utils.data_io.save_h5

utils.data_io.save_h5

hdf5_arraylist.py

Utilities to store and retrieve a Python list of NumPy arrays with variable shapes in a single HDF5 file (chunked + compressed).

Dependencies

• numpy
• h5py

Classes

Name	Description
H5ArrayListSpec	Naming spec for datasets in the group.

H5ArrayListSpec

utils.data_io.save_h5.H5ArrayListSpec(
    self,
    group='solutions',
    dataset_prefix='sol',
    digits=6,
)

Naming spec for datasets in the group.

Functions

Name	Description
iter_array_list_h5	Iterate arrays one-by-one (reduces peak RAM versus loading all at once).
list_keys_h5	Return dataset keys under the target group.
load_array_h5	Load a single array by index.
load_array_list_h5	Load the full list into memory.
load_metadata_h5	Load optional metadata dict stored at save time; returns {} if absent.
save_array_list_h5	Save a list of arrays (variable shapes allowed) into a single HDF5 file.

iter_array_list_h5

utils.data_io.save_h5.iter_array_list_h5(filepath, *, spec=H5ArrayListSpec())

Iterate arrays one-by-one (reduces peak RAM versus loading all at once).

list_keys_h5

utils.data_io.save_h5.list_keys_h5(filepath, *, spec=H5ArrayListSpec())

Return dataset keys under the target group.

load_array_h5

utils.data_io.save_h5.load_array_h5(filepath, index, *, spec=H5ArrayListSpec())

Load a single array by index.

load_array_list_h5

utils.data_io.save_h5.load_array_list_h5(filepath, *, spec=H5ArrayListSpec())

Load the full list into memory.

load_metadata_h5

utils.data_io.save_h5.load_metadata_h5(filepath, *, spec=H5ArrayListSpec())

Load optional metadata dict stored at save time; returns {} if absent.

save_array_list_h5

utils.data_io.save_h5.save_array_list_h5(
    filepath,
    arrays,
    *,
    spec=H5ArrayListSpec(),
    compression='gzip',
    compression_opts=4,
    chunks=True,
    overwrite=True,
    metadata=None,
)

Save a list of arrays (variable shapes allowed) into a single HDF5 file.

Parameters

Name	Type	Description	Default
filepath	Union[str, 'Path']	Target .h5/.hdf5 path.	required
arrays	Sequence[ArrayLike]	Sequence of array-like objects. Each entry is converted by np.asarray.	required
spec	H5ArrayListSpec	Group name and dataset naming scheme.	H5ArrayListSpec()
compression	Optional[str]	HDF5 compression filter (e.g., “gzip”, “lzf”, None).	'gzip'
compression_opts	Optional[int]	Compression level for gzip (0-9). Ignored for None and some filters.	4
chunks	Union[bool, tuple, None]	Use chunking (True/False) or pass an explicit chunk shape tuple. For compression, chunking must be enabled.	True
overwrite	bool	If True, recreate the file. If False, update/replace the target group.	True
metadata	Optional[Dict[str, Any]]	Optional dict stored as JSON in group attrs (“metadata_json”).	None

Notes

• Do not load untrusted HDF5 files if your workflow executes code based on stored metadata. The array payload itself is data-only.