Adding new integrations¶

Though there are some integrations in library you are not limited to use them.

The main points are:

Find a way to pass a global container instance. Often it is attached to application instance or passed by a middleware.
Find a place to enter and exit request scope and how to pass the container to a handler. Usually, it is entered in a middleware and container is stored in some kind of request context.

Alternatively, you can use the wrap_injection function with manage_scope=True to automate entering and exiting the request scope without relying on middleware. When enabled, manage_scope ensures that the container passed to wrap_injection enters and exits the next scope. Custom Scope Targeting

For more complex cases, you can pass a scope argument to wrap_injection to specify a custom scope to enter on function call. For example, you can create decorators that target specific scopes in the dependency hierarchy:
```
@inject(scope=Scope.STEP)
async def handler(step_dep: StepDep = FromDishka[StepDep]):
    ...

@inject(scope=Scope.ACTION)
def process_action(action_dep: ActionDep = FromDishka[ActionDep]):
    ...
```
When you use manage_scope=True or specify a custom scope, you can pass provide_context function that allows to populate the container context with passed arguments. This is useful when you want to pass the function scoped arguments to other dependencies without relying on middleware.
Configure a decorator. The main option here is to provide a way for retrieving container. Often, need to modify handler signature adding additional parameters. It is also available.
Check if you can apply decorator automatically.

While writing middlewares and working with scopes is done by your custom code, we have a helper for creating @inject decorators - a wrap_injection function.

container_getter is a function with two params (args, kwargs) which is called to get a container used to retrieve dependencies within scope.
additional_params is a list of inspect.Parameter which should be added to handler signature.

For more details, check existing integrations.