BlueArchiveToolkit/crates/bat-patch/src/binary.rs

//! Deterministic binary hunk patch.

use crate::PatchError;
use serde::{Deserialize, Serialize};

/// Current binary patch schema version.
pub const BINARY_PATCH_VERSION: u32 = 1;

/// Binary patch made of deterministic copy/insert hunks.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub struct BinaryPatch {
    /// Patch schema version.
    pub version: u32,
    /// Expected BLAKE3 hash of the source bytes.
    pub source_blake3: String,
    /// Expected BLAKE3 hash of the target bytes.
    pub target_blake3: String,
    /// Source byte length.
    pub source_size: u64,
    /// Target byte length.
    pub target_size: u64,
    /// Ordered hunks.
    pub hunks: Vec<BinaryPatchHunk>,
}

/// One binary patch hunk.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
#[serde(tag = "kind", rename_all = "snake_case")]
pub enum BinaryPatchHunk {
    /// Copy a byte range from the source.
    Copy {
        /// Source offset.
        offset: u64,
        /// Number of bytes to copy.
        length: u64,
    },
    /// Insert literal bytes.
    Insert {
        /// Literal bytes.
        bytes: Vec<u8>,
    },
}

/// Creates a deterministic hunk patch.
///
/// The first implementation optimizes for correctness and stable output. It
/// emits copy hunks for equal runs and insert hunks for changed runs; more
/// compact suffix/prefix matching can be added later without changing the
/// manifest/integrity contract.
pub fn diff(old: &[u8], new: &[u8]) -> BinaryPatch {
    let mut hunks = Vec::new();
    let mut index = 0usize;
    while index < new.len() {
        if index < old.len() && old[index] == new[index] {
            let start = index;
            while index < new.len() && index < old.len() && old[index] == new[index] {
                index += 1;
            }
            hunks.push(BinaryPatchHunk::Copy {
                offset: start as u64,
                length: (index - start) as u64,
            });
            continue;
        }

        let start = index;
        while index < new.len() && (index >= old.len() || old[index] != new[index]) {
            index += 1;
        }
        hunks.push(BinaryPatchHunk::Insert {
            bytes: new[start..index].to_vec(),
        });
    }

    BinaryPatch {
        version: BINARY_PATCH_VERSION,
        source_blake3: blake3_hex(old),
        target_blake3: blake3_hex(new),
        source_size: old.len() as u64,
        target_size: new.len() as u64,
        hunks,
    }
}

/// Applies a structured binary patch.
pub fn apply_binary_patch(old: &[u8], patch: &BinaryPatch) -> crate::Result<Vec<u8>> {
    if patch.version != BINARY_PATCH_VERSION {
        return Err(PatchError::ApplyFailed(format!(
            "unsupported binary patch version {}",
            patch.version
        )));
    }
    if patch.source_size != old.len() as u64 || patch.source_blake3 != blake3_hex(old) {
        return Err(PatchError::ApplyFailed(
            "binary patch source integrity mismatch".to_string(),
        ));
    }

    let target_capacity = usize::try_from(patch.target_size)
        .map_err(|_| PatchError::ApplyFailed("binary patch target too large".to_string()))?;
    let mut output = Vec::with_capacity(target_capacity);
    for hunk in &patch.hunks {
        match hunk {
            BinaryPatchHunk::Copy { offset, length } => {
                let start = usize::try_from(*offset).map_err(|_| {
                    PatchError::ApplyFailed("binary patch copy offset overflow".to_string())
                })?;
                let length = usize::try_from(*length).map_err(|_| {
                    PatchError::ApplyFailed("binary patch copy length overflow".to_string())
                })?;
                let end = start.checked_add(length).ok_or_else(|| {
                    PatchError::ApplyFailed("binary patch copy range overflow".to_string())
                })?;
                let bytes = old.get(start..end).ok_or_else(|| {
                    PatchError::ApplyFailed(format!(
                        "binary patch copy range {start}..{end} exceeds source {}",
                        old.len()
                    ))
                })?;
                output.extend_from_slice(bytes);
            }
            BinaryPatchHunk::Insert { bytes } => output.extend_from_slice(bytes),
        }
    }

    if output.len() as u64 != patch.target_size || blake3_hex(&output) != patch.target_blake3 {
        return Err(PatchError::ApplyFailed(
            "binary patch target integrity mismatch".to_string(),
        ));
    }
    Ok(output)
}

/// Serializes and applies a binary patch.
pub fn apply_patch(old: &[u8], patch: &[u8]) -> crate::Result<Vec<u8>> {
    let patch: BinaryPatch = serde_json::from_slice(patch)
        .map_err(|error| PatchError::ApplyFailed(format!("invalid binary patch JSON: {error}")))?;
    apply_binary_patch(old, &patch)
}

fn blake3_hex(bytes: &[u8]) -> String {
    blake3::hash(bytes).to_hex().to_string()
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn binary_patch_round_trips_changed_bytes() {
        let old = b"abcdef012345";
        let new = b"abcXYZ012345!";
        let patch = diff(old, new);
        let patch_json = serde_json::to_vec(&patch).unwrap();

        assert_eq!(apply_binary_patch(old, &patch).unwrap(), new);
        assert_eq!(apply_patch(old, &patch_json).unwrap(), new);
        assert!(patch
            .hunks
            .iter()
            .any(|hunk| matches!(hunk, BinaryPatchHunk::Insert { .. })));
    }

    #[test]
    fn binary_patch_rejects_wrong_source() {
        let patch = diff(b"old", b"new");
        let error = apply_binary_patch(b"bad", &patch).unwrap_err();
        assert!(matches!(error, crate::PatchError::ApplyFailed(_)));
    }
}