Style/MixinUsage false positive in specs

[tagliala]

The Style/MixinUsage finds false positive inside specs on the current master branch of RuboCop

---

Expected behavior

Does not report a lint

Actual behavior

Reports a lint

Steps to reproduce the problem

my_class_spec.rb

# frozen_string_literal: true

require 'spec_helper'

# Top level docs
class MyClass
  def self.descendants
    [String]
  end
end

describe MyClass do
  it 'includes class' do
    described_class.descendants.should include String
  end
end

$ rubocop my_class_spec.rb 
Inspecting 1 file
C

Offenses:

my_class_spec.rb:14:40: C: include is used at the top level. Use inside class or module.
    described_class.descendants.should include String
                                       ^^^^^^^^^^^^^^

1 file inspected, 1 offense detected

Real world use case: https://github.com/brigade/scss-lint/blob/master/spec/scss_lint/reporter_spec.rb

RuboCop version

Include the output of rubocop -V. Here's an example:

$ rubocop -V
0.51.0 (using Parser 2.4.0.2, running on ruby 2.4.2 x86_64-darwin16)

[koic]

Thanks for opening issue. I can reproduce it.

I think that this minimum case will be below.

do_something(include(M))

The difficulty is that it can not be judged by receiver object whether it is Module#include or other include methods.

However, as indicated in the issue example, there are some cases where it becomes false positive with RSpec.
On the other hand, I think that there are not many use cases to use Module#include as method arguments.

So it may be a practical solution to make do_something(include(M)) case false negative.

[Drenmi]

I agree @koic. AFAIK, if the include call is a method argument, it's safe to assume it's not a mixin.

This would also mean the fix should be trivial:

return if include_node.argument?

[koic]

@Drenmi Thanks for comment and advice. I will open a PR to resolve this issue.

[tagliala]

Thanks!