A code snippet for scraping with Casper.js.

Web scraping with casper.js (built on Phantom.js, a headless webkit browser) is pretty simple. Casper provides a nice API allowing us to easily time and target tests. We just collect information from the DOM and save it to analyze later.

After installing Phantom.js, you can install Casper.js globally on your computer through npm. Just type npm install -g casperjs in your terminal.

Once Casper.js is installed, you can run a file through the command line with casperjs filename.js.

Let's say we want to scrape Twitter and list out who someone follows. Maybe we want to this to collect data in order to graph the levels of influence a certain person has. Below is some rough code illustrating for how we might approach this:


// Scrapes twitter users' follows. With great power comes great responsibility Runs on Casper.js (Phantom JS).

// Define a twitter ID to look up.
var twitterId = 'ChaptrApp';

// Define your own e-mail and password to get access to Twitter
var email = 'YourEmail';
var auth = 'YourPassword'

// Define a selector to look for before doing an action.
var findSelector = '.GridTimeline-items';

var urls = [];

var links = [];
var casper = require('casper').create();

function getLinks() {
    var links = document.querySelectorAll('div.ProfileCard a.ProfileCard-screennameLink');

    return Array.prototype.map.call(links, function(e) {

        // Return all of these links
        return e.getAttribute('href');
    });
}

// Start at twitter.com with the defined twitter id
casper.start('http://twitter.com/' + twitterId + '/following', function() {

      // Log the site title
      this.echo(this.getTitle());

      // Log the starting URL
      console.log('Starting location is ' + this.getCurrentUrl());
});

casper.then(function() {

      // Sign In To Twitter. Be careful not to abuse this or we run into the captcha.
      // TODO figure out how to run this overtime without getting caught.
      this.fillSelectors('form.signin', {
          'input[name="session[username_or_email]"]':    email,
          'input[name="session[password]"]':             auth
      }, true);
});

casper.then(function() {

    // Log the new URL
    console.log('Authentication ok, new location is ' + this.getCurrentUrl());

    // Log Error if we hit the captcha
    if (/captcha/.test(this.getCurrentUrl())) {
        console.log('Please login and confirm your captcha.');
    }
});

casper.waitForSelector('.GridTimeline-items', function() {

      // Log that we found the right selector to capture. This is looking for the follower grid on the mid 2014 Twitter Redesign.
      console.log('.GridTimeline-items' + ' is loaded.');
      
      // Screenshot
      this.captureSelector('timeline.png', 'div.GridTimeline-items');

      // Grab First Links
      this.emit('results.log');
});

function tryAndScroll(casper) {
    casper.waitFor(function() {
        this.page.scrollPosition = { top: this.page.scrollPosition["top"] + 3333, left: 0 };
        return true;
    }, function() {
        var info = this.getElementInfo('div.GridTimeline-footerIcon span.Icon--logo');

        // Recursively scroll until the bottom nav is visible.
        if (info["visible"] !== true) {
            this.waitWhileVisible('div.GridTimeline-footerIcon span.Icon--logo', function () {
                this.emit('results.loaded');
                this.emit('results.log');

            // We're still missing the last few rows of people.
            }, function () {
                this.echo('Next results not loaded');
            }, 5000);
        }
    }, function() {
        this.echo("Scrolling failed. Sorry.").exit();
    }, 500);
}

casper.on('results.loaded', function () {

    this.echo('Scrollin');

    // Call Scroll Function 
    tryAndScroll(this);
});

casper.on('results.log', function () {
    // Grab Additional Links
    links = this.evaluate(getLinks);
});

casper.then(function() {
    tryAndScroll(this);
});

// Snippet from: https://www.andykelk.net/tech/web-scraping-with-casperjs
casper.run(function() {

    // Run through our array to format the JSON for D3
    for (i in links) {
      links[i] = "{'name':" + "'" + links[i] + "'" + ", 'group': 2}";
    }

    nodeLinks = [];

    var min;

    for (i in links) {
      // A crude buffer
      min = Math.min((parseInt(i) * 10), 450);

      nodeLinks.push("{'source':" + (parseInt(i) + 1) + "," + "'target':0,'value':" + min + "}");
    }

    // Format it
    var re = /\//g;
    var name = links.join(' ');
 
    // Change to a string
    var jsonString = JSON.stringify(links);

    // Do some gross replaces just for D3
    var finalJson = ("\"nodes\" :[ " + "{'name': '" + twitterId + "', " + "'group':1}\", " + jsonString.replace(re," ") + " ]");

    finalJson = finalJson.replace(/"\{/g,"{");
    finalJson = finalJson.replace(/\}"/g,"}");

    finalJson = finalJson.replace(/\[\{/g,"{");
    finalJson = finalJson.replace(/\}\]/g,"}");

    finalJson = "{ " + finalJson + ", " + "\n \n'links':[" + nodeLinks.toString() + "]" + " }"; 

    finalJson = finalJson.replace(/'/g,"\"");

    console.log(finalJson);

    // Export as JSON
    require('fs').write("finalJson.json", finalJson, 'w');

    // Close Casper
    this.exit();
});

From here you could use the JSON to visualize something like small designer communities on Twitter – based on the assumption that people follow who they look up to and occasionally prune their follower list. You could recursively go through each of the followers from the original user of interest and then map out overlap based off who they follow.

Here, a super simple example just shows who the target user has followed. The more recent ones are are closer to the center of the circle (They're not proportional – the code above doesn't have access to any sort of time through an API).

7th Gear Animated Title