Is it safe (in matter of uniqueness) to use UUID to generate a unique identifier for specific string?

String myText;
UUID.nameUUIDFromBytes((myText).getBytes()).toString();

I am using above code to generate a representative for specific texts. For example 'Moien' should always be represeted with "e9cad067-56f3-3ea9-98d2-26e25778c48f", not any changes like project rebuild should be able to change that UUID.

The reason why I'm doing this is so that I don't want those specific texts to be readable(understandable) to human.

Note: I don't need the ability to regenerate the main text (e.g "Moien") after hashing .

I have an alternative way too : 

            MessageDigest digest = MessageDigest.getInstance("SHA-256");
            byte[] hash = digest.digest((matcher.group(1)).getBytes("UTF-8"));
            String a = Base64.encode(hash);

Which od you think is better for my problem?

Jon Skeet
people
quotationmark

UUID.nameUUIDFromBytes appears to basically just be MD5 hashing, with the result being represented as a UUID.

It feels clearer to me to use a base64-encoded hash explicitly, partly as you can then control which hash gets used - which could be relevant if collisions pose any sort of security risk. (SHA-256 is likely a better option than MD5 for exactly that reason.) The string will be longer from SHA-256 of course, but hopefully that's not a problem.

Note that in either case, I'd convert the string to text using a fixed encoding via StandardCharsets. Don't use the platform default (as per your first snippet) and prefer StandardCharsets over magic string values (as per your second snippet).

people

See more on this question at Stackoverflow