Unique ID generator

Anon

I'm working on an algorithm which causes some difficulties...

The Problem:
I'm trying to make a function. This functions input is a sequence of characters (letters) and its output is a number. The number is supposed to be generated on the basis of the input, so that each time the same sequence of characters occur the same number is generated. The number has to be unique, so that different sequences of characters generates different numbers. And now comes the really tricky part: The number has to be max five digits (99999). I know that the theoretical number of combinations generated on the basis of these characters (30 in length) are much higher, but in the practical world all these combinations will not occur. I know exactly how many combinations there are before the function is supposed to run (currently there are ca. 35000).

Does anyone know how to create such a algorithm, or have any tips on any principles... or any idea of how I can find out?

I would really appreciate some help here, so if anyone... :-)

Anon

MD5 does something similar to what you're trying to accomplish but with 32 hex digits (i.e. ffffffffffffffffffffffffffffff). Here's a URL for the MD5 algorithm and whitepapers on it: http://www.enteract.com/~lspitz/md5.html

Anon

Thanks for your suggestion! ...but I've tried MD5 and it doesn't work for my purposes :-(

I need straight decimal numbers ... and they have to be small.

...ay other suggestions?

Anon

Well, you could add up the ASCII value of each character, but you won't guarantee a unique number (two different words could add up to the same value).

Anon

What I'm saying is read up on the MD5 algorithm and implement one yourself that uses less digits. AFAIK there are no other alternatives.

Anon

Hi.
First of all, sorry for my bas english.
For your problem, I suggest this:

Sum (I suposse that multiply will work fine, try it) the ASCII values of all the characters of the string. It will generate a number of unknow lenght.
Add a number (for example, "99999" for be sure that the number have enought lenght.
Use the % operator until the number lenght became five.
Choose the last five ciphers
of the number. Here is your number.
If you want to test the algorithm, make a bucle that test all the combinations of characters and check if the function returns are duplicates.
Another thing you can do, is apply the MD5 function and transform all characters in chipers.

Anon

Your requirements are pretty much impossible to reach. If you only have a range of 00000 to 99999 to represent the output, then you have a range of 100,000. It is highly likely that you will get the same number for two or more words.

If that's ok every now and then, then what I'd do is md5 hash it, then do something quick and dirty based on the hash to get a number out of it, like adding the ascii value of each letter times some power that increases as you go across, then grabbing the last 5 digits (or first) and using that.

Anon

Thank you for your suggestions. It seems as if most of you do not believe that it is possible due to the requirements, but you have given me something to work on so i'll be doing that for a couple of days now.

If anyone else has something to say, their comments would be very welcome though.

Anon

A gift:

Here's something that might work:

<form>
Nugget:<input type=text name=nugget>
</form>
<?
if ($nugget){
$temp=tempnam("/tmp","md5");
$fp=fopen($temp,"w");
fwrite ($fp,$nugget,strlen($nugget));
fclose($fp);
$nugget=EscapeShellCmd($nugget);
$answer=exec("md5sum $temp");
$p=split(" ",$answer);
unlink($temp);
$test=base_convert($p[0],36,10);
$test=substr($test,0,5);
print $test;

}

?>

Anon

oops. the escapeshellcmd line should go right before fwrite...

Anon

What about using the ideas of adding the ascii values of the characters in some what to come up with a number within your range, then writing the original sequence of characters and the ID to a 2 column database?

Now whenever someone enters a sequence, an ID is generated and compared to the existing IDs. If this new ID is a duplicate, add 1 and search again. Once it is unique, write the new sequence and ID to the database.

Rob