从Firebase Storage获取ReadStream传入OpenAI API遇400错误

核心差异 #

Firebase Storage（底层基于Google Cloud Storage）的ReadStream和Node.js原生fs.ReadStream有两个关键区别：

• fs.ReadStream自带path属性，OpenAI SDK可通过该属性自动推断文件的MIME类型和文件名，从而正确构建multipart/form-data请求。
• GCS的ReadStream默认没有这些元数据属性，且属于远程流，初始化存在延迟，直接传递给OpenAI接口时，SDK无法识别文件类型，或接收到不完整的数据流，最终触发400 Bad Request错误。

解决方案1：为GCS流补充元数据属性 #

手动给GCS流添加文件名和MIME类型，让OpenAI SDK能正确识别文件信息：

const fileStorage = await getStorageAdmin(uid, "/sample.png");
// 获取文件元数据
const [fileMetadata] = await fileStorage.getMetadata();
const file = fileStorage.createReadStream();
// 补充文件标识信息
file.name = "sample.png";
file.type = fileMetadata.contentType;

const maskStorage = await getStorageAdmin(uid, "/mask.png");
const [maskMetadata] = await maskStorage.getMetadata();
const mask = maskStorage.createReadStream();
mask.name = "mask.png";
mask.type = maskMetadata.contentType;

const res = await openai.createImageEdit(
    file,
    prompt,
    mask,
    1,
    "512x512",
    "url"
).catch((e) => {
    console.log("ERROR: " + e)
})

解决方案2：将GCS流转换为Buffer #

把远程流转换为本地Buffer后再传递，彻底避免流初始化延迟和元数据缺失问题：

import { pipeline } from 'stream/promises';
import { Buffer } from 'buffer';

// 流转Buffer工具函数
const streamToBuffer = async (stream) => {
  const chunks = [];
  await pipeline(stream, async function* (source) {
    for await (const chunk of source) {
      chunks.push(chunk);
    }
  });
  return Buffer.concat(chunks);
};

const fileStorage = await getStorageAdmin(uid, "/sample.png");
const fileBuffer = await streamToBuffer(fileStorage.createReadStream());
// 包装成带元数据的对象
const file = {
  data: fileBuffer,
  name: "sample.png",
  type: "image/png" // 也可从文件元数据中获取真实类型
};

const maskStorage = await getStorageAdmin(uid, "/mask.png");
const maskBuffer = await streamToBuffer(maskStorage.createReadStream());
const mask = {
  data: maskBuffer,
  name: "mask.png",
  type: "image/png"
};

const res = await openai.createImageEdit(
    file,
    prompt,
    mask,
    1,
    "512x512",
    "url"
).catch((e) => {
    console.log("ERROR: " + e)
})

验证方法 #

打印两种流的属性对比，直观确认元数据缺失问题：

console.log("FS流属性:", fs.createReadStream("./images/anna.png"));
console.log("GCS流属性:", fileStorage.createReadStream());

内容的提问来源于stack exchange，提问作者Duc Tran