你当前正在访问 Microsoft Azure Global Edition 技术文档网站。 如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站,请访问 https://docs.azure.cn。
单元测试是现代软件开发实践的重要组成部分。 单元测试验证业务逻辑行为,并防止将来引入未注意到的中断性变更。 Durable Functions 可以轻松增加复杂性,因此引入单元测试有助于避免重大更改。 以下部分介绍如何对三种函数类型(业务流程客户端、业务流程协调程序和实体函数)进行单元测试。
注释
本指南仅适用于以 Python v2 编程模型编写的 Durable Functions 应用。
先决条件
本文中的示例需要了解以下概念和框架:
- 单元测试
- 持久性函数
- Python unittest
- unittest.mock
设置测试环境
若要测试 Durable Functions,设置适当的测试环境至关重要。 这包括创建测试目录并将 Python 模块 unittest
安装到 Python 环境中。 有关详细信息,请参阅 Azure Functions Python 单元测试概述。
对触发器函数进行单元测试
触发器函数(通常称为 客户端 函数)启动业务流程和外部事件。 若要测试以下函数,请执行以下步骤:
- 模拟
DurableOrchestrationClient
以仿真业务流程执行和状态管理。 - 将
DurableOrchestrationClient
等方法与start_new
、get_status
或raise_event
的模拟函数结合,使其返回预期值。 - 通过模拟客户端和其他必要输入(如 HTTP 触发器客户端函数的
req
(HTTP 请求对象))直接调用客户端函数。 - 使用断言和
unittest.mock
工具验证预期的业务流程启动行为、参数和 HTTP 响应。
import asyncio
import unittest
import azure.functions as func
from unittest.mock import AsyncMock, Mock, patch
from function_app import start_orchestrator
class TestFunction(unittest.TestCase):
@patch('azure.durable_functions.DurableOrchestrationClient')
def test_HttpStart(self, client):
# Get the original method definition as seen in the function_app.py file
func_call = http_start.build().get_user_function().client_function
req = func.HttpRequest(method='GET',
body=b'{}',
url='/api/my_second_function',
route_params={"functionName": "my_orchestrator"})
client.start_new = AsyncMock(return_value="instance_id")
client.create_check_status_response = Mock(return_value="check_status_response")
# Execute the function code
result = asyncio.run(func_call(req, client))
client.start_new.assert_called_once_with("my_orchestrator")
client.create_check_status_response.assert_called_once_with(req, "instance_id")
self.assertEqual(result, "check_status_response")
对业务流程协调程序函数进行单元测试
协调器函数管理多个活动函数的执行。 测试业务流程协调程序:
- 模拟
DurableOrchestrationContext
以控制函数执行。 - 将业务流程协调程序执行所需的
DurableOrchestrationContext
方法(如call_activity
或create_timer
)替换为模拟函数。 这些函数通常返回具有result
属性的 TaskBase 类型的对象。 - 以递归方式调用业务流程协调程序,将上一个 yield 语句生成的 Task 结果传递给下一个。
- 使用业务流程协调程序返回的结果和
unittest.mock
验证业务流程协调程序结果。
import unittest
from unittest.mock import Mock, patch, call
from datetime import timedelta
from azure.durable_functions.testing import orchestrator_generator_wrapper
from function_app import my_orchestrator
class TestFunction(unittest.TestCase):
@patch('azure.durable_functions.DurableOrchestrationContext')
def test_chaining_orchestrator(self, context):
# Get the original method definition as seen in the function_app.py file
func_call = my_orchestrator.build().get_user_function().orchestrator_function
# The mock_activity method is defined above with behavior specific to your app.
# It returns a TaskBase object with the result expected from the activity call.
context.call_activity = Mock(side_effect=mock_activity)
# Create a generator using the method and mocked context
user_orchestrator = func_call(context)
# Use orchestrator_generator_wrapper to get the values from the generator.
# Processes the orchestrator in a way that is equivalent to the Durable replay logic
values = [val for val in orchestrator_generator_wrapper(user_orchestrator)]
expected_activity_calls = [call('say_hello', 'Tokyo'),
call('say_hello', 'Seattle'),
call('say_hello', 'London')]
self.assertEqual(context.call_activity.call_count, 3)
self.assertEqual(context.call_activity.call_args_list, expected_activity_calls)
self.assertEqual(values[3], ["Hello Tokyo!", "Hello Seattle!", "Hello London!"])
对实体函数进行单元测试
实体函数通过操作管理有状态对象。 测试一个实体函数:
- 模拟
DurableEntityContext
以仿真实体的内部状态和操作输入。 - 将
DurableEntityContext
、get_state
和set_state
等方法替换为返回受控值的operation_name
模拟。 - 借助模拟上下文直接调用实体函数。
- 使用断言来验证状态更改、返回值和
unittest.mock
工具的正确性。
import unittest
from unittest.mock import Mock, patch
from function_app import Counter
class TestEntityFunction(unittest.TestCase):
@patch('azure.durable_functions.DurableEntityContext')
def test_entity_add_operation(self, context_mock):
# Get the original method definition as seen in function_app.py
func_call = Counter.build().get_user_function().entity_function
# Setup mock context behavior
state = 0
result = None
def set_state(new_state):
nonlocal state
state = new_state
def set_result(new_result):
nonlocal result
result = new_result
context_mock.get_state = Mock(return_value=state)
context_mock.set_state = Mock(side_effect=set_state)
context_mock.operation_name = "add"
context_mock.get_input = Mock(return_value=5)
context_mock.set_result = Mock(side_effect=lambda x: set_result)
# Call the entity function with the mocked context
func_call(context_mock)
# Verify the state was updated correctly
context_mock.set_state.assert_called_once_with(5)
self.assertEqual(state, 5)
self.assertEqual(result, None)
对活动函数进行单元测试
活动函数无需进行任何特定于 Durable 的修改即可测试。 Azure Functions Python 单元测试概述中找到的指南足以测试这些函数。