Reputation: 6281
I want to create a byte sequence with a fixed length out of a string which has a variable length. What is the best way to archive this. All bytes should be as different as possible.
This has been my first approach for the generation of the bytes:
static byte[] GenerateBytes(string password, Int32 strength)
{
Byte[] result = new byte[strength];
Byte[] pwBytes = Encoding.ASCII.GetBytes(password);
Int32 prime = GetLowerPrime(pwBytes.Length);
// Offset count to avoid values
Int32 count = prime;
Int32 sum = 0;
for (int i = 0; i < result.Length; i++) {
sum += (result[i] = pwBytes[(count++ % pwBytes.Length)]);
}
count += prime;
Int32 pcount = prime;
for (int i = 0; i < result.Length * 7; i++) {
result[(i % result.Length)] ^= (Byte)(pwBytes[(count++ % pwBytes.Length)] ^ ((pcount += pwBytes[(count % pwBytes.Length)]) % 255));
}
return result;
}
And generated some samples with 256 / 128 / 64 generated bytes and counted the unique bytes:
Password "Short": 170 103 60
Password "LongerX": 173 101 55
Password "Really Long": 169 100 57
Password "Unbelivable Safe!0§$": 162 101 56
Password "MCV": 119 113 61
Password "AAA": 50 51 50
Password "BBB": 67 67 52
Password "AAAAAA": 48 48 48
I tried to change the prime selector a bit this improves the generation with short keys but has partly a impact on long ones. I also tracked some statistics of the bytes. Generated and each byte value is used between 9 and 30 times.
Upvotes: 2
Views: 419
Reputation: 4097
well if you really want to roll your own solution that has no real practical use other than theoretical interest, (because this sounds like a homework question) just start off with a one-time pad of random bytes and XOR the pwd with the first few bytes, should give you reasonably high entropy for short pwds.
Upvotes: 2
Reputation: 14160
You seems to be reinventing the wheel. If you need to make key from the password, use hashing function, or, the best way - one of the standard password-based key derivation function. Search for PBKDF2.
Upvotes: 2