upfetch - advanced fetch client builder

upfetch is an advanced fetch client builder with standard schema validation, automatic response parsing, smart defaults and more. Designed to make data fetching type-safe and developer-friendly while keeping the familiar fetch API.

中文文档 (AI 翻译)

Table of Contents

• Highlights
• QuickStart
• Key Features
  • Request Configuration
  • Simple Query Parameters
  • Automatic Body Handling
  • Schema Validation
  • Lifecycle Hooks
  • Timeout
  • Retry
  • Error Handling
• Usage
  • Authentication
  • Delete a default option
  • FormData
  • Multiple fetch clients
  • Streaming
  • Progress
• Advanced Usage
  • Error as value
  • Custom response parsing
  • Custom response errors
  • Custom params serialization
  • Custom body serialization
  • Defaults based on the request
• API Reference
• Feature Comparison
• Environment Support

➡️ Highlights

• 🚀 Lightweight - 1.6kB gzipped, no dependency
• 🔒 Typesafe - Validate API responses with zod, valibot or arktype
• 🛠️ Practical API - Use objects for params and body, get parsed responses automatically
• 🎨 Flexible Config - Set defaults like baseUrl or headers once, use everywhere
• 🎯 Comprehensive - Built-in retries, timeouts, progress tracking, streaming, lifecycle hooks, and more
• 🤝 Familiar - same API as fetch with additional options and sensible defaults

➡️ QuickStart

1npm i up-fetch

Create a new upfetch instance:

1import { up } from 'up-fetch'
2
3export const upfetch = up(fetch)

Make a fetch request with schema validation:

1import { upfetch } from './upfetch'
2import { z } from 'zod'
3
4const user = await upfetch('https://a.b.c/users/1', {
5   schema: z.object({
6      id: z.number(),
7      name: z.string(),
8      avatar: z.string().url(),
9   }),
10})

The response is already parsed and properly typed based on the schema.

upfetch extends the native fetch API, which means all standard fetch options are available.

➡️ Key Features

✔️ Request Configuration

Set defaults for all requests when creating an instance:

1const upfetch = up(fetch, () => ({
2   baseUrl: 'https://a.b.c',
3   timeout: 30000,
4}))

Check out the the API Reference for the full list of options.

✔️ Simple Query Parameters

👎 With raw fetch:

1fetch(
2   `https://api.example.com/todos?search=${search}&skip=${skip}&take=${take}`,
3)

👍 With upfetch:

1upfetch('/todos', {
2   params: { search, skip, take },
3})

Use the serializeParams option to customize the query parameter serialization.

✔️ Automatic Body Handling

👎 With raw fetch:

1fetch('https://api.example.com/todos', {
2   method: 'POST',
3   headers: { 'Content-Type': 'application/json' },
4   body: JSON.stringify({ title: 'New Todo' }),
5})

👍 With upfetch:

1upfetch('/todos', {
2   method: 'POST',
3   body: { title: 'New Todo' },
4})

upfetch also supports all fetch body types.

Check out the serializeBody option to customize the body serialization.

✔️ Schema Validation

Since upfetch follows the Standard Schema Specification it can be used with any schema library that implements the spec.
See the full list here.

👉 With zod 3.24+

1import { z } from 'zod'
2
3const posts = await upfetch('/posts/1', {
4   schema: z.object({
5      id: z.number(),
6      title: z.string(),
7   }),
8})

👉 With valibot 1.0+

1import { object, string, number } from 'valibot'
2
3const posts = await upfetch('/posts/1', {
4   schema: object({
5      id: number(),
6      title: string(),
7   }),
8})

✔️ Lifecycle Hooks

Control request/response lifecycle with simple hooks:

1const upfetch = up(fetch, () => ({
2   onRequest: (options) => {
3      // Called before the request is made, options might be mutated here
4   },
5   onSuccess: (data, options) => {
6      // Called when the request successfully completes
7   },
8   onError: (error, options) => {
9      // Called when the request fails
10   },
11}))

✔️ Timeout

Set a timeout for one request:

1upfetch('/todos', {
2   timeout: 3000,
3})

Set a default timeout for all requests:

1const upfetch = up(fetch, () => ({
2   timeout: 5000,
3}))

✔️ Retry

The retry functionality allows you to automatically retry failed requests with configurable attempts, delay, and condition.

