trisongz/lazycls

When writing data classes becomes too much work

lazy

Python Utilities for the Super Lazy

---

Motivation

This library is forked from an internal project that works with a lot of dataclasses, (AWS API) and I got tired of writing data classes to work with and manipulate them. This library is a wrapper around the main pydantic.create_model function that recursively parses a dict object and transforms them into subclasses. So nested dict objects within dicts get transformed into their own dataclass.

---

Quickstart

pip install --upgrade lazycls

from lazycls import LazyCls, BaseLazy

data = {
    'x': ...,
    'y': ...
}

obj = LazyCls(
    name: str = 'CustomCls',
    data: Dict[str, Any] = data, 
    modulename: str = 'lazycls', # your module name
    basecls: Type[BaseModel] = BaseLazy # A custom Base Model class that is used to generate the model
    ) -> Type[BaseModel]:

"""
obj =   lazycls.CustomCls
        lazycls.CustomCls.x = ...
        lazycls.CustomCls.y = ...
"""

---

Utilities

Some additional enhancements/utilities include:

• set_modulename(name) - set the default module name - useful when included in other libs

• clear_lazy_models - clears all the currently created lazy models. Memory management

• classproperty - allows for usage of @classproperty which isn't available for Python < 3.9

• BaseCls - A wrapper around BaseModel with:

  • arbitrary_types_allowed = True
  • .get(name, default) function to retain dict-like properties

• BaseLazy - Another wrapper around BaseModel with:

  • arbitrary_types_allowed = True
  • extra = 'allow'
  • alias_generator = to_camelcase
  • orjson serializer by default
  • .get(name, default) function to retain dict-like properties