tensogram 0.21.0

Fast binary N-tensor message format for scientific data

Tensogram Python Bindings

Python bindings for the Tensogram N-tensor message format.

Native extension built with PyO3 + maturin. Supports NumPy arrays, async I/O, and GIL-free operation on free-threaded Python (3.13t / 3.14t).

Installation

pip install tensogram
# or, with the xarray and Zarr backends:
pip install tensogram[all]

Usage

import numpy as np
import tensogram

data = np.random.randn(100, 200).astype(np.float32)
msg = tensogram.encode(
    {"version": 3},
    [(
        {"type": "ntensor", "shape": [100, 200], "dtype": "float32",
         "compression": "szip"},
        data,
    )],
)
result = tensogram.decode(msg)
arr = result.objects[0][1]   # numpy array

Features

• NumPy integration across every supported dtype (float / complex / int / uint, plus bitmask and bfloat16)
• Sync and async file APIs (TensogramFile / AsyncTensogramFile)
• GIL-free parallel encode / decode on free-threaded Python
• Partial-range decode (decode_range)
• Full codec support: szip, zstd, lz4, blosc2, zfp, sz3
• Validation (tensogram.validate, tensogram.validate_file)
• GRIB / NetCDF conversion (when the wheel is built with the matching feature)

Documentation

• Full documentation: https://sites.ecmwf.int/docs/tensogram/main/
• Python user guide: https://sites.ecmwf.int/docs/tensogram/main/guide/python-api.html
• Repository: https://github.com/ecmwf/tensogram

License

Copyright 2026- ECMWF. Licensed under Apache-2.0. See LICENSE.