//! Dependency-free, `no_std`, deterministic SHA-256 (FIPS 180-4).
//!
//! WHY A HAND-ROLLED SHA-256. This crate is the offline-testable REFERENCE model
//! of the per-validator inner logic the aggregation zkVM guest runs. Keeping it at
//! zero external dependencies means `cargo test` verifies the RFC 8391 XMSS core
//! against the official known-answer vector with no network and no toolchain
//! surprises, on any machine.
//!
//! WHAT THE REAL SP1 GUEST USES INSTEAD. Inside the SP1 zkVM the production guest
//! would replace this module with the SP1-patched `sha2` crate (the accelerated
//! SHA-256 precompile), exactly as `../zk-light-client/guest/Cargo.toml` patches
//! `sha2` and `k256`. The BYTES are identical either way, so this module is a
//! faithful stand-in for the circuit's hash gate; only the in-circuit cost differs.
//! This is single-block-friendly, allocation-free, and float-free by construction.

const H0: [u32; 8] = [
    0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19,
];

const K: [u32; 64] = [
    0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
    0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
    0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
    0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
    0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
    0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
    0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
    0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2,
];

/// Incremental SHA-256 state. `no_std`, no allocation, deterministic.
pub struct Sha256 {
    h: [u32; 8],
    buf: [u8; 64],
    buf_len: usize,
    len_bits: u64,
}

impl Sha256 {
    #[inline]
    pub fn new() -> Self {
        Sha256 { h: H0, buf: [0u8; 64], buf_len: 0, len_bits: 0 }
    }

    pub fn update(&mut self, mut data: &[u8]) {
        self.len_bits = self.len_bits.wrapping_add((data.len() as u64) * 8);
        // Fill any partial buffer first.
        if self.buf_len > 0 {
            let take = core::cmp::min(64 - self.buf_len, data.len());
            self.buf[self.buf_len..self.buf_len + take].copy_from_slice(&data[..take]);
            self.buf_len += take;
            data = &data[take..];
            if self.buf_len == 64 {
                let block = self.buf;
                self.compress(&block);
                self.buf_len = 0;
            }
        }
        // Compress full 64-byte blocks straight from the input.
        while data.len() >= 64 {
            let mut block = [0u8; 64];
            block.copy_from_slice(&data[..64]);
            self.compress(&block);
            data = &data[64..];
        }
        // Stash the remainder.
        if !data.is_empty() {
            self.buf[..data.len()].copy_from_slice(data);
            self.buf_len = data.len();
        }
    }

    pub fn finalize(mut self) -> [u8; 32] {
        let len_bits = self.len_bits;
        // Padding: 0x80, then zeros, then the 64-bit big-endian bit length.
        let mut pad = [0u8; 72];
        pad[0] = 0x80;
        // total padded region so that (buf_len + 1 + zeros + 8) % 64 == 0
        let pad_zeros = if self.buf_len < 56 { 56 - self.buf_len } else { 120 - self.buf_len };
        let total = pad_zeros + 8; // includes the 0x80 byte within pad_zeros count below
        // Note: pad_zeros here already counts the 0x80 byte position; write length at the end.
        pad[pad_zeros..pad_zeros + 8].copy_from_slice(&len_bits.to_be_bytes());
        self.update_no_len(&pad[..total]);

        let mut out = [0u8; 32];
        for (i, word) in self.h.iter().enumerate() {
            out[i * 4..i * 4 + 4].copy_from_slice(&word.to_be_bytes());
        }
        out
    }

    /// Same as `update` but does NOT advance the tracked message length (used only
    /// to feed the precomputed padding block in `finalize`).
    fn update_no_len(&mut self, mut data: &[u8]) {
        if self.buf_len > 0 {
            let take = core::cmp::min(64 - self.buf_len, data.len());
            self.buf[self.buf_len..self.buf_len + take].copy_from_slice(&data[..take]);
            self.buf_len += take;
            data = &data[take..];
            if self.buf_len == 64 {
                let block = self.buf;
                self.compress(&block);
                self.buf_len = 0;
            }
        }
        while data.len() >= 64 {
            let mut block = [0u8; 64];
            block.copy_from_slice(&data[..64]);
            self.compress(&block);
            data = &data[64..];
        }
        if !data.is_empty() {
            self.buf[..data.len()].copy_from_slice(data);
            self.buf_len = data.len();
        }
    }

    fn compress(&mut self, block: &[u8; 64]) {
        let mut w = [0u32; 64];
        for i in 0..16 {
            w[i] = u32::from_be_bytes([
                block[i * 4],
                block[i * 4 + 1],
                block[i * 4 + 2],
                block[i * 4 + 3],
            ]);
        }
        for i in 16..64 {
            let s0 = w[i - 15].rotate_right(7) ^ w[i - 15].rotate_right(18) ^ (w[i - 15] >> 3);
            let s1 = w[i - 2].rotate_right(17) ^ w[i - 2].rotate_right(19) ^ (w[i - 2] >> 10);
            w[i] = w[i - 16]
                .wrapping_add(s0)
                .wrapping_add(w[i - 7])
                .wrapping_add(s1);
        }

        let mut a = self.h[0];
        let mut b = self.h[1];
        let mut c = self.h[2];
        let mut d = self.h[3];
        let mut e = self.h[4];
        let mut f = self.h[5];
        let mut g = self.h[6];
        let mut hh = self.h[7];

        for i in 0..64 {
            let s1 = e.rotate_right(6) ^ e.rotate_right(11) ^ e.rotate_right(25);
            let ch = (e & f) ^ ((!e) & g);
            let t1 = hh
                .wrapping_add(s1)
                .wrapping_add(ch)
                .wrapping_add(K[i])
                .wrapping_add(w[i]);
            let s0 = a.rotate_right(2) ^ a.rotate_right(13) ^ a.rotate_right(22);
            let maj = (a & b) ^ (a & c) ^ (b & c);
            let t2 = s0.wrapping_add(maj);
            hh = g;
            g = f;
            f = e;
            e = d.wrapping_add(t1);
            d = c;
            c = b;
            b = a;
            a = t1.wrapping_add(t2);
        }

        self.h[0] = self.h[0].wrapping_add(a);
        self.h[1] = self.h[1].wrapping_add(b);
        self.h[2] = self.h[2].wrapping_add(c);
        self.h[3] = self.h[3].wrapping_add(d);
        self.h[4] = self.h[4].wrapping_add(e);
        self.h[5] = self.h[5].wrapping_add(f);
        self.h[6] = self.h[6].wrapping_add(g);
        self.h[7] = self.h[7].wrapping_add(hh);
    }
}

impl Default for Sha256 {
    fn default() -> Self {
        Self::new()
    }
}

/// One-shot SHA-256 over `data`.
#[inline]
pub fn sha256(data: &[u8]) -> [u8; 32] {
    let mut h = Sha256::new();
    h.update(data);
    h.finalize()
}