Type annotations used for functions strategy

[lengau]

When you accept a function as a parameter to another function, if you're using mypy and typing you may well have that method described already using Callable etc. In this case, it's redundant to create a lambda function for the like attribute of hypothesis.strategies.functions. It would be nice to be able to reuse these objects in hypothesis.

[Zac-HD]

It's not quite clear to me what you're asking for. If you mean "st.functions(like=Callable[[int, int], int]) should work", you're better off using st.from_type().

If you mean "st.functions() ought to infer a strategy for returns from the return-type annotation on the like function", that sounds pretty good to me! I'd be happy to help out if you'd like to open a PR: hypothesis.internal.compat.get_type_hints(func).get("return", type(None)) and st.from_type() should get you started 😄

[akash-suresh]

Hey @Zac-HD ! I would like to pick this up. But I would need some guidance on how to get this done. 😄

[Zac-HD]

Sure! I'll keep the tips a little vague, because figuring out how to do it is part of the fun. Getting stuck isn't fun for long though, so if you're getting frustrated just push what you have up and ask me for more detailed advice.

1. Find the definition of the functions() strategy.
2. We're going to change the default value of the returns argument to None (instead of st.none()).
3. If a strategy is passed, same logic as now.
4. If returns is None, we want to get the return-type annotation of the like argument using our get_type_hints function.
5. Use st.from_type() to get the strategy, defaulting to st.none() if there is no return type hint (again, the current behaviour)
6. Write some tests in tests/cover/test_functions.py to check that it's working.
7. Write release notes in a new file hypothesis-python/RELEASE.rst for a minor release. Check some of the other PRs for examples of this.
8. Add your name to the list of contributors in CONTRIBUTING.rst

and that's it!