2016-05-03 23:42:34 +02:00
|
|
|
<?php
|
|
|
|
class NakedSecurityBridge extends BridgeAbstract {
|
|
|
|
|
2016-08-30 11:23:55 +02:00
|
|
|
const MAINTAINER = 'ORelio';
|
|
|
|
const NAME = 'Naked Security';
|
|
|
|
const URI = 'https://nakedsecurity.sophos.com/';
|
|
|
|
const DESCRIPTION = 'Returns the newest articles.';
|
2016-05-03 23:42:34 +02:00
|
|
|
|
2016-08-25 01:24:53 +02:00
|
|
|
public function collectData(){
|
2016-05-03 23:42:34 +02:00
|
|
|
|
|
|
|
function StripRecursiveHTMLSection($string, $tag_name, $tag_start) {
|
|
|
|
$open_tag = '<'.$tag_name;
|
|
|
|
$close_tag = '</'.$tag_name.'>';
|
|
|
|
$close_tag_length = strlen($close_tag);
|
|
|
|
if (strpos($tag_start, $open_tag) === 0) {
|
|
|
|
while (strpos($string, $tag_start) !== false) {
|
|
|
|
$max_recursion = 100;
|
|
|
|
$section_to_remove = null;
|
|
|
|
$section_start = strpos($string, $tag_start);
|
|
|
|
$search_offset = $section_start;
|
|
|
|
do {
|
|
|
|
$max_recursion--;
|
|
|
|
$section_end = strpos($string, $close_tag, $search_offset);
|
|
|
|
$search_offset = $section_end + $close_tag_length;
|
|
|
|
$section_to_remove = substr($string, $section_start, $section_end - $section_start + $close_tag_length);
|
|
|
|
$open_tag_count = substr_count($section_to_remove, $open_tag);
|
|
|
|
$close_tag_count = substr_count($section_to_remove, $close_tag);
|
|
|
|
} while ($open_tag_count > $close_tag_count && $max_recursion > 0);
|
|
|
|
$string = str_replace($section_to_remove, '', $string);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return $string;
|
|
|
|
}
|
|
|
|
|
|
|
|
$feedUrl = 'https://feeds.feedburner.com/nakedsecurity?format=xml';
|
2016-07-08 19:06:35 +02:00
|
|
|
$html = $this->getSimpleHTMLDOM($feedUrl) or $this->returnServerError('Could not request '.$this->getName().': '.$feedUrl);
|
2016-05-03 23:42:34 +02:00
|
|
|
$limit = 0;
|
|
|
|
|
|
|
|
foreach ($html->find('item') as $element) {
|
|
|
|
if ($limit < 10) {
|
|
|
|
|
|
|
|
//Retrieve article Uri and get that page
|
|
|
|
$article_uri = $element->find('guid', 0)->plaintext;
|
2016-07-08 19:06:35 +02:00
|
|
|
$article_html = $this->getSimpleHTMLDOM($article_uri) or $this->returnServerError('Could not request '.$this->getName().': '.$article_uri);
|
2016-05-03 23:42:34 +02:00
|
|
|
|
|
|
|
//Build article contents from corresponding elements
|
|
|
|
$article_title = trim($element->find('title', 0)->plaintext);
|
|
|
|
$article_image = $article_html->find('img.wp-post-image', 0)->src;
|
|
|
|
$article_summary = strip_tags(html_entity_decode($element->find('description', 0)->plaintext));
|
|
|
|
$article_content = $article_html->find('div.entry-content', 0)->innertext;
|
|
|
|
$article_content = StripRecursiveHTMLSection($article_content , 'div', '<div class="entry-prefix"');
|
|
|
|
$article_content = StripRecursiveHTMLSection($article_content , 'script', '<script');
|
|
|
|
$article_content = StripRecursiveHTMLSection($article_content , 'aside', '<aside');
|
|
|
|
$article_content = '<p><img src="'.$article_image.'" /></p><p><b>'.$article_summary.'</b></p>'.$article_content;
|
|
|
|
|
|
|
|
//Build and add final item
|
2016-08-22 18:55:59 +02:00
|
|
|
$item = array();
|
|
|
|
$item['uri'] = $article_uri;
|
|
|
|
$item['title'] = $article_title;
|
|
|
|
$item['author'] = $article_html->find('a[rel=author]', 0)->plaintext;
|
|
|
|
$item['timestamp'] = strtotime($element->find('pubDate', 0)->plaintext);
|
|
|
|
$item['content'] = $article_content;
|
2016-05-03 23:42:34 +02:00
|
|
|
$this->items[] = $item;
|
|
|
|
$limit++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2016-07-08 19:06:35 +02:00
|
|
|
}
|