Safer tar extraction

safer_tar_extract.py

# SPDX-FileCopyrightText: 2023 geisserml <[email protected]>
# SPDX-License-Identifier: Apache-2.0 OR BSD-3-Clause OR MPL-2.0

# Safer tar extraction (hopefully) preventing CVE-2007-4559 etc.
# Tries to use the most elegant strategy available in the caller's python version (>= 3.6)

__all__ = ["safer_tar_unpack"]

import sys
if sys.version_info >= (3, 11, 4):  # PEP 706
    import shutil
    def safer_tar_unpack(archive_path, dest_dir):
        shutil.unpack_archive(archive_path, dest_dir, format="tar", filter="data")

else:  # workaround
    import tarfile
    from pathlib import Path
    
    def safer_tar_unpack(archive_path, dest_dir):
        dest_dir = Path(dest_dir).resolve()
        with tarfile.open(archive_path) as tar:
            for m in tar.getmembers():
                if not ((m.isfile() or m.isdir()) and dest_dir in (dest_dir/m.name).resolve().parents):
                    raise RuntimeError("Path traversal, symlink or non-file member in tar archive (probably malicious).")
            tar.extractall(dest_dir)

Also note, a known limitation of the above code is that existing files in dir may be overwritten by the archive. But this is expected.

Re Comment 2: I updated the code accordingly, however the latest revision is now untested.