Getting started

[MarcoGorelli]

The testing tasks involve the following:

1. choose a linting rule (found in pandas_dev_flaker/_plugins_tree or pandas_dev_flaker/_plugins_tokens). E.g. pandas_dev_flaker/_plugins_tree/builtin_filter.py.
2. From the content of MSG, try to understand what it's meant to check for. If it's not clear, please ask :) The above example checks that the builtin function filter isn't used
3. Make a test file in tests, e.g. for the above, it would be tests/builtin_filter_test.py

For your test, follow this template:

import ast
import tokenize
from io import StringIO

import pytest

from pandas_dev_flaker.__main__ import run


def results(s):
    return {
        "{}:{}: {}".format(*r)
        for r in run(
            ast.parse(s),
            list(tokenize.generate_tokens(StringIO(s).readline)),
        )
    }


@pytest.mark.parametrize(
    "source",
    (
        pytest.param(
            <CODE WHICH SHOULD NOT ERROR GOES HERE>,
            id=<NAME FOR TEST CASE>,
        ),
    ),
)
def test_noop(source):
    assert not results(source)


@pytest.mark.parametrize(
    "source, expected",
    (
        pytest.param(
            <CODE WHICH SHOULD ERROR>,
            <EXPECTED ERROR MESSAGE>,
            id=<NAME FOR TEST CASE>
        ),
    ),
)
def test_violation(source, expected):
    (result,) = results(source)
    assert result == expected

So, in the above example, I would expect

filter(lambda x: x, [0, 1])

to error (because we're using the built-in filter function), while

import foo

foo.filter(lambda x: x, [0, 1])

shouldn't error, because it's a function filter which was imported from somewhere rather than being the builtin filter.