[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — ran pylint/mypy on the two changed source files only: no new messages on changed lines, no mypy errors in them; not the make targets
• make test passes, and a test case covers the change — ran only the modules listed below
• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible — N/A, added centrally after the wave

What is the purpose of your pull request?

• 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 #1165. The PCAP engine passes the input header's thiszone, sigfigs and snaplen to Extractor._open_output, and PCAPIO._dump_header hands them to Header (which already accepts them as make() keywords). test_runtime_engines.py pinned the old _open_output arguments and is updated.

Sweep, every examples/captures/*.pcap + http6.cap, default engine, format='pcap': before 1/17 byte-identical (16 differ only at header bytes 16–19); after 17/17.

Tests: new tests/foundation/test_pcap_output_header_fields_unit.py (in.pcap rewritten to thiszone -3600 / sigfigs 7 / snaplen 1514: whole file, split mode, zero frames). Without the fix: 3 failed; with: 3 passed. Also test_runtime_engines.py 5 passed, test_extraction_pcap_output_unit.py 7 passed, test_extraction.py 22 passed, dumpkit/test_common_unit.py 14 passed, integration/test_files_output_naming.py 4 passed, tests/project 389 passed, 1 skipped.

[JarryShaw]

Cross-review verdict on 77aeb27ec: GOOD TO GO (ran on Sonnet; author Opus)

• Root cause confirmed: on main, PCAPIO._dump_header builds Header from network, byteorder and nanosecond only. I re-read this myself. Header.make already accepts thiszone=0, sigfigs=0 and snaplen=0x40000 (header.py:189-190).
• Sweep: default engine, format='pcap', over all 17 pcap inputs. Byte-identical output was 1/17 before and is 17/17 now. All 16 earlier diffs were header bytes 16–19 (snaplen).
• Split mode and zero-frame output both carry the input header. Big-endian nanosecond input with a negative thiszone round-trips byte-identically.
• No other path breaks. PCAP-NG input still raises FormatError, the non-PCAP dumpers take the new keywords through **kwargs, and traceflow never calls _dump_header.
• Fails without the fix: 3 failed. The edited test_runtime_engines.py pin adds non-default values, so swapping two fields would also fail.
• Follow-ups, not blocking:
  • The input's version fields are still written as 2.4. Every capture in the repo is 2.4.
  • The new test covers only little-endian input. Big-endian was probed by hand but is not pinned.

[github-actions]

Coverage: 88.88% (unit tier, Python 3.14, 77aeb27ec, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18757	1034	2342	845	90.24%
pcapkit/corekit	1878	91	580	22	94.34%
pcapkit/dumpkit	142	0	42	0	100.00%
pcapkit/foundation	2442	102	852	39	94.38%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	15696	188	3970	165	98.17%
pcapkit/toolkit	487	65	144	1	85.42%
pcapkit/utilities	429	4	122	4	98.55%
pcapkit/vendor	4409	2359	1006	158	42.84%

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