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

Impact

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.

Patches

Users employing the HTML cleaner in a security-sensitive context should upgrade to lxml 0.4.0, which addresses this issue.

Workarounds

As a temporary mitigation, users can configure lxml_html_clean with the following settings to prevent the exploitation of this vulnerability:

• remove_tags: Specify tags to remove - their content is moved to their parents' tags.
• kill_tags: Specify tags to be removed completely.
• allow_tags: Restrict the set of permissible tags, excluding context-switching tags like <svg>, <math> and <noscript>.

References

• https://github.com/fedora-python/lxml_html_clean/pull/19
• https://github.com/fedora-python/lxml_html_clean/pull/19/commits/c5d816f86eb3707d72a8ecf5f3823e0daa1b3808