#!/usr/bin/perl -w

use Getopt::Long qw(:config no_ignore_case);
use strict;

our ($showall, $help);
our $threshold = 10;
our $spamfile;

GetOptions (
    'a'				=> \$showall,
    'threshold|t=f'	=> \$threshold,
    'help|h|?'      => \$help,
);

&usage if $help;

if ($ARGV[0]) {
  $spamfile = $ARGV[0];
}
else {
  die "FATAL: Specify MBOX\n";
}

our $emails = 0;
our (%TO, %FROM);
our ($totalTo, $totalFrom);
our ($lackingTo, $lackingFrom);
our ($percTo, $percFrom, $percTot);
our $header;

open SPAMBOX, $spamfile;

while (<SPAMBOX>) {
  if (/^From\s+/) {
    $emails++;
    $header = 1;
  }
  
  if (/^\s*$/) {
    $header = 0;
  }
  
  if (/^To:\s+(?:[[:alnum:]\s"']+\s+)?[\<]?([^[:space:]@]+@(?:(?:[[:alnum:]\-])+\.)+[[:alpha:]]{2,6})[\>]?$/ && $header != 0) {
    $TO{$1}++;
    $totalTo++;
  }
  
  if (/^From:\s+(?:[[:alnum:]\s"']+\s+)?[\<]?([^[:space:]@]+@(?:(?:[[:alnum:]\-])+\.)+[[:alpha:]]{2,6})[\>]?$/ && $header != 0) {
    $FROM{$1}++;
    $totalFrom++;
  }
}

close SPAMBOX;

$lackingTo = ($emails - $totalTo);
$lackingFrom = ($emails - $totalFrom);
print "Number of emails: $emails\n";
print "Threshold: $threshold\%\n";

print "Number of unique recipients: " . (keys %TO) . "\n";
print "Total number of To: lines: $totalTo\n";
print "$lackingTo emails had no To: line\n";
print "Highest receivers (%age To:) (%age Total):\n\n";

foreach my $addr (sort bypercTO keys %TO) {
  #$totalto += $TO{$addr};
  $percTot = ($TO{$addr} / $emails) * 100;
  $percTot = sprintf("%.2f", $percTot);
  $percTo  = ($TO{$addr} / $totalTo) * 100; 
  $percTo = sprintf("%.2f", $percTo);
  if (!$showall) {
    print "$addr: $TO{$addr} ($percTo\%) ($percTot\%)\n" if ($percTot > $threshold or $percTo > $threshold);
  }
  else {
    print "$addr: $TO{$addr} ($percTo\%) ($percTot\%)\n";
  }
}

print "\nTotal number of From: lines: $totalFrom\n";
print "$lackingFrom emails had no From: line\n";
print "Highest senders:\n\n";
foreach my $addr (sort bypercFROM keys %FROM) {
  $percTot = ($FROM{$addr} / $emails) * 100;
  $percTot = sprintf("%.2f", $percTot);
  $percFrom  = ($FROM{$addr} / $totalFrom) * 100; 
  $percFrom = sprintf("%.2f", $percFrom);
  if (!$showall) {
    print "$addr: $FROM{$addr} ($percFrom\%) ($percTot\%)\n" if ($percTot > $threshold or $percFrom > $threshold);
  }
  else {
    print "$addr: $FROM{$addr} ($percFrom\%) ($percTot\%)\n";
  }
}

sub bypercTO { $TO{$b} <=> $TO{$a} }
sub bypercFROM { $FROM{$b} <=> $FROM{$a} }

sub usage {
    print "Usage: $0 [-a] [-t <limit>] [-h] MBOX\n";
    print "  -a - show all addresses (cancels out -t)\n";
    print "  -t - threshold (\%age) of items to show\n";
    print "  -h - this help\n";

    exit 0;
}