Attribute Error when finding modules

[peteboothroyd]

Describe the bug
When using Pyfakefs with pytest an AttributeError is thrown within the _find_modules function within fake_filesystem_unittest.py. This seems to be because the __module__ Attribute does not exist on the EagerTensor from 'tensorflow.python.framework.ops'.

StackTrace:

request = <SubRequest 'fs' for <Function 'test_my_fakefs'>>

    @pytest.fixture
    def fs(request):
        """ Fake filesystem. """
        patcher = Patcher()
>       patcher.setUp()

venv/lib/python3.6/site-packages/pyfakefs/pytest_plugin.py:35:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
venv/lib/python3.6/site-packages/pyfakefs/fake_filesystem_unittest.py:495: in setUp
    self._find_modules()
venv/lib/python3.6/site-packages/pyfakefs/fake_filesystem_unittest.py:458: in _find_modules
    modules = {name: mod for name, mod in module.__dict__.items()
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

.0 = <dict_itemiterator object at 0x10f372458>

    modules = {name: mod for name, mod in module.__dict__.items()
               if inspect.ismodule(mod) and
               mod.__name__ in module_names
               or inspect.isclass(mod) and
>              mod.__module__ in self._class_modules.get(name, [])}
E   AttributeError: __module__

venv/lib/python3.6/site-packages/pyfakefs/fake_filesystem_unittest.py:462: AttributeError

How To Reproduce
Run with tensorflow imported

** Your enviroment**

python -c "import platform; print(platform.platform())": Darwin-18.2.0-x86_64-i386-64bit
python -c "import sys; print('Python', sys.version)": [GCC 4.2.1 Compatible Apple LLVM 9.1.0 (clang-902.0.39.2)]
python -c "from pyfakefs.fake_filesystem import __version__; print('pyfakefs', __version__)": pyfakefs 3.5.3

[peteboothroyd]

Note that for recent releases of tensorflow this has been fixed however I can't currently move to the most recent TensorFlow version.

[mrbean-bremen]

Thanks for the report - I will have a look in the evening. Should be an easy fix - I wasn't aware that this may happen (e.g. missing __module__).
For your use case, are you comfortable with the fix in master, or do you need a new PyPi release?

[peteboothroyd]

Great thanks for the quick response! A new PyPi release would be ideal if possible.

[mrbean-bremen]

Ok, I uploaded a new version to PyPi. Please check if that works for you.

[peteboothroyd]

Thank you I can confirm that this fixes the issue on my end. Appreciate the super quick response!