将本地可用的Multer上传函数适配为Google Cloud Functions的Busboy实现

问题背景 #

我在本地使用multer的memoryStorage实现了以下文件上传代码，运行正常，但迁移到Google Cloud Functions时遇到问题，尝试用busboy但理不清思路，有两个疑问：

1. 直接用req.rawBody替代file.buffer可行吗？
2. 网上建议用os和fs，不知道原因是什么？

本地代码如下：

上传函数 #

async function uploadFile(file: any, fileName: string, targetFolder: string) {
  return new Promise((resolve, reject) => {
    const destinationFilepath: string = `${targetFolder ? targetFolder + '/' : ''}${fileName}`;
    const storage = new Storage();

    const bucket = storage.bucket(UPLOAD_BUCKET);
    const blob = bucket.file(destinationFilepath);
    const blobStream = blob.createWriteStream();

    blobStream.on('error', (err: Error) => reject(err));

    blobStream.on('finish', () => {
      resolve({ message: `Success! File uploaded to ${blob.publicUrl()}`, publicUrl: blob.publicUrl(), bucket: bucket.name, fileName: blob.name });
    });

    blobStream.end(file.buffer);
  });
}

Express POST路由 #

app.post('/gcs/upload/:folder?', async (req: Request, res: Response) => {
  const folder: string = req.params.folder || '';
  const fileName: string =  req?.file?.originalname ||  'FileWithoutFileName.txt';

  res.status(200).send(await uploadFile(req.rawBody, fileName, folder));
});

解决方案 #

1. 核心原理 #

Google Cloud Functions的请求处理逻辑和本地Express存在差异，multer的memoryStorage无法直接适配GCF的请求对象，必须用busboy来解析multipart/form-data类型的请求（文件上传的标准格式）。

2. 具体修改步骤 #

（1）安装依赖

先安装busboy包：

npm install busboy

（2）复用并优化上传函数

你的uploadFile函数几乎不需要改动，仅需调整类型定义让逻辑更严谨：

import { Storage } from '@google-cloud/storage';

// 替换为你的实际存储桶名称
const UPLOAD_BUCKET = 'your-bucket-name';

async function uploadFile(fileBuffer: Buffer, fileName: string, targetFolder: string) {
  return new Promise((resolve, reject) => {
    const destinationFilepath = `${targetFolder ? `${targetFolder}/` : ''}${fileName}`;
    const storage = new Storage();

    const bucket = storage.bucket(UPLOAD_BUCKET);
    const blob = bucket.file(destinationFilepath);
    const blobStream = blob.createWriteStream();

    blobStream.on('error', (err: Error) => reject(err));

    blobStream.on('finish', () => {
      resolve({ 
        message: `Success! File uploaded to ${blob.publicUrl()}`, 
        publicUrl: blob.publicUrl(), 
        bucket: bucket.name, 
        fileName: blob.name 
      });
    });

    blobStream.end(fileBuffer);
  });
}

（3）修改GCF请求处理函数

替换原Express路由，改为GCF的导出函数，用busboy解析请求：

import * as Busboy from 'busboy';
import { Request, Response } from 'express';

export const uploadToGCS = async (req: Request, res: Response) => {
  // 拦截非POST请求
  if (req.method !== 'POST') {
    return res.status(405).send('Method Not Allowed');
  }

  const busboy = new Busboy({ headers: req.headers });
  const folder = req.params.folder || '';
  let uploadedFile: { buffer: Buffer, filename: string };

  // 解析上传的文件字段
  busboy.on('file', (fieldname, file, filename) => {
    const chunks: Buffer[] = [];
    file.on('data', (chunk) => chunks.push(chunk));
    file.on('end', () => {
      uploadedFile = {
        buffer: Buffer.concat(chunks),
        filename: filename || 'FileWithoutFileName.txt'
      };
    });
  });

  // 解析完成后执行上传逻辑
  busboy.on('finish', async () => {
    if (!uploadedFile) {
      return res.status(400).send('No file uploaded');
    }

    try {
      const result = await uploadFile(uploadedFile.buffer, uploadedFile.filename, folder);
      res.status(200).send(result);
    } catch (err) {
      res.status(500).send(`Upload failed: ${err.message}`);
    }
  });

  // 启动请求体解析
  busboy.end(req.rawBody);
};

疑问解答 #

1. 直接用req.rawBody可行吗？
   不行。multipart/form-data类型的请求体包含分隔符（boundary）、文件头和文件内容等额外信息，req.rawBody是整个请求的原始二进制数据，直接写入GCS会导致文件包含冗余信息而损坏，必须通过busboy解析后才能得到纯净的文件buffer。

2. 为什么网上建议用os和fs？
   这类方案是把文件先写入GCF的临时目录（os.tmpdir()），再通过fs读取后上传，适合处理超大文件（避免内存溢出）。但如果你的文件体积不大，直接用内存buffer更高效，无需写入本地文件。

内容的提问来源于stack exchange，提问作者Spurious