Accessing CWB corpora in backend tests

[janiemi]

It has been decided to use pytest for writing tests for the Korp backend.

I have no experience in using pytest for testing a Flask/WSGI application, but I’ll see if I have time to investigate how to write rudimentary tests, maybe first for the modifications I’m making myself.

However, at least I find that it complicates endpoint-level backend functionality testing that the tests should be able to access some CWB corpora. I suppose the tests can assume CWB to be available but I think they should use dedicated corpus data to be self-contained. Maybe the test setup should create the corpora and place them in a dedicated place. Would it be easiest to have some pre-encoded corpus data files that would just be copied to an appropriate place? Or should something perhaps be mocked to make testing easier (or possible)?

I intend to continue this discussion myself once I have taken a look at the options in practice, but I’d also appreciate any kind of help or hints on the issue.