Test

Project setup

First, create a new project and install the required testing dependencies:

uv init test
cd test
uv add fastapi[standard]
uv add --dev pytest httpx

FastAPI testing is built on HTTPX, which shares a similar design with the popular Requests library. This makes testing familiar and intuitive.

Since HTTPX is compatible with PyTest, you can use pytest directly to test your FastAPI applications.

TestClient

Create your FastAPI application in main.py:

python

main.py

from fastapi import FastAPI

app = FastAPI()

@app.get("/")
async def get():
    return {"message": "Hello World"}

Then create a test file test_main.py:

python

test_main.py

from fastapi.testclient import TestClient
from main import app

client = TestClient(app)


def test_get():
    response = client.get("/")
    assert response.status_code == 200
    assert response.json() == {"message": "Hello World"}

Writing Tests

Create a TestClient instance by passing your FastAPI application to it
Define test functions with names starting with test_ (standard pytest convention)
Make HTTP requests using the TestClient object the same way as you would with httpx
Add assertions using standard Python assert statements to verify the expected behavior

Note

Notice that test functions use regular def, not async def.

Client calls are also synchronous (no await needed).

This allows you to use pytest directly without complications.

Extended example

Let’s extend this example with more details to demonstrate testing different scenarios.

Suppose your main.py file now includes additional path operations:

A GET operation that may return errors
A POST operation that may return multiple types of errors
Both operations require an X-Token header for authentication

python

main.py

from typing import Annotated

from fastapi import FastAPI, Header, HTTPException
from pydantic import BaseModel

secret_token = "V2o3cTZwZ0ZyVnNqVXA5a0xkU1RNVWJ6cGJ3Z1ZqQ3c="

class Item(BaseModel):
    id: int
    title: str
    description: str | None = None


db: dict[int, Item] = {
    1: Item(id=1, title="Laptop", description="A high-performance laptop for development"),
    2: Item(id=2, title="Mouse", description="Wireless ergonomic mouse"),
    3: Item(id=3, title="Keyboard"),
}

app = FastAPI()


@app.get("/item/{item_id}", response_model=Item)
async def item_get(item_id: int, x_token: Annotated[str, Header()]):
    if x_token != secret_token:
        raise HTTPException(status_code=400, detail="Invalid X-Token header")
    if item_id not in db:
        raise HTTPException(status_code=404, detail="Item not found")
    return db[item_id]


@app.post("/item", response_model=Item)
async def item_post(item: Item, x_token: Annotated[str, Header()]):
    if x_token != secret_token:
        raise HTTPException(status_code=400, detail="Invalid X-Token header")
    if item.id in db:
        raise HTTPException(status_code=409, detail="Item already exists")
    db[item.id] = item
    return item

Here are comprehensive tests covering both success and error cases:

python

test_main.py

from fastapi.testclient import TestClient
from main import app

client = TestClient(app)


def test_item_get():
    response = client.get("/item/1", headers={"X-Token": "V2o3cTZwZ0ZyVnNqVXA5a0xkU1RNVWJ6cGJ3Z1ZqQ3c="})
    assert response.status_code == 200
    assert response.json() == {"id": 1, "title": "Laptop", "description": "A high-performance laptop for development"}


def test_item_get_bad_token():
    response = client.get("/item/1", headers={"X-Token": "Z1l5R0xqQk1yR1ZrU2RwX1NMS09mV2Ryb1p3b1h2b1E="})
    assert response.status_code == 400
    assert response.json() == {"detail": "Invalid X-Token header"}


def test_item_get_nonexistent_item():
    response = client.get("/item/10", headers={"X-Token": "V2o3cTZwZ0ZyVnNqVXA5a0xkU1RNVWJ6cGJ3Z1ZqQ3c="})
    assert response.status_code == 404
    assert response.json() == {"detail": "Item not found"}


def test_item_post():
    response = client.post(
        "/item",
        headers={"X-Token": "V2o3cTZwZ0ZyVnNqVXA5a0xkU1RNVWJ6cGJ3Z1ZqQ3c="},
        json={"id": 4, "title": "Monitor", "description": "27-inch 4K IPS display"},
    )
    assert response.status_code == 200
    assert response.json() == {"id": 4, "title": "Monitor", "description": "27-inch 4K IPS display"}


def test_item_post_bad_token():
    response = client.post(
        "/item",
        headers={"X-Token": "Z1l5R0xqQk1yR1ZrU2RwX1NMS09mV2Ryb1p3b1h2b1E="},
        json={"id": 5, "title": "USB-C Hub", "description": "7-in-1 hub with HDMI and Ethernet"}
    )
    assert response.status_code == 400
    assert response.json() == {"detail": "Invalid X-Token header"}


def test_item_post_existing_item():
    response = client.post(
        "/item",
        headers={"X-Token": "V2o3cTZwZ0ZyVnNqVXA5a0xkU1RNVWJ6cGJ3Z1ZqQ3c="},
        json={"id": 4, "title": "Monitor", "description": "27-inch 4K IPS display"},
    )
    assert response.status_code == 409
    assert response.json() == {"detail": "Item already exists"}

