axios vs react-query vs fetch-mock vs use-http
前端数据获取与状态管理方案对比
axiosreact-queryfetch-mockuse-http类似的npm包:

前端数据获取与状态管理方案对比

axios 是一个基于 Promise 的 HTTP 客户端,用于浏览器和 Node.js 环境中发起网络请求。fetch-mock 并非用于生产环境的数据获取,而是一个专门用于模拟 fetch API 调用的测试工具,帮助开发者在单元测试或集成测试中拦截并模拟网络响应。react-query(现为 TanStack Query)是一个强大的数据同步库,专注于服务端状态管理,提供自动缓存、去重、后台更新和错误重试等能力。use-http 是一个轻量级 React Hook,封装了 fetch,旨在简化组件内的 HTTP 请求逻辑,同时提供加载状态、错误处理等基础功能。这四个包分别服务于不同场景:通用 HTTP 客户端、测试模拟、高级数据同步、轻量级 React 数据请求。

npm下载趋势

3 年

GitHub Stars 排名

统计详情

npm包名称
下载量
Stars
大小
Issues
发布时间
License
axios144,484,005109,3152.22 MB1187 天前MIT
react-query2,115,73050,5212.26 MB1304 年前MIT
fetch-mock1,587,7071,307157 kB91 年前MIT
use-http02,297224 kB923 年前MIT

前端数据获取方案深度对比:axios、fetch-mock、react-query 与 use-http

在现代前端工程中,如何高效、可靠地获取和管理远程数据是核心问题之一。axios、fetch-mock、react-query 和 use-http 这四个 npm 包常被提及,但它们的角色截然不同 —— 有的用于生产请求,有的专攻测试,有的专注状态同步。本文将从真实开发场景出发,深入比较它们的技术实现、适用边界和关键差异。

🧩 核心定位:各司其职,不可混用

首先必须明确:fetch-mock 不是生产库,它仅用于测试。其余三者虽都涉及“发请求”,但抽象层级和目标完全不同。

  • axios:通用 HTTP 客户端(与框架无关)
  • fetch-mock:fetch API 的测试模拟工具
  • react-query:React 专属的服务端状态管理库
  • use-http:轻量级 React Hook 封装 fetch

混淆它们的用途会导致架构混乱。例如,试图用 fetch-mock 替代 axios 发请求,或用 use-http 实现 react-query 的缓存策略,都是错误方向。

📡 基础请求能力:如何发起一个 GET 请求?

axios:跨平台、配置丰富

// axios
import axios from 'axios';

const response = await axios.get('/api/user', {
  headers: { 'Authorization': 'Bearer token' }
});
console.log(response.data);

axios 自动解析 JSON,支持超时、拦截器、取消令牌等,且在浏览器和 Node.js 中行为一致。

use-http:React Hook 风格,状态内置

// use-http
import useFetch from 'use-http';

function UserProfile() {
  const { loading, error, data } = useFetch('/api/user');
  
  if (loading) return <div>加载中...</div>;
  if (error) return <div>出错了</div>;
  return <div>{data.name}</div>;
}

use-http 将请求与 React 组件生命周期绑定,自动管理 loading/error/data 状态,但底层仍基于 fetch。

react-query:声明式数据依赖

// react-query
import { useQuery } from '@tanstack/react-query';

function UserProfile() {
  const { isLoading, isError, data } = useQuery({
    queryKey: ['user'],
    queryFn: () => fetch('/api/user').then(res => res.json())
  });
  
  if (isLoading) return <div>加载中...</div>;
  if (isError) return <div>出错了</div>;
  return <div>{data.name}</div>;
}

注意:react-query 本身不提供 HTTP 客户端,需配合 fetch 或 axios 使用。它的核心价值在于 queryKey 驱动的缓存和状态复用。

fetch-mock:仅用于测试!

// fetch-mock (测试文件中)
import fetchMock from 'fetch-mock';

beforeEach(() => {
  fetchMock.get('/api/user', { name: 'Alice' });
});

afterEach(() => {
  fetchMock.restore();
});

// 测试中调用 use-http 或原生 fetch 会命中 mock

fetch-mock 拦截全局 fetch 调用,返回预设响应。它不能用于组件逻辑,只应在测试套件中使用。

🔄 缓存与状态管理:谁负责记住数据?

这是区分 react-query 与其他方案的关键。

axios 与 use-http:无内置缓存

每次调用都会发起新请求:

// 两次独立请求,即使 URL 相同
const user1 = await axios.get('/api/user');
const user2 = await axios.get('/api/user'); // 再次网络请求

若需缓存,必须自行实现(如用 Map 存储响应),或结合其他库。

react-query:自动缓存 + 智能失效

// 第一次:网络请求
const { data: user1 } = useQuery({ queryKey: ['user'], queryFn: fetchUser });

// 第二次(同一 queryKey):直接返回缓存,无网络请求
const { data: user2 } = useQuery({ queryKey: ['user'], queryFn: fetchUser });

缓存默认 5 分钟有效,可通过 staleTime、cacheTime 精细控制。还支持后台刷新(stale-while-revalidate):当数据“过期”后,先返回缓存,再静默拉取新数据更新 UI。

fetch-mock:无缓存概念

因其仅用于测试,缓存由被测代码决定,fetch-mock 只负责返回固定响应。

🧪 测试支持:如何模拟网络请求?

若使用 axios:需 axios-mock-adapter

import MockAdapter from 'axios-mock-adapter';
import axios from 'axios';

const mock = new MockAdapter(axios);
mock.onGet('/api/user').reply(200, { name: 'Alice' });

若使用 fetch / use-http:用 fetch-mock

fetchMock.get('/api/user', { name: 'Alice' });

若使用 react-query:可直接 mock queryFn

// 在测试中替换 queryFn
const { result } = renderHook(() =>
  useQuery({
    queryKey: ['user'],
    queryFn: () => Promise.resolve({ name: 'Alice' })
  })
);

也可结合 fetch-mock 或 axios-mock-adapter,取决于底层客户端。

⚙️ 高级功能对比

功能axiosfetch-mockreact-queryuse-http
请求/响应拦截器✅❌❌❌
自动 JSON 解析✅❌❌(需手动)✅
取消请求✅❌✅(自动)✅
自动缓存❌❌✅❌
后台刷新❌❌✅❌
乐观更新❌❌✅❌
分页/无限滚动❌❌✅❌
仅用于测试❌✅❌❌
React 状态集成❌❌✅✅

🛠️ 典型使用场景

场景 1:构建企业级 React 应用,含复杂数据交互

  • 推荐:react-query + axios(或原生 fetch)
  • 理由:react-query 处理缓存、同步、重试;axios 提供可靠的底层请求能力。两者职责分离,架构清晰。

场景 2:快速原型或小型内部工具

  • 推荐:use-http
  • 理由:零配置,一行 Hook 获取数据+状态,无需学习复杂概念。

场景 3:非 React 项目(如 Vue、Angular 或 Node.js 脚本)

  • 推荐:axios
  • 理由:框架无关,功能完备,社区支持广泛。

场景 4:编写单元测试,需模拟 API 响应

  • 推荐:若用 fetch → fetch-mock;若用 axios → axios-mock-adapter
  • 警告:切勿在生产代码中引入 fetch-mock。

⚠️ 常见误区

  1. “用 use-http 替代 react-query”:use-http 无缓存,重复请求同一数据会多次触发网络调用,导致性能问题和 UI 闪烁。
  2. “在生产代码中用 fetch-mock”:fetch-mock 会污染全局 fetch,且无实际请求能力,仅用于测试。
  3. “react-query 是 HTTP 客户端”:它不发请求,只管理请求的结果。你仍需 fetch 或 axios 作为 transport layer。

💡 总结建议

  • 需要强大缓存和状态同步? → 选 react-query,搭配你喜欢的 HTTP 客户端。
  • 追求简单,项目小? → 选 use-http,但警惕重复请求问题。
  • 跨平台或非 React 项目? → 选 axios。
  • 写测试? → 根据底层客户端选 fetch-mock(fetch)或 axios-mock-adapter(axios)。

记住:没有“最好”的库,只有“最合适”当前场景的工具。理解每个包的设计意图,才能避免过度工程或功能缺失。

如何选择: axios vs react-query vs fetch-mock vs use-http

  • axios:

    选择 axios 如果你需要一个成熟、功能全面且跨平台(浏览器 + Node.js)的 HTTP 客户端,支持拦截器、请求/响应转换、取消请求、自动 JSON 解析等特性。它适用于需要精细控制网络层或与旧项目兼容的场景,尤其当你不依赖 React 特定生态时。

  • react-query:

    选择 react-query 如果你正在构建复杂的 React 应用,需要高效管理服务端状态(如用户资料、列表数据),并希望获得自动缓存、后台刷新、乐观更新、分页/无限滚动等高级功能。它能显著减少手动管理加载/错误状态的样板代码,并提升应用性能和用户体验。

  • fetch-mock:

    选择 fetch-mock 仅用于测试环境,当你使用原生 fetch 或基于 fetch 的封装(如 use-http)时,需要在 Jest、Vitest 等测试框架中模拟网络请求。它不应出现在生产代码中,且如果你使用 axios,应改用 axios-mock-adapter 等对应工具。

  • use-http:

    选择 use-http 如果你希望在 React 组件中以极简方式发起 fetch 请求,并自动获得 loading、error、data 等状态,且项目对 bundle size 敏感、不需要 react-query 的复杂功能。它适合小型应用或原型开发,但缺乏缓存策略和高级同步能力。

axios的README

[!IMPORTANT] Get Security Updates for Axios v0 →

💎 Platinum sponsors

Thanks.dev

We're passionate about making open source sustainable. Scan your dependency tree to better understand which open source projects need funding.

thanks.dev

💜 Become a sponsor
💜 Become a sponsor 💜 Become a sponsor

🥇 Gold sponsors

Principal Financial Group

Free tools to help with your financial planning needs!

principal.com

SAP

BSAP SE, a global software company, is one of the largest vendors of ERP and other enterprise applications.

