tarslip-guard

Find and fix tar-slip / zip-slip path-traversal bugs (CWE-22). Two complementary, dependency-free tools:

1. A static checker that flags extractall() calls with no filter= or members= argument — the pattern that lets a malicious archive write files outside the intended directory.
2. A safe runtime extractor (safe_extractall) that validates every member before writing it, so you can extract untrusted archives without traversal.

Install

pip install tarslip-guard

Static checking

from tarslip_guard import check_source, check_path

check_source("import tarfile\ntarfile.open('x').extractall('out')\n")
# -> [Finding(... rule='TARSLIP-EXTRACTALL', cwe='CWE-22', line=2 ...)]

for finding in check_path("src/"):
    print(finding)

tarslip-guard src/ tests/      # exits 1 if any unsafe call is found

extractall(..., filter="data") (Python 3.12+) and extractall(members=...) are treated as safe and not flagged.

Safe extraction

from tarslip_guard import safe_extractall, UnsafeArchiveError

try:
    safe_extractall("untrusted.tar.gz", "output_dir")
except UnsafeArchiveError as exc:
    print("refused:", exc)

• Accepts an open tarfile.TarFile / zipfile.ZipFile, or a path (type is auto-detected).
• Rejects absolute paths and .. traversal for every member.
• Rejects symlink/hardlink members that escape the destination (pass allow_symlinks=True to permit links that stay inside it).
• safe_extract_to_temp(archive) extracts into a fresh temp dir and returns it.

Why

TarFile.extractall() has no path sanitization before Python 3.12's data filter, and even on newer versions a lot of code passes untrusted archives without opting in. This library gives you a linter to find those call sites and a safe function to replace them.

License

MIT