调用OpenAI API遇类型错误：ChatCompletion无'data'属性如何解决？

你遇到的错误是因为当前使用的OpenAI SDK（v4版本）中，chat.completions.create方法返回的response本身就是ChatCompletion类型实例，该类型不存在data属性，直接解构data会触发类型错误。

正确解决方案 #

直接从response的choices数组中提取模型生成的内容即可，由于你设置了n:1，数组只会包含一条结果。修改后的完整代码如下：

import openai from "@/openai";
import { NextResponse } from "next/server";

export async function POST(request: Request){
    const {todos} = await request.json();

    const response = await openai.chat.completions.create({
        model: "gpt-3.5-turbo",
        temperature: 0.8,
        n: 1,
        stream: false,
        messages: [
            {
                role: "system",
                content: `When responding, welcome the user always as Hello User and say Welcome to the Todo List App!
                Limit text to 200 characters.`
            },
            {
                role: "user",
                content: `Hi there, provide a summary of the following todos. Count how many todos are in each category such as To do, in progress and done, 
                then tell the user to have a productive day! Here's the data: ${JSON.stringify(
                    todos
                )}`
            }
        ]
    });

    // 提取模型生成的总结内容
    const { choices: [{ message: { content: summary } }] } = response;

    // 返回JSON结果给前端
    return NextResponse.json({ summary });
}

关键说明 #

• ChatCompletion类型的核心结构中，choices字段存储模型的返回结果数组，每个元素包含message对象，message.content就是生成的文本内容。
• 如果后续设置n大于1，需要遍历choices数组处理每条返回结果。

内容的提问来源于stack exchange，提问作者user9187915