#!/usr/bin/ruby

#
# Train dspam by checking if users moved any messages
# between the inbox and the junkbox
#

JUNK_BOX = '.Junk/'
DSPAM = 'dspam'

require 'rubygems'
require_gem 'activerecord'

`logger -p mail.info Training dspam`

class User < ActiveRecord::Base
  establish_connection   :adapter => 'mysql',
  :database => 'cura_prod',
  :host => 'localhost',
  :username => 'cura',
  :password => 'yakubrub'
end

def log(msg)
  `logger -t train_dspam -p mail.info #{msg}`
end

User.find(:all).each do |user|
  # Look at the user's inbox for mail
  # incorrectly labeled as spam
  log "Processing #{user.email} - inbox"
  Dir.glob(user.maildir + '{cur,new}/*').each do |file|
    `grep 'X-DSPAM-Result: Spam' #{file} 1>/dev/null`
    if $?.to_i == 0
      # If such a message is found, tell dspam about it and
      # change the header so we won't process it again
      log "Correcting a false positive"
      `#{DSPAM} --user #{user.email} --class=innocent --source=error < #{file}`
      `sed 's/X-DSPAM-Result: Spam/X-DSPAM-Result: Innocent/g' #{file} > #{file}_dspam_tmp && mv #{file}_dspam_tmp #{file}`
    end
  end

  # Now look at the junk folder
  # for mail which failed to be identified as spam
  log "Processing #{user.email} - junkbox"
  Dir.glob(user.maildir + "#{JUNK_BOX}/{cur,new}/*").each do |file|
    `grep "X-DSPAM-Result: Innocent" #{file} 1>/dev/null`
    if $?.to_i == 0
      # If such a message is found, tell dspam about it and
      # since it is spam we can just delete it
      log "Correcting a false negative"
      `#{DSPAM} --user #{user.email} --class=spam --source=error < #{file}`
      `rm -f #{file}`
    end
  end
end