[search][utils] Strip database_name from split chars

If the database name ends with a dot, the split method will return
a list with one empty char entry which means that in the case of
the dataset name generation we might end up with an empty string
and trigger an invalid index error. This patch avoids this by
ignoring leading and trailing split chars.

beat/web/search/utils.py

@@ -198,7 +198,7 @@ class FilterGenerator(object):
     # beginning and/or end of the relevant dataset name parts.
     #---------------------------------------------------------------------------
     def parse_dataset_name(self, dataset_name, exact=False):
-        parts = re.split(r'[\.@]', dataset_name)
+        parts = re.split(r'[\.@]', dataset_name.strip(".@"))
 
         database = None
         version  = None