Let train_dspam_from_folder generate debug messages.

2 files changed, 60 insertions, 2 deletions

diff --git a/train_bogofilter b/train_bogofilter
new file mode 100755
index 0000000..d4e5bb2
--- /dev/null
+++ b/train_bogofilter
@@ -0,0 +1,49 @@
+#!/usr/bin/python
+import imaplib, subprocess, email, sys
+import logging
+import email.Parser
+from ConfigParser import ConfigParser
+
+cmd_string = "/usr/bin/dspam --user dspam --class=spam "\
+     + "--source=error --deliver=summary --stdout --signature='%s'"
+
+DEBUG=False
+
+def debug(msg):
+    if DEBUG:
+        print >>sys.stderr,msg
+
+hparser = email.Parser.Parser()
+config = ConfigParser()
+config.read("/etc/dspam-imap-train.cfg")
+
+login = config.get("imap-training","login")
+password = config.get("imap-training","password")
+client = imaplib.IMAP4_SSL("luther.ceplovi.cz")
+client.login(login, password)
+client.select("Public folders/Junk")
+status, resp = client.search(None, "ALL")
+messages = resp[0].split()
+processedCounter = 0
+debug(messages)
+
+for msgId in messages:
+    # or no .PEEK ... do I want to mark a message as Seen?
+    typ, msg_data = client.fetch(msgId, '(BODY.PEEK[HEADER.FIELDS (SUBJECT FROM X-DSPAM-SIGNATURE)])')
+    debug("msgId = %s" % msgId)
+    headers = hparser.parsestr(msg_data[0][1],headersonly=True)
+    debug("headers:\n%s" % headers)
+    if 'X-Dspam-Signature' in headers.keys():
+        ret = subprocess.Popen(cmd_string \
+            % headers['X-Dspam-Signature'], shell=True).wait()
+        if ret == 0:
+            typ, response = client.store(msgId, '+FLAGS', r'(\Deleted)')
+            processedCounter += 1
+        else:
+            raise OSError, "dspam finished with failure code: %d" % ret
+
+client.expunge()
+client.close()
+client.logout()
+if processedCounter > 0:
+    debug("Processed %d spam messages." % processedCounter)
diff --git a/train_dspam_from_folder b/train_dspam_from_folder
index 2aed4b7..6cc9323 100755
--- a/train_dspam_from_folder
+++ b/train_dspam_from_folder
@@ -1,12 +1,18 @@
 #!/usr/bin/python
 # Mind you, we have only python 2.4.3 on RHEL-5
-import imaplib, subprocess, email
+import imaplib, subprocess, email, sys
 import email.Parser
 from ConfigParser import ConfigParser
 
 cmd_string = "/usr/bin/dspam --user dspam --class=spam "\
      + "--source=error --deliver=summary --stdout --signature='%s'"
 
+DEBUG=False
+
+def debug(msg):
+    if DEBUG:
+        print >>sys.stderr,msg
+
 hparser = email.Parser.Parser()
 config = ConfigParser()
 config.read("/etc/dspam-imap-train.cfg")
@@ -19,11 +25,14 @@ client.select("Public folders/Junk")
 status, resp = client.search(None, "ALL")
 messages = resp[0].split()
 processedCounter = 0
+debug(messages)
 
 for msgId in messages:
     # or no .PEEK ... do I want to mark a message as Seen?
     typ, msg_data = client.fetch(msgId, '(BODY.PEEK[HEADER.FIELDS (SUBJECT FROM X-DSPAM-SIGNATURE)])')
+    debug("msgId = %s" % msgId)
     headers = hparser.parsestr(msg_data[0][1],headersonly=True)
+    debug("headers:\n%s" % headers)
     if 'X-Dspam-Signature' in headers.keys():
         ret = subprocess.Popen(cmd_string \
             % headers['X-Dspam-Signature'], shell=True).wait()
@@ -37,4 +46,4 @@ client.expunge()
 client.close()
 client.logout()
 if processedCounter > 0:
-    print "Processed %d spam messages." % processedCounter
+    debug("Processed %d spam messages." % processedCounter)