[FIX] mail: more robust parsing of In-Reply-To/References (OPW 608919)

When parsing incoming messages, ignore white-space around
In-Reply-To headers, and extract message-id items inside
the References header using a regex.
This actually serves as a workaround for broken MTAs
mangling References (such as outlook.com nesting past ones
with commas, violating RFC2822).

Closes #516 as a manual rebase.

addons/mail/mail_thread.py

@@ -29,6 +29,7 @@ import re
 import socket
 import time
 import xmlrpclib
+import re
 from email.message import Message
 
 from openerp import tools
@@ -42,6 +43,8 @@ from openerp.tools.translate import _
 _logger = logging.getLogger(__name__)
 
 
+mail_header_msgid_re = re.compile('<[^<>]+>')
+
 def decode_header(message, header, separator=' '):
     return separator.join(map(decode, filter(None, message.get_all(header, []))))
 
@@ -916,13 +919,13 @@ class mail_thread(osv.AbstractModel):
             msg_dict['date'] = stored_date.strftime(tools.DEFAULT_SERVER_DATETIME_FORMAT)
 
         if message.get('In-Reply-To'):
-            parent_ids = self.pool.get('mail.message').search(cr, uid, [('message_id', '=', decode(message['In-Reply-To']))])
+            parent_ids = self.pool.get('mail.message').search(cr, uid, [('message_id', '=', decode(message['In-Reply-To'].strip()))])
             if parent_ids:
                 msg_dict['parent_id'] = parent_ids[0]
 
         if message.get('References') and 'parent_id' not in msg_dict:
-            parent_ids = self.pool.get('mail.message').search(cr, uid, [('message_id', 'in',
-                                                                         [x.strip() for x in decode(message['References']).split()])])
+            msg_list =  mail_header_msgid_re.findall(decode(message['References']))
+            parent_ids = self.pool.get('mail.message').search(cr, uid, [('message_id', 'in', [x.strip() for x in msg_list])])
             if parent_ids:
                 msg_dict['parent_id'] = parent_ids[0]