@fastify/multipart

Fastify plugin to parse the multipart content-type. Supports:

• Async / Await
• Async iterator support to handle multiple parts
• Stream & Disk mode
• Accumulating the entire file in memory
• Mode to attach all fields to the request body
• Tested across Linux/Mac/Windows

Under the hood, it uses @fastify/busboy.

Install

1npm i @fastify/multipart

Usage

1const fastify = require('fastify')()
2const fs = require('node:fs')
3const { pipeline } = require('node:stream/promises')
4
5fastify.register(require('@fastify/multipart'))
6
7fastify.post('/', async function (req, reply) {
8  // process a single file
9  // also, consider that if you allow to upload multiple files
10  // you must consume all files otherwise the promise will never fulfill
11  const data = await req.file()
12
13  data.file // stream
14  data.fields // other parsed parts
15  data.fieldname
16  data.filename
17  data.encoding
18  data.mimetype
19
20  // to accumulate the file in memory! Be careful!
21  //
22  // await data.toBuffer() // Buffer
23  //
24  // or
25
26  await pipeline(data.file, fs.createWriteStream(data.filename))
27
28  // be careful of permission issues on disk and not overwrite
29  // sensitive files that could cause security risks
30
31  // also, consider that if the file stream is not consumed, the promise will never fulfill
32
33  reply.send()
34})
35
36fastify.listen({ port: 3000 }, err => {
37  if (err) throw err
38  console.log(`server listening on ${fastify.server.address().port}`)
39})

Note about data.fields: busboy consumes the multipart in serial order (stream). Therefore, the order of form fields is VERY IMPORTANT to how @fastify/multipart can display the fields to you. We would recommend you place the value fields first before any of the file fields. It will ensure your fields are accessible before it starts consuming any files. If you cannot control the order of the placed fields, be sure to read data.fields AFTER consuming the stream, or it will only contain the fields parsed at that moment.

You can also pass optional arguments to @fastify/busboy when registering with Fastify. This is useful for setting limits on the content that can be uploaded. A full list of available options can be found in the @fastify/busboy documentation.

1fastify.register(require('@fastify/multipart'), {
2  limits: {
3    fieldNameSize: 100, // Max field name size in bytes
4    fieldSize: 100,     // Max field value size in bytes
5    fields: 10,         // Max number of non-file fields
6    fileSize: 1000000,  // For multipart forms, the max file size in bytes
7    files: 1,           // Max number of file fields
8    headerPairs: 2000,  // Max number of header key=>value pairs
9    parts: 1000         // For multipart forms, the max number of parts (fields + files)
10  }
11});

For security reasons, @fastify/multipart sets the limit for parts and fileSize being 1000 and 1048576 respectively.

Note: if the file stream that is provided by data.file is not consumed, like in the example below with the usage of pipeline, the promise will not be fulfilled at the end of the multipart processing. This behavior is inherited from @fastify/busboy.

Note: if you set a fileSize limit and you want to know if the file limit was reached you can:

• listen to data.file.on('limit')
• or check at the end of the stream the property data.file.truncated
• or call data.file.toBuffer() and wait for the error to be thrown

1const data = await req.file()
2await pipeline(data.file, fs.createWriteStream(data.filename))
3if (data.file.truncated) {
4  // you may need to delete the part of the file that has been saved on disk
5  // before the `limits.fileSize` has been reached
6  reply.send(new fastify.multipartErrors.FilesLimitError());
7}
8
9// OR
10const data = await req.file()
11try {
12  const buffer = await data.toBuffer()
13} catch (err) {
14  // fileSize limit reached!
15}
16

Additionally, you can pass per-request options to the req.file, req.files, req.saveRequestFiles or req.parts function.

1fastify.post('/', async function (req, reply) {
2  const options = { limits: { fileSize: 1000 } };
3  const data = await req.file(options)
4  await pipeline(data.file, fs.createWriteStream(data.filename))
5  reply.send()
6})

Handle multiple file streams

1fastify.post('/', async function (req, reply) {
2  const parts = req.files()
3  for await (const part of parts) {
4    await pipeline(part.file, fs.createWriteStream(part.filename))
5  }
6  reply.send()
7})

Handle multiple file streams and fields

1fastify.post('/upload/raw/any', async function (req, reply) {
2  const parts = req.parts()
3  for await (const part of parts) {
4    if (part.type === 'file') {
5      await pipeline(part.file, fs.createWriteStream(part.filename))
6    } else {
7      // part.type === 'field
8      console.log(part)
9    }
10  }
11  reply.send()
12})

Accumulating the entire file in memory

1fastify.post('/upload/raw/any', async function (req, reply) {
2  const data = await req.file()
3  const buffer = await data.toBuffer()
4  // upload to S3
5  reply.send()
6})

Upload files to disk and work with temporary file paths

This will store all files in the operating system's default directory for temporary files. As soon as the response ends all files are removed.

