✨ Add support for `lifespan` async context managers (superseding `startup` and `shutdown` events) (#2944)

Co-authored-by: Mike Shantz <[email protected]>
Co-authored-by: Jonathan Plasse <[email protected]>
Co-authored-by: Sebastián Ramírez <[email protected]>

docs/en/docs/advanced/events.md

@@ -1,13 +1,108 @@
-# Events: startup - shutdown
+# Lifespan Events
 
-You can define event handlers (functions) that need to be executed before the application starts up, or when the application is shutting down.
+You can define logic (code) that should be executed before the application **starts up**. This means that this code will be executed **once**, **before** the application **starts receiving requests**.
 
-These functions can be declared with `async def` or normal `def`.
+The same way, you can define logic (code) that should be executed when the application is **shutting down**. In this case, this code will be executed **once**, **after** having handled possibly **many requests**.
+
+Because this code is executed before the application **starts** taking requests, and right after it **finishes** handling requests, it covers the whole application **lifespan** (the word "lifespan" will be important in a second 😉).
+
+This can be very useful for setting up **resources** that you need to use for the whole app, and that are **shared** among requests, and/or that you need to **clean up** afterwards. For example, a database connection pool, or loading a shared machine learning model.
+
+## Use Case
+
+Let's start with an example **use case** and then see how to solve it with this.
+
+Let's imagine that you have some **machine learning models** that you want to use to handle requests. 🤖
+
+The same models are shared among requests, so, it's not one model per request, or one per user or something similar.
+
+Let's imagine that loading the model can **take quite some time**, because it has to read a lot of **data from disk**. So you don't want to do it for every request.
+
+You could load it at the top level of the module/file, but that would also mean that it would **load the model** even if you are just running a simple automated test, then that test would be **slow** because it would have to wait for the model to load before being able to run an independent part of the code.
+
+That's what we'll solve, let's load the model before the requests are handled, but only right before the application starts receiving requests, not while  the code is being loaded.
+
+## Lifespan
+
+You can define this *startup* and *shutdown* logic using the `lifespan` parameter of the `FastAPI` app, and a "context manager" (I'll show you what that is in a second).
+
+Let's start with an example and then see it in detail.
+
+We create an async function `lifespan()` with `yield` like this:
+
+```Python hl_lines="16  19"
+{!../../../docs_src/events/tutorial003.py!}
+```
+
+Here we are simulating the expensive *startup* operation of loading the model by putting the (fake) model function in the dictionary with machine learning models before the `yield`. This code will be executed **before** the application **starts taking requests**, during the *startup*.
+
+And then, right after the `yield`, we unload the model. This code will be executed **after** the application **finishes handling requests**, right before the *shutdown*. This could, for example, release resources like memory or a GPU.
+
+!!! tip
+    The `shutdown` would happen when you are **stopping** the application.
+
+    Maybe you need to start a new version, or you just got tired of running it. 🤷
+
+### Lifespan function
+
+The first thing to notice, is that we are defining an async function with `yield`. This is very similar to Dependencies with `yield`.
+
+```Python hl_lines="14-19"
+{!../../../docs_src/events/tutorial003.py!}
+```
+
+The first part of the function, before the `yield`, will be executed **before** the application starts.
+
+And the part after the `yield` will be executed **after** the application has finished.
+
+### Async Context Manager
+
+If you check, the function is decorated with an `@asynccontextmanager`.
+
+That converts the function into something called an "**async context manager**".
+
+```Python hl_lines="1  13"
+{!../../../docs_src/events/tutorial003.py!}
+```
+
+A **context manager** in Python is something that you can use in a `with` statement, for example, `open()` can be used as a context manager:
+
+```Python
+with open("file.txt") as file:
+    file.read()
+```
+
+In recent versions of Python, there's also an **async context manager**. You would use it with `async with`:
+
+```Python
+async with lifespan(app):
+    await do_stuff()
+```
+
+When you create a context manager or an async context manager like above, what it does is that, before entering the `with` block, it will execute the code before the `yield`, and after exiting the `with` block, it will execute the code after the `yield`.
+
+In our code example above, we don't use it directly, but we pass it to FastAPI for it to use it.
+
+The `lifespan` parameter of the `FastAPI` app takes an **async context manager**, so we can pass our new `lifespan` async context manager to it.
+
+```Python hl_lines="22"
+{!../../../docs_src/events/tutorial003.py!}
+```
+
+## Alternative Events (deprecated)
 
 !!! warning
-    Only event handlers for the main application will be executed, not for [Sub Applications - Mounts](./sub-applications.md){.internal-link target=_blank}.
+    The recommended way to handle the *startup* and *shutdown* is using the `lifespan` parameter of the `FastAPI` app as described above.
 
-## `startup` event
+    You can probably skip this part.
+
+There's an alternative way to define this logic to be executed during *startup* and during *shutdown*.
+
+You can define event handlers (functions) that need to be executed before the application starts up, or when the application is shutting down.
+
+These functions can be declared with `async def` or normal `def`.
+
+### `startup` event
 
 To add a function that should be run before the application starts, declare it with the event `"startup"`:
 
@@ -21,7 +116,7 @@ You can add more than one event handler function.
 
