Add MatMul operator to patchedast

[lieryan]

Description

Fix issue with parsing code with MatMul operator.

Fixes #496

Checklist (delete if not relevant):

• I have added tests that prove my fix is effective or that my feature works
• I have updated CHANGELOG.md

[lieryan]

Hi @eivindjahren, I wasn't able to replicate the issue, but this PR should fix this issue. Let me know if you would be able to test this branch, or when 1.3.0 is released.

[eivindjahren]

Unfortunately it doesn't seem to fix the problem as the name of the operator is MatMult and not MatMul.

The following regression test (when added to ropetest/refactor/patchedasttest.py reproduces the bug:

    def test_matmul_node(self):
        source = (
            "import numpy as np\n"
            "a=np.array([[1,2]])\n"
            "b=np.array([[1],[2]])\n"
            "a @ b\n"
        )
        ast_frag = patchedast.get_patched_ast(source, True)
        checker = _ResultChecker(self, ast_frag)
        checker.check_children("BinOp", ["Name", " ", "@", " ", "Name"])

I think you probably can skip the import and use of numpy by just using lists. As I understand it, that would just result in a runtime error where it is discovered that a and b do not implement __matmul__. However, adding the numpy import perhaps is more clear when it comes to how this is used.

[eivindjahren]

Should be "MatMult": "@",

[eivindjahren]

Fixed now

[eivindjahren]

The bug is fixed now!🎉 I prefer to add a regression test with every bug-fix. I find its a good way to grow test coverage. The test I pasted above should suffice. But its just my preference and you call the shots 🔫 , so feel free to just ignore that :P

[eivindjahren]

Nice 👍