Resolving Issue #699 - Fix runtime error

[lorena-b]

Motivation and Context

This pull request is required to resolve #699. This change resolves a runtime error which occurs when running the UnnecessaryIndexingChecker on a block of code where an assignment node has an empty scope.

This pull request also fixes functions that no longer worked correctly due to changes in the astroid node representation, which caused the example file cases to fail.

Your Changes

Description:

This change handles the case where the second argument of the tuple returned by node.lookup() is an empty tuple so that an index error does not occur.

_is_load_subscript() helper function was rewritten to produce the correct output.

Type of change (select all that apply):

• Bug fix (non-breaking change which fixes an issue)

Testing

In order to test this change a test suite was created for this checker using pylints testing capabilities. The checker was tested with the example code and the code that raised the original error.

Questions and Comments (if applicable)

The tests for the code examples are failing because the checker seems to not have been implemented correctly. I am trying to figure out what is wrong with the current implementation that makes the checker not work even in basic cases.

Update:

• There was an issue with the _is_load_subscript() helper function due to changes in the astroid node representation. When indexing the iterable the Name node now directly has a Subscript as the parent, instead of having an Index node in between.
• Checker now works to pass all of the example cases, issue unnecessary-indexing runtime error #699 case, and some extra cases such as unused iteration variables and looping through the iterable directly.

Checklist

• I have performed a self-review of my own code.
• I have verified that the TravisCI tests have passed.
• I have reviewed the test coverage changes reported on Coveralls.
• I have added tests for my changes.

[david-yz-liu]

@lorena-b great detective work. For the issue of the checker no longer working, you determined that the Index node is no longer available, and replaced that with accessing the Subscript directly.

However, this is a change in Python 3.9, so the tests are now failing in Python 3.8. In cases like this, we can check sys.version (imported from the sys module) to see what the current version of Python is, and then based on the result execute different branches of code. You should do this for the is_load_subscript function.