opensource.sap.com

Descope

Reduce user friction, prevent account takeover, and get a 360° view of your customer and agentic identities with the Descope External IAM platform.

descope.com

RxDB

RxDB is a NoSQL database for JavaScript that runs directly in your app.

rxdb.info

Poprey

Buy Instagram Likes

poprey.com

Fun 88

Fun88 is a global online gambling and betting brand founded in 2009, offering a wide range of services including sports betting, live casino games, slots, and virtual gaming.

global.fun88.com

JBO Vietnam

JBO Vietnam is a prominent online entertainment brand in Vietnam, offering sports betting, esports, online casino games, and a wide range of other exciting games.

jbo88b.com

JBO Thailand

JBO Thailand is a prominent online entertainment brand in Thailand, offering sports betting, esports, online casino games, and a wide range of other exciting games.

jbo88b.com

ScrapingBee

Scrape websites with a powerful web scraping API without managing proxies, browsers, or anti-bot defenses.

scrapingbee.com

HeroDevs

Security and long-term support for end-of-life open source software.

herodevs.com

Sweepico Promo Code

Social Gaming Platform

Sweepico Promo Code



Axios

Promise based HTTP client for the browser and node.js

Website • Documentation

npm version Build status Gitpod Ready-to-Code install size npm bundle size npm downloads gitter chat code helpers Contributors Agent Friendly

Table of contents

Features

  • Make XMLHttpRequests from the browser.
  • Make http requests from Node.js.
  • Use the Promise API for asynchronous request handling.
  • Intercept requests and responses to add custom logic or transform data.
  • Transform request and response data.
  • Cancel requests with built-in cancellation APIs.
  • Serialize and parse JSON data.
  • Serialize data objects to multipart/form-data or application/x-www-form-urlencoded.
  • Add client-side protection against Cross-Site Request Forgery.

Browser support

ChromeFirefoxSafariOperaEdge
Chrome browser logoFirefox browser logoSafari browser logoOpera browser logoEdge browser logo
Latest ✔Latest ✔Latest ✔Latest ✔Latest ✔

Browser Matrix

Installing

Package manager

Using npm:

$ npm install axios

Using yarn:

$ yarn add axios

Using pnpm:

$ pnpm add axios

Using bun:

$ bun add axios

Using Deno:

$ deno add axios

Once the package is installed, import it with import or require:

import axios, { isCancel, AxiosError } from 'axios';

You can also use the default export, since the named export is just a re-export from the Axios factory:

import axios from 'axios';

console.log(axios.isCancel('something'));

If you use require for importing, only the default export is available:

const axios = require('axios');

console.log(axios.isCancel('something'));

Some bundlers and ES6 linters need this form:

import { default as axios } from 'axios';

In custom or legacy environments, you can import the bundle directly:

const axios = require('axios/dist/browser/axios.cjs'); // browser commonJS bundle (ES2017)
// const axios = require('axios/dist/node/axios.cjs'); // node commonJS bundle (ES2017)

CDN

Using jsDelivr CDN (ES5 UMD browser module):

<script src="https://cdn.jsdelivr.net/npm/[email protected]/dist/axios.min.js"></script>

Using unpkg CDN:

<script src="https://unpkg.com/[email protected]/dist/axios.min.js"></script>

Example

import axios from 'axios';
//const axios = require('axios'); // legacy way

try {
  const response = await axios.get('/user?ID=12345');
  console.log(response);
} catch (error) {
  console.error(error);
}

// Optionally the request above could also be done as
axios
  .get('/user', {
    params: {
      ID: 12345,
    },
    timeout: 5000, // 5 seconds. See "Handling Timeouts" below for matching error handling
  })
  .then(function (response) {
    console.log(response);
  })
  .catch(function (error) {
    console.log(error);
  })
  .finally(function () {
    // always executed
  });

// Want to use async/await? Add the `async` keyword to your outer function/method.
async function getUser() {
  try {
    // Example: GET request with query parameters
    const response = await axios.get('/user', {
      params: {
        ID: 12345,
      },
    });

    // Using the `params` option improves readability and automatically formats query strings

    console.log(response);
  } catch (error) {
    console.error(error);
  }
}

Note: Set a timeout in production. Without one, a stalled request can hang indefinitely. See Handling Timeouts for the matching error handling.

Note: async/await is part of ECMAScript 2017 and is not supported in Internet Explorer and older browsers, so use with caution.

Performing a POST request

const response = await axios.post('/user', {
  firstName: 'Fred',
  lastName: 'Flintstone',
});
console.log(response);

Performing multiple concurrent requests

function getUserAccount() {
  return axios.get('/user/12345');
}

function getUserPermissions() {
  return axios.get('/user/12345/permissions');
}

Promise.all([getUserAccount(), getUserPermissions()]).then(function (results) {
  const acct = results[0];
  const perm = results[1];
});

axios API

Requests can be made by passing the relevant config to axios.

axios(config)
// Send a POST request
axios({
  method: 'post',
  url: '/user/12345',
  data: {
    firstName: 'Fred',
    lastName: 'Flintstone',
  },
});
// GET request for remote image in node.js
import fs from 'fs';
const response = await axios({
  method: 'get',
  url: 'https://bit.ly/2mTM3nY',
  responseType: 'stream',
});
response.data.pipe(fs.createWriteStream('ada_lovelace.jpg'));
axios(url[, config])
// Send a GET request (default method)
axios('/user/12345');

Request method aliases

For convenience, aliases have been provided for all common request methods.

axios.request(config)
axios.get(url[, config])
axios.delete(url[, config])
axios.head(url[, config])
axios.options(url[, config])
axios.post(url[, data[, config]])
axios.put(url[, data[, config]])
axios.patch(url[, data[, config]])
Note

When using the alias methods url, method, and data properties don't need to be specified in config.

Concurrency (deprecated)

Use Promise.all instead of these helpers.

Helper functions for dealing with concurrent requests.

axios.all(iterable) axios.spread(callback)

Creating an instance

You can create a new instance of axios with a custom config.

axios.create([config])
const instance = axios.create({
  baseURL: 'https://some-domain.com/api/',
  timeout: 1000,
  headers: { 'X-Custom-Header': 'foobar' },
});

Instance methods

The following instance methods are available. Axios merges the specified config with the instance config.

axios#request(config)
axios#get(url[, config])
axios#delete(url[, config])
axios#head(url[, config])
axios#options(url[, config])
axios#post(url[, data[, config]])
axios#put(url[, data[, config]])
axios#patch(url[, data[, config]])
axios#getUri([config])

Request config

Security notice: decompression-bomb protection is opt-in

By default maxContentLength and maxBodyLength are -1 (unlimited). A malicious or compromised server can return a tiny gzip/deflate/brotli/zstd body that expands to gigabytes and exhaust the Node.js process.

If you call servers you do not fully trust, set a cap:

axios.defaults.maxContentLength = 10 * 1024 * 1024; // 10 MB
axios.defaults.maxBodyLength = 10 * 1024 * 1024;

See the security guide for details.

These config options are available for requests. Only url is required. Requests default to GET when method is not set.

{
  // `url` is the server URL for the request
  url: '/user',

  // `method` is the request method to be used when making the request
  method: 'get', // default

  // Axios prepends `baseURL` to `url` unless `url` is absolute and `allowAbsoluteUrls` is set to true.
  // It can be convenient to set `baseURL` for an instance of axios to pass relative URLs
  // to the methods of that instance.
  // `baseURL` is not a security boundary. If `url` is attacker-controlled, validate it
  // before passing it to axios. Relative URLs can contain `..` segments that resolve
  // outside an intended path prefix after the final URL is parsed.
  // HTTP(S) URLs must include `//` after the protocol. Values such as
  // `https:example.com` and `https:/example.com` are rejected with ERR_INVALID_URL.
  baseURL: 'https://some-domain.com/api/',

  // `allowAbsoluteUrls` determines whether or not absolute URLs will override a configured `baseUrl`.
  // When set to true (default), absolute values for `url` will override `baseUrl`.
  // When set to false, absolute values for `url` will always be prepended by `baseUrl`.
  allowAbsoluteUrls: true,

  // `transformRequest` allows changes to the request data before it is sent to the server
  // This is only applicable for request methods 'PUT', 'POST', 'PATCH' and 'DELETE'
  // The last function in the array must return a string or an instance of Buffer, ArrayBuffer,
  // FormData or Stream
  // You may modify the headers object.
  transformRequest: [function (data, headers) {
    // Do whatever you want to transform the data

    return data;
  }],

  // `transformResponse` allows changes to the response data to be made before
  // it is passed to then/catch
  transformResponse: [function (data) {
    // Do whatever you want to transform the data

    return data;
  }],

  // `parseReviver` is an optional function passed as the
  // second argument (reviver) to JSON.parse()
  parseReviver: function (key, value, context) {
    // In modern environments, context.source provides the raw JSON string
    // allowing for precision-safe parsing of BigInt
    if (typeof value === 'number' && context?.source) {
      const isInteger = Number.isInteger(value);
      const isUnsafe = !Number.isSafeInteger(value);
      const isValidIntegerString = /^-?\d+$/.test(context.source);

      if (isInteger && isUnsafe && isValidIntegerString) {
        try {
          return BigInt(context.source);
        } catch {
          // Fallback: return original value if parsing fails
        }
      }
    }
    return value;
  },

  // `headers` are custom headers to be sent
  headers: {'X-Requested-With': 'XMLHttpRequest'},

  // `params` are the URL parameters to be sent with the request
  // Must be a plain object or a URLSearchParams object
  params: {
    ID: 12345
  },

  // `paramsSerializer` is an optional config that allows you to customize serializing `params`.
  paramsSerializer: {

    // Custom encoder function which sends key/value pairs in an iterative fashion.
    encode?: (param: string): string => { /* Do custom operations here and return transformed string */ },

    // Custom serializer function for the entire parameter. Allows the user to mimic pre 1.x behaviour.
    serialize?: (params: Record<string, any>, options?: ParamsSerializerOptions ),

    // Configuration for formatting array indexes in the params.
    indexes: false, // Three available options: (1) indexes: null (leads to no brackets), (2) (default) indexes: false (leads to empty brackets), (3) indexes: true (leads to brackets with indexes).

