kbbi_grabber.php

<?php

/**
 * @author Ammar Faizi <[email protected]>
 *
 * KBBI Grabber.
 */

define("PDO_CONNECT", "mysql:host=localhost;dbname=kamus_bindo;");
define("PDO_USER", "icetea");
define("PDO_PASS", "");


$pdo = new PDO(PDO_CONNECT, PDO_USER, PDO_PASS);
$pdo->exec("DROP TABLE IF EXISTS `kb_kamus`");
$pdo->exec("CREATE TABLE `kb_kamus` (`id` int(11) NOT NULL AUTO_INCREMENT,`kata` varchar(225) NOT NULL,PRIMARY KEY (`id`)) ENGINE=InnoDB AUTO_INCREMENT=0 DEFAULT CHARSET=latin1;");
$st = $pdo->prepare("INSERT INTO `kb_kamus` (id, kata) VALUES (:id, :kata);");

$pg = 1;
do {
    $a = curl("http://www.kbbi.co.id/daftar-kata?page=".$pg++);
    $a = explode('<div class="row"><div class="col-md-2 col-sm-3 col-xs-4">', $a, 2);
    if (isset($a[1])) {
    $a = explode('</nav>', $a[1])[0];
        $a = explode('<nav', $a)[0];
        $a = explode("<li>", $a);
        $ct = count($a);
        for ($i=1; $i < $ct; $i++) {
            $w = preg_match_all("#<a href=\"(.+)\">(.*)</a></li>#", $a[$i], $n);
            print $n[2][0]." | ".($st->execute([
                "id" => null,
                "kata" => trim(strtolower(html_entity_decode($n[2][0], ENT_QUOTES, 'UTF-8')), "-")
            ])?"true":"false")."\n";
        }
    } else {
        // Finished.
        break;
    }
} while (true);


// Simple curl.
function curl($url)
{
    $ch = curl_init($url);
    curl_setopt_array($ch, [
            CURLOPT_RETURNTRANSFER => true
        ]);
    $out = curl_exec($ch);
    curl_close($ch);
    return $out;
}