#!/usr/bin/perl -w

use strict;

my $basedir = "$ENV{HOME}/mail";
my (%unfiltered_spam, %msent, %mread, %mspam, %monthhash, %monthnumhash);

$monthhash{"jan"} = "01";
$monthhash{"feb"} = "02";
$monthhash{"mar"} = "03";
$monthhash{"apr"} = "04";
$monthhash{"may"} = "05";
$monthhash{"jun"} = "06";
$monthhash{"jul"} = "07";
$monthhash{"aug"} = "08";
$monthhash{"sep"} = "09";
$monthhash{"oct"} = "10";
$monthhash{"nov"} = "11";
$monthhash{"dec"} = "12";

$monthnumhash{"01"} = "jan";
$monthnumhash{"02"} = "feb";
$monthnumhash{"03"} = "mar";
$monthnumhash{"04"} = "apr";
$monthnumhash{"05"} = "may";
$monthnumhash{"06"} = "jun";
$monthnumhash{"07"} = "jul";
$monthnumhash{"08"} = "aug";
$monthnumhash{"09"} = "sep";
$monthnumhash{"10"} = "oct";
$monthnumhash{"11"} = "nov";
$monthnumhash{"12"} = "dec";

###################################################################

chdir("$ENV{HOME}/www/mailstats") || die "Could not change directory";
init_hardcoded_data();
scandir($basedir);
dumpdata();
gengraph();
printHTML();
print "...Done\nMake sure to update the graph scale to include this month\n";
###################################################################



sub scandir {
  my ($basedir) = @_;
  my (@dentries, $entry, $num, $val);
  my (@dirs, $mon, $year, $type, $dir);

  @dirs = ("trash", "sent", "spam");

  foreach $type (@dirs){
      $dir = "$basedir/$type";
#      print "Processing $dir...\n";
      defined($dir) || die "Huh? Null dir request";

      opendir(DHANDLE, $dir) || die "Could not open $dir";
      @dentries = readdir(DHANDLE);
      closedir(DHANDLE);
      

      foreach $entry (@dentries) {
		  if ($entry =~ /^mail-[0-9][0-9][0-9][0-9]-[0-9][0-9]$/) {
			  ($year, $mon) = ($entry =~ /^mail-(....)-(..)$/);
              $mon = $monthnumhash{$mon};

			  $val = "$year-$mon";
#			  print $val . "\n";
			  if($type eq "trash"){
				  if(!$mread{$val}){
                      print "$dir/$entry\n";
					  $num = `grep "^From\ " $dir/$entry | wc -l` - 1;
					  $mread{$val} = $num;
				  }
			  }
			  elsif($type eq "sent"){
				  if(!$msent{$val}){
                      print "$dir/$entry\n";
					  $num = `grep "^From\ " $dir/$entry | wc -l` - 1;
					  # get heather's directory too
					  # HACK - do it here, because of the lost mail
					  # and hard-coded data.
					  #$num += `grep "^From\ " $basedir/heather/sent/$entry | wc -l` - 1;
					  $msent{$val} += $num;
				  }
			  }
			  elsif($type eq "spam"){
				  if(!$mspam{$val}){
                      print "$dir/$entry\n";
					  $num = `grep "^From\ " $dir/$entry | wc -l` - 1;
					  $mspam{$val} = $num;
				  }
			  }
			  else{
                  print "$dir/$entry\n";
				  die "Invalid type: $type\n";
			  }
		  } elsif ($entry =~ /^(\.|\.\.|mail|.imap|[0-9]{4})$/) {
			  # nothing
		  } else {
			  print STDERR "Unrecognized entry $entry\n";
		  }
      }
  }
  return;
}

sub msort {
    my ($amonth, $ayear, $bmonth, $byear);
    ($ayear, $amonth) = ($a =~ /^(....)-(...)$/);
    ($byear, $bmonth) = ($b =~ /^(....)-(...)$/);
    if($ayear == $byear) {
        $monthhash{$amonth} <=> $monthhash{$bmonth};
    } else {
        $ayear <=> $byear;
    }
}

