#!/usr/bin/perl

# works with perl 4

# sendmail_stats Version 0.08
# (c) 2000-2002 Jeremy C. Reed
# this is free, but use at your own risk

## still needs a lot of work; it is probably only 10 percent finished
## and about 90 percent of this code needs to be rewritten.

# use like in your crontab:
#30 5 * * *      nobody /usr/contrib/bin/gunzip -c /var/log/maillog.0.gz |\
# /usr/local/bin/sendmail_stats | /usr/bin/mail -s "sendmail stats" postmaster

#for appending "@" and host name
$default_hostname = 'EDIT-THIS-default-domain.net';

$top_amt = 20;

$debug = 0; # 2 is more verbose


## MAIN routine

while ($line = <>) {

  if ($line =~ 
   /^(\w+)\s+(\d+)\s(\d+):(\d+:\d+)\s+([\w-]+)\s+([\w-]+)\[(\d+)\]:\s(\w+\d+):\s(.+)$/) {

    if ($debug) { print "log line in expected format\n"; }
    if ($debug > 1) { print "$line\n"; }

    $month = $1;
    $day = $2;
    $hour = $3;
    $min_sec = $4;
    $host = $5;
    $daemon = $6;
    $pid = $7;
    $id = $8;
    $data = $9;

    # get log start time
    if (! $log_start_time) { $log_start_time = "$month $day $hour $min_sec"; }

    if ($daemon eq 'sendmail' || $daemon eq 'sm-mta') {

      if ($data =~ /from=(.+),\ssize=/) {
        $from = $1;
        $from =~ s/[><]//g;
        if ($from eq "") {  # from a Mailer-Daemon
          $from = "<>";
        }

        $senders{$from}++;
        if ($debug) { print "sender is $from\n"; }

        if ($data =~ / size=(\d+)\,\s/) {
          $size = $1;
          $sizes{$id} += $size; 
          $total_size += $size;
        }
#$from_count++;

        if ($data =~ / relay=(.+)[$:]/) {
          $relay = $1;
          if ($relay =~ /may be/) { $relay =~ s/(.*) \(may be/$1/; }
          $sending_hosts{$relay}++;
          if ($debug) { print "sending host is $relay\n"; }
        }

      } # from=
      elsif (($data =~ /to=(.+)\, ctladdr=(.+), delay=/) ||
             ($data =~ /to=(.+)\, delay=/)) {
        $to = $1;
        if ($2) {
          $to = $2;
          $to =~ s/ \(\d+\/\d+\)//g;
        }

        if ($data =~ / stat=([^ ]+)/) {
          $stat = $1;
        }

        if ($stat eq "Sent") {
          if ($data =~ / mailer=([^,]+)\,\s/) {
            $mailer = $1;

            if ($mailer eq "relay") {
              $local_size += $sizes{$id};
              delete ($sizes{$id});
              $received_messages++;
            }
            if ($mailer eq "local" || $mailer eq "prog" ||
                $mailer eq "virtual" || $mailer eq "*file*") {

              $local_size += $sizes{$id};
              delete ($sizes{$id});
              $received_messages++;
              $to =~ s/[<"](.+)[>"]/$1/; 
              if ($to =~ /\,/) {
                foreach $username (split (/\,/, $to)) {
                  if ($username !~ /@/) {
                    $username = $username . '@' . $default_hostname;
                  }
                  $local_deliveries{$username}++;
                  if ($debug) { print "local delivery to $username\n"; }
                }
              }
              else {
                $local_deliveries{$to}++;
                if ($debug) { print "local delivery to $to\n"; }
              }
            } # a local delivery
            elsif ($mailer =~ /smtp|^relay$/i) {
              $sent_messages++;

              if ($data =~ / relay=([^,]+),\s.+$/) {
                $relay = $1;
                $destination_hosts{$relay}++;
                if ($debug) { print "destination host is $relay\n"; }
              }

            } # sent out
 
          } # mailer=

          $hours{$hour}++;

        } # stat is Sent
        else {
          $sending_problems++;
          if ($debug) {
            print "sending problem \"stat=$stat\" ($sending_problems)\n";
          }
        }
      } # to=