    // Maximum object nesting depth when serializing params. Payloads deeper than this throw an
    // AxiosError with code ERR_FORM_DATA_DEPTH_EXCEEDED. Default: 100. Set to Infinity to disable.
    maxDepth: 100

  },

  // `data` is the data to be sent as the request body
  // Only applicable for request methods 'PUT', 'POST', 'DELETE', and 'PATCH'
  // `data` is request-specific: axios does not inherit or deep-merge it from defaults.
  // To add shared body fields, use a request interceptor or transformRequest.
  // When no `transformRequest` is set, it must be of one of the following types:
  // - string, plain object, ArrayBuffer, ArrayBufferView, URLSearchParams
  // - Browser only: FormData, File, Blob
  // - React Native: FormData
  // - Node only: Stream, Buffer, FormData (form-data package)
  data: {
    firstName: 'Fred'
  },

  // `formDataHeaderPolicy` controls how node.js FormData#getHeaders() is copied.
  // 'legacy' (default) copies all returned headers for v1 compatibility.
  // 'content-only' copies only Content-Type and Content-Length.
  formDataHeaderPolicy: 'legacy',

  // syntax alternative to send data into the body
  // method post
  // only the value is sent, not the key
  data: 'Country=Brasil&City=Belo Horizonte',

  // `timeout` specifies the number of milliseconds before the request times out.
  // If the request takes longer than `timeout`, Axios aborts it.
  timeout: 1000, // default is `0` (no timeout)

  // `withCredentials` indicates whether or not cross-site Access-Control requests
  // should be made using credentials
  // This only controls whether the browser sends credentials.
  // It does not control whether the XSRF header is added.
  withCredentials: false, // default

  // `adapter` allows custom handling of requests which makes testing easier.
  // Return a promise and supply a valid response (see lib/adapters/README.md)
  adapter: function (config) {
    /* ... */
  },
  // Also, you can set the name of the built-in adapter, or provide an array with their names
  // to choose the first available in the environment
  adapter: 'xhr', // 'fetch' | 'http' | ['xhr', 'http', 'fetch']

  // `auth` indicates that HTTP Basic auth should be used, and supplies credentials.
  // This will set an `Authorization` header, overwriting any existing
  // `Authorization` custom headers you have set using `headers`.
  // If `auth` is omitted, the Node.js HTTP and fetch adapters can read
  // HTTP Basic auth credentials from the request URL, for example
  // `https://user:[email protected]`. Axios decodes percent-encoded URL
  // credentials, and `auth` takes precedence over URL-embedded credentials.
  // The Node.js HTTP adapter preserves Basic auth on same-origin redirects
  // and strips it on cross-origin redirects.
  // Please note that only HTTP Basic auth is configurable through this parameter.
  // For Bearer tokens and such, use `Authorization` custom headers instead.
  auth: {
    username: 'janedoe',
    password: 's00pers3cret'
  },

  // `responseType` indicates the type of data that the server will respond with
  // options are: 'arraybuffer', 'document', 'json', 'text', 'stream'
  //   browser only: 'blob'
  responseType: 'json', // default

  // `responseEncoding` indicates encoding to use for decoding responses (Node.js only)
  // Note: Ignored for `responseType` of 'stream' or client-side requests
  // options are: 'ascii', 'ASCII', 'ansi', 'ANSI', 'binary', 'BINARY', 'base64', 'BASE64', 'base64url',
  // 'BASE64URL', 'hex', 'HEX', 'latin1', 'LATIN1', 'ucs-2', 'UCS-2', 'ucs2', 'UCS2', 'utf-8', 'UTF-8',
  // 'utf8', 'UTF8', 'utf16le', 'UTF16LE'
  responseEncoding: 'utf8', // default

  // `xsrfCookieName` is the name of the cookie to use as a value for the xsrf token
  xsrfCookieName: 'XSRF-TOKEN', // default

  // `xsrfHeaderName` is the name of the http header that carries the xsrf token value
  xsrfHeaderName: 'X-XSRF-TOKEN', // default

  // `withXSRFToken` defines whether to send the XSRF header in browser requests.
  // `undefined` (default) - set XSRF header only for the same origin requests
  // `true` - always set XSRF header, including for cross-origin requests
  // `false` - never set XSRF header
  // function - resolve with custom logic; receives the internal config object
  withXSRFToken: boolean | undefined | ((config: InternalAxiosRequestConfig) => boolean | undefined),

  // `withXSRFToken` controls whether Axios reads the XSRF cookie and sets the XSRF header.
  // - `undefined` (default): the XSRF header is set only for same-origin requests.
  // - `true`: attempt to set the XSRF header for all requests (including cross-origin).
  // - `false`: never set the XSRF header.
  // - function: a callback that receives the request `config` and returns `true`,
  //   `false`, or `undefined` to decide per-request behavior.
  //
  // Note about `withCredentials`: `withCredentials` controls whether cross-site
  // requests include credentials (cookies and HTTP auth). In older Axios versions,
  // setting `withCredentials: true` implicitly caused Axios to set the XSRF header
  // for cross-origin requests. Newer Axios separates these concerns: to allow the
  // XSRF header to be sent for cross-origin requests you should set both
  // `withCredentials: true` and `withXSRFToken: true`.
  //
  // Example:
  // axios.get('/user', { withCredentials: true, withXSRFToken: true });

  // `onUploadProgress` allows handling of progress events for uploads
  // browser & node.js
  onUploadProgress: function ({loaded, total, progress, bytes, estimated, rate, upload = true}) {
    // Do whatever you want with the Axios progress event
  },

  // `onDownloadProgress` allows handling of progress events for downloads
  // browser & node.js
  onDownloadProgress: function ({loaded, total, progress, bytes, estimated, rate, download = true}) {
    // Do whatever you want with the Axios progress event
  },

  // `maxContentLength` defines the max size of the response content in bytes.
  // It is enforced by the Node.js HTTP adapter and the fetch adapter.
  maxContentLength: 2000,

  // `maxBodyLength` defines the max size of the request content in bytes.
  // It is enforced by the Node.js HTTP adapter and the fetch adapter when the body length can be determined.
  maxBodyLength: 2000,

  // `redact` masks matching config keys when AxiosError#toJSON() is called.
  // Matching is case-insensitive and recursive. It does not change the request.
  redact: ['authorization', 'password'],

  // `validateStatus` defines whether to resolve or reject the promise for a given
  // HTTP response status code. If `validateStatus` returns `true` or is set to
  // `null`, Axios resolves the promise; otherwise, Axios rejects it.
  // Explicit `validateStatus: undefined` resolves every status by default for
  // backward compatibility. Set `transitional.validateStatusUndefinedResolves`
  // to `false` to make explicit `undefined` behave as if this option was omitted.
  validateStatus: function (status) {
    return status >= 200 && status < 300; // default
  },

  // `maxRedirects` defines the maximum number of redirects to follow in node.js.
  // If set to 0, Axios follows no redirects.
  maxRedirects: 21, // default

  // `sensitiveHeaders` (Node only option) lists custom secret-bearing headers
  // (such as `X-API-Key`) to remove from cross-origin redirects. Matching is
  // case-insensitive. Same-origin redirects keep these headers. If
  // `maxRedirects` is 0, this option is not used.
  sensitiveHeaders: ['X-API-Key'],

  // `beforeRedirect` defines a function that Axios calls before redirect.
  // Use this to adjust the request options upon redirecting,
  // to inspect the latest response headers,
  // or to cancel the request by throwing an error
  // If maxRedirects is set to 0, `beforeRedirect` is not used.

  beforeRedirect: (options, { headers }) => {
    if (
      options.hostname === "example.com" &&
      options.protocol === "https:"
    ) {
      options.auth = "user:password";
    }
  },
  // Security note:
  // The `beforeRedirect` hook runs after sensitive headers are stripped during redirects.
  // `follow-redirects` removes credentials on protocol downgrades
  // (HTTPS to HTTP). Because `beforeRedirect` runs after that step,
  // re-injecting credentials without checking the destination can expose
  // sensitive data. Only add credentials for trusted HTTPS destinations.

  // `socketPath` defines a UNIX Socket to be used in node.js.
  // e.g. '/var/run/docker.sock' to send requests to the docker daemon.
  // Only either `socketPath` or `proxy` can be specified.
  // If both are specified, `socketPath` is used.
  //
  // Security: when `socketPath` is set, hostname/port of the URL are ignored,
  // which bypasses hostname-based SSRF protections. Never derive `socketPath`
  // from untrusted input. Use `allowedSocketPaths` (below) to restrict accepted
  // socket paths for defense-in-depth.
  socketPath: null, // default

  // `allowedSocketPaths` restricts which `socketPath` values are accepted.
  // Accepts a string or array of strings. Entries and the incoming socketPath
  // are compared after path.resolve(). A mismatch throws AxiosError with code
  // `ERR_BAD_OPTION_VALUE`. When null/undefined, no restriction is applied.
  allowedSocketPaths: null, // default

  // `transport` determines the transport method for the request.
  // If defined, Axios uses it. Otherwise, if `maxRedirects` is 0,
  // Axios uses the default `http` or `https` library, depending on the protocol specified in `protocol`.
  // Otherwise, Axios uses the `httpFollow` or `httpsFollow` library, again depending on the protocol,
  // which can handle redirects.
  transport: undefined, // default

  // `httpAgent` and `httpsAgent` define a custom agent to be used when performing http
  // and https requests, respectively, in node.js. This allows options to be added like
  // `keepAlive` that are not enabled by default before Node.js v19.0.0. After Node.js
  // v19.0.0, you no longer need to customize the agent to enable `keepAlive` because
  // `http.globalAgent` has `keepAlive` enabled by default.
  httpAgent: new http.Agent({ keepAlive: true }),
  httpsAgent: new https.Agent({ keepAlive: true }),