sub dumpdata {
    dumphash("read", \%mread, undef);
    dumphash("sent", \%msent, undef);
    dumphash("spam", \%mspam, undef);
    dumphash("unfiltered_spam", \%unfiltered_spam, undef);
}

sub dumphash {
    my ($mname, $phash, $smonth) = @_;
    my ($key, $month, $year);
    
    open(MAIL, "> $mname-mail.dat");
    foreach $key (sort msort keys(%$phash)) {
        ($year, $month) = ($key =~ /^(....)-(...)$/);
        if(!defined($smonth) || ($smonth eq $month)) {
            print(MAIL "$key $$phash{$key}\n");
        }
    }
    close(MAIL);
    
    return;
}


sub gengraph {
    my $graph;
    my @graphs = ("read-mail", "sent-mail", "spam-mail", "unfiltered_spam-mail");

    foreach $graph (@graphs){
        open(GNUPLOT, "| gnuplot") || die "Could not spawn gnuplot";
        print GNUPLOT <<DONE;
        set key on left top
        set grid
        set xdata time
        set timefmt "%Y-%b"
        set xtics border autofreq
        set xrange ["2003-May":"2012-Jan"]
        set format x "%b%y"
  
        set output "${graph}.png"
        set term png xCCCCCC size 640,300
#        set size 0.7,0.3  # old method
        plot "${graph}.dat" using 1:2 title "" with lines
DONE
        close(GNUPLOT);
    }

    return;
}


sub commaizeNumber { 
    my $number = shift @_; 
    $number =~ s/^(-?\d+)(\d\d\d)/$1,$2/; 
    $number;
}

sub printHTML{
    my ($tot_sent, $tot_rec, $tot_spam, $tot_unfil);
    my ($avg_sent, $avg_rec, $avg_spam, $avg_unfil);
    my ($percent, $str);

    $tot_sent = gettotal(\%msent);
    $tot_rec = gettotal(\%mread);
    $tot_spam = gettotal(\%mspam) - gettotal(\%unfiltered_spam);
    $tot_unfil = gettotal(\%unfiltered_spam);

    $avg_sent = getavg(\%msent);
    $avg_rec = getavg(\%mread);
    $avg_spam = getavg(\%mspam) - getavg(\%unfiltered_spam);
    $avg_unfil = getavg(\%unfiltered_spam);

    $percent = 100-(100*$tot_unfil/($tot_rec+$tot_spam));

    # add commas for nicer printing
    $tot_sent = commaizeNumber($tot_sent);
    $tot_rec = commaizeNumber($tot_rec);
    $tot_spam = commaizeNumber($tot_spam);
    $tot_unfil = commaizeNumber($tot_unfil);
    $avg_sent = commaizeNumber($avg_sent);
    $avg_rec = commaizeNumber($avg_rec);
    $avg_spam = commaizeNumber($avg_spam);
    $avg_unfil = commaizeNumber($avg_unfil);

