蘋果官網整修品頁面新品取得機器人

Apple_special_deal_crawler_demo.php

<?php
require_once __DIR__ . '/../vendor/autoload.php';

use Goutte\Client as Goutte_Client;

$client = new Goutte_Client();
$records_path = __DIR__ . '/../logs/records.log';
$apple_url = 'https://www.apple.com';
$country_code = '/tw';
$specialdeals_url = '/shop/browse/home/specialdeals/mac';
$request_uri = "{$apple_url}{$country_code}{$specialdeals_url}";

$content_selector = 'div#primary div.box.refurb-list > div.box-content';

$crawler = $client->request('GET', $request_uri)
                  ->filter($content_selector);

$new_items = [];
$products_selector = 'table > tbody > tr';
$crawler->filter($products_selector)
        ->each(function ($node) use (&$new_items) {
            $item = trim($node->filter('td.specs > h3 > a')->text());
            $price  = trim(
                substr(trim($node->filter('td.purchase-info > p')->text()), -10)
            );
            $url = $node->filter('td.purchase-info > a')->attr('href');
            $url_attributes = explode('/', $url);
            $product_id = trim($url_attributes[4]);

            $new_items[$product_id] = [
                'Item'  => $item,
                'Price' => $price,
                'Url'   => $url,
            ];
        });

$new_record = json_encode($new_items);

$previous_record = file_get_contents($records_path);

if (
    $previous_record !== false
    && $new_record != $previous_record
) {
    send_alert($new_items, json_decode($previous_record, true));
}

file_put_contents($records_path, $new_record);

function send_alert($new, $previous) {
    // do something
    print_r(array_diff_key($new, $previous));
}