fix(protocols): PCAPNG.register restores the shipped descriptor; a zero snaplen drops SPB padding (#1504, #1558)

[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) — Makefile flags on the changed modules, no new findings

• make test passes, and a test case covers the change — narrow selections only, not the full suite

• 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

• fix(protocols): restoring a protocol registration puts back the class, not the shipped ModuleDescriptor #1504: PCAPNG.register (pcapkit/protocols/misc/pcapng.py:1021-1035) resolved a descriptor before storing it. It now shares ProtocolBase._next_layer_entry, as the other six registrars do since fix(protocols): restoring a registration puts back the shipped ModuleDescriptor (#1504) #1533. The Gap(1504) row is deleted, and PCAPNG joins the fix(protocols): restoring a protocol registration puts back the class, not the shipped ModuleDescriptor #1504 per-site tests.
• fix(engines): a Simple Packet Block keeps its padding when the interface snaplen is 0 #1558: the engine passed interface 0's snaplen of 0 through as __packet__['snaplen'] (pcapkit/foundation/engines/pcapng.py:209, :413-414). _spb_data_length then read it as a limit of no octets and took the whole block body, padding included. _get_snaplen now hands over no limit, and PCAPNG.unpack reads a zero snaplen handed in as no limit too. The new tests check captured_len/packet = 57 for snaplen 0 and 65535, and a byte-exact rebuild from both info and info.to_dict().
• fix(protocols): the default engine does not dissect LINKTYPE_RAW (101) in PCAP and PCAP-NG #1502 is not in this PR. The branch name keeps it so it can be added here later. The PCAP-NG dispatch is ready, but landing it changes the dpkt and scapy rows owned by fix(engines): dpkt float timestamps, Simple Packet Blocks and raw-IP tracing (#1501, #1520, #1548) #1550 and fix(engines): take a scapy packet's link type and clock from its interface (#1503, #1517, #1549) #1563, and its dpkt half needs pcapkit/foundation/engines/dpkt.py.

Closes #1504
Closes #1558

[JarryShaw]

Verdict: GOOD TO GO at 0f4652b1f. Cross-reviewed on Sonnet; the author ran Opus. It merges cleanly into current main (cc084357f).

• fix(engines): a Simple Packet Block keeps its padding when the interface snaplen is 0 #1558: 248 cases across both byte orders, covering original lengths 1–100 and snaplens 0 to 0xFFFFFFFF, plus multi-SPB and second-section cases. Every one gives captured_len = min(orig, snaplen), with snaplen 0 meaning unlimited, and re-packs byte-identical, padding included, through from_data(info), from_data(info.to_dict()) and make. On main, 54 of the 248 fail. I checked one myself: an SPB of 57 octets under snaplen 0 reads 57 here and 60 on main.
• The "no limit" sentinel: 0xFFFF_FFFF_FFFF_FFFF stays internal and never appears in repr/to_dict(). The IDB still reads snaplen 0. A direct PCAPNG(...) given snaplen 0 gives the same result.
• fix(protocols): restoring a protocol registration puts back the class, not the shipped ModuleDescriptor #1504: for all 3 shipped codes, re-registering the shipped descriptor puts back the identical object, through register_pcapng and register_linktype as well. _next_layer_entry still raises the same three RegistryErrors.
• Fail-before:
  • reverting fix(protocols): restoring a protocol registration puts back the class, not the shipped ModuleDescriptor #1504: 10 failed;
  • reverting both fix(engines): a Simple Packet Block keeps its padding when the interface snaplen is 0 #1558 halves: 4 failed;
  • reverting only the protocol half: 2 failed;
  • reverting only the engine half: 1 failed.
• Tests: registry (43), all 22 PCAP-NG modules (226), the agreement module (identical to main) and the capture round trip (20,492 subtests) all pass.

Not this PR's: a truncated SPB (snaplen 10, original length 57) emits SchemaWarning: packet length < 0: -2 and -4 while parsing. I measured it identically on main, and I'm filing it separately.

[github-actions]

Coverage: 89.64% (unit tier, Python 3.14, 0f4652b1f, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18858	957	2366	871	91.03%
pcapkit/corekit	2240	67	722	34	96.32%
pcapkit/dumpkit	258	2	90	2	98.85%
pcapkit/foundation	2766	129	980	47	94.23%
pcapkit/interface	108	7	36	5	91.67%
pcapkit/protocols	16990	230	4542	191	98.02%
pcapkit/toolkit	697	100	232	7	84.61%
pcapkit/utilities	431	4	124	4	98.56%
pcapkit/vendor	4409	2342	1006	157	43.25%

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