#!/usr/bin/perl -w

# This code was written by Felix Buenemann.
#
# It was posted to the spamdyke-users mailing list on Nov 4 2008.
#
# It has been modified by Brent Gardner:
# - removed commented-out code
# - fixed some indenting 'cause I have OCD :P
# - fixed a bug that prevented processing of Spamdyke info in Qmail logs
#


my $build = "2010022601";
use diagnostics;
use strict;
use Getopt::Long;
my $tldtop = 0;
my $detailed = 1;
my $syslog = 1;
GetOptions (
        "tld=i"   => \$tldtop,
        "detail!" => \$detailed
) or exit 1;

# Usage:  # cat /var/log/qmail/smtpd/current | ./this_file

my %status = ();        # hash of status connections
my %origin = ();        # hash of tld per status code
my %originsum = ();     # hash of tld per status code sums
my %rblstat = ();       # hash of DNSBL lists matched
my %rhsblstat = ();     # hash of RHSBL lists matched
my %rdnsblstat = ();    # hash of patterns in IP_IN_RDNS_BLACKLIST  matched
my ($allow, $deny, $timeout, $error, $allowpercentage, $timeoutpercentage, 
$errorpercentage, $spampercentage, $sum, $rblsum, $rhsblsum, $rdnsblsum);

sub percentage {
        my $num = pop;
        my $sum = pop;
        return "  0.00%" unless $sum;
        $num = $num/$sum*100.0;
        return sprintf("%3d.%02d%%", $num, ($num - int($num))*100);
}

#print "spamdyke-stats build $build\n\n";

while(<>){
        if( substr($_,0,1) eq '@' ) {
                # this gets rid of the leading 25-char timestamp and a space
                #
                $_ = substr $_,26;

                # BG: this gets rid of all lines that don't start with "spamdyke[<pid>]: "
                #
                next unless ( m/(^spamdyke\[[0-9]+\]: )(.*)/i );

                # BG: this in conjunction with the previous line gets rid of the leading "spamdyke[<pid>]: "
                #
                $_ = $2;

        } else {
                my ($hostname,$id,$line) = split / /, substr($_,16), 3;
                next unless substr($id,0,9) eq 'spamdyke[';
                my ($line2,$line3) = split /:/, substr($line,0), 2;
                next unless index($line2,"call]") > 0;
                $_ = $line3;
        }
        if( m/^(ALLOWED|ERROR|TIMEOUT|((DENIED|FILTER)_[^ ]+))/ ) {
                my $line = substr $_,length $1;
                $_ = $1;
                if( $detailed ) {
                        if( m/FILTER_RBL_MATCH/ ){
                                $line =~ m/rbl: (\S+)/;
                                $rblstat{$1}++;
                                $rblsum++;
                        }
                        elsif( m/FILTER_RHSBL_MATCH/ ){
                                $line =~ m/rhsbl: (\S+)/;
                                $rhsblstat{$1}++;
                                $rhsblsum++;
                        }
                        elsif( m/FILTER_IP_IN_RDNS_BLACKLIST/ ){
                                $line =~ m/keyword: (\S+)/;
                                $rdnsblstat{$1}++;
                                $rdnsblsum++;
                        }
                }
                next if m/^FILTER_/;
                $status{$_}++;
                if($tldtop and $line =~ m/ origin_rdns: ([^ ]+)/) {
                        my $rdns = $1;
                        $originsum{$_}++;
                        if($rdns =~ m/^\(unknown\)$/){
                                next;
                        } elsif($rdns =~ m/\.(com|net)$/){
                                $origin{$_}{$1}++;
                        } elsif($rdns =~ m/\.([a-z]{2,2}\.[a-z]{2,2})$/){ # co.uk
                                $origin{$_}{$1}++;
                        } elsif($rdns =~ m/\.([a-z]{2,})$/){ # de, ru, ...
                                $origin{$_}{$1}++
                        } else {
                                next;
                        }
                }
        }
}

