2014-05-26 21:45:10 +04:00
|
|
|
<?php
|
|
|
|
class NiceMatinBridge extends BridgeAbstract{
|
|
|
|
|
2016-08-27 22:03:26 +03:00
|
|
|
public $maintainer = "pit-fgfjiudghdf";
|
|
|
|
public $name = "NiceMatin";
|
|
|
|
public $uri = "http://www.nicematin.com/";
|
|
|
|
public $description = "Returns the 10 newest posts from NiceMatin (full text)";
|
2015-11-04 12:47:21 +03:00
|
|
|
|
2016-08-06 17:00:56 +03:00
|
|
|
private function NiceMatinExtractContent($url) {
|
2016-07-08 20:06:35 +03:00
|
|
|
$html = $this->getSimpleHTMLDOM($url);
|
2016-08-03 23:39:03 +03:00
|
|
|
if(!$html)
|
2016-08-29 15:03:17 +03:00
|
|
|
return 'Could not acquire content from url: ' . $url . '!';
|
2016-07-08 20:06:35 +03:00
|
|
|
|
2016-08-03 23:39:03 +03:00
|
|
|
$content = $html->find('article', 0);
|
|
|
|
if(!$content)
|
2016-08-29 15:03:17 +03:00
|
|
|
return 'Could not find \'section\'!';
|
2016-07-08 20:06:35 +03:00
|
|
|
|
2016-08-03 23:39:03 +03:00
|
|
|
$text = preg_replace('#<script(.*?)>(.*?)</script>#is', '', $content->innertext);
|
|
|
|
$text = strip_tags($text, '<p><a><img>');
|
|
|
|
return $text;
|
2015-11-04 12:47:21 +03:00
|
|
|
}
|
|
|
|
|
2016-08-25 02:24:53 +03:00
|
|
|
public function collectData(){
|
2016-08-29 15:03:17 +03:00
|
|
|
$html = $this->getSimpleHTMLDOM($this->uri.'derniere-minute/rss')
|
|
|
|
or $this->returnServerError('Could not request NiceMatin.');
|
2016-08-03 23:39:03 +03:00
|
|
|
$limit = 0;
|
|
|
|
|
|
|
|
foreach($html->find('item') as $element) {
|
2016-08-29 15:03:17 +03:00
|
|
|
if($limit >= 10) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// We need to fix the 'link' tag as simplehtmldom cannot parse it (just rename it and load back as dom)
|
|
|
|
$element_text = $element->outertext;
|
|
|
|
$element_text = str_replace('<link>', '<url>', $element_text);
|
|
|
|
$element_text = str_replace('</link>', '</url>', $element_text);
|
|
|
|
$element = str_get_html($element_text);
|
2016-08-03 23:39:03 +03:00
|
|
|
|
2016-08-29 15:03:17 +03:00
|
|
|
$item = array();
|
|
|
|
$item['title'] = $element->find('title', 0)->innertext;
|
|
|
|
$item['uri'] = $element->find('url', 0)->innertext;
|
|
|
|
$item['timestamp'] = strtotime($element->find('pubDate', 0)->plaintext);
|
|
|
|
$item['content'] = $this->NiceMatinExtractContent($item['uri']);
|
|
|
|
$this->items[] = $item;
|
|
|
|
$limit++;
|
2016-08-03 23:39:03 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|