Tee should have a public encoding attr

[MatthijsBurgh]

As also in the original nose library, the Tee object should have a public encoding attribute instead of it being hidden. See https://github.com/nose-devs/nose/blob/7c26ad1e6b7d308cafa328ad34736d34028c122a/nose/plugins/xunit.py#L129

As we use nose in combination with doctests, we get an attribute error when doctest tries to get the encoding of what doctest thinks is just a stdout, but because we use nose, it is a Tee object. Which would be fine, if it had the same attributes.
See https://github.com/python/cpython/blob/v3.8.20/Lib/doctest.py#L1452

[MatthijsBurgh]

Please create a new PyPi release after merging this PR.

[mdmintz]

If there's a bug, I need to see a full example that reproduces the issue with the current state of pynose.
Then I will confirm that your proposed update solves the issue.

[MatthijsBurgh]

I don't have a simple example that can be tested easily. You can the CI logs https://dev.azure.com/tue-robotics/tue_robocup/_build/results?buildId=5859&view=logs&j=ce00c4f3-9970-58be-17a6-619446bb3f5a&t=c1c22edd-5a48-59aa-c27e-44463b4e95bc&l=92

Let me know whether this enough for you or not

[MatthijsBurgh]

I was able to reproduce it in a simple example.

Using pynose 1.5.3. I have tested in on python 3.8.18 and 3.10.12.

File structure:

├── main.py
└── tests
    ├── __init__.py
    ├── lib
    │   ├── __init__.py
    │   └── factorial.py
    └── test_case.py

main.py

from tests.test_case import _TestDocTests
import unittest

class DocTests(_TestDocTests):
    def __init__(self, method_name="test_doctests"):
        super().__init__(method_name=method_name)


if __name__ == "__main__":
    unittest.main()

tests/test_case.py

import doctest
from pathlib import Path
import importlib
import os
import unittest

class _TestDocTests(unittest.TestCase):
    def __init__(self, method_name: str = "test_doctests"):
        """
        Constructor

        :param method_name: Name of the member variable to run, this should be "test_doctests" and shouldn't
            be changed.
        """
        assert method_name == "test_doctests", "The method_name should be 'test_doctests'. This is the function which" \
                                               "implements the functionality of this TestCase."
        super().__init__(method_name)

    def test_doctests(self):
        """
        Iterates over all Python files in a path and runs doctest.testmod
        """

        path = Path(__file__).parent

        for root, dirs, files in os.walk(path):
            for filename in files:
                if filename.endswith("factorial.py") and "__init__" not in filename:
                    filepath = os.path.join(root, filename)
                    module_name = "tests"
                    module_name = module_name + filepath.split(module_name)[-1]
                    # module_name = module_name[:-3]
                    module_name = module_name.replace("/", ".")
                    module_name = module_name.rstrip(".py")
                    print(f"{module_name=}")
                    mod = importlib.import_module(module_name)
                    print(mod)
                    doctest.testmod(mod, report=1)

        failed_count, attempted_count = doctest.master.summarize(True)
        print("Attempted count: {}".format(attempted_count))
        self.assertEqual(failed_count, 0, "{} out of {} tests failed".format(failed_count, attempted_count))

tests/lib/factorial.py

"""
This is the "example" module.

The example module supplies one function, factorial().  For example,

>>> factorial(5)
120
"""

def factorial(n):
    """Return the factorial of n, an exact integer >= 0.

    >>> [factorial(n) for n in range(6)]
    [1, 1, 2, 6, 24, 120]
    >>> factorial(30)
    265252859812191058636308480000000
    >>> factorial(-1)
    Traceback (most recent call last):
        ...
    ValueError: n must be >= 0

    Factorials of floats are OK, but the float must be an exact integer:
    >>> factorial(30.1)
    Traceback (most recent call last):
        ...
    ValueError: n must be exact integer
    >>> factorial(30.0)
    265252859812191058636308480000000

    It must also not be ridiculously large:
    >>> factorial(1e100)
    Traceback (most recent call last):
        ...
    OverflowError: n too large
    """

    import math
    if not n >= 0:
        raise ValueError("n must be >= 0")
    if math.floor(n) != n:
        raise ValueError("n must be exact integer")
    if n+1 == n:  # catch a value like 1e300
        raise OverflowError("n too large")
    result = 1
    factor = 2
    while factor <= n:
        result *= factor
        factor += 1
    return result


if __name__ == "__main__":
    import doctest
    doctest.testmod()

Run nosetests main.py --with-xunit

It should give you the following error:

module_name='tests.lib.factorial'
<module 'tests.lib.factorial' from '/ROOT_DIR/tests/lib/factorial.py'>
E
======================================================================
ERROR: Iterates over all Python files in a path and runs doctest.testmod
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/ROOT_DIR/tests/test_case.py", line 38, in test_doctests
    doctest.testmod(mod, report=1)
  File "/home/USER/.pyenv/versions/3.8.18/lib/python3.8/doctest.py", line 1956, in testmod
    runner.run(test)
  File "/home/USER/.pyenv/versions/3.8.18/lib/python3.8/doctest.py", line 1452, in run
    encoding = save_stdout.encoding
AttributeError: 'Tee' object has no attribute 'encoding'

----------------------------------------------------------------------
Ran 1 test in 0.001s

FAILED (errors=1)

[mdmintz]

OK, merged. Will be shipped to PyPI sometime today.

[mdmintz]

Shipped in 1.5.4 - https://pypi.org/project/pynose/1.5.4/