Fix VimPathFinder.find_spec (Python 3.4+)

[blueyed]

The fix is similar to what VimModuleLoader.find_module does already.

[blueyed]

__spec__ might be None here.. seen when using jedi-vim, which apparently also comes here.

Need to investigate..

[blueyed]

@davidhalter
Any idea maybe?
Maybe AttributeError could be handled like KeyError, but it would be good to know what is going on.

I do not have the details anymore, but am using this branch again now (to fix reload(jedi_vim) in Neovim).

[blueyed]

While #297 (comment) is not really clear, I think this PR is better than the current situation, where reload() does not work at all:

:py3 from importlib import reload; reload(jedi_vim) results in:

Error detected while processing function provider#python3#Call:
line   18:
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/lib/python3.6/importlib/__init__.py", line 165, in reload
    spec = module.__spec__ = _bootstrap._find_spec(name, pkgpath, target)
  File "<frozen importlib._bootstrap>", line 894, in _find_spec          
  File "<frozen importlib._bootstrap_external>", line 1157, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1129, in _get_spec
  File "…/neovim-python-client/neovim/plugin/script_host.py", line 247, in find_spec
    return PathFinder.find_spec(fullname, path or _get_paths(), target)
  File "<frozen importlib._bootstrap_external>", line 1157, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1123, in _get_spec
TypeError: 'module' object is not iterable                              

find_spec gets called with the following with reload:

find_spec: fullname='jedi_vim', path=<module 'jedi_vim' from '/home/user/.vim/plugged/jedi-vim/pythonx/jedi_vim.py'>, target=None

[blueyed]

Found the issue: it was not using the correct interface (path is only used with meta path finders!).

[justinmk]

LGTM. The flake8 failure happens on master, seems like a flake8 bug.

flake8 version:

checkqa installed: flake8==3.5.0

Did a web search but didn't find much.

[blueyed]

@justinmk

The flake8 failure happens on master, seems like a flake8 bug.

Could you reproduce it locally?
tox -e checkqa works for me.

[justinmk]

@blueyed yes, in a virtualenv:

$ ./venv/bin/pip3 install .

$ ./venv/bin/python --version
Python 3.5.2
$ ./venv/bin/tox --version
3.0.0 imported from /home/vagrant/python-client/venv/lib/python3.5/site-packages/tox/__init__.py
$ ./venv/bin/flake8 --version
3.5.0 (mccabe: 0.6.1, pycodestyle: 2.3.1, pyflakes: 1.6.0) CPython 3.5.2 on Linux

$ 


TOXENV=checkqa ./venv/bin/tox
...
AttributeError: 'Attribute' object has no attribute 'id'
ERROR: InvocationError for command '/home/vagrant/python-client/.tox/checkqa/bin/flake8 neovim' (exited with code 1)