boyska/rss-bridge
1
0
Fork 0
forked from blallo/rss-bridge
Code Releases Activity

[WordPressBridge] make it FeedExpander

Browse source 
Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
Pierre Mazière 2016-09-14 23:01:54 +02:00
parent 2744c13735
commit b5358d6cea
1 changed files with 34 additions and 100 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

114
bridges/WordPressBridge.php
View file

@ -1,10 +1,7 @@
<?php <?php
define('WORDPRESS_TYPE_ATOM', 1); // Content is of type ATOM define('WORDPRESS_TYPE_ATOM', 1); // Content is of type ATOM
define('WORDPRESS_TYPE_RSS', 2); // Content is of type RSS define('WORDPRESS_TYPE_RSS', 2); // Content is of type RSS
class WordPressBridge extends BridgeAbstract { class WordPressBridge extends FeedExpander {
public $sitename; // Name of the site
const MAINTAINER = "aledeg"; const MAINTAINER = "aledeg";
const NAME = "Wordpress Bridge"; const NAME = "Wordpress Bridge";
const URI = "https://wordpress.org/"; const URI = "https://wordpress.org/";
@ -17,112 +14,49 @@ class WordPressBridge extends BridgeAbstract {
) )
)); ));
// Replaces all 'link' tags with 'url' for simplehtmldom to actually find 'links' ('url') private function clearContent($content) {
private function ReplaceLinkTagsWithUrlTags($element){
// We need to fix the 'link' tag as simplehtmldom cannot parse it (just rename it and load back as dom)
$element_text = $element->outertext;
$element_text = str_replace('<link>', '<url>', $element_text);
$element_text = str_replace('</link>', '</url>', $element_text);
$element_text = str_replace('<link ', '<url ', $element_text);
return str_get_html($element_text);
}
private function StripCDATA($string) {
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
private function ClearContent($content) {
$content = preg_replace('/<script[^>]*>[^<]*<\/script>/', '', $content); $content = preg_replace('/<script[^>]*>[^<]*<\/script>/', '', $content);
$content = preg_replace('/<div class="wpa".*/', '', $content); $content = preg_replace('/<div class="wpa".*/', '', $content);
$content = preg_replace('/<form.*\/form>/', '', $content); $content = preg_replace('/<form.*\/form>/', '', $content);
return $content; return $content;
} }
public function collectData(){ protected function parseItem($newItem){
$item=parent::parseItem($newItem);
$html = $this->getSimpleHTMLDOM($this->getURI().'/feed/atom')
or $this->returnServerError("Could not request ".$this->getURI().'/feed/atom');
// Notice: We requested an ATOM feed, however some sites return RSS feeds instead!
if($html->find('entry')){
$type=WORDPRESS_TYPE_ATOM;
}else if($html->find('item')){
$type=WORDPRESS_TYPE_RSS;
}else{
$type=WORDPRESS_TYPE_ATOM; // Make ATOM default
}
if($type === WORDPRESS_TYPE_RSS)
$posts = $html->find('item');
else
$posts = $html->find('entry');
if(!empty($posts) ) {
$this->sitename = $html->find('title', 0)->plaintext;
$i=0;
foreach ($posts as $article) {
if($i < 3) {
$item = array();
$article = $this->ReplaceLinkTagsWithUrlTags($article);
if($type === WORDPRESS_TYPE_RSS){
$item['uri'] = $article->find('url', 0)->innertext; // 'link' => 'url'!
$item['title'] = $article->find('title', 0)->plaintext;
$item['author'] = trim($this->StripCDATA($article->find('dc:creator', 0)->innertext));
$item['timestamp'] = strtotime($article->find('pubDate', 0)->innertext);
} else {
$item['uri'] = $article->find('url', 0)->getAttribute('href'); // 'link' => 'url'!
$item['title'] = $this->StripCDATA($article->find('title', 0)->plaintext);
$item['author'] = trim($article->find('author', 0)->innertext);
$item['timestamp'] = strtotime($article->find('updated', 0)->innertext);
}
$article_html = $this->getSimpleHTMLDOMCached($item['uri']); $article_html = $this->getSimpleHTMLDOMCached($item['uri']);
// Attempt to find most common content div $article=null;
if(!isset($item['content'])){ switch(true){
case !is_null($article_html->find('article',0)):
// most common content div
$article = $article_html->find('article', 0); $article = $article_html->find('article', 0);
if(!empty($article)){ break;
$item['content'] = $this->ClearContent($article->innertext); case !is_null($article_html->find('.single-content',0)):
}
}
// another common content div // another common content div
if(!isset($item['content'])){
$article = $article_html->find('.single-content', 0); $article = $article_html->find('.single-content', 0);
if(!empty($article)){ break;
$item['content'] = $this->ClearContent($article->innertext); case !is_null($article_html->find('.post',0)):
}
}
// for old WordPress themes without HTML5 // for old WordPress themes without HTML5
if(!isset($item['content'])){
$article = $article_html->find('.post', 0); $article = $article_html->find('.post', 0);
if(!empty($article)){ break;
$item['content'] = $this->ClearContent($article->innertext);
}
} }
$this->items[] = $item; if(!is_null($article)){
$i++; $item['content'] = $this->clearContent($article->innertext);
}
}
} else {
$this->returnServerError("Sorry, ".$this->getURI()." doesn't seem to be a Wordpress blog.");
}
} }
public function getURI(){ return $item;
return $this->getInput('url');
} }
public function getName() { public function collectData(){
return "{$this->sitename} - Wordpress Bridge"; if(substr($this->getInput(url),0,srlen('http'))!=='http'){
// just in case someone find a way to access local files by playing with the url
returnClientError('The url parameter must either refer to http or https protocol.');
}
$this->collectExpandableDatas($this->getURI().'/feed/atom');
} }
public function getCacheDuration() { public function getCacheDuration() {