Bug pylint 2436

[hippo91]

Steps

• [Ok ] For new features or bug fixes, add a ChangeLog entry describing what your PR does.
• [Ok] Write a good description on what the PR does.

Description

This PR solve the bug reported as pylint-dev/pylint#2436.
In fact two problems were present.
The first one was a lack of unary operators definition for some numeric types in brain_numpy. Thus the inference of such operators failed and a message invalid-unary-operand-type was emitted.

The second one was a wrong inference of call to some numpy functions. A List or Tupleinstance was present in the inference generator. It has been corrected by adding some transformations in the brain_numpy module. Such transformation creates a closure in which the call to classical inference method is made and then filtered to remove undesirable types.

Type of Changes

	Type
✓	🐛 Bug fix

Related Issue

Close pylint-dev/pylint#2436

[PCManticore]

@hippo91 The first one with the missing methods makes sense. Can you split that up into its own separate PR? I am not sure about the first one though, as you mentioned in the comment from the original Pylint issue, the type you get depends on what argument you pass into a function. As such, it is entirely possible to get back a list or a tuple depending on the value that went into the function call. I don't think we should be doing anything here, in terms of modelling the types that those functions return, it would still be an Union[tuple, list, recarray, ...].

[hippo91]

@PCManticore, no pb i'll do a PR just for the first point.
For the second however i'wonder how it will be possible to suppress spurious message such as E1130: bad operand type for unary -: list (invalid-unary-operand-type) without filtering List or Tuple from inference results.
Moreover, from the numpy documentation, i understand, but i'm maybe wrong, that those function (array and linspace) return ndarray and not another type.

[hippo91]

@PCManticore , the PR for the first part is available under #645.

[PCManticore]

@hippo91 Gotcha, that makes sense. I approved the PR, let's fix the conflicts and merge it.

[hippo91]

@PCManticore i fixed the conflicts. Only two unit tests are failing but they are identical to the ones that failed in previous commits. I merged the branch.