dataclass-persistence 0.0.14

This package enables to persist information stored in dataclasses.

This program can be used to make dataclasses persistent by adding store and load functionality. The dataclass is stored in .json format which is by default compressed inside of a .zip file.

What makes dataclass-persistence special?

• Support for numpy arrays

• Support for nested dataclasses

• Human readable storage format with small file size

Usage

Let your dataclass inherit from PersistentDataclass. Then the dataclass can be stored on disk using .store_to_disk() and loaded from disk using .load_from_disk().

In the example below, we create an instance of dataclass, which is stored to and loaded from disk.

from dataclass_persistence import PersistentDataclass
from dataclasses import dataclass
import numpy as np


@dataclass
class SomeData(PersistentDataclass):
    parameter_a: str
    array: np.ndarray


data = SomeData('my_string', np.array([0, 0]))
file = 'my_file'
data.store_to_disk(file)
data_reconstructed = SomeData.load_from_disk(file)

On disk the code above produces my_file.zip which contains my_file.json:

{
  "parameter_a": "my_string",
  "array": {"data": [0, 0], "dtype": "int32"}
}