test(integration): skip upstream-pinned PCAP-NG counts on a stand-in fixture (#1478)

[JarryShaw]

Please follow the guide below

• You will be asked some questions, please read them carefully and answer honestly

• Put an x into all the boxes [ ] relevant to your pull request (like that [x])

• Use Preview tab to see how your pull request will actually look like

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort)

• make test passes, and a test case covers the change

• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible — N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

What is the purpose of your pull request?

Tick the commit type your subject line carries.

• fix — corrects a defect
• feat — adds a feature
• perf — changes performance, not behaviour
• refactor — changes neither behaviour nor performance
• test — tests only
• docs — documentation only
• ci — workflows or build tooling
• release — bumps the version or rolls up a distribution
• chore — anything else

---

Description of your pull request and other information

Closes #1478. When a download fails, examples/generators/pcapng.py writes a synthesised stand-in under the upstream name, so the counts pinned to upstream went red on a network blip (job 113845878658).

Design. The generator already pins each download's SHA-256 in FIXTURES. That pinned digest is the marker: tests/_support.stand_in_reason() hashes the capture on disk and compares. A sidecar file could get out of step with the bytes it describes; this check can't. skip_if_stand_in() skips only the assertions pinned to upstream, so they still run whenever the real file is present. No stand-in numbers are hard-coded.

Scope. Only pcapng.py downloads. Of its two downloads:

• many_interfaces.pcapng: guarded at the three failing sites (:70, :88, :96 on main).
• dhcp_big_endian.pcapng: its stand-in matches upstream's 4-frame inventory, so the integration test still asserts it. The SHB option layout check in test_pcapng_unit.py is pinned to upstream ([9, 12, 15, 7, 0] against the stand-in's [37, 9, 22, 29, 0]), so that check is guarded too.

Proof (stand-ins written by generate() with _download patched to raise URLError):

• main's tests on the stand-ins: 4 failed. These are the three from the issue plus the unit check.
• This branch on the stand-ins: 138 passed, 5 skipped. Each skip gives a reason that names the digest.
• This branch on the real files: the integration test gives 6 passed, 1 skipped (the existing dictdumper skip), so the 64-frame assertions run.
• New StandInReasonTests cover the helpers, including a simulated failed download through the real generator.
• tests/project: 419 passed, 1 skipped.

[JarryShaw]

Verdict: GOOD TO GO at 34bd7b13b. Cross-reviewed on Sonnet; the author ran on Opus.

• Real fixtures still assert the upstream counts. Re-run by me: the downloaded many_interfaces.pcapng matches the pinned digest. The integration file plus tests/test_support_helpers.py give 43 passed, 1 skipped (the existing dictdumper skip). In a scratch copy, changing the expected 64 to 65 fails all three guarded tests.
• The stand-in path. With the download forced to fail, the integration, test_pcapng_unit and test_pcapng_regression files give 112 passed and 5 skipped, with no failures. Only the upstream-pinned assertions are skipped; dhcp_big_endian still checks 4 frames.
• The skip cannot hide a parser regression. It depends only on the file's SHA-256 against the generator's pin. The digest helper leaves sys.path, os.environ and logging unchanged, and imports nothing from pcapkit.
• No pinned count is left unguarded. A grep of tests/ for both fixtures found five count assertions, all guarded. tests/project and the tier/docstring guards: 572 passed.

Not audited, out of scope: other downloaded fixtures, such as http6.cap.

[JarryShaw]

Verdict carried to 2167d38c5: GOOD TO GO. This is a pure rebase onto main at bbfd4bf64, after #1475 merged. range-diff shows the one commit as =, and the stable patch-id is unchanged (3dbf97b98e02).

[github-actions]

Coverage: 89.55% (unit tier, Python 3.14, 2167d38c5, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18795	958	2356	871	90.99%
pcapkit/corekit	2107	93	672	27	94.60%
pcapkit/dumpkit	242	2	88	2	98.79%
pcapkit/foundation	2648	98	930	48	94.80%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	17049	240	4610	193	97.95%
pcapkit/toolkit	539	74	168	3	84.58%
pcapkit/utilities	429	4	122	4	98.55%
pcapkit/vendor	4409	2342	1006	157	43.25%

Per-file detail: the coverage-html artifact of this run.