chore: mark vermin's guarded-import false positives, close the engine handle

[JarryShaw]

Two small fixes and one measurement worth recording.

close_extractor leaked the engine handle

tests/_support.close_extractor closed _ifile only, while both Extractor._cleanup and Extractor.__exit__ close _ifile then _exeng.

That matters because Engine.close() is a no-op on the base class but not on the subclasses — pcap_ct and pypcap close a live pcap.pcap handle, pyshark closes a temp file. So an extractor abandoned mid-file in a test teardown leaked a real OS handle.

Now closes both, in the library's own order, each guarded separately and inside a try/finally so a failing stream close cannot strand the engine. getattr probing keeps it tolerant of an extractor that died before _exeng was assigned — which is the state teardown most often sees.

No test depended on the leak: all ~30 call sites use it purely as teardown and never touch the extractor afterwards. New tests/test_support_helpers.py pins the contract, following the tests/test_tier_guard.py precedent for covering test-support code.

# novermin on three guarded imports

Vermin reported a 3.11 minimum from enum.StrEnum and enum.show_flag_values, and 3.10 from typing.TypeAlias — all three in the else: branch of a sys.version_info guard, so they never execute on an older interpreter. False positives, and # novermin is Vermin's documented escape hatch for exactly this shape.

Verified the combined # type: ignore[attr-defined] # novermin still works: zero mypy errors in the file, and mypy does report unused ignores elsewhere in the same run, so the check has teeth.

targets = 3.6 left alone — and what that costs

Per your decision to keep the 3.6 claim. Recording the measurement so it doesn't have to be rediscovered:

After these three markers the report lands on 3.9, and what remains is real code, not false positives — 65 named-expression (walrus) sites across 57 files, unguarded and at module level, plus a positional-only / parameter in foundation/engines/engine.py. Walrus is 3.8+, so no amount of marking reaches 3.6.

Vermin isn't wired into any workflow, so nothing is red either way. Flagging it only because the gap between the declared floor and the actual one is now quantified.

Also measured, not changed

The packaging question, settled by building both artefacts: the sdist carries all 91 test modules, the wheel carries none. That is the right split — the suite could not run from an installed package anyway, since the generated captures aren't shipped and tests/_tiers.py resolves paths from a repository root. MANIFEST.in's prune test (singular) does not match tests, which is why the sdist includes them; changing it to prune tests would remove the suite from the sdist, the opposite of what was documented. Left as a deliberate decision, with the prose corrected on #408 instead.

Verification

tests/test_support_helpers.py + tests/utilities: 96 passed, 100 subtests.

[Copilot]

🟡 Changes recommended

_close_quietly can currently raise from getattr(target, "close", ...), which is a teardown-safety regression compared to the previous hasattr(...) behavior and should be guarded.

Get a fresh assessment by requesting another Copilot review.

Pull request overview

This PR tightens test teardown correctness by ensuring tests._support.close_extractor releases both the extractor’s input stream and its engine, and it reduces Vermin noise by marking version-guarded imports as false positives.

Changes:

• Extend tests._support.close_extractor to close _ifile and _exeng in library order, with independent failure isolation.
• Add a dedicated unit test module to pin teardown-helper behavior and prevent regressions.
• Annotate pcapkit.utilities.compat guarded imports with # novermin to avoid static false positives.

File summaries

File	Description
tests/test_support_helpers.py	Adds contract tests covering teardown helper tolerance and close-order/independence.
tests/_support.py	Updates teardown helper to close both stream and engine with guarding.
pcapkit/utilities/compat.py	Marks guarded imports as Vermin false positives using # novermin.

Review details

• Files reviewed: 3/3 changed files
• Comments generated: 2
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.