[Suggestion] Use a namedtuple for capsys results

[samueldg]

Currently, the capsys fixture's readouterr() method returns a tuple with (stdout, stderr) results.

Usage will look something like this:

def test_print_1(capsys):
    print('something')
    out, err = capsys.readouterr()  # Named, unused variable
    assert 'something' in out

def test_print_2(capsys):
    print('something ')
    out, _ = capsys.readouterr()  # Anonymous, ignored variable
    assert 'something' in out

def test_print_3(capsys):
    print('something')
    assert 'something' in capsys.readouterr()[0]  # Index access (not super readable)

My suggestion would be to return a namedtuple instead, so the attributes can be accessed by name

def test_print_4(capsys):
    print('something')
    assert 'something' in capsys.readouterr().out  # Attribute access

Which could be achieved by wrapping the result with something like:

from collections import namedtuple
CapsysResult = namedtuple('CapsysResult', ['out', 'err'])

This would be pretty straightforward, and still fully compatible with existing code using tuples either for index access or unpacking.

[samueldg]

Let me know if that's something you'd want in, I'd gladly submit a PR.

[nicoddemus]

@samueldg that's a perfect case where changing to a namedtuple is a great idea. A PR would be greatly appreciated, thanks! (Make sure to target the features branch)

[samueldg]

@nicoddemus There you go: #2880

[nicoddemus]

Awesome, thanks!

[RonnyPfannschmidt]

closed by #2880