Fix Python 2 to 3 Unicode issues for string join

Revisited earlier commit where email subject lines with potentially
mixed encoding are joined into a single string. This fix brings in
the 'codecs' import to decode bytes to string using either the passed
encoding (if provided) or otherwise utf-8.

Changed the test_donor_import to no longer convert the Header to a
string and instead leave it as a byte array.

2 files changed, 3 insertions, 2 deletions

diff --git a/main/management/commands/donor_import.py b/main/management/commands/donor_import.py
index b8bdbf54..cc0e59f2 100644
--- a/main/management/commands/donor_import.py
+++ b/main/management/commands/donor_import.py
@@ -14,6 +14,7 @@ Subject: Receipt [$25.00] By: John Doe [john.doe@archlinux.org]
 Usage: ./manage.py donor_import path/to/maildir/
 """
 
+import codecs
 import logging
 import mailbox
 import sys
@@ -46,7 +47,7 @@ class Command(BaseCommand):
         default_charset = 'utf-8'
         # Convert the list of tuples containing the decoded string and encoding to
         # UTF-8
-        return u''.join([s[0].encode(default_charset, 'replace').decode(default_charset, 'replace') for s in subject])
+        return ''.join([codecs.decode(s[0], s[1] or default_charset) for s in subject])
 
 
     def parse_subject(self, subject):
diff --git a/main/tests/test_donor_import.py b/main/tests/test_donor_import.py
index f31ae099..0f27eee7 100644
--- a/main/tests/test_donor_import.py
+++ b/main/tests/test_donor_import.py
@@ -38,7 +38,7 @@ class DonorImportTest(TransactionTestCase):
 
     def test_decode_subject(self):
         text = u'メイル'
-        subject = str(Header(text, 'utf-8'))
+        subject = Header(text, 'utf-8')
         self.assertEqual(self.command.decode_subject(subject), text)
 
     def test_invalid_args(self):