Cross-Site Scripting (XSS) vulnerability through unescaped HTML attribute values

Impact

The prosemirror_to_html gem is vulnerable to Cross-Site Scripting (XSS) attacks through malicious HTML attribute values. While tag content is properly escaped, attribute values are not, allowing attackers to inject arbitrary JavaScript code.

Who is impacted:

• Any application using prosemirror_to_html to convert ProseMirror documents to HTML
• Applications that process user-generated ProseMirror content are at highest risk
• End users viewing the rendered HTML output could have malicious JavaScript executed in their browsers

Attack vectors include:

• href attributes with javascript: protocol: <a href="javascript:alert(document.cookie)">
• Event handlers: <div onclick="maliciousCode()">
• onerror attributes on images: <img src=x onerror="alert('XSS')">
• Other HTML attributes that can execute JavaScript

Patches

A fix is currently in development. Users should upgrade to version 0.2.1 or later once released.

The patch escapes all HTML attribute values using CGI.escapeHTML to prevent injection attacks.

Workarounds

Until a patched version is available, users can implement one or more of these mitigations:

1. Sanitize output: Pass the HTML output through a sanitization library like Sanitize or Loofah:

   html = ProsemirrorToHtml.render(document)
   safe_html = Sanitize.fragment(html, Sanitize::Config::RELAXED)

2. Implement Content Security Policy (CSP): Add strict CSP headers to prevent inline JavaScript execution:

   Content-Security-Policy: default-src 'self'; script-src 'self'

3. Input validation: If possible, validate and sanitize ProseMirror documents before conversion to prevent malicious content from entering the system.

References

• Vulnerable code: https://github.com/etaminstudio/prosemirror_to_html/blob/ea8beb32f6c37f29f042ba4155ccf18504da716e/lib/prosemirror_to_html.rb#L249
• OWASP XSS Prevention Cheat Sheet
• CWE-79: Improper Neutralization of Input During Web Page Generation