{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Openai Function Call \n",
    "https://platform.openai.com/docs/guides/function-calling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "User>\t How's the weather in Hangzhou?\n",
      "[ChatCompletionMessageToolCall(id='call_zvxBS1zoCa6X9fJzE48BYo2z', function=Function(arguments='{\"location\":\"Hangzhou, China\"}', name='get_weather', parameters=None), type='function')]\n",
      "https://api.weatherapi.com/v1/current.json?key=b7e69154e60b427586983619250502&q=Hangzhou, China\n",
      "Tool call>\t Temperature: 13.0°C, Condition: Partly cloudy\n",
      "Model>\t The current weather in Hangzhou, China is partly cloudy with a temperature of 13.0°C.\n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "import json\n",
    "import requests\n",
    "\n",
    "# 定义获取天气的工具\n",
    "def get_weather(location: str):\n",
    "    \"\"\"调用天气 API 获取指定地点的天气信息\"\"\"\n",
    "    url = f\"https://api.weatherapi.com/v1/current.json?key=b7e69154e60b427586983619250502&q={location}\"\n",
    "    print(url)\n",
    "    response = requests.get(url)\n",
    "    if response.status_code == 200:\n",
    "        data = response.json()\n",
    "        return f\"Temperature: {data['current']['temp_c']}°C, Condition: {data['current']['condition']['text']}\"\n",
    "    return \"Weather data not found\"\n",
    "\n",
    "def send_messages(messages):\n",
    "    response = client.chat.completions.create(\n",
    "        model=\"gpt-4o\",\n",
    "        messages=messages,\n",
    "        tools=tools,  # 指定可调用的函数\n",
    "    )\n",
    "    return response.choices[0].message\n",
    "\n",
    "client = OpenAI(\n",
    "    api_key=\"sk-lXFW7Bl1ruw2qmHu287e979847354601A07fE2D85a567bD7\",             # pass litellm proxy key, if you're using virtual keys\n",
    "    base_url=\"https://yunwu.ai/v1/\" # litellm-proxy-base url\n",
    ")\n",
    "\n",
    "tools = [{\n",
    "    \"type\": \"function\",\n",
    "    \"function\": {\n",
    "        \"name\": \"get_weather\",\n",
    "        \"description\": \"Get current temperature for a given location.\",\n",
    "        \"parameters\": {\n",
    "            \"type\": \"object\",\n",
    "            \"properties\": {\n",
    "                \"location\": {\n",
    "                    \"type\": \"string\",\n",
    "                    \"description\": \"City and country e.g. Bogotá, Colombia\"\n",
    "                }\n",
    "            },\n",
    "            \"required\": [\n",
    "                \"location\"\n",
    "            ],\n",
    "            \"additionalProperties\": False\n",
    "        },\n",
    "        \"strict\": True\n",
    "    }\n",
    "}]\n",
    "# step1\n",
    "messages = [{\"role\": \"user\", \"content\": \"How's the weather in Hangzhou?\"}]\n",
    "message = send_messages(messages)\n",
    "print(f\"User>\\t {messages[0]['content']}\")\n",
    "print(message.tool_calls)\n",
    "\n",
    "# step2\n",
    "tool_call = message.tool_calls[0]\n",
    "\n",
    "#step3 tool call chioce to local function call\n",
    "args = json.loads(tool_call.function.arguments)\n",
    "result = get_weather(args[\"location\"])\n",
    "print(\"Tool call>\\t\", result)\n",
    "\n",
    "# step4\n",
    "messages.append(message)  # append model's function call message\n",
    "messages.append({                               # append result message\n",
    "    \"role\": \"tool\",\n",
    "    \"tool_call_id\": tool_call.id,\n",
    "    \"content\": result\n",
    "})\n",
    "\n",
    "message = send_messages(messages)\n",
    "\n",
    "# step5\n",
    "print(\"Model>\\t\", message.content)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Qwen Tool Use Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "------------------------------------------------------------\n",
      "\n",
      "第1轮大模型输出信息：{'id': 'chatcmpl-91008e14-9254-98b9-8b14-da6632f6c711', 'choices': [{'finish_reason': 'tool_calls', 'index': 0, 'logprobs': None, 'message': {'content': '', 'refusal': None, 'role': 'assistant', 'audio': None, 'function_call': None, 'tool_calls': [{'id': 'call_5ae24ff6a07e4a1093fff6', 'function': {'arguments': '{}', 'name': 'get_current_time'}, 'type': 'function', 'index': 0}]}}], 'created': 1738762202, 'model': 'qwen-max-2025-01-25', 'object': 'chat.completion', 'service_tier': None, 'system_fingerprint': None, 'usage': {'completion_tokens': 13, 'prompt_tokens': 219, 'total_tokens': 232, 'completion_tokens_details': None, 'prompt_tokens_details': None}}\n",
      "\n",
      "工具输出信息：当前时间：2025-02-05 21:29:58。\n",
      "\n",
      "------------------------------------------------------------\n",
      "第2轮大模型输出信息：{'content': '现在的时间是2025年2月5日，晚上9点29分58秒。', 'refusal': None, 'role': 'assistant', 'audio': None, 'function_call': None, 'tool_calls': None}\n",
      "\n",
