#248 Fixed the UTF-8 decode for special characteres

2 files changed, 10 insertions, 4 deletions

diff --git a/service/pixelated/adapter/model/mail.py b/service/pixelated/adapter/model/mail.py
index 65e87f54..85f2bed4 100644
--- a/service/pixelated/adapter/model/mail.py
+++ b/service/pixelated/adapter/model/mail.py
@@ -16,6 +16,7 @@
 import json
 from uuid import uuid4
 from email.mime.text import MIMEText
+from email.header import decode_header
 
 from leap.mail.imap.fields import fields
 import leap.mail.walk as walk
@@ -288,7 +289,8 @@ class PixelatedMail(Mail):
             _headers[header] = [head.strip() for head in compact(_headers[header])]
 
         for header in ['From', 'Subject']:
-            _headers[header] = hdoc_headers.get(header)
+            # _headers[header] = hdoc_headers.get(header)
+            _headers[header] = self._decode_header(hdoc_headers.get(header))
 
         _headers['Date'] = self._get_date()
 
@@ -302,6 +304,10 @@ class PixelatedMail(Mail):
 
         return _headers
 
+    def _decode_header(self, header):
+        arr_header = decode_header(header)
+        return arr_header[0][0]
+
     def _get_date(self):
         date = self.hdoc.content.get('date', None)
         if not date:
diff --git a/service/pixelated/adapter/search/__init__.py b/service/pixelated/adapter/search/__init__.py
index 33259e63..331ea8fe 100644
--- a/service/pixelated/adapter/search/__init__.py
+++ b/service/pixelated/adapter/search/__init__.py
@@ -128,8 +128,8 @@ class SearchEngine(object):
         tags.append(mail.mailbox_name.lower())
         bounced = mail.bounced if mail.bounced else ['']
         index_data = {
-            'sender': unicode(header.get('from', '')),
-            'subject': unicode(header.get('subject', '')),
+            'sender': unicode(header.get('from', '').decode('utf-8')),
+            'subject': unicode(header.get('subject', '').decode('utf-8')),
             'date': milliseconds(header.get('date', '')),
             'to': u','.join(header.get('to', [''])),
             'cc': u','.join(header.get('cc', [''])),
@@ -139,7 +139,7 @@ class SearchEngine(object):
             'body': unicode(mdict['textPlainBody']),
             'ident': unicode(mdict['ident']),
             'flags': unicode(','.join(unique(mail.flags))),
-            'raw': unicode(mail.raw)
+            'raw': unicode(mail.raw.decode('utf-8'))
         }
 
         writer.update_document(**index_data)