Ensure UTF-8 surogates escaped on save - fix #159

.gitignore

@@ -11,4 +11,5 @@ include/*
 dummy_project/*
 .cache/
 .tox/
-messages
+/messages
+

django_mailbox/models.py

@@ -346,7 +346,6 @@ def _get_dehydrated_message(self, msg, record):
     def _process_message(self, message):
         msg = Message()
         settings = utils.get_settings()
-
         if settings['store_original_message']:
             self._process_save_original_message(message, msg)
         msg.mailbox = self
@@ -386,10 +385,14 @@ def _process_message(self, message):
 
     def _process_save_original_message(self, message, msg):
         settings = utils.get_settings()
+        if six.PY3:
+            content = message.as_string().encode('ascii', 'surrogateescape')
+        else:
+            content = message.as_string()
         if settings['compress_original_message']:
             with NamedTemporaryFile(suffix=".eml.gz") as fp_tmp:
                 with gzip.GzipFile(fileobj=fp_tmp, mode="w") as fp:
-                    fp.write(message.as_string().encode('utf-8'))
+                    fp.write(content)
                 msg.eml.save(
                     "%s.eml.gz" % (uuid.uuid4(), ),
                     File(fp_tmp),
@@ -399,7 +402,7 @@ def _process_save_original_message(self, message, msg):
         else:
             msg.eml.save(
                 '%s.eml' % uuid.uuid4(),
-                ContentFile(message.as_string()),
+                ContentFile(content),
                 save=False
             )
 

django_mailbox/tests/messages/message_with_utf8_surrogates.eml

@@ -0,0 +1,46 @@
+Return-path: <SRS0=HZOvtU=DM=REDACTED@porady.REDACTED>
+Envelope-to: sprawa-1418@porady.REDACTED
+Delivery-date: Sat, 16 Dec 2017 16:22:42 +0100
+Received: from mx1.wp.pl ([212.77.101.6])
+ by s50.hekko.net.pl with esmtps (TLSv1.2:ECDHE-RSA-AES256-GCM-SHA384:256)
+ (Exim 4.89) (envelope-from <REDACTED@wp.pl>) id 1eQEII-0005Fu-Cs
+ for sprawa-1418@porady.REDACTED; Sat, 16 Dec 2017 16:22:42 +0100
+Received: (wp-smtpd smtp.wp.pl 33592 invoked from network);
+ 16 Dec 2017 16:22:11 +0100
+DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=wp.pl; s=1024a;
+ t=1513437731; bh=6Ox0FVna7vMxBu5CbF0s6HvpkNalENwSSgDxNJ/Rsdc=;
+ h=From:To:Subject;
+ b=QnrdNDBDNLuENN9QS0Pvn85/bDE1Fc6jJrvKUdnApFrykwzbHxXxCG4qX7g3sS3Qj
+ xxHGNf8UXZh3zyCln2EZpUD03LgkppMpTbv3tLKA4HSnaT7txr6AWHq2y8A/YQo7EY
+ 2806CYWtFCKYoVolzDN9lctM2nEoZpD5jOVZqYsM=
+Received: from public-gprs394416.centertel.pl (HELO REDACTED)
+ (REDACTED@wp.pl@[37.47.171.241]) (envelope-sender <REDACTED@wp.pl>)
+ by smtp.wp.pl (WP-SMTPD) with SMTP
+ for <REDACTED@sejm.pl>; 16 Dec 2017 16:22:11 +0100
+Message-ID: <BD6D8EECA3A74E9A9ABD7A1EEA593F76@REDACTED>
+From: <REDACTED@wp.pl>
+To: <sprawa-1418@porady.REDACTED>
+Subject: =?windows-1250?Q?Do_czego_te=BF_s=B9_zdolni_Polscy_s=EAdziowie_..._?=
+Date: Sat, 16 Dec 2017 16:21:04 +0100
+MIME-Version: 1.0
+Content-Type: multipart/mixed;
+ boundary="----=_NextPart_000_0018_01D37689.E058CEF0"
+X-Spam-Status: No, message_size=7331926 larger than 200K
+
+To jest wielocz�ciowa wiadomo�� w formacie MIME.
+
+------=_NextPart_000_0018_01D37689.E058CEF0
+Content-Type: multipart/alternative;
+ boundary="----=_NextPart_001_0019_01D37689.E058CEF0"
+
+
+------=_NextPart_001_0019_01D37689.E058CEF0
+Content-Type: text/plain;
+	charset="windows-1250"
+Content-Transfer-Encoding: quoted-printable
+
+REDACTED
+
+------=_NextPart_001_0019_01D37689.E058CEF0--
+
+------=_NextPart_000_0018_01D37689.E058CEF0--

django_mailbox/tests/test_process_email.py

@@ -147,6 +147,40 @@ def test_message_with_utf8_attachment_header(self):
             u'odpowied\u017a Burmistrza.jpg'
         )
 
+    def test_message_with_utf8_surrogates(self):
+        """Ensure that we properly handle UTF-8 surrogates
+
+        The problem observed in Python 3.5. It safes from regress of #159.
+        """
+
+        email_object = self._get_email_object(
+            'message_with_utf8_surrogates.eml',
+        )
+        mailbox = Mailbox.objects.create()
+        default_settings = utils.get_settings()
+        with mock.patch('django_mailbox.utils.get_settings') as get_settings:
+            altered = copy.deepcopy(default_settings)
+            altered['store_original_message'] = True
+
+            get_settings.return_value = altered
+
+            # This call throws the UnicodeEncodeError exception.
+            msg = mailbox.process_incoming_message(email_object)
+
+        self.assertEqual(
+            msg.subject,
+            u'Do czego te\u017c s\u0105 zdolni Polscy s\u0119dziowie ... '
+        )
+
+        self.assertEqual(
+            msg.attachments.count(),
+            0
+        )
+
+        with open(msg.eml.name, 'rb') as f:
+            self.assertEqual(f.read(),
+                             self._get_email_as_text('message_with_utf8_surrogates.eml'))
+
     def test_message_get_text_body(self):
         message = self._get_email_object('multipart_text.eml')
 
@@ -468,4 +502,4 @@ def test_message_compressed(self):
 
         with gzip.open(msg.eml.name, 'rb') as f:
             self.assertEqual(f.read(),
-                             self._get_email_as_text('generic_message.eml'))
+                             self._get_email_as_text('generic_message.eml'))