diff --git a/bridges/DeveloppezDotComBridge.php b/bridges/DeveloppezDotComBridge.php index c32a0d6..bc868ae 100644 --- a/bridges/DeveloppezDotComBridge.php +++ b/bridges/DeveloppezDotComBridge.php @@ -19,23 +19,41 @@ class DeveloppezDotComBridge extends BridgeAbstract{ return $string; } + function convert_smart_quotes($string)//F***ing quotes from Microsoft Word badly encoded, here was the trick: http://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php + { + $search = array(chr(145), + chr(146), + chr(147), + chr(148), + chr(151)); + + $replace = array("'", + "'", + '"', + '"', + '-'); + + return str_replace($search, $replace, $string); + } + function DeveloppezDotComExtractContent($url) { $articleHTMLContent = file_get_html($url); - $text = $text.$articleHTMLContent->find('div.content', 0)->innertext; - $text = strip_tags($text, '



  • '); - return $text; + $text = convert_smart_quotes($articleHTMLContent->find('div.content', 0)->innertext); + $text = utf8_encode($text); + return trim($text); } $rssFeed = file_get_html('http://www.developpez.com/index/rss') or $this->returnError('Could not request http://www.developpez.com/index/rss', 404); $limit = 0; foreach($rssFeed->find('item') as $element) { - if($limit < 15) { + if($limit < 2) { $item = new \Item(); $item->title = DeveloppezDotComStripCDATA($element->find('title', 0)->innertext); $item->uri = DeveloppezDotComStripCDATA($element->find('guid', 0)->plaintext); $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext); - $item->content = DeveloppezDotComExtractContent($item->uri); + $content = DeveloppezDotComExtractContent($item->uri); + $item->content = strlen($content) ? $content : $element->description;//In case of it is a tutorial, we just keep the original description $this->items[] = $item; $limit++; }