  // `proxy` defines the hostname, port, and protocol of the proxy server.
  // You can also define your proxy using the conventional `http_proxy` and
  // `https_proxy` environment variables. If you are using environment variables
  // for your proxy configuration, you can also define a `no_proxy` environment
  // variable as a comma-separated list of domains that should not be proxied.
  // Use `false` to disable proxies, ignoring environment variables.
  // On Node.js versions with native environment proxy support, axios defers
  // environment proxy handling to Node when the selected agent has `proxyEnv`
  // enabled, including processes started with `NODE_USE_ENV_PROXY=1`,
  // `--use-env-proxy`, or `NODE_OPTIONS=--use-env-proxy`. Custom agents without
  // `proxyEnv` continue to use axios environment proxy resolution. Explicit
  // `proxy` config is still handled by axios.
  // `auth` indicates that HTTP Basic auth should be used to connect to the proxy, and
  // supplies credentials.
  // For `http://` targets, axios sends the request to the proxy in
  // forward-proxy mode and stamps `Proxy-Authorization` onto the request
  // headers (overwriting any user-supplied `Proxy-Authorization` header).
  // For `https://` targets, axios establishes a CONNECT tunnel through the
  // proxy and performs TLS end-to-end with the origin; `Proxy-Authorization`
  // is sent on the CONNECT request only, never on the wrapped TLS request,
  // so the proxy never sees the URL, headers, or body. Axios forwards
  // `httpsAgent` TLS options such as `ca`, `cert`, `key`, and
  // `rejectUnauthorized` to the generated tunneling agent, so they still apply
  // to the origin TLS connection.
  // If you supply an `HttpsProxyAgent`, axios leaves tunneling to that agent.
  // If the proxy server uses HTTPS, then you must set the protocol to `https`.
  // A user-supplied `Host` header in `headers` is preserved when forwarding
  // through a proxy (case-insensitive match on `host`/`Host`/`HOST`); this
  // lets you target a virtual host that differs from the request URL, for
  // example, hitting `127.0.0.1:4000` while having the proxy treat the
  // request as `example.com`. If no `Host` header is supplied, axios
  // defaults it to the request URL's `hostname:port` as before. The Host
  // header is only set in forward-proxy mode (HTTP targets); for HTTPS
  // tunneling the Host header is sent inside the TLS connection, not seen
  // by the proxy.
  proxy: {
    protocol: 'https',
    host: '127.0.0.1',
    // hostname: '127.0.0.1' // Takes precedence over 'host' if both are defined
    port: 9000,
    auth: {
      username: 'mikeymike',
      password: 'rapunz3l'
    }
  },

  // `cancelToken` specifies a cancel token that can be used to cancel the request
  // (see Cancellation section below for details)
  cancelToken: new CancelToken(function (cancel) {
  }),

  // an alternative way to cancel Axios requests using AbortController
  signal: new AbortController().signal,

  // `decompress` indicates whether or not the response body should be decompressed
  // automatically. If set to `true` will also remove the 'content-encoding' header
  // from the responses objects of all decompressed responses
  // Axios supports gzip, deflate, brotli, and zstd when the current Node.js
  // runtime provides the corresponding zlib decompressor.
  // - Node only (XHR cannot turn off decompression)
  decompress: true, // default

  // `insecureHTTPParser` boolean.
  // Indicates where to use an insecure HTTP parser that accepts invalid HTTP headers.
  // This may allow interoperability with non-conformant HTTP implementations.
  // Using the insecure parser should be avoided.
  // see options https://nodejs.org/dist/latest-v12.x/docs/api/http.html#http_http_request_url_options_callback
  // see also https://nodejs.org/en/blog/vulnerability/february-2020-security-releases/#strict-http-header-parsing-none
  insecureHTTPParser: undefined, // default

  // transitional options for backward compatibility that may be removed in the newer versions
  transitional: {
    // silent JSON parsing mode
    // `true`  - ignore JSON parsing errors and set response.data to null if parsing failed (old behaviour)
    // `false` - throw SyntaxError if JSON parsing failed
    // Important: this option only takes effect when `responseType` is explicitly set to 'json'.
    // When `responseType` is omitted (defaults to no value), axios uses `forcedJSONParsing`
    // to attempt JSON parsing, but will silently return the raw string on failure regardless
    // of this setting. To have invalid JSON throw errors, use:
    //   { responseType: 'json', transitional: { silentJSONParsing: false } }
    silentJSONParsing: true, // default value for the current Axios version

    // try to parse the response string as JSON even if `responseType` is not 'json'
    forcedJSONParsing: true,

    // throw ETIMEDOUT error instead of generic ECONNABORTED on request timeouts
    clarifyTimeoutError: false,

    // keep explicit `validateStatus: undefined` resolving every response status
    // for backward compatibility. Set to false to make explicit undefined behave
    // as if validateStatus was omitted.
    validateStatusUndefinedResolves: true,

    // advertise `zstd` in the default Accept-Encoding header when the current
    // Node.js runtime supports zstd decompression. Axios still decompresses
    // zstd responses when support exists and `decompress` is true.
    advertiseZstdAcceptEncoding: false,

    // use the legacy interceptor request/response ordering
    legacyInterceptorReqResOrdering: true, // default
  },

  env: {
    // The FormData class to be used to automatically serialize the payload into a FormData object
    FormData: window?.FormData || global?.FormData
  },

  formSerializer: {
      visitor: (value, key, path, helpers) => {}; // custom visitor function to serialize form values
      dots: boolean; // use dots instead of brackets format
      metaTokens: boolean; // keep special endings like {} in parameter key
      indexes: boolean; // array indexes format null - no brackets, false - empty brackets, true - brackets with indexes
      maxDepth: 100; // maximum object nesting depth; throws AxiosError (ERR_FORM_DATA_DEPTH_EXCEEDED) if exceeded. Set to Infinity to disable.
  },

  // http adapter only (node.js)
  maxRate: [
    100 * 1024, // 100KB/s upload limit,
    100 * 1024  // 100KB/s download limit
  ]
}

For custom secret-bearing headers in Node.js, list them in sensitiveHeaders so Axios removes them when following a redirect to another origin:

axios.get('https://api.example.com/users', {
  headers: { 'X-API-Key': 'secret' },
  sensitiveHeaders: ['X-API-Key'],
});

Strict RFC 3986 percent-encoding for query params

By default, axios decodes %3A, %24, %2C and %20 back to :, $, , and + for readability (the + follows the application/x-www-form-urlencoded convention for spaces in query strings). These characters are valid in a query component under RFC 3986, so the default output is correct, but some backends require strict percent-encoding and reject the readable form.

Override the default encoder via paramsSerializer.encode:

// Per-request: emit strict RFC 3986 percent-encoding for query values
axios.get('/foo', {
  params: { filter: JSON.stringify({ startedAt: '2026-01-23' }) },
  paramsSerializer: { encode: encodeURIComponent },
});

// Or set it on the instance defaults
const client = axios.create({
  paramsSerializer: { encode: encodeURIComponent },
});

HTTP/2 support

Axios has experimental HTTP/2 support in the Node.js HTTP adapter.

Support depends on the runtime environment and Node.js version. Redirects and some adapter behavior may differ from HTTP/1.1.

Options like httpVersion and http2Options are adapter-specific and may not work the same way in every environment.

If you need HTTP/2, check runtime support or use a custom adapter.

Response schema

The response to a request contains the following information.

{
  // `data` is the response that was provided by the server
  data: {},

  // `status` is the HTTP status code from the server response
  status: 200,

  // `statusText` is the HTTP status message from the server response
  statusText: 'OK',

  // `headers` the HTTP headers that the server responded with
  // All header names are lowercase and can be accessed using the bracket notation.
  // Example: `response.headers['content-type']`
  headers: {},

  // `config` is the config that was provided to `axios` for the request
  config: {},

  // `request` is the request that generated this response
  // It is the last ClientRequest instance in node.js (in redirects)
  // and an XMLHttpRequest instance in the browser
  request: {}
}

When using then, you receive the response like this:

const response = await axios.get('/user/12345');
console.log(response.data);
console.log(response.status);
console.log(response.statusText);
console.log(response.headers);
console.log(response.config);

When using catch, or passing a rejection callback as the second parameter of then, read the response from the error object. See Handling errors.

Config defaults

Config defaults apply to every request.

Global axios defaults

axios.defaults.baseURL = 'https://api.example.com';

// Important: If you use axios with multiple domains, Axios sends AUTH_TOKEN to all of them.
// See below for an example using Custom instance defaults instead.
axios.defaults.headers.common['Authorization'] = AUTH_TOKEN;

axios.defaults.headers.post['Content-Type'] = 'application/x-www-form-urlencoded';

Custom instance defaults

// Set config defaults when creating the instance
const instance = axios.create({
  baseURL: 'https://api.example.com',
});

// Alter defaults after instance has been created
instance.defaults.headers.common['Authorization'] = AUTH_TOKEN;

Config order of precedence

Axios merges config in this order: library defaults from lib/defaults/index.js, the instance defaults property, and the request config argument. Later values take precedence over earlier ones.

Some options are request-specific and are only taken from the request config. data is one of those options: axios does not inherit or deep-merge request bodies from global or instance defaults. If every request needs shared body fields, add them with a request interceptor or transformRequest, and scope that logic carefully so sensitive values are not sent to the wrong endpoint.

