Email headers and non-ASCII characters

[Christoph Haas]

Hello, everyone...

I'm trying to send an email to people with non-ASCII characters in their 
names. A recpient's address may look like:

"Jörg Nørgens" <joerg at nowhere>

My example code:

=================================
def sendmail(sender, recipient, body, subject):
   message = MIMEText(body)
   message['Subject'] = Header(subject, 'iso-8859-1')
   message['From'] = Header(sender, 'iso-8859-1')
   message['To'] = Header(recipient, 'iso-8859-1')

   s = smtplib.SMTP()
   s.connect()
   s.sendmail(sender, recipient, message.as_string())
   s.close()
=================================

However the Header() method encodes the whole expression in ISO-8859-1:

=?iso-8859-1?q?=22J=C3=B6rg_N=C3=B8rgens=22_=3Cjoerg=40nowhere=3E?=

However I had expected something like:

"=?utf-8?q?J=C3=B6rg?= =?utf-8?q?_N=C3=B8rgens?=" <joerg at nowhere>

Of course my mail transfer agent is not happy with the first string 
although I see that Header() is just doing its job. I'm looking for a way 
though to encode just the non-ASCII parts like any mail client does. Does 
anyone have a recipe on how to do that? Or is there a method in 
the "email" module of the standard library that does what I need? Or 
should I split by regular expression to extract the email address 
beforehand? Or a list comprehension to just look for non-ASCII character 
and Header() them? Sounds dirty.

Hints welcome.

Regards
 Christoph