1const upfetch = up(fetch, () => ({
2   retry: {
3      attempts: 3,
4      delay: 1000,
5   },
6}))

Examples:

1await upfetch('/api/data', {
2   method: 'DELETE',
3   retry: {
4      attempts: 2,
5   },
6})

1const upfetch = up(fetch, () => ({
2   retry: {
3      attempts: 3,
4      delay: (ctx) => ctx.attempt ** 2 * 1000,
5   },
6}))

1const upfetch = up(fetch, () => ({
2   retry: {
3      // One retry for GET requests, no retries for other methods:
4      attempts: (ctx) => (ctx.request.method === 'GET' ? 1 : 0),
5      delay: 1000,
6   },
7}))

1const upfetch = up(fetch, () => ({
2   retry: {
3      when: (ctx) =>
4         [408, 413, 429, 500, 502, 503, 504].includes(ctx.response?.status),
5      attempts: 1,
6      delay: 1000,
7   },
8}))

1const upfetch = up(fetch, () => ({
2   retry: {
3      attempts: 2,
4      delay: 1000,
5      when: (ctx) => {
6         // Retry on timeout errors
7         if (ctx.error) return ctx.error.name === 'TimeoutError'
8         // Retry on 429 server errors
9         if (ctx.response) return ctx.response.status === 429
10         return false
11      },
12   },
13}))

✔️ Error Handling

👉 ResponseError

Raised when response.ok is false.
Use isResponseError to identify this error type.

1import { isResponseError } from 'up-fetch'
2
3try {
4   await upfetch('/todos/1')
5} catch (error) {
6   if (isResponseError(error)) {
7      console.log(error.status)
8   }
9}

• Use the parseRejected option to throw a custom error instead.
• Use the reject option to decide when to throw.

👉 ValidationError

Raised when schema validation fails.
Use isValidationError to identify this error type.

1import { isValidationError } from 'up-fetch'
2
3try {
4   await upfetch('/todos/1', { schema: todoSchema })
5} catch (error) {
6   if (isValidationError(error)) {
7      console.log(error.issues)
8   }
9}

➡️ Usage

✔️ Authentication

You can easily add authentication to all requests by setting a default header.

Retrieve the token from localStorage before each request:

1const upfetch = up(fetch, () => ({
2   headers: { Authorization: localStorage.getItem('bearer-token') },
3}))

Retrieve an async token:

1const upfetch = up(fetch, async () => ({
2   headers: { Authorization: await getToken() },
3}))

✔️ Delete a default option

Simply pass undefined:

1upfetch('/todos', {
2   signal: undefined,
3})

Also works for single params and headers:

1upfetch('/todos', {
2   headers: { Authorization: undefined },
3})

✔️ FormData

Grab the FormData from a form.

1const form = document.querySelector('#my-form')
2
3upfetch('/todos', {
4   method: 'POST',
5   body: new FormData(form),
6})

Or create FormData from an object:

1import { serialize } from 'object-to-formdata'
2
3const upfetch = up(fetch, () => ({
4   serializeBody: (body) => serialize(body),
5}))
6
7upfetch('https://a.b.c', {
8   method: 'POST',
9   body: { file: new File(['foo'], 'foo.txt') },
10})

✔️ Multiple fetch clients

You can create multiple upfetch instances with different defaults:

1const fetchMovie = up(fetch, () => ({
2   baseUrl: 'https://api.themoviedb.org',
3   headers: {
4      accept: 'application/json',
5      Authorization: `Bearer ${process.env.API_KEY}`,
6   },
7}))
8
9const fetchFile = up(fetch, () => ({
10   parseResponse: async (res) => {
11      const name = res.url.split('/').at(-1) ?? ''
12      const type = res.headers.get('content-type') ?? ''
13      return new File([await res.blob()], name, { type })
14   },
15}))

✔️ Streaming

upfetch provides powerful streaming capabilities through onRequestStreaming for upload operations, and onResponseStreaming for download operations.

Both handlers receive the following event object plus the request/response:

1type StreamingEvent = {
2   chunk: Uint8Array // The current chunk of data being streamed
3   totalBytes: number // Total size of the data
4   transferredBytes: number // Amount of data transferred so far
5}

The totalBytes property of the event is read from the "Content-Length" header.
For request streaming, if the header is not present, the total bytes are read from the request body.

Here's an example of processing a streamed response from an AI chatbot:

1const decoder = new TextDecoder()
2
3upfetch('/ai-chatbot', {
4   onResponseStreaming: (event, response) => {
5      const text = decoder.decode(event.chunk)
6      console.log(text)
7   },
8})

✔️ Progress

Upload progress:

1upfetch('/upload', {
2   method: 'POST',
3   body: new File(['large file'], 'foo.txt'),
4   onRequestStreaming: ({ transferredBytes, totalBytes }) => {
5      console.log(`Progress: ${transferredBytes} / ${totalBytes}`)
6   },
7})

Download progress:

1upfetch('/download', {
2   onResponseStreaming: ({ transferredBytes, totalBytes }) => {
3      console.log(`Progress: ${transferredBytes} / ${totalBytes}`)
4   },
5})

➡️ Advanced Usage

✔️ Error as value

While the Fetch API does not throw an error when the response is not ok, upfetch throws a ResponseError instead.

