Encode String (rfc4122) to Number in Java, decode in PHP

Question

In my use case, a javascript tracker generate a unique ID for a visitor whenever he/she visits the site, using the following formula:

function generateUUID(){
    return 'xxxxxxxx-xxxx-4xxx-yxxx-xxxxxxxxxxxx'.replace(/[xy]/g, function(c) {
        var r = Math.random()*16|0, v = c == 'x' ? r : (r&0x3|0x8);
        return v.toString(16);
    });
}

It generates strings like this (rfc4122):

"3314891e-285e-40a7-ac59-8b232863bead"

Now I need to encode that string in a Number (e.g. BigInteger in Java) that can be read by Mahout. And likewise, restore it (in PHP) to display results. Is there any fast, consistent and reliable way to do that?

Some solutions are:

Mapping each possible char (alphanumeric + '-') to a number [1..M] and summing each char position accordingly
get 2 longs from md5 hash
keep a hash map in memory

Any ideas appreciated!

I know, that's why I post the function above. Maybe someone will come up with a better solution. — ksiomelo, Jan 16 '14 at 16:10
You could convert it to a BigInteger in Java.What do you want to use it for? — Elliott Frisch, Jan 16 '14 at 16:10
@ksiomelo: It is _fundamentally impossible_ to extract 128 bits of data from a 64-bit long. Just use a string. — SLaks, Jan 16 '14 at 16:13
Mahout only accepts IDs in Long format to generate recommendations — ksiomelo, Jan 16 '14 at 16:50

score 1 · Accepted Answer · answered Jan 16 '14 at 19:23

If Mahout can use a compound ID of two longs, you can use:

UUID uuid = UUID.fromString(string);
long l1 = uuid.getMostSignificantBits();
long l2 = uuid.getLeastSignificantBits();

If you really are stuck with one long, then I'd agree with your idea to use a portion of a hash based on the entire UUID

Encode String (rfc4122) to Number in Java, decode in PHP

1 Answers1