Checking if two strings are anagrams

Question

I have a piece of code that checks if two strings are anagrams. I would like this to be as pythonic as possible.

def are_two_string_anagram(string1, string2):
    if len(string1) != len(string2):
        return False

    def char_count(string):
        char_count = {}
        for s in string:
            char_count[s] = char_count.get(s, 0) + 1
        return char_count

    chr_count = char_count(string1)
    chr1_count = char_count(string2)
    return chr_count == chr1_count

string1 = "mary"
string2 = "yram"
print(are_two_string_anagram(string1, string2))

string1 = "albert"
string2 = "abelrt"
print(are_two_string_anagram(string1, string2))

string1 = "something"
string2 = "anotherus"
print(are_two_string_anagram(string1, string2))

Answer 1

Most python users would just use sorted(string1) == sorted(string2). That's \$O(n log n)\$, which is not that much worse than your \$O(n)\$ solution.

But the code you've actually written is just a reimplementation of:

collections.Counter(string1) == collections.Counter(string2)

That said, your short-circuit if the lengths mismatch is a good idea no matter which implementation you use.

There are some style issues that jump out with your implementation:

• char_count doesn't have a pressing need to be a nested function, I would put it at top level.
• Don't have a local variable named char_count inside the same-named function.
• Don't cross numbers like chr_count = char_count(string1);chr1_count = char_count(string2). If you need numbered variables, use the same number for all transformed forms also.

For your demonstration code, I would have written:

def test_two_string_anagrams():
    assert are_two_string_anagram("mary", "yram")
    assert are_two_string_anagram("albert", "abelrt")
    assert are_two_string_anagram("something", "anotherus")

Or possibly:

def test_two_string_anagrams():
    for s1, s2 in [
        ("mary", "yram"),
        ("albert", "abelrt"),
        ("something", "anotherus"),

    ]:
        assert are_two_string_anagram(s1, s2)

That way it can be run directly with py.test (and possibly other harnesses, I know the standard unittest requires wrapping them in a dummy class, but I haven't used any others).