 And your application won't start receiving requests until all the `startup` event handlers have completed.
 
-## `shutdown` event
+### `shutdown` event
 
 To add a function that should be run when the application is shutting down, declare it with the event `"shutdown"`:
 
@@ -45,3 +140,21 @@ Here, the `shutdown` event handler function will write a text line `"Application
 
 !!! info
     You can read more about these event handlers in <a href="https://www.starlette.io/events/" class="external-link" target="_blank">Starlette's  Events' docs</a>.
+
+### `startup` and `shutdown` together
+
+There's a high chance that the logic for your *startup* and *shutdown* is connected, you might want to start something and then finish it, acquire a resource and then release it, etc.
+
+Doing that in separated functions that don't share logic or variables together is more difficult as you would need to store values in global variables or similar tricks.
+
+Because of that, it's now recommended to instead use the `lifespan` as explained above.
+
+## Technical Details
+
+Just a technical detail for the curious nerds. 🤓
+
+Underneath, in the ASGI technical specification, this is part of the <a href="https://asgi.readthedocs.io/en/latest/specs/lifespan.html" class="external-link" target="_blank">Lifespan Protocol</a>, and it defines events called `startup` and `shutdown`.
+
+## Sub Applications
+
+🚨 Have in mind that these lifespan events (startup and shutdown) will only be executed for the main application, not for [Sub Applications - Mounts](./sub-applications.md){.internal-link target=_blank}.

docs_src/events/tutorial003.py

@@ -0,0 +1,28 @@
+from contextlib import asynccontextmanager
+
+from fastapi import FastAPI
+
+
+def fake_answer_to_everything_ml_model(x: float):
+    return x * 42
+
+
+ml_models = {}
+
+
+@asynccontextmanager
+async def lifespan(app: FastAPI):
+    # Load the ML model
+    ml_models["answer_to_everything"] = fake_answer_to_everything_ml_model
+    yield
+    # Clean up the ML models and release the resources
+    ml_models.clear()
+
+
+app = FastAPI(lifespan=lifespan)
+
+
+@app.get("/predict")
+async def predict(x: float):
+    result = ml_models["answer_to_everything"](x)
+    return {"result": result}

fastapi/applications.py

@@ -1,6 +1,7 @@
 from enum import Enum
 from typing import (
     Any,
+    AsyncContextManager,
     Awaitable,
     Callable,
     Coroutine,
@@ -71,6 +72,7 @@ class FastAPI(Starlette):
         ] = None,
         on_startup: Optional[Sequence[Callable[[], Any]]] = None,
         on_shutdown: Optional[Sequence[Callable[[], Any]]] = None,
+        lifespan: Optional[Callable[["FastAPI"], AsyncContextManager[Any]]] = None,
         terms_of_service: Optional[str] = None,
         contact: Optional[Dict[str, Union[str, Any]]] = None,
         license_info: Optional[Dict[str, Union[str, Any]]] = None,
@@ -125,6 +127,7 @@ class FastAPI(Starlette):
             dependency_overrides_provider=self,
             on_startup=on_startup,
             on_shutdown=on_shutdown,
+            lifespan=lifespan,
             default_response_class=default_response_class,
             dependencies=dependencies,
             callbacks=callbacks,

fastapi/routing.py

@@ -7,6 +7,7 @@ from contextlib import AsyncExitStack
 from enum import Enum, IntEnum
 from typing import (
     Any,
+    AsyncContextManager,
     Callable,
     Coroutine,
     Dict,
@@ -492,6 +493,7 @@ class APIRouter(routing.Router):
         route_class: Type[APIRoute] = APIRoute,
         on_startup: Optional[Sequence[Callable[[], Any]]] = None,
         on_shutdown: Optional[Sequence[Callable[[], Any]]] = None,
+        lifespan: Optional[Callable[[Any], AsyncContextManager[Any]]] = None,
         deprecated: Optional[bool] = None,
         include_in_schema: bool = True,
         generate_unique_id_function: Callable[[APIRoute], str] = Default(
@@ -504,6 +506,7 @@ class APIRouter(routing.Router):
             default=default,
             on_startup=on_startup,
             on_shutdown=on_shutdown,
+            lifespan=lifespan,
         )
         if prefix:
             assert prefix.startswith("/"), "A path prefix must start with '/'"

tests/test_router_events.py

@@ -1,3 +1,7 @@
+from contextlib import asynccontextmanager
+from typing import AsyncGenerator, Dict
+
+import pytest
 from fastapi import APIRouter, FastAPI
 from fastapi.testclient import TestClient
 from pydantic import BaseModel
@@ -12,57 +16,49 @@ class State(BaseModel):
     sub_router_shutdown: bool = False
 
 
-state = State()
+@pytest.fixture
-
+def state() -> State:
-app = FastAPI()
+    return State()
-
-
-@app.on_event("startup")
-def app_startup():
-    state.app_startup = True
-
-
-@app.on_event("shutdown")
-def app_shutdown():
-    state.app_shutdown = True
-
 
-router = APIRouter()
 