// Create an instance using the config defaults provided by the library
// At this point the timeout config value is `0` as is the default for the library
const instance = axios.create();

// Override timeout default for the library
// Now all requests using this instance will wait 2.5 seconds before timing out
instance.defaults.timeout = 2500;

// Override timeout for this request as it's known to take a long time
instance.get('/longRequest', {
  timeout: 5000,
});

Interceptors

You can intercept requests or responses before methods like .get() or .post() resolve their promises (before code inside then or catch, or after await)

const instance = axios.create();

// Add a request interceptor
instance.interceptors.request.use(
  function (config) {
    // Do something before the request is sent
    return config;
  },
  function (error) {
    // Do something with the request error
    return Promise.reject(error);
  }
);

// Add a response interceptor
instance.interceptors.response.use(
  function (response) {
    // Any status code that lies within the range of 2xx causes this function to trigger
    // Do something with response data
    return response;
  },
  function (error) {
    // Any status codes that fall outside the range of 2xx cause this function to trigger
    // Do something with response error
    return Promise.reject(error);
  }
);

If you need to remove an interceptor later you can.

const instance = axios.create();
const myInterceptor = instance.interceptors.request.use(function () {
  /*...*/
});
instance.interceptors.request.eject(myInterceptor);

You can also clear all interceptors for requests or responses.

const instance = axios.create();
instance.interceptors.request.use(function () {
  /*...*/
});
instance.interceptors.request.clear(); // Removes interceptors from requests
instance.interceptors.response.use(function () {
  /*...*/
});
instance.interceptors.response.clear(); // Removes interceptors from responses

You can add interceptors to a custom instance of axios.

const instance = axios.create();
instance.interceptors.request.use(function () {
  /*...*/
});

When you add request interceptors, they are presumed to be asynchronous by default. This can cause a delay in the execution of your axios request when the main thread is blocked (a promise is created under the hood for the interceptor and your request gets put at the bottom of the call stack). If your request interceptors are synchronous you can add a flag to the options object that will tell axios to run the code synchronously and avoid any delays in request execution.

axios.interceptors.request.use(
  function (config) {
    config.headers.test = 'I am only a header!';
    return config;
  },
  null,
  { synchronous: true }
);

If a synchronous request interceptor throws, Axios calls that interceptor's paired rejection handler and stops running the remaining request interceptors. A rejection handler that returns normally, including one that returns undefined or a fulfilled Promise, treats the error as handled: Axios dispatches the request with the last valid config, and ignores the handler's return value. To block dispatch, omit the rejection handler or throw/return a rejected Promise from it. The final error then continues through response rejection interceptors.

Use a rejected Promise when validation must prevent the request:

axios.interceptors.request.use(
  function validate(config) {
    if (!config.headers.has('Authorization')) {
      throw new Error('Authorization is required');
    }
    return config;
  },
  function rejectInvalidRequest(error) {
    return Promise.reject(error);
  },
  { synchronous: true }
);

A logging-only rejection handler can return normally to preserve the existing request-continuation behavior:

axios.interceptors.request.use(
  function prepare(config) {
    throw new Error('Optional preparation failed');
  },
  function logPreparationFailure(error) {
    console.warn(error);
    // Returning normally dispatches with the last valid config.
  },
  { synchronous: true }
);

If you want to execute a particular interceptor based on a runtime check, you can add a runWhen function to the options object. The request interceptor will not run if and only if the return of runWhen is false. Axios calls the function with the config object (don't forget that you can bind your own arguments to it as well.) This can be handy when you have an asynchronous request interceptor that only needs to run at certain times.

function onGetCall(config) {
  return config.method === 'get';
}
axios.interceptors.request.use(
  function (config) {
    config.headers.test = 'special get headers';
    return config;
  },
  null,
  { runWhen: onGetCall }
);

Note: The options parameter (with synchronous and runWhen properties) is only supported for request interceptors at the moment.

Interceptor execution order

Request and response interceptors use different execution orders.

Request interceptors run in reverse order (LIFO: last in, first out). The last interceptor added runs first.

Response interceptors run in the order they were added (FIFO: first in, first out). The first interceptor added runs first.

Example:

const instance = axios.create();

const interceptor = (id) => (base) => {
  console.log(id);
  return base;
};

instance.interceptors.request.use(interceptor('Request Interceptor 1'));
instance.interceptors.request.use(interceptor('Request Interceptor 2'));
instance.interceptors.request.use(interceptor('Request Interceptor 3'));
instance.interceptors.response.use(interceptor('Response Interceptor 1'));
instance.interceptors.response.use(interceptor('Response Interceptor 2'));
instance.interceptors.response.use(interceptor('Response Interceptor 3'));

// Console output:
// Request Interceptor 3
// Request Interceptor 2
// Request Interceptor 1
// [HTTP request is made]
// Response Interceptor 1
// Response Interceptor 2
// Response Interceptor 3

Multiple interceptors

When a response is fulfilled and multiple response interceptors are registered:

  • Each interceptor runs in registration order.
  • Each interceptor receives the result from the previous interceptor.
  • The chain returns the result from the last interceptor.
  • If a fulfillment interceptor throws, Axios skips the next fulfillment interceptor and calls the next rejection interceptor.
  • After the error is caught, later fulfillment interceptors run again, just like in a promise chain.

Read the interceptor tests to see all this in code.

Error types

Axios error messages include details that can help you debug the request.

Axios errors use this structure:

PropertyDefinition
messageA quick summary of the error message and the status it failed with.
nameThis defines where the error originated from. For axios, it will always be an 'AxiosError'.
stackStack trace for the error.
configAn axios config object with specific instance configurations defined by the user from when the request was made
codeAxios error code. The table below lists internal Axios error codes.
statusHTTP response status code. See here for common HTTP response status code meanings.

These are the internal Axios error codes:

CodeDefinition
ERR_BAD_OPTION_VALUEInvalid value provided in axios configuration.
ERR_BAD_OPTIONInvalid option provided in axios configuration.
ERR_NOT_SUPPORTFeature or method not supported in the current axios environment.
ERR_DEPRECATEDDeprecated feature or method used in axios.
ERR_INVALID_URLInvalid URL provided for axios request.
ECONNABORTEDTypically indicates that the request has been timed out (unless transitional.clarifyTimeoutError is set) or aborted by the browser or its plugin.
ERR_CANCELEDThe user explicitly canceled the request with an AbortSignal or CancelToken.
ETIMEDOUTRequest timed out after exceeding the configured Axios timeout. Set transitional.clarifyTimeoutError to true; otherwise Axios throws a generic ECONNABORTED error.
ERR_NETWORKNetwork-related issue. In the browser, this error can also be caused by a CORS or Mixed Content policy violation. The browser does not allow the JS code to clarify the real reason for the error caused by security issues, so please check the console.
ERR_FR_TOO_MANY_REDIRECTSRequest exceeded the configured maximum number of redirects.
ERR_BAD_RESPONSEResponse cannot be parsed properly or is in an unexpected format. Usually related to a response with 5xx status code.
ERR_BAD_REQUESTThe request has an unexpected format or is missing required parameters. Usually related to a response with 4xx status code.

Handling errors

By default, Axios rejects responses with status codes outside the 2xx range.

axios.get('/user/12345').catch(function (error) {
  if (error.response) {
    // The request was made and the server responded with a status code
    // that falls out of the range of 2xx
    console.log(error.response.data);
    console.log(error.response.status);
    console.log(error.response.headers);
  } else if (error.request) {
    // The request was made but no response was received
    // `error.request` is an instance of XMLHttpRequest in the browser and an instance of
    // http.ClientRequest in node.js
    console.log(error.request);
  } else {
    // Something happened in setting up the request that triggered an Error
    console.log('Error', error.message);
  }
  console.log(error.config);
});

Axios rejects an http: or https: request url or baseURL that omits // after the protocol. For example, https:example.com and https:/example.com reject with AxiosError code ERR_INVALID_URL instead of being silently normalized. Pass a well-formed URL such as https://example.com.

The error identifies the malformed URL, for example Invalid URL "https:example.com": missing "//" after protocol. To keep logs useful without leaking secrets, Axios reports the control-character-normalized URL with credentials, query parameter values, and fragment contents redacted; the scheme, host, path, and query parameter names remain visible.

Use validateStatus to override the default condition (status >= 200 && status < 300) and choose which HTTP status codes should reject.

axios.get('/user/12345', {
  validateStatus: function (status) {
    return status < 500; // Resolve only if the status code is less than 500
  },
});

By default, explicit validateStatus: undefined keeps legacy behavior and resolves every response status because transitional.validateStatusUndefinedResolves defaults to true. Set it to false to make explicit validateStatus: undefined behave like the option was omitted, so Axios uses the configured/default validator and rejects non-2xx responses by default.

validateStatus: null still accepts every response status. If you disable the transitional behavior and intentionally want all statuses to resolve, use null or () => true.

axios.get('/user/12345', {
  validateStatus: undefined,
  transitional: {
    validateStatusUndefinedResolves: false,
  },
});

Use toJSON to get more information about the HTTP error.

axios.get('/user/12345').catch(function (error) {
  console.log(error.toJSON());
});

To avoid logging secrets from error.config, pass a redact array in the request config. Matching config keys are masked case-insensitively at any depth when AxiosError#toJSON() is called.

axios
  .get('/user/12345', {
    headers: { Authorization: 'Bearer token' },
    redact: ['authorization'],
  })
  .catch(function (error) {
    console.log(error.toJSON().config.headers.Authorization); // [REDACTED ****]
  });

Handling timeouts

async function fetchWithTimeout() {
  try {
    const response = await axios.get('https://example.com/data', {
      timeout: 5000, // 5 seconds
      transitional: {
        // set to true if you prefer ETIMEDOUT over ECONNABORTED
        clarifyTimeoutError: false,
      },
    });

    console.log('Response:', response.data);
  } catch (error) {
    if (axios.isAxiosError(error)) {
      if (error.code === 'ECONNABORTED' || error.code === 'ETIMEDOUT') {
        console.error('Request timed out. Please try again.');
        return;
      }

      console.error('Axios error:', error.message);
      return;
    }

    console.error('Unexpected error:', error);
  }
}

Cancellation

AbortController

Since v0.22.0, Axios supports AbortController:

const controller = new AbortController();

axios
  .get('/foo/bar', {
    signal: controller.signal,
  })
  .then(function (response) {
    //...
  });
// cancel the request
controller.abort();

CancelToken (deprecated)

You can also cancel a request using a CancelToken.

The axios cancel token API is based on the withdrawn cancellable promises proposal.

This API is deprecated since v0.22.0 and should not be used in new projects.

