@tus/server

👉 Note: since 1.0.0 packages are split and published under the @tus scope. The old package, tus-node-server, is considered unstable and will only receive security fixes. Make sure to use the new package.

Contents

• Install
• Use
• API
  • new Server(options)
  • EVENTS
• Examples
  • Example: integrate tus into Express
  • Example: integrate tus into Koa
  • Example: integrate tus into Fastify
  • Example: integrate tus into Next.js
  • Example: validate metadata when an upload is created
  • Example: store files in custom nested directories
  • Example: use with Nginx
• Types
• Compatibility
• Contribute
• License

Install

In Node.js (16.0+), install with npm:

npm install @tus/server

Use

A standalone server which stores files on disk.

const {Server} = require('@tus/server')
const {FileStore} = require('@tus/file-store')
const host = '127.0.0.1'
const port = 1080

const server = new Server({
  path: '/files',
  datastore: new FileStore({directory: './files'}),
})
server.listen({host, port})

API

This package exports Server and all constants, types, models, and kvstores. There is no default export. You should only need the Server, EVENTS, and KV store exports.

new Server(options)

Creates a new tus server with options.

options.path

The route to accept requests (string).

options.maxSize

Max file size (in bytes) allowed when uploading (number | ((req, id: string | null) => Promise<number> | number)). When providing a function during the OPTIONS request the id will be null.

options.allowedCredentials

Sets Access-Control-Allow-Credentials (boolean, default: false).

options.allowedOrigins

Trusted origins (string[]).

Sends the client's origin back in Access-Control-Allow-Origin if it matches.

options.postReceiveInterval

Interval in milliseconds for sending progress of an upload over POST_RECEIVE_V2 (number).

options.relativeLocation

Return a relative URL as the Location header to the client (boolean).

options.respectForwardedHeaders

Allow Forwarded, X-Forwarded-Proto, and X-Forwarded-Host headers to override the Location header returned by the server (boolean).

options.allowedHeaders

Additional headers sent in Access-Control-Allow-Headers (string[]).

options.generateUrl

Control how the upload URL is generated ((req, { proto, host, path, id }) => string))

This only changes the upload URL (Location header). If you also want to change the file name in storage use namingFunction. Returning prefix-1234 in namingFunction means the id argument in generateUrl is prefix-1234.

@tus/server expects everything in the path after the last / to be the upload id. If you change that you have to use getFileIdFromRequest as well.

A common use case of this function and getFileIdFromRequest is to base65 encode a complex id into the URL.

Checkout the example how to store files in custom nested directories.

options.getFileIdFromRequest

Control how the Upload-ID is extracted from the request ((req, lastPath) => string | void)

By default, it expects everything in the path after the last / to be the upload id. lastPath is everything after the last /.

Checkout the example how to store files in custom nested directories.

options.namingFunction

Control how you want to name files ((req, metadata) => string | Promise<string>)

In @tus/server, the upload ID in the URL is the same as the file name. This means using a custom namingFunction will return a different Location header for uploading and result in a different file name in storage.

It is important to make these unique to prevent data loss. Only use it if you need to. Default uses crypto.randomBytes(16).toString('hex').

Checkout the example how to store files in custom nested directories.

options.locker

The locker interface to manage locks for exclusive access control over resources (Locker).

By default it uses an in-memory locker (MemoryLocker) for safe concurrent access to uploads using a single server. When running multiple instances of the server, you need to provide a locker implementation that is shared between all instances (such as a RedisLocker).

options.disableTerminationForFinishedUploads

Disallow the termination extension for finished uploads. (boolean)

options.onUploadCreate

onUploadCreate will be invoked before a new upload is created. ((req, res, upload) => Promise<{ res: http.ServerResponse, metadata?: Record<string, string>}>).

• If the function returns the (modified) response the upload will be created.
• You can optionally return metadata which will override (not merge!) upload.metadata.
• You can throw an Object and the HTTP request will be aborted with the provided body and status_code (or their fallbacks).

