AI大模型的使用-让AI帮你写单元测试

1.体验多步提示语

我们本节就让AI帮我们写一个单元测试，全程用AI给我们答案，首先单元测试前需要有代码，那么我们让AI给我们生成一个代码，要求如下：

用Python写一个函数，进行时间格式化输出，比如：
输入  输出
1  1s
61  1min1s
要求仅需要格式化到小时(?h?min?s)，即可

假如AI给了我们这样一段代码，比如，输入 1 就返回 1s，输入 61 就返回 1min1s，代码如下：

def format_time(seconds):minutes, seconds = divmod(seconds, 60)hours, minutes = divmod(minutes, 60)if hours > 0:return f"{hours}h{minutes}min{seconds}s"elif minutes > 0:return f"{minutes}min{seconds}s"else:return f"{seconds}s"
print(format_time(1))

我们可以用单元测试一下下面代码，单元测试包采取pytest

!pip install pytest
import pytestdef test_format_time():assert format_time(1) == "1s"assert format_time(59) == "59s"assert format_time(60) == "1min0s"assert format_time(61) == "1min1s"assert format_time(3600) == "1h0min0s"assert format_time(3661) == "1h1min1s"

ok，没有问题，但是关注点放入这个单元测试，我们发现这个单元测试过于简单，很多边界情况或者容易出现异常的情况都没有测试到，如，没有包含异常情况以及类型不一致或者空情况的测试。

我们想让AI生成完美的单元测试就需要一步一步引导，让它理解代码是干什么的，才能进行单元测试的编写。

下面我们就给出代码让它解释出我们的代码是干嘛的，看解释的是否准确，代码比较简单，

1.定义代码变量code母的是将代码给到openAI让它解释这段代码，

2.定义prompt跟Ai说明需求，这里需求是“让它写单元测试，并用Python 3.10和pytest高级特性写代码出合适的单元测试代码验证有效性，写代码之前，先去了解作者的意图”将需求promot和代码传给 AI即可，

