a unique number for a word

[dkaip]

Hello all. I am trying to take a unique value for a wide string, with the function.

long int UniqueCode(wchar_t* ch)
{
long power=127.0;
        long int code=0;
        int res,j=0;
        for (;*ch;ch++) {
                res=*ch;
                code+=res*pow(power,j++);
        }
        return code;
}

But at some different words i have the same resaults.
Also when the word is big the resault is negative, bacause of limits.
Any ideas?

[helios]

Google CRC32. It's still quite likely to get collisions, though. If you want to have a little more certainty that there won't be any collisions, Google MD5 or SHA-1.

[csiz]

wchar_t has more biytes I think

now power=2^(sizeof(wchar_t)*8) and if its signed than you need to make it unsigned(at leas make it appear unsigned) so after res=*ch; res+=power/2;

[dkaip]

Here is the all routine, but it seems have a bug...
In this routine there is a unique subcode for the same string.
Any improovment?
To use an md5 etc.. is an all story...
Difficault for me.

unsigned long int UniqueCode(wchar_t* ch, int sbCode)
{
        long power=2^(sizeof(wchar_t)*8);
        long int code=0;
        int res,j=0;
        for (;*ch;ch++) {
                res=*ch;
                res+=power/2;
                code+=res*pow(power+0.0,j++);
        }
        code+=sbCode*2^j;
        return code;
}

[Duthomhas]

You are trying to take N significant bits and compress them into <N significant bits. There will always be collisions. It is not a 1:1 function.

[dkaip]

New idea, function etc ?

[Duthomhas]

What are you trying to do with this?

[dkaip]

It is a very complicate project, and i think i am finish, and without this function. It is about control in Greek anssient text, that means utf8. I convert the hole project to c++, without knowing STL, and reading and asq.
Thanks all for the great help.