Craig said:
Nowadays, most of spams are coming in html format.
Not from what I hear from sysadmins having to deal with the crap. Most of
it is text format.
So message rules do not work on those kinds of spam.
Primarily because the rules that look INSIDE the body of the mail don't work
well because they look for words OUTSIDE the HTML tags, so they will miss
something like "via said:
How can I kill those spams (in html)?
http://www.spampal.org
I suggest you use the following DNSBLs with SpamPal:
Spamhaus SBL+XBL
ORDB
SpamCop
NJABL
Don't use SPEWS since they do *not* target spam sources but instead rate a
domain regarding their spamminess, spam laziness, or spam friendliness.
SORBS would be okay except they are very slow to update their blocklist (and
they are a bit too SPEWS-bent in their methodology).
I suggest also using the following add-ins:
Bayesian
HTML-Modify (*)
UserLogfile (**)
(*) Up the criteria threshold from its default to something like 90. Don't
bother with the virus support since you should already be using something
that is more up-to-date and covers way more than just viral e-mails.
Disable the "HTML without alternative" option: it will tag e-mails that are
HTML-formatted but do not provide a plain-text part, and there are too many
HTML clients and services that still do that (e.g., Hotmail). Disable the
"illegal HTML tags" criteria option since I've reported and the author has
acknowledged problems in his parser. Disable "unneeded coding of HTML" if
you expect to get HTML-formatted e-mails from sites, like for newsletters.
Disable "many HTML comments" because the author's parser has a problem
there, too. Disable the "usage of security holes" criteria option since
those are very old exploits that you be patched for by now and can cause
false triggers when the old exploit isn't even used.
(**) The User Logfile plug-in keeps a plain text version of any spam-tagged
mails so you can go look at them in case there was a false positive. If
instead you move spam-tagged e-mails into the Junk folder and configure
auto-archive to permanently delete items older than, say, 2 days then you
have there to check for false positives, like when you register for a site
and are expecting a confirmation e-mail to complete that registration.
The Bayesian filter needs time to learn your particular history of e-mails
regarding spam and ham, unless you have old mails that you can feed into it
to pre-learn that filter (most Bayes filters don't let you train their
filter except over time depending on how many and how fast you get e-mails
for it to interrogate).
I use Outlook Express (and WinXP Pro-SP2).
Spampal runs as a local proxy to which any POP3/IMAP client can connect.