    $str = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\">\n";
    $str .= "<html>\n";
    $str .= " <head>\n";
    $str .= "  <title>The Daley's Email Statistics</title>\n";
    $str .= "  <link href='/.style.css' rel='stylesheet' type='text/css'>\n";
    $str .= " </head>\n";
    $str .= " <body>\n";
    $str .= "  <h1 align=center>Email Stats</h1>\n";
    $str .= "  <p align='center'>Thanks to <a href='http://www.spamassassin.org/'>SpamAssassin</a>,\n";
    $str .= "   <a href='http://razor.sourceforge.net/'>Razor</a>,\n";
    $str .= "   <a href='http://isg.ee.ethz.ch/tools/postgrey/'>Postgrey</a>\n";
    $str .= "   and <a href='http://www.spamcop.net/'>SpamCop</a><br/>\n";
    $str .= "   (<i>Postgrey installed August 2005, and stops ~2000 spams monthly</i>)<br/>\n";
    $str .= "   Spam Filter Accuracy ";
    $str = sprintf("%s%.1f%%", $str, $percent);
    $str .= "  </p>\n";
    $str .= "  <div class='center'>\n";
    $str .= "  <table><tr>\n";
    $str .= "    <td>Received (non-spam): $tot_rec total, $avg_rec average<br/>\n";
    $str .= "      <img width='100%' src='read-mail.png'></td>\n";
    $str .= "    <td>Sent Mail: $tot_sent total, $avg_sent monthly average<br/>\n";
    $str .= "      <img width='100%' src='sent-mail.png'></td>\n";
    $str .= "   </tr><tr>\n";
    $str .= "    <td>Filtered Spam: $tot_spam total, $avg_spam average<br/>\n";
    $str .= "      <img width='100%' src='spam-mail.png'></td>\n";
    $str .= "    <td>Unfiltered Spam: $tot_unfil total, $avg_unfil average<br/>\n";
    $str .= "      <img width='100%' src='unfiltered_spam-mail.png'>\n";
    $str .= "    </td></tr></table>\n";
    $str .= "   </div>\n";
    $str .= " <p align=center>\n";
    $str .= "  <a href='mailgraph'>Graph utility</a> heavily modified from a script\n";
    $str .= "   by <a href='http://www.snurgle.org/~griffon/'>Chris Chiappa</a>\n";
    $str .= " </p>\n";
    $str .= "</body></html>\n";
    
    open (OUTHTML, "> index.html");
    print OUTHTML $str;
    close (OUTHTML);
}


sub gettotal{
    my ($phash) = @_;
    my ($key, $total);
    $total = 0;
    foreach $key (keys(%$phash)){
        $total += $$phash{$key};
    }
    return $total;
}

sub getavg{
    my ($phash) = @_;
    my ($key, $average, $total, $count);
    $count = 0;
    $total = 0;
    
    foreach $key (keys(%$phash)){
        $count++;
        $total += $$phash{$key};
    }
    if($count){
        $average = int($total / $count);
    }
    else{
        $average = 0;
    }
    return $average;
}


