How to get hypothesmith to generate a FunctionDef with arguments?

[julian-goettingen]

I have been experimenting with this library for a bit for purposes of testing refactoring tools. I can generate lots of extremely weird identifiers for all kinds of things, but the structure of everything generated seems fairly simple - too simple.

For example, I am trying to generate function signatures. I imagine many possibilities with the arguments - they can have annotations, positional-only-args, the **kwargs-keyword, default-arguments, etc etc.

Here is my code to generate function-definitions that have at least something in their argument brackets (execute with pytests -s flag to show the prints):

import hypothesmith
from hypothesis import given, settings, HealthCheck, assume
import libcst
import re

s = hypothesmith.from_node(node=libcst.FunctionDef, auto_target=True)

@given(code = s)
@settings(suppress_health_check=[HealthCheck.filter_too_much, HealthCheck.too_slow], max_examples=5000)
def test(code):
    assume(re.search(r"\([^)(]+\)\s*:", code) is not None) # should match 'def foo(x):pass' but not 'def foo():pass'
    print("------")
    print(code)
    print("------")

This calculates for a long time but finds absolutely nothing. What am I doing wrong? Or did I misunderstand the purpose of this library?

[Zac-HD]

Hi Julian! This is definitely something that I'd like to work, but as the readme says

This is definitely pre-alpha, but if you want to play with it feel free! You can even keep the shiny pieces when - not if - it breaks.

so I think the answer is just that we're currently defaulting to a pretty poor/narrow automatically-inferred strategy for FunctionDef nodes. The obvious solution is just to add a hand-written strategy to https://github.com/Zac-HD/hypothesmith/blob/master/src/hypothesmith/cst.py - want to open a PR?

Unfortunately hypothesmith is currently a tier-three oss project for me - priority is something like {work, phd}, {hypothesis}, {hypofuzz, shed, pytest, flake8-trio}, {hypothesmith, hypothesis-jsonschema, ......}, so I'm unlikely to write it myself any time soon; very happy to review and help with PRs though!

[julian-goettingen]

Thank you for the quick response and the pointer on how to go about implementing it. I'll put it in my own personal backlog, though I don't suppose that PR is happening this week.