1

Problem Statement

Below is a toy example that is close to what I am trying to do.

@given(
    idx_start=integers(min_value=0, max_value=100000),
    idx_window=integers(min_value=0, max_value=100000),
)
def test_calc_max(conftest_df, idx_start, idx_window):
    row_idxs = conftest_df.index[idx_start : (idx_start + idx_window)]
    assert calc_max(conftest_df.loc[row_idxs, "my_column"]) >= 0

conftest_df is a Dataframe that I am making available in my conftest.py fixture file, which represents a portion of real data that I am using for my package.

This dataframe has very few NaN values in it. I want to use hypothesis because, well, it's awesome and I strongly believe it is the right way to do things.

But I also want to make sure that the methods and functions under test work for NaN's. I don't really want to just say NaNs, something else might come up in the future (say, a number that uses a comma instead of a period to represent a decimal).

Ideal Solution via hypothesis

I would rather be able to do something like this:

@given(
    idx_start=integers(min_value=0, max_value=100000, includes=[5, 4000, 80000]),
    idx_window=integers(min_value=0, max_value=100000, includes=[20]),
)
.
.
.

And have a way to ensure that certain values are considered via the includes argument.

I know that hypothesis keeps track of failing values, but it does not seem to guarantee their use, from my experience.

Is there a way to do what I would like?

Mike Williamson
  • 4,915
  • 14
  • 67
  • 104
  • 1
    we can combine `strategie.sampled_from([list of edge cases]) | strategy_for_common_case` and according to the rule examples from the first strategy will be generated more likely than the second one – Azat Ibrakov Sep 25 '20 at 19:08
  • 1
    also we can decorate specific test with [`example`](https://hypothesis.readthedocs.io/en/latest/reproducing.html#hypothesis.example) – Azat Ibrakov Sep 25 '20 at 19:12

1 Answers1

1

You can use the @example decorator to ensure that certain examples are being tested.

This is the example from hypothesis's quick start guide:

from hypothesis import given, example
from hypothesis.strategies import text

@given(text())
@example('')
def test_decode_inverts_encode(s):
    assert decode(encode(s)) == s

Here, @example('') is ensuring that the test will be run with the empty string as an example.

I haven't fully understood the details of your scenario, but maybe you are able to adapt this example to your needs on your own.

Jonathan Scholbach
  • 4,925
  • 3
  • 23
  • 44