Multiple exceptions and code coverage when unit testing python

Question

The Problem:

Here is an artificial example of the code under test:

from datetime import datetime

def f(s):
    try:
        date = s.split(":")[1]
        return datetime.strptime(date, "%Y%m%d")
    except (ValueError, IndexError) as e:
        # some code here
        raise

Here is a set of tests I currently have:

from datetime import datetime
import unittest

from test_module import f

class MyTestCase(unittest.TestCase):
    def test_valid_date(self):
        self.assertEqual(f("1:20130101"), datetime(2013, 1, 1))

    def test_invalid_date(self):
        self.assertRaises(ValueError, f, "1:invalid")

The test passes and, if I run the coverage with the --branch flag, I would get 100% line and branch coverage:

$ coverage run --branch -m unittest test
..
----------------------------------------------------------------------
Ran 2 tests in 0.003s

OK
$ coverage report
Name            Stmts   Miss Branch BrPart  Cover
--------------------------------------------
test_module.py      7      0      0      0   100%
--------------------------------------------
TOTAL               7      0      0      0   100%

However, note that the test currently examines only two cases - when there is no exception thrown, and there is a ValueError exception raised.

The question:

Is there a way for coverage to report that I have not tested a case when IndexError is raised?

You could have a different except block for IndexError. I think coverage just counts which lines that have been executed. — Håken Lid
– Håken Lid, Commented May 26, 2016 at 14:43

Ned Batchelder · Accepted Answer · 2016-05-27 10:57:04Z

9

+50

Coverage.py can only measure which execution paths (statements or branches) were run. It has no means of tracking what values were used, including what exception types were raised.

As I see it, your options are:

Separate the exception clauses. In the code you've shown, the two exceptions could be raised by separate lines anyway, though perhaps in your real code they are not so separable.
Don't worry about the two exceptions. Your tests for this code will likely consider a number of different inputs, designed to exercise different edge cases. Coverage.py can't help you with distinguishing among them all, or ensuring that you've written enough cases. Use other criteria to decide you've written enough test cases.

edited May 27, 2016 at 10:57

answered May 27, 2016 at 0:53

Ned Batchelder

378k77 gold badges583 silver badges675 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

alecxe Over a year ago

Okay, thanks so much. Good luck with your talk on PyCon!

Nikign · Accepted Answer · 2016-05-26 14:43:23Z

2

I think you can try with two separate excepts for the two exceptions. In that case, line coverage shows that you haven't tested one condition.

from datetime import datetime

def f(s):
    try:
        date = s.split(":")[1]
        return datetime.strptime(date, "%Y%m%d")
    except ValueError as e:
        # some code here
        raise
    except IndexError as e:
        # some code
        raise

If you don't want to repeat your some code, you might be able to use a function for that.

answered May 26, 2016 at 14:43

Nikign

3793 silver badges9 bronze badges

1 Comment

alecxe Over a year ago

That is definitely an option, but my curiosity still has that question if coverage (or other package) would determine the non-tested execution path without changing the code under test. Thanks!

Collectives™ on Stack Overflow

Multiple exceptions and code coverage when unit testing python

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related