This can be used to implement validation of upload metadata or add headers.

options.onUploadFinish

onUploadFinish will be invoked after an upload is completed but before a response is returned to the client ((req, res, upload) => Promise<{ res: http.ServerResponse, status_code?: number, headers?: Record<string, string | number>, body?: string }>).

• You can optionally return status_code, headers and body to modify the response. Note that the tus specification does not allow sending response body nor status code other than 204, but most clients support it. Use at your own risk.
• You can throw an Object and the HTTP request will be aborted with the provided body and status_code (or their fallbacks).

This can be used to implement post-processing validation.

options.onIncomingRequest

onIncomingRequest is a middleware function invoked before all handlers ((req, res) => Promise<void>)

This can be used for things like access control. You can throw an Object and the HTTP request will be aborted with the provided body and status_code (or their fallbacks).

options.onResponseError

onResponseError will be invoked when an error response is about to be sent by the server. you use this function to map custom errors to tus errors or for custom observability. ((req, res, err) => Promise<{status_code: number; body: string} | void> | {status_code: number; body: string} | void)

server.handle(req, res)

The main server request handler invoked on every request. You only need to use this when you integrate tus into an existing Node.js server.

server.get(req, res)

You can implement your own GET handlers. For instance, to return all files.

const fs = require('node:fs/promises')
const {Server} require('@tus/server')
const {FileStore} require('@tus/file-store')

const server = new Server({
  path: '/files',
  datastore: new FileStore({ directory: './files' }),
})

server.get('/uploads', async (req, res) => {
  const files = await fs.readdir(server.datastore.directory)
  // Format and return
})

server.listen()

Start the tus server. Supported arguments are the same as server.listen() from node:net.

server.cleanUpExpiredUploads()

Clean up expired uploads. Your chosen datastore must support the expiration extension for this to work.

EVENTS

Events to subscribe to (Object<string>).

You can listen for events by using the .on() method on the Server instance.

POST_CREATE

Called after an upload has been created but before it's written to a store.

const {EVENTS} = require('@tus/server')
// ...
server.on(EVENTS.POST_CREATE, (req, res, upload => {})

POST_RECEIVE

Deprecated.

Called every time an upload finished writing to the store. This event is emitted whenever the request handling is completed (which is the same as onUploadFinish, almost the same as POST_FINISH), whereas the POST_RECEIVE_V2 event is emitted while the request is being handled.

const {EVENTS} = require('@tus/server')
// ...
server.on(EVENTS.POST_RECEIVE, (req, res, upload => {})

POST_RECEIVE_V2

Called every postReceiveInterval milliseconds for every upload while it‘s being written to the store.

This means you are not guaranteed to get (all) events for an upload. For instance if postReceiveInterval is set to 1000ms and an PATCH request takes 500ms, no event is emitted. If the PATCH request takes 2500ms, you would get the offset at 2000ms, but not at 2500ms.

Use POST_FINISH if you need to know when an upload is done.

const {EVENTS} = require('@tus/server')
// ...
server.on(EVENTS.POST_RECEIVE_V2, (req, upload => {})

POST_FINISH

Called an upload has completed and after a response has been sent to the client.

const {EVENTS} = require('@tus/server')
// ...
server.on(EVENTS.POST_FINISH, (req, res, upload => {})

POST_TERMINATE

Called after an upload has been terminated and a response has been sent to the client.

const {EVENTS} = require('@tus/server')
// ...
server.on(EVENTS.POST_TERMINATE, (req, res, id => {})

Key-Value Stores

All stores (as in the datastore option) save two files, the uploaded file and an info file with metadata, usually adjacent to each other.

In @tus/file-store the FileKvStore is used to persist upload info but the KV stores can also be used as a cache in other stores, such as @tus/s3-store.

