Or, for better time complexity (with a bit of extra space) making a map of char -> frequency and comparing the results.