I have implemented John's idea and it works like a charm!
I use postfix in my mail gateway at the perimiter, I have :
always_bcc = [EMAIL PROTECTED]
The reason I'm bcc-ing is because my mail-gateway is not suitable to
store large quantity of emails, very plain server.
I wrote a simple perl script below to automate all the processing.
/home/spamtrapcc/.maildir/.UndetectedSPAMS/cur/ is the folder where all
users drop the undetected spams.
/home/spamtrapcc/.maildir/.WhitelistSPAMS/cur/ users drop their
whitelist email request here
/home/ccmail/.maildir/new is the [EMAIL PROTECTED] mailbox
- this folder grows very quickly, I have a cron job to clean this
directory on a daily basis:
find /home/ccmail/.maildir/new -type f -mtime +3 -exec rm {} \;
Btw as you already see, I use Maildir format instead of Mailbox, just
love maildir!!
My perl is not good, the script only works in my environment. Feel free
to modify and post it back to this list if you have improvements.
Cheers
Christian.
--
#!/usr/bin/perl
chdir("/home/spamtrapcc/.maildir/.UndetectedSPAMS/cur/") or die("Failed to
chdir: $!\n");
opendir(UND, ".") or die("Opendir failed: $!\n");
my @undfiles = grep {!/^\.+$/} readdir(UND);
closedir(UND);
my %badmessageids = ();
foreach $undfile (@undfiles) {
open(UFILE, $undfile) or die("Failed to open $undfile for reading: $!\n");
while (<UFILE>) {
if (/^Message-ID: <(.*?)>/) {
$badmessageids{"$1"} = 1;
last;
} }
close(UFILE);
}
chdir("/home/spamtrapcc/.maildir/.WhitelistSPAMS/cur/") or die("Failed to
chdir: $!\n");
opendir(WHITE, ".") or die("Opendir failed: $!\n");
my @whitefiles = grep {!/^\.+$/} readdir(WHITE);
closedir(WHITE);
my %goodmessageids = ();
foreach $whitefile (@whitefiles) {
open(WFILE, $whitefile) or die("Failed to open $whitefile for reading: $!\n");
while (<WFILE>) {
if (/^Message-ID: <(.*?)>/) {
$goodmessageids{"$1"} = 1;
last;
} }
close(WFILE);
}
chdir("/home/ccmail/.maildir/new") or die("chdir failed: $!\n");
opendir(CCMAIL, ".") or die("Failed to opendir CCMAIL: $!\n");
my @ccmailfiles = grep {!/^\.+$/} readdir(CCMAIL);
closedir(CCMAIL);
foreach $cm (@ccmailfiles) {
open(CCFILE, $cm) or die("Failed to open $cm for reading: $!");
while (<CCFILE>) {
if (/^Message-ID: <(.*?)>/i) {
if (defined($badmessageids{$1})) {
print "Copying $cm to blackspam...\n";
system "cp $cm /home/ccmail/.maildir/.blackspams/tmptmp";
chown 1008, 1008, "/home/ccmail/.maildir/.blackspams/tmptmp/$cm";
} elsif (defined($goodmessageids{$1})) {
print "Copying $cm to whitespam...\n";
system "cp $cm /home/ccmail/.maildir/.whitespams/tmptmp";
chown 1008, 1008, "/home/ccmail/.maildir/.whitespams/tmptmp/$cm";
}
}
}
}
Subject: Re: Exchange/Outlook - how do you learn spam?
Date: Wed, Jun 22, 2005 at 04:25:03PM +0100
Quoting John Hall ([EMAIL PROTECTED]):
: Sadly 'script' is rather a grand term. I haven't got around to automating
: things properly, so I manually do the following roughly once a week or so:
:
: # get a list of message-ids from the imap folder
: $ grep -i "^message-id: " imap-folder | sed -e 's/.*<\(.*\)>.*/\1/' > msgids
:
: # get e-mails from the current archive folder
: $ cat /var/local/archive/mail | spamlearn msgids > mbox
:
: # get e-mails from the last 3 weeks of archives
: $ zcat /var/local/archive/mail.[123].gz | spamlearn msgids >> mbox
:
: # feed to SA
: $ sa-learn --spam --mbox < mbox
:
:
: The archive I keep is rotated once a week and gzipped. The perl script,
: spamlearn, is attached. This just scans stdin for messages in the supplied
: file and writes them to stdout. Any that it finds it removes from the
: supplied file. Hopefully, by the end, the file 'msgids' is empty.
:
: This could obviously be vastly improved and automated - I just haven't got
: round to it yet.
