fix(pcap): Header.pack() writes the same magic for every endian/nanosecond setting; _make_data returns magic_number

[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — Makefile flags run on header.py only: isort clean, mypy 0 errors in the file, pylint only the existing W0201 on _byte/_nsec
• make test passes, and a test case covers the change — the modules listed below, not the full suite
• Added a changelog entry — N/A, added centrally after the wave

What is the purpose of your pull request?

• fix — corrects a defect

---

Description of your pull request and other information

Closes #1125. Header._make_data returned magic_number, which make() does not accept. from_data dropped it (with an UnknownFieldWarning) and rebuilt every header as the host default, d4c3b2a1. It now returns byteorder and nanosecond. Calling pack() with no arguments rebuilds from make() defaults for every protocol (Ethernet(src=...).pack() does the same), so I left that alone.

Probe: Header.from_data(Header(**kw).info).data == Header(**kw).data, with four byteorder×nanosecond cases plus the lilendian/bigendian flags. Before: only the d4c3b2a1 cases match, and every call warns. After: all 8 match with no warnings, and big-endian packs a1b2c3d4 00020004.

The new tests/protocols/misc/pcap/test_header_make_data_magic_unit.py passes 2 with the fix and fails 5 (incl. subtests) with it reverted. I also deleted the recorded-mismatch test from test_construction_keyword_check_unit.py: its only entry was Header's. And test_header_frame_unit.py's stub MagicNumber now carries the byteorder/nanosecond fields the real one has.

Results: new module 2 passed · keyword_check 21 passed · header_magic_protos 3 passed · header_frame 6 passed · option_roundtrip 6 passed · tests/project 385 passed, 1 skipped.

[JarryShaw]

Cross-review verdict on ae5fffc90: GOOD TO GO (ran on Sonnet; author Opus)

• The issue's part 1 is a misreading. ProtocolBase.pack() calls make(**kwargs) and ignores the instance's state. A bare pack() therefore rebuilds from make() defaults for every protocol, not just Header. Ethernet(src=…).pack() gives zeros too. This is an undocumented behaviour affecting all protocols, not a Header defect, so I have noted it for the docs: #719-style prose sweep after the #1090–#1102 defect wave #1103 docs sweep.
• The real fix: _make_data now returns byteorder and nanosecond instead of the magic_number key that make() drops. All 8 constructions now round-trip with no warnings. On main, 3 of 8 do, and every call warns.
• Tests:
  • with header.py reverted, the new module fails 5;
  • the deleted recorded-mismatch test has no references left;
  • test_construction_keyword_check_unit (21) and test_header_frame_unit (6) pass.

The single-flag cases pass only because both sides ignore a lone flag. That is #1144.

[github-actions]

Coverage: 88.75% (unit tier, Python 3.14, ae5fffc90, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18757	1035	2342	846	90.23%
pcapkit/corekit	1873	91	578	22	94.33%
pcapkit/dumpkit	136	0	40	0	100.00%
pcapkit/foundation	2425	145	842	34	92.56%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	15659	188	3944	163	98.18%
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.