Depends on the actual situation, but just from looking at it I would guess that named functions instead of nebulous unrelated nested branches would be a start.
Edit: it’s also just kind of a code smell that suggests the overall structure is not well thought out, with that particular collection of tests. Could be fine in that respect in context tho
As far as our manual testing goes, there was no bug in it. But also there were no tests for it. Writing tests for it is horrible. You need to prepare test for each switch/case branch and it adds a lot of mocking, faking data, setting up variables...
Edit: it’s also just kind of a code smell that suggests the overall structure is not well thought out, with that particular collection of tests. Could be fine in that respect in context tho