From 2b5ab82d00fb7ebe746f0df6ef66b014fd4e50f6 Mon Sep 17 00:00:00 2001
From: Folker Bernitt <fbernitt@thoughtworks.com>
Date: Thu, 9 Apr 2015 11:38:07 +0200
Subject: Added fallback to us ascii for invalid Content-Type values.

- Issue #347
---
 service/pixelated/adapter/model/mail.py | 13 ++++++++++---
 service/test/unit/adapter/test_mail.py  |  9 +++++++++
 2 files changed, 19 insertions(+), 3 deletions(-)

(limited to 'service')

diff --git a/service/pixelated/adapter/model/mail.py b/service/pixelated/adapter/model/mail.py
index 618b980a..7e0d5ad1 100644
--- a/service/pixelated/adapter/model/mail.py
+++ b/service/pixelated/adapter/model/mail.py
@@ -240,21 +240,28 @@ class PixelatedMail(Mail):
 
         try:
             decoding_map = {
-                'quoted-printable': lambda content, content_type: unicode(content.decode('quopri'), content_type),
+                'quoted-printable': lambda content, content_type: content.decode('quopri').decode(content_type),
                 'base64': lambda content, content_type: content.decode('base64').decode('utf-8'),
                 '7bit': lambda content, content_type: content.encode(content_type),
                 '8bit': lambda content, content_type: content.encode(content_type)
             }
             if encoding:
-                return decoding_map[encoding](part['content'], content_type)
+                return self._decode_content_with_fallback(part['content'], decoding_map[encoding], content_type)
             else:
-                return part['content']
+                return self._decode_content_with_fallback(part['content'], lambda content, content_type: content.encode(content_type), 'ascii')
         except Exception:
             logger.error('Failed to decode mail part with:')
             logger.error('Content-Transfer-Encoding: %s' % encoding)
             logger.error('Content-Type: %s' % part['headers'].get('Content-Type'))
             raise
 
+    def _decode_content_with_fallback(self, content, decode_func, content_type):
+        try:
+            return decode_func(content, content_type)
+            # return content.encode(content_type)
+        except ValueError:
+            return content.encode('ascii', 'ignore')
+
     @property
     def alternatives(self):
         return self.parts.get('alternatives')
diff --git a/service/test/unit/adapter/test_mail.py b/service/test/unit/adapter/test_mail.py
index d77816cd..dbad72b3 100644
--- a/service/test/unit/adapter/test_mail.py
+++ b/service/test/unit/adapter/test_mail.py
@@ -232,6 +232,15 @@ class TestPixelatedMail(unittest.TestCase):
         self.assertEquals(u'H\xe4llo', mail.text_plain_body)
         self.assertEquals(u'<p>H\xe4llo</p>', mail.html_body)
 
+    def test_broken_content_type_defaults_to_usascii(self):
+        plain_headers = {'Content-Type': 'I lie to you', 'Content-Transfer-Encoding': 'quoted-printable'}
+        html_headers = {'Content-Type': 'text/html;\ncharset=utf-8', 'Content-Transfer-Encoding': 'quoted-printable'}
+        parts = {'alternatives': [{'content': 'H=E4llo', 'headers': plain_headers}, {'content': '<p>H=C3=A4llo</p>', 'headers': html_headers}]}
+
+        mail = PixelatedMail.from_soledad(None, None, self._create_bdoc(raw='some raw body'), parts=parts, soledad_querier=None)
+
+        self.assertEquals(u'H=E4llo', mail.text_plain_body)
+
     def test_clean_line_breaks_on_address_headers(self):
         many_recipients = 'One <one@mail.com>,\nTwo <two@mail.com>, Normal <normal@mail.com>,\nalone@mail.com'
         headers = {'Cc': many_recipients,
-- 
cgit v1.2.3