twitter-scrape-account-stats/lib/index.js

scrape public Twitter data w/out API access

// Generated by CoffeeScript 1.10.0
(function() {
  var STAT_NAME_MAP, STAT_TITLE_RE, cheerio, getAccountStats, getStatInt, request;

  request = require('request-promise');

  cheerio = require('cheerio');

  STAT_NAME_MAP = {
    following: 'following_stats',
    followers: 'follower_stats',
    posts: 'tweet_stats'
  };

  STAT_TITLE_RE = {
    'following_stats': /^[0-9,]+ Following$/,
    'follower_stats': /^[0-9,]+ Followers$/,
    'tweet_stats': /^[0-9,]+ Tweets$/
  };

  getStatInt = function($, statName) {
    var stat;
    stat = $("[data-element-term=\"" + statName + "\"]").attr('title');
    if (stat == null) {
      throw new Error("couldn't get " + statName);
    }
    if (!STAT_TITLE_RE[statName].test(stat)) {
      throw new Error('unexpected format in title attribute');
    }
    return parseInt(stat.replace(/[^0-9]/g, ''));
  };

  getAccountStats = function(arg) {
    var data, queryString, userId, username;
    username = arg.username, userId = arg.userId;
    if ((username == null) && (userId == null)) {
      throw new Error('A username or userId needs to be passed');
    }
    queryString = {
      'wants_hovercard': true,
      '_': (new Date()).getTime()
    };
    if (userId != null) {
      queryString['user_id'] = userId;
    } else {
      queryString['screen_name'] = username;
    }
    data = void 0;
    return request.get({
      uri: 'https://twitter.com/i/profiles/popup',
      qs: queryString
    }).then(function(resp) {
      data = JSON.parse(resp);
      return cheerio.load(data.html);
    }).then(function($) {
      var e, error, field, res, statName;
      res = {
        description: $('.bio.profile-field').text(),
        isVerified: $('.Icon--verified').length > 0,
        name: $('.ProfileCard-avatarLink').attr('title'),
        userId: data['user_id'],
        username: data['screen_name']
      };
      for (field in STAT_NAME_MAP) {
        statName = STAT_NAME_MAP[field];
        try {
          res[field] = getStatInt($, statName);
        } catch (error) {
          e = error;
          throw new Error("Couldn't get " + field + " stat for " + res.username);
        }
      }
      return res;
    });
  };

  module.exports = {
    getAccountStats: getAccountStats
  };

}).call(this);