#!/usr/bin/perl

# If this script fails to find your SpamAssassin.cache.db file then put
# the filename on the command-line.

use DBI;

# Connect to the database
my $cachefile = shift @ARGV || '/var/spool/MailScanner/incoming/SpamAssassin.cache.db';

my $dbh = DBI->connect("dbi:SQLite:$cachefile");

my($sql, $hash);

# Get total records
$sql = "SELECT COUNT(*) AS total, strftime('%s','now')-MIN(first) AS oldest_first, strftime('%s','now')-MAX(first) AS newest_first, strftime('%s','now')-MIN(last) AS oldest_last, strftime('%s','now')-MAX(last) AS newest_last FROM cache";
$hash = $dbh->selectrow_hashref($sql);
# Get hit rate percent
$sql = "SELECT (((SELECT SUM(count) FROM cache WHERE count>1)-(SELECT COUNT(*) FROM cache WHERE count>1))*100)/(SUM(count)-(SELECT count(*) FROM cache WHERE COUNT>1)) AS hitrate_percent FROM cache";
$hash1 = $dbh->selectrow_hashref($sql);
printf("--------- TOTALS ---------\n");
printf("Total records:       %d\nFirst seen (oldest): %d sec\nFirst seen (newest): %d sec\nLast seen (oldest):  %d sec\nLast seen (newest):  %d sec\nCache Hit Rate       %d%%\n",$hash->{total},$hash->{'oldest_first'},$hash->{'newest_first'},$hash->{'oldest_last'},$hash->{'newest_last'},$hash1->{hitrate_percent});
$hash = undef;
$hash1 = undef;

# Get oldest and newest non-spam records
printf("-------- NON-SPAM --------\n");
$sql = "SELECT COUNT(*) AS total, strftime('%s','now')-MIN(first) AS oldest_first, strftime('%s','now')-MAX(first) AS newest_first, strftime('%s','now')-MIN(last) AS oldest_last, strftime('%s','now')-MAX(last) AS newest_last FROM cache WHERE sasaysspam=0 AND virusinfected<1";
$hash = $dbh->selectrow_hashref($sql);
printf("Total records:       %d\nFirst seen (oldest): %d sec\nFirst seen (newest): %d sec\nLast seen (oldest):  %d sec\nLast seen (newest):  %d sec\n",$hash->{total},$hash->{oldest_first},$hash->{newest_first},$hash->{oldest_last},$hash->{newest_last});
$hash = undef;

# Get oldest and newest low-spam records
printf("-------- LOW-SPAM --------\n");
$sql = "SELECT COUNT(*) AS total, strftime('%s','now')-MIN(first) AS oldest_first, strftime('%s','now')-MAX(first) AS newest_first, strftime('%s','now')-MIN(last) AS oldest_last, strftime('%s','now')-MAX(last) AS newest_last FROM cache WHERE sasaysspam>0 AND sahighscoring=0 and virusinfected<1";
$hash = $dbh->selectrow_hashref($sql);
printf("Total records:       %d\nFirst seen (oldest): %d sec\nFirst seen (newest): %d sec\nLast seen (oldest):  %d sec\nLast seen (newest):  %d sec\n",$hash->{total},$hash->{oldest_first},$hash->{newest_first},$hash->{oldest_last},$hash->{newest_last});
$hash = undef;


# Get oldest and newest high-spam records
printf("------- HIGH-SPAM --------\n");
$sql = "SELECT COUNT(*) AS total, strftime('%s','now')-MIN(first) AS oldest_first, strftime('%s','now')-MAX(first) AS newest_first, strftime('%s','now')-MIN(last) AS oldest_last, strftime('%s','now')-MAX(last) AS newest_last FROM cache WHERE sasaysspam>0 AND sahighscoring>0 AND virusinfected<1";
$hash = $dbh->selectrow_hashref($sql);
printf("Total records:       %d\nFirst seen (oldest): %d sec\nFirst seen (newest): %d sec\nLast seen (oldest):  %d sec\nLast seen (newest):  %d sec\n",$hash->{total},$hash->{oldest_first},$hash->{newest_first},$hash->{oldest_last},$hash->{newest_last});
$hash = undef;

# Get oldest and newest virus records
printf("-------- VIRUSES  --------\n");
$sql = "SELECT COUNT(*) AS total, strftime('%s','now')-MIN(first) AS oldest_first, strftime('%s','now')-MAX(first) AS newest_first, strftime('%s','now')-MIN(last) AS oldest_last, strftime('%s','now')-MAX(last) AS newest_last FROM cache WHERE virusinfected>0";
$hash = $dbh->selectrow_hashref($sql);
printf("Total records:       %d\nFirst seen (oldest): %d sec\nFirst seen (newest): %d sec\nLast seen (oldest):  %d sec\nLast seen (newest):  %d sec\n",$hash->{total},$hash->{oldest_first},$hash->{newest_first},$hash->{oldest_last},$hash->{newest_last});
$hash = undef;

# List top 5 hashes and count
printf("----- TOP 5 HASHES -------\n");
$sql = "SELECT md5, count, strftime('%s','now')-first AS first, strftime('%s','now')-last AS last FROM cache WHERE count>1 ORDER BY count DESC LIMIT 5";
$sth = $dbh->prepare($sql);
$sth->execute();
printf("MD5\t\t\t\t\tCOUNT\tFIRST\tLAST\n");
while($hash = $sth->fetchrow_hashref) {
 printf("%s\t%d\t%d\t%d\n",$hash->{md5},$hash->{count},$hash->{first},$hash->{last});
} 

$dbh->disconnect;