Encode message body to bytes prior to reconstituting e-mail message object. Added additional test to ensure that messages can make it the full cycle.

django_mailbox/models.py

@@ -394,7 +394,21 @@ class Message(models.Model):
         ).replace('=\n', '').rstrip('\n')
 
     def get_email_object(self):
-        return email.message_from_string(self.body)
+        """ Returns an `email.message.Message` instance for this message.
+
+        Note: Python 2.6's version of email.message_from_string requires
+        bytes and will incorrectly create an e-mail object if a unicode
+        object is passed-in.
+
+        Note that in reality 7-bit ascii *should* be an acceptable encoding
+        here per RFC-2822 (2.1), but given that `message_from_string` really
+        only cares that bytes are incoming rather than a unicode object, and
+        django will dutifully, if you were to edit the message in code or
+        via the admin, allow you to store unicode characters, returning UTF-8
+        encoded bytes is probably the safest answer.
+
+        """
+        return email.message_from_string(self.body.encode('utf-8'))
 
     def delete(self, *args, **kwargs):
         for attachment in self.attachments.all():

django_mailbox/tests/__init__.py

@@ -155,3 +155,25 @@ class TestGetMessage(EmailMessageTestCase):
             actual_text,
             expected_text
         )
+
+
+class TestMessageGetEmailObject(TestCase):
+    def test_get_body_properly_handles_unicode_body(self):
+        with open(
+            os.path.join(
+                os.path.dirname(__file__),
+                'generic_message.eml'
+            )
+        ) as f:
+            unicode_body = unicode(f.read())
+
+        message = Message()
+        message.body = unicode_body
+
+        expected_body = unicode_body
+        actual_body = message.get_email_object().as_string()
+
+        self.assertEquals(
+            expected_body,
+            actual_body
+        )