Merge pull request #136 from Riduidel/TheOatmeal

[new] The oatmeal

bridges/RssExpander.php

@@ -16,7 +16,7 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
         if (empty($param['url'])) {
             $this->returnError('There is no $param[\'url\'] for this RSS expander', 404);
         }
- //       $this->message("Loading from ".$param['url']);
+//       $this->message("Loading from ".$param['url']);
         // Notice WE DO NOT use cache here on purpose : we want a fresh view of the RSS stream each time
         $rssContent = simplexml_load_file($param['url']) or $this->returnError('Could not request '.$param['url'], 404);
 //        $this->message("loaded RSS from ".$param['url']);
@@ -25,7 +25,7 @@ abstract class RssExpander extends HttpCachingBridgeAbstract{
         $this->collect_RSS_2_0_data($rssContent);
     }
     
-    private function collect_RSS_2_0_data($rssContent) {
+    protected function collect_RSS_2_0_data($rssContent) {
         $rssContent = $rssContent->channel[0];
 //        $this->message("RSS content is ===========\n".var_export($rssContent, true)."===========");
         $this->load_RSS_2_0_feed_data($rssContent);

bridges/TheOatMealBridge.php

@@ -0,0 +1,62 @@
+<?php
+/**
+*
+* @name The Oatmeal
+* @description Un petit site de dessins assez rigolos
+* @update 20/02/201403/07/2015
+*/
+require_once 'bridges/RssExpander.php';
+define("THE_OATMEAL", "http://theoatmeal.com/");
+define("RSS", "http://feeds.feedburner.com/oatmealfeed");
+class TheOatmealBridge extends RssExpander{
+
+    public function collectData(array $param){
+        $param['url'] = RSS;
+        parent::collectData($param);
+    }
+
+
+    /**
+     * Since the oatmeal produces a weird RSS feed, I have to fix it by loading the items separatly from the feed infos
+     */
+    protected function collect_RSS_2_0_data($rssContent) {
+        $rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/");
+        $rssHeaderContent = $rssContent->channel[0];
+//        $this->message("RSS content is ===========\n".var_export($rssHeaderContent, true)."===========");
+        $this->load_RSS_2_0_feed_data($rssHeaderContent);
+        foreach($rssContent->item as $item) {
+            $this->message("parsing item ".var_export($item, true));
+            $this->items[] = $this->parseRSSItem($item);
+        }
+    }
+
+
+    protected function parseRSSItem($newsItem) {
+        $namespaces = $newsItem->getNameSpaces(true);
+        $dc = $newsItem->children($namespaces['dc']);
+        $rdf = $newsItem->children($namespaces['rdf']);
+        $item = new Item();
+        $item->title = trim($newsItem->title);
+        $this->message("browsing Oatmeal item ".var_export($newsItem, true));
+        $item->uri=$newsItem->attributes($namespaces['rdf'])->about;
+        // now load that uri from cache
+        $this->message("now loading page ".$item->uri);
+        $articlePage = str_get_html($this->get_cached($item->uri));
+
+        $content = $articlePage->find('#comic', 0);
+		if($content==null) {
+			$content = $articlePage->find('#blog');
+		}
+        $item->content = $content->innertext;
+        
+        $this->message("dc content is ".var_export($dc, true));
+        $item->name = $dc->creator;
+        $item->timestamp = DateTime::createFromFormat(DateTime::ISO8601, $dc->date)->getTimestamp();
+        $this->message("writtem by ".$item->name." on ".$item->timestamp);
+        return $item;
+    }
+    
+    public function getCacheDuration(){
+        return 7200; // 2h hours
+    }
+}

index.php

@@ -13,7 +13,8 @@ TODO :
 
 date_default_timezone_set('UTC');
 error_reporting(0);
-//ini_set('display_errors','1'); error_reporting(E_ALL);  // For debugging only.
+//ini_set('display_errors','1');
+//error_reporting(E_ALL);  // For debugging only.
 
 // extensions check
 if (!extension_loaded('openssl'))
@@ -220,7 +221,7 @@ $formats = Format::searchInformation();
 
     <header>
         <h1>RSS-Bridge</h1>
-        <h2>·Reconnecting the Web·</h2>
+        <h2>·Reconnecting the Web·</h2>
     </header>
 	<?php
 	    $activeFoundBridgeCount = 0;