rss-bridge/bridges/NextgovBridge.php

71 lines
2.6 KiB
PHP
Raw Normal View History

2016-07-26 22:20:13 +03:00
<?php
class NextgovBridge extends FeedExpander {
2016-07-26 22:20:13 +03:00
const MAINTAINER = 'ORelio';
const NAME = 'Nextgov Bridge';
const URI = 'https://www.nextgov.com/';
const DESCRIPTION = 'USA Federal technology news, best practices, and web 2.0 tools.';
2016-07-26 22:20:13 +03:00
const PARAMETERS = array( array(
'category'=>array(
'name'=>'Category',
'type'=>'list',
'values'=>array(
'All'=>'all',
'Technology News'=>'technology-news',
'CIO Briefing'=>'cio-briefing',
'Emerging Tech'=>'emerging-tech',
'Cloud'=>'cloud-computing',
'Cybersecurity'=>'cybersecurity',
'Mobile'=>'mobile',
'Health'=>'health',
'Defense'=>'defense',
'Big Data'=>'big-data'
)
)
));
2016-07-26 22:20:13 +03:00
public function collectData(){
$this->collectExpandableDatas(self::URI . 'rss/' . $this->getInput('category') . '/', 10);
}
2016-07-26 22:20:13 +03:00
protected function parseItem($newsItem){
$item = parent::parseItem($newsItem);
2016-07-26 22:20:13 +03:00
$item['content'] = '';
2016-07-26 22:20:13 +03:00
$namespaces = $newsItem->getNamespaces(true);
if(isset($namespaces['media'])){
$media = $newsItem->children($namespaces['media']);
if(isset($media->content)){
$attributes = $media->content->attributes();
$item['content'] = '<img src="' . $attributes['url'] . '">';
}
}
2016-07-26 22:20:13 +03:00
$item['content'] .= $this->ExtractContent($item['uri']);
return $item;
}
private function StripWithDelimiters($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 $string;
}
2016-07-26 22:20:13 +03:00
private function ExtractContent($url){
$article = $this->getSimpleHTMLDOMCached($url)
or $this->returnServerError('Could not request Nextgov: ' . $url);
2016-07-26 22:20:13 +03:00
$contents = $article->find('div.wysiwyg', 0)->innertext;
$contents = $this->StripWithDelimiters($contents, '<div class="ad-container">', '</div>');
$contents = $this->StripWithDelimiters($contents, '<div', '</div>'); //ad outer div
return $this->StripWithDelimiters($contents, '<script', '</script>');
$contents = ($article_thumbnail == '' ? '' : '<p><img src="'.$article_thumbnail.'" /></p>')
.'<p><b>'.$article_subtitle.'</b></p>'
.trim($contents);
2016-07-26 22:20:13 +03:00
}
}