From c07eacfd6af5646c795e5739e8d39964002fa87f Mon Sep 17 00:00:00 2001
From: Mitsukarenai <mitsu@suumitsu.eu>
Date: Mon, 26 May 2014 19:45:10 +0200
Subject: [PATCH] Add extra bridges, second batch (TPB inside)

---
 bridges/CollegeDeFranceBridge.php | 41 ++++++++++++++
 bridges/FlickrTagBridge.php       | 53 +++++++++++++++++
 bridges/KoreusBridge.php          | 56 ++++++++++++++++++
 bridges/MsnMondeBridge.php        | 55 ++++++++++++++++++
 bridges/NiceMatinBridge.php       | 62 ++++++++++++++++++++
 bridges/PlanetLibreBridge.php     | 49 ++++++++++++++++
 bridges/RaymondBridge.php         | 52 +++++++++++++++++
 bridges/Sexactu.php               | 91 ++++++++++++++++++++++++++++++
 bridges/ThePirateBayBridge.php    | 49 ++++++++++++++++
 bridges/WordPressBridge.php       | 94 +++++++++++++++++++++++++++++++
 10 files changed, 602 insertions(+)
 create mode 100644 bridges/CollegeDeFranceBridge.php
 create mode 100644 bridges/FlickrTagBridge.php
 create mode 100644 bridges/KoreusBridge.php
 create mode 100644 bridges/MsnMondeBridge.php
 create mode 100644 bridges/NiceMatinBridge.php
 create mode 100644 bridges/PlanetLibreBridge.php
 create mode 100644 bridges/RaymondBridge.php
 create mode 100644 bridges/Sexactu.php
 create mode 100644 bridges/ThePirateBayBridge.php
 create mode 100644 bridges/WordPressBridge.php

diff --git a/bridges/CollegeDeFranceBridge.php b/bridges/CollegeDeFranceBridge.php
new file mode 100644
index 0000000..6526eae
--- /dev/null
+++ b/bridges/CollegeDeFranceBridge.php
@@ -0,0 +1,41 @@
+<?php
+/**
+* RssBridgeCollegeDeFrance
+* Returns the 10 newest posts from http://www.college-de-france.fr
+*
+* @name CollegeDeFrance
+* @homepage http://www.college-de-france.fr/
+* @description Returns the 10 newest posts from CollegeDeFrance
+* @maintainer pit-fgfjiudghdf
+* @update 2014-05-26
+*/
+class CollegeDeFranceBridge extends BridgeAbstract{
+    public function collectData(array $param){
+$find = array('janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'novembre', 'décembre');
+$replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December');
+        $html = file_get_html('http://www.college-de-france.fr/site/audio-video/_audiovideos.jsp?index=0&prompt=&fulltextdefault=mots-cles...&fulltext=mots-cles...&fields=TYPE2_ACTIVITY&fieldsdefault=0_0&TYPE2=0') or $this->returnError('Could not request CollegeDeFrance.', 404);
+        $limit = 0;
+        foreach($html->find('li.audio') as $element) {
+         if($limit < 10) {
+         $item = new \Item();
+         $item->title = $element->find('span.title', 0)->plaintext;
+         $item->timestamp = strtotime(str_replace($find, $replace, $element->find('span.date', 0)->plaintext));
+         $item->content =  $element->find('span.lecturer', 0)->innertext . ' - ' . $element->find('span.title', 0)->innertext;
+         $item->uri = $element->find('a', 0)->href;
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+    public function getName(){
+        return 'CollegeDeFrance';
+    }
+    public function getURI(){
+        return 'http://www.college-de-france.fr/';
+    }
+    public function getCacheDuration(){
+        return 3600*3; // 3 hour
+    }
+}
+
diff --git a/bridges/FlickrTagBridge.php b/bridges/FlickrTagBridge.php
new file mode 100644
index 0000000..fa274b1
--- /dev/null
+++ b/bridges/FlickrTagBridge.php
@@ -0,0 +1,53 @@
+<?php
+/**
+* RssBridgeFlickrTagUser 
+* Returns the tagged images from http://www.flickr.com/
+* 2014-05-26
+*
+* @name Flickr TagUser
+* @homepage http://www.flickr.com/
+* @description Returns the tagged or user images from Flickr
+* @maintainer erwang
+* @use1(q="keyword")
+* @use2(u="username")
+*/
+class FlickrTagBridge extends BridgeAbstract{
