rss-bridge/bridges/LeMondeInformatiqueBridge.php
Eugene Molotov 37cb4091d4
bridges: remove redundant "or returnServerError" after getContents/getSimpleHTMLDom/getSimpleHTMLDomCached (#2398)
When fetching website contents, exceptions already raise on fetching error
2022-01-02 14:36:09 +05:00

39 lines
1.3 KiB
PHP

<?php
class LeMondeInformatiqueBridge extends FeedExpander {
const MAINTAINER = 'ORelio';
const NAME = 'Le Monde Informatique';
const URI = 'https://www.lemondeinformatique.fr/';
const DESCRIPTION = 'Returns the newest articles.';
public function collectData(){
$this->collectExpandableDatas(self::URI . 'rss/rss.xml', 10);
}
protected function parseItem($newsItem){
$item = parent::parseItem($newsItem);
$article_html = getSimpleHTMLDOMCached($item['uri']);
//Deduce thumbnail URL from article image URL
$item['enclosures'] = array(
str_replace(
'/grande/',
'/petite/',
$article_html->find('.article-image > img, figure > img', 0)->src
)
);
//No response header sets the encoding, explicit conversion is needed or subsequent xml_encode() will fail
$content_node = $article_html->find('div.col-primary, div.col-sm-9', 0);
$item['content'] = $this->cleanArticle($content_node->innertext);
$item['author'] = $article_html->find('div.author-infos', 0)->find('b', 0)->plaintext;
return $item;
}
private function cleanArticle($article_html){
$article_html = stripWithDelimiters($article_html, '<script', '</script>');
$article_html = explode('<p class="contact-error', $article_html)[0] . '</div>';
return $article_html;
}
}