Minor PR cleanup and reorganization.

django_mailbox/admin.py

@@ -5,6 +5,7 @@ from django.contrib import admin
 
 from django_mailbox.models import MessageAttachment, Message, Mailbox
 from django_mailbox.signals import message_received
+from django_mailbox.utils import decode_header
 
 logger = logging.getLogger(__name__)
 
@@ -44,23 +45,19 @@ class MessageAttachmentInline(admin.TabularInline):
     model = MessageAttachment
     extra = 0
 
-default_charset = 'ASCII'
-import email.header
-def decode(m):
-    return ''.join([ unicode(t[0], t[1] or default_charset) for t in email.header.decode_header(m) ])
 
 class MessageAdmin(admin.ModelAdmin):
     def attachment_count(self, msg):
         return msg.attachments.count()
 
-    def Subject(self, msg):
+    def subject(self, msg):
-        return decode(msg.subject)
+        return decode_header(msg.subject)
 
     inlines = [
         MessageAttachmentInline,
     ]
     list_display = (
-        'Subject',
+        'subject',
         'processed',
         'read',
         'mailbox',

django_mailbox/models.py

@@ -6,7 +6,6 @@ import urllib
 import mimetypes
 import os.path
 from quopri import encode as encode_quopri
-from quopri import decodestring as decodestring_quopri
 import sys
 import uuid
 
@@ -258,7 +257,7 @@ class Mailbox(models.Model):
             msg.to_header = message['to']
         msg.save()
         message = self._get_dehydrated_message(message, msg)
-        msg.body = unicode(message.as_string(),errors='ignore')
+        msg.body = message.as_string()
         if message['in-reply-to']:
             try:
                 msg.in_reply_to = Message.objects.filter(
@@ -406,21 +405,16 @@ class Message(models.Model):
     def get_text_body(self):
         def get_body_from_message(message):
             body = ''
-            charset = 'ascii'
             for part in message.walk():
                 if (
                     part.get_content_maintype() == 'text'
                     and part.get_content_subtype() == 'plain'
                 ):
-                    # body = body + part.get_payload()
+                    charset = part.get_content_charset()
-                    if part.get_content_charset():
+                    this_part = part.get_payload(decode=True)
-                        charset = part.get_content_charset()
+                    if charset:
-                    encoding = part['Content-Transfer-Encoding']
+                        this_part = this_part.decode(charset)
-                    # print encoding
+                    body += this_part
-                    if encoding and encoding.lower() == 'base64':
-                        body = body + unicode(part.get_payload().decode('base64'), charset)
-                    else:
-                        body = body + unicode(decodestring_quopri(part.get_payload()), charset)
             return body
 
         return get_body_from_message(
@@ -552,23 +546,21 @@ class MessageAttachment(models.Model):
     def get_filename(self):
         file_name = self._get_rehydrated_headers().get_filename()
         if file_name:
-            try:
+            encoded_subject, encoding = email.Header.decode_header(file_name)[0]
-                d_subject = email.Header.decode_header(file_name)
+            if encoding:
-            except UnicodeEncodeError, e:
+                encoded_subject = encoded_subject.decode(encoding)
-                print e
+            return encoded_subject
-                return file_name
-            if d_subject[0][1]:
-                return unicode(d_subject[0][0], d_subject[0][1])
-            else:
-                return d_subject[0][0]
         else:
-            return ''
+            return None
 
     def items(self):
         return self._get_rehydrated_headers().items()
 
-    # def __getitem__(self, name):
+    def __getitem__(self, name):
-    #     return self._get_rehydrated_headers()[name]
+        value = self._get_rehydrated_headers()[name]
+        if value is None:
+            raise KeyError('Header %s does not exist' % name)
+        return value
 
     def __unicode__(self):
         return self.document.url

django_mailbox/utils.py

@@ -0,0 +1,31 @@
+import email.header
+import logging
+
+from django.conf import settings
+
+
+logger = logging.getLogger(__name__)
+
+
+DEFAULT_CHARSET = getattr(
+    settings,
+    'DJANGO_MAILBOX_DEFAULT_CHARSET',
+    'ascii',
+)
+
+
+def decode_header(header):
+    try:
+        return ''.join(
+            [
+                unicode(t[0], t[1] or DEFAULT_CHARSET)
+                for t in email.header.decode_header(header)
+            ]
+        )
+    except UnicodeDecodeError:
+        logger.exception(
+            'Errors encountered decoding header %s into encoding %s.',
+            header,
+            DEFAULT_CHARSET,
+        )
+        return unicode(header, DEFAULT_CHARSET, 'replace')