Comprendre les dépendances dans FastAPI

Ever wished your functions could magically get the tools they need? That's what Dependency Injection (DI) helps with!

In FastAPI, DI is a powerful design pattern where components (like your API endpoints) don't create their dependencies. Instead, they declare what they need, and FastAPI "injects" them.

FastAPI uses DI extensively because it makes your code:

• Reusable: Write logic once, use everywhere.
• Testable: Easily swap real services with mock versions for tests.
• Clean: Keeps path operation functions focused on their main task.
• Maintainable: Changes to dependencies don't ripple through your entire app.

A dependency in FastAPI is usually just a regular Python function.

This function can perform any setup, validation, or resource acquisition needed before your main API logic runs. It can even return a value.

FastAPI will automatically call this function and pass its result to your endpoint.

To use a dependency, you declare it as a parameter in your path operation function, using FastAPI's special Depends function.

When a request comes in, FastAPI checks your endpoint's parameters, resolves any dependencies, and then calls your endpoint with the results.

Let's see a basic dependency that returns a simple string. Notice how get_message is called by FastAPI before read_root.

Dependencies aren't limited to simple functions. They can also take their own parameters, which FastAPI will also resolve.

This means a dependency can itself depend on other dependencies, forming a chain! It's super powerful for building complex logic.

Here, greet_user is a dependency that takes a name. If name is not provided in the query, it defaults to 'Guest'.

The magical piece connecting your endpoint to a dependency is the Depends() function.

• It tells FastAPI: "Hey, before running this function, please resolve this dependency."
• You pass your dependency function (or callable) directly to Depends(), not its result.
• FastAPI handles calling your dependency function for you.

Dependencies are perfect for:

• Database connections: Get a session for each request.
• Authentication: Verify user credentials.
• Authorization: Check user roles/permissions.
• Shared logic: Any code that multiple endpoints need.

They keep your API clean and focused.

Consider the following FastAPI code snippet:

from fastapi import FastAPI, Depends

app = FastAPI()

def get_db_session():
    # Imagine this connects to a database
    return "Database Session Object"

@app.get("/items/")
async def read_items(session: str = Depends(get_db_session)):
    return {"message": f"Using {session}"}

What will be the value of session inside the read_items function when a request is made to /items/?

You've taken a crucial step in mastering FastAPI by understanding Dependency Injection!

• Dependencies are functions FastAPI runs before your endpoint.
• They keep your code modular, reusable, and testable.
• You declare them using param: Type = Depends(your_dependency_func).

Next, we'll dive into more common and advanced dependency patterns!