MNT: add handling of mpod questionnaire

[cristinasewell]

Description

Add support to create mpod channels from the questionnaire entries.

Motivation and Context

This change goes with the additions from this PR: pcdshub/pcdsdevices#690 . This PR was recently merged, and I don't know if this one should get merged too or not yet, but I wanted to open it so I don't forget about it.

How Has This Been Tested?

Only locally.

Where Has This Been Documented?

N/A

[cristinasewell]

Anybody knows how to 'cleanup' or 'close' in a way a mock when using with patch?
It seems like the first patch is leaking over to the second one, even if I make a new test function.

I can avoid using with and use the monkeypatch or the mocker.patch.object instead and I think it would work fine, but I did not want to have to import pcdsdevices module in this file 😀
Or I can just leave only one test with either Low Voltage or High Voltage

[klauer]

I like to use a pattern similar to:

@pytest.mark.fixture(scope='function')
def my_modified_obj(monkeypatch):
   obj = ...
   monkeypatch.setattr(obj, 'attr', 'value')
   yield obj
   # any cleanup if necessary, the `yield` above makes this fixture a context manager effectively

def test_mytest(my_modified_obj):
    assert my_modified_obj.attr == 'value'

It keeps the monkeypatch scoped to a specific test and is pretty explicit about it.

That said, the with block you use should have reset the original attribute value - not sure why it didn't, or if maybe something else is going on there.

[cristinasewell]

Thank you Ken, yeah I think something else is going on, only if I test them separately it works though.
I am not calling this function that I need to mock (get_max_voltage) anywhere directly here - I probably have to mock something else instead to be able to test both option.
I think it is related to mockqsbackend having a module scope.... so it just mocks my function once per module with the first mock I define - maybe not....

[cristinasewell]

Coming back to this guy, I tested it with some dummy PVs and no mocking get_max_voltage and it seems like it is working, which is good:

'led': MPODChannelLV(csewell:mpod:l:CH:001, name=led), 'blower': MPODChannelHV(csewell:mpod:h:CH:001, name=blower)}
MPODChannelLV(csewell:mpod:l:CH:001, name=led)
MPODChannelHV(csewell:mpod:h:CH:001, name=blower)

caget csewell:mpod:h:CH:001:GetMaxVoltage
csewell:mpod:h:CH:001:GetMaxVoltage 51

caget csewell:mpod:l:CH:001:GetMaxVoltage
csewell:mpod:l:CH:001:GetMaxVoltage 49

Later I realized that I had to use use_cash=False in order for it to work fine for testing:

d = load_devices(*c.all_devices, pprint=False, use_cache=False).__dict__

I was getting here:
https://github.com/cristinasewell/happi/blob/d5c24dc8c4416d852f98d4e1c0b3dc62d9a4c6a5/happi/loader.py#L122

if I did not set use_cash=False , and the second mpod was created the same as the first one (E.g.: if the first one was HV the second one was HV too ...

I am not sure if I understand why was this happening exactly ... it is probably a combination of how the mockqsbackend is setup and the fact that I am looking at the voltage to create the mpod channel ....