rss-bridge/bridges/ArstechnicaBridge.php

67 lines
2.3 KiB
PHP
Raw Normal View History

2016-08-02 14:24:54 +02:00
<?php
#ini_set('display_errors', 'On');
#error_reporting(E_ALL);
class ArstechnicaBridge extends BridgeAbstract {
const MAINTAINER = "prysme";
const NAME = "ArstechnicaBridge";
const URI = "http://arstechnica.com";
const DESCRIPTION = "The PC enthusiast's resource. Power users and the tools they love, without computing religion";
2016-08-02 14:24:54 +02:00
function StripWithDelimiters($string, $start, $end) {
while (strpos($string, $start) !== false) {
$section_to_remove = substr($string, strpos($string, $start));
$section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
$string = str_replace($section_to_remove, '', $string);
} return $string;
}
2016-08-02 14:24:54 +02:00
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
2016-08-02 14:24:54 +02:00
function ExtractContent($url) {
#echo $url;
$html2 = $this->getSimpleHTMLDOM($url);
$text = $html2->find("section[id='article-guts']", 0);
2016-08-02 14:24:54 +02:00
/*foreach ($text->find('<aside id="social-left">') as $node)
{ $node = NULL; }*/
$text = $this->StripWithDelimiters($text->innertext,'<aside id="social-left">','</aside>');
$text = $this->StripWithDelimiters($text,'<figcaption class="caption">','</figcaption>');
$text = $this->StripWithDelimiters($text,'<div class="gallery shortcode-gallery">','</div>');
//error_log("ICI", 0);
//error_log($text, 0);
2016-08-02 14:24:54 +02:00
return $text;
}
2016-08-02 14:24:54 +02:00
public function collectData(){
2016-08-02 14:24:54 +02:00
$html = $this->getSimpleHTMLDOM('http://feeds.arstechnica.com/arstechnica/index') or $this->returnServerError('Could not request NextInpact.');
$limit = 0;
2016-08-02 14:24:54 +02:00
foreach($html->find('item') as $element) {
if($limit < 5) {
$item = array();
$item['title'] = $this->StripCDATA($element->find('title', 0)->innertext);
$item['uri'] = $this->StripCDATA($element->find('guid', 0)->plaintext);
$item['author'] = $this->StripCDATA($element->find('author', 0)->innertext);
$item['timestamp'] = strtotime($element->find('pubDate', 0)->plaintext);
$item['content'] = $this->ExtractContent($item['uri']);
//$item['content'] = $item['uri'];
$this->items[] = $item;
$limit++;
}
}
}
2016-08-02 14:24:54 +02:00
public function getCacheDuration() {
return 7200; // 2h
}
2016-08-02 14:24:54 +02:00
}