vecs.md

vecs file format

A binary format to store the vectors.

| dim (4 bytes) | vector (4 * dim bytes) |
| dim (4 bytes) | vector (4 * dim bytes) |
...
| dim (4 bytes) | vector (4 * dim bytes) |

vecs.py

from struct import unpack, pack
import numpy as np


def read_vec(filepath: str, vec_type: np.dtype = np.float32):
    """Read vectors from a file. Support `fvecs`, `ivecs` and `bvecs` format.

    Args:
        filepath: The path of the file.
        vec_type: The type of the vectors.
    """
    size = np.dtype(vec_type).itemsize
    with open(filepath, "rb") as f:
        vecs = []
        while True:
            try:
                buf = f.read(4)
                if len(buf) == 0:
                    break
                dim = unpack("<i", buf)[0]
                vecs.append(np.frombuffer(f.read(dim * size), dtype=vec_type))
            except Exception as err:
                print(err)
                break
    return np.array(vecs)
  
  
def write_vec(filepath: str, vecs: np.ndarray, vec_type: np.dtype = np.float32):
    """Write vectors to a file. Support `fvecs`, `ivecs` and `bvecs` format."""
    with open(filepath, "wb") as f:
        for vec in vecs:
            f.write(pack("<i", len(vec)))
            f.write(vec.tobytes())

vecs.rs

use std::fs::File;
use std::io::{BufReader, BufWriter, Read, Write};
use std::path::Path;

use num_traits::{FromBytes, ToBytes};


/// Read the fvces/ivces file.
pub fn read_vecs<T>(path: &Path) -> std::io::Result<Vec<Vec<T>>>
where
    T: Sized + FromBytes<Bytes = [u8; 4]>,
{
    let file = File::open(path)?;
    let mut reader = BufReader::new(file);
    let mut buf = [0u8; 4];
    let mut count: usize;
    let mut vecs = Vec::new();
    loop {
        count = reader.read(&mut buf)?;
        if count == 0 {
            break;
        }
        let dim = u32::from_le_bytes(buf) as usize;
        let mut vec = Vec::with_capacity(dim);
        for _ in 0..dim {
            reader.read_exact(&mut buf)?;
            vec.push(T::from_le_bytes(&buf));
        }
        vecs.push(vec);
    }
    Ok(vecs)
}

/// Write the fvecs/ivecs file.
pub fn write_vecs<T>(path: &Path, vecs: &[&Vec<T>]) -> std::io::Result<()>
where
    T: Sized + ToBytes,
{
    let file = File::create(path)?;
    let mut writer = BufWriter::new(file);
    for vec in vecs.iter() {
        writer.write_all(&(vec.len() as u32).to_le_bytes())?;
        for v in vec.iter() {
            writer.write_all(T::to_le_bytes(v).as_ref())?;
        }
    }
    writer.flush()?;
    Ok(())
}