PyDIT (Python Dependency Injection with Typing)

Description

With PyDit you can use ABC and Protocol to make interfaces and use the power of Dependency Inversion Principle.

PyDit allow your code to depend only abstract things, not the real implementation.

Install

With poetry

poetry add python-pydit

With pip

pip install python-pydit

Usage

Create the PyDit instance:

app/configs/di.py

from pydit import PyDit

pydit = PyDit()

Import the pydit instance and register your project's dependencies:

app/configs/dependencies.py

from typing import Any
from app.configs.di import pydit
from app.adapters.repositories.sqlalchemy.user_repository import SqlalchemyUserRepository
from app.configs.database import get_db_config


def setup_dependencies():
  """
  This is only a suggestion, you're free to configure it like you want
  """

  dependencies: list[dict[str, Any]] = [
    {
      "dependency": SqlalchemyUserRepository,
      "token": "sqlalchemy_user_repository"
    },
    {
      "dependency": get_db_config,
      "token": "database_config",
    },
    {
      "dependency": "HELLO WORLD",
      "token": "test"
    }
  ]

  for dependency in dependencies:
    pydit.add_dependency(dependency["dependency"], dependency.get("token"))

Call the setup_dependencies in the main file

app/main.py

from app.config.di import setup_dependencies()

setup_dependencies()

Injecting dependencies

Inject by property

Pydit supports python's @property decorator.

app/domain/user/services/create.py

from typing import cast, Any
from app.configs.di import pydit
# This class can be a Protocol or a clas that inherits from ABC
from app.adapters.repositories.interfaces.user_repositgory import IUserRepository

class CreateUserService:
  @pydit.inject()
  def user_repository(self) -> IUserRepository:
    return cast(IUserRepository, None)

  @pydit.inject(token="test")
  def other_property(self) -> str:
    return ""

  def execute(self, data: dict[str, Any]):
    self.user_repository.create(data)

    # Prints HELLO WORLD
    print(self.other_property)

How you can see, we're depending on the intarface IUserRepository, not the real SqlalchemyUserRepository implementation.

Inject dependencies by constuctor

PyDIT needs that each parameter of class/function have default value to be resolved

app/domain/user/services/create.py

from pydit import FunctionInject
from app.adapters.repositories.interfaces.user_repositgory import IUserRepository

class CreateUserService:
  def __init__(
    self,
    user_repository: IUserRepository = FunctionInject(),
    other_property: str = FunctionInject(token="test")
  ):
    self.user_repository = user_repository

  def execute(self, data):
    self.user_repository.create(data)

    # Prints HELLO WORLD
    print(other_property)

Inject dependencies by function

PyDIT can handle user defined functions too.

app/domain/user/services/create.py

from typing import Any
from pydit import FunctionInject
from app.adapters.repositories.interfaces.user_repositgory import IUserRepository

def create_user(user_data: dict[str, Any], repository: IUserRepository = FunctionInject()):
  repository.create(user_data)

app/config/dependencies.py

from app.configs.di import pydit
from app.domain.user.services.create import create_user

def setup_dependencies():
  pydit.add_dependency(create_user, "create_user")

app/routers/v1/user.py

from fastapi import APIRouter
from app.configs.di import pydit
from app.domain.user.dto.create import CreateUserDTO


router = APIRouter(prefix="/users", tags=["Users"])

@router.post("")
def create(data: CreateUserDTO):
  create_service = pydit.get_value(token="create_user")
  create_service(data.dict())

  return "OK"

Singleton

To use singleton approach, pass true to singleton parameter in inject fn.

from typing import cast, Any
from app.configs.di import pydit
# This class can be a Protocol or a clas that inherits from ABC
from app.adapters.repositories.interfaces.user_repositgory import IUserRepository

class CreateUserService:
  @pydit.inject(singleton=True)
  def user_repository(self) -> IUserRepository:
    return cast(IUserRepository, None)

  @pydit.inject(token="test")
  def other_property(self) -> str:
    return ""

  def execute(self, data: dict[str, Any]):
    self.user_repository.create(data)

    # Prints HELLO WORLD
    print(self.other_property)

Features:

• Inject values based on type signature
• Inject values based on inheritance
• Inject values via token
• Resolves function dependencies, calling and injecting the call result
• Singleton support
• Inject values in function calls or class constructor __init__ based on the arguments' signatures
• Add support for kwargs when inject values in functions