Intro

Dependency Injection is the key pattern used to implement Inversion of Control and Dependency Inversion principles (check D in SOLID).

It makes it simple to standardize the way objects are initialized in the application.

This package provides you with the set of components to automate dependency injection through the application.

Installation

Add oop-di as a dependency in pyproject.toml of your project. Or, do

pip install oop-di

Usage

To make it work do the following steps:

• provide definitions for all the services
• compile the container
• use container to inject services

# Init the `ContainerDefinition` component
container_definition = ContainerDefinition()
# fill the service definitions
container_definition.add_service(SomeService)
# compile the container
container = container_definition.compile()

# get an instance of the service from the container ...
service = container.get(SomeService)

# ... or inject it. Kwonly arguments will be injected
@container.inject()
def something(*, service: SomeService):
    ...

something()

# partial injection is also possible
def something_else(*, wont_inject: UnknownService, service: SomeService):
    ...

something_else(wont_inject=UnknownService())

Extensions

EnvExtension

You can add all environment variables to your container by adding EnvExtension to the container definition:

container_definition = ContainerDefinition()
container_definition.add_extension(EnvExtension())

JsonExtension

You can add define parameters/aliases/services in a JSON file. Check examples/s4_json.py for more details.

container_definition = ContainerDefinition()
container_definition.add_extension(JsonExtension(Path(__file__).parent / "config.json"))

Config example:

{
  "parameters": {
    "test1_email": "test@example.com",
    "test2_email": "test2@example.com",
    "test3_email": "test3@example.com"
  },
  "services": {
    "examples.s4_module": [
      {"class": "ProductService"},
      {"class": "Mailer", "name": "m1","tags": ["admin_mailers"], "parameters": {"from_email": "test1_email"}},
      {"class": "Mailer", "name": "m2","tags": ["admin_mailers"], "parameters": {"from_email": "test2_email"}},
      {"class": "Mailer", "name": "m3","tags": ["admin_mailers"], "parameters": {"from_email": "test3_email"}},
      {"class": "MultiMailer", "name": "@examples.s4_module.MailerInterface","parameters": {"mailers": "#admin_mailers"}}
    ]
  }
}

Why not python-dependency-injector?

python-dependency-injector is the most popular Python DI library with a large community, excellent docs, and tons of features that are not available in oop-di. Probably, python-dependency-injector is the best choice for you. I've built this small library to handle the DI of my small experiments/pet projects.

What I don't like in python-dependency-injector and tried to mitigate in this library:

• it requires a name for every service. oop-di can use class itself as the name of the service;
• it makes you pollute your function's params with = Provide[Container.service] defaults to inject something. oop-di just auto-wires the services by checking the parameter's types;
• oop-di has the tagging feature that can gather services into an array without an explicit definition of that array.

Check examples/ folder for more info:

• s1_simple.py - shows the basics: how to define services and how to inject the services somewhere.
• s2_extensions.py - how to split the definitions between modules.
• s3_tags.py, s3_tags2_dict.py - how to use tags to group services (list or dict grouping is possible).
• s4_json.py - JsonExtension usage example.