Merge pull request #328 from LogMANOriginal/ReporterreBridge

Reporterre bridge
This commit is contained in:
Mitsu 2016-08-04 13:49:28 +02:00 committed by GitHub
commit 68be407564

View file

@ -2,54 +2,56 @@
class ReporterreBridge extends BridgeAbstract{ class ReporterreBridge extends BridgeAbstract{
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "nyutag"; $this->maintainer = "nyutag";
$this->name = "Reporterre Bridge"; $this->name = "Reporterre Bridge";
$this->uri = "http://www.reporterre.net/"; $this->uri = "http://www.reporterre.net/";
$this->description = "Returns the newest articles."; $this->description = "Returns the newest articles.";
$this->update = "2015-04-07"; $this->update = "2016-08-04";
} }
public function collectData(array $param){
function ExtractContentReporterre($url) { function ExtractContentReporterre($url) {
$html2 = $this->file_get_html($url); $html2 = $this->file_get_html($url);
foreach($html2->find('div[style=text-align:justify]') as $e) {
$text = $e->outertext; foreach($html2->find('div[style=text-align:justify]') as $e) {
} $text = $e->outertext;
$html2->clear(); }
unset ($html2);
return $text; $html2->clear();
unset ($html2);
// Replace all relative urls with absolute ones
$text = preg_replace('/(href|src)(\=[\"\'])(?!http)([^"\']+)/ims', "$1$2" . $this->getURI() . "$3", $text);
$text = strip_tags($text, '<p><br><a><img>');
return $text;
} }
public function collectData(array $param){
$html = $this->file_get_html('http://www.reporterre.net/spip.php?page=backend') or $this->returnError('Could not request Reporterre.', 404); $html = $this->file_get_html('http://www.reporterre.net/spip.php?page=backend') or $this->returnError('Could not request Reporterre.', 404);
$limit = 0; $limit = 0;
foreach($html->find('item') as $element) { foreach($html->find('item') as $element) {
if($limit < 5) { if($limit < 5) {
$item = new \Item(); $item = new \Item();
$item->title = html_entity_decode($element->find('title', 0)->plaintext); $item->title = html_entity_decode($element->find('title', 0)->plaintext);
$item->timestamp = strtotime($element->find('dc:date', 0)->plaintext); $item->timestamp = strtotime($element->find('dc:date', 0)->plaintext);
$item->uri = $element->find('guid', 0)->innertext; $item->uri = $element->find('guid', 0)->innertext;
$item->content = html_entity_decode(ExtractContentReporterre($item->uri)); $item->content = html_entity_decode($this->ExtractContentReporterre($item->uri));
$this->items[] = $item; $this->items[] = $item;
$limit++; $limit++;
} }
} }
}
} public function getName(){
return 'Reporterre Bridge';
}
public function getName(){ public function getURI(){
return 'Reporterre Bridge'; return 'http://www.reporterre.net/';
} }
public function getURI(){ public function getCacheDuration(){
return 'http://www.reporterre.net/'; return 3600; // 1 hours
} }
public function getCacheDuration(){
return 3600; // 1 hours
// return 0;
}
} }