| Crates.io | blosc2 |
| lib.rs | blosc2 |
| version | 0.2.2 |
| created_at | 2025-07-18 13:17:56.25896+00 |
| updated_at | 2025-10-15 08:43:04.248495+00 |
| description | Safe Rust bindings for blosc2 - a fast, compressed, persistent binary data store library |
| homepage | |
| repository | https://github.com/barakugav/blosc2-rs |
| max_upload_size | |
| id | 1758947 |
| size | 322,295 |
Rust bindings for blosc2 - a fast, compressed, persistent binary data store library.
Provide a safe interface to the blosc2 library, which is a new major version of the original blosc library.
The library provides Ndarray, an n-dimensional array implementation with compressed
storage, and some lower level utilities for working with compressed data such as Chunk and SChunk.
To use this library, add the following to your Cargo.toml:
[dependencies]
blosc2 = "0.2"
The Ndarray is an n-dimensional array with compressed storage, that support random access
to items or slices.
In the following example, we create a new Ndarray from an array from the ndarray crate, and than access its
elements and slice:
use blosc2::nd::{Ndarray, NdarrayParams};
let arr = Ndarray::from_ndarray(
&ndarray::array!([1_i32, 2, 3], [4, 5, 6], [7, 8, 9]),
NdarrayParams::default()
.chunkshape(Some(&[2, 2]))
.blockshape(Some(&[1, 1])),
).unwrap();
assert_eq!(4, arr.get::<i32>(&[1, 0]).unwrap());
assert_eq!(9, arr.get::<i32>(&[2, 2]).unwrap());
let slice_arr: ndarray::Array2<i32> = arr.slice(&[0..2, 1..3]).unwrap();
assert_eq!(slice_arr, ndarray::array![[2, 3], [5, 6]]);
The library provides a high level Ndarray struct, which is built upon the
SChunk, that in itself is built upon Chunk.
Most users will probably interact only with the Ndarray, but the chunks types are also available.
zlib: Enable support for the zlib compression codec.zstd: Enable support for the zstd compression codec.ndarray: Enable conversions between blosc2's Ndarray and the ndarray crate's
ArrayBase types.half: Add a dependency to the half crate, and implement the Dtyped trait for
half::f16. See the util module for the info.num-complex: Add a dependency to the num-complex crate, and implement the Dtyped trait
for num_complex::Complex<f32> and num_complex::Complex<f64>. See the util module for the info.The library follow the C API and returns error codes. In addition, if the environment variable
BLOSC_TRACE is set, it will print detailed trace during failures which is useful for
debugging.