I just got an email saying that my TEncryption program is going to be hosted on softpedia (http://mac.softpedia.com/get/Utilities/TEncryption.shtml).
It was 3 years ago that I created this program, and completely forgot about it until today. I made it open source, and wrote it in java (so it's platform-independent). It's my own algorithm up to 2^31 bit encryption (only because of memory issues); feel free to look at the source code to learn how it works.
I basically just want to say a little bit about an idea I developed for extending fixed-length hash algorithms to be variable-length hash algorithms. TEncryption used MD5, which I will probably have to change to SHA2 due to the fact that MD5 has been shown to be weak. However, the idea I had (outlined below) is generalizable to any hash algorithm, so that shouldn't be a problem.
The basic flow of symmetric private key encryption is as follows:
A problem arises, however, when you want an 8-bit private key, but only have a 4-bit hash function available. The steps outlined below are how I overcame this problem.
The final variable-length hash algorithm is only as strong as the underlying fixed-length hash algorithm used. In my case, I used this idea to extend the 128-bit MD5 hash algorithm to create variable-length keys. Some implementation choices exist when the hash algorithm's bitlength is not a power of 2 (or a fraction of the desired key's bitlength), and when the fixed-length hash algorithm has a greater bitlength than the variable-length key (not recommended). Check out the hash.java source file in the TEncryption program above to see my implementation of this.
Perhaps in another post I'll explain my underlying encryption algorithm I created but I'm not in the mood right now. Like I said, the source code is all there.