      "最终答案：现在的时间是2025年2月5日，晚上9点29分58秒。\n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "from datetime import datetime\n",
    "import json\n",
    "import os\n",
    "\n",
    "client = OpenAI(\n",
    "    # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n",
    "    api_key=\"sk-487d366becc14df0a58ff4d4559fabf4\",\n",
    "    base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",  # 填写DashScope SDK的base_url\n",
    ")\n",
    "\n",
    "# 定义工具列表，模型在选择使用哪个工具时会参考工具的name和description\n",
    "tools = [\n",
    "    # 工具1 获取当前时刻的时间\n",
    "    {\n",
    "        \"type\": \"function\",\n",
    "        \"function\": {\n",
    "            \"name\": \"get_current_time\",\n",
    "            \"description\": \"当你想知道现在的时间时非常有用。\",\n",
    "            # 因为获取当前时间无需输入参数，因此parameters为空字典\n",
    "            \"parameters\": {}\n",
    "        }\n",
    "    },  \n",
    "    # 工具2 获取指定城市的天气\n",
    "    {\n",
    "        \"type\": \"function\",\n",
    "        \"function\": {\n",
    "            \"name\": \"get_current_weather\",\n",
    "            \"description\": \"当你想查询指定城市的天气时非常有用。\",\n",
    "            \"parameters\": {  \n",
    "                \"type\": \"object\",\n",
    "                \"properties\": {\n",
    "                    # 查询天气时需要提供位置，因此参数设置为location\n",
    "                    \"location\": {\n",
    "                        \"type\": \"string\",\n",
    "                        \"description\": \"城市或县区，比如北京市、杭州市、余杭区等。\"\n",
    "                    }\n",
    "                }\n",
    "            },\n",
    "            \"required\": [\n",
    "                \"location\"\n",
    "            ]\n",
    "        }\n",
    "    }\n",
    "]\n",
    "\n",
    "# 模拟天气查询工具。返回结果示例：“北京今天是雨天。”\n",
    "def get_current_weather(location):\n",
    "    return f\"{location}今天是雨天。 \"\n",
    "\n",
    "# 查询当前时间的工具。返回结果示例：“当前时间：2024-04-15 17:15:18。“\n",
    "def get_current_time():\n",
    "    # 获取当前日期和时间\n",
    "    current_datetime = datetime.now()\n",
    "    # 格式化当前日期和时间\n",
    "    formatted_time = current_datetime.strftime('%Y-%m-%d %H:%M:%S')\n",
    "    # 返回格式化后的当前时间\n",
    "    return f\"当前时间：{formatted_time}。\"\n",
    "\n",
    "# 封装模型响应函数\n",
    "def get_response(messages):\n",
    "    completion = client.chat.completions.create(\n",
    "        model=\"qwen-max-2025-01-25\",\n",
    "        messages=messages,\n",
    "        tools=tools\n",
    "        )\n",
    "    return completion.model_dump()\n",
    "\n",
    "def call_with_messages():\n",
    "    print('\\n')\n",
    "    messages = [\n",
    "            {\n",
    "                \"content\": input('现在几点了？'),  # 提问示例：\"现在几点了？\" \"一个小时后几点\" \"北京天气如何？\"\n",
    "                \"role\": \"user\"\n",
    "            }\n",
    "    ]\n",
    "    print(\"-\"*60)\n",
    "    # 模型的第一轮调用\n",
    "    i = 1\n",
    "    first_response = get_response(messages)\n",
    "    assistant_output = first_response['choices'][0]['message']\n",
    "    print(f\"\\n第{i}轮大模型输出信息：{first_response}\\n\")\n",
    "    if  assistant_output['content'] is None:\n",
    "        assistant_output['content'] = \"\"\n",
    "    messages.append(assistant_output)\n",
    "    # 如果不需要调用工具，则直接返回最终答案\n",
    "    if assistant_output['tool_calls'] == None:  # 如果模型判断无需调用工具，则将assistant的回复直接打印出来，无需进行模型的第二轮调用\n",
    "        print(f\"无需调用工具，我可以直接回复：{assistant_output['content']}\")\n",
    "        return\n",
    "    # 如果需要调用工具，则进行模型的多轮调用，直到模型判断无需调用工具\n",
    "    while assistant_output['tool_calls'] != None:\n",
    "        # 如果判断需要调用查询天气工具，则运行查询天气工具\n",
    "        if assistant_output['tool_calls'][0]['function']['name'] == 'get_current_weather':\n",
    "            tool_info = {\"name\": \"get_current_weather\", \"role\":\"tool\"}\n",
    "            # 提取位置参数信息\n",
    "            location = json.loads(assistant_output['tool_calls'][0]['function']['arguments'])['location']\n",
    "            tool_info['content'] = get_current_weather(location)\n",
    "        # 如果判断需要调用查询时间工具，则运行查询时间工具\n",
    "        elif assistant_output['tool_calls'][0]['function']['name'] == 'get_current_time':\n",
    "            tool_info = {\"name\": \"get_current_time\", \"role\":\"tool\"}\n",
    "            tool_info['content'] = get_current_time()\n",
    "        print(f\"工具输出信息：{tool_info['content']}\\n\")\n",
    "        print(\"-\"*60)\n",
    "        messages.append(tool_info)\n",
    "        assistant_output = get_response(messages)['choices'][0]['message']\n",
    "        if  assistant_output['content'] is None:\n",
    "            assistant_output['content'] = \"\"\n",
    "        messages.append(assistant_output)\n",
    "        i += 1\n",
    "        print(f\"第{i}轮大模型输出信息：{assistant_output}\\n\")\n",
    "    print(f\"最终答案：{assistant_output['content']}\")\n",
    "\n",
    "call_with_messages()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}