Recursing python dictionaries with yield from to generate list of nested dictionary keys

I want to generate a list of unique nested keys for each value in a dictionary such that:

input_dict = {"a": {"b": "c", "d": {"e": "f"}}, "f": "g"}

expected_result = [["a", "b"], ["a", "d", "e"], ["f"]]

I thought something along these lines would work, appending each key to a list and recursing until a value is reached. At which point I yield a list and carry on.

def _i(itr, list_of_keys):
    if list_of_keys is None:
        list_of_keys = []

    if isinstance(itr, dict):
        # For each dict, add the key to the list, and recurse
        for key, value in itr.items():
            list_of_keys.append(key)
            yield from _i(value, list_of_keys)
    else:
        # If not a dict, then at the end of a series of keys, yield the list
        yield list_of_keys
        list_of_keys = []

However when running, the result is all the unique keys

x = _i(input_dict, list_of_keys=None)
list(x)

 [['a', 'b', 'd', 'e', 'f'],
 ['a', 'b', 'd', 'e', 'f'],
 ['a', 'b', 'd', 'e', 'f']]

Think I must be missing something here on how yielding / input arguments work

Answer

You keep modying the same list object (the reassignment of a local variable at the end there has no effect for any recursive call up or down the stack)! A simpler approach without a carry-over variable would be:

def _i(obj):
    if isinstance(obj, dict):
        for k, v in obj.items():
            for keys in _i(v):
                yield [k] + keys
    else:
        yield []

>>> list(_i(input_dict))
[['a', 'b'], ['a', 'd', 'e'], ['f']]