+def test_router_events(state: State) -> None:
+    app = FastAPI()
 
-@router.on_event("startup")
+    @app.get("/")
-def router_startup():
+    def main() -> Dict[str, str]:
-    state.router_startup = True
+        return {"message": "Hello World"}
 
+    @app.on_event("startup")
+    def app_startup() -> None:
+        state.app_startup = True
 
-@router.on_event("shutdown")
+    @app.on_event("shutdown")
-def router_shutdown():
+    def app_shutdown() -> None:
-    state.router_shutdown = True
+        state.app_shutdown = True
 
+    router = APIRouter()
 
-sub_router = APIRouter()
+    @router.on_event("startup")
+    def router_startup() -> None:
+        state.router_startup = True
 
+    @router.on_event("shutdown")
+    def router_shutdown() -> None:
+        state.router_shutdown = True
 
-@sub_router.on_event("startup")
+    sub_router = APIRouter()
-def sub_router_startup():
-    state.sub_router_startup = True
 
+    @sub_router.on_event("startup")
+    def sub_router_startup() -> None:
+        state.sub_router_startup = True
 
-@sub_router.on_event("shutdown")
+    @sub_router.on_event("shutdown")
-def sub_router_shutdown():
+    def sub_router_shutdown() -> None:
-    state.sub_router_shutdown = True
+        state.sub_router_shutdown = True
 
+    router.include_router(sub_router)
+    app.include_router(router)
 
-@sub_router.get("/")
-def main():
-    return {"message": "Hello World"}
-
-
-router.include_router(sub_router)
-app.include_router(router)
-
-
-def test_router_events():
     assert state.app_startup is False
     assert state.router_startup is False
     assert state.sub_router_startup is False
@@ -85,3 +81,28 @@ def test_router_events():
     assert state.app_shutdown is True
     assert state.router_shutdown is True
     assert state.sub_router_shutdown is True
+
+
+def test_app_lifespan_state(state: State) -> None:
+    @asynccontextmanager
+    async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
+        state.app_startup = True
+        yield
+        state.app_shutdown = True
+
+    app = FastAPI(lifespan=lifespan)
+
+    @app.get("/")
+    def main() -> Dict[str, str]:
+        return {"message": "Hello World"}
+
+    assert state.app_startup is False
+    assert state.app_shutdown is False
+    with TestClient(app) as client:
+        assert state.app_startup is True
+        assert state.app_shutdown is False
+        response = client.get("/")
+        assert response.status_code == 200, response.text
+        assert response.json() == {"message": "Hello World"}
+    assert state.app_startup is True
+    assert state.app_shutdown is True

tests/test_tutorial/test_events/test_tutorial003.py

@@ -0,0 +1,86 @@
+from fastapi.testclient import TestClient
+
+from docs_src.events.tutorial003 import (
+    app,
+    fake_answer_to_everything_ml_model,
+    ml_models,
+)
+
+openapi_schema = {
+    "openapi": "3.0.2",
+    "info": {"title": "FastAPI", "version": "0.1.0"},
+    "paths": {
+        "/predict": {
+            "get": {
+                "summary": "Predict",
+                "operationId": "predict_predict_get",
+                "parameters": [
+                    {
+                        "required": True,
+                        "schema": {"title": "X", "type": "number"},
+                        "name": "x",
+                        "in": "query",
+                    }
+                ],
+                "responses": {
+                    "200": {
+                        "description": "Successful Response",
+                        "content": {"application/json": {"schema": {}}},
+                    },
+                    "422": {
+                        "description": "Validation Error",
+                        "content": {
+                            "application/json": {
+                                "schema": {
+                                    "$ref": "#/components/schemas/HTTPValidationError"
+                                }
+                            }
+                        },
+                    },
+                },
+            }
+        }
+    },
+    "components": {
+        "schemas": {
+            "HTTPValidationError": {
+                "title": "HTTPValidationError",
+                "type": "object",
+                "properties": {
+                    "detail": {
+                        "title": "Detail",
+                        "type": "array",
+                        "items": {"$ref": "#/components/schemas/ValidationError"},
+                    }
+                },
+            },
+            "ValidationError": {
+                "title": "ValidationError",
+                "required": ["loc", "msg", "type"],
+                "type": "object",
+                "properties": {
+                    "loc": {
+                        "title": "Location",
+                        "type": "array",
+                        "items": {"anyOf": [{"type": "string"}, {"type": "integer"}]},
+                    },
+                    "msg": {"title": "Message", "type": "string"},
+                    "type": {"title": "Error Type", "type": "string"},
+                },
+            },
+        }
+    },
+}
+
+
+def test_events():
+    assert not ml_models, "ml_models should be empty"
+    with TestClient(app) as client:
+        assert ml_models["answer_to_everything"] == fake_answer_to_everything_ml_model
+        response = client.get("/openapi.json")
+        assert response.status_code == 200, response.text
+        assert response.json() == openapi_schema
+        response = client.get("/predict", params={"x": 2})
+        assert response.status_code == 200, response.text
+        assert response.json() == {"result": 84.0}
+    assert not ml_models, "ml_models should be empty"