Join Us

We welcome all pull requests. Suggestions and feedback are also welcomed here (opens new window).

# Join the discussion

  1. Telegram Group (opens new window)
  2. GitHub Issues (opens new window)

# Submit new RSS rule

Before you start writing an RSS rule, please make sure that the source site does not provide RSS. Some web pages will include a link element with type application/atom+xml or application/rss+xml in the HTML header to indicate the RSS link.

# Debug

First, yarn or npm install to install dependencies, then execute yarn dev or npm run dev, open http://localhost:1200 to see the effect, and the page will refresh automatically if files are modified.

# Add route

Firstly, add a route in /lib/router.js (opens new window)

# Code the script

Create a new js script in /lib/routes/ (opens new window) corresponding to route path

# Acquiring Data

  • Typically, the data are acquired via HTTP requests (via API or webpage) sent by got (opens new window)

  • Occasionally puppeteer (opens new window) is required for browser stimulation and page rendering in order to acquire the data

  • The acquired data are most likely in JSON or HTML format

  • For HTML format, cheerio (opens new window) is used for further processing

  • Below is a list of data acquisition methods, ordered by the 「level of recommendation」

    1. Acquire data via API using got

    Example: /lib/routes/bilibili/coin.js (opens new window)

    Acquiring data via the official API provided by the data source using got:

    // Initiate a HTTP GET request
    const response = await got({
        method: 'get',
        url: `${uid}&jsonp=jsonp`,
        headers: {
            Referer: `${uid}/`,
    const data =; // is the data object returned from the previous request
    // The object contains a nested object called data, thus is the actual data needed here

    One of the leaf objects ([0]):

        "aid": 33614333,
        "videos": 2,
        "tid": 20,
        "tname": "宅舞",
        "copyright": 1,
        "pic": "",
        "title": "【赤九玖】响喜乱舞【和我一起狂舞吧,团长大人(✧◡✧)】",
        "pubdate": 1539259203,
        "ctime": 1539249536,
        "desc": "编舞出处:av31984673\n真心好喜欢这个舞和这首歌,居然恰巧被邀请跳了,感谢《苍之纪元》官方的邀请。这次cos的是游戏的新角色缪斯。然而时间有限很多地方还有很多不足。也没跳够,以后私下还会继续练习,希望能学到更多动作,也能为了有机会把它跳的更好。 \n摄影:绯山圣瞳九命猫 \n后期:炉火"
        // some more data....

    Processing the data further to generate objects in accordance with RSS specification, mainly title, link, description, publish time, then assign them to, produce RSS feed: = {
        // the source title
        title: `${name} 的 bilibili 投币视频`,
        // the source link
        link: `${uid}`,
        // the source description
        description: `${name} 的 bilibili 投币视频`,
        // iterate through all leaf objects
        item: => ({
            // the article title
            title: item.title,
            // the article content
            description: `${item.desc}<br><img src="${item.pic}">`,
            // the article publish time
            pubDate: new Date(item.time * 1000).toUTCString(),
            // the article link
            link: `${item.aid}`,
    // the route is now done
    1. Acquire data via HTML webpage using got

    Data have to be acquired via HTML webpage if no API was provided, for example: /lib/routes/douban/explore.js (opens new window)

    Acquiring data by scrapping the HTML using got:

    // Initiate a HTTP GET request
    const response = await got({
        method: 'get',
        url: '',
    const data =; // is the entire HTML source of the target page, returned from the previous request

    Parsing the HTML using cheerio:

    const $ = cheerio.load(data); // Load the HTML returned into cheerio
    const list = $('div[data-item_id]');
    // use cheerio selector, select all 'div' elements with 'data-item_id' attribute, the result is an array of cheerio node objects
    // use cheerio get() method to transform a cheerio node object array into a node array
    // PS: every cheerio node is a HTML DOM
    // PPS: cheerio selector is almost identical to jquery selector
    // Refer to cheerio docs:

    Use map to traverse the array and parse out the result of each item = {
        title: '豆瓣-浏览发现',
        link: '',
            list &&
                .map((index, item) => {
                    item = $(item);
                    itemPicUrl = item.find('a.cover').attr('style').replace('background-image:url(', '').replace(')', '');
                    return {
                        title: item.find('.title a').first().text(),
                        description: `作者:${item.find('.usr-pic a').last().text()}<br>描述:${item.find('.content p').text()}<br><img src="${itemPicUrl}">`,
                        link: item.find('.title a').attr('href'),
    // the route is now done
    1. Acquire data via page rendering using puppeteer


    This method consumes more resources and is less performant, use only when the above methods are failed to acquire data. Otherwise, your pull requests will be rejected!

