Fix bridge to work with new layout

This commit is contained in:
logmanoriginal 2016-08-03 22:39:03 +02:00
parent 6ff73d47bb
commit ec5cb657aa

View file

@ -2,49 +2,48 @@
class NiceMatinBridge extends BridgeAbstract{ class NiceMatinBridge extends BridgeAbstract{
public function loadMetadatas() { public function loadMetadatas() {
$this->maintainer = "pit-fgfjiudghdf"; $this->maintainer = "pit-fgfjiudghdf";
$this->name = "NiceMatin"; $this->name = "NiceMatin";
$this->uri = "http://www.nicematin.com/"; $this->uri = "http://www.nicematin.com/";
$this->description = "Returns the 10 newest posts from NiceMatin (full text)"; $this->description = "Returns the 10 newest posts from NiceMatin (full text)";
$this->update = "2014-05-26"; $this->update = "2016-08-03";
}
public function collectData(array $param){
function NiceMatinUrl($string) {
$string = str_replace('</link>', '', $string);
//$string = str_replace('.+', '', $string);
$string = preg_replace('/html.*http.*/i','html',$string);
$string = preg_replace('/.*http/i','http',$string);
return $string;
} }
function NiceMatinExtractContent($url) { function NiceMatinExtractContent($url) {
$html2 = $this->file_get_html($url); $html = $this->file_get_html($url);
$text = $html2->find('figure[itemprop=associatedMedia]', 0)->innertext; if(!$html)
$text .= $html2->find('div[id=content-article]', 0)->innertext; $this->returnError('Could not acquire content from url: ' . $url . '!', 404);
$content = $html->find('article', 0);
if(!$content)
$this->returnError('Could not find \'section\'!', 404);
$text = preg_replace('#<script(.*?)>(.*?)</script>#is', '', $content->innertext);
$text = strip_tags($text, '<p><a><img>');
return $text; return $text;
} }
public function collectData(array $param){
$html = $this->file_get_html('http://www.nicematin.com/derniere-minute/rss') or $this->returnError('Could not request NiceMatin.', 404); $html = $this->file_get_html('http://www.nicematin.com/derniere-minute/rss') or $this->returnError('Could not request NiceMatin.', 404);
$limit = 0; $limit = 0;
foreach($html->find('item') as $element) { foreach($html->find('item') as $element) {
if($limit < 10) { if($limit < 10) {
$item = new \Item(); // We need to fix the 'link' tag as simplehtmldom cannot parse it (just rename it and load back as dom)
//$item->title = NiceMatinStripCDATA($element->find('title', 0)->innertext); $element_text = $element->outertext;
$item->title = $element->find('title', 0)->innertext; $element_text = str_replace('<link>', '<url>', $element_text);
$item->uri = NiceMatinUrl($element->plaintext); $element_text = str_replace('</link>', '</url>', $element_text);
$element = str_get_html($element_text);
$item = new \Item();
$item->title = $element->find('title', 0)->innertext;
$item->uri = $element->find('url', 0)->innertext;
$item->timestamp = strtotime($element->find('pubDate', 0)->plaintext); $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
$item->content = NiceMatinExtractContent($item->uri); $item->content = $this->NiceMatinExtractContent($item->uri);
$this->items[] = $item; $this->items[] = $item;
$limit++; $limit++;
} }
} }
} }
public function getName(){ public function getName(){
@ -59,4 +58,3 @@ class NiceMatinBridge extends BridgeAbstract{
return 3600; // 1 hour return 3600; // 1 hour
} }
} }