/drf-service-layer

Simple package which supports service-layered architecture for Django REST Framework.

Primary LanguagePythonMIT LicenseMIT

DRF-Service-Layer

Simple package which supports service-layered architecture for Django REST Framework.


Why service layer?

Have you ever wondered where to put your business logic when you use Django & DRF? There are several solutions with their pros and cons. Let's check them one by one.

  1. Fat Models, Skinny Views
    This is one of the most popular ways to split business logic from views. To keep your views light, all the heavy codes go into "fat" models. The problem is that, as your project gets bigger, models become fatter and have too many responsibilities. They become "god" models over "fat" ones in the end. Besides, there are some cases when your business logic doesn't require any database access or they refers to multiple tables. In these cases, it's ambiguous which model should manage the logic.

  2. QuerySet/Managers
    It might be preferable to move your business logic from models to queryset or managers. But still, this solution is not a good choice like the "Fat Models approach" if your business logic doesn't need to communicate with databases.

  3. Forms or Serializers
    Probably the worst option. They each have their own purpose. Please don't do this.

  4. Fat Views
    If all of your business logic stays in views, you'll have trouble understanding the flow of your views in a very short period of time. And if you inherit one of your fat views, the parent view and the child view are too strongly coupled. So you'll have a hard time when you want to extract the legacy apis from your project.

  5. Service layer
    Split your business logic into functions and put them in a separate layer which ties models and views. To manage functions efficiently and improve the cohesion of codes, combine them into classes. In this way, views become easier to read and business logic becomes much more maintainable. Even though this may not be a standard design pattern from Django convention, some big companies like Doordash are already using this pattern by implementing it on their own.


How to use DRF-Service-Layer in View

Steps

  1. Install package

    pip install drf-service-layer

  2. Decide a type of DTO.

    💁 What is DTO?
    DTO(Data Transfer Object) is an object that carries data between processes.
    In DRF-Service-Layer, DTO is an object used for transferring data necessary for your business logic.

    DTO works between views and the service layer. If you want to transfer any data from a view to a service, implement dto property in your view that inherits GenericServiceAPIView from DRF-Service-Layer. We'll cover this shortly.

    Let's implement DTO. There are several container types you can choose. If you want to validate your DTO before transfer, you can use 3rd party library like Pydantic.

    • DTO as dataclass

      # services.py
from dataclasses import dataclass
from typing import Optional


@dataclass
class OrderDTO:
    user_id: int
    order_id: int
    sort: Optional[str] = None

    • DTO as dictionary

    • DTO as list

    • or any type you want...

  3. Implement dto property in views.

    If you decide to use dataclass as DTO:

    # views.py
from drf_service_layer.views import GenericServiceAPIView


class OrderAPIView(GenericServiceAPIView):

    @property
    def dto(self) -> OrderDTO:
        return OrderDTO(
            user_id=self.request.user.id,
            order_id=self.kwargs['order_id'],
            sort=self.request.query_params.get("sort"),
        )

  4. Create a service class and implement business logic as an instance method.

    # services.py
from drf_service_layer.services import Service


class OrderService(Service):
    dto: OrderDTO

    def any_business_method(self):
        user_id = self.dto.user_id
        order_id = self.dto.order_id
        sort = self.dto.sort

        # business logic goes here.

  5. Specify a service class into a view as service_class.

    # views.py
class OrderAPIView(GenericServiceAPIView):
    service_class = OrderService  # new

    @property
    def dto(self) -> OrderDTO:
        # ...

  6. Use service layer in a view.

    # views.py
class OrderAPIView(GenericServiceAPIView):
    service_class = OrderService

    @property
    def dto(self) -> OrderDTO:
        # ...

    def get(self, request, *args, **kwargs):  # new
        # ...
        self.service.any_business_method()
        # ...
        return Response(...)

Description

When a view is initialized by DRF's initial() method, dto property is used as an argument for instantiating the service layer. DTO is already injected into the service layer as an instance variable(self.dto), so you don't need to care about parameters when implementing business logic and using them. You can get editor support from type hinting(self.dto: OrderDTO) when you deal with self.dto. After all, you can call any method from the service layer using self.service in your views.


(WIP)How to use DRF-Service-Layer in Serializer

from drf_service_layer.services import service_layer


@service_layer(FooService, FooDTO)
class FooSerializer(serializers.ModelSerializer):
    # ...

    def foo(self, obj):
        # ...
        return self.service.any_business_method()

Description

If you add @service_layer() decorator to your serializer, you can access the service layer through self.serivce. The decorator explicitly notifies you that the serializer is connected to the service layer.

Inspired by