python-injection 0.10.0

Fast and easy dependency injection framework.

Basic usage

Register an injectable

Note: If the class needs dependencies, these will be resolved when the instance is retrieved.

If you wish to inject a singleton, use singleton decorator.

from injection import singleton

@singleton
class ServiceA:
    """ class implementation """

If you wish to inject a new instance each time, use injectable decorator.

from injection import injectable

@injectable
class ServiceB:
    """ class implementation """

If you have a constant (such as a global variable) and wish to register it as an injectable, use set_constant function.

from injection import set_constant

class ServiceC:
    """ class implementation """

service_c = ServiceC()
set_constant(service_c)

Or here is the decorator constant which is equivalent:

Unlike @singleton, dependencies will not be resolved.

from injection import constant

@constant
class ServiceC:
    """ class implementation """

Inject an instance

To inject one or several instances, use inject decorator. Don't forget to annotate type of parameter to inject.

from injection import inject

@inject
def some_function(service_a: ServiceA):
    """ function implementation """

If inject decorates a class, it will be applied to the __init__ method. Especially useful for dataclasses:

Note: Doesn't work with Pydantic BaseModel because the signature of the __init__ method doesn't contain the dependencies.

from dataclasses import dataclass

from injection import inject

@inject
@dataclass
class SomeDataClass:
    service_a: ServiceA = ...

Get an instance

Example with get_instance function:

from injection import get_instance

service_a = get_instance(ServiceA)

Example with get_lazy_instance function:

from injection import get_lazy_instance

lazy_service_a = get_lazy_instance(ServiceA)
# ...
service_a = ~lazy_service_a

Inheritance

In the case of inheritance, you can use the decorator parameter on to link the injection to one or several other classes.

Warning: if the child class is in another file, make sure that file is imported before injection. See load_packages function.

Example with one class:

class AbstractService(ABC):
    ...

@injectable(on=AbstractService)
class ConcreteService(AbstractService):
    ...

Example with several classes:

class AbstractService(ABC):
    ...

class ConcreteService(AbstractService):
    ...

@injectable(on=(AbstractService, ConcreteService))
class ConcreteServiceOverload(ConcreteService):
    ...

If a class is registered in a package, and you want to override it, there is the mode parameter:

@injectable
class InaccessibleService:
    ...

# ...

@injectable(on=InaccessibleService, mode="override")
class ServiceOverload(InaccessibleService):
    ...

Recipes

A recipe is a function that tells the injector how to construct the instance to be injected. It is important to specify the return type annotation when defining the recipe.

from injection import injectable

@injectable
def service_d_recipe() -> ServiceD:
    """ recipe implementation """

Working with type aliases

from injection import injectable, set_constant

type APIKey = str

set_constant("<secret_api_key>", APIKey, alias=True)

@injectable
class Client:
    def __init__(self, api_key: APIKey):
        ...