Twitter Archive Analyzer

analyze-twitter-archive.php

<?php

// Twitter Archive Analyzer
// @tmdpw, 2013-01-18
// https://github.com/timdp

// Instructions:
// 1. Request your Twitter archive from twitter.com.
// 2. Extract the zip file you obtained.
// 3. Run this script with the extraction path as its argument.

error_reporting(E_ALL ^ E_NOTICE);

if (count($argv) != 2) {
    die("Usage: php $argv[0] PATH");
}
$base_dir = $argv[1];

$mentions_freq = array();
$hashtag_freq = array();
$word_freq = array();

$files = scandir($base_dir);
foreach ($files as $index => $file) {
    if (!preg_match('/^([0-9]{4})_([0-9]{2})\.js$/', $file, $matches)) {
        continue;
    }
    printf("[%" . len(count($files)) . "d/%d] Loading %s ..." . PHP_EOL,
        $index + 1, count($files), $file);
    $year = $matches[1];
    $month = @intval($matches[2]);
    $contents = file_get_contents("$base_dir/$file");
    $contents = preg_replace('/^.*?=\s*/s', '', $contents);
    $data = json_decode($contents);
    foreach ($data as $tweet) {
        foreach ($tweet->entities->hashtags as $ht) {
            $hashtag_freq[lower($ht->text)]++;
        }
        foreach ($tweet->entities->user_mentions as $m) {
            $mentions_freq[lower($m->screen_name)]++;
        }
        preg_match_all('/\b(\S+)\b/', $tweet->text, $matches);
        foreach ($matches[1] as $word) {
            $word = lower($word);
            if (!array_key_exists($word, $hashtag_freq)
                    && !array_key_exists($word, $mentions_freq)) {
                $word_freq[$word]++;
            }
        }
    }
}

echo PHP_EOL;

print_top('Hashtags', $hashtag_freq, 10);
print_top('Mentions', $mentions_freq, 10);
print_top('Words', $word_freq, 10);

function print_top($what, &$data, $max) {
    arsort($data, SORT_NUMERIC);
    array_splice($data, $max);
    print_title($what, 78);
    $i = 0;
    foreach ($data as $item => $count) {
        printf('%5s. %-60s %10s' . PHP_EOL,
            number_format(++$i), $item, number_format($count));
    }
    echo PHP_EOL;
}

function print_title($title, $width) {
    $line = str_repeat('=', $width);
    echo $line, PHP_EOL;
    printf('%' . ($width - len($title)) / 2 . 's', '');
    echo $title, PHP_EOL, $line, PHP_EOL, PHP_EOL;
}

function lower($str) {
    return extension_loaded('mbstring')
        ? mb_strtolower($str, 'UTF-8')
        : strtolower($str);
}

function len($str) {
    return extension_loaded('mbstring')
        ? mb_strlen($str, 'UTF-8')
        : strlen($str);
}