Übertyped

Convert dataclass to TypedDict for fun and type safety 🪄

uber- /ˈuːbə/ to a great or extreme degree

Requirements

• python>=3.11
• mypy>=1.0.1

Installation

pip install ubertyped

Mypy plugin

Add ubertyped.mypy_plugin to the list of plugins in your mypy config file (for example pyproject.toml)

[tool.mypy]
python_version = "3.11"
plugins = ["ubertyped.mypy_plugin"]

Features

• ✅ AsTypedDict[T] generic type converting dataclasses to TypeDicts
• ✅ as_typed_dict utility function wrapping dataclasses.asdict
• ✅ Support for usage with TypeVars
• ✅ Nested dataclasses
• ✅ Compatibility with other typecheckers such as Pylance and Pyright
• ✅ Zero dependencies

Usage

from dataclasses import asdict, dataclass
from typing import Self, reveal_type

from ubertyped import AsTypedDict, as_typed_dict


@dataclass
class Base:
    base: bool


@dataclass
class IntWrapper:
    value: int


@dataclass
class Data(Base):
    version: IntWrapper
    command: str

    def as_typed_dict(self) -> AsTypedDict[Self]:
        return as_typed_dict(self)


data = Data(version=IntWrapper(1), command="c", base=False)

# 🎉 Type-safe conversion!
td = as_typed_dict(data)
reveal_type(td)
# Revealed type is "TypedDict({'version': TypedDict({'value': builtins.int}), 'command': builtins.str, 'base': builtins.bool})"

# 🎉 Works with nested dataclasses too!
reveal_type(td["version"]["value"])
# Revealed type is "builtins.int"


# 🎉 Binding `Self` in methods is resolved correctly!
reveal_type(data.as_typed_dict)
# Revealed type is "def () -> TypedDict({'version': TypedDict({'value': builtins.int}), 'command': builtins.str, 'base': builtins.bool})"

# 🎉 In runtime, `as_typed_dict` is just a wrapper around `asdict`!
if asdict(data) == data.as_typed_dict():
    print("✅")
else:
    print("❌")