Hopefully working version of train_bogofilter.

1 files changed, 32 insertions, 36 deletions

diff --git a/train_bogofilter b/train_bogofilter
index 56d1086..00f444d 100755
--- a/train_bogofilter
+++ b/train_bogofilter
@@ -1,55 +1,49 @@
 #!/usr/bin/python
-import imaplib
-import subprocess
 import email
+import imaplib
 import logging
-import email.Parser
+import os.path
+import subprocess
 from ConfigParser import ConfigParser
 logging.basicConfig(format='%(levelname)s:%(funcName)s:%(message)s',
                     level=logging.DEBUG)
 
 CMD_STR = "/usr/bin/bogofilter -%s"
+imaplib.Debug = 4
 
 def process_folder(proc_fld_name, bogofilter_param, end_fld_name):
     client.select(proc_fld_name)
     _, resp = client.search(None, "ALL")
     messages = resp[0].split()
+    logging.debug('messages = {0}'.format(messages))
     proc_msg_count = 0
-    logging.debug(messages)
 
     for msgId in messages:
-        # FIXME or no .PEEK ... do I want to mark a message as Seen?
-        # If not, I have to unset \Seen flag
-        # fetch(msgId, '(RFC822)') gets whole message as it is
-        _, msg_data = client.fetch(msgId,
-                                   '(BODY.PEEK[HEADER.FIELDS ' +
-                                   '(SUBJECT FROM X-DSPAM-SIGNATURE)])')
-        logging.debug("msgId = %s", msgId)
-        headers = hparser.parsestr(msg_data[0][1], headersonly=True)
-        logging.debug("headers:\n%s", headers)
-        if 'X-Dspam-Signature' in headers.keys():
-            ret = subprocess.Popen(CMD_STR % bogofilter_param,
-                                   stdin=subprocess.Popen,
-                                   shell=True)
-            ret.communicate(input=msg_data)  # FIXME Whole message
+        logging.debug('msgId = {0}'.format(msgId))
+        typ, msg_data = client.fetch(msgId, '(RFC822)')
+
+        msg = hparser.parsestr(msg_data[0][1])
+
+        ret = subprocess.Popen(CMD_STR % bogofilter_param,
+                               stdin=subprocess.PIPE,
+                               shell=True)
+        ret.communicate(input=msg_data[0][1])
 
-            if ret.returncode == 0:
-                typ, _ = client.copy(msgId, end_fld_name)
-                # Also
-                # http://bytes.com/topic/python/answers\
-                #        /41900-copying-moving-mail-message-using-imaplib
-                # Also RFC 2060, section 6.4.8, UID command
-                # Also http://pymotw.com/2/imaplib/
-                # folder.copy(msg_ids, where_folder)
-                if typ != 'OK':
-                    raise IOError("Cannot store a message to the folder %s"
-                                  % end_fld_name)
-                else:
-                    client.store(msgId, '+FLAGS', r'(\Deleted)')
-                    proc_msg_count += 1
+        logging.debug("ret.returncode = %s", ret.returncode)
+        if ret.returncode == 0:
+            del msg['X-Bogosity']
+            typ, _ = client.append(end_fld_name, '', '',
+                                   msg.as_string(True))
+            logging.debug("typ = %s", typ)
+            if typ != 'OK':
+                raise IOError("Cannot store a message to the folder %s"
+                              % end_fld_name)
             else:
-                raise OSError("bogofilter finished with the returncode: %d"
-                              % ret)
+                client.store(msgId, '+FLAGS', r'(\Deleted)')
+                proc_msg_count += 1
+        else:
+            raise OSError("bogofilter finished with the returncode: %d"
+                          % ret)
 
     client.expunge()
     client.close()
@@ -59,14 +53,16 @@ def process_folder(proc_fld_name, bogofilter_param, end_fld_name):
 processedCounter = 0
 hparser = email.Parser.Parser()
 config = ConfigParser()
-config.read("/etc/bogofilter-imap-train.cfg")
+config.read(os.path.expanduser("~/.bogofilter-imap-train-rc"))
 
 login = config.get("imap-training", "login")
 password = config.get("imap-training", "password")
-client = imaplib.IMAP4_SSL("luther.ceplovi.cz")
+server = config.get("imap-training", "server")
+client = imaplib.IMAP4_SSL(server)
 client.login(login, password)
 
 for box in [('Spam', 's', 'Deleted Items'), ('Ham', 'n', 'INBOX')]:
+    logging.debug('box = {0}'.format(box))
     processedCounter += process_folder(box[0], box[1], box[2])
 
 client.logout()