$allow = 0;
$deny = 0;
$error = 0;
$timeout = 0;

foreach my $stat (sort keys %status){
        if( $stat =~ m/ALLOWED/ ){
                $allow = $status{$stat};
        }
        elsif( $stat =~ m/TIMEOUT/ ){
                $timeout += $status{$stat};
        }
        elsif( $stat =~ m/ERROR/ ){
                $error += $status{$stat};
        }
        else{
                $deny += $status{$stat};
        }
}
$sum = ($deny + $error + $timeout + $allow);

foreach my $key (sort { $status{$b} <=> $status{$a} || $a cmp $b; } keys %status){
        printf "%8d  %s  $key\n", $status{$key}, percentage($sum, $status{$key});
        if(length scalar(keys %rblstat)  and $key eq "DENIED_RBL_MATCH" ){
                print "--------------- Breakdown ---------------\n";
                foreach my $key (sort { $rblstat{$b} <=> $rblstat{$a} || $a cmp $b; } keys %rblstat){
                        printf "%8d  %s  $key\n", $rblstat{$key}, percentage($rblsum,$rblstat{$key});
                }
                print "-----------------------------------------\n";
        }
        elsif(length scalar(keys %rhsblstat) and $key eq "DENIED_RHSBL_MATCH" ){
                print "--------------- Breakdown ---------------\n";
                foreach my $key (sort { $rhsblstat{$b} <=> $rblstat{$a} || $a cmp $b; } keys %rhsblstat){
                        printf "%8d  %s  $key\n", $rhsblstat{$key}, percentage($rhsblsum,$rhsblstat{$key});
                }
                print "-----------------------------------------\n";
        }
        elsif(length scalar(keys %rdnsblstat) and $key eq "DENIED_IP_IN_RDNS" ){
                print "--------------- Breakdown ---------------\n";
                foreach my $key (sort { $rdnsblstat{$b} <=> $rdnsblstat{$a} || $a cmp $b; } keys %rdnsblstat){
                        printf "%8d  %s  $key\n", $rdnsblstat{$key}, percentage($rdnsblsum,$rdnsblstat{$key});
                }
                print "-----------------------------------------\n";
        }
        if($tldtop && $origin{$key}) {
                my $top = $tldtop;
                print "--------------- Top $top TLD ---------------\n";
                my $tldsum = 0;
                my $lastsum = 0;
                my @tldgroup = ();
                my %neworigin = ();
                foreach my $tld (sort { $origin{$key}{$a} <=> $origin{$key}{$b} } keys %{$origin{$key}}){
                        if(($origin{$key}{$tld}/$originsum{$key}*100) == $lastsum) {
                                push(@tldgroup, $tld);
                        } else {
                                if(scalar @tldgroup) {
                                        $neworigin{join(', ', @tldgroup)} = $lastsum;
                                        @tldgroup = ();
                                }
                                push(@tldgroup, $tld);
                        }
                        $lastsum = $origin{$key}{$tld}/$originsum{$key}*100;
                        $tldsum += $origin{$key}{$tld};
                }
                if(scalar @tldgroup) {
                        $neworigin{join(', ', scalar(@tldgroup))} = $lastsum * length scalar(@tldgroup);
                }
                foreach my $tld (sort { $neworigin{$b} <=> $neworigin{$a} } keys %neworigin){
                        printf "%s\t$tld\n", percentage($originsum{$key}, $neworigin{$tld}/100.0*$originsum{$key});
                        last unless --$top;
                }
                print "-----------------------------------------\n";
        }
        
}
my $format_summary = "%8d  %s";
print "\n";
print "---------------- Summary ----------------\n";
printf "Allowed: $format_summary\n", $allow, percentage($sum, $allow);
printf "Timeout: $format_summary\n", $timeout, percentage($sum, $timeout);
printf "Errors : $format_summary\n", $error, percentage($sum, $error);
printf "Denied : $format_summary\n", $deny, percentage($sum, $deny);
printf "Total  : $format_summary\n", $sum, percentage($sum, $sum);