fix(const): keep AppType/OptionType.register loud inside a quiet block (#1916)

[JarryShaw]

• 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

• Added a 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

AppType.register and OptionType.register now run _validate_value inside with _CaughtByCaller():, as EnumRegistry.register does (#1845), so a rejection stays loud inside a quiet get block. The vendor templates are edited, and both const modules are regenerated from the crawled data already committed, without a re-crawl.

The new test checks that TCP.register(70000, 'x') and OptionType.register(-1, 'x') raise and log on the pcapkit logger inside _CaughtByCaller(ValueError). Both cases fail without the fix.

Closes #1916

[JarryShaw]

GOOD TO GO at 2d50c6bb0. A Sonnet cross-review checked every _validate_value( call in pcapkit/vendor and pcapkit/const; these four sites are all of them and all are now wrapped, and Command.register goes through the base register. Each const diff mirrors its template diff. Restoring the const files from main makes both new tests fail, and the fix makes them pass. The CRITICAL level matches exceptions.py and #1873's test, and the test uses no stdlib API newer than 3.8. The new test, test_const_enum_no_mint.py, tests/vendor and tests/corekit/test_enum* all pass: 378 passed.

[github-actions]

Coverage: 90.07% (unit tier, Python 3.15, 2d50c6bb0, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	2	20	2	96.77%
pcapkit/const	19597	881	2618	867	92.11%
pcapkit/corekit	2504	70	792	36	96.54%
pcapkit/dumpkit	276	3	100	3	98.40%
pcapkit/foundation	2916	164	1016	34	93.59%
pcapkit/interface	138	7	52	5	93.68%
pcapkit/protocols	17411	224	4580	189	98.10%
pcapkit/toolkit	872	177	298	8	77.18%
pcapkit/utilities	522	2	158	4	99.12%
pcapkit/vendor	4431	2341	1018	156	43.64%

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

[JarryShaw]

Wondering if we can actually rename _CaughtByCaller as CaughtByCaller - it's being a library internal public context manager already.

[JarryShaw]

Yes. _CaughtByCaller came in with #1845 during this beta, and v1.4.1.post2 has no such name, so it can be renamed outright without a re-export. I'll fold the rename into this PR, because its templates and const modules already import it. That covers pcapkit/utilities/exceptions.py, corekit/enum.py, the four vendor/const files here, the tests and the docs. The class also gets a documented entry and an __all__ slot as CaughtByCaller.

[JarryShaw]

ready to go or pending changes?

[JarryShaw]

Pending changes. The _CaughtByCaller → CaughtByCaller rename you asked for is still in progress: a rebase onto current main (which now includes #1935's edits to the same templates), the rename across exceptions.py, enum.py, the templates and const modules, the tests and the docs, plus a documented entry. The GOOD TO GO above applies to 2d50c6bb0 without the rename. I'll post a new verdict once the rename is pushed and re-reviewed.

[JarryShaw]

GOOD TO GO at 985da2dd9, rebased onto 3c67a3890, now including the CaughtByCaller rename. _CaughtByCaller appears nowhere in the tree, and v1.4.1.post2 has neither name, so no re-export is needed. A Sonnet re-review checked the new docstring against BaseError.__init__. Inside the block the named classes skip both the log and the hook, and others stay loud. The state is a ContextVar, which is what makes it per thread and per context. The guard commit survived the rebase intact. Each const diff mirrors its template diff, and exceptions.rst has a single autoclass. The #1916, #1845, builtin-parity and isort tests pass, and merge-tree against main is clean.