mail: handle unicode in subject for python3

diff --git a/pym/portage/mail.py b/pym/portage/mail.py
index f87efe2623b9f1e62b5ecd93364538fe88e95454..598c1f9b74d86ade9e7729b9d3bb2c394aafde8d 100644 (file)
--- a/pym/portage/mail.py
+++ b/pym/portage/mail.py
@@ -67,9 +67,19 @@ def create_message(sender, recipient, subject, body, attachments=None):
        mymessage.set_unixfrom(sender)
        mymessage["To"] = recipient
        mymessage["From"] = sender
-       # Use Header as a workaround so that long subject lines are wrapped
-       # correctly by <=python-2.6 (gentoo bug #263370, python issue #1974).
-       mymessage["Subject"] = Header(subject)
+
+       if sys.hexversion >= 0x3000000:
+               # Avoid UnicodeEncodeError in python3 with non-ascii characters.
+               #  File "/usr/lib/python3.1/email/header.py", line 189, in __init__
+               #    self.append(s, charset, errors)
+               #  File "/usr/lib/python3.1/email/header.py", line 262, in append
+               #    input_bytes = s.encode(input_charset, errors)
+               #UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-9: ordinal not in range(128)
+               mymessage["Subject"] = subject
+       else:
+               # Use Header as a workaround so that long subject lines are wrapped
+               # correctly by <=python-2.6 (gentoo bug #263370, python issue #1974).
+               mymessage["Subject"] = Header(subject)
        mymessage["Date"] = time.strftime("%a, %d %b %Y %H:%M:%S %z")
        
        return mymessage