Refactoring imports for Python3 support; refactoring unicode payload

handling.  Refs #13.

django_mailbox/models.py

@@ -4,18 +4,12 @@
 from email.message import Message as EmailMessage
 from email.utils import formatdate, parseaddr
 from email.encoders import encode_base64
-import urllib
 import mimetypes
 import os.path
 from quopri import encode as encode_quopri
 import sys
 import uuid
 
-try:
-    import urllib.parse as urlparse
-except ImportError:
-    import urlparse
-
 from django.conf import settings
 from django.core.mail.message import make_msgid
 from django.core.files.base import ContentFile
@@ -25,6 +19,7 @@
     MMDFTransport
 from django_mailbox.signals import message_received
 import six
+from six.moves.urllib.parse import unquote, urlparse
 
 
 STRIP_UNALLOWED_MIMETYPES = getattr(
@@ -117,7 +112,7 @@ class Mailbox(models.Model):
 
     @property
     def _protocol_info(self):
-        return urlparse.urlparse(self.uri)
+        return urlparse(self.uri)
 
     @property
     def _domain(self):
@@ -129,11 +124,11 @@ def port(self):
 
     @property
     def username(self):
-        return urllib.unquote(self._protocol_info.username)
+        return unquote(self._protocol_info.username)
 
     @property
     def password(self):
-        return urllib.unquote(self._protocol_info.password)
+        return unquote(self._protocol_info.password)
 
     @property
     def location(self):
@@ -249,6 +244,21 @@ def _get_dehydrated_message(self, msg, record):
             placeholder[ATTACHMENT_INTERPOLATION_HEADER] = str(attachment.pk)
             new = placeholder
         else:
+            content_charset = msg.get_content_charset()
+            if not content_charset:
+                content_charset = 'ascii'
+            try:
+                # Make sure that the payload can be properly decoded in the
+                # defined charset, if it can't, let's mash some things
+                # inside the payload :-\
+                msg.get_payload(decode=True).decode(content_charset)
+            except UnicodeDecodeError:
+                msg.set_payload(
+                    msg.get_payload(decode=True).decode(
+                        content_charset,
+                        'replace'
+                    ).encode(content_charset, 'replace')
+                )
             new = msg
         return new
 
@@ -495,18 +505,15 @@ def get_body(self):
         return self.body.encode('utf-8')
 
     def set_body(self, body):
-        if sys.version_info >= (3, 0):
+        if six.PY3:
             body = body.encode('utf-8')
         self.encoded = True
         self.body = base64.b64encode(body).decode('ascii')
 
     def get_email_object(self):
         """ Returns an `email.message.Message` instance for this message."""
         body = self.get_body()
-        if sys.version_info < (3, 0):
-            flat = email.message_from_string(body)
-        else:
-            flat = email.message_from_bytes(body)
+        flat = email.message_from_bytes(body)
         return self._rehydrate(flat)
 
     def delete(self, *args, **kwargs):

django_mailbox/tests/messages/message_with_invalid_content_for_declared_encoding.eml

@@ -1,5 +1,5 @@
 Date: Fri, 30 Jan 2012 11:30:01 PST
-Subject: Tjest
+Subject: Tjest Wrong
 From: "Somebody" <[email protected]>
 To: [email protected]
 MIME-Version: 1.0

setup.py

@@ -41,6 +41,6 @@
         'django_mailbox.tests',
     ],
     install_requires=[
-        'six'
+        'six>=1.6.1'
     ]
 )