OpenAI 库 工具调用

代码实现，两个工具调用

import openai
import json
from openai import OpenAI 


# 定义工具函数
def get_temperature(city):
    """模拟获取城市温度的函数"""
    # 这里应该是实际的 API 调用，这里用模拟数据代替
    temperature_data = {
        "北京": 25,
        "上海": 28,
        "广州": 30,
        "深圳": 29,
        "成都": 26
    }
    return temperature_data.get(city, 20)  # 默认返回20度

def get_fit_cloth(temperature):
    """根据温度推荐合适的衣物"""
    if temperature > 28:
        return "短袖T恤和短裤"
    elif 24 <= temperature <= 28:
        return "短袖T恤和薄长裤"
    elif 18 <= temperature < 24:
        return "长袖衬衫和长裤"
    elif 10 <= temperature < 18:
        return "毛衣或薄外套"
    else:
        return "厚外套或羽绒服"

# 定义工具描述，供模型了解可用的工具
tools = [  
    {
        "type": "function",
        "function": {
            "name": "get_temperature", 
            "description": "获取指定城市的当前温度",
            "parameters": {
                "type": "object",
                "properties": {
                    "city": {
                        "type": "string",
                        "description": "城市名称，如北京、上海",
                    }
                },
                "required": ["city"],
            },
        },
    },
    {
        "type": "function",
        "function": {
            "name": "get_fit_cloth",
            "description": "根据温度获取适合的衣物推荐",
            "parameters": {
                "type": "object",
                "properties": {
                    "temperature": {
                        "type": "number",
                        "description": "温度值，单位是摄氏度",
                    }
                },
                "required": ["temperature"],
            },
        },
    }
]

base_url = "https://open.bigmodel.cn/api/paas/v4/" 
api_key = "..."
model_name = 'glm-4-flash'

import time 
t1 = time.time() 

client = OpenAI(api_key=api_key ,base_url=base_url ) 


def run_conversation(user_input):
    # 第一步：发送用户输入和工具描述给模型
    messages = [{"role": "user", "content": user_input}]

    response = client.chat.completions.create(  
        model=model_name,   
        messages=messages,
        tools=tools,
        # tool_choice="auto",  
    )                               
    
    print('-- response : ', response )  

    response_message = response.choices[0].message
    tool_calls = response_message.tool_calls
    
    # 第二步：检查模型是否想要调用工具
    if tool_calls:
        messages.append(response_message)  # 添加模型的响应到消息历史
        
        # 处理每个工具调用
        for tool_call in tool_calls:
            function_name = tool_call.function.name
            function_args = json.loads(tool_call.function.arguments)
            
            # 调用相应的函数
            if function_name == "get_temperature":
                function_response = get_temperature(
                    city=function_args.get("city")
                )
            elif function_name == "get_fit_cloth":
                function_response = get_fit_cloth(
                    temperature=function_args.get("temperature")
                )
            else:
                function_response = "未知工具调用"
            
            tool_message = {
                    "tool_call_id": tool_call.id,
                    "role": "tool",
                    "name": function_name,
                    "content": str(function_response),
                }
            
            messages.append( tool_message )   
        
        second_response = client.chat.completions.create(  
            model=model_name,   
            messages=messages,
            tools=tools,
            # tool_choice="auto",  
        )       
        
        print('\n-- second_response : ', second_response)
        return second_response.choices[0].message.content
    else:
        return response_message.content

# 示例使用
if __name__ == "__main__":
    # 示例1: 直接询问天气和衣物推荐
    print(run_conversation("北京现在适合穿什么衣服？"))
    
    # 示例2: 只询问温度
    # print(run_conversation("上海现在的温度是多少？"))
    
    # # 示例3: 已知温度询问衣物推荐
    # print(run_conversation("现在温度是22度，我该穿什么？"))

response 日志

--response: 
	ChatCompletion(
		id = '20250619170027a299743fa79c436f',
		choices = [
			Choice(
				finish_reason = 'tool_calls', 
        index = 0,
				logprobs = None,
				message = ChatCompletionMessage(
					content = None, refusal = None, role = 'assistant',
					annotations = None, audio = None,
					function_call = None,
					tool_calls = [
						ChatCompletionMessageToolCall(
							id = 'call_-8598006517705558361',
							function = Function(
								arguments = '{"city": "北京"}', name = 'get_temperature'
							),
							type = 'function', index = 0)
					]))
		],

		created = 1750323628,
		model = 'glm-4-flash',
		object = None,
		service_tier = None,
		system_fingerprint = None,
		usage = CompletionUsage(
			completion_tokens = 10,
			prompt_tokens = 270,
			total_tokens = 280,
			completion_tokens_details = None,
			prompt_tokens_details = None
		),
		request_id = '20250619170027a299743fa79c436f'
	)

	--second_response:
	ChatCompletion(
		id = '202506191700284b5faf039ce14317',
		choices = [
			Choice(
				finish_reason = 'tool_calls',
				index = 0,
				logprobs = None,
				message = ChatCompletionMessage(
					content = None,
					refusal = None,
					role = 'assistant',
					annotations = None, audio = None,
					function_call = None,
					tool_calls = [
						ChatCompletionMessageToolCall(
							id = 'call_-8598014111208728724',
							function = Function(
								arguments = '{"temperature": 25}',
								name = 'get_fit_cloth'
							),

							type = 'function',
							index = 0
						)
					]
        )
      )
		],
		created = 1750323628,
		model = 'glm-4-flash',
		object = None,
		service_tier = None, system_fingerprint = None,
		usage = CompletionUsage(
			completion_tokens = 12,
			prompt_tokens = 283,
			total_tokens = 295, 
      completion_tokens_details = None,
			prompt_tokens_details = None
		),
		request_id = '202506191700284b5faf039ce14317'
	)

tool_message

ChatCompletionMessage(
  content=None, 
  refusal=None, 
  role='assistant', 
  annotations=None, 
  audio=None, 
  function_call=None,  
  
  tool_calls=[
    ChatCompletionMessageToolCall(
      id='call_-8598006517705558361', 
      function=Function(
        arguments='{"city": "北京"}', 
        name='get_temperature'
      ), 
      type='function', 
      index=0
    )
  ]
  
)

2025-06-19（四）