On 02Nov2012 20:15, Russell L. Harris <[email protected]> wrote:
| * Jamie Paul Griffin <[email protected]> [121102 19:36]:
| > I have set up macros that bind keys to pass messages to spamassassin
| > using sa-learn and then puts the message into the spam mailbox. Is
| > this the type thing you mean? The spam mailbox can later be used to
| > train spamassassin for future filtering, using procmail.
|
| Is the "return path" the true address of the sender?
No.
In legitimate messages it is where errors should go. For a person, that
would normally be themselves. But for a mailing list it would be the
list admin.
| If so, I would
| like to blacklist such addresses.
Not a lot of point.
| The "From:" field shown in the
| message index of Mutt almost always is rewritten -- sometimes to an
| address from which valid messages may originate -- so I hesitate to
| blacklist such addresses.
Indeed.
I've just implemented a script for this (adding a message subject as a
spam filter rule), and I have this macro:
macro index,pager SS "<save-message>+spool-spam-subj<enter><next-undeleted>"
"delete message as spam by subject"
I have arranged that messages saved to "+spool-spam-subj" get their subjects
saved to my "spam-subj" mail filing rules file. Details below.
I am not using spamassassin myself, but have a fairly effective
strategy:
- I catch "important" messages as being to "me" and "from" a
whitelist of known addresses (actually a whitelist of address
groups - the SO, family, friends and an assortment of business
entities like my credit union)
- I catch and alert on a short list of very specific messages from
monitoring systems, based on to: and from: _and_ subject:
- an ad hoc list of other special rules
- I have a zillion rules for various mailing lists, generally based on
to/cc: or sender:
- anything else lands in my "UNKNOWN" folder; it is 99% spam
I always sort that folder on subject when I visit it;
it makes tossing it much easier because a lot of spam gets repeated
in big chunks
This keeps my inbox fairly clean without nebulous bayesian filters etc.
The SO also keeps a blacklist of subject lines; I've been meaning to do
the same and your post has kicked me to do so.
How it works:
My mail setup is as follows:
- I fetch with getmail, delivering to my "spool" maildir
- I filter messaging using my mailfiler program, which monitors
a list of maildirs once a second, thus:
mailfiler monitor -d 1 ~/mail/spool ~/mail/spool-in ~/mail/spool-out
~/mail/spool-spam-subj
The "spool" rules are meant to divert the spam and deposit probably
nonspam into the "spool-in" maildir, which runs all the rules for the
mailing lists etc.
The rules for "spool" sources my "spam-subj" rule file, currently saying
this:
# spam matching by subject line
=spam SPAM-SUBJ subject:/^Anz e-banking alert
which files messages to the "spam" folder with the X-Label "SPAM-SUBJ"
if they match my (sole, so far) common spam subject.
The reason it is a separate file is that the script to add a new subject
line rewrites that file.
The script itself is here:
https://bitbucket.org/cameron_simpson/css/src/tip/bin-cs/email-add-spam-subject
The filter rules for the "spool-spam-subj" maildir (monitored above by the
mailfiler) read:
< env
# add this message's subject to the spam subject list
"|email-add-spam-subject" . .
spam . .
It pipes the message to the email-add-spam-subject script (which
rewrites the spam-suj rules file) and saves the message in my "spam"
folder where it all accumulates, like a septic tank.
email-add-spam-subject emits a new rule and sorts it into the existing
rules, rewriting the file if it changes.
Cheers,
--
Cameron Simpson <[email protected]>
Is it true, Sen. Bedfellow, that your wife rides with bikers? - Milo Bloom
