Problem
Due to a parsing issue in the upstream package masterminds/html5
, malicious markup used in a sequence with special HTML CDATA sections cannot be filtered and sanitized. This allows bypassing the cross-site scripting mechanism of typo3/html-sanitizer
.
Besides that, the upstream package masterminds/html5
provides HTML raw text elements (script
, style
, noframes
, noembed
and iframe
) as DOMText nodes, which were not processed and sanitized further. None of the mentioned elements were defined in the default builder configuration, that’s why only custom behaviors, using one of those tag names, were vulnerable to cross-site scripting.
Solution
Update to typo3/html-sanitizer
versions 1.5.0 or 2.1.1 that fix the problem described.
References
- https://github.com/TYPO3/html-sanitizer/security/advisories/GHSA-hvwx-qh2h-xcfj
- https://github.com/TYPO3/html-sanitizer/pull/105
- https://github.com/TYPO3/html-sanitizer/pull/106
- https://github.com/TYPO3/html-sanitizer/releases/tag/v1.5.0
- https://github.com/TYPO3/html-sanitizer/releases/tag/v2.1.1
- https://github.com/FriendsOfPHP/security-advisories/blob/master/typo3/cms/CVE-2022-23499.yaml
- https://typo3.org/security/advisory/typo3-core-sa-2022-017
- https://nvd.nist.gov/vuln/detail/CVE-2022-23499
- https://github.com/advisories/GHSA-hvwx-qh2h-xcfj