3.最后我们防止直接给出代测试代码（不完美的单元测试），我们需要用stop及时停止，让它只返回代码解释说明，看看它有没有理解。

#! pip install openai
import openai
openai.api_key='sk'
# 使用text-davinci-002模型，是一个通过监督学习微调的生成文本的模型。因为这里我们希望生成目标明确的文本的代码解释，所以选用了这个模型。
# openAI调用
def gpt35(prompt, model="text-davinci-002", temperature=0.4, max_tokens=1000, top_p=1, stop=["\n\n", "\n\t\n", "\n    \n"]):response = openai.Completion.create(model=model,prompt = prompt,temperature = temperature,max_tokens = max_tokens,top_p = top_p,stop = stop)message = response["choices"][0]["text"]return message# 代码
code = """
def format_time(seconds):minutes, seconds = divmod(seconds, 60)hours, minutes = divmod(minutes, 60)if hours > 0:return f"{hours}h{minutes}min{seconds}s"elif minutes > 0:return f"{minutes}min{seconds}s"else:return f"{seconds}s"
"""# promot跟AI说明要做什么
def explain_code(function_to_test, unit_test_package="pytest"):prompt = f""""# How to write great unit tests with {unit_test_package}In this advanced tutorial for experts, we'll use Python 3.10 and `{unit_test_package}` to write a suite of unit tests to verify the behavior of the following function.
```python
{function_to_test}Before writing any unit tests, let's review what each element of the function is doing exactly and what the author's intentions may have been.
- First,"""response = gpt35(prompt)return response, prompt
# 在最后一行用 “- First” 开头，引导 GPT 模型，逐步分行描述要测试的代码干了什么。# 封装调用
code_explaination, prompt_to_explain_code = explain_code(code)-
print(code_explaination)

结果：

the function `format_time` takes in a single parameter `seconds` which should be a `float` or an `int`.
- The function then uses `divmod` to calculate the number of `minutes` and `seconds` from the total number of `seconds`.
- Next, `divmod` is used again to calculate the number of `hours` and `minutes` from the total number of `minutes`.
- Finally, the function returns a `string` formatted according to the number of `hours`, `minutes`, and `seconds`.

接下来根据上面的反馈以及进一步的需求，让它帮忙写出一个简单的但是包含多种情况的单元测试例子，我们提出测试单元的需求，看看它能否给出可靠的全面的测试要求，代码如下，

# 要求如下
# 1.我们要求测试用例，尽量考虑输入的范围广一些。
# 2.我们要求 AI 想一些连代码作者没有想到过的边界条件。
# 3.我们希望 AI 能够利用好 pytest 这个测试包的特性。
# 4.希望测试用例清晰易读，测试的代码要干净。
# 5.我们要求测试代码的输出结果是确定的，要么通过，要么失败，不要有随机性。
def generate_a_test_plan(full_code_explaination, unit_test_package="pytest"):prompt_to_explain_a_plan = f"""A good unit test suite should aim to:
- Test the function's behavior for a wide range of possible inputs
- Test edge cases that the author may not have foreseen
- Take advantage of the features of `{unit_test_package}` to make the tests easy to write and maintain
- Be easy to read and understand, with clean code and descriptive names
- Be deterministic, so that the tests always pass or fail in the same way`{unit_test_package}` has many convenient features that make it easy to write and maintain unit tests. We'll use them to write unit tests for the function above.For this particular function, we'll want our unit tests to handle the following diverse scenarios (and under each scenario, we include a few examples as sub-bullets):
-"""prompt = full_code_explaination + prompt_to_explain_a_planresponse = gpt35(prompt)return response, prompttest_plan, prompt_to_get_test_plan = generate_a_test_plan(prompt_to_explain_code + code_explaination)
print(test_plan)

结果：当然每次执行返回不一样，因为大模型的随机性。可以看到它能够考虑到异常情况，以及字符不符合会出现什么问题。

 Normal inputs- `format_time(0)` should return `"0s"`- `format_time(1)` should return `"1s"`- `format_time(59)` should return `"59s"`- `format_time(60)` should return `"1min0s"`- `format_time(61)` should return `"1min1s"`- `format_time(3599)` should return `"59min59s"`- `format_time(3600)` should return `"1h0min0s"`- `format_time(3601)` should return `"1h0min1s"`- `format_time(3660)` should return `"1h1min0s"`- `format_time(7199)` should return `"1h59min59s"`- `format_time(7200)` should return `"2h0min0s"`
- Invalid inputs- `format_time(None)` should raise a `TypeError`- `format_time("1")` should raise a `TypeError`- `format_time(-1)` should raise a `ValueError`

可以根据它提供的修改为可执行的单元测试：

import pytestdef test_format_time():assert format_time(0) == "0s"assert format_time(1) == "1s"assert format_time(59) == "59s"assert format_time(60) == "1min0s"assert format_time(61) == "1min1s"assert format_time(3600) == "1h0min0s"assert format_time(3661) == "1h1min1s"assert format_time(7200) == "2h0min0s"assert format_time(None) == "TypeError"assert format_time("1") == "TypeError"assert format_time(-1) == "ValueError"

没问题继续往下走，要是生成的测试计划数不满意，可以判断用例数是否小于指定的数，如果是那么就继续promot让其再生成按照如下的要求。

not_enough_test_plan = """The function is called with a valid number of seconds- `format_time(1)` should return `"1s"`- `format_time(59)` should return `"59s"`- `format_time(60)` should return `"1min"`
"""approx_min_cases_to_cover = 7
# 上下文用例数是否小于指定数字
elaboration_needed = test_plan.count("\n-") +1 < approx_min_cases_to_cover 
# 是的话就提需求，在调用下AI
if elaboration_needed:prompt_to_elaborate_on_the_plan = f"""In addition to the scenarios above, we'll also want to make sure we don't forget to test rare or unexpected edge cases (and under each edge case, we include a few examples as sub-bullets):
