protected function customScraper($xpath, $urlComponents)
{
// This sample is really just a copy of the code from the default Handler
// with arbitrary values filled in for the parent's private functions
// All logic should be implemented in this function
echo "Bam! We're in the custom Handler.";
$xpathQuery = '/html/head/title';
$title = parent::getTitle($xpath, $xpathQuery);
$price = 10000000.0;
//arbitrary
$xpathQuery = '/html/head/meta';
$descriptionNodeArray = parent::getDescription($xpath, $xpathQuery);
$description = "a description metatag has not been found";
for ($i = 0; $i < $descriptionNodeArray->length; $i++) {
$descriptionNode = $descriptionNodeArray->item($i);
$descriptionNode->getAttribute('name') == 'description' || $descriptionNode->getAttribute('name') == 'Description' ? $description = $descriptionNode->getAttribute('content') : 0;
}
$imageWidthThreshold = 200;
$xpathQuery = '/html/body/descendant::img';
$productImages = parent::getImages($xpath, $xpathQuery, $imageWidthThreshold);
$normalizedUrl = "This is a fake normalized URL. Unimplemented";
//arbitrary
//assemble return array
$scrapedValues = array();
$scrapedValues[] = $title;
$scrapedValues[] = $price;
$scrapedValues[] = $description;
$scrapedValues[] = $productImages;
$scrapedValues[] = $normalizedUrl;
return $scrapedValues;
}
<?php
require_once 'ProductScraper.php';
// Test urls
/*
http://www.zappos.com/lucky-brand-abbey-road-dune?zlfid=111
http://store.americanapparel.net/rsa642sg.html
http://www.llbean.com/webapp/wcs/stores/servlet/CategoryDisplay?storeId=1&catalogId=1&langId=-1&categoryId=18584&productId=91702&qs=3009652
*/
$scrapedValues = array();
$scrapedValues = ProductScraper::getInfo('http://store.americanapparel.net/rsa642sg.html ');
//testing output
echo "Title: {$scrapedValues['0']} <br />";
echo "Price: " . $scrapedValues[1] . "<br />";
echo "Description: {$scrapedValues['2']} <br />";
echo "Images: ";
print_r($scrapedValues[3]);
echo '<br />';
echo "Normalized url: {$scrapedValues['4']} <br />";
?>
