/**
* getProducts()
* Scrapes the products and returns them in a unified JSON format
*/
public function getProducts()
{
// Load HTML into scraper
$scraper = new Crawler($this->source);
// Get products
$results = $scraper->filter('#productLister .product');
$products = [];
foreach ($results as $product) {
// Create new crawler for each product iteration
$product = new Crawler($product);
// Get title
$title = trim($product->filter('h3')->first()->text());
// Get link
$link = trim($product->filter('h3 a')->first()->attr('href'));
// Get unit_price
$unit_price = preg_replace('/[^\\d+\\.]/', '', $product->filter('.pricePerUnit')->first()->text());
try {
// Get secondary resource
Console::log('Following product link: ' . $link);
$resource = $this->client->request('GET', $link);
// Get size / Content-Length
$size = number_format($resource->getHeader('Content-Length')[0] / 1024, 2);
// Create tertiary crawler to get description
$productDetails = new Crawler($resource->getBody()->getContents());
// Get description
$description = trim($productDetails->filter('#information .productText')->first()->text());
} catch (RequestException $e) {
Console::log('An error occurred while trying to fetch details for a product. I skipped this one for now. Error: ' . $e->getMessage());
continue;
}
// Collect item
$products[] = ['title' => $title, 'unit_price' => $unit_price, 'size' => $size . 'kb', 'description' => $description];
}
// Get total value and namespace the array
$totalSum = array_sum(array_column($products, 'unit_price'));
try {
// Transform collection
$products = $this->transformCollection($products);
// Namespace collection and encode to JSON
$json = $this->toJson(['results' => $products, 'totals' => $totalSum]);
} catch (JsonException $e) {
// JSON transformation failed, I'd usually also log this incident here with Monolog or similar
Console::log('An error occurred while transforming the products into the appropriate JSON format: ' . $e->getMessage());
return false;
}
return $json;
}
<?php
// Mainly for PSR-4 auto-loading
require_once dirname(__FILE__) . '/vendor/autoload.php';
// Initiate Scraper
$scraper = new \Sainsbury\HtmlScraper();
$client = new \GuzzleHttp\Client();
// Logging
\Sainsbury\Console::log('Getting initial source...');
$source = $client->request('GET', 'http://hiring-tests.s3-website-eu-west-1.amazonaws.com/2015_Developer_Scrape/5_products.html');
// Get Products
$products = $scraper->setSource($source->getBody()->getContents())->getProducts();
// Show Products
print_r($products);
