Add option to ignore missing message ID

parse

@@ -16,6 +16,9 @@ parser.add_argument('-a', dest='attachment_dir', action='store', default='attach
 parser.add_argument('-f', '--forced', dest='forced', action='store_true',
                    help='force insertion into database, even if entries already exist')
                    
+parser.add_argument('-I', '--ignore-invalid', dest='ignore_invalid', action='store_true',
+                   help='process invalid e-mail files anyway, for example when missing message-id headers')
+
 args = parser.parse_args()
 options = vars(args)
 
@@ -100,8 +103,16 @@ for email_file in file_list:
 	message = email.message_from_file(open(email_file, 'r'))
 	
 	if message['message-id'] is None:
+		if options['ignore_invalid'] == True:
+			message_id = ""
+			print "WARNING: %s does not contain a valid message-id header. Empty message-id assumed." % email_file
+		else:
 			print "%s is not a valid e-mail file." % email_file
+			finished += 1
+			continue
 	else:
+		message_id = message['message-id']
+		
 	if 'subject' not in message or message['subject'] is None:
 		subject = ""
 	else:
@@ -111,7 +122,7 @@ for email_file in file_list:
 	htmlbody = ""
 	attachment_list = []
 	
-		sha1_hash = hashlib.sha1("%s/%s/%s/%s" % (message['from'], message['to'], message['message-id'], subject)).hexdigest()
+	sha1_hash = hashlib.sha1("%s/%s/%s/%s" % (message['from'], message['to'], message_id, subject)).hexdigest()
 	
 	if options['forced'] == False:
 		cursor.execute("SELECT * FROM emails WHERE `Hash` = ?", (sha1_hash,))
@@ -176,7 +187,7 @@ for email_file in file_list:
 		print "WARNING: Failed to determine unix timestamp for %s." % sha1_hash
 
 	try:
-			new_row = (getheader(message['message-id']), getheader(message['from']), getheader(message['to']), getheader(subject), timestamp, textbody, htmlbody, sha1_hash)
+		new_row = (getheader(message_id), getheader(message['from']), getheader(message['to']), getheader(subject), timestamp, textbody, htmlbody, sha1_hash)
 	except UnicodeDecodeError:
 		print "ERROR: Failed parsing %s, headers could not be decoded." % sha1_hash
 		continue

rename

@@ -7,6 +7,9 @@ parser = argparse.ArgumentParser(description='Parses emails into an SQLite datab
 parser.add_argument('-p', dest='pattern', action='store', default='*',
                    help='glob pattern (including path) that has to be matched for a file to be parsed')
 
+parser.add_argument('-I', '--ignore-invalid', dest='ignore_invalid', action='store_true',
+                   help='process invalid e-mail files anyway, for example when missing message-id headers')
+
 args = parser.parse_args()
 options = vars(args)
 
@@ -62,14 +65,22 @@ for email_file in file_list:
 	message = email.message_from_file(open(email_file, 'r'))
 
 	if message['message-id'] is None:
+		if options['ignore_invalid'] == True:
+			message_id = ""
+			print "WARNING: %s does not contain a valid message-id header. Empty message-id assumed." % email_file
+		else:
 			print "%s is not a valid e-mail file." % email_file
+			finished += 1
+			continue
 	else:
+		message_id = message['message-id']
+	
 	if 'subject' not in message or message['subject'] is None:
 		subject = ""
 	else:
 		subject = message['subject']
 	
-		sha1_hash = hashlib.sha1("%s/%s/%s/%s" % (message['from'], message['to'], message['message-id'], subject)).hexdigest()
+	sha1_hash = hashlib.sha1("%s/%s/%s/%s" % (message['from'], message['to'], message_id, subject)).hexdigest()
 	
 	new_path = "%s/%s.eml" % (os.path.dirname(email_file), sha1_hash)