// AES specs
// https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.197.pdf

// S box, 16*16 Matrix with fixed values
#[rustfmt::skip]
static SBOX: [[u8; 16]; 16] = [
    [0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5, 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,],
    [0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0, 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,],
    [0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc, 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,],
    [0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a, 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,],
    [0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0, 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,],
    [0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b, 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,],
    [0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85, 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,],
    [0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5, 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,],
    [0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17, 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,],
    [0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88, 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,],
    [0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c, 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,],
    [0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9, 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,],
    [0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6, 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,],
    [0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e, 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,],
    [0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94, 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,],
    [0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68, 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16,],
];

#[rustfmt::skip]
static INVERSE_SBOX: [[u8; 16]; 16] = [
    [0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38, 0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,],
    [0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87, 0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,],
    [0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d, 0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,],
    [0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2, 0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,],
    [0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16, 0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,],
    [0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda, 0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,],
    [0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a, 0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,],
    [0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02, 0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,],
    [0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea, 0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,],
    [0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85, 0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,],
    [0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89, 0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,],
    [0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20, 0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,],
    [0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31, 0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,],
    [0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d, 0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,],
    [0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0, 0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,],
    [0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26, 0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d,],
];

// for key schedule
pub static RC: [u8; 11] = [
    0x00, 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1B, 0x36,
];

pub fn clone_into_array<A, T>(slice: &[T]) -> A
where
    A: Default + AsMut<[T]>,
    T: Clone,
{
    let mut a = A::default();
    <A as AsMut<[T]>>::as_mut(&mut a).clone_from_slice(slice);
    a
}

pub fn xor(bits1: &[u8; 4], bits2: &[u8; 4]) -> [u8; 4] {
    let mut result = [0u8; 4];
    for i in 0..4 {
        result[i] = bits1[i] ^ bits2[i];
    }
    result
}

pub fn substitute_word(bytes: &[u8; 4]) -> [u8; 4] {
    let mut result = [0u8; 4];
    for i in 0..4 {
        result[i] = substitute(bytes[i], true);
    }
    result
}
fn inverse_substitute_word(bytes: &[u8; 4]) -> [u8; 4] {
    let mut result = [0u8; 4];
    for i in 0..4 {
        result[i] = substitute(bytes[i], false);
    }
    result
}

fn substitute_state(state: &mut [[u8; 4]; 4]) {
    for row in state.iter_mut().take(4) {
        for item in row.iter_mut().take(4) {
            *item = substitute(*item, true);
        }
    }
}

pub fn inverse_substitute_state(state: &mut [[u8; 4]; 4]) {
    for row in state.iter_mut().take(4) {
        for item in row.iter_mut().take(4) {
            *item = substitute(*item, false);
        }
    }
}

pub fn substitute(byte: u8, encryption: bool) -> u8 {
    let i: usize = ((byte >> 4) & 0xF).into();
    let j: usize = (byte & 0xF).into();
    if encryption {
        SBOX[i][j]
    } else {
        INVERSE_SBOX[i][j]
    }
}

pub fn shift_word(bytes: &[u8; 4]) -> [u8; 4] {
    let mut result = [0u8; 4];
    for i in 0..4 {
        result[i] = bytes[(i + 1) % 4];
    }
    result
}

fn inverse_shift_word(bytes: &[u8; 4]) -> [u8; 4] {
    let mut result = [0u8; 4];
    for i in 0..4 {
        result[i] = bytes[(i + 3) % 4];
    }
    result
}

fn shift_rows(state: &mut [[u8; 4]; 4]) {
    for i in 1..4 {
        let mut tmp = vec![0u8; i];
        tmp[..i].copy_from_slice(&state[i][..i]);
        for j in 0..4 - i {
            state[i][j] = state[i][j + i];
        }
        for j in 0..i {
            state[i][3 - j] = tmp[i - j - 1];
        }
    }
}

pub fn inverse_shift_rows(state: &mut [[u8; 4]; 4]) {
    for i in (1..4).rev() {
        let mut tmp = vec![0u8; i];
        tmp[..i].copy_from_slice(&state[4 - i][..i]);
        for j in 0..4 - i {
            state[4 - i][j] = state[4 - i][j + i];
        }
        for j in 0..i {
            state[4 - i][3 - j] = tmp[i - j - 1];
        }
    }
}

