Render parts of mails more safely

We can't safely assume that mails have correct charsets declared,
including for headers. If using the declared charset fails, just
use repr for now.

1 files changed, 8 insertions, 2 deletions

diff --git a/distixlib/message_renderer.py b/distixlib/message_renderer.py
index 340e40b..9c8d900 100644
--- a/distixlib/message_renderer.py
+++ b/distixlib/message_renderer.py
@@ -36,10 +36,16 @@ class MessageRenderer(object):
     def _get_header(self, msg, name):
         decoded = email.header.decode_header(msg[name])
         combined = u' '.join(
-            value.decode(encoding or 'us-ascii')
+            self._decode_safely(value, encoding)
             for value, encoding in decoded)
         return combined
 
+    def _decode_safely(self, s, encoding):
+        try:
+            return s.decode(encoding or 'us-ascii')
+        except UnicodeDecodeError:
+            return repr(s)
+
     def _plain_text_body(self, msg):
         body = self._find_first_plain_text_part(msg)
         if body is None:
@@ -56,7 +62,7 @@ class MessageRenderer(object):
     def _get_payload_as_unicode(self, body):
         charset = body.get_param('charset', 'us-ascii')
         text = body.get_payload(decode=True)
-        return text.decode(charset)
+        return self._decode_safely(text, charset)
 
     def _strip_signature(self, text):
         sigsep = u'\n-- \n'