1fastify.post('/upload/files', async function (req, reply) {
2  // stores files to tmp dir and return files
3  const files = await req.saveRequestFiles()
4  files[0].type // "file"
5  files[0].filepath
6  files[0].fieldname
7  files[0].filename
8  files[0].encoding
9  files[0].mimetype
10  files[0].fields // other parsed parts
11
12  reply.send()
13})

Handle file size limitation

If you set a fileSize limit, it can throw a RequestFileTooLargeError error when limit reached.

1fastify.post('/upload/files', async function (req, reply) {
2  try {
3    const file = await req.file({ limits: { fileSize: 17000 } })
4    //const files = req.files({ limits: { fileSize: 17000 } })
5    //const parts = req.parts({ limits: { fileSize: 17000 } })
6    //const files = await req.saveRequestFiles({ limits: { fileSize: 17000 } })
7    reply.send()
8  } catch (error) {
9    // error instanceof fastify.multipartErrors.RequestFileTooLargeError
10  }
11})

If you want to fallback to the handling before 4.0.0, you can disable the throwing behavior by passing throwFileSizeLimit. Note: It will not affect the behavior of saveRequestFiles()

1// globally disable
2fastify.register(fastifyMultipart, { throwFileSizeLimit: false })
3
4fastify.post('/upload/file', async function (req, reply) {
5  const file = await req.file({ throwFileSizeLimit: false, limits: { fileSize: 17000 } })
6  //const files = req.files({ throwFileSizeLimit: false, limits: { fileSize: 17000 } })
7  //const parts = req.parts({ throwFileSizeLimit: false, limits: { fileSize: 17000 } })
8  //const files = await req.saveRequestFiles({ throwFileSizeLimit: false, limits: { fileSize: 17000 } })
9  reply.send()
10})

Parse all fields and assign them to the body

This allows you to parse all fields automatically and assign them to the request.body. By default, files are accumulated in memory (Be careful!) to buffer objects. Uncaught errors are handled by Fastify.

1fastify.register(require('@fastify/multipart'), { attachFieldsToBody: true })
2
3fastify.post('/upload/files', async function (req, reply) {
4  const uploadValue = await req.body.upload.toBuffer() // access files
5  const fooValue = req.body.foo.value                  // other fields
6  const body = Object.fromEntries(
7    Object.keys(req.body).map((key) => [key, req.body[key].value])
8  ) // Request body in key-value pairs, like req.body in Express (Node 12+)
9
10  // On Node 18+
11  const formData = await req.formData()
12  console.log(formData)
13})

Request body key-value pairs can be assigned directly using attachFieldsToBody: 'keyValues'. Field values, including file buffers, will be attached to the body object.

1fastify.register(require('@fastify/multipart'), { attachFieldsToBody: 'keyValues' })
2
3fastify.post('/upload/files', async function (req, reply) {
4  const uploadValue = req.body.upload // access file as buffer
5  const fooValue = req.body.foo       // other fields
6})

You can also define an onFile handler to avoid accumulating all files in memory.

1async function onFile(part) {
2  // you have access to original request via `this`
3  console.log(this.id)
4  await pipeline(part.file, fs.createWriteStream(part.filename))
5}
6
7fastify.register(require('@fastify/multipart'), { attachFieldsToBody: true, onFile })
8
9fastify.post('/upload/files', async function (req, reply) {
10  const fooValue = req.body.foo.value // other fields
11})

The onFile handler can also be used with attachFieldsToBody: 'keyValues' in order to specify how file buffer values are decoded.

1async function onFile(part) {
2  const buff = await part.toBuffer()
3  const decoded = Buffer.from(buff.toString(), 'base64').toString()
4  part.value = decoded // set `part.value` to specify the request body value
5}
6
7fastify.register(require('@fastify/multipart'), { attachFieldsToBody: 'keyValues', onFile })
8
9fastify.post('/upload/files', async function (req, reply) {
10  const uploadValue = req.body.upload // access file as base64 string
11  const fooValue = req.body.foo       // other fields
12})

Note: if you assign all fields to the body and don't define an onFile handler, you won't be able to read the files through streams, as they are already read and their contents are accumulated in memory. You can only use the toBuffer method to read the content. If you try to read from a stream and pipe to a new file, you will obtain an empty new file.

JSON Schema body validation

When the attachFieldsToBody parameter is set to 'keyValues', JSON Schema validation on the body will behave similarly to application/json and application/x-www-form-urlencoded content types. Additionally, uploaded files will be attached to the body as Buffer objects.

