Recipe: bounded ZIP manifest without extracting to disk

Intended use and acceptance criteria

Before extracting an offline learning pack, inspect its members and produce a stable manifest of name, uncompressed byte count and SHA-256. Reject duplicate member names, absolute or parent paths and archive content above a small predefined total. The fixture is created entirely in memory and no untrusted file is written onto disk.

Complete runnable example

Save this as 09_Archive_manifest.py and execute with Python 3. The fixture is disposable and has no network dependency.

import hashlib, io, zipfile
def manifest(zip_bytes, total_limit=2048):
    entries, seen, total = [], set(), 0
    with zipfile.ZipFile(io.BytesIO(zip_bytes)) as archive:
        for entry in archive.infolist():
            name = entry.filename
            if (name in seen or name.startswith("/") or "\\" in name or
                    any(x in ("", ".", "..") for x in name.split("/"))):
                raise ValueError("unsafe or duplicate member")
            seen.add(name)
            total += entry.file_size
            if total > total_limit or entry.file_size > total_limit:
                raise ValueError("archive content over limit")
            data = archive.read(entry)
            entries.append((name,len(data),hashlib.sha256(data).hexdigest()))
    return sorted(entries)
buffer = io.BytesIO()
with zipfile.ZipFile(buffer,"w",zipfile.ZIP_STORED) as out:
    out.writestr("docs/a.txt",b"alpha")
    out.writestr("assets/icon.txt",b"icon")
items = manifest(buffer.getvalue())
assert [x[0] for x in items] == ["assets/icon.txt","docs/a.txt"]
assert items[1][1] == 5
try: manifest(buffer.getvalue(),3)
except ValueError: pass
else: raise AssertionError("archive limit ignored")
print("PASS bounded ZIP manifest")

Guarantees and limitations

ZipInfo.file_size provides declared uncompressed size. The example also checks actual bytes read and hashes each file, but it deliberately remains limited to tiny archives. A malicious ZIP may include symlinks encoded in metadata, unsupported compression, deceptive encoding or extreme compression ratios; safe extraction needs additional checks and strict streaming decompression caps. Member names should be validated before creating any output file.

Adaptation lab

Add a third valid member and check deterministic manifest ordering. Insert a ../outside.txt member and prove the tool rejects it. Test repeated member names, a directory entry, an archive above the total limit and an invalid ZIP. Distinguish inspecting ZIP metadata from proving extracted files are safe to open or execute.

Troubleshooting and release check

If a member is rejected, display its escaped name, not a normalized guess at its intended path. A ZIP that looks small on disk may expand dramatically; cap total uncompressed bytes and member counts. Do not call extractall on untrusted input without a separate extraction security review.