Function build_sidecar

nodedb-query/src/msgpack_scan/sidecar.rs:132–184

Build a sidecar-indexed document from raw msgpack bytes. Scans the map keys once and appends the sidecar index. Returns `None` if the buffer is not a valid msgpack map or has too many fields (> u16::MAX). Entries are stored sorted by `field_hash` to enable binary search.

(msgpack: &[u8])

/// the buffer is not a valid msgpack map or has too many fields (> u16::MAX).
/// Entries are stored sorted by `field_hash` to enable binary search.
pub fn build_sidecar(msgpack: &[u8]) -> Option<Vec<u8>> {
    let (count, mut pos) = map_header(msgpack, 0)?;
    if count > u16::MAX as usize {
        return None;
    }

    let mut entries: Vec<SidecarEntry> = Vec::with_capacity(count);

    for _ in 0..count {
        // Read key string bounds (start of str content, length of content).
        let (key_data_start, key_data_len) = str_bounds(msgpack, pos)?;
        let key_bytes = msgpack.get(key_data_start..key_data_start + key_data_len)?;
        let field_hash = fnv1a_hash(key_bytes);

        // Skip over the key to reach the value.
        pos = skip_value(msgpack, pos)?;
        let value_start = pos;
        let value_end = skip_value(msgpack, pos)?;

        let value_len = value_end.checked_sub(value_start)?;
        if value_start > u32::MAX as usize || value_len > u32::MAX as usize {
            return None;
        }

        entries.push(SidecarEntry {
            field_hash,
            value_offset: value_start as u32,
            value_len: value_len as u32,
        });

        pos = value_end;
    }

    // Sort by hash to allow binary search during lookup.
    entries.sort_unstable_by_key(|e| e.field_hash);

    let entry_count = entries.len();
    let total_len = msgpack.len() + entry_count * ENTRY_SIZE + TRAILER_SIZE;
    let mut out = Vec::with_capacity(total_len);
    out.extend_from_slice(msgpack);

    for e in &entries {
        out.extend_from_slice(&e.field_hash.to_le_bytes());
        out.extend_from_slice(&e.value_offset.to_le_bytes());
        out.extend_from_slice(&e.value_len.to_le_bytes());
    }

    // Trailer: entry_count (u16 LE) + magic (u32 LE).
    out.extend_from_slice(&(entry_count as u16).to_le_bytes());
    out.extend_from_slice(&SIDECAR_MAGIC_LE);

    Some(out)
}

/// Look up a field's byte range `(start, end)` using the sidecar index.
///