An encryption/decryption program with an algorithm I came up on my own. It encrypts files based on successive rotations of their bytes.
The algorithm is designed to hinder frequency analysis and plain text attacks, by rotating bytes not only based on passwords, but based on their position in the file as well.
I came up with it on my own lacking a background in cryptography. Some cryptography know-hows told me it looks like a toughened version of an existing algorithm (Vigenere's cipher) with extra steps.
Here is a verbal description of it:
- The user gives an input and a password.
- The characters of the password are treated as u8, and successively XORed or XNORed, depending on whether their are an odd or an even number of characters. Let T the result of this operation. T is a byte treated as a three-digit integer between 000-255. The first digit defines whether the first rotation is to the left (0 or 1), or to the right (2). The second digit defines the number of successive rotations to the right. The third, the number of rotations to the left.
- The bytes of the input are rotated to the right or the left, for as many bits as the "ones" of the password bytes XOR T. i.e. We XOR the first char of the password with T, and use the resulting number of 1s to rotate the first char of the input. We XOR the second char of the password with T, and use the resulting number of 1s to rotate the second char of the input. e.t.c
- Padding is added (see algorithm below)
The core algorithm (1-3) is periodical and rather simple -by no means a strong cipher by modern standards. The padding step makes it tougher to break, at least as long as nothing is known about the passworg (e.g. length).
Padding algorithm:
- Generate N random bytes.
- Create a vector of size [input_size + N].
- The first random byte will be placed at position P1=R1+1 (to always skip the first byte of the input). The second random byte at position P2 = P1+R2. etc. If some position is equal to or greater than the input size, we place the rest of the input bytes after the previous one and the rest of the random bytes will all be appended to the end of the file.
- The beginning and the end of the file is padded with extra random bytes equal to the password length.
N: Check the function determine_padding_size().
R1: Passwords first byte AND 00000111. R2: Passwords second byte AND 00000111. Etc.
There is a TODO of changing N to 256 + (number_of_kbs * 64). Recovering the original size with byte precision is less trivial.
The algorithm has no official name, but if I had to give it one, that would be SYNTOM (from the Greek word 'syntomos' (=quick/brief) ).
Encrytion: run --package crypto-min --bin crypto-min -- -e {path_to_unencrypted_file} -p {password}
Decryption: run --package crypto-min --bin crypto-min -- -d {path_to_encrypted_file} -p {password}
- Add a flag enabling double encryption.
- Use a proper algorithm to determine the padding size.