test: run the suite in CI and cover the 0.5.0 fixes - #280
Merged
Conversation
The repository had pytest, coverage and Ruff configured in pyproject.toml
but nothing that ran them, no pytest dependency, and a root `test.py` whose
entire body was `print("fake test")`. The one real test, added with #269,
has never been executed anywhere.
Add pytest to the dev group, narrow `testpaths` to `tests`, drop the
placeholder, and run the suite on every pull request and on pushes to master
across Python 3.11 to 3.14. `uv sync --locked` doubles as the lockfile check
the pull request workflows otherwise lack.
Ruff is deliberately not gated. It currently reports 431 findings on the
existing code, so a blocking job would fail on arrival. Cleaning that up is
its own piece of work.
The tests themselves cover the bugs fixed for 0.5.0, each written so that it
fails again if the fix is reverted:
test_utils_datetime both API timestamp shapes, offsets other than Z,
rejection of ambiguous values, UTC normalization
of the --start-date/--end-date option type
test_models_dates the date filter across both shapes and naive and
aware bounds, items carrying no usable date at all,
is_published including the AudioPart parent
fallback
test_download_queue the deadlock: failing jobs must not strand the
queue, the abort must stop queued work while
letting running downloads finish, and a run that
saw failures must raise
test_download_cli the command itself has to raise on a failed job;
testing raise_for_errors() alone would stay green if
the call were dropped from the command
test_exceptions ItemNotPublished must not fail while building its
own message
mkb79
added a commit
that referenced
this pull request
Aug 7, 2026
Ruff has been configured in pyproject.toml for a while but never ran anywhere. Turning it on as-is would fail immediately: 442 findings in 29 files, and 29 files the formatter has never touched. Both jobs run over the whole repository rather than a list of directories, so `pyi_entrypoint.py`, which the release build uses, `utils/` and the docs config are covered too, and a new top-level file cannot slip past. Rather than switch the rules off globally, pin what each file already breaks. A global ignore list would be an amnesty: a new violation of the same rule would then pass anywhere, including in files written from scratch. With `per-file-ignores` a new file, and a rule a file does not already break, both still fail. The one thing this does not catch is another violation of a rule already listed for that same file — `per-file-ignores` keys on file and rule, not on individual findings. Ruff has no native baseline for that yet. `ruff format` is gated the same way, through `[tool.ruff.format].exclude`. That leaves the 25 legacy files alone while every new file is checked from the start, and avoids a 1400-line reformat that would collide with the eight external pull requests still open. The three test files added in #280 were not in that state for long, so they are formatted here instead of listed. Ruff itself is pinned. The enabled rule families pull in new stable rules on an upgrade, which would turn the gate red without anyone touching the code. Both run as separate jobs so a failure names which one, without the Python matrix the tests need. Shrinking this is the point: fix a rule in a file, drop its code from the list, or run `ruff format` on one file and remove it from the exclude list. 197 of the findings are safe autofixes, which is the obvious next step.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The repository had pytest, coverage and Ruff configured in
pyproject.tomlbut nothing that ran them, no pytest dependency, and a roottest.pywhose entire body wasprint("fake test"). The one real test —tests/test_cmd_manage.py, contributed with #269 — has never been executed anywhere.Infrastructure
pytestadded to thedevgroup,testpathsnarrowed totests, the placeholder deletedtests.yml: runs on pull requests and pushes to master, across Python 3.11 – 3.14, withpermissions: contents: read, aconcurrencygroup that cancels superseded runs, andfail-fast: falseso one version's failure does not hide the othersuv sync --lockeddoubles as the lockfile check the pull request workflows otherwise lack — apyproject.tomlchange without a matchinguv.locknow fails hereRuff is deliberately not gated. It reports 431 findings on the existing code, so a blocking job would fail on arrival. Cleaning that up is its own piece of work.
tests/itself is Ruff-clean.Tests
56 tests, 0.3s, covering the bugs fixed for 0.5.0:
test_utils_datetimeZ, rejection of ambiguous values, UTC normalization of the--start-date/--end-datetypetest_models_datesis_publishedincluding theAudioPartparent fallbacktest_download_queuetest_download_clitest_exceptionsItemNotPublishedmust not fail while building its own messageWhy these tests are worth having
Green tests prove nothing on their own, so each fix was reverted in turn to check the suite goes red:
parse_api_datetimeback to a single format (#264)library_statusguard removed (#268)AudioPartpublication fallback removed (#268)ItemNotPublishedcatch narrowed (#268)run.raise_for_errors()call deleted (#256)consume()restored (#235)That last row is the point of
test_download_cli: testingraise_for_errors()in isolation would have stayed green if the command stopped calling it, so the wiring is exercised through the realcliwith only the network edges mocked.Not covered
The download paths that only log a failure rather than raising — an unknown ASIN, a cover without a URL, a non-success
Statusfrom the downloader. Those are the remaining half of #256 and need the fix before they can be tested.