2015-05-05 21:18:12 +03:00
|
|
|
<?php
|
2017-02-11 18:16:56 +03:00
|
|
|
class ReporterreBridge extends BridgeAbstract {
|
2015-11-05 14:20:11 +03:00
|
|
|
|
2017-02-11 18:16:56 +03:00
|
|
|
const MAINTAINER = 'nyutag';
|
|
|
|
const NAME = 'Reporterre Bridge';
|
2019-10-16 22:34:28 +03:00
|
|
|
const URI = 'https://www.reporterre.net/';
|
2017-02-11 18:16:56 +03:00
|
|
|
const DESCRIPTION = 'Returns the newest articles.';
|
2015-11-05 14:20:11 +03:00
|
|
|
|
2017-02-11 18:16:56 +03:00
|
|
|
private function extractContent($url){
|
2016-09-26 00:22:33 +03:00
|
|
|
$html2 = getSimpleHTMLDOM($url);
|
2016-08-04 13:34:40 +03:00
|
|
|
|
2017-07-29 20:28:00 +03:00
|
|
|
foreach($html2->find('div[style=text-align:justify]') as $e) {
|
2016-08-04 13:11:10 +03:00
|
|
|
$text = $e->outertext;
|
|
|
|
}
|
2016-08-04 13:34:40 +03:00
|
|
|
|
2016-08-04 13:11:10 +03:00
|
|
|
$html2->clear();
|
2017-02-11 18:16:56 +03:00
|
|
|
unset($html2);
|
2016-08-04 13:34:19 +03:00
|
|
|
|
|
|
|
// Replace all relative urls with absolute ones
|
2017-02-11 18:16:56 +03:00
|
|
|
$text = preg_replace(
|
|
|
|
'/(href|src)(\=[\"\'])(?!http)([^"\']+)/ims',
|
2018-06-30 00:55:33 +03:00
|
|
|
'$1$2' . self::URI . '$3',
|
2017-02-11 18:16:56 +03:00
|
|
|
$text
|
|
|
|
);
|
2016-08-04 13:34:19 +03:00
|
|
|
|
2016-08-04 13:34:40 +03:00
|
|
|
$text = strip_tags($text, '<p><br><a><img>');
|
2016-08-04 13:11:10 +03:00
|
|
|
return $text;
|
2015-05-05 21:18:12 +03:00
|
|
|
}
|
|
|
|
|
2016-08-25 02:24:53 +03:00
|
|
|
public function collectData(){
|
2017-02-11 18:16:56 +03:00
|
|
|
$html = getSimpleHTMLDOM(self::URI . 'spip.php?page=backend')
|
|
|
|
or returnServerError('Could not request Reporterre.');
|
2015-05-05 21:18:12 +03:00
|
|
|
$limit = 0;
|
|
|
|
|
2017-07-29 20:28:00 +03:00
|
|
|
foreach($html->find('item') as $element) {
|
2016-08-04 13:34:40 +03:00
|
|
|
if($limit < 5) {
|
2016-08-22 19:55:59 +03:00
|
|
|
$item = array();
|
|
|
|
$item['title'] = html_entity_decode($element->find('title', 0)->plaintext);
|
|
|
|
$item['timestamp'] = strtotime($element->find('dc:date', 0)->plaintext);
|
|
|
|
$item['uri'] = $element->find('guid', 0)->innertext;
|
2017-02-11 18:16:56 +03:00
|
|
|
$item['content'] = html_entity_decode($this->extractContent($item['uri']));
|
2016-08-04 13:34:40 +03:00
|
|
|
$this->items[] = $item;
|
|
|
|
$limit++;
|
|
|
|
}
|
2015-05-05 21:18:12 +03:00
|
|
|
}
|
2016-08-04 13:34:40 +03:00
|
|
|
}
|
2015-05-05 21:18:12 +03:00
|
|
|
}
|