如何在Pytest中复用不同用户Fixture生成对应认证Token？

我在conftest.py中定义了三个session级的用户Fixture：

@pytest.fixture(scope="session")
def user1(db: Session):
    user_create = UserCreate(name="u1", email="u1@email.com", password="pass")
    return Users.create(db=db, user_create=user_create)


@pytest.fixture(scope="session")
def user2(db: Session):
    user_create = UserCreate(name="u2", email="u2@email.com", password="pass")
    return Users.create(db=db, user_create=user_create)


@pytest.fixture(scope="session")
def user3(db: Session):
    user_create = UserCreate(name="u3", email="u3@email.com", password="pass")
    return Users.create(db=db, user_create=user_create)

同时有一个工具函数user_auth_test_token，可根据用户凭证生成认证Bearer Token：

def user_auth_test_token(
    client: TestClient, db: Session, email: EmailStr, name: str, password: str
):
    user = Users.get(db=db, email=email)
    if not user:
        user = Users.create(
            db=db, user_create=UserCreate(name=name, email=email, password=password)
        )

    return _user_auth_header(client=client, email=email, password=password)

我有多个需认证的FastAPI接口，测试时需传入对应认证头，部分测试场景还需多用户同时登录。当前测试时需手动调用user_auth_test_token生成认证头，代码示例如下：

def test_auth(client: TestClient, user1: Users):
    header = user_auth_test_token(client=client, db=db, email=user1.email, name=user1.name, password='pass')
    r = client.post(f"{settings.API_STR}/users/me", headers=header)
    ...

这类样板代码重复繁琐，请问能否创建参数化的user_token Fixture，接收上述任意用户Fixture并自动生成对应认证头？有没有更优的实现方案？

方案1：参数化user_token Fixture #

通过pytest的间接参数化，实现接收任意用户Fixture并生成对应token的功能。在conftest.py中添加：

@pytest.fixture(scope="session")
def user_token(request, client: TestClient, db: Session):
    user = request.param
    return user_auth_test_token(
        client=client,
        db=db,
        email=user.email,
        name=user.name,
        password="pass"
    )

测试用例中通过parametrize指定用户Fixture：

@pytest.mark.parametrize("user_token", [pytest.lazy_fixture("user1")], indirect=True)
def test_auth(client: TestClient, user_token):
    r = client.post(f"{settings.API_STR}/users/me", headers=user_token)
    assert r.status_code == 200
    assert r.json()["email"] == "u1@email.com"

多用户场景可叠加参数化：

@pytest.mark.parametrize("user1_token", [pytest.lazy_fixture("user1")], indirect=True)
@pytest.mark.parametrize("user2_token", [pytest.lazy_fixture("user2")], indirect=True)
def test_multi_user(client: TestClient, user1_token, user2_token):
    r1 = client.post(f"{settings.API_STR}/users/me", headers=user1_token)
    assert r1.json()["name"] == "u1"
    
    r2 = client.post(f"{settings.API_STR}/users/me", headers=user2_token)
    assert r2.json()["name"] == "u2"

方案2：为每个用户单独生成token Fixture #

如果常用用户数量固定，直接为每个用户创建对应token Fixture，使用更直观：

@pytest.fixture(scope="session")
def user1_token(client: TestClient, db: Session, user1: Users):
    return user_auth_test_token(
        client=client,
        db=db,
        email=user1.email,
        name=user1.name,
        password="pass"
    )

@pytest.fixture(scope="session")
def user2_token(client: TestClient, db: Session, user2: Users):
    return user_auth_test_token(
        client=client,
        db=db,
        email=user2.email,
        name=user2.name,
        password="pass"
    )

@pytest.fixture(scope="session")
def user3_token(client: TestClient, db: Session, user3: Users):
    return user_auth_test_token(
        client=client,
        db=db,
        email=user3.email,
        name=user3.name,
        password="pass"
    )

测试用例直接注入即可：

def test_auth(client: TestClient, user1_token):
    r = client.post(f"{settings.API_STR}/users/me", headers=user1_token)
    assert r.status_code == 200

def test_multi_user(client: TestClient, user1_token, user2_token):
    # 多用户逻辑实现
    pass

方案3：工厂Fixture（灵活度最高） #

适合动态生成任意用户token或测试临时用户的场景，在conftest.py中添加：

@pytest.fixture(scope="session")
def get_user_token(client: TestClient, db: Session):
    def _get_token(user: Users, password: str = "pass"):
        return user_auth_test_token(
            client=client,
            db=db,
            email=user.email,
            name=user.name,
            password=password
        )
    return _get_token

测试用例中调用工厂函数生成token：

def test_auth(client: TestClient, user1: Users, get_user_token):
    user1_token = get_user_token(user1)
    r = client.post(f"{settings.API_STR}/users/me", headers=user1_token)
    assert r.status_code == 200

# 临时用户测试场景
def test_temp_user(client: TestClient, db: Session, get_user_token):
    temp_user = Users.create(db=db, user_create=UserCreate(name="temp", email="temp@test.com", password="temp_pass"))
    temp_token = get_user_token(temp_user, password="temp_pass")
    r = client.post(f"{settings.API_STR}/users/me", headers=temp_token)
    assert r.json()["name"] == "temp"

优化建议 #

1. 简化user_auth_test_token：因为用户Fixture已提前创建用户，可去掉if not user的判断逻辑，减少不必要的数据库查询。
2. 提取密码常量：如果所有测试用户密码固定为pass，可将其定义为常量，避免硬编码。
3. 场景适配：固定用户测试选方案2，批量同逻辑测试选方案1，动态用户测试选方案3。

内容的提问来源于stack exchange，提问作者Shiladitya Bose