padelgids.php

<?php

use GuzzleHttp\Client; // guzzlehttp/guzzle
use rdx\jsdom\Node; // rdx/js-dom

require 'vendor/autoload.php';

$guzzle = new Client();

if (file_exists($file = 'cache/padelgids.html')) {
	$html = file_get_contents($file);
}
else {
	$rsp = $guzzle->get('https://padelgids.nl/booking/knltb-meet-play/?type=&display=table');
	$html = (string) $rsp->getBody();
	file_put_contents($file, $html);
}

$dom = Node::create($html, 'utf-8');

$links = $dom->queryAll('tr div + a');
$total = count($links);
var_dump($total);

$done = 0;
foreach ($links as $link) {
	$path = explode('?', $link['href'])[0];
	$url = 'https://padelgids.nl' . $path;
	echo $url, "...\n";

	$file = preg_replace('#[^0-9a-z-]+#i', '-', $path);
	$file = preg_replace('#\-+#', '-', $file);
	$file = 'cache/' . trim($file, '-') . '.html';
	// echo $file, "\n";

	if (file_exists($file)) {
		$html = file_get_contents($file);
	}
	else {
		if ($done > 0) {
			usleep(rand(700, 1400) * 1000);
		}

		$rsp = $guzzle->get($url);
		$html = (string) $rsp->getBody();
		file_put_contents($file, $html);
	}

	$dom = Node::create($html, 'utf-8');

	$h1 = $dom->query('h1');
	// var_dump($h1->textContent);
	$mail = $dom->query('.social-link a[href^="mailto:"]');
	// var_dump($mail->textContent);
	$phone = $dom->query('.social-link a[href^="tel:"]');
	// var_dump($phone->textContent);
	$city = $dom->query('.social-link br + a.text-dark');
	// var_dump($city->textContent);

	file_put_contents('output.csv', sprintf("%s\t%s\t%s\t%s\t\t%s\n", $h1?->textContent, $city?->textContent, $mail?->textContent, $phone?->textContent, $url), FILE_APPEND);

	$done++;

	if (($done % 10) == 0) {
		echo "  $done / $total\n";
	}
}