MemoryKvStore

import {MemoryKvStore} from '@tus/server'
import S3Store, {type MetadataValue} from '@tus/s3-store'

new S3Store({
  // ...
  cache: new MemoryKvStore<MetadataValue>(),
})

FileKvStore

import {FileKvStore} from '@tus/server'
import S3Store, {type MetadataValue} from '@tus/s3-store'

const path = './uploads'

new S3Store({
  // ...
  cache: new FileKvStore<MetadataValue>(path),
})

RedisKvStore

import {RedisKvStore} from '@tus/server'
import S3Store, {type MetadataValue} from '@tus/s3-store'
import {createClient} from '@redis/client'

const client = await createClient().connect()
const path = './uploads'
const prefix = 'foo' // prefix for the key (foo${id})

new S3Store({
  // ...
  cache: new RedisKvStore<MetadataValue>(client, prefix),
})

Examples

Example: integrate tus into Express

const {Server} = require('@tus/server')
const {FileStore} = require('@tus/file-store')
const express = require('express')

const host = '127.0.0.1'
const port = 1080
const app = express()
const uploadApp = express()
const server = new Server({
  path: '/uploads',
  datastore: new FileStore({directory: '/files'}),
})

uploadApp.all('*', server.handle.bind(server))
app.use('/uploads', uploadApp)
app.listen(port, host)

Example: integrate tus into Koa

const http = require('node:http')
const url = require('node:url')
const Koa = require('koa')
const {Server} = require('@tus/server')
const {FileStore} = require('@tus/file-store')

const app = new Koa()
const appCallback = app.callback()
const port = 1080
const tusServer = new Server({
  path: '/files',
  datastore: new FileStore({directory: '/files'}),
})

const server = http.createServer((req, res) => {
  const urlPath = url.parse(req.url).pathname

  // handle any requests with the `/files/*` pattern
  if (/^\/files\/.+/.test(urlPath.toLowerCase())) {
    return tusServer.handle(req, res)
  }

  appCallback(req, res)
})

server.listen(port)

Example: integrate tus into Fastify

const fastify = require('fastify')({logger: true})
const {Server} = require('@tus/server')
const {FileStore} = require('@tus/file-store')

const tusServer = new Server({
  path: '/files',
  datastore: new FileStore({directory: './files'}),
})

/**
 * add new content-type to fastify forewards request
 * without any parser to leave body untouched
 * @see https://www.fastify.io/docs/latest/Reference/ContentTypeParser/
 */
fastify.addContentTypeParser(
  'application/offset+octet-stream',
  (request, payload, done) => done(null)
)

/**
 * let tus handle preparation and filehandling requests
 * fastify exposes raw nodejs http req/res via .raw property
 * @see https://www.fastify.io/docs/latest/Reference/Request/
 * @see https://www.fastify.io/docs/latest/Reference/Reply/#raw
 */
fastify.all('/files', (req, res) => {
  tusServer.handle(req.raw, res.raw)
})
fastify.all('/files/*', (req, res) => {
  tusServer.handle(req.raw, res.raw)
})
fastify.listen(3000, (err) => {
  if (err) {
    fastify.log.error(err)
    process.exit(1)
  }
})

Example: integrate tus into Next.js

Attach the tus server handler to a Next.js route handler in an optional catch-all route file

/pages/api/upload/[[...file]].ts

import type {NextApiRequest, NextApiResponse} from 'next'
import {Server, Upload} from '@tus/server'
import {FileStore} from '@tus/file-store'

/**
 * !Important. This will tell Next.js NOT Parse the body as tus requires
 * @see https://nextjs.org/docs/api-routes/request-helpers
 */
export const config = {
  api: {
    bodyParser: false,
  },
}

const tusServer = new Server({
  // `path` needs to match the route declared by the next file router
  // ie /api/upload
  path: '/api/upload',
  datastore: new FileStore({directory: './files'}),
})

export default function handler(req: NextApiRequest, res: NextApiResponse) {
  return tusServer.handle(req, res)
}

Example: validate metadata when an upload is created

const {Server} = require('@tus/server')
// ...

