find('div[class=listArticles]', 0)->children();
foreach ($list as $element) {
if ($element->tag == 'a') {
$articleURL = self::URI . $element->href;
$article = getSimpleHTMLDOM($articleURL);
$this->rewriteAudioPlayers($article);
// Reload the modified content
$article = str_get_html($article->save());
$textDOM = $article->find('article', 0);
// Initialise arrays
$item = [];
$audio = [];
$picture = [];
// Get the Main picture URL
$picture[] = self::URI . $article->find('figure[class=photoviewer]', 0)->find('img', 0)->src;
$audioHTML = $article->find('audio');
// Add the audio element to the enclosure
foreach ($audioHTML as $audioElement) {
$audioURL = $audioElement->src;
$audio[] = $audioURL;
}
// Rewrite pictures URL
$imgs = $textDOM->find('img[src^="http://www.radiomelodie.com/image.php]');
foreach ($imgs as $img) {
$img->src = $this->rewriteImage($img->src);
$article->save();
}
// Remove Google Ads
$ads = $article->find('div[class=adInline]');
foreach ($ads as $ad) {
$ad->outertext = '';
$article->save();
}
// Extract the author
$author = $article->find('div[class=author]', 0)->children(1)->children(0)->plaintext;
// Handle date to timestamp
$dateHTML = $article->find('div[class=author]', 0)->children(1)->plaintext;
preg_match('/([a-z]{4,10}[ ]{1,2}[0-9]{1,2} [\p{L}]{3,10} [0-9]{4} à [0-9]{2}:[0-9]{2})/mus', $dateHTML, $matches);
$dateText = $matches[1];
$timestamp = $this->parseDate($dateText);
$item['enclosures'] = array_merge($picture, $audio);
$item['author'] = $author;
$item['uri'] = $articleURL;
$item['title'] = $article->find('meta[property=og:title]', 0)->content;
if ($timestamp !== false) {
$item['timestamp'] = $timestamp;
}
// Remove the share article part
$textDOM->find('div[class=share]', 0)->outertext = '';
// Rewrite relative Links
$textDOM = defaultLinkTo($textDOM, self::URI . '/');
$article->save();
$text = $textDOM->innertext;
$item['content'] = '
' . $item['title'] . '
' . $dateText . ' ' . $text;
$this->items[] = $item;
}
}
}
/*
* Function to rewrite image URL to use the real Image URL and not the resized one (which is very slow)
*/
private function rewriteImage($url)
{
$parts = explode('?', $url);
parse_str(html_entity_decode($parts[1]), $params);
return self::URI . '/' . $params['image'];
}
/*
* Function to rewrite Audio Players to use the