Computes a hash by using an algorithm based on the XOR-cipher. Basically, it splits the file in chunks whose byte-length is the same as the digest size (padding with 0s), and XORs all those chunks together into a new chunk of the same size, the resulting chunk is printed as an array of hexadecimal bytes.
This isn't a good hash function, it's only good for checksums, because it lacks the Avalanche Effect, flipping 1 input bit flips 1 output bit. It is intended to be a simple/basic, educational, and fast checksum algorithm.
The digest size is 128bit (16Byte) by default, but can be set to any valid usize value with the --len option. The IV is hardcoded to be 0x0, it'll allow custom values in the future.
The naming is based on Unix and GNU-coreutils naming conventions, like cksum and md5sum. The behavior of the program is also intended to be similar (but not identical) to those checksum programs, with some inspiration from b3sum.
I'm still trying to fix the formatting of the output to be a single sequence of hex nibbles without delimiter.
sh
cargo install xorsum
xorsum [OPTION]... [FILE]...
If you want to build from source:
sh
cd [REPO] #path to cloned/downloaded repo
cargo build --release
Let's create an empty text file named a. The output of xorsum --len 4 a should be:
[00, 00, 00, 00] a
If we write "aaaa" to this file and rehash it with xorsum a -l 4, the output will be:
[61, 61, 61, 61] a
Because "61" is the hex value of UTF-8 char "a"
Rehashing the file with xorsum a -b yields:
[61, 61, 61, 61, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00, 00]
This is because the IV is all 0s, and the padding is 0 too.
For more info, run:
sh
xorsum --help
DO NOT USE FOR ðCRYPTOGRAPHIC PURPOSES. The algorithm is ðnot crypto-secure