Respond to review:

- Refactor "fully" decoding the payload of a text/* part.  In Python 3,
  decode=True only means to decode according to Content-Transfer-Encoding, not
  according to any charset in the Content-Type header.  So do that.

1 files changed, 1 insertions, 11 deletions

diff --git a/cloudinit/user_data.py b/cloudinit/user_data.py
index bf5642a5..5fdc46f2 100644
--- a/cloudinit/user_data.py
+++ b/cloudinit/user_data.py
@@ -108,17 +108,7 @@ class UserDataProcessor(object):
 
             ctype = None
             ctype_orig = part.get_content_type()
-            ctype_main = part.get_content_maintype()
-            payload = part.get_payload(decode=True)
-            # In Python 3, decoding the payload will ironically hand us a
-            # bytes object.  'decode' means to decode according to
-            # Content-Transfer-Encoding, not according to any charset in the
-            # Content-Type.  So, if we end up with bytes, first try to decode
-            # to str via CT charset, and failing that, try utf-8 using
-            # surrogate escapes.
-            if six.PY3 and ctype_main == 'text' and isinstance(payload, bytes):
-                charset = part.get_charset() or 'utf-8'
-                payload = payload.decode(charset, errors='surrogateescape')
+            payload = util.fully_decoded_payload(part)
             was_compressed = False
 
             # When the message states it is of a gzipped content type ensure