mirror of
https://github.com/RSS-Bridge/rss-bridge.git
synced 2025-08-11 11:04:36 +02:00
bridges: use BridgeAbstract::getSimpleHTMLDOM
instead of BridgeAbstract::file_get_html Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
@@ -12,49 +12,49 @@ class CourrierInternationalBridge extends BridgeAbstract{
|
||||
}
|
||||
|
||||
public function collectData(array $param){
|
||||
|
||||
|
||||
$html = '';
|
||||
|
||||
$html = $this->file_get_html('http://www.courrierinternational.com/') or $this->returnServerError('Error.');
|
||||
|
||||
$html = $this->getSimpleHTMLDOM('http://www.courrierinternational.com/') or $this->returnServerError('Error.');
|
||||
|
||||
|
||||
|
||||
|
||||
$element = $html->find("article");
|
||||
|
||||
$article_count = 1;
|
||||
$article_count = 1;
|
||||
|
||||
foreach($element as $article) {
|
||||
|
||||
|
||||
$item = new \Item();
|
||||
|
||||
|
||||
$item->uri = $article->parent->getAttribute("href");
|
||||
|
||||
if(strpos($item->uri, "http") === FALSE) {
|
||||
$item->uri = "http://courrierinternational.fr/".$item->uri;
|
||||
}
|
||||
|
||||
$page = $this->file_get_html($item->uri);
|
||||
|
||||
$page = $this->getSimpleHTMLDOM($item->uri);
|
||||
|
||||
$cleaner = new HTMLSanitizer();
|
||||
|
||||
|
||||
$item->content = $cleaner->sanitize($page->find("div.article-text")[0]);
|
||||
$item->title = strip_tags($article->find(".title")[0]);
|
||||
|
||||
$dateTime = date_parse($page->find("time")[0]);
|
||||
|
||||
$item->timestamp = mktime(
|
||||
$dateTime['hour'],
|
||||
$dateTime['minute'],
|
||||
$dateTime['second'],
|
||||
$dateTime['month'],
|
||||
$dateTime['day'],
|
||||
$dateTime['hour'],
|
||||
$dateTime['minute'],
|
||||
$dateTime['second'],
|
||||
$dateTime['month'],
|
||||
$dateTime['day'],
|
||||
$dateTime['year']
|
||||
);
|
||||
|
||||
|
||||
$this->items[] = $item;
|
||||
$article_count ++;
|
||||
if($article_count > 5) break;
|
||||
|
||||
|
||||
}
|
||||
|
||||
|
||||
|
Reference in New Issue
Block a user