STYLE disallow Series | AnyArrayLike

[MarcoGorelli]

PR should be made to https://github.com/pandas-dev/pandas-dev-flaker

idea is to disallow foo: Series | AnyArrayLike, as Series is array-like already

This came out of this review: #41955 (comment)

[jramirez857]

Hello, I'd like to try taking this on!

[MarcoGorelli]

Awesome @jramirez857 - a PR should probably go to https://github.com/pandas-dev/pandas-dev-flaker , let me know if you want / need any help

[jramirez857]

Thank you, @MarcoGorelli for offering help. Yes, this is my first contribution and if you could point me in the right direction that would be very helpful. Here is what I currently have.

import ast
from typing import Iterator, Tuple

from pandas_dev_flaker._data_tree import State, register

MSG = "PDF026 found union between State and AnyArrayLike in type hint"

@register(ast.FunctionDef)
def visit_arg(
    state: State,
    node: ast.FunctionDef,
    parent: ast.AST,
) -> Iterator[Tuple[int, int, str]]:

    arguments = node.args
    for arg in arguments:
        if (isinstance(arg.annotation, ast.BinOp)):
            yield node.lineno, node.col_offset, MSG

I am currently trying out using registering ast.FunctionDef using the decorator and I think I will need to traverse a binary tree of BinOp of argument nodes. I have written a test case shown below:

@pytest.mark.parametrize(
   "source, expected",
   (
       pytest.param(
           "def dot(foo, other: Series | DataFrame | AnyArrayLike): pass",
           "1:0: PDF026 Both State and AnyArrayLike in type hint",
           id="Series DataFrame AnyArrayLike type hint",
       ),
   ),
)
def test_violation(source, expected):
   (result,) = results(source)
   assert result == expected

I am currently figuring out how I can try to step through it in the repl. If you have any tips on how I could step through the tree traversal within the pandas-dev-flake environment it would be a huge help for my debugging purposes.

[MarcoGorelli]

You might want to have a look at pyupgrade, they do some similar things (for example to rewrite Optional[int] as int | None)

[Pydare]

hello, is it too late to try this out?

[MarcoGorelli]

Hey @Pydare - I don't think anyone's working on this, though @jramirez857 please do speak up if you still are

[jramirez857]

Feel free to take this @Pydare. I made some progress initially but have not been actively working on this recently.

[Pydare]

take

[MarcoGorelli]

this is done, thanks to Pydare - all that needs doing next is to make a new release and bump the version - I'll do that when I get a chance, hopefully soon