parse.php

<?php

/**
 * PHP Version 7
 * Melakukan crawling dan parsing data berita.
 * @category Berita
 * @package  Komunitas-PMO
 * @author   Putu Arya Pradipta <13515017@std.stei.itb.ac.id>
 * @license  http://www.gnu.org/copyleft/lesser.html  LGPL License 2.1
 * @link     http://pear.php.net/package/komunitas-pmo
 */

require 'dbConnect.php';
require 'sendNotif.php';

$url = 'http://pmo.itb.ac.id/feed/';
// header('Content-Type: text/html; charset=utf-8');

$rss = new DOMDocument('1.0', 'UTF-8');
$rss->load($url);
$feed = array();

$rssLastBuild =  $rss->getElementsByTagName('lastBuildDate')->item(0)->nodeValue;
// cek apakah last build lebih baru
$lastBuild = file_get_contents("lastNewsUpdate.txt");
if (strtotime($lastBuild) < strtotime($rssLastBuild)) {
    foreach ($rss->getElementsByTagName('item') as $node) {
        // ambil judul
        $judul = htmlspecialchars(
            $node->getElementsByTagName('title')->item(0)->nodeValue
        );
        
        // cek apakah judul ada di dalam database
        $sql = "SELECT * FROM `berita` WHERE `judul` like '%" . $judul . "%'";
        $stmt = $conn->query($sql);
        $numRows = $stmt->num_rows;
        $stmt->close();
        if ($numRows == 0) {
            $stmt2 = $conn->prepare(
                "INSERT INTO berita(judul, tgl, konten, foto) VALUES (?,?,?,?)"
            );
            echo "<br> <h1> " . $judul . "</h1>";
            $tgl = $node->getElementsByTagName('pubDate')->item(0)->nodeValue;
            $tgl = date("Y-m-d", strtotime($tgl));
            
            $konten = $node->getElementsByTagName('encoded')->item(0)->nodeValue;
            
            $dom = new DOMDocument();

            // suppress error
            libxml_use_internal_errors(true);
            $dom->loadHTML('<?xml encoding="utf-8" ?>' . $konten);
            libxml_clear_errors();
            
            $imgarray = $dom->getElementsByTagName('img');
            
            $allData = '';
            // ambil gambar pertama untuk thumbnail
            // $thumbnail = null;
            // if ($imgarray->length > 0) {
            //   $thumbnail = $imgarray->item(0)->getAttribute("src");
            //   print_r($thumbnail);
            // }
            $arrayimage = array();
            while ($imgarray->length > 0) {
                $img = $imgarray->item(0);
                // $img->setAttribute("style","display: block; width: 90%");
                array_push($arrayimage, $img->getAttribute("src"));
                // $allData .= $img->ownerDocument->saveHTML($img);
                $img->parentNode->removeChild($img);
            }
            $arrayimage = json_encode($arrayimage);
            // $parray = $dom->getElementsByTagName('p');
            // while ($parray->length > 0) {
            //   $p = $parray->item(0);
            //   // $p->setAttribute("style","display: block; width: 90%");
            //   $allData .= $p->ownerDocument->saveHTML($p);
            //   $p->parentNode->removeChild($p);
            // }
            $allData .= $dom->saveHTML();
            print_r($allData);
            $stmt2->bind_param("ssss", $judul, $tgl, $allData, $arrayimage);
            $stmt2->execute();
            $stmt2->close();
        }
    }
    // update
    file_put_contents("lastNewsUpdate.txt", $rssLastBuild);
    \pmotraining\FirebaseNotif::sendNotif("Ada berita baru", "Klik untuk membuka", 1, 0);
}
// closing
$conn->close();