Create a cancel token with the CancelToken.source factory:

const CancelToken = axios.CancelToken;
const source = CancelToken.source();

axios
  .get('/user/12345', {
    cancelToken: source.token,
  })
  .catch(function (thrown) {
    if (axios.isCancel(thrown)) {
      console.log('Request canceled', thrown.message);
    } else {
      // handle error
    }
  });

axios.post(
  '/user/12345',
  {
    name: 'new name',
  },
  {
    cancelToken: source.token,
  }
);

// cancel the request (the message parameter is optional)
source.cancel('Operation canceled by the user.');

You can also pass an executor function to the CancelToken constructor:

const CancelToken = axios.CancelToken;
let cancel;

axios.get('/user/12345', {
  cancelToken: new CancelToken(function executor(c) {
    // An executor function receives a cancel function as a parameter
    cancel = c;
  }),
});

// cancel the request
cancel();

CancelToken also exposes low-level helpers for legacy integrations:

const source = axios.CancelToken.source();

const listener = (cancel) => {
  console.log(cancel.message);
};

source.token.subscribe(listener);

const signal = source.token.toAbortSignal();
// Pass `signal` to APIs that accept AbortSignal.

source.cancel('Operation canceled by the user.');
source.token.unsubscribe(listener);

Canceled requests reject with axios.CanceledError. The legacy axios.Cancel export is an alias of axios.CanceledError, and cancellation errors include __CANCEL__ for axios.isCancel compatibility.

Note: You can cancel several requests with the same cancel token or abort controller. If a cancellation token is already cancelled when an Axios request starts, Axios cancels the request immediately without making a real request.

During the transition period, you can use both cancellation APIs, even for the same request:

const controller = new AbortController();
const source = axios.CancelToken.source();

axios.get('/user/12345', {
  cancelToken: source.token,
  signal: controller.signal,
});

controller.abort();
source.cancel('Operation canceled by the user.');

Using application/x-www-form-urlencoded format

URLSearchParams

By default, axios serializes JavaScript objects to JSON. To send data as application/x-www-form-urlencoded, use the URLSearchParams API. It works in most browsers and in Node v10 and later.

const params = new URLSearchParams({ foo: 'bar' });
params.append('extraparam', 'value');
axios.post('/foo', params);

Query string (older browsers)

For very old browsers, use a polyfill and make sure it patches the global environment.

Alternatively, you can encode data using the qs library:

const qs = require('qs');
axios.post('/foo', qs.stringify({ bar: 123 }));

With ES modules:

import qs from 'qs';
const data = { bar: 123 };
const options = {
  method: 'POST',
  headers: { 'content-type': 'application/x-www-form-urlencoded' },
  data: qs.stringify(data),
  url,
};
axios(options);

Older Node.js versions

For older Node.js engines, use the querystring module:

const querystring = require('querystring');
axios.post('https://something.com/', querystring.stringify({ foo: 'bar' }));

You can also use the qs library.

Note: The qs library is preferable if you need to stringify nested objects, as the querystring method has known issues with that use case.

Automatic serialization to URLSearchParams

Axios automatically serializes the data object to urlencoded format if the content-type header is set to "application/x-www-form-urlencoded".

const data = {
  x: 1,
  arr: [1, 2, 3],
  arr2: [1, [2], 3],
  users: [
    { name: 'Peter', surname: 'Griffin' },
    { name: 'Thomas', surname: 'Anderson' },
  ],
};

await axios.postForm('https://postman-echo.com/post', data, {
  headers: { 'content-type': 'application/x-www-form-urlencoded' },
});

The server receives these fields:

  {
    x: '1',
    'arr[]': [ '1', '2', '3' ],
    'arr2[0]': '1',
    'arr2[1][0]': '2',
    'arr2[2]': '3',
    'arr3[]': [ '1', '2', '3' ],
    'users[0][name]': 'Peter',
    'users[0][surname]': 'griffin',
    'users[1][name]': 'Thomas',
    'users[1][surname]': 'Anderson'
  }

If your backend body parser, such as body-parser for express.js, supports nested object decoding, the server receives the same object structure:

const app = express();

app.use(bodyParser.urlencoded({ extended: true })); // support encoded bodies

app.post('/', function (req, res, next) {
  // echo body as JSON
  res.send(JSON.stringify(req.body));
});

server = app.listen(3000);

Using multipart/form-data format

FormData

To send data as multipart/form-data, pass a FormData instance as the payload. You do not need to set the Content-Type header. Axios detects it from the payload type. For browser, web worker, and React Native FormData, leave Content-Type unset so the runtime can add the multipart boundary.

const formData = new FormData();
formData.append('foo', 'bar');

axios.post('https://httpbin.org/post', formData);

In node.js, use the form-data library:

const FormData = require('form-data');

const form = new FormData();
form.append('my_field', 'my value');
form.append('my_buffer', Buffer.alloc(10));
form.append('my_file', fs.createReadStream('/foo/bar.jpg'));

axios.post('https://example.com', form);

In node.js, when a FormData object provides getHeaders(), axios copies all returned headers by default for v1 compatibility. If the FormData object is custom or not fully trusted, set formDataHeaderPolicy: 'content-only' to copy only Content-Type and Content-Length, and set any other request headers explicitly with the request headers config.

Automatic serialization to FormData

Since v0.27.0, Axios can serialize an object to FormData if the request Content-Type header is set to multipart/form-data.

This request submits data as FormData in browsers and Node.js:

import axios from 'axios';

axios
  .post(
    'https://httpbin.org/post',
    { x: 1 },
    {
      headers: {
        'Content-Type': 'multipart/form-data',
      },
    }
  )
  .then(({ data }) => console.log(data));

The Node.js build uses the form-data polyfill by default.

You can override the FormData class with the env.FormData config option, but most applications do not need this:

const axios = require('axios');
var FormData = require('form-data');

axios
  .post(
    'https://httpbin.org/post',
    { x: 1, buf: Buffer.alloc(10) },
    {
      headers: {
        'Content-Type': 'multipart/form-data',
      },
    }
  )
  .then(({ data }) => console.log(data));

The Axios FormData serializer supports these special endings:

  • {} - serialize the value with JSON.stringify
  • [] - unwrap the array-like object as separate fields with the same key

Note: Arrays and FileList objects are unwrapped by default.

FormData serializer supports additional options via config.formSerializer: object property to handle rare cases:

  • visitor: Function - user-defined visitor function that Axios calls recursively to serialize the data object to a FormData object by following custom rules.

  • dots: boolean = false - use dot notation instead of brackets to serialize arrays and objects;

  • metaTokens: boolean = true - add the special ending (e.g user{}: '{"name": "John"}') in the FormData key. A backend body parser can use this meta-information to parse the value as JSON.

  • indexes: null|false|true = false - controls how Axios adds indexes to unwrapped keys of flat array-like objects.

    • null - don't add brackets (arr: 1, arr: 2, arr: 3)
    • false(default) - add empty brackets (arr[]: 1, arr[]: 2, arr[]: 3)
    • true - add brackets with indexes (arr[0]: 1, arr[1]: 2, arr[2]: 3)
  • maxDepth: number = 100 - maximum object nesting depth the serializer will recurse into. If the input object exceeds this depth, an AxiosError with code: 'ERR_FORM_DATA_DEPTH_EXCEEDED' is thrown instead of overflowing the call stack. This protects server applications from DoS attacks via deeply nested payloads. Set to Infinity to disable the limit and restore pre-fix behaviour.

  • Blob: typeof Blob - Blob constructor used when converting ArrayBuffer-like values for spec-compliant FormData. Override it only for runtimes that provide a compatible Blob constructor under a different binding.

// Raise the limit for a schema that genuinely nests deeper than 100 levels:
axios.postForm('/api', data, { formSerializer: { maxDepth: 200 } });

// Same protection applies to params serialization:
axios.get('/api', { params: data, paramsSerializer: { maxDepth: 200 } });

Given this object:

const obj = {
  x: 1,
  arr: [1, 2, 3],
  arr2: [1, [2], 3],
  users: [
    { name: 'Peter', surname: 'Griffin' },
    { name: 'Thomas', surname: 'Anderson' },
  ],
  'obj2{}': [{ x: 1 }],
};

The Axios serializer appends these fields:

const formData = new FormData();
formData.append('x', '1');
formData.append('arr[]', '1');
formData.append('arr[]', '2');
formData.append('arr[]', '3');
formData.append('arr2[0]', '1');
formData.append('arr2[1][0]', '2');
formData.append('arr2[2]', '3');
formData.append('users[0][name]', 'Peter');
formData.append('users[0][surname]', 'Griffin');
formData.append('users[1][name]', 'Thomas');
formData.append('users[1][surname]', 'Anderson');
formData.append('obj2{}', '[{"x":1}]');

Converting FormData to JSON

axios.formToJSON() converts FormData field names that use dot or bracket notation into nested objects and arrays. Only ., [, and ] are structural separators. Other characters, including -, spaces, +, *, and &, remain part of the literal key.

const form = new FormData();
form.append('user-name', 'johndoe');
form.append('user.name', 'john');

console.log(axios.formToJSON(form));
// {
//   'user-name': 'johndoe',
//   user: { name: 'john' }
// }

user[name] also creates { user: { name: ... } }, while items[] creates an array.

Axios supports postForm, putForm, and patchForm as shortcuts for the matching HTTP methods with the Content-Type header preset to multipart/form-data.

Posting files

Submit a single file:

await axios.postForm('https://httpbin.org/post', {
  myVar: 'foo',
  file: document.querySelector('#fileInput').files[0],
});

or multiple files as multipart/form-data:

await axios.postForm('https://httpbin.org/post', {
  'files[]': document.querySelector('#fileInput').files,
});

FileList object can be passed directly:

await axios.postForm('https://httpbin.org/post', document.querySelector('#fileInput').files);

Axios sends all files with the same field name: files[].

