GHSA-gj48-438w-jh9v
ADVISORY - githubSummary
Summary
Bleach clean() / Cleaner() fails to sanitize dangerous URI schemes in allowed formaction attributes.
Bleach applies URI protocol sanitization only to attributes listed in attr_val_is_uri. While URI-bearing attributes such as action, href, src, and poster are included in that set, formaction is not. As a result, if a downstream application explicitly allows formaction on submit-capable controls in untrusted HTML, Bleach preserves dangerous values such as javascript:alert(1) instead of stripping them.
This can lead to submit-triggered JavaScript execution in applications that rely on Bleach to sanitize untrusted HTML and allow the relevant tag/attribute combination.
Details
The issue appears to be a URI-sanitization coverage gap in Bleach’s sanitizer logic.
Relevant code paths:
bleach/sanitizer.py—BleachSanitizerFilter.allow_token(around line 553)bleach/_vendor/html5lib/filters/sanitizer.py—attr_val_is_uri(around line 525)
In BleachSanitizerFilter.allow_token, URI protocol sanitization is only applied when:
if namespaced_name in self.attr_val_is_uri:
However, (None, 'formaction') is currently missing from attr_val_is_uri.
This creates an inconsistency where action is protocol-sanitized, but formaction is not.
As a result, if a downstream application allows:
- tags such as
<button>or<input> - the
formactionattribute
then Bleach preserves dangerous URI schemes such as javascript: in formaction.
Examples of affected submit-capable controls include:
<button>(default submit behavior unlesstype="button"is set)<input type="submit"><input type="image">
This appears to be a real library-side sanitizer gap rather than only an application misuse issue, because Bleach already treats similar URI-bearing attributes (such as action) as protocol-sensitive and sanitizes them.
Suggested minimal fix:
Add:
(None, 'formaction')
to attr_val_is_uri in:
bleach/_vendor/html5lib/filters/sanitizer.py
I also prepared a minimal patch and focused regression tests if helpful.
PoC
Below are minimal reproductions using bleach.clean().
1) <button>
from bleach import clean
print(clean(
'<form><button formaction="javascript:alert(1)">go</button></form>',
tags={'form', 'button'},
attributes={'button': ['formaction']},
))
Actual output:
<form><button formaction="javascript:alert(1)">go</button></form>
Expected output:
<form><button>go</button></form>
2) <input type="submit">
print(clean(
'<form><input type="submit" formaction="javascript:alert(1)" value="go"></form>',
tags={'form', 'input'},
attributes={'input': ['type', 'formaction', 'value']},
))
Actual output:
<form><input type="submit" formaction="javascript:alert(1)" value="go"></form>
Expected output:
<form><input type="submit" value="go"></form>
3) <input type="image">
print(clean(
'<form><input type="image" formaction="javascript:alert(1)" src="/foo.png"></form>',
tags={'form', 'input'},
attributes={'input': ['type', 'formaction', 'src']},
))
Actual output:
<form><input type="image" formaction="javascript:alert(1)" src="/foo.png"></form>
Expected output:
<form><input type="image" src="/foo.png"></form>
Impact
This is a client-side HTML sanitization bypass / dangerous URI preservation issue.
If an application relies on Bleach to sanitize untrusted HTML and explicitly allows:
formaction- and submit-capable controls such as
<button>or<input>
then Bleach can emit sanitized output that still contains a dangerous javascript: URI in formaction.
That can lead to submit-triggered JavaScript execution when the user activates the control.
Impact is limited to configurations that explicitly allow the relevant tag/attribute combination, but the issue is still security-relevant because:
formactionis a real browser sink- Bleach already protocol-sanitizes similar URI-bearing attributes like
action - the omission creates inconsistent sanitizer coverage for dangerous URI schemes
I would currently assess this as Medium severity.
If useful, I also have:
a minimal patch
focused regression tests for:
<button formaction="javascript:..."><input type="submit" formaction="javascript:..."><input type="image" formaction="javascript:...">- a safe control case where
formaction="/submit"is preserved
Common Weakness Enumeration (CWE)
Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')
GitHub
2.8
CVSS SCORE
6.1medium| Package | Type | OS Name | OS Version | Affected Ranges | Fix Versions |
|---|---|---|---|---|---|
| bleach | pypi | - | - | <6.4.0 | 6.4.0 |
CVSS:3 Severity and metrics
The CVSS metrics represent different qualitative aspects of a vulnerability that impact the overall score, as defined by the CVSS Specification.
The vulnerable component is bound to the network stack, but the attack is limited at the protocol level to a logically adjacent topology. This can mean an attack must be launched from the same shared physical (e.g., Bluetooth or IEEE 802.11) or logical (e.g., local IP subnet) network, or from within a secure or otherwise limited administrative domain (e.g., MPLS, secure VPN to an administrative network zone). One example of an Adjacent attack would be an ARP (IPv4) or neighbor discovery (IPv6) flood leading to a denial of service on the local LAN segment (e.g., CVE-2013-6014).
Specialized access conditions or extenuating circumstances do not exist. An attacker can expect repeatable success when attacking the vulnerable component.
The attacker is unauthorized prior to attack, and therefore does not require any access to settings or files of the vulnerable system to carry out an attack.
Successful exploitation of this vulnerability requires a user to take some action before the vulnerability can be exploited. For example, a successful exploit may only be possible during the installation of an application by a system administrator.
An exploited vulnerability can affect resources beyond the security scope managed by the security authority of the vulnerable component. In this case, the vulnerable component and the impacted component are different and managed by different security authorities.
There is some loss of confidentiality. Access to some restricted information is obtained, but the attacker does not have control over what information is obtained, or the amount or kind of loss is limited. The information disclosure does not cause a direct, serious loss to the impacted component.
Modification of data is possible, but the attacker does not have control over the consequence of a modification, or the amount of modification is limited. The data modification does not have a direct, serious impact on the impacted component.
There is no impact to availability within the impacted component.
Chainguard
CGA-587j-pvj5-v7h4
-