# hard code these numbers, in order to save time
# when running the script.  Also, doing this means
# I can erase old listserv emails that I don't care
# about without messing up the stats.
sub init_hardcoded_data{
  $msent{"2003-may"} = 332;
  $msent{"2003-jun"} = 522;
  $msent{"2003-jul"} = 625;
  $msent{"2003-aug"} = 365;
  $msent{"2003-sep"} = 553;
  $msent{"2003-oct"} = 439;
  $msent{"2003-nov"} = 293;
  $msent{"2003-dec"} = 346;
  $msent{"2004-jan"} = 386;
  $msent{"2004-feb"} = 406;
  $msent{"2004-mar"} = 662;
  $msent{"2004-apr"} = 709;
  $msent{"2004-may"} = 627;
  $msent{"2004-jun"} = 429;
  $msent{"2004-jul"} = 334;
  $msent{"2004-aug"} = 467;
  $msent{"2004-sep"} = 680;
  $msent{"2004-oct"} = 533;
  $msent{"2004-nov"} = 493;
  $msent{"2004-dec"} = 372;
  $msent{"2005-jan"} = 553;
  $msent{"2005-feb"} = 653;
  $msent{"2005-mar"} = 717;
  $msent{"2005-apr"} = 698;
  $msent{"2005-may"} = 685;
  $msent{"2005-jun"} = 545;
  $msent{"2005-jul"} = 294;
  $msent{"2005-aug"} = 405;
  $msent{"2005-sep"} = 538;
  $msent{"2005-oct"} = 375;
  $msent{"2005-nov"} = 417;
  $msent{"2005-dec"} = 386;
  $msent{"2006-jan"} = 679;
  $msent{"2006-feb"} = 549;
  $msent{"2006-mar"} = 544;
  $msent{"2006-apr"} = 626;
  $msent{"2006-may"} = 888;
  $msent{"2006-jun"} = 677;
  $msent{"2006-jul"} = 700;
  $msent{"2006-aug"} = 694;
  $msent{"2006-sep"} = 653;
  $msent{"2006-oct"} = 470;
  $msent{"2006-nov"} = 565;
  $msent{"2006-dec"} = 483;
  $msent{"2007-jan"} = 760;
  $msent{"2007-feb"} = 522;
  $msent{"2007-mar"} = 717;
  $msent{"2007-apr"} = 541;
  $msent{"2007-may"} = 666;
  $msent{"2007-jun"} = 594;
  $msent{"2007-jul"} = 548;
  $msent{"2007-aug"} = 628;
  $msent{"2007-sep"} = 682;
  $msent{"2007-oct"} = 812;
  $msent{"2007-nov"} = 517;
  $msent{"2007-dec"} = 726;
  $msent{"2008-jan"} = 830;
  $msent{"2008-feb"} = 490;
  $msent{"2008-mar"} = 522;
  $msent{"2008-apr"} = 670;
  $msent{"2008-may"} = 770;
  $msent{"2008-jun"} = 577;
  $msent{"2008-jul"} = 459;
  $msent{"2008-aug"} = 360;
  $msent{"2008-sep"} = 373;
  $msent{"2008-oct"} = 598;
  $msent{"2008-nov"} = 415;
  $msent{"2008-dec"} = 634;
  $msent{"2009-jan"} = 407;
  $msent{"2009-feb"} = 549;
  $msent{"2009-mar"} = 529;
  $msent{"2009-apr"} = 477;
  $msent{"2009-may"} = 544;
  $msent{"2009-jun"} = 543;
  $msent{"2009-jul"} = 493;
  $msent{"2009-aug"} = 487;
  $msent{"2009-sep"} = 420;
  $msent{"2009-oct"} = 547;
  $msent{"2009-nov"} = 398;
  $msent{"2009-dec"} = 376;
  $msent{"2010-jan"} = 490;
  $msent{"2010-feb"} = 567;
  $msent{"2010-mar"} = 399;
  $msent{"2010-apr"} = 383;
  $msent{"2010-may"} = 476;
  $msent{"2010-jun"} = 593;
  $msent{"2010-jul"} = 410;
  $msent{"2010-aug"} = 308;
  $msent{"2010-sep"} = 507;
  $msent{"2010-oct"} = 563;
  $msent{"2010-nov"} = 407;
  $msent{"2010-dec"} = 372;
  $msent{"2011-jan"} = 467;
  $msent{"2011-feb"} = 453;
  $msent{"2011-mar"} = 506;
  $msent{"2011-apr"} = 470;
  $msent{"2011-may"} = 411;
  $msent{"2011-jun"} = 337;
  $msent{"2011-jul"} = 342;
  $msent{"2011-aug"} = 400;
  $msent{"2011-sep"} = 496;
  $msent{"2011-oct"} = 400;
  $msent{"2011-nov"} = 354;
  $msent{"2011-dec"} = 480;
  $msent{"2012-jan"} = 492;

