What happened to hammie.py?
It was deprecated a long while back. You can still access it if you want to, by using the hammie.py module (in the spambayes directory) as a script, or by using the hammiebulk.py module. The recommended method is to use sb_filter.py and sb_mboxtrain.py or sb_server.py with sb_upload.py (the former in your case, I would imagine).
Regardless, you're saying that there's an open ticket which would fix my dilemma?
Yes.
Iff true, this alone would allow me to switch from SA to SB. Learning Python seems like fun, maybe I'll idle the devel list and see if there's anything I can do to help.
Good plan :) The change isn't that difficult to make, but I can't see that I would be able to get to it before mid November (I'm going away for a while). One of the other developers might, though. Actually, although that is the correct long-term solution, you could add in subject modification to a local copy yourself without much trouble. I've only done a bare minimum of testing this, but it appears to work: """ *** hammie.py Mon Oct 20 19:51:31 2003 --- hammie2.py Mon Oct 20 19:47:20 2003 *************** *** 116,121 **** --- 116,125 ---- else: is_spam = False disp = options["Headers", "header_unsure_string"] + if disp in options["Headers", "notate_subject"]: + subj = msg["Subject"] + del msg["Subject"] + msg["Subject"] = "%s %s" % (disp, subj) if train: self.train(msg, is_spam, True) disp += ("; %."+str(options["Headers", "header_score_digits"])+"f") % Prob """
I greatly appreciate your assistance, Tony. Have a great weekend.
No worries, and thanks, I did :) =Tony Meyer
On Mon, Oct 20, 2003 at 07:52:36PM +1300, Tony Meyer wrote:
""" *** hammie.py Mon Oct 20 19:51:31 2003 --- hammie2.py Mon Oct 20 19:47:20 2003 *************** *** 116,121 **** --- 116,125 ---- else: is_spam = False disp = options["Headers", "header_unsure_string"] + if disp in options["Headers", "notate_subject"]: + subj = msg["Subject"] + del msg["Subject"] + msg["Subject"] = "%s %s" % (disp, subj) if train: self.train(msg, is_spam, True) disp += ("; %."+str(options["Headers", "header_score_digits"])+"f") % Prob """
First delve into Python, but here goes. I by no means mean to nitpick, but it would be very nice to see the Prob score in the Subject as well. I couldn't get it to work and also look to the config file for which disp to notate, so I hard coded it to make only messages classified as 'spam' appear in the subject with the score. Here is how I do it: """ --- hammie.py.orig Fri Nov 7 09:20:37 2003 +++ hammie.py Tue Nov 11 17:40:55 2003 @@ -130,6 +130,10 @@ disp += " (%d)"%x del msg[header] msg.add_header(header, disp) + if is_spam: + subj = msg["Subject"] + del msg["Subject"] + msg["Subject"] = "[%s] %s" % (disp, subj) if debug: disp = self.formatclues(clues) del msg[debugheader] """ So far, I am VERY happy with how this filter is performing. I am now rolling it out to the rest of my clients, and so far it is well received.
Hank, A couple of weeks ago you and Tony posted patches to make sb_filter obey the notate_to and notate_subject options. I've finally applied a modified version of them, that respects both of those options. I've committed it to the head, so it may not be in the next bugfix release, but will be in the next feature release.
I by no means mean to nitpick, but it would be very nice to see the Prob score in the Subject as well.
I didn't do this, partly because none of the other header-modifying spambayes applications does it, and partly because it made the To / Subject very messy looking. And I find that the scores get less interesting over time - they're fun to look at at first, and useful when debugging, but day to day I don't look at them. YMMV. 8-)
So far, I am VERY happy with how this filter is performing. I am now rolling it out to the rest of my clients, and so far it is well received.
Hope its still going well? -- Richie Hindle richie@entrian.com
On Tue, Nov 25, 2003 at 10:35:26PM +0000, Richie Hindle wrote:
A couple of weeks ago you and Tony posted patches to make sb_filter obey the notate_to and notate_subject options. I've finally applied a modified version of them, that respects both of those options. I've committed it to the head, so it may not be in the next bugfix release, but will be in the next feature release.
That's great to hear!
I didn't do this, partly because none of the other header-modifying spambayes applications does it, and partly because it made the To / Subject very messy looking. And I find that the scores get less interesting over time - they're fun to look at at first, and useful when debugging, but day to day I don't look at them. YMMV. 8-)
I agree with you here. The spam scores I am noticing are usually '1.00', and while people understand the "spam" part, they have a hard time understanding what this percentage really means. I see nothing valuable gained, and only used bandwidth.
Hope its still going well?
The configuration is IMAP (courier) server-based (Exim router), and the setup only takes a few minutes over the phone. The people in the office love it, and the best part is they can now keep it trained. I'm done running tests with domain-based filtering in addition to the user-based and default filter, but not many corporate clients are using it yet. In addition, I'm still awaiting feedback from several residential users concerning their satisfaction. My happiness may stem from ignorance of others' perception, but I'm hoping my effort to curb the tide of spam is both an effective and simple solution. So, no major headaches (yet). ;) Thanks for your time Richie. Have a good one, hank
participants (3)
-
hank＠marinar.com -
Richie Hindle -
Tony Meyer