const server = new Server({
  // ..
  async onUploadCreate(req, res, upload) {
    const {ok, expected, received} = validateMetadata(upload) // your logic
    if (!ok) {
      const body = `Expected "${expected}" in "Upload-Metadata" but received "${received}"`
      throw {status_code: 500, body} // if undefined, falls back to 500 with "Internal server error".
    }
    // You can optionally return metadata to override the upload metadata,
    // such as `{ storagePath: "/upload/123abc..." }`
    const extraMeta = getExtraMetadata(req) // your logic
    return {res, metadata: {...upload.metadata, ...extraMeta}}
  },
})

Example: access control

Access control is opinionated and can be done in different ways. This example is psuedo-code for what it could look like with JSON Web Tokens.

const {Server} = require('@tus/server')
// ...

const server = new Server({
  // ..
  async onIncomingRequest(req, res) {
    const token = req.headers.authorization

    if (!token) {
      throw {status_code: 401, body: 'Unauthorized'}
    }

    try {
      const decodedToken = await jwt.verify(token, 'your_secret_key')
      req.user = decodedToken
    } catch (error) {
      throw {status_code: 401, body: 'Invalid token'}
    }

    if (req.user.role !== 'admin') {
      throw {status_code: 403, body: 'Access denied'}
    }
  },
})

Example: store files in custom nested directories

You can use namingFunction to change the name of the stored file. If you’re only adding a prefix or suffix without a slash (/), you don’t need to implement generateUrl and getFileIdFromRequest.

Adding a slash means you create a new directory, for which you need to implement all three functions as we need encode the id with base64 into the URL.

const path = '/files'
const server = new Server({
  path,
  datastore: new FileStore({directory: './test/output'}),
  namingFunction(req) {
    const id = crypto.randomBytes(16).toString('hex')
    const folder = getFolderForUser(req) // your custom logic
    return `users/${folder}/${id}`
  },
  generateUrl(req, {proto, host, path, id}) {
    id = Buffer.from(id, 'utf-8').toString('base64url')
    return `${proto}://${host}${path}/${id}`
  },
  getFileIdFromRequest(req, lastPath) {
    // lastPath is everything after the last `/`
    // If your custom URL is different, this might be undefined
    // and you need to extract the ID yourself
    return Buffer.from(lastPath, 'base64url').toString('utf-8')
  },
})

Example: use with Nginx

In some cases, it is necessary to run behind a reverse proxy (Nginx, HAProxy etc), for example for TLS termination or serving multiple services on the same hostname. To properly do this, @tus/server and the proxy must be configured appropriately.

Firstly, you must set respectForwardedHeaders indicating that a reverse proxy is in use and that it should respect the X-Forwarded-*/Forwarded headers:

const {Server} = require('@tus/server')
// ...

const server = new Server({
  // ..
  respectForwardedHeaders: true,
})

Secondly, some of the reverse proxy's settings should be adjusted. The exact steps depend on the used proxy, but the following points should be checked:

• Disable request buffering. Nginx, for example, reads the entire incoming HTTP request, including its body, before sending it to the backend, by default. This behavior defeats the purpose of resumability where an upload is processed and saved while it's being transferred, allowing it be resumed. Therefore, such a feature must be disabled.

• Adjust maximum request size. Some proxies have default values for how big a request may be in order to protect your services. Be sure to check these settings to match the requirements of your application.

• Forward hostname and scheme. If the proxy rewrites the request URL, the tusd server does not know the original URL which was used to reach the proxy. This behavior can lead to situations, where tusd returns a redirect to a URL which can not be reached by the client. To avoid this issue, you can explicitly tell tusd which hostname and scheme to use by supplying the X-Forwarded-Host and X-Forwarded-Proto headers. Configure the proxy to set these headers to the original hostname and protocol when forwarding requests to tusd.

You can also take a look at the Nginx configuration from tusd which is used to power the tusd.tusdemo.net instance.

Types

This package is fully typed with TypeScript.

Compatibility

This package requires Node.js 16.0+.

Contribute

See contributing.md.

License

MIT © tus