rss-bridge/bridges/SiliconBridge.php

76 lines
2.8 KiB
PHP
Raw Normal View History

2015-10-22 14:52:34 +02:00
<?php
/**
* Silicon Bridge
* Returns the newest articles
* 2015-09-08
*
* @name Silicon.fr
* @homepage http://www.silicon.fr/
* @description Returns the newest articles.
* @maintainer ORelio
* @update 2015-09-08
*/
class SiliconBridge extends BridgeAbstract {
public function collectData(array $param) {
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
$feedUrl = 'http://www.silicon.fr/feed';
2015-10-22 15:20:42 +02:00
$html = file_get_html($feedUrl) or $this->returnError('Could not request Silicon: '.$feedUrl, 500);
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>'));
2015-10-22 15:20:42 +02:00
$article_html = file_get_html($article_uri) or $this->returnError('Could not request Silicon: '.$article_uri, 500);
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 = new \Item();
$item->uri = $article_uri;
$item->thumbnailUri = $thumbnailUri;
$item->title = StripCDATA($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;
$this->items[] = $item;
$limit++;
}
}
}
public function getName() {
return 'Silicon Bridge';
}
public function getURI() {
return 'http://www.silicon.fr/';
}
public function getCacheDuration() {
return 1800; // 30 minutes
// return 0;
}
}