cuqmbr/rss-bridge
1
0
Fork 0
mirror of https://github.com/RSS-Bridge/rss-bridge.git synced 2025-04-23 23:16:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

Extend support for wordpress.com

Browse Source 
Refactor the code to make it more readable.
Change parameters to be more flexible.
Add crawling through articles to get the full content of it.
Add a cleaning method on content to remove scripts (bye bye google tracking)
This commit is contained in:
Alexis Degrugillier 2014-04-25 19:36:33 -04:00
parent 077f0e3fb1
commit cdca40c2bb
1 changed files with 39 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
bridges/WordPressBridge.php
View File

@ -5,62 +5,75 @@
*
* @name Wordpress Bridge
* @description Returns the newest articles of a blog hosted on wordpress
* @use1(s="subdomain", f="folder")
* @use1(url="blog URL", name="blog name")
*/
class WordpressBridge extends BridgeAbstract{
private $subdomain;
private $folder;
private $url;
private $name;
public function collectData(array $param){
$this->processParams($param);
if (!$this->hasSubdomain()) {
$this->returnError('You must specify a subdomain', 400);
if (!$this->hasUrl()) {
$this->returnError('You must specify a URL', 400);
}
$html = file_get_html($this->getSiteURI()) or $this->returnError("Could not request {$this->getSiteURI()}.", 404);
$html = file_get_html($this->url) or $this->returnError("Could not request {$this->url}.", 404);
foreach($html->find('.post') as $article) {
$item = new \Item();
$uri = $article->find('a[rel=bookmark]',0)->href;
$item->uri = $uri;
$item->title = $article->find('h2',0)->innertext;
$item->content = $article->find('.entry',0)->innertext;
preg_match('/\d{4}\/\d{2}\/\d{2}/', $uri, $matches);
$date = new \DateTime($matches[0]);
$item->timestamp = $date->format('U');
$this->items[] = $item;
$uri = $article->find('a',0)->href;
$this->items[] = $this->getDetails($uri);
}
}
private function getDetails($uri) {
$html = file_get_html($uri) or exit;
$item = new \Item();
$article = $html->find('.post', 0);
$item->uri = $uri;
$item->title = $article->find('h1',0)->innertext;
$item->content = $this->clearContent($article->find('.entry-content,.entry',0)->innertext);
$item->timestamp = $this->getDate($uri);
return $item;
}
private function clearContent($content) {
$content = preg_replace('/<script.*\/script>/', '', $content);
return $content;
}
private function getDate($uri) {
preg_match('/\d{4}\/\d{2}\/\d{2}/', $uri, $matches);
$date = new \DateTime($matches[0]);
return $date->format('U');
}
public function getName(){
return 'Wordpress.com Bridge';
return "{$this->name} - Wordpress.com Bridge";
}
public function getURI(){
return 'http://%s.wordpress.com/%s';
return $this->url;
}
public function getCacheDuration(){
return 21600; // 6 hours
}
private function getSiteURI(){
return sprintf($this->getURI(), $this->subdomain, $this->folder);
}
private function hasSubdomain(){
if (empty($this->subdomain)){
private function hasUrl(){
if (empty($this->url)){
return false;
}
return true;
}
private function processParams($param){
$this->subdomain = $param['s'];
$this->folder = $param['f'];
$this->url = $param['url'];
$this->name = $param['name'];
}
}