fn galois_multiplication(ap: u8, bp: u8) -> u8 {
    let mut p = 0u8;
    let mut high_bit;
    let mut a = ap;
    let mut b = bp;
    for _i in 0..8 {
        if b & 1 == 1 {
            p ^= a
        }
        high_bit = a & 0x80;
        a <<= 1;
        if high_bit == 0x80 {
            a ^= 0x1b;
        }
        b >>= 1;
    }
    p
}

fn mix_columns(state: &mut [[u8; 4]; 4]) {
    for i in 0..4 {
        let mut temp = [0u8; 4];
        for j in 0..4 {
            temp[j] = state[j][i];
        }
        state[0][i] = galois_multiplication(temp[0], 2)
            ^ galois_multiplication(temp[3], 1)
            ^ galois_multiplication(temp[2], 1)
            ^ galois_multiplication(temp[1], 3);
        state[1][i] = galois_multiplication(temp[1], 2)
            ^ galois_multiplication(temp[0], 1)
            ^ galois_multiplication(temp[3], 1)
            ^ galois_multiplication(temp[2], 3);
        state[2][i] = galois_multiplication(temp[2], 2)
            ^ galois_multiplication(temp[1], 1)
            ^ galois_multiplication(temp[0], 1)
            ^ galois_multiplication(temp[3], 3);
        state[3][i] = galois_multiplication(temp[3], 2)
            ^ galois_multiplication(temp[2], 1)
            ^ galois_multiplication(temp[1], 1)
            ^ galois_multiplication(temp[0], 3);
    }
}

fn inverse_mix_columns(state: &mut [[u8; 4]; 4]) {
    for i in 0..4 {
        let mut temp = [0u8; 4];
        for j in 0..4 {
            temp[j] = state[j][i];
        }

        state[0][i] = galois_multiplication(temp[0], 14)
            ^ galois_multiplication(temp[3], 9)
            ^ galois_multiplication(temp[2], 13)
            ^ galois_multiplication(temp[1], 11);
        state[1][i] = galois_multiplication(temp[1], 14)
            ^ galois_multiplication(temp[0], 9)
            ^ galois_multiplication(temp[3], 13)
            ^ galois_multiplication(temp[2], 11);
        state[2][i] = galois_multiplication(temp[2], 14)
            ^ galois_multiplication(temp[1], 9)
            ^ galois_multiplication(temp[0], 13)
            ^ galois_multiplication(temp[3], 11);
        state[3][i] = galois_multiplication(temp[3], 14)
            ^ galois_multiplication(temp[2], 9)
            ^ galois_multiplication(temp[1], 13)
            ^ galois_multiplication(temp[0], 11);
    }
}

pub fn add_round_key(state: &mut [[u8; 4]; 4], key: &[[u8; 4]; 4]) {
    // for i in 0..4 {
    //     for j in 0..4 {
    //         state[i][j] ^= key[j][i];
    //     }
    // }
    for (i, row) in state.iter_mut().enumerate().take(4) {
        for (j, item) in row.iter_mut().enumerate().take(4) {
            *item ^= key[j][i];
        }
    }
}

pub struct Aes {
    n_turn: usize,
    pub expanded_key: [[u8; 4]; 44],
}

impl Aes {
    pub fn new(&key: &[u8; 16], &n_turn: &usize) -> Self {
        Aes {
            n_turn,
            expanded_key: Self::key_schedule(&key),
        }
    }

    pub fn encrypt_block(&self, block: &[u8; 16]) -> [u8; 16] {
        let mut result = [0u8; 16];

        let mut state = [[0u8; 4]; 4];
        for i in 0..16 {
            state[i % 4][i / 4] = block[i];
        }

        add_round_key(&mut state, &clone_into_array(&self.expanded_key[0..4]));

        for i in 1..self.n_turn {
            substitute_state(&mut state);
            shift_rows(&mut state);
            mix_columns(&mut state);
            add_round_key(
                &mut state,
                &clone_into_array(&self.expanded_key[i * 4..(i + 1) * 4]),
            );
        }

        substitute_state(&mut state);
        shift_rows(&mut state);
        add_round_key(
            &mut state,
            &clone_into_array(&self.expanded_key[(self.n_turn) * 4..(self.n_turn + 1) * 4]),
        );