  $mread{"2003-jun"} = 954;
  $mread{"2003-jul"} = 1561;
  $mread{"2003-aug"} = 835;
  $mread{"2003-sep"} = 1264;
  $mread{"2003-oct"} = 1143;
  $mread{"2003-nov"} = 1121;
  $mread{"2003-dec"} = 1214;
  $mread{"2004-jan"} = 1218;
  $mread{"2004-feb"} = 1197;
  $mread{"2004-mar"} = 1332;
  $mread{"2004-apr"} = 2198;
  $mread{"2004-may"} = 1297;
  $mread{"2004-jun"} = 1171;
  $mread{"2004-jul"} = 1079;
  $mread{"2004-aug"} = 1280;
  $mread{"2004-sep"} = 1981;
  $mread{"2004-oct"} = 1837;
  $mread{"2004-nov"} = 1759;
  $mread{"2004-dec"} = 1401;
  $mread{"2005-jan"} = 2126;
  $mread{"2005-feb"} = 2775;
  $mread{"2005-mar"} = 3753;
  $mread{"2005-apr"} = 4482;
  $mread{"2005-may"} = 4210;
  $mread{"2005-jun"} = 3591;
  $mread{"2005-jul"} = 2998;
  $mread{"2005-aug"} = 2826;
  $mread{"2005-sep"} = 2767;
  $mread{"2005-oct"} = 2464;
  $mread{"2005-nov"} = 2340;
  $mread{"2005-dec"} = 2621;
  $mread{"2006-jan"} = 3793;
  $mread{"2006-feb"} = 3165;
  $mread{"2006-mar"} = 3731;
  $mread{"2006-apr"} = 4000;
  $mread{"2006-may"} = 4508;
  $mread{"2006-jun"} = 4140;
  $mread{"2006-jul"} = 4703;
  $mread{"2006-aug"} = 4644;
  $mread{"2006-sep"} = 4911;
  $mread{"2006-oct"} = 4816;
  $mread{"2006-nov"} = 5669;
  $mread{"2006-dec"} = 3623;
  $mread{"2007-jan"} = 4856;
  $mread{"2007-feb"} = 5199;
  $mread{"2007-mar"} = 5306;
  $mread{"2007-apr"} = 4651;
  $mread{"2007-may"} = 5141;
  $mread{"2007-jun"} = 4962;
  $mread{"2007-jul"} = 4742;
  $mread{"2007-aug"} = 5010;
  $mread{"2007-sep"} = 5155;
  $mread{"2007-oct"} = 4872;
  $mread{"2007-nov"} = 3755;
  $mread{"2007-dec"} = 4304;
  $mread{"2008-jan"} = 4724;
  $mread{"2008-feb"} = 3914;
  $mread{"2008-mar"} = 4004;
  $mread{"2008-apr"} = 4182;
  $mread{"2008-may"} = 5401;
  $mread{"2008-jun"} = 4479;
  $mread{"2008-jul"} = 4073;
  $mread{"2008-aug"} = 3576;
  $mread{"2008-sep"} = 3287;
  $mread{"2008-oct"} = 3844;
  $mread{"2008-nov"} = 3534;
  $mread{"2008-dec"} = 3779;
  $mread{"2009-jan"} = 2932;
  $mread{"2009-feb"} = 3731;
  $mread{"2009-mar"} = 3816;
  $mread{"2009-apr"} = 3844;
  $mread{"2009-may"} = 4703;
  $mread{"2009-jun"} = 5638;
  $mread{"2009-jul"} = 3425;
  $mread{"2009-aug"} = 3619;
  $mread{"2009-sep"} = 3710;
  $mread{"2009-oct"} = 4556;
  $mread{"2009-nov"} = 3330;
  $mread{"2009-dec"} = 4170;
  $mread{"2010-jan"} = 3889;
  $mread{"2010-feb"} = 3675;
  $mread{"2010-mar"} = 3160;
  $mread{"2010-apr"} = 2832;
  $mread{"2010-may"} = 2966;
  $mread{"2010-jun"} = 3305;
  $mread{"2010-jul"} = 2998;
  $mread{"2010-aug"} = 2885;
  $mread{"2010-sep"} = 3384;
  $mread{"2010-oct"} = 3142;
  $mread{"2010-nov"} = 2544;
  $mread{"2010-dec"} = 2229;
  $mread{"2011-jan"} = 2397;
  $mread{"2011-feb"} = 2681;
  $mread{"2011-mar"} = 2580;
  $mread{"2011-apr"} = 2844;
  $mread{"2011-may"} = 4345;
  $mread{"2011-jun"} = 3170;
  $mread{"2011-jul"} = 2765;
  $mread{"2011-aug"} = 3066;
  $mread{"2011-sep"} = 3112;
  $mread{"2011-oct"} = 3205;
  $mread{"2011-nov"} = 3224;
  $mread{"2011-dec"} = 3648;
  $mread{"2012-jan"} = 4156;