    Seldomly, data source provides no API and the page requires rendering to acquire data, for example: /lib/routes/sspai/series.js (opens new window)

    // use puppeteer util class, initialise a browser instance
    const browser = await require('@/utils/puppeteer')();
    // open a new page
    const page = await browser.newPage();
    // access the target link
    const link = '';
    await page.goto(link);
    // render the page
    const html = await page.evaluate(
        () =>
            // process on the rendered page
    // shutdown the browser

    Parsing the HTML using cheerio:

    const $ = cheerio.load(html); // Load the HTML returned into cheerio
    const list = $('div.item'); //  // use cheerio selector, select all 'div class="item"' elements, the result is an array of cheerio node objects

    Assign the value to = {
        title: '少数派 -- 最新上架付费专栏',
        description: '少数派 -- 最新上架付费专栏',
        item: list
            .map((i, item) => ({
                // the article title
                title: $(item).find('.item-title a').text().trim(),
                // the article link
                link: url.resolve(link, $(item).find('.item-title a').attr('href')),
                // the article author
                author: $(item).find('.item-author').text().trim(),
            .get(), // use cheerio get() method to transform a cheerio node object array into a node array
    // the route is now done
    // PS: the route acts as a notifier of new articles. It does not provide access to the content behind the paywall. Thus no content was fetched
    1. Use general configuration routing

    A large number of websites can generate RSS through a configuration paradigm.

    The general configuration is to generate RSS with ease by reading json data through cheerio (CSS selector, jQuery function)

    First, we need a few data:

    1. RSS source link
    2. Data source link
    3. RSS title (not item title)
    const buildData = require('@/utils/common-config');
    module.exports = async (ctx) => { = await buildData({
            link: '', // RSS source link
            url: '', // Data source link
            title: '%title%', // Variables are used here, such as **% xxx%** will be parsed into variables with values of the same name under **params**
            params: {
                title: '', // RSS title

    Our RSS does not have any content for now. The content needs to be completed by item. Here is an example

    const buildData = require('@/utils/common-config');
    module.exports = async (ctx) => {
        const link = ``; = await buildData({
            url: link,
            title: `%title%`,
            params: {
                title: '裏垢女子まとめ',
            item: {
                item: '.content-main .stream .stream-item',
                title: `$('.post-account-group').text() + ' - %title%'`, // Only supports js statements like $().xxx()
                link: `$('.post-account-group').attr('href')`, // .text() means get the text of the element, .attr() means get the specified attribute
                description: `$('.post .context').html()`, // .text() means get the text of the the html code
                pubDate: `new Date($('.post-time').attr('datetime')).toUTCString()`,
                guid: `new Date($('.post-time').attr('datetime')).getTime()`,

    So far we have completed a simplest route

# Use Cache

All routes have a cache, the global cache time is set in lib/config.js, but the content returned by some interfaces is updated less frequently. For that, you should set a longer cache time for these data like the full text of an article.

For example, the bilibili column needs to get the full text of the article: /lib/routes/bilibili/followings_article.js (opens new window)

Since the full text of all articles cannot be got from one API, each article needs to be requested once, and these data are generally unchanged, so these data should be stored in the cache to avoid requesting repeatedly

const description = await ctx.cache.tryGet(link, async () => {
    const result = await got.get(link);

    const $ = cheerio.load(;
    $('img').each(function (i, e) {
        $(e).attr('src', $(e).attr('data-src'));

    return $('.article-holder').html();

The implementation of tryGet can be seen here (opens new window). The 1st parameter is the cache key; the 2nd parameter is the cache data acquisition method (executed when cache miss); the 3rd parameter is the cache time, it should not be passed in normally and defaults to CACHE_CONTENT_EXPIRE; the 4th parameter determines whether to recalculate the expiration time ("renew" the cache) when the current attempt cache hits, true is on, false is off, default is on