Passing data to the TestClient

When you need to send data in requests, and you’re unsure of the syntax, you can reference the HTTPX documentation or even search for examples using the Requests library, since HTTPX’s API is based on Requests.

Common patterns:

Data Type	How to Pass It
Path or query parameters	Add them directly to the URL
JSON body	Pass a Python object (e.g., a `dict`) to the `json` parameter
Form data	Use the `data` parameter instead of `json`
Headers	Pass a `dict` to the `headers` parameter
Cookies	Pass a `dict` to the `cookies` parameter

Note

Note that the TestClient receives data that can be converted to JSON, not Pydantic models.

If you have a Pydantic model in your test and you want to send its data to the application during testing, you can use the jsonable_encoder described in JSON Compatible Encoder.

Async tests

Why async tests?

Asynchronous tests are essential when your application interacts with async components like databases, external APIs, or message queues. For example, if you’re using an async database library, you’ll want to verify that your FastAPI endpoints correctly write data to the database.

However, this introduces a challenge: how do you test async operations effectively?

Installing dependencies

First, add the required testing dependency for async support:

uv add --dev trio

Using `pytest.mark.anyio`

To test asynchronous code, your test functions must also be asynchronous. The AnyIO library provides a pytest plugin that enables async test execution.

Mark your test functions with @pytest.mark.anyio to run them asynchronously.

Why you can’t use TestClient for async tests

Even if your FastAPI application uses regular def functions instead of async def, FastAPI is fundamentally asynchronous under the hood.

The TestClient uses internal mechanisms to call your asynchronous FastAPI application from synchronous test functions. However, this approach breaks down when your test function itself is asynchronous. When running tests with async def, you cannot use TestClient.

Instead, you’ll use HTTPX’s AsyncClient directly.

Using AsyncClient

Since TestClient is built on top of HTTPX, you can use HTTPX’s AsyncClient to test your API directly in async tests.

Update your test_main.py file:

kotlin

test_main.py

import pytest
from httpx import ASGITransport, AsyncClient

@pytest.mark.anyio
async def test_get_async():
    async with AsyncClient(
            transport=ASGITransport(app=app), base_url="http://test"
    ) as async_client:
        response = await async_client.get("/")
    assert response.status_code == 200
    assert response.json() == {"message": "Hello World"}

Key differences from synchronous tests:

Test function declaration: Use async def instead of def
Decorator: Add @pytest.mark.anyio before the test function
Client: Use AsyncClient instead of TestClient
Requests: Use await when making requests
Context manager: Use async with for proper resource management

Caution

The AsyncClient doesn’t automatically trigger lifespan events (startup/shutdown). If your application depends on these events, use LifespanManager from florimondmanca/asgi-lifespan.

Testing with other async operations

Now that your test function is asynchronous, you can freely await other async operations within your tests. This is particularly useful for:

Querying async databases
Making external API calls
Verifying background tasks
Checking message queues

For example, you might test an endpoint and then verify the database state:

python

@pytest.mark.anyio
async def test_item_creates_db_entry():
    async with AsyncClient as async_client: response = await async_client.post
    # Now you can await async database operations
    item = await db.get_item(10)  # Async database call
    assert item.title == "New Item"

Note

If you encounter a RuntimeError: Task attached to a different loop when integrating asynchronous function calls in your tests (e.g. when using MongoDB’s MotorClient), remember to instantiate objects that need an event loop only within async functions, e.g. an @app.on_event("startup") callback.

Test

#Project setup

#TestClient

#Writing Tests

#Extended example

#Passing data to the TestClient

#Async tests

#Why async tests?

#Installing dependencies

#Using pytest.mark.anyio

#Why you can’t use TestClient for async tests

#Using AsyncClient

#Testing with other async operations

#Pending

Project setup

TestClient

Writing Tests

Extended example

Passing data to the TestClient

Async tests

Why async tests?

Installing dependencies

Using `pytest.mark.anyio`

Why you can’t use TestClient for async tests

Using AsyncClient

Testing with other async operations

Pending