#      $months{$month}++;
#      $days{$day}++;
#      $total++;
    }
  }
  else {
    if ($debug) { print "log line not in expected format:\n$line\n"; }
  }
}

# get log start time
$log_end_time = "$month $day $hour $min_sec";

## 1048576 is one meg
$total_size_mb = sprintf ("%.1f", ($total_size/1048576));
$local_size_mb = sprintf ("%.1f", ($local_size/1048576));

print <<END_OF_DATA;
Sendmail log analysis

Log starts at $log_start_time and ends at $log_end_time

Total bytes transferred: $total_size ($total_size_mb MB)
Total bytes In: $local_size ($local_size_mb MB)
Messages Out: $sent_messages
Messages In: $received_messages
Sending Problems: $sending_problems

END_OF_DATA

#print "from_count $from_count\n";

$max = 0;
for ($c = 0; $c < 24; $c++) {
  if ($c < 10) { $i = "0$c"; }
  else { $i = $c; }
  if ($hours{$i} > $max) { $max = $hours{$i}; }
}

$scale = int($max/50);
$scale = 1 if $scale == 0;

print "Messages per hour (each dot is $scale messages)\n________________\n";

for ($c = 0; $c < 24; $c++) {
  if ($c < 10) { $i = "0$c"; }
  else { $i = $c; }
  printf("%3d: %6d %s\n", $i, $hours{$i}, "." x int($hours{$i}/$scale));
}

&show_sending_hosts;
&show_destination_hosts;
&show_local_deliveries;
&show_senders;

print "\n";

## end

sub show_sending_hosts {
  $count = 0;
  print "\nTop $top_amt Sending Hosts\n____________________\n";

  foreach $key (sort {
                  $sending_hosts{$b} <=> $sending_hosts{$a} }
                keys %sending_hosts) {
    $count++;
    if ($count <= $top_amt) {
      print "$sending_hosts{$key}\t$key\n";
    }
    $total_sending_hosts++;
    $total_sending_host_messages += $sending_hosts{$key};
  }

  print "\n$total_sending_hosts sending hosts attempted $total_sending_host_messages messages\n"; 
} # show_sending_hosts

sub show_destination_hosts {
  $count = 0;
  print "\nTop $top_amt Destination Hosts\n________________________\n";

  foreach $key (sort {
                  $destination_hosts{$b} <=> $destination_hosts{$a} }
                keys %destination_hosts) {
    $count++;
    if ($count <= $top_amt) {
      print "$destination_hosts{$key}\t$key\n";
    }
    $total_destination_hosts++;
    $total_destination_host_messages += $destination_hosts{$key};
  }

  print "\n$total_destination_hosts destination hosts received $total_destination_host_messages messages\n"; 

} # show_destination_hosts

sub show_local_deliveries {
  $count = 0;
  print "\nTop $top_amt Local Deliveries\n_______________________\n";

  foreach $key (sort {
                  $local_deliveries{$b} <=> $local_deliveries{$a} }
                keys %local_deliveries) {
    $count++;
    if ($count <= $top_amt) {
      print "$local_deliveries{$key}\t$key\n";
    }
    $local_accounts++;
    $total_local_messages += $local_deliveries{$key};
  }
  print
 "\n$local_accounts local accounts received mail ($total_local_messages messages combined).\n";

} # show_local_deliveries

sub show_senders {
  $count = 0;
  print "\nTop $top_amt Senders\n______________\n";

  foreach $key (sort {
                  $senders{$b} <=> $senders{$a} }
                keys %senders) {
    $count++;
    if ($count <= $top_amt) {
      print "$senders{$key}\t$key\n";
    }
    $total_senders++;
    $total_senders_messages += $senders{$key}; 
  }
  print
 "\n$total_senders senders sent a combined $total_senders_messages messages.\n";

} # show_senders

sub show_data {

  print <<END_OF_DATA;
$month $day $hour $min_sec 
$host $daemon
$pid pid
$id
$data

END_OF_DATA

} #show data