V1: Enabling Experimental Features - rnag/dataclass-wizard GitHub Wiki

[!TIP]
Early access to V1 is available starting from v0.33.0! This enables performance optimizations and introduces experimental changes that will be part of the upcoming major v1 release. 🚀

To opt-in to experimental v1 features, users can pass v1 = True in the Meta setting.

Example: Opting In To v1 Features

from dataclasses import dataclass
from dataclass_wizard import JSONPyWizard
from dataclass_wizard.v1 import Alias

@dataclass
class A(JSONPyWizard):
    class _(JSONPyWizard.Meta):
        v1 = True

    my_str: str
    version_info: float = Alias(load='v-info')

a = A.from_dict({'my_str': 'test', 'v-info': '1.0'})
print(a)  #> A(my_str='test', version_info=1.0)

assert a.version_info == 1.0
assert a.to_dict() == {'my_str': 'test', 'version_info': 1.0}

[!NOTE] DataclassWizard is now the recommended approach for v1 opt-in, as it (1) applies the @dataclass decorator to subclasses automatically and (2) enables v1=True in the Meta settings.

from dataclass_wizard import DataclassWizard
from dataclass_wizard.v1 import Alias

class A(DataclassWizard):
    my_str: str
    version_info: float = Alias(load='v-info')

Without Inheritance

For simple dataclasses that do not subclass DataclassWizard or JSONPyWizard, you can use LoadMeta (and similar) utilities:

from dataclasses import dataclass
from dataclass_wizard import DumpMeta, LoadMeta, asdict, fromdict
from dataclass_wizard.v1 import Alias

@dataclass
class A:
    my_str: str
    version_info: float = Alias(load='v-info')

LoadMeta(v1=True).bind_to(A)
DumpMeta(key_transform='NONE').bind_to(A)

a = fromdict(A, {'my_str': 'test', 'v-info': '1.0'})
print(a)  #> A(my_str='test', version_info=1.0)

assert a.version_info == 1.0
assert asdict(a) == {'my_str': 'test', 'version_info': 1.0}

Pass case on Sub-class

This is a lesser-known (but equally valid) approach to opting in to experimental v1 features and worth documenting.

When sub-classing from JSONWizard (or a subclass thereof), pass in case (alternatively, load_case and dump_case for de/serialization respectively) which determines the letter case for JSON keys in de-serialization. Read more about Key Case Transformation.

For example, set the AUTO key casing mode:

from dataclass_wizard import DataclassWizard
from dataclass_wizard.class_helper import get_meta
from dataclass_wizard.v1 import Alias


class A(DataclassWizard, case='AUTO'):
    my_str: str
    version_info: float = Alias(load='v-info')


# Confirm `v1` is automatically enabled
assert get_meta(A).v1  # > True

a = A.from_dict({'My-Str': 'test', 'v-info': '1.0'})
print(a)  #> A(my_str='test', version_info=1.0)

assert a.version_info == 1.0
assert a.to_dict() == {'my_str': 'test', 'version_info': 1.0}

Discovering New Settings

All new settings related to v1 opt-in are prefixed with v1_ for convenience and discoverability. For example:

v1_unsafe_parse_dataclass_in_union = False

Note: The v1_unsafe_parse_dataclass_in_union setting is documented separately. Refer to the Enhanced Union or | Type Parsing page for details.

---

Important Notes on Experimental Features

[!WARNING]
Experimental Nature: The v1 features are experimental and may change in future releases. Exercise caution when using these features in production environments.

Backward Compatibility: Enabling v1 introduces new features and optimizations that may come with breaking changes. However, the core functionality should remain the same.

For more details, refer to the Dataclass Wizard Documentation.