  $mspam{"2003-aug"} = 41;
  $mspam{"2003-sep"} = 99;
  $mspam{"2003-oct"} = 73;
  $mspam{"2003-nov"} = 61;
  $mspam{"2003-dec"} = 57;
  $mspam{"2004-jan"} = 116;
  $mspam{"2004-feb"} = 261;
  $mspam{"2004-mar"} = 389;
  $mspam{"2004-apr"} = 868;
  $mspam{"2004-may"} = 768;
  $mspam{"2004-jun"} = 644;
  $mspam{"2004-jul"} = 511;
  $mspam{"2004-aug"} = 606;
  $mspam{"2004-sep"} = 754;
  $mspam{"2004-oct"} = 867;
  $mspam{"2004-nov"} = 951;
  $mspam{"2004-dec"} = 858;
  $mspam{"2005-jan"} = 773;
  $mspam{"2005-feb"} = 1372;
  $mspam{"2005-mar"} = 1212;
  $mspam{"2005-apr"} = 1296;
  $mspam{"2005-may"} = 1547;
  $mspam{"2005-jun"} = 1433;
  $mspam{"2005-jul"} = 1732;
  $mspam{"2005-aug"} = 64;
  $mspam{"2005-sep"} = 58;
  $mspam{"2005-oct"} = 61;
  $mspam{"2005-nov"} = 62;
  $mspam{"2005-dec"} = 67;
  $mspam{"2006-jan"} = 75;
  $mspam{"2006-feb"} = 79;
  $mspam{"2006-mar"} = 101;
  $mspam{"2006-apr"} = 97;
  $mspam{"2006-may"} = 190;
  $mspam{"2006-jun"} = 204;
  $mspam{"2006-jul"} = 313;
  $mspam{"2006-aug"} = 303;
  $mspam{"2006-sep"} = 250;
  $mspam{"2006-oct"} = 420;
  $mspam{"2006-nov"} = 701;
  $mspam{"2006-dec"} = 816;
  $mspam{"2007-jan"} = 394;
  $mspam{"2007-feb"} = 307;
  $mspam{"2007-mar"} = 308;
  $mspam{"2007-apr"} = 223;
  $mspam{"2007-may"} = 165;
  $mspam{"2007-jun"} = 311;
  $mspam{"2007-jul"} = 378;
  $mspam{"2007-aug"} = 374;
  $mspam{"2007-sep"} = 279;
  $mspam{"2007-oct"} = 665;
  $mspam{"2007-nov"} = 384;
  $mspam{"2007-dec"} = 212;
  $mspam{"2008-jan"} = 180;
  $mspam{"2008-feb"} = 269;
  $mspam{"2008-mar"} = 228;
  $mspam{"2008-apr"} = 185;
  $mspam{"2008-may"} = 160;
  $mspam{"2008-jun"} = 276;
  $mspam{"2008-jul"} = 281;
  $mspam{"2008-aug"} = 196;
  $mspam{"2008-sep"} = 242;
  $mspam{"2008-oct"} = 125;
  $mspam{"2008-nov"} = 156;
  $mspam{"2008-dec"} = 193;
  $mspam{"2009-jan"} = 156;
  $mspam{"2009-feb"} = 133;
  $mspam{"2009-mar"} = 214;
  $mspam{"2009-apr"} = 324;
  $mspam{"2009-may"} = 282;
  $mspam{"2009-jun"} = 241;
  $mspam{"2009-jul"} = 264;
  $mspam{"2009-aug"} = 226;
  $mspam{"2009-sep"} = 203;
  $mspam{"2009-oct"} = 203;
  $mspam{"2009-nov"} = 174;
  $mspam{"2009-dec"} = 140;
  $mspam{"2010-jan"} = 203;
  $mspam{"2010-feb"} = 185;
  $mspam{"2010-mar"} = 177;
  $mspam{"2010-apr"} = 175;
  $mspam{"2010-may"} = 199;
  $mspam{"2010-jun"} = 192;
  $mspam{"2010-jul"} = 276;
  $mspam{"2010-aug"} = 279;
  $mspam{"2010-sep"} = 164;
  $mspam{"2010-oct"} = 167;
  $mspam{"2010-nov"} = 133;
  $mspam{"2010-dec"} = 102;
  $mspam{"2011-jan"} = 91;
  $mspam{"2011-feb"} = 110;
  $mspam{"2011-mar"} = 158;
  $mspam{"2011-apr"} = 141;
  $mspam{"2011-may"} = 100;  # don't really know, deleted the file too quickly
  $mspam{"2011-jun"} = 81;
  $mspam{"2011-jul"} = 82;
  $mspam{"2011-aug"} = 61;
  $mspam{"2011-sep"} = 78;
  $mspam{"2011-oct"} = 87;
  $mspam{"2011-nov"} = 103;
  $mspam{"2011-dec"} = 114;
  $mspam{"2012-jan"} = 77;

