1
0
Fork 0
forked from blallo/rss-bridge
rss-bridge/bridges/SiliconBridge.php

62 lines
2.6 KiB
PHP
Raw Normal View History

2015-10-22 14:52:34 +02:00
<?php
class SiliconBridge extends BridgeAbstract {
public $maintainer = "ORelio";
public $name = 'Silicon Bridge';
public $uri = 'http://www.silicon.fr/';
public $description = "Returns the newest articles.";
public function collectData(){
2015-10-22 14:52:34 +02:00
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
$feedUrl = $this->uri.'feed';
$html = $this->getSimpleHTMLDOM($feedUrl)
or $this->returnServerError('Could not request Silicon: '.$feedUrl);
2015-10-22 14:52:34 +02:00
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
//Retrieve article Uri and get that page
$article_uri = $element->innertext;
$article_uri = substr($article_uri, strpos($article_uri, '<link>') + 6);
$article_uri = substr($article_uri, 0, strpos($article_uri, '</link>'));
$article_html = $this->getSimpleHTMLDOM($article_uri)
or $this->returnServerError('Could not request Silicon: '.$article_uri);
2015-10-22 14:52:34 +02:00
//Build article contents from corresponding elements
$thumbnailUri = $element->find('enclosure', 0)->url;
$article_content = '<p><img src="'.$thumbnailUri.'" /></p>'
.'<p><b>'.$article_html->find('div.entry-excerpt', 0)->plaintext.'</b></p>'
.$article_html->find('div.entry-content', 0)->innertext;
//Remove useless scripts left in the page
while (strpos($article_content, '<script') !== false) {
$script_section = substr($article_content, strpos($article_content, '<script'));
$script_section = substr($script_section, 0, strpos($script_section, '</script>') + 9);
$article_content = str_replace($script_section, '', $article_content);
}
//Build and add final item
$item = array();
$item['uri'] = $article_uri;
$item['title'] = StripCDATA($element->find('title', 0)->innertext);
$item['author'] = StripCDATA($element->find('dc:creator', 0)->innertext);
$item['timestamp'] = strtotime($element->find('pubDate', 0)->plaintext);
$item['content'] = $article_content;
2015-10-22 14:52:34 +02:00
$this->items[] = $item;
$limit++;
}
}
}
public function getCacheDuration() {
return 1800; // 30 minutes
}
}