
$f = filter_xss("<scrscriptipt src=http://www.example.com/a.js>");
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- filter just removing "script".');
$f = filter_xss('<<script>alert(0);//<</script>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- double opening brackets.');
$f = filter_xss('<script src=http://www.example.com/a.js?<b>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- no closing tag.');
// DRUPAL-SA-2008-047: This doesn't seem exploitable, but the filter should
// work consistently.
$f = filter_xss('<script>>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- double closing tag.');
$f = filter_xss('<script src=//www.example.com/.a>');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- no scheme or ending slash.');
$f = filter_xss('<script src=http://www.example.com/.a');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- no closing bracket.');
$f = filter_xss('<script src=http://www.example.com/ <');
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- opening instead of closing bracket.');
$f = filter_xss('<nosuchtag attribute="newScriptInjectionVector">');
$this
->assertNoNormalized($f, 'nosuchtag', 'HTML tag stripping evasion -- unknown tag.');
$f = filter_xss('<?xml:namespace ns="urn:schemas-microsoft-com:time">');
$this
->assertTrue(stripos($f, '<?xml') === FALSE, 'HTML tag stripping evasion -- starting with a question sign (processing instructions).');
$f = filter_xss('<t:set attributeName="innerHTML" to="<script defer>alert(0)</script>">');
$this
->assertNoNormalized($f, 't:set', 'HTML tag stripping evasion -- colon in the tag name (namespaces\' tricks).');
$f = filter_xss('<img """><script>alert(0)</script>', array(
'img',
));
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- a malformed image tag.');
$f = filter_xss('<blockquote><script>alert(0)</script></blockquote>', array(
'blockquote',
));
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- script in a blockqoute.');
$f = filter_xss("<!--[if true]><script>alert(0)</script><![endif]-->");
$this
->assertNoNormalized($f, 'script', 'HTML tag stripping evasion -- script within a comment.');
// Dangerous attributes removal.
$f = filter_xss('<p onmouseover="http://www.example.com/">', array(
'p',
));
$this
->assertNoNormalized($f, 'onmouseover', 'HTML filter attributes removal -- events, no evasion.');
$f = filter_xss('<li style="list-style-image: url(javascript:alert(0))">', array(
'li',
));
$this
->assertNoNormalized($f, 'style', 'HTML filter attributes removal -- style, no evasion.');
$f = filter_xss('<img onerror =alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'onerror', 'HTML filter attributes removal evasion -- spaces before equals sign.');
$f = filter_xss('<img onabort!#$%&()*~+-_.,:;?@[/|\\]^`=alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'onabort', 'HTML filter attributes removal evasion -- non alphanumeric characters before equals sign.');
$f = filter_xss('<img oNmediAError=alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'onmediaerror', 'HTML filter attributes removal evasion -- varying case.');
// Works at least with IE6.
$f = filter_xss("<img o\0nfocus\0=alert(0)>", array(
'img',
));
$this
->assertNoNormalized($f, 'focus', 'HTML filter attributes removal evasion -- breaking with nulls.');
// Only whitelisted scheme names allowed in attributes.
$f = filter_xss('<img src="javascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing -- no evasion.');
$f = filter_xss('<img src=javascript:alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- no quotes.');
// A bit like CVE-2006-0070.
$f = filter_xss('<img src="javascript:confirm(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- no alert ;)');
$f = filter_xss('<img src=`javascript:alert(0)`>', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- grave accents.');
$f = filter_xss('<img dynsrc="javascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing -- rare attribute.');
$f = filter_xss('<table background="javascript:alert(0)">', array(
'table',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing -- another tag.');
$f = filter_xss('<base href="javascript:alert(0);//">', array(
'base',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing -- one more attribute and tag.');
$f = filter_xss('<img src="jaVaSCriPt:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- varying case.');
$f = filter_xss('<img src=javascript:alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- UTF-8 decimal encoding.');
$f = filter_xss('<img src=javascript:alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- long UTF-8 encoding.');
$f = filter_xss('<img src=javascript:alert(0)>', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- UTF-8 hex encoding.');
$f = filter_xss("<img src=\"jav\tascript:alert(0)\">", array(
'img',
));
$this
->assertNoNormalized($f, 'script', 'HTML scheme clearing evasion -- an embedded tab.');
$f = filter_xss('<img src="jav	ascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'script', 'HTML scheme clearing evasion -- an encoded, embedded tab.');
$f = filter_xss('<img src="jav
ascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'script', 'HTML scheme clearing evasion -- an encoded, embedded newline.');
// With 
 this test would fail, but the entity gets turned into
// &#xD;, so it's OK.
$f = filter_xss('<img src="jav
ascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'script', 'HTML scheme clearing evasion -- an encoded, embedded carriage return.');
$f = filter_xss("<img src=\"\n\n\nj\na\nva\ns\ncript:alert(0)\">", array(
'img',
));
$this
->assertNoNormalized($f, 'cript', 'HTML scheme clearing evasion -- broken into many lines.');
$f = filter_xss("<img src=\"jav\0a\0\0cript:alert(0)\">", array(
'img',
));
$this
->assertNoNormalized($f, 'cript', 'HTML scheme clearing evasion -- embedded nulls.');
// @todo This dataset currently fails under 5.4 because of
// https://www.drupal.org/node/1210798. Restore after it's fixed.
if (version_compare(PHP_VERSION, '5.4.0', '<')) {
$f = filter_xss('<img src="  javascript:alert(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'javascript', 'HTML scheme clearing evasion -- spaces and metacharacters before scheme.');
}
$f = filter_xss('<img src="vbscript:msgbox(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'vbscript', 'HTML scheme clearing evasion -- another scheme.');
$f = filter_xss('<img src="nosuchscheme:notice(0)">', array(
'img',
));
$this
->assertNoNormalized($f, 'nosuchscheme', 'HTML scheme clearing evasion -- unknown scheme.');
// Netscape 4.x javascript entities.
$f = filter_xss('<br size="&{alert(0)}">', array(
'br',
));
$this
->assertNoNormalized($f, 'alert', 'Netscape 4.x javascript entities.');
// DRUPAL-SA-2008-006: Invalid UTF-8, these only work as reflected XSS with
// Internet Explorer 6.
$f = filter_xss("", array(
'p',
));
$this
->assertNoNormalized($f, 'style', 'HTML filter -- invalid UTF-8.');
$f = filter_xss("");
$this
->assertEqual($f, '', 'HTML filter -- overlong UTF-8 sequences.');
$f = filter_xss("Who's Online");
$this
->assertNormalized($f, "who's online", 'HTML filter -- html entity number');
$f = filter_xss("Who&#039;s Online");
$this
->assertNormalized($f, "who's online", 'HTML filter -- encoded html entity number');
$f = filter_xss("Who&amp;#039; Online");
$this
->assertNormalized($f, "who&#039; online", 'HTML filter -- double encoded html entity number');
}