We frequently use pydantic at Pluralsight to validate data at the edge of a well typed domain. I've been trying to sell my coworkers on using hypothesis for testing, and thought it might go easier if they could generate test data from existing pydantic schemas. I found that it was almost trivial for data classes, but BaseModel
subclasses (which are unfortunately much more common in our code) don't play as nicely, and deeply nested schemas can get you into trouble. If anyone has any advice on how to get around the errors that come from the last test, I'd be super greatful.
- pydantic
- hypothesis
- pytest