  $unfiltered_spam{"2003-oct"} = 0;
  $unfiltered_spam{"2003-nov"} = 0;
  $unfiltered_spam{"2003-dec"} = 1;
  $unfiltered_spam{"2004-jan"} = 0;
  $unfiltered_spam{"2004-feb"} = 2;
  $unfiltered_spam{"2004-mar"} = 0;
  $unfiltered_spam{"2004-apr"} = 2;
  $unfiltered_spam{"2004-may"} = 1;
  $unfiltered_spam{"2004-jun"} = 0;
  $unfiltered_spam{"2004-jul"} = 1;
  $unfiltered_spam{"2004-aug"} = 3;
  $unfiltered_spam{"2004-sep"} = 0;
  $unfiltered_spam{"2004-oct"} = 1;
  $unfiltered_spam{"2004-nov"} = 0;
  $unfiltered_spam{"2004-dec"} = 8;
  $unfiltered_spam{"2005-jan"} = 10;
  $unfiltered_spam{"2005-feb"} = 12;
  $unfiltered_spam{"2005-mar"} = 11;
  $unfiltered_spam{"2005-apr"} = 12;
  $unfiltered_spam{"2005-may"} = 7;
  $unfiltered_spam{"2005-jun"} = 11;
  $unfiltered_spam{"2005-jul"} = 12;
  $unfiltered_spam{"2005-aug"} = 5;
  $unfiltered_spam{"2005-sep"} = 1;
  $unfiltered_spam{"2005-oct"} = 6;
  $unfiltered_spam{"2005-nov"} = 5;
  $unfiltered_spam{"2005-dec"} = 0;
  $unfiltered_spam{"2006-jan"} = 6;
  $unfiltered_spam{"2006-feb"} = 4;
  $unfiltered_spam{"2006-mar"} = 8;
  $unfiltered_spam{"2006-apr"} = 11;
  $unfiltered_spam{"2006-may"} = 4;
  $unfiltered_spam{"2006-jun"} = 7;
  $unfiltered_spam{"2006-jul"} = 10;
  $unfiltered_spam{"2006-aug"} = 9;
  $unfiltered_spam{"2006-sep"} = 22;
  $unfiltered_spam{"2006-oct"} = 71;
  $unfiltered_spam{"2006-nov"} = 10;
  $unfiltered_spam{"2006-dec"} = 7;
  $unfiltered_spam{"2007-jan"} = 15;
  $unfiltered_spam{"2007-feb"} = 40;
  $unfiltered_spam{"2007-mar"} = 34;
  $unfiltered_spam{"2007-apr"} = 26;
  $unfiltered_spam{"2007-may"} = 27;
  $unfiltered_spam{"2007-jun"} = 47;
  $unfiltered_spam{"2007-jul"} = 20;
  $unfiltered_spam{"2007-aug"} = 7;
  $unfiltered_spam{"2007-sep"} = 15;
  $unfiltered_spam{"2007-oct"} = 21;
  $unfiltered_spam{"2007-nov"} = 113;
  $unfiltered_spam{"2007-dec"} = 28;
