首页 > 解决方案 > AWS SDK 文件通过 Node/Express 使用流 PassThrough 上传到 S3 - 文件总是损坏

问题描述

这很简单。使用此代码,任何上传的图像文件都已损坏且无法打开。PDF 看起来不错,但我注意到它正在将值注入基于文本的文件中。这是 s3 中正确的文件大小,而不是像出了问题那样为零。我不确定这是否是 Express、SDK 或两者兼而有之的问题?是邮递员吗?我在今年 3 月的一个工作项目中构建了类似的东西,它完美无缺。我不再可以访问该代码进行比较。

没有错误,没有任何问题的迹象。

const aws = require("aws-sdk");
const stream = require("stream");
const express = require("express");
const router = express.Router();

const AWS_ACCESS_KEY_ID = "XXXXXXXXXXXXXXXXXXXX";
const AWS_SECRET_ACCESS_KEY = "superSecretAccessKey";
const BUCKET_NAME = "my-bucket";
const BUCKET_REGION = "us-east-1";

const s3 = new aws.S3({
    region: BUCKET_REGION,
    accessKeyId: AWS_ACCESS_KEY_ID,
    secretAccessKey: AWS_SECRET_ACCESS_KEY
});

const uploadStream = key => {
    let streamPass = new stream.PassThrough();
    let params = {
        Bucket: BUCKET_NAME,
        Key: key,
        Body: streamPass
    };
    let streamPromise = s3.upload(params, (err, data) => {
        if (err) {
            console.error("ERROR: uploadStream:", err);
        } else {
            console.log("INFO: uploadStream:", data);
        }
    }).promise();
    return {
        streamPass: streamPass,
        streamPromise: streamPromise
    };
};

router.post("/upload", async (req, res) => {
    try {
        let key = req.query.file_name;
        let { streamPass, streamPromise } = uploadStream(key);
        req.pipe(streamPass);
        await streamPromise;
        res.status(200).send({ result: "Success!" });
    } catch (e) {
        res.status(500).send({ result: "Fail!" });
    }
});

module.exports = router;

这是我的 package.json:

{
  "name": "expresss3streampass",
  "version": "0.0.0",
  "private": true,
  "scripts": {
    "start": "node ./bin/www"
  },
  "dependencies": {
    "aws-sdk": "^2.812.0",
    "cookie-parser": "~1.4.4",
    "debug": "~2.6.9",
    "express": "~4.16.1",
    "morgan": "~1.9.1"
  }
}

更新

经过进一步测试,我注意到 Postman 正在更改纯文本文件。例如,这个源文件:

{
    "question_id": null,
    "position_type_id": 1,
    "question_category_id": 1,
    "position_level_id": 1,
    "question": "Do you test your code before calling it \"done\"?",
    "answer": "Candidate should respond that they at least happy path test every feature and bug fix they write.",
    "active": 1
}

...它落入桶中后看起来像这样:

----------------------------472518836063077482836177
Content-Disposition: form-data; name="file"; filename="question.json"
Content-Type: application/json

{
    "question_id": null,
    "position_type_id": 1,
    "question_category_id": 1,
    "position_level_id": 1,
    "question": "Do you test your code before calling it \"done\"?",
    "answer": "Candidate should respond that they at least happy path test every feature and bug fix they write.",
    "active": 1
}
----------------------------472518836063077482836177--

我不得不认为这是问题所在。邮递员是这个等式中唯一改变的东西,从这段代码第一次为我工作起。我的请求标头如下所示:

在此处输入图像描述

我是最初添加“application/x-www-form-urlencoded”标头的人。如果我现在使用它,我最终会在存储桶中得到一个 0 字节的文件。

标签: javascriptnode.jsexpressamazon-s3aws-sdk

解决方案


穆尔特是要走的路。

它提供了几种不同的模式,但据我所知,您必须编写一个自定义存储处理程序才能访问底层 Stream,否则它将缓冲内存中的所有数据,并且只有在完成后才会回调。

如果您签req.file入路由处理程序,Multer 通常会在该buffer字段下提供一个缓冲区,但它不再存在,因为我没有在回调中传递任何内容,所以我有理由相信这是按预期进行的流式传输。

下面是一个可行的解决方案。

注意:parse.single('image')被传递到路由处理程序中。这是指我使用的多部分字段名称。

const aws = require('aws-sdk');
const stream = require('stream');
const express = require('express');
const router = express.Router();
const multer = require('multer')

const AWS_ACCESS_KEY_ID = "XXXXXXXXXXXXXXXXXXXX";
const AWS_SECRET_ACCESS_KEY = "superSecretAccessKey";
const BUCKET_NAME = "my-bucket";
const BUCKET_REGION = "us-east-1";

const s3 = new aws.S3({
    region: BUCKET_REGION,
    accessKeyId: AWS_ACCESS_KEY_ID,
    secretAccessKey: AWS_SECRET_ACCESS_KEY
});

const uploadStream = key => {
    let streamPass = new stream.PassThrough();
    let params = {
        Bucket: BUCKET_NAME,
        Key: key,
        Body: streamPass
    };
    let streamPromise = s3.upload(params, (err, data) => {
        if (err) {
            console.error('ERROR: uploadStream:', err);
        } else {
            console.log('INFO: uploadStream:', data);
        }
    }).promise();
    return {
        streamPass: streamPass,
        streamPromise: streamPromise
    };
};

class CustomStorage {
    _handleFile(req, file, cb) {
        let key = req.query.file_name;
        let { streamPass, streamPromise } = uploadStream(key);
        file.stream.pipe(streamPass)
        streamPromise.then(() => cb(null, {}))
    }
}

const storage = new CustomStorage();
const parse = multer({storage});

router.post('/upload', parse.single('image'), async (req, res) => {
    try {
        res.status(200).send({ result: 'Success!' });
    } catch (e) {
        console.log(e)
        res.status(500).send({ result: 'Fail!' });
    }
});

module.exports = router;

更新:更好的解决方案

我上面提供的基于Multer的解决方案有点老套。所以我在引擎盖下看了一下,看看它是如何工作的。该解决方案仅使用Busboy来解析和流式传输文件。Multer 实际上只是一个带有一些磁盘 I/O 便利功能的包装器。

const aws = require('aws-sdk');
const express = require('express');
const Busboy = require('busboy');
const router = express.Router();

const AWS_ACCESS_KEY_ID = "XXXXXXXXXXXXXXXXXXXX";
const AWS_SECRET_ACCESS_KEY = "superSecretAccessKey";
const BUCKET_NAME = "my-bucket";
const BUCKET_REGION = "us-east-1";

const s3 = new aws.S3({
    region: BUCKET_REGION,
    accessKeyId: AWS_ACCESS_KEY_ID,
    secretAccessKey: AWS_SECRET_ACCESS_KEY
});

function multipart(request){
    return new Promise(async (resolve, reject) => {
        const headers = request.headers;
        const busboy = new Busboy({ headers });
        // you may need to add cleanup logic using 'busboy.on' events
        busboy.on('error', err => reject(err))
        busboy.on('file', function (fieldName, fileStream, fileName, encoding, mimeType) {
            const params = {
                Bucket: BUCKET_NAME,
                Key: fileName,
                Body: fileStream
            };
            s3.upload(params).promise().then(() => resolve());
        })
        request.pipe(busboy)
    })
}

router.post('/upload', async (req, res) => {
    try {
        await multipart(req)
        res.status(200).send({ result: 'Success!' });
    } catch (e) {
        console.log(e)
        res.status(500).send({ result: 'Fail!' });
    }
});

module.exports = router;

推荐阅读