updates

DMS/content_filter.py

@@ -11,6 +11,8 @@ import logging
 from email import message_from_binary_file
 from email.mime.text import MIMEText
 from email.utils import parseaddr, formatdate, make_msgid
+from datetime import datetime, timedelta
+from io import BytesIO
 
 # Setup logging
 logging.basicConfig(
@@ -30,6 +32,10 @@ DYNAMODB_RULES_TABLE = 'email-rules'
 REINJECT_HOST = 'localhost'
 REINJECT_PORT = 10026
 
+# Cache for DynamoDB rules (TTL 5 minutes)
+RULES_CACHE = {}
+CACHE_TTL = timedelta(minutes=5)
+
 # Initialize boto3 (lazy import to catch errors)
 try:
     import boto3
@@ -42,21 +48,28 @@ except Exception as e:
     logging.error(f"DynamoDB initialization failed: {e}")
 
 def get_email_rules(email_address):
-    """Fetch forwarding and auto-reply rules from DynamoDB"""
+    """Fetch forwarding and auto-reply rules from DynamoDB with caching"""
     if not DYNAMODB_AVAILABLE:
         return {}
     
+    now = datetime.now()
+    cache_key = email_address.lower()
+    if cache_key in RULES_CACHE and now - RULES_CACHE[cache_key]['time'] < CACHE_TTL:
+        logging.debug(f"Cache hit for {email_address}")
+        return RULES_CACHE[cache_key]['rules']
+    
     try:
         response = rules_table.get_item(Key={'email_address': email_address})
         item = response.get('Item', {})
         if item:
             logging.info(f"Rules found for {email_address}: forwards={len(item.get('forwards', []))}, ooo={item.get('ooo_active', False)}")
+        RULES_CACHE[cache_key] = {'rules': item, 'time': now}
         return item
     except Exception as e:
         logging.error(f"DynamoDB error for {email_address}: {e}")
         return {}
 
-def should_send_autoreply(sender_addr):
+def should_send_autoreply(original_msg, sender_addr):
     """Check if we should send auto-reply to this sender"""
     sender_lower = sender_addr.lower()
     
@@ -76,6 +89,11 @@ def should_send_autoreply(sender_addr):
             logging.info(f"Skipping auto-reply to automated sender: {sender_addr}")
             return False
     
+    # Check for auto-submitted header to prevent loops (RFC 3834)
+    if original_msg.get('Auto-Submitted', '').startswith('auto-'):
+        logging.info(f"Skipping auto-reply due to Auto-Submitted header: {sender_addr}")
+        return False
+    
     return True
 
 def send_autoreply(original_msg, recipient_rules, recipient_addr):
@@ -91,7 +109,7 @@ def send_autoreply(original_msg, recipient_rules, recipient_addr):
     # Extract email from "Name <email>" format
     sender_name, sender_addr = parseaddr(sender)
     
-    if not should_send_autoreply(sender_addr):
+    if not should_send_autoreply(original_msg, sender_addr):
         return
     
     subject = original_msg.get('Subject', 'No Subject')
@@ -140,7 +158,6 @@ def send_forwards(original_msg_bytes, recipient_rules, recipient_addr):
     for forward_addr in forwards:
         try:
             # Parse message again for clean forwarding
-            from io import BytesIO
             msg = message_from_binary_file(BytesIO(original_msg_bytes))
             
             # Add forwarding headers
@@ -167,17 +184,12 @@ def main():
     sender = sys.argv[1]
     recipients = sys.argv[2:]
     
-    logging.info(f"Processing email from {sender} to {', '.join(recipients)}")
+    logging.info(f"Processing email from {sender} to {', '.join(recipient)}")
     
     # Read email from stdin
     try:
         msg_bytes = sys.stdin.buffer.read()
-        msg = message_from_binary_file(sys.stdin.buffer)
-        
-        # Parse again from bytes for processing
-        from io import BytesIO
         msg = message_from_binary_file(BytesIO(msg_bytes))
-        
     except Exception as e:
         logging.error(f"Failed to read email: {e}")
         sys.exit(75)  # EX_TEMPFAIL
@@ -185,6 +197,13 @@ def main():
     # Process each recipient
     for recipient in recipients:
         try:
+            # Check if mail is internal (same domain) - skip if external
+            _, recipient_domain = parseaddr(recipient)[1].rsplit('@', 1) if '@' in recipient else ('', '')
+            _, sender_domain = parseaddr(sender)[1].rsplit('@', 1) if '@' in sender else ('', '')
+            if recipient_domain != sender_domain:
+                logging.info(f"Skipping external mail to {recipient} (sender domain: {sender_domain})")
+                continue
+            
             rules = get_email_rules(recipient)
             
             if rules:
@@ -203,7 +222,7 @@ def main():
     try:
         with smtplib.SMTP(REINJECT_HOST, REINJECT_PORT, timeout=30) as smtp:
             smtp.sendmail(sender, recipients, msg_bytes)
-        logging.info(f"✓ Delivered to {', '.join(recipients)}")
+        logging.info(f"✓ Delivered to {', '.join(recipient)}")
         sys.exit(0)
     except Exception as e:
         logging.error(f"✗ Delivery failed: {e}")