add annotations in pylint/checkers.

[sushobhit27]

add mypy in tox.ini for only pylint/checkers.

Fixes / new features

• Annotate pylint code with type annotations #2079

[AWhetter]

Please could this be run in a separate virtualenv to the tests. Seeing as mypy is failing at the moment, I think we should stop the mypy virtualenv running by default as well.

[sushobhit27]

@AWhetter added mypy in seperate virtualenv, however it is failing as many errors are false positive IMO. I will work on it to either fix it or ignore it depending on each error.

[coveralls]

Coverage decreased (-0.6%) to 89.587% when pulling 654447a on sushobhit27:fix_2079 into db63b49 on PyCQA:master.

[PCManticore]

@sushobhit27 I think you might need to do a rebase since some changes that are now in master appear in your branch as well. Didn't have time yet to review this PR though

[PCManticore]

Looks good overall. As mentioned, I think this needs to be rebased to get rid of these new additions. Also left a couple of comments with some places where I'm not sure what we're type ignoring.

[PCManticore]

Does this class behave like None when we define a variable with it?

[sushobhit27]

will check for it.

[asottile]

should these instead be CLASS_NAME_RGX: Pattern[str]

[sushobhit27]

@asottile my bad...yes that should be like that, or CLASS_NAME_RGX: Optional[Pattern[str]], if variable can have either None or Pattern[str] value. Missed to replace = with :

[PCManticore]

Unfortunately that's 3.6 syntax only so won't work for us just yet (we still support 3.4 and 3.5)

[PCManticore]

Why do we have to ignore it here?

[PCManticore]

Ditto

[sushobhit27]

$ python -m mypy --ignore-missing-imports checkers/spelling.py
checkers/spelling.py:33: error: Name 'Filter' already defined (possibly by an import)

python/mypy#2556
All these ignore were generally required to just suppress the false positive. Although now I think, in first pass instead of using ignore we should let the errors comes and filter out any actual errors and have a fix for them.
e.g mypy helped in finding just below two errors:

def is_error(node) -> bool:
     """return true if the function does nothing but raising an exception"""
     for child_node in node.get_children():
         if isinstance(child_node, astroid.Raise):
             return True
         return False

IMO return False at last statement should be de-idented.

Similarly in below example, although in most of cases below code should work but instead of returning None, bool should be returned explicitly.

def _is_property_decorator(decorator) -> bool:
     for infered in decorator.infer():
         if isinstance(infered, astroid.ClassDef):
             if infered.root().name == BUILTINS_NAME and infered.name == 'property':
                 return True
             for ancestor in infered.ancestors():
                 if ancestor.name == 'property' and ancestor.root().name == BUILTINS_NAME:
                     return True
     return None 

[PCManticore]

Ditto. what does this ignore, the iteration or the result of the dictionary creation?

[sushobhit27]

checkers/utils.py:108: error: "object" has no attribute "iter"; maybe "str" or "dir"? (not iterable)
somehow mypy not able to infer that method is a tuple. annotating _SPECIAL_METHODS_PARAMS might have fixed this, but didn't try it.

[PCManticore]

Hey @sushobhit27 Do you still plan to work on this PR? Would love to get some annotations in.

[sushobhit27]

Hey @sushobhit27 Do you still plan to work on this PR? Would love to get some annotations in.

@PCManticore AFAIK it was decided to postpone this activity and have annotations done in astroid first, that's why I stopped working on this PR, however if you want I can continue on it. Let me know any specific files, I need to pick first.

[PCManticore]

@sushobhit27 We can definitely continue with this PR as well, as we'd see at least some minimal benefit in how to set up annotations in a way that works across Python 3.4 - 3.7. So let's bring it to a shippable state and after that we can take a look at astroid as well.

[sushobhit27]

@PCManticore ok, I start working on it again.

[PCManticore]

We can come back later at this.