ModuleNotFoundError for Numpy when pytest.approx fails

Description of the bug:

When using pytest.approx to compare lists in a test, a ModuleNotFoundError is raised for Numpy whenever the test fails. If the test succeeds, there is no such error.

It appears that pytest.approx does not require Numpy to run, but the error message requires Numpy to display the details. Since pytest.approx does not require Numpy to function correctly, it should not require Numpy to display the error.

Output of pip list from the virtual environment:

Package    Version
---------- -------
attrs      21.4.0
iniconfig  1.1.1
packaging  21.3
pip        22.0.3
pluggy     1.0.0
py         1.11.0
pyparsing  3.0.7
pytest     7.0.1
setuptools 49.2.1
tomli      2.0.1

pytest and operating system versions:

$ python --version
Python 3.9.0
$ python -m pytest --version
pytest 7.0.1

macOS Big Sur

Version 11.6.2

Minimal example:

import pytest
def test_approx():
    assert [1, 2] == pytest.approx([1.001, 2.002])

Actual Result:

$ pytest
============================= test session starts ==============================
platform darwin -- Python 3.9.0, pytest-7.0.1, pluggy-1.0.0
rootdir: ****
collected 1 item                                                               

test_approx.py F                                                         [100%]

=================================== FAILURES ===================================
_________________________________ test_approx __________________________________

    def test_approx():
>       assert [1, 2] == pytest.approx([1.001, 2.002])
E       AssertionError: assert [1, 2] == approx([1.001...02 ± 2.0e-06])
E         (pytest_assertion plugin: representation of details failed: /Users/adalessa/Downloads/diffusion-master 2/venv/lib/python3.9/site-packages/_pytest/python_api.py:323: ModuleNotFoundError: No module named 'numpy'.
E          Probably an object has a faulty __repr__.)

test_approx.py:5: AssertionError
=========================== short test summary info ============================
FAILED test_approx.py::test_approx - AssertionError: assert [1, 2] == approx(...
============================== 1 failed in 0.04s ===============================

Expected result:

No ModuleNotFoundError: No module named 'numpy'. which makes the whole error message confusing and leads you to believe it failed because Numpy is not installed instead of the fact it was an assertion error.

[nicoddemus]

@dzht19 please go ahead!

The offending line is here:

https://github.com/pytest-dev/pytest/blob/9318b2cb7f81252fec215e1cce4c5de021bda180/src/_pytest/python_api.py#L343

np.inf should be replaced by math.inf, and the numpy import at the beginning of the function should be removed. Also we should fix our test suite: TestApprox.test_error_messages currently tests scalars, lists and numpy arrays, but it uses importorskip at the beginning, so we skip the tests if numpy is not installed. We should split the test into two: one which tests everything not-numpy related, and one which tests numpy-data and depends on numpy.

[dzht19]

Hi, I am a beginner and I am looking for the first issue to work on. Could I try to work on this one? Is there anyone else who started contributing? Thank you for the answer in advance.