+
+    public function collectData(array $param){
+        $html = file_get_html('http://www.flickr.com/search/?q=vendee&s=rec') or $this->returnError('Could not request Flickr.', 404);
+        if (isset($param['q'])) {   /* keyword search mode */
+            $this->request = $param['q'];
+            $html = file_get_html('http://www.flickr.com/search/?q='.urlencode($this->request).'&s=rec') or $this->returnError('No results for this query.', 404);
+        }
+        elseif (isset($param['u'])) {   /* user timeline mode */
+            $this->request = $param['u'];
+            $html = file_get_html('http://www.flickr.com/photos/'.urlencode($this->request).'/') or $this->returnError('Requested username can\'t be found.', 404);
+        }
+        
+        else {
+            $this->returnError('You must specify a keyword or a Flickr username.', 400);
+        }
+
+        foreach($html->find('span.photo_container') as $element) {
+            $item = new \Item();
+            $item->uri = 'http://flickr.com'.$element->find('a',0)->href;
+            $item->thumbnailUri = $element->find('img',0)->getAttribute('data-defer-src');
+            $item->content = '<a href="' . $item->uri . '"><img src="' . $item->thumbnailUri . '" /></a>'; // FIXME: Filter javascript ?
+            $item->title = $element->find('a',0)->title;
+            $this->items[] = $item;
+        }
+    }
+
+    public function getName(){
+        return 'Flickr Tag';
+    }
+
+    public function getURI(){
+        return 'http://www.flickr.com/search/';
+    }
+
+    public function getCacheDuration(){
+        return 21600; // 6 hours
+    }
+}
+
diff --git a/bridges/KoreusBridge.php b/bridges/KoreusBridge.php
new file mode 100644
index 0000000..8ae90e8
--- /dev/null
+++ b/bridges/KoreusBridge.php
@@ -0,0 +1,56 @@
+<?php
+/**
+* RssBridgeKoreus
+* Returns the 5 newest posts from Koreus (full text)
+*
+* @name Koreus
+* @homepage http://www.koreus.com/
+* @description Returns the 5 newest posts from Koreus (full text)
+* @maintainer pit-fgfjiudghdf
+* @update 2014-05-26
+*/
+class KoreusBridge extends BridgeAbstract{
+
+    public function collectData(array $param){
+
+    function KoreusStripCDATA($string) {
+        $string = str_replace('<![CDATA[', '', $string);
+        $string = str_replace(']]>', '', $string);
+        return $string;
+    }
+    function KoreusExtractContent($url) {
+        $html2 = file_get_html($url);
+        $text = $html2->find('p[class=itemText]', 0)->innertext;
+        $text = utf8_encode(preg_replace('/(Sur le m.+?)+$/i','',$text));
+        return $text;
+    }
+        $html = file_get_html('http://feeds.feedburner.com/Koreus-articles') or $this->returnError('Could not request Koreus.', 404);
+        $limit = 0;
+
+        foreach($html->find('item') as $element) {
+         if($limit < 5) {
+         $item = new \Item();
+         $item->title = KoreusStripCDATA($element->find('title', 0)->innertext);
+         $item->uri = KoreusStripCDATA($element->find('guid', 0)->plaintext);
+         $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
+         $item->content = KoreusExtractContent($item->uri);
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+
+    public function getName(){
+        return 'Koreus';
+    }
+
+    public function getURI(){
+        return 'http://www.koreus.com/';
+    }
+
+    public function getCacheDuration(){
+        return 3600; // 1 hour
+    }
+}
+
diff --git a/bridges/MsnMondeBridge.php b/bridges/MsnMondeBridge.php
new file mode 100644
index 0000000..bd7f5b1
--- /dev/null
+++ b/bridges/MsnMondeBridge.php
@@ -0,0 +1,55 @@
+<?php
+/**
+* RssBridgeMsnMonde 
+* Returns the 10 newest posts from MSN Actualités (full text)
+*
+* @name MSN Actu Monde
+* @homepage http://news.fr.msn.com/m6-actualite/monde/
+* @description Returns the 10 newest posts from MSN Actualités (full text)
+* @maintainer pit-fgfjiudghdf 
+* @update 2014-05-26
+*/
+class MsnMondeBridge extends BridgeAbstract{
+
+    public function collectData(array $param){
+
+    function MsnMondeExtractContent($url) {
+        $html2 = file_get_html($url);
+        $html2->find('div[id=m6_diaponews_placeholder]', 0)->outertext=''; //Supression de la partie "et aussi"
+        $text = $html2->find('div[class=svsubtorabs]', 0)->innertext; // ajout du resume
+        $text .= $html2->find('div[id=page1]', 0)->innertext;   // article
+        $text = preg_replace('/<p><strong>Lire aussi.*/i','',$text); //Supression de la partie "Lire aussi"
+
+        return $text;
+    }
+
+        $html = file_get_html('http://news.fr.msn.com/m6-actualite/RSS/News_RSS_Monde.aspx') or $this->returnError('Could not request MsnMonde.', 404);
+        $limit = 0;
+
+        foreach($html->find('item') as $element) {
+         if($limit < 10) {
+         $item = new \Item();
+         $item->title = $element->find('title', 0)->innertext;
+         $item->uri = $element->find('guid', 0)->plaintext;
+         $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
+         $item->content = MsnMondeExtractContent($item->uri);
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+
+    public function getName(){
+        return 'MSN Actu Monde';
+    }
+
+    public function getURI(){
+        return 'http://news.fr.msn.com/m6-actualite/monde/';
+    }
+
+    public function getCacheDuration(){
+        return 3600; // 1 hour
+    }
+}
+
diff --git a/bridges/NiceMatinBridge.php b/bridges/NiceMatinBridge.php
new file mode 100644
index 0000000..7c0441a
--- /dev/null
+++ b/bridges/NiceMatinBridge.php
@@ -0,0 +1,62 @@
+<?php
+/**
+* RssBridgeNiceMatin 
+* Returns the 10 newest posts from Nice Matin (full text)
+*
+* @name NiceMatin
+* @homepage http://www.nicematin.com/
+* @description Returns the 10 newest posts from NiceMatin (full text)
+* @maintainer pit-fgfjiudghdf
+* @update 2014-05-26
+*/
+class NiceMatinBridge extends BridgeAbstract{
+
+    public function collectData(array $param){
+
+    function NiceMatinUrl($string) {
+        $string = str_replace('</link>', '', $string);
+        //$string = str_replace('.+', '', $string);
+        $string = preg_replace('/html.*http.*/i','html',$string);
+        $string = preg_replace('/.*http/i','http',$string);
+        return $string;
+    }
+
+    function NiceMatinExtractContent($url) {
+        $html2 = file_get_html($url);
+        $text = $html2->find('figure[itemprop=associatedMedia]', 0)->innertext;
+        $text .= $html2->find('div[id=content-article]', 0)->innertext;
+        return $text;
+    }
+
+        $html = file_get_html('http://www.nicematin.com/derniere-minute/rss') or $this->returnError('Could not request NiceMatin.', 404);
+        $limit = 0;
+
+        foreach($html->find('item') as $element) {
+         if($limit < 10) {
+         $item = new \Item();
+         //$item->title = NiceMatinStripCDATA($element->find('title', 0)->innertext);
+         $item->title = $element->find('title', 0)->innertext;
+         $item->uri = NiceMatinUrl($element->plaintext);
+
+         $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
+         $item->content = NiceMatinExtractContent($item->uri);
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+
+    public function getName(){
+        return 'NiceMatin';
+    }
+
+    public function getURI(){
+        return 'http://www.nicematin.com/';
+    }
+
+    public function getCacheDuration(){
+        return 3600; // 1 hour
+    }
+}
+
diff --git a/bridges/PlanetLibreBridge.php b/bridges/PlanetLibreBridge.php
new file mode 100644
index 0000000..073ec28
--- /dev/null
+++ b/bridges/PlanetLibreBridge.php
@@ -0,0 +1,49 @@
+<?php
+/**
+* RssBridgePlanetLibre
+* Returns the 5 newest posts from PlanetLibre (full text)
+*
+* @name PlanetLibre
+* @homepage http://www.www.planet-libre.org
+* @description Returns the 5 newest posts from PlanetLibre (full text)
+* @maintainer pit-fgfjiudghdf 
+* @update 2014-05-26
+*/
+class PlanetLibreBridge extends BridgeAbstract{
+    public function collectData(array $param){
+    function PlanetLibreStripCDATA($string) {
+        $string = str_replace('<![CDATA[', '', $string);
+        $string = str_replace(']]>', '', $string);
+        return $string;
+    }
+    function PlanetLibreExtractContent($url) {
+        $html2 = file_get_html($url);
+        $text = $html2->find('div[class=post-text]', 0)->innertext;
+        return $text;
+    }
+        $html = file_get_html('http://www.planet-libre.org/rss10.php') or $this->returnError('Could not request PlanetLibre.', 404);
+        $limit = 0;
+        foreach($html->find('item') as $element) {
+         if($limit < 5) {
+         $item = new \Item();
+         $item->title = PlanetLibreStripCDATA($element->find('title', 0)->innertext);
+         $item->uri = PlanetLibreStripCDATA($element->find('guid', 0)->plaintext);
+         $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
+         $item->content = PlanetLibreExtractContent($item->uri);
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+    public function getName(){
+        return 'PlanetLibre';
+    }
+    public function getURI(){
+        return 'http://www.planet-libre.org/';
+    }
+    public function getCacheDuration(){
+        return 3600*2; // 1 hour
+    }
+}
+
diff --git a/bridges/RaymondBridge.php b/bridges/RaymondBridge.php
new file mode 100644
index 0000000..2d79ebb
--- /dev/null
+++ b/bridges/RaymondBridge.php
@@ -0,0 +1,52 @@
+<?php
+/**
+* RssBridgeRaymond
+* Returns the 3 newest posts from Raymond.cc (full text)
+*
+* @name Raymond
+* @homepage http://www.raymond.cc
+* @description Returns the 3 newest posts from Raymond.cc (full text)
+* @maintainer pit-fgfjiudghdf
+* @update 2014-05-26
+*/
+class RaymondBridge extends BridgeAbstract{
+    public function collectData(array $param){
+    function raymondStripCDATA($string) {
+        $string = str_replace('<![CDATA[', '', $string);
+        $string = str_replace(']]>', '', $string);
+        return $string;
+    }
+    function raymondExtractContent($url) {
+        $html2 = file_get_html($url);
+        $text = $html2->find('div.entry-content', 0)->innertext;
+	$text = preg_replace('/class="ad".*/', '', $text);
+	$text = strip_tags($text, '<p><a><i><strong><em><img>');
+	$text = str_replace('(adsbygoogle = window.adsbygoogle || []).push({});', '', $text);
+        return $text;
+    }
+        $html = file_get_html('http://www.raymond.cc/blog/feed') or $this->returnError('Could not request raymond.', 404);
+        $limit = 0;
+        foreach($html->find('item') as $element) {
+         if($limit < 3) {
+         $item = new \Item();
+         $item->title = raymondStripCDATA($element->find('title', 0)->innertext);
+         $item->uri = raymondStripCDATA($element->find('guid', 0)->plaintext);
+         $item->timestamp = strtotime($element->find('pubDate', 0)->plaintext);
+         $item->content = raymondExtractContent($item->uri);
+         $this->items[] = $item;
+         $limit++;
+         }
+        }
+
+    }
+    public function getName(){
+        return 'raymond';
+    }
+    public function getURI(){
+        return 'http://www.raymond.cc/blog';
+    }
+    public function getCacheDuration(){
+        return 3600*12; // 12 hour
+    }
+}
+
diff --git a/bridges/Sexactu.php b/bridges/Sexactu.php
new file mode 100644
index 0000000..6339095
--- /dev/null
+++ b/bridges/Sexactu.php
@@ -0,0 +1,91 @@
+<?php
+/**
+* 2014-05-26
+*
+* @name Sexactu
+* @homepage http://www.gqmagazine.fr/sexactu
+* @description Sexactu
+* @maintainer Riduidel
+* @update 04/02/2014
+*/
+define("GQ", "http://www.gqmagazine.fr");
+class Sexactu extends BridgeAbstract{
+
+    public function collectData(array $param){
+$find = array('janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'novembre', 'décembre');
+$replace = array('January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December');
+
+    $html = file_get_html($this->getURI()) or $this->returnError('Could not request '.$this->getURI(), 404);
+
+        foreach($html->find('.content-holder') as $contentHolder) {
+            // only use first list as second one only contains pages numbers 
+            $articles = $contentHolder->find('ul', 0);
+            foreach($articles->find('li') as $element) {
+                // if you ask about that method_exists, there seems to be a bug in simple html dom
+                // see stackoverflow for more details : http://stackoverflow.com/a/10828479/15619
+                if(is_object($element)) {
+                    $item = new Item();
+                    // various metadata
+                    $titleBlock = $element->find('.title-holder', 0);
+                    if(is_object($titleBlock)) {
+                        $titleDetails = $titleBlock->find('.article-title',0);
+                        $titleData = $titleDetails->find('h2', 0)->find('a',0);
+                        $titleTimestamp =$titleDetails->find('h4',0);
+                        $item->title = $this->correctCase(trim($titleData->innertext));
+                        $item->uri = GQ.$titleData->href;
+
+                        // Fugly date parsing due to the fact my DNS-323 doesn't support php intl extension
+                        $dateText = $titleTimestamp->innertext;
+                        $dateText = substr($dateText, strpos($dateText,',')+1);
+                        $dateText = str_replace($find, $replace, strtolower($dateText));
+                        $date = strtotime($dateText); 
+                        $item->timestamp = $date;
+
+                        $item->name = "Maïa Mazaurette";
+                        $elementText = $element->find('.text-container', 0);
+                        // don't forget to replace images server url with gq one
+                        foreach($elementText->find('img') as $image) {
+                            $image->src = GQ.$image->src;
+                        }
+                        $item->content = $elementText->innertext;
+                        $this->items[] = $item;
+                    }
+                    
+                }
+                
+            }
+        }
+    }
+
+    public function getName(){
+        return 'Sexactu';
+    }
+
+    public function getURI(){
+        return GQ.'/sexactu';
+    }
+
+    public function getCacheDuration(){
+        return 7200; // 2h hours
+    }
+    public function getDescription(){
+        return "Sexactu";
+    }
+    
+    public function correctCase($str) {
+        $sentences=explode('.', mb_strtolower($str, "UTF-8"));
+        $str="";
+        $sep="";
+        foreach ($sentences as $sentence)
+        {
+           //upper case first char
+           $sentence=ucfirst(trim($sentence));
+        
+           //append sentence to output
+           $str=$str.$sep.$sentence;
