CODEWITHSUNDEEP
java
snow8261
snow8261

Reputation: 1009

Unique ID Generation in java

Giving a Input String, the string length will not more than 30, the output will be a unique id number. Is there a way in Java can do this?the same string will always generate the same id, different string can not generate the same id. the java HashCode() can do this ?

thanks

Upvotes: 2

Views: 1382

Answers (5)

Petr
Petr

Reputation: 63409

To satisfy the requirement

the same string will always generate the same id, different string can not generate the same id

you'll get pretty huge numbers. You require that the function will be injective so you need as many numbers as the number of possible Strings, which is something like $255^30$ in your case (or something like $65536^30$ if you allow arbitrary Unicode characters). So you'll need BigIntegers for that and certainly you use int (simply there are more Strings of length up to 30 than numbers in int). For example, new BigInteger(theString.getBytes("")) satisfies your requirement.

If you use hashCode, you'll lose injectivity, but the chance that two Strings will have the same hashCode is very low in most cases (it's actually the purpose of hashing to make this chance low). If you want to be extra sure that the number of collisions uniformly small, you can use some cryptographic hash function, but still, the mapping won't be injective.

Perhaps explaining the reasons for your requirements would help finding the best solution.

Upvotes: 3

codebox
codebox

Reputation: 20264

If you really truly need to be sure there are no collisions then the numbers you get will be enormous, certainly bigger than any of the primitive numeric types. My advice would be to use a SHA-1 hash which is almost certainly good enough (the Git version control system relies on the uniqueness of these hash values, are your requirements really stricter than theirs?)

If you want to guarantee uniqueness then I suppose one way you could do it would be to take the ASCII (or Unicode, depending on your input) numeric value for each character, padding it with zeroes so that all values have the same length, and then concatenating them all together into one big number. As stated previously there is no way you will fit this into a long value, so you will need to use the BigInteger class. The padding will be necessary to avoid collisions like '12 + 34' and '123 + 4'.

Upvotes: 0

Nicola Musatti
Nicola Musatti

Reputation: 18236

Unless you can restrict the set of valid strings in a deterministic way your only option is to interpret the whole string as your id. As you're going to obtain very large id's you should use the BigInteger class to represent them.

I'd convert characters in reverse order, so as not to have to worry about strings of different length.

Upvotes: 0

hsz
hsz

Reputation: 152294

Maybe it is a bit silly, but how about converting every character to ASCII number ?

Upvotes: 0

srini.venigalla
srini.venigalla

Reputation: 5145

Java hashcode is not guaranteed to be unique. You need to look into UUID.

public static UUID nameUUIDFromBytes(byte[] name)

http://docs.oracle.com/javase/1.5.0/docs/api/java/util/UUID.html

Upvotes: 1

Related Questions