        for i in 0..4 {
            for j in 0..4 {
                result[4 * j + i] = state[i][j]
            }
        }
        result
    }

    pub fn encrypt_string(&self, text: &String) -> [u8; 16] {
        let mut block = [0u8; 16];
        let bytes = text.as_bytes();
        if bytes.len() > 16 {
            println!("string too long, will be cut.")
        }
        let len = bytes.len().min(16);

        block[..len].copy_from_slice(&bytes[..len]);

        self.encrypt_block(&block)
    }

    pub fn decrypt_block(&self, block: &[u8; 16]) -> [u8; 16] {
        let mut result = [0u8; 16];

        let mut state = [[0u8; 4]; 4];
        for i in 0..16 {
            state[i % 4][i / 4] = block[i];
        }

        add_round_key(
            &mut state,
            &clone_into_array(&self.expanded_key[(self.n_turn) * 4..(self.n_turn + 1) * 4]),
        );
        inverse_shift_rows(&mut state);
        inverse_substitute_state(&mut state);

        for i in (1..self.n_turn).rev() {
            add_round_key(
                &mut state,
                &clone_into_array(&self.expanded_key[i * 4..(i + 1) * 4]),
            );
            inverse_mix_columns(&mut state);
            inverse_shift_rows(&mut state);
            inverse_substitute_state(&mut state);
        }

        add_round_key(&mut state, &clone_into_array(&self.expanded_key[0..4]));

        for i in 0..4 {
            for j in 0..4 {
                result[4 * j + i] = state[i][j]
            }
        }

        result
    }

    pub fn decrypt_string(&self, text: &String) -> [u8; 16] {
        let mut block = [0u8; 16];
        let bytes = text.as_bytes();
        if bytes.len() > 16 {
            println!("string too long, will be cut.")
        }
        let len = bytes.len().min(16);

        block[..len].copy_from_slice(&bytes[..len]);

        self.decrypt_block(&block)
    }

    pub fn key_schedule(key_bytes: &[u8; 16]) -> [[u8; 4]; 44] {
        let mut original_key = [[0u8; 4]; 4];
        let mut expanded_key = [[0u8; 4]; 44];
        let n = 4;

        for i in 0..16 {
            original_key[i / 4][i % 4] = key_bytes[i];
        }

        // AES128, 11 rounds, i ∈ [0;4*rounds-1]
        for i in 0..44 {
            if i < n {
                expanded_key[i] = original_key[i];
            } else if i >= n && i % n == 0 {
                let mut rcon = [0u8; 4];
                rcon[0] = RC[i / n];
                expanded_key[i] = xor(
                    &xor(
                        &expanded_key[i - n],
                        &substitute_word(&shift_word(&expanded_key[i - 1])),
                    ),
                    &rcon,
                );
            } else {
                expanded_key[i] = xor(&expanded_key[i - n], &expanded_key[i - 1]);
            }
        }

