HTML Cleaner allows crafted scripts in special contexts like svg or math to pass through

Vulnerability Description

lxml_html_clean is a project for HTML cleaning functionalities copied from `lxml.html.clean`. Prior to version 0.4.0, the HTML Parser in lxml does not properly handle context-switching for special HTML tags such as `<svg>`, `<math>` and `<noscript>`. This behavior deviates from how web browsers parse and interpret such tags. Specifically, content in CSS comments is ignored by lxml_html_clean but may be interpreted differently by web browsers, enabling malicious scripts to bypass the cleaning process. This vulnerability could lead to Cross-Site Scripting (XSS) attacks, compromising the security of users relying on lxml_html_clean in default configuration for sanitizing untrusted HTML content. Users employing the HTML cleaner in a security-sensitive context should upgrade to lxml 0.4.0, which addresses this issue. As a temporary mitigation, users can configure lxml_html_clean with the following settings to prevent the exploitation of this vulnerability. Via `remove_tags`, one may specify tags to remove - their content is moved to their parents' tags. Via `kill_tags`, one may specify tags to be removed completely. Via `allow_tags`, one may restrict the set of permissible tags, excluding context-switching tags like `<svg>`, `<math>` and `<noscript>`.

References

• https://github.com/fedora-python/lxml_html_clean/security/advisories/GHSA-5jfw-gq64-q45f
• https://github.com/fedora-python/lxml_html_clean/pull/19
• https://github.com/fedora-python/lxml_html_clean/commit/c5d816f86eb3707d72a8ecf5f3823e0daa1b3808