numcodecs.ZFPY() fails

[afj-qeye]

Zarr version

v3.1.3

Numcodecs version

v0.16.1

Python Version

3.13.7

Operating System

Linux

Installation

using conda

Description

Using numcodecs.Zstd() works, but not numcodecs.ZFPY() which produces the following error message:

Traceback (most recent call last):
  File "zarr_compression.py", line 4, in <module>
    array = zarr.create_array(
        store="/tmp/data.zarr",
    ...<3 lines>...
        compressors=[numcodecs.ZFPY()],
        overwrite=True)
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/api/synchronous.py", line 945, in create_array
    sync(
    ~~~~^
        zarr.core.array.create_array(
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ...<20 lines>...
        )
        ^
    )
    ^
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/sync.py", line 163, in sync
    raise return_result
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/sync.py", line 119, in _runner
    return await coro
           ^^^^^^^^^^
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/array.py", line 4786, in create_array
    return await init_array(
           ^^^^^^^^^^^^^^^^^
    ...<16 lines>...
    )
    ^
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/array.py", line 4572, in init_array
    array_array, array_bytes, bytes_bytes = _parse_chunk_encoding_v3(
                                            ~~~~~~~~~~~~~~~~~~~~~~~~^
        compressors=compressors,
        ^^^^^^^^^^^^^^^^^^^^^^^^
    ...<2 lines>...
        dtype=zdtype,
        ^^^^^^^^^^^^^
    )
    ^
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/array.py", line 5108, in _parse_chunk_encoding_v3
    out_bytes_bytes = tuple(_parse_bytes_bytes_codec(c) for c in maybe_bytes_bytes)
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/core/array.py", line 5108, in <genexpr>
    out_bytes_bytes = tuple(_parse_bytes_bytes_codec(c) for c in maybe_bytes_bytes)
                            ~~~~~~~~~~~~~~~~~~~~~~~~^^^
  File "/anaconda3/2025.06/envs/qeye3/lib/python3.13/site-packages/zarr/registry.py", line 214, in _parse_bytes_bytes_codec
    raise TypeError(f"Expected a BytesBytesCodec. Got {type(data)} instead.")
TypeError: Expected a BytesBytesCodec. Got <class 'zarr.codecs.numcodecs._codecs.ZFPY'> instead.

Steps to reproduce

import zarr
import zarr.codecs.numcodecs as numcodecs

array = zarr.create_array(
    store="/tmp/data.zarr",
    shape=(1024, 1024),
    chunks=(64, 64),
    dtype="float32",
    compressors=[numcodecs.ZFPY()],
    overwrite=True)

Additional output

No response

[d-v-b]

Our error message should be better here. zfpy is a array-bytes codec, so try declaring it via the serializer parameter of create_array:

# /// script
# requires-python = ">=3.11"
# dependencies = [
#   "zarr@git+https://github.com/zarr-developers/zarr-python.git@main",
#   "zfpy"
# ]
# ///

import zarr
import zarr.codecs.numcodecs as numcodecs

array = zarr.create_array(
    store={},
    shape=(1024, 1024),
    chunks=(64, 64),
    dtype="float32",
    serializer=numcodecs.ZFPY(),
    overwrite=True)```

[afj-qeye]

Writing data gives the following traceback with error: numcodecs.errors.UnknownCodecError: codec not available: ''zfpy''.

And in the numcodecs/zfpy.py I found the problem. It appears the problem is that there is no version info to be found in the zfpy module, see #LLNL/zfp#280. Thanks for the help.

if _zfpy_version:
    # Check NumPy version
    _numpy_version: tuple = tuple(map(int, version("numpy").split('.')))
    if _numpy_version >= (2, 0, 0) and _zfpy_version < (1, 0, 1):  # pragma: no cover
        _zfpy_version = ()
        warnings.warn(
            "NumPy version >= 2.0.0 detected. The zfpy library is incompatible with this version of NumPy. "
            "Please downgrade to NumPy < 2.0.0 or wait for an update from zfpy.",
            UserWarning,
            stacklevel=2,
        )
    else:
        with suppress(ImportError):
            import zfpy as _zfpy  # type: ignore[no-redef]

[afj-qeye]

I have hacked the above numcodecs to load the zfpy version 1.0.1 and using serializer=numcodecs.ZFPY()

import numpy as np
import zarr
import zarr.codecs.numcodecs as numcodecs

array = zarr.create_array(
    store="/qdata/tmp/data.zarr",
    shape=(1024, 1024),
    chunks=(64, 64),
    dtype="float32",
    serializer=numcodecs.ZFPY(),
    overwrite=True)

array[:] = 1.0 # OK
array[:] = np.ones(array.shape, dtype="float32")  # FAILS (despite being C_CONTIGUOUS)

ValueError: The zfp codec does not support F order arrays. Your arrays flags were   C_CONTIGUOUS : False
  F_CONTIGUOUS : False
  OWNDATA : False
  WRITEABLE : True
  ALIGNED : True
  WRITEBACKIFCOPY : False