mirror of
https://github.com/jbranchaud/til
synced 2026-08-31 16:51:46 +00:00
56 lines
2.5 KiB
Markdown
56 lines
2.5 KiB
Markdown
# Resurface Exceptions Swallowed By Click Under Test
|
|
|
|
The testing utilities provided by
|
|
[Click](https://click.palletsprojects.com/en/stable/) are generally very nice to
|
|
work with. One particular aspect of makes the testing feedback loop almost
|
|
non-existent. Click stashes any actual exceptions that occur during a run of the
|
|
`CliRunner` and instead shows an opaque failed assertion.
|
|
|
|
Here is the primary output of a failing test I'm dealing with for
|
|
[`py-vmt`](https://github.com/jbranchaud/py-vmt):
|
|
|
|
```bash
|
|
tests/src/py_vmt/test_cli.py:61: AssertionError
|
|
======================================== short test summary info ========================================
|
|
FAILED tests/src/py_vmt/test_cli.py::test_start_status_stop_flow[sqlite] - assert "Tracking 'my-project' for 30m (since 10:05AM)" in '• Not tracking\n'
|
|
=========================================== 1 failed in 0.07s ===========================================
|
|
```
|
|
|
|
What I'd rather see when this test fails is something a little closer to the
|
|
root cause of the failure, like this:
|
|
|
|
```bash
|
|
result = cursor.execute(fetch_active_session_sql)
|
|
if result.fetchone() is None:
|
|
return None
|
|
> _id, _active, project_name, start_time, end_time = result.fetchone()
|
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
|
E TypeError: cannot unpack non-iterable NoneType object
|
|
|
|
src/py_vmt/cli.py:46: TypeError
|
|
======================================= short test summary info ========================================
|
|
FAILED tests/src/py_vmt/test_cli.py::test_start_status_stop_flow[sqlite] - TypeError: cannot unpack non-iterable NoneType object
|
|
========================================== 1 failed in 0.11s ===========================================
|
|
```
|
|
|
|
That's more like it. `TypeError: cannot unpack non-iterable NoneType object` and
|
|
a specific pointer to the line where this happened is way more helpful.
|
|
|
|
To get this improvement in my test failure output, I had to create a small
|
|
wrapper around `CliRunner` that re-raises the exception I care about. I add this
|
|
in my `conftest.py` in the same directory as my `test_cli.py` file.
|
|
|
|
```python
|
|
from click.testing import CliRunner
|
|
|
|
class BetterCliRunner(CliRunner):
|
|
def invoke(self, *args, **kwargs):
|
|
result = super().invoke(*args, **kwargs)
|
|
if result.exception and not isinstance(result.exception, SystemExit):
|
|
raise result.exception
|
|
return result
|
|
```
|
|
|
|
Then in `test_cli.py` I import the `BetterCliRunner` class and replace all
|
|
instantiations of `CliRunner()` with `BetterCliRunner()`.
|