Dependency

Dependencies are simple functions that produce results. They allow you to declaratively split code into preparation and business logic.

By default they are called only once during injection, so it guarantees no unnecessary repetition. This is useful when there is much to do and you don’t want to pass all things inside arguments. This behavior may be disabled using caching=False parameter in from_(...) and scan(...) functions.

Note: disabling caching in from_(...) and scan(...) works only for specified function, not for its dependencies.

Example of dependency that generates one random name:

import random

def require_random_name() -> str:
    return random.choice(
        ("Bob", "Steve", "Petro", "Yevhen", "Stepan", "Vitaliy", "Volodymyr", "Tom", "Jack", "Jerry")
    )

In real apps, you can override this dependency for testing. See overriding section

Asynchronous dependency is defined in the same way:

import random
import asyncio

async def require_random_name() -> str:
    await asyncio.sleep(0.4)  # simulate web request
    return random.choice(
        ("Bob", "Steve", "Petro", "Yevhen", "Stepan", "Vitaliy", "Volodymyr", "Tom", "Jack", "Jerry")
    )

Naming convention

Because get_user_or_die_trying is a little too honest.

I’d recommend to give to dependency the name that actually represent what they provide and do.

Prefixes are good way to tell how dependency behaves on injection. Some of the prefixes I use:

• require_ — Dependency may raise error whenever something had failed

• optional_ — Dependency may return None whenever it fails

• acquire_ — Dependency acquires some resources. Used for lifespan-dependencies

Also, prefixes can help you to separate dependency name from the parameter name where it’s result goes. Simplest example is admin_user and require_admin_user