:
: regards,
: John
:
:
: begin 666 spamlearn.dat
: M(R$O=7-R+V)I;B]P97)L"@IU<V4@<W1R:6-T.PH*;7D@)6US9VED<SL*;7D@
: M)'-T871E.PH*;7D@)%-4051%7U-405)4(#T@,#L*;7D@)%-4051%7TA%041%
: M4E,@/2 Q.PIM>2 D4U1!5$5?3U544%545$E.1R ](#(["FUY("135$%415]4
: M2%)/5TE.1U]!5T%9(#T@,SL*"B,@;&]O:R!A="!A<F=U;65N=',*:[EMAIL PROTECTED]'-C
: M86QA<B! 05)'5B A/2 Q*2!["@EP<FEN=" B57-A9V4Z('-P86UL96%R;B!M
: M<V=I9'-<;B(["@EE>&ET(#$["GT*"FUY("1M<V=I9'-?9FEL92 ]('-H:69T
: M($!!4D=6.PII9B H(2 M92 D;7-G:61S7V9I;&4I('L*"7!R:6YT("(D;7-G
: M:61S7V9I;&5<.B!F:6QE(&YO="!F;W5N9%QN(CL*"65X:70@,3L*?0H*(R!G
: M970@;&ES="!O9B!M<V=I9',*;W!E;B!-4T=)1%,L("(\("1M<V=I9'-?9FEL
: M92(@;W(@9&EE(")5;F%B;&[EMAIL PROTECTED]&\@;W!E;B D;7-G:61S7V9I;&5<.B D(5QN
: M(CL*=VAI;&[EMAIL PROTECTED])1%,^*2!["@EC:&]M<#L*"21M<V=I9'-[)%]](#T@
: M,3L*?0IC;&]S92!-4T=)1%,["@IM>2! 8W5R<F5N=%]M<V<["B1S=&%T92 ]
: M("135$%415]35$%25#L*"G=H:6QE("@D7R ](#P^*2!["@EI9B H)'-T871E
: M(#T]("135$%415]35$%25"D@>PH)"6EF("@O7D9R;[EMAIL PROTECTED]@>PH)"0DD<W1A
: M=&4@/2 D4U1!5$5?2$5!1$524SL*"0D)0&-U<G)E;G1?;7-G([EMAIL PROTECTED]"1?*3L*
: M"0E]"@E]"@EE;'-I9B H)'-T871E(#T]("135$%415](14%$15)3*2!["@D)
: M:[EMAIL PROTECTED]"]>365S<V%G92U)1#I<<RH\*"XJ*3XO:2D@>PH)"0EP=7-H($!C=7)R
: M96YT7VUS9RP@)%\["@D)"6UY("1M<V=I9" ]("0Q.PH)"0DC('!R:6YT(%-4
: M1$524B B365S<V%G93H@)#$@(CL*"0D):[EMAIL PROTECTED]"1M<V=I9'-[)&US9VED?2D@
: M>PH)"0D)<')I;G0@(B1?(B!F;W)E86-H("A 8W5R<F5N=%]M<V<I.PH)"0D)
: M)'-T871E(#T@)%-4051%7T]55%!55%1)3D<["@D)"0ED96QE=&4@)&US9VED
: M<WLD;7-G:61].PH)"0D)(R!P<FEN="!35$1%4E(@(E-P86U<;B(["@D)"7T*
: M"0D)96QS92!["@D)"0DD<W1A=&4@/2 D4U1!5$5?5$A23U=)3D=?05=!63L*
: M"0D)"2,@<')I;[EMAIL PROTECTED])2(")/2UQN(CL*"0D)?0H)"7T*"0EE;'-I9B H
: M+UY<<RHD+RD@>PH)"0DD<W1A=&4@/2 D4U1!5$5?5$A23U=)3D=?05=!63L*
: M"0E]"@D)96QS92!["@D)"7!U<V@@0&-U<G)E;G1?;7-G+" D7SL*"0E]"@E]
: M"@EE;'-I9B H)'-T871E(#T]("135$%415]/5510551424Y'*2!["@D):68@
: M*"]>1G)O;2 O*2!["@D)"21S=&%T92 ]("135$%415](14%$15)3.PH)"0E
: M8W5R<F5N=%]M<V<@/2 H)%\I.PH)"7T*"0EE;'-E('L*"0D)<')I;G0["@D)
: M?0H)?0H)96QS:[EMAIL PROTECTED]"1S=&%T92 ]/2 D4U1!5$5?5$A23U=)3D=?05=!62D@
: M>PH)"6EF("@O7D9R;[EMAIL PROTECTED]@>PH)"0DD<W1A=&4@/2 D4U1!5$5?2$5!1$52
: M4SL*"0D)0&-U<G)E;G1?;7-G([EMAIL PROTECTED]"1?*3L*"0E]"@E]"GT*"B,@<F5W<FET
: M92!M<V=I9',*;W!E;B!-4T=)1%,L("(^("1M<V=I9'-?9FEL92(@;W(@9&EE
: M(")5;F%B;&[EMAIL PROTECTED]&\@;W!E;B D;7-G:61S7V9I;&5<.B D(5QN(CL*9F]R96%C
: M:" H:V5Y<R E;7-G:61S*2!["@EP<FEN="!-4T=)1%,@(B1?7&XB.PI]"F-L
: 5;W-E($U31TE$4SL*"F5X:70@,#L*
: `
: end
:
:
