Update (rss): add domain blacklist

app/Jobs/Tasks/BrowseRSSLatestNewsTask.php

@@ -11,6 +11,7 @@ public static function handleMulti($hours = 3)
     {
         $rss_urls = config('platform.global.rss');
 
+
         $raw_posts = [];
 
         foreach ($rss_urls as $rss_url) {
@@ -26,6 +27,8 @@ public static function handleMulti($hours = 3)
 
     public static function handleSingle($rss_url, $hours = 3)
     {
+        $blacklist_rss_post_domain = config('platform.global.blacklist_rss_post_domain');
+
 
         $f = FeedReader::read($rss_url);
 
@@ -41,6 +44,17 @@ public static function handleSingle($rss_url, $hours = 3)
             $title = trim($item->get_title());
             $description = trim($item->get_content());
 
+
+            $domain = get_domain_from_url($item->get_link());
+
+            if (in_array($domain, $blacklist_rss_post_domain))
+            {
+              continue ;
+            }
+
+
+
+
             $raw_posts[] = (object) [
                 'source' => $f->get_title(),
                 'source_url' => $rss_url,

app/Jobs/Tasks/ParseRssPostMetadataTask.php

@@ -141,9 +141,23 @@ public static function handle(int $rss_post_id)
 
         $rss_post->keyword_list = implode(',', $words_to_add_in_keyword_list);
 
+
+        if (is_empty($rss_post->bites))
+        {
+          $rss_post->status = 'blocked';
+        }
+        else
+        {
           $rss_post->status = 'published';
+        }
+
 
         if ($rss_post->save()) {
+
+
+            if ($rss_post->status == 'published')
+            {
+
                 $has_saved_keyword = false;
 
                 $deleted_rpk = RssPostKeyword::where('rss_post_id', $rss_post->id)->delete();
@@ -169,6 +183,9 @@ public static function handle(int $rss_post_id)
                     $rss_post->keyword_saved = true;
                     $rss_post->save();
                 }
+
+            }
+
         }
 
     }

config/platform/global.php

@@ -10,6 +10,13 @@
         'https://u.today',
     ],
 
+    'blacklist_rss_post_domain' => [
+      'www.techinasia.com',
+      'www.marktechpost.com',
+      'twitter.com',
+      'www.youtube.com',
+    ],
+
     'rss' => [
         'http://news.ycombinator.com/rss',
         // 'http://blog.samaltman.com/posts.atom',

routes/tests.php

@@ -57,6 +57,8 @@
 
     $last_record = RssPost::whereNull('post_domain')->orderBy('id', 'DESC')->first();
 
+    dd($last_record);
+
     for ($i = 1; $i <= $last_record->id; $i++) {
         $rss_post = RssPost::find($i);