Goal

Filter spam and virus from email messages

Requirements

1. Get mail from eBox mail module, filter and reinject it
2. Get mail from external server (if working as a mail gateway), filter and reinject it
3. Filtering parameters may be set up by domain
   • Parameters may be set up on installation
   • Parameters may be modified by sysadmin, and changes will take effect on new and existing accounts
4. Set up different policies on "spam", "virus", "malformed header" or "banned file"
   • PASS: message will be delivered
   • DISCARD: message will not be delivered. User will not be notified.
   • BOUNCE: message will not be delivered. User will be notified.
5. Stores information about mail accounts and virtual domains in eBox userandgroups service.
6. Quarantines are not supported
7. Module parameters will affect single or global domains
   • Global parameters
     • Switch on mail filtering
     • Switch on bayesian filtering
       • Upload "spam" messages for bayesian learning
       • Upload "ham" messages for bayesian learning
       • Clean up bayesian DB (remove all tokens and train it again)
       • Specific accounts to REDIRECT bogus positive and bogus negative and reinject them in autolearning system
       • Switch on autolearning
         • Set up "spam" threshold level (when a message detected as spam enters autolearning system)
         • Set up "ham" threshold level (when a message detected as ham enters autolearning system)
     • Set up string to add to subject when spam detected
       • If it has been set up by domain
     • Hit level required to:
       • Add informative headers to processed message
       • Add informative "SPAM DETECTED" header to message
       • Activate positive policy
     • Policy in:
       • Virus filtering positive
       • Spam filtering positive
       • Header check positive
       • Banned extension positive
       • Virus list autoupdate
   • By domain parameters
     • White list
       • Virus filtering
       • Spam filtering
       • Malformed header check
       • Banned extension check
     • Hit level required to:
       • Add informative headers to processed message
       • Add informative "SPAM DETECTED" header to message
       • Activate positive policy
     • Modify message subject (boolean)
     • Soft whitelist (with hit rate which modifies global hit account)
     • Soft blacklist (with hit rate which modifies global hit account)
8. Module parameters will affect single or global domains
9. By domain parameters may take its value from global ones