Rss-Bridge/bridges/ReporterreBridge.php
Pierre Mazière 3c0d13c1bb bridges: use BridgeAbstract::getSimpleHTMLDOM
instead of BridgeAbstract::file_get_html

Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
2016-08-19 10:20:27 +02:00

45 lines
1.4 KiB
PHP

<?php
class ReporterreBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "nyutag";
$this->name = "Reporterre Bridge";
$this->uri = "http://www.reporterre.net/";
$this->description = "Returns the newest articles.";
$this->update = '2016-08-17';
}
private function ExtractContentReporterre($url) {
$html2 = $this->getSimpleHTMLDOM($url);
foreach($html2->find('div[style=text-align:justify]') as $e) {
$text = $e->outertext;
}
$html2->clear();
unset ($html2);
// Replace all relative urls with absolute ones
$text = preg_replace('/(href|src)(\=[\"\'])(?!http)([^"\']+)/ims', "$1$2" . $this->uri . "$3", $text);
$text = strip_tags($text, '<p><br><a><img>');
return $text;
}
public function collectData(array $param){
$html = $this->getSimpleHTMLDOM('http://www.reporterre.net/spip.php?page=backend') or $this->returnServerError('Could not request Reporterre.');
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
$item = new \Item();
$item->title = html_entity_decode($element->find('title', 0)->plaintext);
$item->timestamp = strtotime($element->find('dc:date', 0)->plaintext);
$item->uri = $element->find('guid', 0)->innertext;
$item->content = html_entity_decode($this->ExtractContentReporterre($item->uri));
$this->items[] = $item;
$limit++;
}
}
}
}