fix(corekit): quiet get's fallback over an out-of-range value (#1845)

[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. Narrow runs only, listed below.

• 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

---

• exceptions: add a _CaughtByCaller block, backed by a ContextVar, inside
  which a BaseError of a named class is constructed quiet; BaseError
  consults it alongside quiet=True.
• enum: EnumLookup.get runs _validate_value and cls(key) in a
  _CaughtByCaller(ValueError) block when default is usable, so a generated
  override's loud rejection logs nothing and installs no hook there; with no
  usable default the miss stays loud.
• enum: EnumRegistry.register validates in a block declaring nothing caught,
  so its rejection stays loud even when reached from inside get.
• docs: get's docstring gains a 1.5.0 versionchanged and no longer says the
  value miss stays loud; registry-protocol.rst says what is reported now.
• tests: cover quiet fallback, the loud paths, nesting and per-thread scope.

Closes #1845

Open scope question, asked on #1845: protocol parsers never pass default to get, so this quiet path does not change what parsing logs. Whether a no-default get should also raise quietly is pending the owner.

[JarryShaw]

Pushed 2da6b1be7 per the ruling on #1845. get is now quiet for an out-of-range value with or without a default. An uncaught miss still prints the interpreter's own traceback, with no CRITICAL log. register stays loud. The tests that pinned the no-default get as loud are flipped, including the #923 one, and a loud control is added in its place. The review is extended to this commit.

[JarryShaw]

Verdict: GOOD TO GO at 2da6b1be7. A Sonnet cross-review covered the final head and the earlier d019b8368, and found no defects.

• get is quiet. ToSPrecedence.get(8) raises EnumValueError with no log record and leaves the excepthooks and sys.tracebacklimit alone. I reproduced this myself. get(8, 0) returns the default quietly, and get_all behaves the same way.
• An uncaught miss prints the interpreter's own traceback (exit 1, no [CRITICAL] line), with or without forced colour.
• register stays loud, including when it is reached from inside a quiet get. Removing the _CaughtByCaller() block fails three new tests. A non-ValueError from the hook still propagates loudly.
• Fail-before. Against d019b8368's enum.py, the new tests fail for the right reasons. The fix(corekit,utilities): raise pcapkit exceptions from EnumLookup.get, following stdlib Enum's shape #923 control fails if BaseError stops logging, so the quiet tests cannot pass vacuously.
• Isolation. The quiet state is isolated per thread and per asyncio task, and swapping the ContextVar for a global fails the thread test.
• Tests: tests/project gave 988 passed; tests/const, corekit, utilities, vendor and interface pass.
• Unverified. End-to-end parsing of real captures, because the review tree had no generated samples.
• Notes, not blockers:
  • The generated AppType.register and OptionType.register call _validate_value without the loud block. Inside a quiet get they would log nothing, but nothing reaches them that way today.
  • A non-str get now costs about 900 ns instead of about 405 ns.

[github-actions]

Coverage: 90.05% (unit tier, Python 3.15, 2da6b1be7, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	2	20	2	96.77%
pcapkit/const	19595	881	2618	867	92.11%
pcapkit/corekit	2464	69	780	35	96.55%
pcapkit/dumpkit	276	3	100	3	98.40%
pcapkit/foundation	2915	164	1016	34	93.59%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17406	224	4580	189	98.10%
pcapkit/toolkit	872	177	298	8	77.18%
pcapkit/utilities	522	2	158	4	99.12%
pcapkit/vendor	4407	2342	1010	157	43.27%

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