2014-07-14 21:41:09 +04:00
|
|
|
<?php
|
|
|
|
class DeveloppezDotComBridge extends BridgeAbstract{
|
|
|
|
|
2016-08-27 22:03:26 +03:00
|
|
|
public $maintainer = "polopollo";
|
|
|
|
public $name = "Developpez.com Actus (FR)";
|
|
|
|
public $uri = "http://www.developpez.com/";
|
|
|
|
public $description = "Returns the 15 newest posts from DeveloppezDotCom (full text).";
|
2015-11-04 01:28:44 +03:00
|
|
|
|
2016-08-06 17:00:56 +03:00
|
|
|
private function DeveloppezDotComStripCDATA($string) {
|
2016-08-03 13:37:56 +03:00
|
|
|
$string = str_replace('<![CDATA[', '', $string);
|
|
|
|
$string = str_replace(']]>', '', $string);
|
|
|
|
return $string;
|
|
|
|
}
|
2014-07-14 21:41:09 +04:00
|
|
|
|
2016-07-08 20:06:35 +03:00
|
|
|
// F***ing quotes from Microsoft Word badly encoded, here was the trick:
|
2016-08-03 13:42:57 +03:00
|
|
|
// http://stackoverflow.com/questions/1262038/how-to-replace-microsoft-encoded-quotes-in-php
|
2016-08-06 17:00:56 +03:00
|
|
|
private function convert_smart_quotes($string)
|
2016-08-03 13:37:56 +03:00
|
|
|
{
|
|
|
|
$search = array(chr(145),
|
|
|
|
chr(146),
|
|
|
|
chr(147),
|
|
|
|
chr(148),
|
|
|
|
chr(151));
|
2014-07-14 21:41:09 +04:00
|
|
|
|
2016-08-03 13:37:56 +03:00
|
|
|
$replace = array("'",
|
|
|
|
"'",
|
|
|
|
'"',
|
|
|
|
'"',
|
|
|
|
'-');
|
2014-07-16 04:31:54 +04:00
|
|
|
|
2016-08-03 13:37:56 +03:00
|
|
|
return str_replace($search, $replace, $string);
|
|
|
|
}
|
2014-07-16 04:31:54 +04:00
|
|
|
|
2016-08-06 17:00:56 +03:00
|
|
|
private function DeveloppezDotComExtractContent($url) {
|
2016-07-08 20:06:35 +03:00
|
|
|
$articleHTMLContent = $this->getSimpleHTMLDOM($url);
|
2016-08-03 13:37:56 +03:00
|
|
|
$text = $this->convert_smart_quotes($articleHTMLContent->find('div.content', 0)->innertext);
|
|
|
|
$text = utf8_encode($text);
|
|
|
|
return trim($text);
|
|
|
|
}
|
2014-07-16 04:31:54 +04:00
|
|
|
|
2016-08-25 02:24:53 +03:00
|
|
|
public function collectData(){
|
2016-08-28 15:00:40 +03:00
|
|
|
$rssFeed = $this->getSimpleHTMLDOM($this->uri.'index/rss')
|
|
|
|
or $this->returnServerError('Could not request '.$this->uri.'index/rss');
|
2016-08-03 13:42:57 +03:00
|
|
|
$limit = 0;
|
2014-07-14 21:41:09 +04:00
|
|
|
|
2016-08-03 13:42:57 +03:00
|
|
|
foreach($rssFeed->find('item') as $element) {
|
|
|
|
if($limit < 10) {
|
2016-08-22 19:55:59 +03:00
|
|
|
$item = array();
|
|
|
|
$item['title'] = $this->DeveloppezDotComStripCDATA($element->find('title', 0)->innertext);
|
|
|
|
$item['uri'] = $this->DeveloppezDotComStripCDATA($element->find('guid', 0)->plaintext);
|
|
|
|
$item['timestamp'] = strtotime($element->find('pubDate', 0)->plaintext);
|
|
|
|
$content = $this->DeveloppezDotComExtractContent($item['uri']);
|
|
|
|
$item['content'] = strlen($content) ? $content : $element->description; //In case of it is a tutorial, we just keep the original description
|
2016-08-03 13:42:57 +03:00
|
|
|
$this->items[] = $item;
|
|
|
|
$limit++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2014-07-14 21:41:09 +04:00
|
|
|
|
2016-08-03 13:42:57 +03:00
|
|
|
public function getCacheDuration(){
|
|
|
|
return 1800; // 30min
|
|
|
|
}
|
2014-07-14 21:41:09 +04:00
|
|
|
}
|