+           $sep=". ";
+        }
+        return $str;
+    }
+}
+
diff --git a/bridges/ThePirateBayBridge.php b/bridges/ThePirateBayBridge.php
new file mode 100644
index 0000000..c0ce493
--- /dev/null
+++ b/bridges/ThePirateBayBridge.php
@@ -0,0 +1,49 @@
+<?php
+/**
+* RssBridgeThePirateBay
+* Returns the newest interesting images from http://www.flickr.com/explore
+* 2014-05-25
+*
+* @name The Pirate Bay
+* @homepage https://thepiratebay.se/
+* @description Returns results for the keywords
+* @maintainer mitsukarenai
+* @update 2014-05-26
+* @use1(q="keywords")
+*/
+class ThePirateBayBridge extends BridgeAbstract{
+
+	public function collectData(array $param){
+
+		if (!isset($param['q']))
+			$this->returnError('You must specify a keyword (?q=...)', 400);
+
+        	$html = file_get_html('https://thepiratebay.se/search/'.rawurlencode($param['q']).'/0/99/0') or $this->returnError('Could not request TPB.', 404);
+
+		if($html->find('table#searchResult', 0) == FALSE)
+			$this->returnError('No result for this query', 404);
+
+		foreach($html->find('tr') as $element) {
+			$item = new \Item();
+			$item->uri = 'https://thepiratebay.se/'.$element->find('a.detLink',0)->href;
+			$item->id = $item->uri;
+			$item->timestamp = time();
+			$item->title = $element->find('a.detLink',0)->plaintext;
+			$item->content = $element->find('font',0)->plaintext.'<br><a href="'.$element->find('a',3)->href.'">download</a>';
+			if(!empty($item->title))
+				$this->items[] = $item;
+		}
+	}
+
+    public function getName(){
+        return 'The Pirate Bay';
+    }
+
+    public function getURI(){
+        return 'https://thepiratebay.se/';
+    }
+
+    public function getCacheDuration(){
+        return 3600; // 1 hour
+    }
+}
diff --git a/bridges/WordPressBridge.php b/bridges/WordPressBridge.php
new file mode 100644
index 0000000..c61fbb7
--- /dev/null
+++ b/bridges/WordPressBridge.php
@@ -0,0 +1,94 @@
+<?php
+
+/**
+ * RssBridgeWordpress
+ * Returns the 3 newest full posts of a Wordpress blog
+ *
+ * @name Wordpress Bridge
+* @homepage https://wordpress.com/
+ * @description Returns the 3 newest full posts of a Wordpress blog
+* @maintainer aledeg 
+* @update 2014-05-26
+ * @use1(url="blog URL (required)", name="blog name")
+ */
+class WordpressBridge extends BridgeAbstract {
+
+	private $url;
+	private $name;
+
+	public function collectData(array $param) {
+		$this->processParams($param);
+
+		if (!$this->hasUrl()) {
+			$this->returnError('You must specify a URL', 400);
+		}
+
+		$html = file_get_html($this->url) or $this->returnError("Could not request {$this->url}.", 404);
+
+		if(!empty($html->find('.post')) ) {
+			$i=0;
+			foreach ($html->find('.post') as $article) {
+				if($i < 3) {
+					$uri = $article->find('a', 0)->href;
+					$this->items[] = $this->getDetails($uri);
+					$i++;
+				}
+			}
+		}
+		else {
+			$this->returnError("Sorry, {$this->url} doesn't seem to be a Wordpress blog.", 404);
+		}
+	}
+
+	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);
+		$content = preg_replace('/<div class="wpa".*/', '', $content);
+		return $content;
+	}
+
+	private function getDate($uri) {
+		preg_match('/\d{4}\/\d{2}\/\d{2}/', $uri, $matches);
+		$date = new \DateTime($matches[0]);
+		return $date->format('U');
+	}
+
+	public function getName() {
+		return "{$this->name} - Wordpress Bridge";
+	}
+
+	public function getURI() {
+		return $this->url;
+	}
+
+	public function getCacheDuration() {
+		return 3600*3; // 3 hours
+	}
+
+	private function hasUrl() {
+		if (empty($this->url)) {
+			return false;
+		}
+		return true;
+	}
+
+	private function processParams($param) {
+		$this->url = $param['url'];
+		$this->name = $param['name'];
+	}
+
+}
+