fix(ospf): rebuild cryptographic authentication from its dict form (#1464)

[JarryShaw]

• Searched for similar pull requests

• Followed the coding style (make pylint, make mypy, make isort): ran mypy and isort on ospf.py only; no new findings, pylint not run

• make test passes, and a test case covers the change: ran the modules listed below

• Changelog entry: N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657

• fix: corrects a defect

---

Closes #1464

to_dict() flattens the AuType 2 auth field (a CryptographicAuthentication info) into a plain dict. from_dict() cannot restore it, because the field is declared bytes | CryptographicAuthentication. _make_encrypt_auth now accepts that dict. A missing reserved defaults to zero, and any other missing key raises ProtocolError.

AuType	object form	dict form, before → after
0, 1, 7, 65535	exact	exact → exact
2 (incl. non-zero reserved)	exact	ProtocolError: invalid type for auth_data → exact

Sweep: the dict-form rebuild of every edge-harness case and every option-generator case is exact. Only OSPF has a top-level nested info that from_dict leaves as a dict. HIP HostIDParameter.hi and Locator.locator are typed the same way, but they sit inside the parameter list, which to_dict does not flatten, so from_data never hands them a dict.

Tests: the new module passes 6 tests and 16 subtests; without the fix, 3 tests and 3 subtests fail. Also passing: the protocols --exclude protocols/internet leg (1158 tests, OK), test_capture_roundtrip_runtime.py (4 passed), the four *_edge_roundtrip_unit.py modules, and tests/project (419 passed, 1 skipped).

[JarryShaw]

Verdict: GOOD TO GO at 7fc3b9bde (Sonnet cross-review; the author ran on Opus).

• Dict-form rebuild is exact for all 162 OSPF packets tried, versus 144 failing on main. They cover AuType 2 with 4 reserved, 3 key_id, 4 len and 3 seq values, plus AuTypes 0, 1, 3, 7, 255 and 65535. The same holds for six Ethernet/IPv4/OSPF frames, in both info and dict form. I re-ran the issue repro myself on this head, and it is exact.
• The reserved default (0000) only affects hand-built dicts. to_dict() always emits the key. A missing key_id, len or seq raises an in-library ProtocolError that names the missing key.
• The author's HIP claim holds. HIP.from_data(info.to_dict()) is exact for HOST_ID and LOCATOR_SET.
• The tests fail without the fix. With ospf.py from main, 3 subtests fail, all AuType 2.
• Legs pass: rest 1158 OK, all four edge modules, capture roundtrip at 20501 subtests, and tests/project.

Nit, no change needed. A wrong value type in the dict, such as key_id='x', raises a raw TypeError from struct. The object form behaves the same way.

[github-actions]

Coverage: 89.56% (unit tier, Python 3.14, 7fc3b9bde, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18795	957	2356	870	91.00%
pcapkit/corekit	2043	89	636	25	94.77%
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	16900	222	4564	187	98.06%
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.