Python/Pandas: Unit testing with hypothesis - reproducing falsifying example

Question

Using the hypothesis library for unit testing, I am wondering how I can reproduce a falsifying example pd.DataFrame?

The output looks like this:

Falsifying example: test_data_frame_data(
    data=                            sec_1  sec_2  sec_3
    2020-01-01 00:00:00.000001    0.0   -0.0    0.0
    2020-01-01 00:00:00.000000    0.0   -0.0    0.0
    2020-01-01 00:00:00.000257    0.0   -0.0    0.0
    2020-01-01 00:00:00.000258    0.0   -0.0    0.0
    2020-01-01 00:00:00.000261    0.0   -0.0    0.0
    ...                           ...    ...    ...
    2020-01-01 00:00:01.526858    0.0   -0.0    0.0
    2020-01-01 01:00:01.065796    0.0   -0.0    0.0
    2020-01-01 01:00:01.065797    0.0   -0.0    0.0
    2020-01-01 01:01:01.065795    0.0   -0.0    0.0
    2020-01-01 00:01:01.000020    0.0   -0.0    0.0

    [300 rows x 3 columns],
)

Hypothesis didn't print a seed or a hint to @reproduce_failure.

I simply would like to use the falsifying example to debug my code.

score 0 · Accepted Answer · answered Sep 07 '20 at 08:33

I got a hint on the hypothesis GitHub issues page (https://github.com/HypothesisWorks/hypothesis/issues/2584). Credits to @Zac-HD.

First solution:
Put @settings(print_blob=True) before a test. This will ensure that you'll get a hint how to reproduce a potential error.

Second solution:
Register a settings profile (e.g. in your conftest.py):

from hypothesis import settings
settings.register_profile(name="dev", max_examples=20, print_blob=True)

All about settings can be found in the official docs: https://hypothesis.readthedocs.io/en/latest/settings.html

Python/Pandas: Unit testing with hypothesis - reproducing falsifying example

1 Answers1