cli-scraper is a Node web scraper library at its core, it tries to make it easy for you to scrape and consume static web pages in your terminal, if you're like me, living in the terminal world, it gives you another reason to stay. 😂 p.s. I know the name sounds bit strange, it does not scrape anything from cli, scraper-for-cli maybe better? Naming is hard!

This README is also available in: English, 中文.

To install cli-scraper globally on your machine, run:

$ yarn global add cli-scraper
# or "npm i -g cli-scraper"

Once cli-scraper is installed, you'll be able to interact with clis command in your terminal, to get cli-scraper to work, all you need to do is to:

1. generate a new configuration file with command $ clis init hello.js.
2. let cli-scraper know how to locate the content you'd like to extract by providing it with the css selectors.
3. lastly, run $ clis process hello.js.

Example - navigating to https://bing.com and scraping logo text:

Generate configuration file with init command by running $ clis init bing.js

// here's the completed configuration, copy paste into your bing.js, try it out.
module.exports = {
  url: 'https://www.bing.com/',     // target url
  process: function ({ $ }) {
    return $('.hp_sw_logo').text()  // grab target element via css selector, then get the text out of it
  },
  finally: function (res) {
    console.log(res   'go :)')      // what you want to do with the result
  }
}

Process the configuration with process command by running $ clis process bing.js

This is the bare minimum configuration you will need for scraping, you may notice there are more properties in the default configuration than the example above, I'll explain them in detail later, before we do that, I'd like to show you another example.

Let's set the stage first, you want to get a sneak peek at your favorite news site, not only the most recent news list, but also some content (maybe the publish date) within each news article, the final output would be something like news title [at] publish date

Example - get a sneak peek at your favorite news site:

Generate configuration file with init command by running $ clis init news.js

// here's the completed configuration, copy paste into your news.js, try it out.
module.exports = {
  url: 'http://www.news.cn/world/index.htm',
  requestOptions: {
    timeout: 10000                     // give up if request takes more than 10 seconds
  },
  randomUserAgent: true,               // set random user-agent header when requesting
  printRequestUrl: true,
  randomWait: 5,                       // wait randomly between 1 to 5 seconds before another request
  process: function ({ $ }) {
    return $('.firstPart ul.newList01 > li > a').map((i, el) => {
      return {
        articleUrl: $(el).prop('href'),
        title: $(el).text()
      }
    })
  },
  next: {
    key: 'articleUrl',                 // where the next to-be-processed article url is stored
    process: function ({ $, prevRes }) {
      return Object.assign(prevRes, {  // merge previous (list) result with new result (article page content)
        date: $('.h-news > .h-info > .h-time').text()
      })
    }
  },
  finally: function (res) {
    for (let item of res) {
      if (item.date) {
        console.log(`${item.title} [at] ${item.date}`)
      }
    }
  },
  catch: function (err) {
    console.log(err)                   // if there's an error, console log it out
  }
}

Process it with process command by running $ clis process news.js

This is the default configuration you'll get after running $ clis init yetAnotherConfig.js, I'll explain them one by one next.

module.exports = {
  url: '',
  urls: [],
  requestOptions: {
    timeout: 10000,
    gzip: true
  },
  beforeRequest: function () { return Promise.resolve({}) },
  afterProcessed: function (res) { return res },
  debugRequest: false,
  randomUserAgent: false,
  printRequestUrl: true,
  promiseLimit: 3,
  randomWait: 5,
  process: function ({ $, url, error, createdAt }) {
    if (error) throw Error(error)
    throw Error('Missing implementation')
  },
  next: {
    key: '',
    process: function ({ $, url, error, createdAt, prevRes }) {
      if (error) throw Error(error)
      throw Error('Missing implementation')
    }
  },
  finally: function (res, _) {
    throw Error('Missing implementation')
  },
  catch: function (err) {
    console.error(err)
  }
}

Required

• url: String target url you'd like to scrape.
• process: Function process function, it receives an object as argument, within that object, you have access to:
  • $: Function the scraped html data wrapped by cheerio, which allows you to do all kinds of extractions, please refer to cheerio's documentation for more information.
  • url: String target url.
  • error: String error message (if we encountered an error while scraping).
  • createdAt: String scraped datetime (ISO-8601).
  • _: Function the handy lodash function.
• finally: Function result handler function, it receives the processed result as well as the lodash function.
• catch: Function exception handler function, it's not required, however you should have it taken care of.

Next

• next: Object this is the most interesting feature cli-scraper offers, it has two properties:
  • key: String whatever key you defined in your outer process function's return object, which points to the next target url.
  • process: Function same as the outer process function, it receives one more property inside of the object called prevRes, which keeps hold of the previous processed result object, as what we did in the second example (merged news title with its publish date), you can do Object.assign(prevRes, { /* new result */ }) to merge the results.

Be a good bot respect Robots.txt

• promiseLimit: Number (default: 3) imaging you scraped a list with 10 items in the first process function, then move on to next cycle, which will handle those 10 urls in parallel, and this setting limits the parallel requests.
• randomWait: Number (default: 5) it waits randomly between 1 to 5 seconds before starting off another request.

Utilities

• debugRequest: Boolean (default: false) console log out request detail.
• printRequestUrl: Boolean (default: true) console log out target url before starting each scrape.
• randomUserAgent: Boolean (default: false) randomly (out of 5 most commonly used user-agent) set a user-agent in request header.

Go infinity and beyond

• urls: Array|Function yes, cli-scraper can work with more than one target urls, either an array or function which returns an array would be fine. Note, urls will be scraped sequentially.
• requestOptions: Object cli-scraper uses request library under the hood, hence it accepts pretty much all options that request offers, please take a look at request documentation for more information
• beforeRequest: Function before request hook async function, it will be triggered before every request, you can use it to for example set different proxy for sending out the request. Remember cli-scraper expects you to resolve an object out of it.
• afterProcessed: Function after request process hook function, it receives the processed result as the argument.

In order to debug your configuration (maybe the content you're expecting was not extracted), first, run $ which clis to determine where clis command is installed, you may get this as return /usr/local/bin/clis

Next you can start clis with devtool by running $ devtool /usr/local/bin/clis process bing.js, set a break point, then dive in.

Happy coding :)