Better detection for wrong file extensions

This fixes issues where previously downloaded
episodes would not get their extension fixed,
because the mime type does not change, although
the filename is still wrong.

src/gpodder/download.py

@@ -687,15 +687,17 @@ class DownloadTask(object):
 
             new_mimetype = headers.get('content-type', self.__episode.mimetype)
             old_mimetype = self.__episode.mimetype
-            if new_mimetype != old_mimetype:
+            _basename, ext = os.path.splitext(self.filename)
+            if new_mimetype != old_mimetype or util.wrong_extension(ext):
                 log('Correcting mime type: %s => %s', old_mimetype, new_mimetype, sender=self)
                 old_extension = self.__episode.extension()
                 self.__episode.mimetype = new_mimetype
                 new_extension = self.__episode.extension()
 
-                # If the desired filename extension changed due to the new mimetype,
-                # we force an update of the local filename to fix the extension
-                if old_extension != new_extension:
+                # If the desired filename extension changed due to the new
+                # mimetype, we force an update of the local filename to fix the
+                # extension.
+                if old_extension != new_extension or util.wrong_extension(ext):
                     self.filename = self.__episode.local_filename(create=True, force_update=True)
 
             # TODO: Check if "real_url" is different from "url" and if it is,

src/gpodder/util.py

@@ -529,6 +529,52 @@ def remove_html_tags(html):
     return result.strip()
 
 
+def wrong_extension(extension):
+    """
+    Determine if a given extension looks like it's
+    wrong (e.g. empty, extremely long or spaces)
+
+    Returns True if the extension most likely is a
+    wrong one and should be replaced.
+
+    >>> wrong_extension('.mp3')
+    False
+    >>> wrong_extension('.divx')
+    False
+    >>> wrong_extension('mp3')
+    True
+    >>> wrong_extension('')
+    True
+    >>> wrong_extension('.12 - Everybody')
+    True
+    >>> wrong_extension('.mp3 ')
+    True
+    >>> wrong_extension('.')
+    True
+    >>> wrong_extension('.42')
+    True
+    """
+    if not extension:
+        return True
+    elif len(extension) > 5:
+        return True
+    elif ' ' in extension:
+        return True
+    elif extension == '.':
+        return True
+    elif not extension.startswith('.'):
+        return True
+    else:
+        try:
+            # ".<number>" is an invalid extension
+            float(extension)
+            return True
+        except:
+            pass
+
+    return False
+
+
 def extension_from_mimetype(mimetype):
     """
     Simply guesses what the file extension should be from the mimetype