try..except on UTF8Type serialize for pre-encoded strings

nisanharamati · nisanharamati · commit 61a1c71aa9c1 · 2014-02-07T16:30:21.000-08:00
Wrap ustr.decode() with a try..except block to handle failure to encode strings of already-encoded unicode.
diff --git a/cassandra/cqltypes.py b/cassandra/cqltypes.py
@@ -528,15 +528,11 @@ def deserialize(byts):
 
     @staticmethod
     def serialize(ustr):
-        # ustr.encode('utf8') fails when the string is already encoded
-        # this is common if your data comes through other database drivers (e.g. odbc, psycopg2, etc.)
-        if isinstance(ustr, unicode): # check type explicitly. Unicode will encode successfuly.
-            return ustr.encode('utf8')
-        # otherwise, our input string is either already encoded or not unicode to begin with.
-        # since all cassandra strings are utf-8, we can validate that the ustr is already encoded utf-8 by decoding it
-        else:
-            ustr.decode('utf-8') # will raise UnicodeDecodeError if not utf8 encoded byte string.
-            return ustr # definitely valid :)
+        try:
+            return ustr.encode('utf-8')
+        except UnicodeDecodeError:
+            # already utf-8
+            return ustr
 
 
 class VarcharType(UTF8Type):
