#!/usr/bin/perl

#
# a program to print out statistics on our junkmail filter
#
# usage: jstat starttime endtime
#
#   where starttime and endtime are of the form:
#         MM/DD/YYYY [HH:MM[:SS]]
#
# jstat looks at all messages from that time interval
# and prints the following results
#
#   # of messages in interval
#   # and % of total messages filed as good
#   # and % of total messages filed as junk
#   # and % of total messages improperly filed as good
#   # and % of total messages improperly filed as junk
#   # and % of total messages that are good
#   # and % of total messages that are junk
#   % accuracy of filter 
#   % of false positives (filed as junk)
#   % of false negatives (filed as good)
#  

use Time::Local;

$logfile = "$ENV{'HOME'}/bin/junkmail.log";

$hour1=0; $minute1=0; $second1=0;
$hour2=0; $minute2=0; $second2=0;

if ($#ARGV < 1 || $#ARGV > 3) {
  print "usage: $0 startdate [starttime] enddate [endtime]\n";
  print "\twhere date is MM/DD/YYYY and time is HH:MM[:SS]\n";
  exit -1;
}

# grab the first argument, the starting date
$_ = shift;
($mon1, $day1, $year1) = split (/\//);

# grab the second argument
$_ = shift;
if (/:/) {
  # starting time is specified
  ($hour1, $minute1, $second1) = split(/:/, $_, 3);
  # grab the next argument
  $_ = shift;
} 

# grab the ending date
($mon2, $day2, $year2) = split (/\//);

if (defined($_ = shift)) {
  # end time is specified
  ($hour2, $minute2, $second2) = split(/:/, $_, 3);
};

$starttime = timelocal($second1, $minute1, $hour1, 
		       $day1, $mon1-1, $year1);
$endtime = timelocal($second2, $minute2, $hour2, 
		       $day2, $mon2-1, $year2);

print "$starttime $endtime \n";

# step through the logfile
open (LOG, "$logfile") || die "can't open $logfile for read";
while (<LOG>) {
  /(\d+) (-?\d+) ?(\d+)?/;
  if ($starttime <= $1 && $1 <= $endtime) {
    if ($2 == -1) {
      $junk++;
      $badjunk++ if ($3 == 1);
    } elsif ($2 == 0) {
      $good++;
      $badgood++ if ($3 == 1);
    }
  }
}

close LOG;

# good is actually good + misfiled junk - misfiled good
# junk is actually junk - misfiled junk + misfiled good
$comp = $badjunk - $badgood;
$total = $junk + $good;

# print the statistics
printf ("%d total messages\n", $total);
printf ("%d filed as good, %d filed as junk\n", $good, $junk);
printf ("%d misfiled as good, %d misfiled as junk\n", $badgood, $badjunk);
printf ("%d actual good, %d actual junk\n", $good + $comp,  $junk - $comp);
printf ("Junkmail is %5.2f%% of total mail\n", 
	100.0*($junk - $comp)/($total));
printf ("False positive is: %5.2f%%, false negative is: %5.2f%%\n",
	100.0*$badjunk/($good+$comp), 100.0*$badgood/($junk-$comp));