fields.buffers.core.HDF5Buffer.from_array#

classmethod HDF5Buffer.from_array(obj: Any, file: str | Path | File | None = None, path: str | None = None, *args, dtype: Any | None = None, units: Unit | str | None = None, **kwargs) → HDF5Buffer[source]#

Construct an HDF5-backed buffer from an array-like object or existing HDF5 dataset.

This method wraps either:

  • an existing h5py.Dataset, or

  • arbitrary in-memory data by creating a new dataset in the specified HDF5 file.

Parameters:

  • obj (array-like or Dataset) – The data to store. If already a h5py.Dataset, it is wrapped directly. Otherwise, it is coerced to a unyt_array and stored in a new dataset.

  • file (str, Path, or File, optional) – Path to the HDF5 file or an open file object. Required if obj is not already a dataset.

  • path (str, optional) – Internal path for the new dataset in the HDF5 file. Required if obj is not already a dataset.

  • dtype (data-type, optional) – Desired data type of the dataset. Defaults to the dtype inferred from obj.

  • units (str or Unit, optional) – Units to attach to the dataset. If obj already has units, this overrides them.

  • overwrite (bool, default False) – If True, any existing dataset at the given path will be deleted and replaced.

  • **kwargs – Additional keyword arguments forwarded to create_dataset().

Returns:

A lazily-loaded buffer backed by an HDF5 dataset.

Return type:

HDF5Buffer

Raises:

  • ValueError – If file or path is not specified when required.

  • TypeError – If input types are not supported or invalid.

See also

HDF5Buffer.create_dataset

Core method used for new dataset creation.

buffer_from_array

Registry-based entry point to buffer resolution.

On this page

This Page