If you'd rather handle errors as values, set reject to return false.
This allows you to customize the parseResponse function to return both successful data and error responses in a structured format.

1const upfetch = up(fetch, () => ({
2   reject: () => false,
3   parseResponse: async (response) => {
4      const json = await response.json()
5      return response.ok
6         ? { data: json, error: null }
7         : { data: null, error: json }
8   },
9}))

Usage:

1const { data, error } = await upfetch('/users/1')

✔️ Custom response parsing

By default upfetch is able to parse json and text sucessful responses automatically.

The parseResponse method is called when reject returns false. You can use that option to parse other response types.

1const upfetch = up(fetch, () => ({
2   parseResponse: (response) => response.blob(),
3}))

💡 Note that the parseResponse method is called only when reject returns false.

✔️ Custom response errors

By default upfetch throws a ResponseError when reject returns true.

If you want to throw a custom error instead, you can pass a function to the parseRejected option.

1const upfetch = up(fetch, () => ({
2   parseRejected: async (response) => {
3      const status = response.status
4      const data = await response.json()
5      return new CustomError(status, data)
6   },
7}))

✔️ Custom params serialization

By default upfetch serializes the params using URLSearchParams.

You can customize the params serialization by passing a function to the serializeParams option.

1import queryString from 'query-string'
2
3const upfetch = up(fetch, () => ({
4   serializeParams: (params) => queryString.stringify(params),
5}))

✔️ Custom body serialization

By default upfetch serializes the plain objects using JSON.stringify.

You can customize the body serialization by passing a function to the serializeBody option. It lets you:

• restrict the valid body type by typing its first argument
• transform the body in a valid BodyInit type

The following example show how to restrict the valid body type to Record<string, any> and serialize it using JSON.stringify:

1// Restrict the body type to Record<string, any> and serialize it
2const upfetch = up(fetch, () => ({
3   serializeBody: (body: Record<string, any>) => JSON.stringify(body),
4}))
5
6// ❌ type error: the body is not a Record<string, any>
7upfetch('https://a.b.c/todos', {
8   method: 'POST',
9   body: [['title', 'New Todo']],
10})
11
12// ✅ works fine with Record<string, any>
13upfetch('https://a.b.c/todos', {
14   method: 'POST',
15   body: { title: 'New Todo' },
16})

The following example uses superjson to serialize the body. The valid body type is inferred from SuperJSON.stringify.

1import SuperJSON from 'superjson'
2
3const upfetch = up(fetch, () => ({
4   serializeBody: SuperJSON.stringify,
5}))

✔️ Defaults based on the request

The default options receive the fetcher arguments, this allows you to tailor the defaults based on the actual request.

1const upfetch = up(fetch, (input, options) => ({
2   baseUrl: 'https://example.com/',
3   // Add authentication only for protected routes
4   headers: {
5      Authorization:
6         typeof input === 'string' && input.startsWith('/api/protected/')
7            ? `Bearer ${getToken()}`
8            : undefined,
9   },
10   // Add tracking params only for public endpoints
11   params: {
12      trackingId:
13         typeof input === 'string' && input.startsWith('/public/')
14            ? crypto.randomUUID()
15            : undefined,
16   },
17   // Increase timeout for long-running operations
18   timeout:
19      typeof input === 'string' && input.startsWith('/export/') ? 30000 : 5000,
20}))

➡️ API Reference

up(fetch, getDefaultOptions?)

Creates a new upfetch instance with optional default options.

1function up(
2   fetchFn: typeof globalThis.fetch,
3   getDefaultOptions?: (
4      input: RequestInit,
5      options: FetcherOptions,
6   ) => DefaultOptions | Promise<DefaultOptions>,
7): UpFetch

upfetch(url, options?)

Makes a fetch request with the given options.

1function upfetch(
2   url: string | URL | Request,
3   options?: FetcherOptions,
4): Promise<any>

Options:

RetryOptions

isResponseError(error)

Checks if the error is a ResponseError.

isValidationError(error)

Checks if the error is a ValidationError.

isJsonifiable(value)

Determines whether a value can be safely converted to json.

Are considered jsonifiable:

• plain objects
• arrays
• class instances with a toJSON method

➡️ Feature Comparison

Check out the Feature Comparison table to see how upfetch compares to other fetching libraries.

➡️ Environment Support

• ✅ Browsers (Chrome, Firefox, Safari, Edge)
• ✅ Node.js (18.0+)
• ✅ Bun
• ✅ Deno
• ✅ Cloudflare Workers
• ✅ Vercel Edge Runtime