/**
* Given a source (URL), retrieve raw data
* and turn it into an array of arrays of product data.
*
* Each sub array should contain the following keys:
* - title (string) - the title of the product
* - size (int) - bytes of raw data
* - unit_price (float) - the price of the product
* - description - description of the product
*
* @param string $source Information about where to load data from.
*
* @throws InvalidDataSourceException If the source is invalid.
* @throws MalformedDataException If the data is malformed.
*
* @return array
*/
public function getProductData($source)
{
// We will save errors to an array.
libxml_use_internal_errors(true);
if (false === $this->looksLikeUrl($source)) {
throw new InvalidDataSourceException("This isn't a URL.");
}
// Create a DOM parser & load the HTML.
$dom = new DOMParser();
$listHtml = $this->scraper->getHtml($source);
$return = array();
$dom->loadHTML($listHtml);
foreach ($dom->querySelectorAll("div.product") as $product) {
// Get the title element & price element
$titleElement = $product->querySelector("div.productInfo")->querySelector("a");
$priceElement = $product->querySelector("p.pricePerUnit");
// Load the product-specific page.
$productUrl = $titleElement->getAttributeNode("href")->textContent;
$productHtml = $this->scraper->getHtml($productUrl);
$productDom = new DOMParser();
$productDom->loadHTML($productHtml);
// Get the product description.
// Note: this does not have a unique ID, this is very volatile.
$descriptionElement = $productDom->querySelector("div#information")->querySelector("div.productText");
// Set up the product data.
$productData = array();
$productData['title'] = trim($titleElement->textContent);
$productData['description'] = trim($descriptionElement->textContent);
$productData['size'] = strlen($productHtml);
$productData['unit_price'] = (double) preg_replace('/[^0-9.]/', '', $priceElement->textContent);
$return[] = $productData;
}
// Put errors in $this->errors.
$this->errors = libxml_get_errors();
libxml_clear_errors();
return $return;
}
private function loadFragmentIntoDom(\DOMDocument $document, $source, $loadOptions)
{
$htmlDom = new Document();
$htmlDom->loadHTML('<html-fragment>' . $source . '</html-fragment>', $loadOptions);
$nodes = $htmlDom->evaluate('//html-fragment[1]/node()');
foreach ($nodes as $node) {
if ($importedNode = $document->importNode($node, TRUE)) {
$document->appendChild($importedNode);
}
}
}
