Don't read the saved mail in binary mode.

The first reason for doing this is that atomic_write doesn't write in binary mode, and I think we should be symmetrical in reading and writing.

The second reason is that I don't think a received email is really binary data in its raw form. I think it will be ascii with non ascii parts converted to ascii using content-transfer-encodings (http://en.wikipedia.org/wiki/MIME).

This matters in Ruby 1.9 because we have a default internal and external encoding (http://blog.grayproductions.net/articles/ruby_19s_three_default_encodings).

UTF-8 seems like a good initial default to set. If we retrieve this data from a file as binary data, we have a problem treating it as text, as we can't convert from binary to text in Ruby 1.9. Given that this is ascii data, I think we can safely treat it as UTF-8 (which should be a superset of ascii), until we convert it into a mail object, which should interpret the mime encodings correctly.

1 files changed, 2 insertions, 2 deletions

diff --git a/app/models/raw_email.rb b/app/models/raw_email.rb
index bae144931..de7978b82 100644
--- a/app/models/raw_email.rb
+++ b/app/models/raw_email.rb
@@ -22,7 +22,7 @@ class RawEmail < ActiveRecord::Base
         if request_id.empty?
             raise "Failed to find the id number of the associated request: has it been saved?"
         end
-        
+
         if ENV["RAILS_ENV"] == "test"
             return File.join(Rails.root, 'files/raw_email_test')
         else
@@ -49,7 +49,7 @@ class RawEmail < ActiveRecord::Base
     end
 
     def data
-        File.open(self.filepath, "rb").read
+        File.open(self.filepath, "r").read
     end
 
     def destroy_file_representation!