# Produce RSS Feed

Assign the acquired data to, the middleware template.js (opens new window) will then process the data and render the RSS output /views/ (opens new window), the list of parameters: = {
    title: '', // The feed title
    link: '', // The feed link
    description: '', // The feed description
    language: '', // The language of the channel
    allowEmpty: false, // default to false, set to true to allow empty item
    item: [
        // An article of the feed
            title: '', //  The article title
            author: '', // Author of the article
            category: '', // Article category
            // category: [''], // Multiple category
            description: '', // The article summary or content
            pubDate: '', // The article publishing datetime
            guid: '', // The article unique identifier, optional, default to the article link below
            link: '', // The article link


title, subtitle (only for atom), author (only for atom), item.title, and should not contain linebreaks, consecutive white spaces, or start/end with white space(s).
Most RSS readers will automatically trim them, so they make no sense. However, some readers may not process them properly, so we will trim them before outputting to ensure these fields contain no linebreaks, consecutive white spaces, or start/end with white space(s).
If the route you are writing can not tolerate these trimmings, you should consider change the format of these fields.

In addition, although other fields will not be forced trimmed, you should also try to avoid violations of the above rules. Especially when using Cheerio to extract web pages, you need to keep in mind that Cheerio will retain wraps and indentation. In particular, for item.description, any intended linebreaks should be converted to <br>, otherwise the RSS reader is likely to trim them; especially if you extract the RSS feed from JSON, the JSON returned by the source website is very likely to contain linebreaks that need to be displayed, so it must be converted in this case.

# Podcast feed

Used for audio feed, these additional data are in accordance with many podcast players' subscription format: = {
    itunes_author: '', // The channel's author, you must fill this data.
    itunes_category: '', // Channel category
    image: '', // Channel's image
    item: [
            itunes_item_image: '', // The item image
            itunes_duration: '', // The audio length in seconds (or H:mm:ss), optional
            enclosure_url: '', // The item's audio link
            enclosure_length: '', // The file size in Bytes, optional
            enclosure_type: '', // Common types are: 'audio/mpeg' for .mp3, 'audio/x-m4a' for .m4a 'video/mp4' for .mp4
# BT/Magnet feed

Used for downloader feed, these additional data are in accordance with many downloaders' subscription format to trigger automated download: = {
    item: [
            enclosure_url: '', // Magnet URI
            enclosure_length: '', // The file size in Bytes, optional
            enclosure_type: 'application/x-bittorrent', // Fixed to 'application/x-bittorrent'
# Media RSS

These additional data are in accordance with many Media RSS (opens new window) softwares' subscription format:

For example: = {
    item: [
            media: {
                content: {
                    url: post.file_url,
                    type: `image/${mime[post.file_ext]}`,
                thumbnail: {
                    url: post.preview_url,
# Interactions

These additional data are in accordance with some softwares' subscription format: = {
    item: [
            upvotes: 0, // default to undefined, how many upvotes for this article,
            downvotes: 0, // default to undefined, how many downvotes for this article,
            comments: 0, // default to undefined, how many comments for this article

# Add the documentation

  1. Update the corresponding file in documentation (/docs/en/) (opens new window) directory, preview the docs via npm run docs:dev

    • Documentation uses vue component:

      • author: route authors, separated by a single space
      • example: route example
      • path: route path
      • :paramsDesc: route parameters description, in array, supports markdown
        1. parameter description must be in the order of its appearance in route path
        2. missing description will cause errors in npm run docs:dev
        3. ' " must be escaped as \' \"
        4. route parameters ending with ?, *, +, and a word represent optional, zero or more, one or more, and mandatory respectively. They are automatically determined by Vue component and do not need to be explicitly mentioned in the description
    • Documentation examples:

      1. No parameter:
      <RouteEn author="HenryQW" example="/sspai/series" path="/sspai/series" />


      Author: @HenryQW


      Route: /sspai/series

      Parameters: N/A

      1. Multiple parameters:
      <RouteEn author="HenryQW" example="/github/issue/DIYgod/RSSHub" path="/github/issue/:user/:repo" :paramsDesc="['GitHub username', 'GitHub repo name']" />


