Scan your Twitter archive to get a basic count of the people you've talked to and the words that you've used.

tweet-counts.js

/*
 * Scan your official Twitter archive to get a basic count of the
 * people you've talked to and the words that you've used.
 *
 * $ node tweet-counts.js ~/path/to/tweets-archive/
 * 
 */
var fs = require('fs');
var base = process.argv[2].replace(/\/$/, ''); // Strip trailing slash.
var Grailbird = { data: {} }; // The Twitter archive uses this.

var statistics = {
  total: 0,
  text: {},
  mentions: {},
  hashtags: {},
  sources: {},
  hourly: {},
  daily: {},
}

function addOne(statistic, key) {
  statistic[key] = (statistic[key] || 0) + 1;
}

// Gross eval... load up the index.
eval(fs.readFileSync(base + '/data/js/tweet_index.js').toString());
tweet_index.forEach(function(tweet_file) {
  // Gross eval... load up a monthly tweet archive.
  eval(fs.readFileSync(base + '/' + tweet_file.file_name).toString());
  Grailbird.data[tweet_file.var_name].forEach(function(tweet) {
    // Total.
    statistics.total++;
    // Text.
    tweet.text.split(/\s+/).forEach(function(word) {
      addOne(statistics.text, word.toLowerCase());
    });
    // Mentions.
    tweet.entities.user_mentions.forEach(function(user) {
      addOne(statistics.mentions, user.screen_name.toLowerCase());
    });
    // Hashtags.
    tweet.entities.hashtags.forEach(function(hashtag) {
      addOne(statistics.hashtags, hashtag.text.toLowerCase());
    });
    // Sources.
    addOne(statistics.sources, tweet.source.toLowerCase());
    // Hourly and Daily.
    var created_at = new Date(tweet.created_at);
    addOne(statistics.hourly, created_at.getHours());
    addOne(statistics.daily, created_at.getDay());
  });
});

function printSortedByValues(statistic, threshold) {
  Object.keys(statistic).sort(function(a, b) { return statistic[a] - statistic[b]; }).forEach(function(key) {
    if (statistic[key] > threshold) {
      console.log(statistic[key] + ' ' + (statistic[key] * 100 / statistics.total).toPrecision(2) + '% ' + key);
    }
  });
}

function printSortedByKeys(statistic) {
  Object.keys(statistic).forEach(function(key) {
    console.log(key + ' ' + statistic[key] + ' ' + (statistic[key] * 100 / statistics.total).toPrecision(2) + '%');
  });
}

function printDays(statistic) {
  var days = ['Sunday','Monday','Tuesday','Wednesday','Thursday','Friday','Saturday'];
  for (var day = 0; day < days.length; day++) {
    console.log(statistic[day] + ' ' + (statistic[day] * 100 / statistics.total).toPrecision(2) + '% ' + days[day]);
  }
}

console.log('Popular words:');
printSortedByValues(statistics.text, 10);

console.log('');
console.log('Who you talk to:');
printSortedByValues(statistics.mentions, 2);

console.log('');
console.log('Hashtags used:');
printSortedByValues(statistics.hashtags, 1);

console.log('');
console.log('Tweet sources:');
printSortedByValues(statistics.sources, 0);

console.log('');
console.log('What time of the day:');
printSortedByKeys(statistics.hourly);

console.log('');
console.log('What day of the week:');
printDays(statistics.daily);