Some optimizations on the suggestion by @Eastsun are:

• Use a single mutable list to store each permutation. This is faster than repeatedly copying immutable strings as a result of the '+' operator, even when converting the list to string at the end.

• Swap values in the items list to avoid expensive items[:i]+items[i+1:] copies with O(n) complexity, and then an undo-swap is required when backtracking.

from typing import Iterator

def permutations(items: str) -> Iterator[list[str]]:
    yield from _permutations_rec(list(items), 0)

def _permutations_rec(items: list[str], index: int) -> Iterator[list[str]]:
    if index == len(items):
        yield items
    else:
        for i in range(index, len(items)):
            items[index], items[i] = items[i], items[index]  # swap
            yield from _permutations_rec(items, index + 1)
            items[index], items[i] = items[i], items[index]  # backtrack

# print permutations
for x in permutations('abc'):
    print(''.join(x))