HDF5

This module contains HDF5Writer for writing HDF5 datasets.

morphocut.hdf5.HDF5Writer(file_name: str, data: Mapping[str, Any | Variable[Any]], *, file_mode='w', dataset_mode: Literal['create', 'append', 'extend'] | Variable[Literal['create', 'append', 'extend']] = 'create', compression=None, chunk_size: int | None = None)[source]

Write arrays to a HDF5 file.

Parameters:

  • file_name (str) – Location of the output file.

  • data (Mapping or Variable) – Data to write (name => array).

  • file_mode (str, optional) – Opening mode of the HDF5 file.

  • dataset_mode (str, optional) – Dataset behavior.

    “append”: Append an array of shape \((*)\) to a dataset of shape \((N,*)\).

    The data must have the same shape as the existing data, without the first dimension.

    “extend”: Extend a dataset of shape \((N,*)\) with an array of shape \((M,*)\).

    The data must have the same shape as the existing data, except in the first dimension.

  • compression – Compression strategy. See h5py.Group.create_dataset().

  • chunk_size – Chunk size. See h5py.Group.create_dataset().

Example

with Pipeline() as pipeline:
    image_fn = ...
    image = ImageReader(image_fn)
    HDF5Writer("path/to/file.h5", {image_fn: image})
pipeline.transform_stream()