find('div[class=displayList]', 0)->children(); foreach($list as $element) { if($element->tag == 'a') { $articleURL = self::URI . $element->href; $article = getSimpleHTMLDOM($articleURL); $textDOM = $article->find('article', 0); // Initialise arrays $item = array(); $audio = array(); $picture = array(); // Get the Main picture URL $picture[] = $this->rewriteImage($article->find('div[id=pictureTitleSupport]', 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); } // Remove Google Ads $ads = $article->find('div[class=adInline]'); foreach($ads as $ad) { $ad->remove(); } // Remove Radio Melodie Logo $logoHTML = $article->find('div[id=logoArticleRM]', 0); $logoHTML->remove(); $author = $article->find('p[class=AuthorName]', 0)->plaintext; $item['enclosures'] = array_merge($picture, $audio); $item['author'] = $author; $item['uri'] = $articleURL; $item['title'] = $article->find('meta[property=og:title]', 0)->content; $date = $article->find('p[class*=date]', 0)->plaintext; // Header Image $header = ''; // Remove the Date and Author part $textDOM->find('div[class=AuthorDate]', 0)->remove(); $text = $textDOM->innertext; $item['content'] = '

' . $item['title'] . '

' . $date . '
' . $header . $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']; } }