        expanded_key
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    #[test]
    fn shift_word_test() {
        let test: u32 = 0x09cf4f3c;
        let test_bytes: [u8; 4] = test.to_be_bytes();
        let expected: u32 = 0xcf4f3c09;
        let expected_bytes: [u8; 4] = expected.to_be_bytes();
        assert_eq!(shift_word(&test_bytes), expected_bytes);
    }
    #[test]
    fn substitute_word_test() {
        let test: u32 = 0xcf4f3c09;
        let test_bytes: [u8; 4] = test.to_be_bytes();
        let expected: u32 = 0x8a84eb01;
        let expected_bytes: [u8; 4] = expected.to_be_bytes();
        assert_eq!(substitute_word(&test_bytes), expected_bytes);
    }
    #[test]
    fn xor_test() {
        let w1: [u8; 4] = [0x00, 0x11, 0x10, 0x10];
        let w2: [u8; 4] = [0x00, 0x11, 0x01, 0x10];
        let expected: [u8; 4] = [0x00, 0x00, 0x11, 0x00];
        assert_eq!(xor(&w1, &w2), expected);
    }
    #[test]
    fn key_schedule_test() {
        let key: [u8; 16] = [
            0x2b, 0x7e, 0x15, 0x16, 0x28, 0xae, 0xd2, 0xa6, 0xab, 0xf7, 0x15, 0x88, 0x09, 0xcf,
            0x4f, 0x3c,
        ];
        let expanded_key = Aes::key_schedule(&key);
        let expected: [u32; 44] = [
            0x2b7e1516, 0x28aed2a6, 0xabf71588, 0x09cf4f3c, 0xa0fafe17, 0x88542cb1, 0x23a33939,
            0x2a6c7605, 0xf2c295f2, 0x7a96b943, 0x5935807a, 0x7359f67f, 0x3d80477d, 0x4716fe3e,
            0x1e237e44, 0x6d7a883b, 0xef44a541, 0xa8525b7f, 0xb671253b, 0xdb0bad00, 0xd4d1c6f8,
            0x7c839d87, 0xcaf2b8bc, 0x11f915bc, 0x6d88a37a, 0x110b3efd, 0xdbf98641, 0xca0093fd,
            0x4e54f70e, 0x5f5fc9f3, 0x84a64fb2, 0x4ea6dc4f, 0xead27321, 0xb58dbad2, 0x312bf560,
            0x7f8d292f, 0xac7766f3, 0x19fadc21, 0x28d12941, 0x575c006e, 0xd014f9a8, 0xc9ee2589,
            0xe13f0cc8, 0xb6630ca6,
        ];
        let mut expected_bytes: [u8; 4];
        for i in 0..11 {
            for j in 0..4 {
                let index = i * 4 + j;
                expected_bytes = expected[index].to_be_bytes();
                for (&byte, &expected_byte) in expanded_key[index].iter().zip(expected_bytes.iter())
                {
                    assert_eq!(byte, expected_byte);
                }
            }
        }
    }
    #[test]
    fn encrypt_test() {
        let cleartext: [u8; 16] = [
            0x32, 0x43, 0xf6, 0xa8, 0x88, 0x5a, 0x30, 0x8d, 0x31, 0x31, 0x98, 0xa2, 0xe0, 0x37,
            0x07, 0x34,
        ];
        let key: [u8; 16] = [
            0x2b, 0x7e, 0x15, 0x16, 0x28, 0xae, 0xd2, 0xa6, 0xab, 0xf7, 0x15, 0x88, 0x09, 0xcf,
            0x4f, 0x3c,
        ];
        let nturn = 10;
        let aescipher = Aes::new(&key, &nturn);
        let ciphertext: [u8; 16] = aescipher.encrypt_block(&cleartext);
        let expected_ciphertext: [u8; 16] = [
            0x39, 0x25, 0x84, 0x1d, 0x02, 0xdc, 0x09, 0xfb, 0xdc, 0x11, 0x85, 0x97, 0x19, 0x6a,
            0x0b, 0x32,
        ];
        assert_eq!(ciphertext, expected_ciphertext);
    }
    #[test]
    fn decrypt_test() {
        let ciphertext: [u8; 16] = [
            0x39, 0x25, 0x84, 0x1d, 0x02, 0xdc, 0x09, 0xfb, 0xdc, 0x11, 0x85, 0x97, 0x19, 0x6a,
            0x0b, 0x32,
        ];
        let key: [u8; 16] = [
            0x2b, 0x7e, 0x15, 0x16, 0x28, 0xae, 0xd2, 0xa6, 0xab, 0xf7, 0x15, 0x88, 0x09, 0xcf,
            0x4f, 0x3c,
        ];
        let nturn = 10;
        let aescipher = Aes::new(&key, &nturn);
        let cleartext: [u8; 16] = aescipher.decrypt_block(&ciphertext);
        let expected_cleartext: [u8; 16] = [
            0x32, 0x43, 0xf6, 0xa8, 0x88, 0x5a, 0x30, 0x8d, 0x31, 0x31, 0x98, 0xa2, 0xe0, 0x37,
            0x07, 0x34,
        ];
        assert_eq!(cleartext, expected_cleartext);
    }
    #[test]
    fn inverse_shift_word_test() {
        let word: [u8; 4] = [0x01, 0x34, 0x76, 0x49];
        let inverse_word: [u8; 4] = shift_word(&word);
        assert_eq!(word, inverse_shift_word(&inverse_word));
    }
    #[test]
    fn inverse_substitute_word_test() {
        let word: [u8; 4] = [0x01, 0x34, 0x76, 0x49];
        let inverse_word: [u8; 4] = substitute_word(&word);
        assert_eq!(word, inverse_substitute_word(&inverse_word));
    }
}