Exclude 'tumblr_' urls from generic image/video matching

- Excludes urls containing 'tumblr_' from the generic image/video pattern matching results.

src/TumblThree/TumblThree.Applications/Crawler/AbstractTumblrCrawler.cs

@@ -252,6 +252,9 @@ protected void AddGenericPhotoUrl(string post)
         {
             foreach (string imageUrl in tumblrParser.SearchForGenericPhotoUrl(post))
             {
+                if (tumblrParser.IsTumblrUrl(imageUrl))
+                    continue;
+
                 if (CheckIfSkipGif(imageUrl))
                     continue;
 
@@ -263,15 +266,10 @@ protected void AddGenericVideoUrl(string post)
         {
             foreach (string videoUrl in tumblrParser.SearchForGenericVideoUrl(post))
             {
-                string url = videoUrl;
-                if (url.Contains("tumblr") && shellService.Settings.VideoSize == 480)
-                {
-                    int indexOfSuffix = url.LastIndexOf('.');
-                    if (indexOfSuffix >= 0)
-                        url = url.Insert(indexOfSuffix, "_480");
-                }
-
-                AddToDownloadList(new VideoPost(url, Guid.NewGuid().ToString("N")));
+                if (tumblrParser.IsTumblrUrl(videoUrl))
+                    continue;
+
+                AddToDownloadList(new VideoPost(videoUrl, Guid.NewGuid().ToString("N")));
             }
         }
     }

src/TumblThree/TumblThree.Applications/Parser/ITumblrParser.cs

@@ -24,5 +24,7 @@ public interface ITumblrParser
         IEnumerable<string> SearchForGenericPhotoUrl(string searchableText);
 
         IEnumerable<string> SearchForGenericVideoUrl(string searchableText);
+
+        bool IsTumblrUrl(string url);
     }
 }

src/TumblThree/TumblThree.Applications/Parser/TumblrParser.cs

@@ -64,5 +64,11 @@ public IEnumerable<string> SearchForGenericVideoUrl(string searchableText)
                 yield return videoUrl;
             }
         }
+
+        public bool IsTumblrUrl(string url)
+        {
+            var regex = new Regex("tumblr_[\\w]*");
+            return regex.IsMatch(url);
+        }
     }
 }

src/TumblThree/TumblThree.Presentation/Properties/Resources.resx

@@ -476,7 +476,7 @@ since the throttling algorithm is too slow to adjust and might negatively impact
     <value>Number of connections used for each scan</value>
   </data>
   <data name="ToolTipScanDescription" xml:space="preserve">
-    <value>Sets the number connections used for scanning.
+    <value>Sets the number of connections used for scanning.
 Since the data is usually small, you should leave this high.
 Note: This setting has no impact if the Limit Tumblr Api Connections setting is turned on.</value>
   </data>
@@ -700,7 +700,7 @@ E.g.  great big car, bears searches for great big cars and bears.</value>
     <value>Number of concurrent connections to the tumblr video host</value>
   </data>
   <data name="ToolTipVideoConnectionsDescription" xml:space="preserve">
-    <value>The vt.tumblr.com host regularly closes connections if the number is too high.</value>
+    <value>The v*.tumblr.com hosts regularly close connections if the number is too high.</value>
   </data>
   <data name="ExternalSettings" xml:space="preserve">
     <value>External</value>
@@ -927,13 +927,15 @@ This value determines the information refresh rate for each individual queued bl
   </data>
   <data name="ToolTipRegExPhotosDescription" xml:space="preserve">
     <value>Uses regular expressions to search for images in everything TumblThree scans.
-This will add plenty of duplicate image urls to the queue, but might gather images from websites that are currently not supported by a specifically written parser.</value>
+This might add plenty of duplicate or random image urls to the queue, but could potentially
+gather images from websites that are currently not supported by a specifically written parser.</value>
   </data>
   <data name="ToolTipRegExVideos" xml:space="preserve">
     <value>Search for videos in the crawl data</value>
   </data>
   <data name="ToolTipRegExVideosDescription" xml:space="preserve">
     <value>Uses regular expressions to search for videos in everything TumblThree scans.
-This will add plenty of duplicate video urls to the queue, but might gather videos from websites that are currently not supported by a specifically written parser.</value>
+This might add plenty of duplicate or random video urls to the queue, but could potentially
+gather videos from websites that are currently not supported by a specifically written parser.</value>
   </data>
 </root>