The Header.decode_header function eats up spaces in
non-encoded part of a header.

See the following source:
# -*- coding: iso-8859-1 -*-
from email.Header import Header, decode_header
h = Header('Essai ', None)
h.append('éè', 'iso-8859-1')
print h
print decode_header(h)

This prints:
Essai =?iso-8859-1?q?=E9=E8?=
[('Test', None), ('\xe9\xe8', 'iso-8859-1')]

This should print:
Essai =?iso-8859-1?q?=E9=E8?=
[('Test ', None), ('\xe9\xe8', 'iso-8859-1')]
       ^ This space disappears

This appears in Python 2.3 but the source code of the
function didn't change in 2.4 so the same problem
should still exist. Bug "[ 1372770 ] email.Header
should preserve original FWS" may be linked to that one
although I'm not sure this is exactly the same.

This patch (not extensively tested though) seems to
solve this problem:

--- /usr/lib/python2.3/email/Header.py  2005-09-05
00:20:03.000000000 +0200
+++ Header.py   2006-04-10 12:27:27.000000000 +0200
@@ -90,7 +90,7 @@
             continue
         parts = ecre.split(line)
         while parts:
-            unenc = parts.pop(0).strip()
+            unenc = parts.pop(0).rstrip()
             if unenc:
                 # Should we continue a long line?
                 if decoded and decoded[-1][1] is None: