HDF5Serializer

class pisces.utilities.io_tools.HDF5Serializer

A JSON-based serialization utility for storing complex Python objects as HDF5 attributes.

This class provides a standardized mechanism for serializing and deserializing Python objects — including extended types like unyt arrays, quantities, and units — to and from JSON strings that can be safely stored as attributes in HDF5 files.

Custom types are registered via the __REGISTRY__ class variable, which maps Python types to a tuple of:

• A unique string tag identifying the type in serialized form,

• A serialization function (returning a JSON-serializable dict),

• A deserialization function (taking that dict and reconstructing the object).

Built-in JSON-compatible types (e.g., int, float, str, list, dict) are serialized directly via json.dumps. All deserialized outputs are reconstructed either via the registry or returned as-is for base types.

This class is intended to be subclassed or extended to support additional custom types.

Methods

__init__()
deserialize_data(data)	Recursively deserialize a JSON string produced by serialize_data.
deserialize_dict(data)	Recursively deserialize a dict produced by serialize_dict.
deserialize_unyt(o)	Deserialize a dictionary into a unyt object.
deserialize_unyt_unit(o)	Deserialize a dictionary into a unyt unit.
serialize_data(data)	Recursively serialize data to a JSON string for HDF5 attributes.
serialize_dict(data)	Recursively serialize a dict’s values (keys must be strings).
serialize_unyt(o)	Serialize a unyt object into a dictionary.
serialize_unyt_unit(o)	Serialize a unyt unit into a dictionary.