db/message: make sure attachment data is always a byte sequence

Encode text strings to UTF-8 and set the charset parameter accordingly.

1 files changed, 10 insertions, 2 deletions

diff --git a/alot/db/message.py b/alot/db/message.py
index 62169356..a6d67dfd 100644
--- a/alot/db/message.py
+++ b/alot/db/message.py
@@ -140,9 +140,17 @@ class _MimeTree:
             cd = part.get_content_disposition()
             fn = part.get_filename()
             if cd == 'attachment' or fn is not None:
-                data = part.get_content()
+                data   = part.get_content()
+                params = part.get_params()
+
+                # make sure data is always a byte-sequence
+                if isinstance(data, str):
+                    data = data.encode('utf-8', errors = 'replace')
+                    params = [p for p in params if p[0] != 'charset'] + \
+                             [('charset', 'utf-8')]
+
                 self.attachment = Attachment(data, self.content_type,
-                                             fn, part.get_params())
+                                             fn, params)
 
     def __str__(self):
         return 'MimePart(%s)' % self.content_type