1fastify.register(require('@fastify/multipart'), { attachFieldsToBody: 'keyValues' })
2
3fastify.post('/upload/files', {
4  schema: {
5    consumes: ['multipart/form-data'],
6    body: {
7      type: 'object',
8      required: ['myFile'],
9      properties: {
10        // file that gets decoded to string
11        myFile: {
12          type: 'object',
13        },
14        hello: {
15          type: 'string',
16          enum: ['world']
17        }
18      }
19    }
20  }
21}, function (req, reply) {
22  console.log({ body: req.body })
23  reply.send('done')
24})

If you enable attachFieldsToBody: true and set sharedSchemaId a shared JSON Schema is added, which can be used to validate parsed multipart fields.

1const opts = {
2  attachFieldsToBody: true,
3  sharedSchemaId: '#mySharedSchema'
4}
5fastify.register(require('@fastify/multipart'), opts)
6
7fastify.post('/upload/files', {
8  schema: {
9    consumes: ['multipart/form-data'],
10    body: {
11      type: 'object',
12      required: ['myField'],
13      properties: {
14        // field that uses the shared schema
15        myField: { $ref: '#mySharedSchema'},
16        // or another field that uses the shared schema
17        myFiles: { type: 'array', items: fastify.getSchema('mySharedSchema') },
18        // or a field that doesn't use the shared schema
19        hello: {
20          properties: {
21            value: {
22              type: 'string',
23              enum: ['male']
24            }
25          }
26        }
27      }
28    }
29  }
30}, function (req, reply) {
31  console.log({ body: req.body })
32  reply.send('done')
33})

If provided, the sharedSchemaId parameter must be a string ID and a shared schema will be added to your fastify instance so you will be able to apply the validation to your service (like in the example mentioned above).

The shared schema, that is added, will look like this:

1{
2  type: 'object',
3  properties: {
4    encoding: { type: 'string' },
5    filename: { type: 'string' },
6    limit: { type: 'boolean' },
7    mimetype: { type: 'string' }
8  }
9}

JSON Schema with Swagger

If you want to use @fastify/multipart with @fastify/swagger and @fastify/swagger-ui you must add a new type called isFile and use a custom instance of a validator compiler Docs.

1
2const fastify = require('fastify')({
3 // ...
4  ajv: {
5    // Adds the file plugin to help @fastify/swagger schema generation
6    plugins: [require('@fastify/multipart').ajvFilePlugin]
7  }
8})
9
10fastify.register(require("@fastify/multipart"), {
11  attachFieldsToBody: true,
12});
13
14fastify.post(
15  "/upload/files",
16  {
17    schema: {
18      consumes: ["multipart/form-data"],
19      body: {
20        type: "object",
21        required: ["myField"],
22        properties: {
23          myField: { isFile: true },
24        },
25      },
26    },
27  },
28  function (req, reply) {
29    console.log({ body: req.body });
30    reply.send("done");
31  }
32);
33

JSON Schema non-file field

When sending fields with the body (attachFieldsToBody set to true), the field might look like this in the request.body:

1{
2  "hello": "world"
3}

The mentioned field will be converted, by this plugin, to a more complex field. The converted field will look something like this:

1{
2  hello: {
3    fieldname: "hello",
4    value: "world",
5    fieldnameTruncated: false,
6    valueTruncated: false,
7    fields: body
8  }
9}

It is important to know that this conversion happens BEFORE the field is validated, so keep that in mind when writing the JSON schema for validation for fields that don't use the shared schema. The schema for validation for the field mentioned above should look like this:

1hello: {
2  properties: {
3    value: {
4      type: 'string'
5    }
6  }
7}

JSON non-file fields

If a non-file field sent has Content-Type header starting with application/json, it will be parsed using JSON.parse.

The schema to validate JSON fields should look like this:

1hello: {
2  properties: {
3    value: {
4      type: 'object',
5      properties: {
6        /* ... */
7      }
8    }
9  }
10}

If you also use the shared JSON schema as shown above, this is a full example that validates the entire field:

1const opts = {
2  attachFieldsToBody: true,
3  sharedSchemaId: '#mySharedSchema'
4}
5fastify.register(require('@fastify/multipart'), opts)
6
7fastify.post('/upload/files', {
8  schema: {
9    consumes: ['multipart/form-data'],
10    body: {
11      type: 'object',
12      required: ['field'],
13      properties: {
14        field: {
15          allOf: [
16            { $ref: '#mySharedSchema' },
17            {
18              properties: {
19                value: {
20                  type: 'object'
21                  properties: {
22                    child: {
23                      type: 'string'
24                    }
25                  }
26                }
27              }
28            }
29          ]
30        }
31      }
32    }
33  }
34}, function (req, reply) {
35  console.log({ body: req.body })
36  reply.send('done')
37})

Access all errors

We export all custom errors via a server decorator fastify.multipartErrors. This is useful if you want to react to specific errors. They are derived from @fastify/error and include the correct statusCode property.

1fastify.post('/upload/files', async function (req, reply) {
2  const { FilesLimitError } = fastify.multipartErrors
3})

Acknowledgments

This project is kindly sponsored by:

• nearForm
• LetzDoIt
• platformatic

License

Licensed under MIT.