jangada.serialization.Persistable._load_data_from_h5py_tree

static Persistable._load_data_from_h5py_tree(value: Any, use_proxy_dataset: bool = False) → Any

Recursively load data from HDF5 structure.

Reconstructs Python objects from HDF5 groups, datasets, and attributes.

Parameters:

valueAny

HDF5 object to load (Group, Dataset, or attribute value).

use_proxy_datasetbool, optional

If True, wrap datasets in ProxyDataset for lazy loading. If False, load datasets fully. Default is False.

Returns:

Any

Reconstructed Python object.

Raises:

ValueError

If __container_type__ has an unknown value.

Notes

This method reverses the mapping from _save_data_in_group: - ‘NoneType:None’ → None - ‘Path:…’ → Path object - Groups with __container_type__ → list or dict - Datasets → numpy arrays, pandas objects, etc. (or ProxyDataset)

The use_proxy_dataset flag enables lazy loading for context manager mode.