HTML form posting (browser)

Pass an HTML Form element as a payload to submit it as multipart/form-data content.

await axios.postForm('https://httpbin.org/post', document.querySelector('#htmlForm'));

FormData and HTMLForm objects can also be posted as JSON by explicitly setting the Content-Type header to application/json:

await axios.post('https://httpbin.org/post', document.querySelector('#htmlForm'), {
  headers: {
    'Content-Type': 'application/json',
  },
});

For example, the Form

<form id="form">
  <input type="text" name="foo" value="1" />
  <input type="text" name="deep.prop" value="2" />
  <input type="text" name="deep prop spaced" value="3" />
  <input type="text" name="baz" value="4" />
  <input type="text" name="baz" value="5" />

  <select name="user.age">
    <option value="value1">Value 1</option>
    <option value="value2" selected>Value 2</option>
    <option value="value3">Value 3</option>
  </select>

  <input type="submit" value="Save" />
</form>

submits this JSON object:

{
  "foo": "1",
  "deep": {
    "prop": "2"
  },
  "deep prop spaced": "3",
  "baz": [
    "4",
    "5"
  ],
  "user": {
    "age": "value2"
  }
}

Only dot and bracket notation create nested paths when a form is converted to JSON. Spaces and other punctuation remain part of literal field names.

Sending Blobs/Files as JSON (base64) is not currently supported.

Progress capturing

Axios can capture request upload and download progress in browsers and Node.js. Progress events are limited to 3 times per second.

await axios.post(url, data, {
  onUploadProgress: function (axiosProgressEvent) {
    /*{
      loaded: number;
      total?: number;
      progress?: number; // in range [0..1]
      bytes: number; // how many bytes have been transferred since the last trigger (delta)
      estimated?: number; // estimated time in seconds
      rate?: number; // upload speed in bytes
      upload: true; // upload sign
    }*/
  },

  onDownloadProgress: function (axiosProgressEvent) {
    /*{
      loaded: number;
      total?: number;
      progress?: number;
      bytes: number;
      estimated?: number;
      rate?: number; // download speed in bytes
      download: true; // download sign
    }*/
  },
});

You can also track stream upload/download progress in node.js:

const { data } = await axios.post(SERVER_URL, readableStream, {
  onUploadProgress: ({ progress }) => {
    console.log((progress * 100).toFixed(2));
  },

  headers: {
    'Content-Length': contentLength,
  },

  maxRedirects: 0, // avoid buffering the entire stream
});

Note: Capturing FormData upload progress is not currently supported in node.js environments.

Warning: Set maxRedirects: 0 when uploading streams in node.js. The follow-redirects package buffers the entire stream in RAM and does not follow the "backpressure" algorithm.

Rate limiting

Download and upload rate limits can only be set for the http adapter (node.js):

const { data } = await axios.post(LOCAL_SERVER_URL, myBuffer, {
  onUploadProgress: ({ progress, rate }) => {
    console.log(`Upload [${(progress * 100).toFixed(2)}%]: ${(rate / 1024).toFixed(2)}KB/s`);
  },

  maxRate: [100 * 1024], // 100KB/s limit
});

AxiosHeaders

Axios includes an AxiosHeaders class for working with headers through a Map-like API. HTTP header names are case-insensitive, but Axios keeps the original header case for style and for servers that incorrectly depend on case. Directly manipulating the headers object still works, but it is deprecated.

Working with headers

An AxiosHeaders instance can contain several internal value types that control setting and merging. Axios gets the final headers object with string values by calling toJSON.

Note: By JSON here we mean an object consisting only of string values intended to be sent over the network.

The header value can be one of the following types:

  • string - normal string value sent to the server
  • null - skip header when rendering to JSON
  • false - skip header when rendering to JSON. Also indicates that the set method must be called with rewrite set to true to overwrite this value (Axios uses this internally to allow users to opt out of installing certain headers like User-Agent or Content-Type)
  • undefined - value is not set

Note: The header value is considered set if it is not equal to undefined.

The headers object is always initialized inside interceptors and transformers:

axios.interceptors.request.use((request: InternalAxiosRequestConfig) => {
  request.headers.set('My-header', 'value');

  request.headers.set({
    'My-set-header1': 'my-set-value1',
    'My-set-header2': 'my-set-value2',
  });

  request.headers.set('User-Agent', false); // prevent Axios from setting this header later

  request.headers.setContentType('text/plain');

  request.headers['My-set-header2'] = 'newValue'; // direct access is deprecated

  return request;
});

You can iterate over an AxiosHeaders instance using a for...of statement:

const headers = new AxiosHeaders({
  foo: '1',
  bar: '2',
  baz: '3',
});

for (const [header, value] of headers) {
  console.log(header, value);
}

// foo 1
// bar 2
// baz 3

Preserving a specific header case

Header names are case-insensitive, but AxiosHeaders keeps the case of the first matching key it sees. If you need a specific case for non-standard case-sensitive servers, define a case preset with undefined and then set the value later:

const api = axios.create();

api.defaults.headers.common = {
  'content-type': undefined,
  accept: undefined,
};

await api.put(url, data, {
  headers: {
    'Content-Type': 'application/octet-stream',
    Accept: 'application/json',
  },
});

You can also compose the same behavior with AxiosHeaders.concat:

const headers = axios.AxiosHeaders.concat(
  { 'content-type': undefined },
  { 'Content-Type': 'application/octet-stream' }
);

await axios.put(url, data, { headers });

new AxiosHeaders(headers?)

Constructs a new AxiosHeaders instance.

constructor(headers?: RawAxiosHeaders | AxiosHeaders | string);

If the headers object is a string, Axios parses it as raw HTTP headers.

const headers = new AxiosHeaders(`
Host: www.bing.com
User-Agent: curl/7.54.0
Accept: */*`);

console.log(headers);

// Object [AxiosHeaders] {
//   host: 'www.bing.com',
//   'user-agent': 'curl/7.54.0',
//   accept: '*/*'
// }

AxiosHeaders#set

set(headerName, value: Axios, rewrite?: boolean);
set(headerName, value, rewrite?: (this: AxiosHeaders, value: string, name: string, headers: RawAxiosHeaders) => boolean);
set(headers?: RawAxiosHeaders | AxiosHeaders | string, rewrite?: boolean);
set(headers?: Iterable<[string, AxiosHeaderValue]>, rewrite?: boolean);

The rewrite argument controls the overwriting behavior:

  • false - do not overwrite if the header's value is set (is not undefined)
  • undefined (default) - overwrite the header unless its value is set to false
  • true - rewrite anyway

The option can also accept a user-defined function that determines whether to overwrite the value.

Empty or whitespace-only header names are ignored.

Iterable key/value pairs, such as a Map, are accepted:

const headers = new AxiosHeaders();

headers.set(
  new Map([
    ['X-Trace-Id', 'abc123'],
    ['Accept', 'application/json'],
  ])
);

Returns this.

AxiosHeaders#get(header)

  get(headerName: string, parser: typeof AxiosHeaders.parseParameters): AxiosHeaderParameters;
  get(headerName: string, parser: RegExp): RegExpExecArray | null;
  get(headerName: string, matcher?: true | AxiosHeaderParser): AxiosHeaderValue;

Returns the internal value of the header. It can take an extra argument to parse the header's value with RegExp.exec, matcher function or internal key-value parser.

const headers = new AxiosHeaders({
  'Content-Type': 'multipart/form-data; boundary=Asrf456BGe4h',
});

console.log(headers.get('Content-Type'));
// multipart/form-data; boundary=Asrf456BGe4h

console.log(headers.get('Content-Type', true)); // parse key-value pairs from a string separated with \s,;= delimiters:
// [Object: null prototype] {
//   'multipart/form-data': undefined,
//    boundary: 'Asrf456BGe4h'
// }

const quotedHeaders = new AxiosHeaders({
  'Content-Type': 'multipart/form-data; boundary="a,b"',
});

console.log({
  ...quotedHeaders.get('Content-Type', AxiosHeaders.parseParameters),
});
// { boundary: 'a,b' }

console.log(
  headers.get('Content-Type', (value, name, headers) => {
    return String(value).replace(/a/g, 'ZZZ');
  })
);
// multipZZZrt/form-dZZZtZZZ; boundZZZry=Asrf456BGe4h

console.log(headers.get('Content-Type', /boundary=(\w+)/)?.[0]);
// boundary=Asrf456BGe4h

AxiosHeaders.parseParameters is the opt-in parser for normalized HTTP parameter values. It returns a null-prototype map, matches parameter names case-insensitively, removes quoted-string delimiters, decodes escaped quotes and backslashes, preserves commas and semicolons inside quoted values, and removes only HTTP optional whitespace around unquoted values. It omits the unsafe keys __proto__, constructor, and prototype.

Passing true remains the legacy tokenizer and keeps its existing output for backward compatibility.

Returns the value of the header.

AxiosHeaders#has(header, matcher?)

has(header: string, matcher?: AxiosHeaderMatcher): boolean;

Returns true if the header is set (has no undefined value).

AxiosHeaders#delete(header, matcher?)

delete(header: string | string[], matcher?: AxiosHeaderMatcher): boolean;

Returns true if at least one header has been removed.

AxiosHeaders#clear(matcher?)

clear(matcher?: AxiosHeaderMatcher): boolean;

Removes all headers. Unlike the delete method matcher, this optional matcher matches the header name rather than the value.

const headers = new AxiosHeaders({
  foo: '1',
  'x-foo': '2',
  'x-bar': '3',
});

console.log(headers.clear(/^x-/)); // true

console.log(headers.toJSON()); // [Object: null prototype] { foo: '1' }

Returns true if at least one header has been cleared.

AxiosHeaders#normalize(format);

If the headers object was changed directly, it can have duplicates with the same name but in different cases. This method normalizes the headers object by combining duplicate keys into one. Axios uses this method internally after calling each interceptor. Set format to true for converting header names to lowercase and capitalizing the initial letters (cOntEnt-type => Content-Type)

const headers = new AxiosHeaders({
  foo: '1',
});

headers.Foo = '2';
headers.FOO = '3';

console.log(headers.toJSON()); // [Object: null prototype] { foo: '1', Foo: '2', FOO: '3' }
console.log(headers.normalize().toJSON()); // [Object: null prototype] { foo: '3' }
console.log(headers.normalize(true).toJSON()); // [Object: null prototype] { Foo: '3' }

Returns this.

