public function testCanDecorateInfo()
{
$productInfo = new ProductInfo();
$productInfo->setTitle('title');
$productInfo->setSize('size');
$productInfo->setDescription('description');
$productInfo->setUnitCost('unit_price');
$expected = ['title' => 'title', 'size' => 'size', 'unit_price' => 'unit_price', 'description' => 'description'];
$this->assertEquals($expected, ScraperLibrary::productInfoArrayDecorator($productInfo));
}
/**
* scrape description from product page and get page size
*
* @return ProductInfo
*/
public function scrape()
{
//get response to page request
$client = new Client(['cookies' => true]);
$response = $client->request('GET', $this->getProductInfo()->getLink());
//try to get size from header, if not found fall back to calculating from page body
if (!($size = $response->getHeader('Content-Length'))) {
$size = $response->getBody()->getSize();
$size = ScraperLibrary::formatBytes($size, 1);
}
$this->getProductInfo()->setSize($size);
$this->getProductInfo()->setDescription($this->scrapeProductDescription($response->getBody()));
return $this->getProductInfo();
}
/**
* scraper to find link to item page from product list item
*/
public function scrapeLink()
{
return ScraperLibrary::findDelimitedContent($this->getString(), $this->getLinkStartMarker(), $this->getLinkEndMarker());
}
public function scrape()
{
//use Guzzle to get page details (cookies on for js load)
$client = new Client(['cookies' => true]);
$response = $client->request('GET', $this->getUrl());
//get the product list string from the page body
$productList = $this->getProductListString($response->getBody());
//convert list to array (removing first element as its before the list elements)
$productListArray = explode('<li>', $productList);
array_shift($productListArray);
$totalCost = 0;
$finalArray = [];
//for each product in list scrape the data and add to total
foreach ($productListArray as $listEntry) {
//send list item html to list scraper which creates a ProductInfo object
$productListItemScraper = new ProductListItemScraper($listEntry);
$productListItemScraper->scrape();
//send ProductInfo object to page scraper
//which uses the link found in the list item to get further information
$productPageScraper = new ProductPageScraper($productListItemScraper->getProductInfo());
$productPageScraper->scrape();
//and finally get complete productInfo object
$productInfo = $productPageScraper->getProductInfo();
//add cost to running total and get formatted array from info object
$totalCost = $totalCost + floatval($productInfo->getUnitCost());
$finalArray[] = ScraperLibrary::productInfoArrayDecorator($productInfo);
}
return ['results' => $finalArray, 'total' => $totalCost];
}