-"""more_test_plan, prompt_to_get_test_plan = generate_a_test_plan(prompt_to_explain_code + code_explaination + not_enough_test_plan + prompt_to_elaborate_on_the_plan)print(more_test_plan)

结果：

The function is called with a valid number of seconds- `format_time(1)` should return `"1s"`- `format_time(59)` should return `"59s"`- `format_time(60)` should return `"1min"`
- The function is called with an invalid number of seconds- `format_time(-1)` should raise a `ValueError`- `format_time("1")` should raise a `TypeError`
- The function is called with a valid number of minutes- `format_time(60)` should return `"1min"`- `format_time(119)` should return `"1min59s"`
- The function is called with an invalid number of minutes- `format_time(-1)` should raise a `ValueError`- `format_time("1")` should raise a `TypeError`
- The function is called with a valid number of hours- `format_time(3600)` should return `"1h"`- `format_time(7199)` should return `"1h59min"`
- The function is called with an invalid number of hours- `format_time(-1)` should raise a `ValueError`- `format_time("1")` should raise a `TypeError`

2.根据测试计划生成测试代码

我们根据上面给的需求以及AI给的答案拼装到新添加的prompt的给到AI，然后让它给我们一个测试单元代码

def generate_test_cases(function_to_test, unit_test_package="pytest"):# 将内容加载到prompt的{starter_comment}中starter_comment = "Below, each test case is represented by a tuple passed to the @pytest.mark.parametrize decorator"# promptprompt_to_generate_the_unit_test = f"""Before going into the individual tests, let's first look at the complete suite                 of unit tests as a cohesive whole. We've added helpful comments to explain what      each line does.```pythonimport {unit_test_package}  # used for our unit tests{function_to_test}#{starter_comment}"""# 将所有的需求和ai给的结果拼接到prompt里，目的为了给的结果更准确full_unit_test_prompt = prompt_to_explain_code + code_explaination + test_plan + prompt_to_generate_the_unit_testreturn gpt35(model="text-davinci-003", prompt=full_unit_test_prompt, stop="```"), prompt_to_generate_the_unit_testunit_test_response, prompt_to_generate_the_unit_test = generate_test_cases(code)
print(unit_test_response)

结果：可以拿过去直接执行

#The first element of the tuple is the name of the test case, and the second element is a dict
#containing the arguments to be passed to the function.
@pytest.mark.parametrize("test_case_name, test_case_args",[("positive_integer", {"seconds": 1}),("positive_integer_60", {"seconds": 60}),("positive_integer_3600", {"seconds": 3600}),("positive_integer_3601", {"seconds": 3601}),("negative_integer", {"seconds": -1}),("negative_integer_60", {"seconds": -60}),("negative_integer_3600", {"seconds": -3600}),("negative_integer_3601", {"seconds": -3601}),("float", {"seconds": 1.5}),("float_60", {"seconds": 60.5}),("float_3600", {"seconds": 3600.5}),("string", {"seconds": "1"}),("string_60", {"seconds": "60"}),("string_3600", {"seconds": "3600"}),("string_3601", {"seconds": "3601"}),("decimal", {"seconds": Decimal("1")}),("decimal_60", {"seconds": Decimal("60")}),("decimal_3600", {"seconds": Decimal("3600")}),("decimal_3601", {"seconds": Decimal("3601")}),],
)
def test_format_time(test_case_name, test_case_args):# Here, we use the test case name to determine the expected output.# This allows us to DRY up our code and avoid repeating ourselves.if "positive_integer" in test_case_name:expected = f"{test_case_args['seconds']}s"elif "positive_integer_60" in test_case_name:expected = "1min"elif "positive_integer_3600" in test_case_name:expected = "1h"elif "positive_integer_3601" in test_case_name:expected = "1h0min1s"elif "negative_integer" in test_case_name:expected = f"-{test_case_args['seconds']}s"elif "negative_integer_60" in test_case_name:expected = "-1min"elif "negative_integer_3600" in test_case_name:expected = "-1h"elif "negative_integer_3601" in test_case_name:expected = "-1h0min1s"elif "float" in test_case_name:expected = f"{test_case_args['seconds']}s"elif "float_60" in test_case_name:expected = "1min0.5s"elif "float_3600" in test_case_name:expected = "1h0.5s"elif "string" in test_case_name:expected = f"{test_case_args['seconds']}s"elif "string_60" in test_case_name:expected = "1min"elif "string_3600" in test_case_name:expected = "1h"elif "string_3601" in test_case_name:expected = "1h0min1s"elif "decimal" in test_case_name:expected = f"{test_case_args['seconds']}s"elif "decimal_60" in test_case_name:expected = "1min"elif "decimal_3600" in test_case_name:expected = "1h"elif "decimal_3601" in test_case_name:expected = "1h0min1s"# Now that we have the expected output, we can call the function and assert that the output is

那么如果全程自动化的话，我并不知道是否给的代码是否是有语句错误等行为，所以我们可以借用AST 库进行语法检查。

3.通过AST 库进行语法检查

可以检测下AI生成的代码，用Python 的 AST 库来完成。代码也是很简单，我们找到代码片段，将代码传给ast.parse()进行校验，有问题则抛出异常

# 语义检查
import ast
# 找到代码的索引
code_start_index = prompt_to_generate_the_unit_test.find("```python\n") + len("```python\n")
# 找到prompt_to_generate_the_unit_test里从187到最后的内容，并拼接单元测试代码的内容
code_output = prompt_to_generate_the_unit_test[code_start_index:] + unit_test_response
print("code_output：",code_output)# 进行语义检查
try:ast.parse(code_output)
except SyntaxError as e:print(f"Syntax error in generated code: {e}")

本节知识资料感谢徐文浩老师的《AI大模型之美》，让我感受它是真的美！