re: pytest REPL

I think the solution here is to write REPL-friendly functions and keep the pytest part to only data assertions.

Of course, you can launch the ipython debugger from pytest like this:

pytest --pdbcls=IPython.terminal.debugger:TerminalPdb --pdb

(and then type interact to go into multi-line REPL mode)

If you have many assertions you can save the expected output to external files; pytest-regressions provides a nice interface for this:

https://github.com/ESSS/pytest-regressions

@pytest.fixture
def assert_unchanged(data_regression, request):
    def assert_unchanged(captured, basename=None):
        data_regression.check(captured, basename=basename if basename else request.node.name.replace("-", " "))

    return assert_unchanged

And here is an example of its use:

@pytest.mark.parametrize("p", [["1970-01-01 00:00:01"], ["--from-unix", "1"]])
@pytest.mark.parametrize("fz", [["-fz", "America/New_York"], ["-fz", "America/Chicago"]])
@pytest.mark.parametrize("tz", [["-tz", "America/New_York"], ["-tz", "America/Chicago"]])
@pytest.mark.parametrize("s", [[], ["-d"], ["-t"]])
@pytest.mark.parametrize("f", [[], ["-TZ"]])
def test_lb_timestamps_tz(assert_unchanged, p, fz, tz, s, f, capsys):
    lb(["timestamps", *p, *fz, *tz, *s, *f])
    captured = capsys.readouterr().out.strip()
    assert_unchanged(captured)