问题
I have a test bench setup where I make heavy use of fixtures to provide resources and "items to test" to the tests I have defined. This is great, it allows my test functions to contain as little code as possible.
However I find myself fighting with pytest quite a lot in the area of finalization.
In pytest, fixture finalizers are conveniently described by making a generator of the form:
@pytest.fixture
def a_fixture():
#some setup code
resource = setup_resource()
yield resource
#finalizer code here
close_resource(resource)
Frequently, I find myself wishing that I could do this:
@pytest.fixture
def a_fixture():
#some setup code
resource = setup_resource()
try:
yield resource
except ValueError as e:
#handle errors in finalization code
handle_error(resource)
#finalizer code here
close_resource()
Some example usecases for this:
- Many of my tests are testing functionality on a piece of hardware connected to a host PC. When an error occurs, I'd like the finalization code to collect more information from the device and log it.
- In the event of a specific kind of error, I'd like to alter the finalization code that executes. For example, the tests have resulted in the device under test crashing and attempting to run the finalization code will just result in wasted time.
I raised a ticket with the pytest project about this, they informed me that this decision is by design and the best option for me would be to create a custom report hook.
I find the concept of fixtures to be extremely useful and powerful. I also find the inability of finalizer code to be informed of errors to be very limiting. A custom report hook doesn't work for me because the finalizer code is too closely linked with its source fixture.
So now to the question: If not propagating exceptions into fixtures is by design, how do you provide conditional resource finalization?
来源:https://stackoverflow.com/questions/62424582/pytest-fixture-finalization-in-the-presence-of-errors