      Author: @HenryQW


      Route: /github/issue/:user/:repo


      • user, required - GitHub username
      • repo, required - GitHub repo name

      1. Use component slot for complicated description:
      <RouteEn author="DIYgod" example="/juejin/category/frontend" path="/juejin/category/:category" :paramsDesc="['分类名']">
      | 前端     | Android | iOS | 后端    | 设计   | 产品    | 工具资源 | 阅读    | 人工智能 |
      | -------- | ------- | --- | ------- | ------ | ------- | -------- | ------- | -------- |
      | frontend | android | ios | backend | design | product | freebie  | article | ai       |


      Author: @DIYgod


      Route: /juejin/category/:category


      • category, required - 分类名
      前端 Android iOS 后端 设计 产品 工具资源 阅读 人工智能
      frontend android ios backend design product freebie article ai

  2. Please be sure to close the tag of <Route>!

  3. Execute npm run format to lint the code before you commit and open a pull request

# Submit new RSSHub Radar rule

# Debug

Open browser's RSSHub Radar extension settings, switch to rules list, scroll down and you will see a text box, copy the new rules into the text box and start debugging

# Code the rule

Add rules to /assets/radar-rules.js (opens new window)

Simplified rules will be used for the following illustration:

    '': {
        _name: 'bilibili',
        www: [{
            title: '分区视频',
            docs: '',
            source: '/v/*tpath',
            target: (params) => {
                let tid;
                switch (params.tpath) {
                    case 'douga/mad':
                        tid = '24';
                        return false;
                return `/bilibili/partion/${tid}`;
    '': {
        _name: 'Twitter',
        '.': [{  // for
            title: 'User timeline',
            docs: '',
            source: '/:id',
            target: (params) => {
                if ( !== 'home') {
                    return '/twitter/user/:id';
    '': {
        _name: 'Pixiv',
        'www': [{
            title: 'User Bookmark',
            docs: '',
            source: '/bookmark.php',
            target: (params, url) => `/pixiv/user/bookmarks/${new URL(url).searchParams.get('id')}`,
    '': {
        _name: '微博',
        '.': [{
            title: '博主',
            docs: '',
            source: ['/u/:id', '/:id'],
            target: (params, url, document) => {
                const uid = document && document.documentElement.innerHTML.match(/\$CONFIG\['oid']='(\d+)'/)[1];
                return uid ? `/weibo/user/${uid}` : '';

The definition and usage of these fields are explained below:

# title

Required, route name

The corresponding name in RSSHub docs, e.g. the title for Twitter User timeline is User timeline

# docs

Required, docs link

e.g. the docs for Twitter User timeline is

Note that it is not, hash should be positioned to the H1 heading

# source

Optional, source path, leave it blank and it will never match, only appears in RSSHub for current website

e.g. the source for Twitter User timeline is /:id

Let's say we are in, which, the resulting params is {id: 'DIYgod'}, the extension will then generate an RSSHub subscription address based on the params target

Please note that source can only match URL Path, if the parameters are in URL Param and URL Hash please use target

# target

Optional, RSSHub path, leave it blank to not generate an RSSHub path

The corresponding path in RSSHub docs, e.g. the target for Twitter User timeline is /twitter/user/:id

The source path in the previous step matches id with DIYgod, the :id in RSSHub path will be replaced with DIYgod, resulting in /twitter/user/DIYgod, which is the result we want

Furthermore, if the source path does not match the desired parameters, we can use target as a function with params, url and document parameters

params are the parameters matched by source in the previous step, url is the page url, document is the page document

Note that the target method runs in a sandbox, any changes to document will not be reflected in the web page

# RSSBud

RSSBud (opens new window) supports RSSHub Radar rules and will also be updated automatically, but please note that:

  • Using '.' subdomain allows RSSBud to support common mobile domains such as m / mobile

  • Using document in target does not apply to RSSBud: RSSBud is not a browser extension, it only fetches and analyzes the URL of a website

# Additional Information

Adding radar="1" in the RSSHub docs will show a support browser extension badge

If RSSBud is also supported, adding rssbud="1" will show a support rssbud badge