Fixture for when you're getting warnings deep in your code during test

fixture_for_catching_warnings.py

"""
Add this fixture to a file containing a test, or to your global `conftest.py`, to turn warnings into errors.

This way the test will break at the warning point, and your traceback will likely contain all functions or methods
responsible for creating the conditions for the warning being raised.

For instance, in the case of a `SettingWithCopyWarning` from pandas, the function that tries to mutate a dataframe
view/slice might not be the function that created the dataframe view/slice in the first place, but that function
is probably in the stack somewhere.

So converting the warning into an error causes a traceback that makes the origin of the problematic dataframe
easier to locate...
"""

@pytest.fixture(autouse=True)
def setting_with_copy_warning_error():
    import warnings
    from pandas.core.common import SettingWithCopyWarning
    with warnings.catch_warnings(record=True) as w:
        warnings.filterwarnings("error", category=SettingWithCopyWarning)
        yield w