Check URLs to see if they are on Shopify or WooCommerce

check-urls.php

<?php
/**
 * Instructions: cd into site directory
 * 
 * Run `wp eval-file ./path/to/script.php`
 */

global $wpdb;

// Query
$sites             = $wpdb->get_col( "SELECT DISTINCT (site_name) FROM wp_edd_license_activations WHERE is_local = 0 AND license_id IN( SELECT id FROM wp_edd_licenses WHERE status = 'expired' )" );
$shopify_sites     = array();
$non_shopify_sites = array();
$woocommerce_sites = array();
$err_sites         = array();
$dead_sites        = array();

foreach ( $sites as $site ) {

	$url = $site;

	if ( false === strpos( $site, 'http://' ) && false === strpos( $site, 'https://' ) ) {
		$url = "https://{$site}";
	}

	// If dev or staging site skip it
	if ( cg_is_local_url( $url ) ) {
		echo "Skipping {$url} because it is a dev or staging site.\n";
		continue;
	}

	// Check if URL has A record
	$dns = checkdnsrr( rtrim( $site, '/' ), 'A' );

	if ( ! $dns ) {
		echo "Skipping {$url} because it has no A record.\n";
		$dead_sites[] = $url;
		continue;
	}


	stream_context_set_default(
		array(
			'http' => array(
				'method'  => 'GET',
				'timeout' => 5,
			),
		)
	);

	// Get the response headers for the URL
	$headers = get_headers( $url, true );

	if ( ! $headers ) {
		echo "Erred on site {$url}\n";
		$err_sites[] = $url;
		continue;
	}

	// print_r( $headers );
	$headers = array_change_key_case( $headers );

	// Check if the "X-ShopId" header is set
	if ( isset( $headers['x-shopid'] ) || isset( $headers['x-sorting-hat-shopid'] ) ) {
		echo "{$url} is on Shopify!\n";
		$shopify_sites[] = $url;
		continue;
	}

	// Check if this file exists wp-content/plugins/woocommerce/license.txt
	$license_file = wp_remote_get( $url . '/wp-content/plugins/woocommerce/assets/js/frontend/add-to-cart.min.js' );

	if ( ! is_wp_error( $license_file ) && 200 === wp_remote_retrieve_response_code( $license_file ) ) {
		echo "{$url} is on WooCommerce!\n";
		$woocommerce_sites[] = $url;

		continue;
	}

	echo "{$url} is not on Shopify or Woo.\n";
	$non_shopify_sites[] = $url;
}

echo 'Shopify sites: ' . count( $shopify_sites ) . "\n";
echo 'WooCommerce sites: ' . count( $woocommerce_sites ) . "\n";
echo 'Non-Shopify and Non-Woo sites: ' . count( $non_shopify_sites ) . "\n";
echo 'Erred sites: ' . count( $err_sites ) . "\n";
echo 'Dead sites: ' . count( $dead_sites ) . "\n";

function cg_is_local_url( $url = '' ) {
	$is_local_url = false;

	// Trim it up
	$url = strtolower( trim( $url ) );

	// Need to get the host...so let's add the scheme so we can use parse_url
	if ( false === strpos( $url, 'http://' ) && false === strpos( $url, 'https://' ) ) {
		$url = 'http://' . $url;
	}

	$url_parts = parse_url( $url );
	$host      = ! empty( $url_parts['host'] ) ? $url_parts['host'] : false;

	if ( empty( $url ) || empty( $host ) ) {
		return false;
	}

	if ( false !== ip2long( $host ) ) {
		if ( ! filter_var( $host, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE ) ) {
			$is_local_url = true;
		}
	} elseif ( 'localhost' === $host ) {
		$is_local_url = true;
	}

	$tlds_to_check = array(
		'.dev',
		'.local',
		'.test',
	);

	foreach ( $tlds_to_check as $tld ) {
		if ( false !== strpos( $host, $tld ) ) {
			$is_local_url = true;
			break;
		}
	}

	if ( substr_count( $host, '.' ) > 1 ) {
		$subdomains_to_check = array(
			'dev.',
			'*.staging.',
			'*.test.',
			'staging-*.',
			'*.wpengine.com',
			'*cloudways*',
		);

		foreach ( $subdomains_to_check as $subdomain ) {

			$subdomain = str_replace( '.', '(.)', $subdomain );
			$subdomain = str_replace( array( '*', '(.)' ), '(.*)', $subdomain );

			if ( preg_match( '/^(' . $subdomain . ')/', $host ) ) {
				$is_local_url = true;
				break;
			}
		}
	}

	// If url has more than 3 forward slashes, it's probably a local url
	if ( substr_count( $url, '/' ) > 3 ) {
		$is_local_url = true;
	}

	return $is_local_url;
}