mirror/php-rss-bridge
1
0
Fork 0
mirror of https://github.com/RSS-Bridge/rss-bridge.git synced 2025-08-26 09:34:46 +02:00
Code Issues Releases Wiki Activity

[bridges] Change to extend from FeedExpander

Browse Source
This commit is contained in:
logmanoriginal
2016-09-05 18:43:56 +02:00
parent 11be7ccb60
commit 179e73fb80
12 changed files with 262 additions and 437 deletions
Download Patch File Download Diff File Expand all files Collapse all files

74
bridges/LeMondeInformatiqueBridge.php
View File

@@ -1,60 +1,42 @@
<?php
class LeMondeInformatiqueBridge extends BridgeAbstract {
class LeMondeInformatiqueBridge extends FeedExpander {
const MAINTAINER = "ORelio";
const NAME = "Le Monde Informatique";
const URI = "http://www.lemondeinformatique.fr/";
const DESCRIPTION = "Returns the newest articles.";
public function collectData(){
public function collectData(){
$this->collectExpandableDatas(self::URI . 'rss/rss.xml');
}
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
protected function parseItem($newsItem){
$item = $this->parseRSS_1_0_Item($newsItem);
$article_html = $this->get_cached($item['uri'])
or $this->returnServerError('Could not request LeMondeInformatique: ' . $item['uri']);
$item['content'] = $this->CleanArticle($article_html->find('div#article', 0)->innertext);
$item['title'] = $article_html->find('h1.cleanprint-title', 0)->plaintext;
return $item;
}
function StripWithDelimiters($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 $string;
}
function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function CleanArticle($article_html) {
$article_html = StripWithDelimiters($article_html, '<script', '</script>');
$article_html = StripWithDelimiters($article_html, '<h1 class="cleanprint-title"', '</h1>');
return $article_html;
}
function StripWithDelimiters($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;
}
$html = $this->getSimpleHTMLDOM(self::URI.'rss/rss.xml')
or $this->returnServerError('Could not request LeMondeInformatique: '
.self::URI.'rss/rss.xml');
$limit = 0;
foreach($html->find('item') as $element) {
if($limit < 5) {
//Retrieve article details
$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 $this->returnServerError('Could not request LeMondeInformatique: '.$article_uri);
$article_content = CleanArticle($article_html->find('div#article', 0)->innertext);
$article_title = $article_html->find('h1.cleanprint-title', 0)->plaintext;
//Build and add final item
$item = array();
$item['uri'] = $article_uri;
$item['title'] = $article_title;
$item['author'] = StripCDATA($element->find('dc:creator', 0)->innertext);
$item['timestamp'] = strtotime($element->find('dc:date', 0)->plaintext);
$item['content'] = $article_content;
$this->items[] = $item;
$limit++;
}
}
function CleanArticle($article_html) {
$article_html = $this->StripWithDelimiters($article_html, '<script', '</script>');
$article_html = $this->StripWithDelimiters($article_html, '<h1 class="cleanprint-title"', '</h1>');
return $article_html;
}
public function getCacheDuration() {