<?php
require_once 'simpletest/browser.php';
$browser =& new SimpleBrowser();
$browser->useCookies();
$source = GetPageSource('https://www.nutritionxcellence.com/protein');
$dom = GetDomObject($source);
$xpath = GetXPathObject($dom);
$subcCategoriesLinksNodes = $xpath->query('//a');
foreach ($subcCategoriesLinksNodes as $node1) {
$ttt = $node1->attributes->item(0)->nodeValue;
echo $ttt . '<br />';
}
echo 'asdf';
function GetPageSource($URL)
{
$GLOBALS['browser']->get($URL);
$output = $GLOBALS['browser']->getContent();
return $output;
}
function GetDomObject($pageSource)
{
$dom = new DOMDocument('1.0');
libxml_use_internal_errors(true);
if (!is_null($pageSource)) {
$dom->loadHTML($pageSource);
return $dom;
} else {
return;
}
}
function GetProductsLinks($catURL, &$products)
{
$listingPageSource = GetPageSource($catURL);
if (empty($listingPageSource)) {
die('Sorry cURL failed to reterive page source');
}
$dom = GetDomObject($listingPageSource);
$xpath = GetXPathObject($dom);
$imgs = $dom->getElementsByTagName('img');
foreach ($imgs as $img) {
$imgParentAnchor = $img->parentNode->getAttribute('href');
$imgParentAnchor = 'http://www.nutritionxcellence.com' . $imgParentAnchor;
if (strpos($imgParentAnchor, '/index.php/') != false) {
array_push($products, $imgParentAnchor);
}
}
return $listingPageSource;
}
function ScrapeProductPage($URL, $catName)
{
$prodPageSource = GetPageSource($URL);
$dom = GetDomObject($prodPageSource);
$xPath = GetXPathObject($dom);
$Category = '';
$Manufacturer = '';
$ProductTitle = '';
$ProductDescription = '';
$SalesPrice = '';
#(USD)
$MSRPPrice = '';
$InStock = 'In stock';
$ProductImages = '';
$NutritionFacts = '';
$OtherIngredients = '';
$AllerginInfo = '';
$Warnings = '';
if (strpos($prodPageSource, 'Notify Me') != false) {
$InStock = "Out of stock";
}
$Category = $catName;
$manufacturerNode = $xPath->query("//*[contains(@class, 'manufacturer')]");
if (!is_null($manufacturerNode)) {
$Manufacturer = $manufacturerNode->item(0)->nodeValue;
$Manufacturer = trim(str_replace('Manufacturer:', '', $Manufacturer));
}
$titleNode = $xPath->query('//*[@id="main"]/div[2]/h1');
if (!is_null($titleNode)) {
$title = $titleNode->item(0)->nodeValue;
$ProductTitle = trim(str_replace("{$Manufacturer}:", '', $title));
}
$descNode = $xPath->query("//*[contains(@class, 'product-description')]");
if (!is_null($descNode)) {
$ProductDescription = $descNode->item(0)->nodeValue;
$ProductDescription = trim(preg_replace('/Description/', '', $ProductDescription));
$ProductDescription = PrepareField($ProductDescription);
}
$SalesPriceNode = $xPath->query("//*[contains(@class, 'PricesalesPrice')]");
if (!is_null($SalesPriceNode)) {
$SalesPrice = $SalesPriceNode->item(0)->nodeValue;
if (preg_match("/\\d+\\.\\d+/", $SalesPrice, $matches)) {
$SalesPrice = $matches[0];
}
}
preg_match("/MSRP:.+?<s>(.+?)<\\/s>/", $prodPageSource, $matchs);
if (!is_null($matchs)) {
$MSRPPrice = $matchs[1];
}
$imageNode = $dom->getElementById('medium-image');
if (!is_null($imageNode)) {
$ProductImages = 'http://www.nutritionxcellence.com' . $imageNode->getAttribute('src');
}
for ($i = 1; $i <= 4; $i++) {
$tabNode = $dom->getElementById("tabs-{$i}");
if (!is_null($tabNode)) {
switch ($i) {
case 1:
$NutritionFacts = PrepareField($tabNode->nodeValue);
break;
case 2:
$OtherIngredients = PrepareField($tabNode->nodeValue);
break;
case 3:
$AllerginInfo = PrepareField($tabNode->nodeValue);
break;
case 4:
$Warnings = PrepareField($tabNode->nodeValue);
}
}
}
# Push data to a csv file.
$headerArray = array("Category", "Manufacturer", "ProductTitle", "ProductDescription", "SalesPrice", "MSRPPrice", "InStock", "ProductImages", "NutritionFacts", "OtherIngredients", "AllerginInfo", "Warnings");
$rowArray = array("{$Category}", "{$Manufacturer}", "{$ProductTitle}", "{$ProductDescription}", "{$SalesPrice}", "{$MSRPPrice}", "{$InStock}", "{$ProductImages}", "{$NutritionFacts}", "{$OtherIngredients}", "{$AllerginInfo}", "{$Warnings}");
PushDataToCSV('data.csv', $headerArray, $rowArray);
echo $title . ' -- Done' . '<br />';
}