  $unfiltered_spam{"2008-jan"} = 16;
  $unfiltered_spam{"2008-feb"} = 16;
  $unfiltered_spam{"2008-mar"} = 15;
  $unfiltered_spam{"2008-apr"} = 17;
  $unfiltered_spam{"2008-may"} = 28;
  $unfiltered_spam{"2008-jun"} = 14;
  $unfiltered_spam{"2008-jul"} = 27;
  $unfiltered_spam{"2008-aug"} = 16;
  $unfiltered_spam{"2008-sep"} = 15;
  $unfiltered_spam{"2008-oct"} = 11;
  $unfiltered_spam{"2008-nov"} = 20;
  $unfiltered_spam{"2008-dec"} = 31;
  $unfiltered_spam{"2009-jan"} = 26;
  $unfiltered_spam{"2009-feb"} = 25;
  $unfiltered_spam{"2009-mar"} = 46;
  $unfiltered_spam{"2009-apr"} = 85;
  $unfiltered_spam{"2009-may"} = 38;
  $unfiltered_spam{"2009-jun"} = 12;
  $unfiltered_spam{"2009-jul"} = 27;
  $unfiltered_spam{"2009-aug"} = 41;
  $unfiltered_spam{"2009-sep"} = 31;
  $unfiltered_spam{"2009-oct"} = 23;
  $unfiltered_spam{"2009-nov"} = 26;
  $unfiltered_spam{"2009-dec"} = 21;
  $unfiltered_spam{"2010-jan"} = 19;
  $unfiltered_spam{"2010-feb"} = 23;
  $unfiltered_spam{"2010-mar"} = 23;
  $unfiltered_spam{"2010-apr"} = 27;
  $unfiltered_spam{"2010-may"} = 16;
  $unfiltered_spam{"2010-jun"} = 13;
  $unfiltered_spam{"2010-jul"} = 17;
  $unfiltered_spam{"2010-aug"} = 13;
  $unfiltered_spam{"2010-sep"} = 13;
  $unfiltered_spam{"2010-oct"} = 21;
  $unfiltered_spam{"2010-nov"} = 16;
  $unfiltered_spam{"2010-dec"} = 5;
  $unfiltered_spam{"2011-jan"} = 7;
  $unfiltered_spam{"2011-feb"} = 15;
  $unfiltered_spam{"2011-mar"} = 15;
  $unfiltered_spam{"2011-apr"} = 5;
  $unfiltered_spam{"2011-apr"} = 20;
  $unfiltered_spam{"2011-may"} = 25;
  $unfiltered_spam{"2011-jun"} = 19;
  $unfiltered_spam{"2011-jul"} = 16;
  $unfiltered_spam{"2011-aug"} = 13;
  $unfiltered_spam{"2011-sep"} = 15;
  $unfiltered_spam{"2011-oct"} = 23;
  $unfiltered_spam{"2011-nov"} = 19;
  $unfiltered_spam{"2011-dec"} = 25;
  $unfiltered_spam{"2012-jan"} = 18;
  $unfiltered_spam{"2012-feb"} = 7;
}