Hello,

So I am working on a Python application which an API using the FastAPI framework and a "watcher" component which will handle some asynchronous things on the backend (checking deployment status, health checks, etc). So what I have right now is 3 containers: ApplicationContainer, APIContainer, and WatcherContainer. The basic package structure is as follows:

.
└── virtual_labs
 ├── api
 │  ├── auth
 │  ├── container.py
 │  ├── data_sources
 │  │  ├── enterprise_dw
 │  │  ├── northwind
 │  │  └── webreports
 │  ├── endpoints
 │  └── services
 ├── container.py
 ├── core
 │  ├── clients
 │  ├── models
 │  ├── mutation_executor
 │  ├── schemas
 │  └── settings
 └── watcher
 └── container.py

Basically the reasoning for this is that I have quite a bit of common code that I would like to be shared between the API and the watcher component. The thing I'm struggling with is how to handle the FastAPI create_app() method and adding the router via dependency injection. Can this be achieve or did I just waste the last several hours of my life?

I was able to get it somewhat working but there's some weird side effects. I started getting warnings that my async resources were not being awaited and seeing this exception get thrown RuntimeError: There is no current event loop in thread 'MainThread' What I was envisioning was something like the following for the main entrypoint:

@inject
async def main(entrypoint: Literal['api', 'watcher'], create_app_factory = Provide[ApplicationContainer.api_package.create_app_factory], watcher_coordinator = Provide[ApplicationContainer.watcher_package.coordinator]):
 if entrypoint == 'api':
 return create_app_factory()
 elif entrypoint == 'watcher':
 await watcher_coordinator.run()
if __name__ == "__main__":
 entrypoint = 'api'
 container = ApplicationContainer()
 settings = get_app_settings()
 settings.configure_logging()
 container.config.from_pydantic(settings)
 container.init_resources()
 
 # would load this via argparser
 entrypoint = "api"
 main(entrypoint)

Any ideas on this topic? After I started I realized it was probably a boneheaded move due to all the complexities introduced but it feels like too late to turn back now.