AxiosHeaders#concat(...targets)

concat(...targets: Array<AxiosHeaders | RawAxiosHeaders | string | undefined | null>): AxiosHeaders;

Merges the instance with targets into a new AxiosHeaders instance. If the target is a string, Axios parses it as raw HTTP headers.

Returns a new AxiosHeaders instance.

AxiosHeaders#toJSON(asStrings?)

toJSON(asStrings: true): Record<string, string>;
toJSON(asStrings?: false): Record<string, string | string[]>;

Resolves all internal header values into a new null prototype object. Set asStrings to true to resolve arrays as a string containing all elements, separated by commas.

AxiosHeaders#toString()

toString(): string;

Returns the headers as a CRLF-free HTTP header block, one name: value pair per line.

AxiosHeaders.from(thing?)

from(thing?: AxiosHeaders | RawAxiosHeaders | string): AxiosHeaders;

Returns a new AxiosHeaders instance created from the raw headers passed in, or returns the given headers object if it's already an AxiosHeaders instance.

AxiosHeaders.concat(...targets)

concat(...targets: Array<AxiosHeaders | RawAxiosHeaders | string | undefined | null>): AxiosHeaders;

Returns a new AxiosHeaders instance created by merging the target objects.

Shortcuts

The following shortcuts are available:

  • setContentType, getContentType, hasContentType

  • setContentLength, getContentLength, hasContentLength

  • setAccept, getAccept, hasAccept

  • setUserAgent, getUserAgent, hasUserAgent

  • setContentEncoding, getContentEncoding, hasContentEncoding

Fetch adapter

Axios introduced the fetch adapter in v1.7.0. By default, Axios uses it when the xhr and http adapters are not available in the build or not supported by the environment. To use it by default, select it explicitly:

const { data } = axios.get(url, {
  adapter: 'fetch', // by default ['xhr', 'http', 'fetch']
});

You can create a separate instance for this:

const fetchAxios = axios.create({
  adapter: 'fetch',
});

const { data } = fetchAxios.get(url);

The adapter supports the same features as the xhr adapter, including upload and download progress capturing. It also supports response types such as stream and formdata when the environment supports them.

When auth is omitted, the fetch adapter can read HTTP Basic auth credentials from the request URL, for example https://user:[email protected]. Percent-encoded URL credentials are decoded before the Authorization header is generated, and auth takes precedence over URL-embedded credentials.

Custom fetch

Since v1.12.0, you can configure the fetch adapter to use a custom fetch API instead of environment globals. Pass a custom fetch function, Request, and Response constructors through env config. This helps in custom environments and app frameworks.

When using a custom fetch, you may also need to set custom Request and Response constructors. If you do not set them, Axios uses the global objects. If your custom fetch API does not provide these objects and the globals are incompatible with it, pass null to disable them inside the fetch adapter.

Note: Setting Request and Response to null prevents the fetch adapter from capturing upload and download progress.

Basic example:

import customFetchFunction from 'customFetchModule';

const instance = axios.create({
  adapter: 'fetch',
  onDownloadProgress(e) {
    console.log('downloadProgress', e);
  },
  env: {
    fetch: customFetchFunction,
    Request: null, // undefined -> use the global constructor
    Response: null,
  },
});

Using with Tauri

A minimal example of setting up Axios for use in a Tauri app with a platform fetch function that ignores CORS policy for requests.

import { fetch } from '@tauri-apps/plugin-http';
import axios from 'axios';

const instance = axios.create({
  adapter: 'fetch',
  onDownloadProgress(e) {
    console.log('downloadProgress', e);
  },
  env: {
    fetch,
  },
});

const { data } = await instance.get('https://google.com');

Using with SvelteKit

SvelteKit uses a custom fetch function for server rendering in load functions. It also uses relative paths, which are incompatible with the standard URL API. Configure Axios to use SvelteKit's custom fetch API:

export async function load({ fetch }) {
  const { data: post } = await axios.get('https://jsonplaceholder.typicode.com/posts/1', {
    adapter: 'fetch',
    env: {
      fetch,
      Request: null,
      Response: null,
    },
  });

  return { post };
}

HTTP/2 support

Axios supports HTTP/2 through the Node.js http adapter, introduced in v1.13.0.

Support depends on the runtime environment. Axios relies on Node.js APIs, so HTTP/2 works in supported Node.js versions but may not work in other environments such as Bun or Deno.

Options like httpVersion and http2Options are adapter-specific and may not behave the same way in every environment.

Note: HTTP/2 redirects are currently not supported by the HTTP/2 adapter.

const form = new FormData();

form.append('foo', '123');

const { data, headers, status } = await axios.post('https://httpbin.org/post', form, {
  onUploadProgress(e) {
    console.log('upload progress', e);
  },
  onDownloadProgress(e) {
    console.log('download progress', e);
  },
  responseType: 'arraybuffer',
});

Semver

Axios follows semver since v1.0.0.

Promises

axios depends on a native ES6 Promise implementation to be supported. If your environment doesn't support ES6 Promises, you can polyfill.

TypeScript

axios includes TypeScript definitions and a type guard for axios errors.

let user: User = null;
try {
  const { data } = await axios.get('/user?ID=12345');
  user = data.userDetails;
} catch (error) {
  if (axios.isAxiosError(error)) {
    handleAxiosError(error);
  } else {
    handleUnexpectedError(error);
  }
}

Use axios.isCancel<T>() to narrow cancellation errors to CanceledError<T>:

const controller = new AbortController();

try {
  await axios.get<User>('/user?ID=12345', { signal: controller.signal });
} catch (error) {
  if (axios.isCancel<User>(error)) {
    handleCancellation(error);
  }
}

Typing request data and query params

AxiosRequestConfig<D = any, P = any> uses D for the request body and P for query parameters. The params serializer receives the same P, and both types are preserved on the default response config:

import axios, {
  type AxiosPromise,
  type AxiosRequestConfig,
  type InternalAxiosRequestConfig,
} from 'axios';

interface RequestBody {
  includeArchived: boolean;
}

interface SearchParams {
  query: string;
  page?: number;
}

interface SearchResponse {
  results: string[];
}

const config: AxiosRequestConfig<RequestBody, SearchParams> = {
  data: { includeArchived: false },
  params: { query: 'axios', page: 1 },
  paramsSerializer: (params) => `${params.query}:${params.page ?? 1}`,
};

const response = await axios.get('/search', config);
response.config.data;   // RequestBody | undefined
response.config.params; // SearchParams | undefined

const invalidConfig: AxiosRequestConfig<RequestBody, SearchParams> = {
  // @ts-expect-error `query` must be a string
  params: { query: 123 },
};

const adapter = (
  adapterConfig: InternalAxiosRequestConfig<RequestBody, SearchParams>
): AxiosPromise<SearchResponse, RequestBody, SearchParams> =>
  Promise.resolve({
    data: { results: [] },
    status: 200,
    statusText: 'OK',
    headers: {},
    config: adapterConfig,
  });

The request methods place P last—<T, R, D, P>—so the existing response data (T), custom response (R), and request data (D) positions do not change. Explicit custom response types still determine the resolved value. RawAxiosRequestConfig, InternalAxiosRequestConfig, defaults, AxiosResponse, AxiosPromise<T, D, P>, AxiosError, CanceledError, callable instances, adapters, and mergeConfig() also carry P. The default is any for backward compatibility.

Use all three public error generics when cancellation narrowing must retain the response, request body, and params types:

declare const error: unknown;

if (axios.isCancel<SearchResponse, RequestBody, SearchParams>(error)) {
  error.config?.data;   // RequestBody | undefined
  error.config?.params; // SearchParams | undefined
}

Symbol-keyed custom request config

Own enumerable symbol properties survive config merging, so module-augmented options can reach request interceptors and adapters:

import axios from 'axios';

export const someFlag: unique symbol = Symbol(
  'some flag used in request interceptor'
);

declare module 'axios' {
  interface AxiosRequestConfig<D = any, P = any> {
    [someFlag]?: boolean;
  }
}

axios.interceptors.request.use((config) => {
  if (config[someFlag]) {
    config.headers.set('X-Some-Flag', 'enabled');
  }
  return config;
});

await axios.get('/users', { [someFlag]: true });

Only own enumerable symbol properties are copied; inherited and non-enumerable symbol properties are not.

Because axios publishes an ESM default export and a CJS module.exports, TypeScript has a few caveats. The recommended setting is "moduleResolution": "node16", which is implied by "module": "node16". This requires TypeScript 4.7 or greater. If you use ESM, your settings should be fine. If you compile TypeScript to CJS and can't use "moduleResolution": "node 16", enable esModuleInterop. If you use TypeScript to type check CJS JavaScript code, your only option is to use "moduleResolution": "node16".

You can also create a custom instance with typed interceptors:

import axios, { AxiosInstance, InternalAxiosRequestConfig } from 'axios';

const apiClient: AxiosInstance = axios.create({
  baseURL: 'https://api.example.com',
  timeout: 10000,
});

apiClient.interceptors.request.use((config: InternalAxiosRequestConfig) => {
  // Add auth token
  return config;
});

Online one-click setup

You can use Gitpod, a free online IDE for open source projects, to contribute or run the examples online.

Open in Gitpod

Contributing

Local setup

As a supply-chain hardening measure, this repository ships a project-level .npmrc that sets ignore-scripts=true. This blocks npm lifecycle scripts (preinstall, install, postinstall, prepare) from any direct or transitive dependency when you run npm install or npm ci inside the repo. See THREATMODEL.md (threat T-S2) for the rationale.

One consequence: the repository's own prepare hook (which installs Husky's git hooks) will not run automatically. After your first install, enable the git hooks manually:

npm ci
npm rebuild husky && npx husky

Run those two commands once per fresh checkout. You do not need to re-run them after every subsequent npm install.

Do not remove ignore-scripts=true from .npmrc to "fix" this. That reopens the lifecycle-script attack surface for every other package in the tree. All CI workflows already invoke npm with --ignore-scripts, so local behaviour matches CI.

Resources

Credits

axios is heavily inspired by the $http service in AngularJS. It provides a standalone $http-like service for use outside AngularJS.

License

License: MIT