# Sobre

与确保应用程序在启动时运行类似，您可能还想确保它在挂掉后**重新启动**。

### 我们会犯错误

作为人类，我们总是会犯**错误**。 软件几乎*总是*在不同的地方隐藏着**bug**。 🐛

作为开发人员，当我们发现这些bug并实现新功能（也可能添加新bug😅）时，我们会不断改进代码。

### 自动处理小错误

使用 FastAPI 构建 Web API 时，如果我们的代码中存在错误，FastAPI 通常会将其包含到触发错误的单个请求中。 🛡

对于该请求，客户端将收到 **500 内部服务器错误**，但应用程序将继续处理下一个请求，而不是完全崩溃。

### 更大的错误 - 崩溃

尽管如此，在某些情况下，我们编写的一些代码可能会导致整个应用程序崩溃，从而导致 Uvicorn 和 Python 崩溃。 💥

import importlib

import pytest
from dirty_equals import IsDict
from fastapi.testclient import TestClient
from inline_snapshot import snapshot

from tests.utils import needs_py39, needs_py310, needs_pydanticv1, needs_pydanticv2


@pytest.fixture(
    name="client",
    params=[
        pytest.param("tutorial002", marks=needs_pydanticv2),
        pytest.param("tutorial002_py39", marks=[needs_py39, needs_pydanticv2]),

from fastapi.testclient import TestClient

from docs_src.path_operation_advanced_configuration.tutorial001 import app

client = TestClient(app)


def test_get():
    response = client.get("/items/")
    assert response.status_code == 200, response.text
    assert response.json() == [{"item_id": "Foo"}]


def test_openapi_schema():
    response = client.get("/openapi.json")
    assert response.status_code == 200, response.text

from dirty_equals import IsDict
from fastapi.testclient import TestClient

from docs_src.dataclasses.tutorial001 import app

client = TestClient(app)


def test_post_item():
    response = client.post("/items/", json={"name": "Foo", "price": 3})
    assert response.status_code == 200
    assert response.json() == {
        "name": "Foo",
        "price": 3,
        "description": None,
        "tax": None,
    }

import pytest
from dirty_equals import IsDict
from fastapi.testclient import TestClient

from docs_src.dependencies.tutorial004 import app

client = TestClient(app)


@pytest.mark.parametrize(
    "path,expected_status,expected_response",
    [
        (
            "/items",
            200,
            {
                "items": [
                    {"item_name": "Foo"},
                    {"item_name": "Bar"},

import pytest
from dirty_equals import IsDict
from fastapi.testclient import TestClient

from docs_src.dependencies.tutorial004_an import app

client = TestClient(app)


@pytest.mark.parametrize(
    "path,expected_status,expected_response",
    [
        (
            "/items",
            200,
            {
                "items": [
                    {"item_name": "Foo"},
                    {"item_name": "Bar"},

from fastapi.testclient import TestClient

from docs_src.behind_a_proxy.tutorial002 import app

client = TestClient(app)


def test_main():
    response = client.get("/app")
    assert response.status_code == 200
    assert response.json() == {"message": "Hello World", "root_path": "/api/v1"}


def test_openapi():
    response = client.get("/openapi.json")
    assert response.status_code == 200
    assert response.json() == {

Durch die Verwendung von `Security` anstelle von `Depends` weiß **FastAPI** jedoch, dass es Sicherheits-Scopes deklarieren, intern verwenden und die API mit OpenAPI dokumentieren kann.

Wenn Sie jedoch `Query`, `Path`, `Depends`, `Security` und andere von `fastapi` importieren, handelt es sich tatsächlich um Funktionen, die spezielle Klassen zurückgeben.

///

# Encoders - `jsonable_encoder`