private static function addByAttribute($img, $attribute, \DOMElement $html, Bag $bag, $domain = null)
{
$src = $img->hasAttribute($attribute);
if ($src) {
$src = new Url($img->getAttribute($attribute));
//Is src relative?
if (!$src->getDomain()) {
$bag->add('images', ['url' => $src->getUrl(), 'alt' => self::_getAltTag($img), 'href' => self::extractA($img, $bag->get('request_url'))]);
return;
}
//Avoid external images or in external links
if ($domain !== null) {
if (!preg_match('~(ebayimg)~i', $src->getDomain())) {
return;
}
$bag->add('images', ['url' => $src->getUrl(), 'alt' => self::_getAltTag($img), 'href' => self::extractA($img, $bag->get('request_url'))]);
}
}
}
/**
* Extract <img> elements
*
* @param \DOMElement $html
* @param Bag $bag
* @param null|string $domain
*/
protected static function extractImages(\DOMElement $html, Bag $bag, $domain = null)
{
foreach ($html->getElementsByTagName('img') as $img) {
if ($img->hasAttribute('src')) {
$src = new Url($img->getAttribute('src'));
//Is src relative?
if (!$src->getDomain()) {
$bag->add('images', ['url' => $src->getUrl(), 'alt' => $img->hasAttribute('alt') ? $img->getAttribute('alt') : '', 'href' => self::extractA($img, $bag->get('request_url'))]);
continue;
}
//Avoid external images or in external links
if ($domain !== null) {
if ($src->getDomain() !== $domain) {
continue;
}
$parent = $img->parentNode;
while ($parent && isset($parent->tagName)) {
if ($parent->tagName === 'a') {
if ($parent->hasAttribute('href')) {
$ahref = $parent->getAttribute('href');
$href = new Url($ahref);
//slow, very slow
// if(in_array(strtolower(pathinfo($ahref, PATHINFO_EXTENSION)), ['jpg', 'jpeg', 'png', 'gif'])) {
// $src = new Url($ahref);
// continue;
// }
if ($href->getDomain() && $src->getDomain() !== $domain) {
continue 2;
}
}
if ($parent->hasAttribute('rel') && (string) $parent->getAttribute('rel') === 'nofollow') {
continue 2;
}
break;
}
$parent = $parent->parentNode;
}
$bag->add('images', ['url' => $src->getUrl(), 'alt' => $img->hasAttribute('alt') ? $img->getAttribute('alt') : '', 'href' => self::extractA($img, $bag->get('request_url'))]);
}
}
}
}
/**
* Extract <img> elements
*
* @param \DOMElement $html
* @param Bag $bag
* @param null|string $domain
*/
protected static function extractImages(\DOMElement $html, Bag $bag, $domain = null)
{
foreach ($html->getElementsByTagName('img') as $img) {
if ($img->hasAttribute('src')) {
$src = new Url($img->getAttribute('src'));
//Is src relative?
if (!$src->getDomain()) {
$bag->add('images', $src->getUrl());
continue;
}
//Avoid external images or in external links
if ($domain !== null) {
if ($src->getDomain() !== $domain) {
continue;
}
$parent = $img->parentNode;
while ($parent && isset($parent->tagName)) {
if ($parent->tagName === 'a') {
if ($parent->hasAttribute('href')) {
$href = new Url($parent->getAttribute('href'));
if ($href->getDomain() && $src->getDomain() !== $domain) {
continue 2;
}
}
if ($parent->hasAttribute('rel') && (string) $parent->getAttribute('rel') === 'nofollow') {
continue 2;
}
break;
}
$parent = $parent->parentNode;
}
$bag->add('images', $src->getUrl());
}
}
}
}
/**
* Extract information from the <meta> elements.
*
* @param \DOMDocument $html
* @param Bag $bag
*/
protected static function extractFromMeta(\DOMDocument $html, Bag $bag)
{
foreach (Utils::getMetas($html) as $meta) {
list($name, $value, $element) = $meta;
if (!$value) {
continue;
}
if ($name) {
$name = strtolower($name);
switch ($name) {
case 'msapplication-tileimage':
$bag->add('icons', $value);
continue 2;
default:
$bag->set($name, $value);
continue 2;
}
}
if ($element->hasAttribute('itemprop')) {
$bag->set($element->getAttribute('itemprop'), $value);
}
if ($element->hasAttribute('http-equiv')) {
$bag->set($element->getAttribute('http-equiv'), $value);
}
}
}