/**
* This will extract individual product information.
* @return array
*/
public function getResponse()
{
//set initial data response
$data = array();
//get curl response
/** @var \Library\Common\Curl $curl */
$curl = $this->getCurl();
$curl->get($this->getUrl());
$response = $curl->getResponse();
//parse response html dom
$parser = new HtmlDomParser();
$parser->loadstring($response);
//get html filesize
$data['size'] = strlen($response);
//get product description information
$strings = array();
foreach ($parser->find('#information .productText p') as $product) {
$strings[] = $product->plaintext;
}
if (!empty($strings[0])) {
$data['description'] = $strings[0];
}
//return html size and description in array
return $data;
}
/**
* This will get the basic product information from the product list html snippet response
* @param $html
* @return array
*/
public function getProductInfo($html)
{
//parse html dom
$parser = new HtmlDomParser();
$parser->loadstring($html);
//init default productInfo array
$productInfo = array();
//set product incrementer
$i = 0;
//iterate over .product classes
foreach ($parser->find('.product') as $product) {
//iterate over .productInfo a elements
foreach ($product->find('.productInfo a') as $a) {
//get product title
$productInfo['title'] = trim($a->plaintext);
//get product href
$productInfo['href'] = trim($a->href);
//use first anchor tag result only
break;
}
//iterate over .pricePerUnit class
foreach ($parser->find('.pricePerUnit') as $pricePerUnit) {
//get product unit price.
$unitPrice = $pricePerUnit->plaintext;
$unitPrice = str_replace('/unit', '', $unitPrice);
$unitPrice = str_replace('£', '', $unitPrice);
$productInfo['unit_price'] = trim($unitPrice);
}
$i++;
}
//return product title, href, and unit price
return $productInfo;
}
use League\Csv\Writer;
# Store url
$url = "http://centrodosuplemento.com.br/suplementos";
$html = new HtmlDomParser();
$html->loadUrl($url);
$page_numbers = [];
foreach ($html->find('.toolbar-bottom .pager .pages ol li') as $page) {
$page_numbers[] = $page->plaintext;
}
$max_page = max($page_numbers);
# Initialize Arrays
$name = [];
$price = [];
for ($i = 1; $i <= $max_page; $i++) {
# Open search results page
$url = "http://centrodosuplemento.com.br/suplementos?mode=list?p={$i}";
$product_html = new HtmlDomParser();
$product_html->loadUrl($url);
# Store data in Arrays
foreach ($product_html->find('.product-name a') as $line) {
$name[] = $line->plaintext;
}
foreach ($product_html->find('.price-box .regular-price .price') as $line) {
$price[] = $line->plaintext;
}
}
$writer = Writer::createFromPath('cds_list.csv', 'w');
$writer->insertOne(["Listing Name", "Price"]);
for ($p = 0; $p < count($name); $p++) {
$writer->insertOne([$name[$p], $price[$p]]);
}