# 获取当前时间 - function calling

当我们询问AI现在几点时，他会回答无法获取当前时间：

<figure><img src="/files/VSpfuBVI4DdtgqGTzItf" alt=""><figcaption></figcaption></figure>

由于没有联网它无法获取当前正确的时间，现在我们可以通过prompt + function calling 让它实现这个功能。

我们先定一个 getCurrentTime的函数

```typescript
const functions = {
    getCurrentTime: function(){
        return new Date().toLocaleTimeString()
    }
}
```

然后在ChatCompletion传入functions参数，写上getCurrentTime函数名以及描述，这个目的是让GPT知道我们有哪些函数可以使用。然后做第一次请求

```typescript
const MODELS = {
    gpt35Turbo0613: "gpt-3.5-turbo-0613",
    gpt4Turbo0613: "gpt-4-0613",
}

const getChatCompletion = async ({messageList}: {
    messageList: Array<ChatCompletionRequestMessage>
}) => { 
    if(!messageList) return;
    let result = ``
    const getCurrentTimeFunObj = {
        name: `getCurrentTime`,
        description: `Get the current time`,
        parameters: {
            type: `object`,
            properties: {},
            required: []
        },
    }
    try {
        const response = await openai.createChatCompletion({
            model: MODELS.gpt35Turbo0613,
            messages: messages,
            functions: [getCurrentTimeFunObj],
            temperature: 0,
        })

        const { choices } = response?.data || {}
        result = choices[0].message || {}
    } catch (e) {
        console.log(`getChatCompletion error Info`, e)
    }
    return result
}

// prompt: "please tell me current time"
let messageList = [{
    role: "user", 
    content: "现在几点了？"
}]
const firstResponse = await getChatCompletion({messageList})
```

当我们在messageList中出现询问当前时间的时候，由于在调用中含有functions，根据描述GPT判断可以使用function。GPT就会以 "function\_call" 来终止并返回function\_call字段，让我们调用 getCurrentTime方法。

<pre class="language-typescript"><code class="lang-typescript"><strong>{
</strong>    id: `chatcmpl-1`,
    object: 'chat.completion',
    created: 1623662570,
    model: 'gpt-3.5-turbo-0613',
    choices: [{
        index: 0,
        message: {
            role: 'assistant',
            content: null,
            function_call: { name: "getCurrentTime", argumentss: '{}' }
        },
        finish_reason: 'function_call',
    }],
    user: { prompt_tokens: 43, completion_tokens: 7, total_tokens: 50 },
}
</code></pre>

在通过GPT指定要调用的方法来处理数据后，我们需要将结果以对应的role类型再次存入messageList进行二次调用

```typescript
const responseCallback = async ({respsone, messageList}: {
    respsone: ChatCompletionResponse,
    messageList: Array<ChatCompletionRequestMessage>
}) => {
    let result = ``;
    const { choices } = respsone?.data || {}
    const {finish_reason, message} = choices?.[0] || {}
    if(finish_reason == "function_call" && message?.function_call){
        const {name: fnName, arguments} = message.function_call || {}
        const fn = functions[fnName]
        if(fn){
            const fnResult = fn(arguments)
            messageList.push({
                role: 'assistant',
                content: null,
                function_call: { name: fnName, argumentss: arguments }
            });
            messageList.push({
                role: 'function',
                name: fnName,
                content: JSON.stringify({result: fnResult}),
            })

            result = await openai.createChatCompletion({messageList})
        }
    } 
    return result
}
```

最后，我们就能得到一个通过 function calling来实现的获取本地时间的结果了

```typescript
{
    message: {
        role: "assistant",
        content: "现在时间是13:57:50",
      }
}
```

总结一下函数调用的基本步骤：

1. 通过用户查询和在函数参数中定义的一组函数来调用模型。
2. 模型可以选择调用函数；如果调用，则内容将是一个符合你预定义模式的字符串化 JSON 对象。（注意：模型可能会生成无效的 JSON 或产生虚假参数）。
3. 在你的代码中将字符串解析为 JSON，并通过入参来调用调用你写的函数。
4. 将函数返回的结果作为新消息附加到模型中，然后让模型将结果汇总回用户。

{% hint style="info" %}
Reference:

[openai-function-calling-nodejs](https://github.com/JohannLai/openai-function-calling-nodejs)

[OpenAI Doc - Function calling](https://platform.openai.com/docs/guides/gpt/function-calling)

[OpenAI Cookbook - How to call functions with chat models](https://github.com/openai/openai-cookbook/blob/main/examples/How_to_call_functions_with_chat_models.ipynb)
{% endhint %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://aiweb.viku.org/practice/openai-function-calling/huo-qu-dang-qian-shi-jian-function-calling.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.