Decode strings to unicode before slicing (bug 1277)

Simplified fix; based on a patch by Justin Forest.

(merged from the gPodder 2.x branch)

src/gpodder/model.py

@@ -340,6 +340,9 @@ class PodcastEpisode(PodcastModelObject):
         if not desc:
             return _('No description available')
         else:
+            # Decode the description to avoid gPodder bug 1277
+            if isinstance(desc, str):
+                desc = desc.decode('utf-8', 'ignore')
             if len(desc) > MAX_LINE_LENGTH:
                 return desc[:MAX_LINE_LENGTH] + '...'
             else: