获取pytest以在测试脚本的基本目录中查找

Question

In pytest, my testing script compares the calculated results with baseline results which are loaded via

SCRIPTLOC = os.path.dirname(__file__)
TESTBASELINE = os.path.join(SCRIPTLOC, 'baseline', 'baseline.csv')
baseline = pandas.DataFrame.from_csv(TESTBASELINE)

Is there a non-boilerplate way to tell pytest to start looking from the root directory of the script rather than get the absolute location through SCRIPTLOC?

Answer 1

If you are simply looking for the pytest equivalent of using __file__ , you can add the request fixture to your test and use request.fspath

From the docs :

 class FixtureRequest ... fspath the file system path of the test module which collected this test. 

So an example might look like:

def test_script_loc(request):
    baseline = os.path.join(request.fspath.dirname, 'baseline', 'baseline.cvs')
    print(baseline)

If you're wanting to avoid boilerplate though, you're not going to gain much from doing this (assuming I understand what you mean by 'non-boilerplate')

Personally, I think using the fixture is more explicit (within pytest idioms), but I prefer to wrap the request manipulation in another fixture so I know that I'm specifically grabbing sample test data just by looking at the method signature of a test.

Here's a snippet I use (modified to match your question, I use a subdirectory hierarchy):

# in conftest.py
import pytest

@pytest.fixture(scope="module")
def script_loc(request):
    '''Return the directory of the currently running test script'''

    # uses .join instead of .dirname so we get a LocalPath object instead of
    # a string. LocalPath.join calls normpath for us when joining the path
    return request.fspath.join('..') 

And sample usage

def test_script_loc(script_loc):
    baseline = script_loc.join('baseline/baseline.cvs')
    print(baseline)