Consolidate CLI testing helpers

[sirosen]

Move common helpers into conftest, so that behaviors can be easily shared.

This change also adds an assert_exit_code=0 kwarg to the test runner invocations, so that running CLI tests always asserts exit codes unless we explicitly opt-out. This prevents a common mistake when tests are added for a CLI tool, in which the exit code is not checked.

---

I write slightly different versions of this same testing tool in each Python CLI that I maintain.
Each one comes out slightly different, and in this case there ws good prior art to build upon. I like this iteration very well: it's short, simple, and effective.

---

📚 Documentation preview 📚: https://dependency-groups--51.org.readthedocs.build/en/51/

[henryiii]

Nice! You can simplify that return value:

diff --git a/tests/conftest.py b/tests/conftest.py
index a138f13..f1f7fd1 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1,6 +1,7 @@
 from __future__ import annotations

 import dataclasses
+import functools
 import typing as t

 import pytest
@@ -44,7 +45,4 @@ class CliRunner:

 @pytest.fixture
 def runner_factory(capsys: pytest.CaptureFixture[str]):
-    def bind(entry_point: CliEntryPoint, /) -> CliRunner:
-        return CliRunner(entry_point, capsys)
-
-    return bind
+    return functools.partial(CliRunner, capsys=capsys)

If you want, I can follow up and add types to the tests; I thought it would be easy (-> missing on runner_factory), so I asked Claude to add them locally, but I see from the diff it's pretty involved as they are missing in a few other places.

[sirosen]

Oh, nice partial usage! I'll go ahead and apply.

Let's save more type coverage for the tests for later? It can get somewhat involved. The last time I worked on this for a package, I found that it's sometimes better to rephrase a fixture entirely rather than trying to describe what's there. (Stuff like protocols with __call__ vs adding a small class.)