MEDIUM 5.5 PyPI

eml_parser: Path Traversal in Official Example Script Leads to Arbitrary File Write

GHSA-389r-rccm-h3h5 · CVE-2026-29780 · PYSEC-2026-2148

Published · Modified

AI SAST

Find this class of vulnerability in your own code

Corgea's AI-native static analysis detects vulnerabilities like this one across your repositories, ranks them by exploitability, and returns review-ready fixes.

Description

Summary

The official example script examples/recursively_extract_attachments.py contains a path traversal vulnerability that allows arbitrary file write outside the intended output directory. Attachment filenames extracted from parsed emails are directly used to construct output file paths without any sanitization, allowing an attacker-controlled filename to escape the target directory.

Details

File: examples/recursively_extract_attachments.py
Lines: 61–64

for a in m['attachment']:
    out_filepath = out_path / a['filename']  # No sanitization
    print(f'\tWriting attachment: {out_filepath}')
    with out_filepath.open('wb') as a_out:
        a_out.write(base64.b64decode(a['raw']))

The value a['filename'] is attacker-controlled via crafted email attachment headers:

Content-Disposition: attachment; filename="../outside/pwned.txt"

No path normalization or boundary validation is performed before writing.

PoC

  1. Create a malicious .eml file:
Content-Disposition: attachment; filename="../outside/pwned.txt"
  1. Run the example script:
python recursively_extract_attachments.py -p ./emails -o ./safe
  1. Expected: ./safe/pwned.txt
  2. Actual: ./outside/pwned.txt ← written outside the intended directory

Verified on Kali Linux with eml_parser installed via pip in a virtual environment.

Impact

This vulnerability is limited to the example script only and does not affect the core eml_parser library. However, as the script is part of the official repository and is likely to be adapted for production use, an attacker supplying a crafted email could achieve arbitrary file write within the execution context.

Potential attack scenarios include:

  • Cron job injection: filename="../../etc/cron.d/backdoor"
  • Web shell upload: filename="../../var/www/html/shell.php"
  • SSH key injection: filename="../../home/user/.ssh/authorized_keys"

Recommended Fix

import os.path

for a in m['attachment']:
    filename = os.path.basename(a['filename'])
    out_filepath = out_path / filename

    if not out_filepath.resolve().is_relative_to(out_path.resolve()):
        print(f'[!] Skipping suspicious filename: {a["filename"]}')
        continue

    print(f'\tWriting attachment: {out_filepath}')
    with out_filepath.open('wb') as a_out:
        a_out.write(base64.b64decode(a['raw']))

Ready to move

Start Securing

Free, no credit card | First findings in minutes