I devised an algorithm where a string is turned into a 26-byte representation of itself, where each byte indicates how many occurrences of a given character are found in the string. For example:

$s = "antidisestablishmentarianism";
#      a b c d e f g h i j k l m n o p q r s t u v w x y z
$r = "\4\1\0\1\2\0\0\1\5\0\0\1\2\3\0\0\0\1\4\3\0\0\0\0\0\0"
[download]

That is then turned into a regex (so \4 becomes [\0-\4]). Then I match the representation of another string against that regex.

I can post the code if you'd like.