array(
'name' => 'ASIN',
'required' => true,
'exampleValue' => 'B071GB1VMQ',
// https://stackoverflow.com/a/12827734
'pattern' => 'B[\dA-Z]{9}|\d{9}(X|\d)',
),
'tld' => array(
'name' => 'Country',
'type' => 'list',
'values' => array(
'Australia' => 'com.au',
'Brazil' => 'com.br',
'Canada' => 'ca',
'China' => 'cn',
'France' => 'fr',
'Germany' => 'de',
'India' => 'in',
'Italy' => 'it',
'Japan' => 'co.jp',
'Mexico' => 'com.mx',
'Netherlands' => 'nl',
'Spain' => 'es',
'Sweden' => 'se',
'Turkey' => 'com.tr',
'United Kingdom' => 'co.uk',
'United States' => 'com',
),
'defaultValue' => 'com',
),
));
const PRICE_SELECTORS = array(
'#priceblock_ourprice',
'.priceBlockBuyingPriceString',
'#newBuyBoxPrice',
'#tp_price_block_total_price_ww',
'span.offer-price',
'.a-color-price',
);
const WHITESPACE = " \t\n\r\0\x0B\xC2\xA0";
protected $title;
/**
* Generates domain name given a amazon TLD
*/
private function getDomainName() {
return 'https://www.amazon.' . $this->getInput('tld');
}
/**
* Generates URI for a Amazon product page
*/
public function getURI() {
if (!is_null($this->getInput('asin'))) {
return $this->getDomainName() . '/dp/' . $this->getInput('asin');
}
return parent::getURI();
}
/**
* Scrapes the product title from the html page
* returns the default title if scraping fails
*/
private function getTitle($html) {
$titleTag = $html->find('#productTitle', 0);
if (!$titleTag) {
return $this->getDefaultTitle();
} else {
return trim(html_entity_decode($titleTag->innertext, ENT_QUOTES));
}
}
/**
* Title used by the feed if none could be found
*/
private function getDefaultTitle() {
return 'Amazon.' . $this->getInput('tld') . ': ' . $this->getInput('asin');
}
/**
* Returns name for the feed
* Uses title (already scraped) if it has one
*/
public function getName() {
if (isset($this->title)) {
return $this->title;
} else {
return parent::getName();
}
}
private function parseDynamicImage($attribute) {
$json = json_decode(html_entity_decode($attribute), true);
if ($json and count($json) > 0) {
return array_keys($json)[0];
}
}
/**
* Returns a generated image tag for the product
*/
private function getImage($html) {
$imageSrc = $html->find('#main-image-container img', 0);
if ($imageSrc) {
$hiresImage = $imageSrc->getAttribute('data-old-hires');
$dynamicImageAttribute = $imageSrc->getAttribute('data-a-dynamic-image');
$image = $hiresImage ?: $this->parseDynamicImage($dynamicImageAttribute);
}
$image = $image ?: 'https://placekitten.com/200/300';
return <<
EOT;
}
/**
* Return \simple_html_dom object
* for the entire html of the product page
*/
private function getHtml() {
$uri = $this->getURI();
return getSimpleHTMLDOM($uri) ?: returnServerError('Could not request Amazon.');
}
private function scrapePriceFromMetrics($html) {
$asinData = $html->find('#cerberus-data-metrics', 0);
//
if ($asinData) {
return array(
'price' => $asinData->getAttribute('data-asin-price'),
'currency' => $asinData->getAttribute('data-asin-currency-code'),
'shipping' => $asinData->getAttribute('data-asin-shipping')
);
}
return false;
}
private function scrapePriceTwister($html) {
$str = $html->find('.twister-plus-buying-options-price-data', 0);
$data = json_decode($str->innertext, true);
if(count($data) === 1) {
$data = $data[0];
return array(
'displayPrice' => $data['displayPrice'],
'currency' => $data['currency'],
'shipping' => '0',
);
}
return false;
}
private function scrapePriceGeneric($html) {
$priceDiv = null;
foreach(self::PRICE_SELECTORS as $sel) {
$priceDiv = $html->find($sel, 0);
if ($priceDiv) {
break;
}
}
if (!$priceDiv) {
return false;
}
$priceString = str_replace(str_split(self::WHITESPACE), '', $priceDiv->plaintext);
preg_match('/(\d+\.\d{0,2})/', $priceString, $matches);
$price = $matches[0];
$currency = str_replace($price, '', $priceString);
if ($price != null && $currency != null) {
return array(
'price' => $price,
'currency' => $currency,
'shipping' => '0'
);
}
return false;
}
private function renderContent($image, $data) {
$price = $data['displayPrice'];
if (!$price) {
$price = "{$data['price']} {$data['currency']}";
}
$html = "$image
Price: $price";
if ($data['shipping'] !== '0') {
$html .= "
Shipping: {$data['shipping']} {$data['currency']}";
}
return $html;
}
/**
* Scrape method for Amazon product page
* @return [type] [description]
*/
public function collectData() {
$html = $this->getHtml();
$this->title = $this->getTitle($html);
$imageTag = $this->getImage($html);
$data = $this->scrapePriceGeneric($html);
$item = array(
'title' => $this->title,
'uri' => $this->getURI(),
'content' => $this->renderContent($imageTag, $data),
// This is to ensure that feed readers notice the price change
'uid' => md5($data['price'])
);
$this->items[] = $item;
}
}