Streaming multer storage engine for AWS S3.

This project is mostly an integration piece for existing code samples from Multer's storage engine documentation with a call to S3 as the substitution piece for file system. Existing solutions I found required buffering the multipart uploads into the actual filesystem which is difficult to scale.

3.x.x releases of multer-s3 use AWS JavaScript SDK v3. Specifically, it uses the Upload class from @aws-sdk/lib-storage which in turn calls the modular S3Client.

2.x.x releases for multer-s3 use AWS JavaScript SDK v2 via a call to s3.upload.

npm install --save multer-s3

const { S3Client } = require('@aws-sdk/client-s3')
const express = require('express')
const multer = require('multer')
const multerS3 = require('multer-s3')

const app = express()

const s3 = new S3Client()

const upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    metadata: function (req, file, cb) {
      cb(null, {fieldName: file.fieldname});
    },
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

app.post('/upload', upload.array('photos', 3), function(req, res, next) {
  res.send('Successfully uploaded ' + req.files.length + ' files!')
})

Each file contains the following information exposed by multer-s3:

ACL values can be set by passing an optional acl parameter into the multerS3 object.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    acl: 'public-read',
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

Available options for canned ACL.

The metadata option is a callback that accepts the request and file, and returns a metadata object to be saved to S3.

Here is an example that stores all fields in the request body as metadata, and uses an id param as the key:

var opts = {
    s3: s3,
    bucket: config.originalsBucket,
    metadata: function (req, file, cb) {
      cb(null, Object.assign({}, req.body));
    },
    key: function (req, file, cb) {
      cb(null, req.params.id + ".jpg");
    }
  };

The optional cacheControl option sets the Cache-Control HTTP header that will be sent if you're serving the files directly from S3. You can pass either a string or a function that returns a string.

Here is an example that will tell browsers and CDNs to cache the file for one year:

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    cacheControl: 'max-age=31536000',
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

The optional contentType option can be used to set Content/mime type of the file. By default the content type is set to application/octet-stream. If you want multer-s3 to automatically find the content-type of the file, use the multerS3.AUTO_CONTENT_TYPE constant. Here is an example that will detect the content type of the file being uploaded.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    contentType: multerS3.AUTO_CONTENT_TYPE,
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

You may also use a function as the contentType, which should be of the form function(req, file, cb).

storageClass values can be set by passing an optional storageClass parameter into the multerS3 object.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    acl: 'public-read',
    storageClass: 'REDUCED_REDUNDANCY',
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

The optional contentDisposition option can be used to set the Content-Disposition header for the uploaded file. By default, the contentDisposition isn't forwarded. As an example below, using the value attachment forces the browser to download the uploaded file instead of trying to open it.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    acl: 'public-read',
    contentDisposition: 'attachment',
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

An overview of S3's server-side encryption can be found in the [S3 Docs] (http://docs.aws.amazon.com/AmazonS3/latest/dev/serv-side-encryption.html); be advised that customer-managed keys (SSE-C) is not implemented at this time.

You may use the S3 server-side encryption functionality via the optional serverSideEncryption and sseKmsKeyId parameters. Full documentation of these parameters in relation to the S3 API can be found [here] (http://docs.aws.amazon.com/AWSJavaScriptSDK/latest/AWS/S3.html#upload-property) and [here] (http://docs.aws.amazon.com/AmazonS3/latest/dev/UsingServerSideEncryption.html).

serverSideEncryption has two valid values: 'AES256' and 'aws:kms'. 'AES256' utilizes the S3-managed key system, while 'aws:kms' utilizes the AWS KMS system and accepts the optional sseKmsKeyId parameter to specify the key ID of the key you wish to use. Leaving sseKmsKeyId blank when 'aws:kms' is specified will use the default KMS key. Note: You must instantiate the S3 instance with signatureVersion: 'v4' in order to use KMS-managed keys [[Docs]] (http://docs.aws.amazon.com/AmazonS3/latest/dev/UsingAWSSDK.html#specify-signature-version), and the specified key must be in the same AWS region as the S3 bucket used.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    acl: 'authenticated-read',
    contentDisposition: 'attachment',
    serverSideEncryption: 'AES256',
    key: function(req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

The optional contentEncoding option can be used to set the Content-Encoding header for the uploaded file. By default, the contentEncoding isn't forwarded. As an example below, using the value gzip, a file can be uploaded as a gzip file - and when it is downloaded, the browser will uncompress it automatically.

var upload = multer({
  storage: multerS3({
    s3: s3,
    bucket: 'some-bucket',
    acl: 'public-read',
    contentEncoding: 'gzip',
    key: function (req, file, cb) {
      cb(null, Date.now().toString())
    }
  })
})

You may also use a function as the contentEncoding, which should be of the form function(req, file, cb).

The tests mock all access to S3 and can be run completely offline.

npm test