Simple hash function (1 byte output from string input)

Question

Hi folks, I'm looking for a simple hash algroithm that will give me one byte of output for a string input (the inputs will be RFC822 email addresses, if that helps).

I'd like it to be simple, fast, and to magnify input differences (so two similar addresses have differnt outputs). (Yes, I am asking for a lot in one byte of output.)

Idealy, I'd like an XSL answer, but I can take it in either Java or Javascript (and then pass the hash as an argument to the XSL processor).

Thanks.

Answer 1

Use a CRC-8, which has 9 bits of information, then drop a bit off either end and call it a day. Otherwise use any of the other common CRC algorithms.

Answer 2

Why not take the most/least significant byte of the standard String hashCode() function ?

Answer 3

Every hash function has its strengths and weaknesses, and fast and easy to compute ones tend to behave badly for certain classes of data. Trial and error needs to be a part of any solution. In addition to the other suggestions, you might try using integer multiplication as part of the hash function, for example

hash = 0
for (int i=0; i<data.length; i++)
    hash = ((37 * hash) + data[i]) & 0xff;

Answer 4

My suggestion would be to simply XOR all the bytes in the string. Every bit of every byte will influence the end result, and any single-bit error will definitely cause the hash to differ.

Very simple, very fast. And probably nearly as good as any other solution, given the small number of result bits.