#! /usr/bin/ksh

# flowsize - a script to do plots based on the distribution of cflowd flow sizes
# Dave Plonka <plonka@doit.wisc.edu>, Jul 24 2000
# set -x

exec flowdumper -cn \
\
-I "@timerange=qw($*);"'
    $base = 2;
    for ($n = 0; $base**$n <= 2**32; $n++) {
       $val = $base**$n;
       $exp{$val} = $n;
       $range{$val} = length($val) > 4? "$base^$n" : $val
    }
    @range = sort { $a <=> $b } keys %range' \
\
-e 'foreach (@range) {
       if ($Cflow::bytes < $_) {
          $lt{$_}++;
	  $bytes{$_} += $Cflow::bytes;
	  last
       }
    }
    foreach (@range) {
       if ($Cflow::pkts < $_) {
          $plt{$_}++;
	  $pkts{$_} += $Cflow::pkts;
	  last
       }
    }
    $flows++;
    $Bytes += $Cflow::bytes;
    $Pkts += $Cflow::pkts' \
\
-E 'if (2 == $base) {
      @xtics = split(", ", "\"1\" 0, \"2\" 1, \"4\" 2, \"8\" 3, \"16\" 4, \"32\" 5, \"64\" 6, \"128\" 7, \"256\" 8, \"512\" 9, \"1K\" 10, \"2K\" 11, \"4K\" 12, \"8K\" 13, \"16K\" 14, \"32K\" 15, \"64K\" 16, \"128K\" 17, \"256K\" 18, \"512K\" 19, \"1M\" 20, \"2M\" 21, \"4M\" 22, \"8M\" 23, \"16M\" 24, \"32M\" 25, \"64M\" 26, \"128M\" 27, \"256M\" 28, \"512M\" 29, \"1G\" 30, \"2G\" 31, \"4G\" 32")
    } else {
       foreach (sort { $a <=> $b } keys(%range)) {
          push(@xtics, "\"<$range{$_}\" $exp{$_}")
       }
    }
    $timerange[0] =~ m/\.([^.]*)$/;
    $timerange = "~$1";
    if (@timerange > 1) {
       $timerange[$#timerange] =~ m/\.([^.]*)$/;
       $timerange .= " -> ~$1"
    }
    $timerange =~ s/_/ /g;
    print "cat > flowsize.gnuplot <<_EOF_\n";
    print "set terminal postscript color solid \"Helvetica\" 10\n";
    print "set title \"Distribution of Internet Flow Sizes, $flows flows, $timerange\"\n";
    print "set size 1,0.75\n";
    print "set xlabel \"Flow Size\"\n";
    print "set ylabel \"Cumulative Percentage of Total Flows\"\n";

    print "set xtics mirror rotate (", join(", ", @xtics), ")\n";
    print "plot \"flowsize.dat\" using 1:2 title \"bytes\" with steps, \"flowsize.dat\" using 1:3 title \"pkts\" with steps\n";
    print "_EOF_\n";
    print "cat > flowsize2.gnuplot <<_EOF_\n";
    print "set terminal postscript color solid \"Helvetica\" 10\n";
    print "set title \"Distribution of Internet Flow Sizes, $flows flows, $timerange\"\n";
    print "set size 1,0.75\n";
    print "set xlabel \"Flow Size\"\n";
    print "set ylabel \"Percentage of Content (IP headers+payload)\"\n";

    print "set xtics mirror rotate (", join(", ", @xtics), ")\n";
    print "set grid\n";
    print "plot \"flowsize.dat\" using 1:4 title \"bytes\" with steps, \"flowsize.dat\" using 1:5 title \"pkts\" with steps\n";
    print "_EOF_\n";
    print "cat > flowsize.dat <<_EOF_\n";
    print "#\n";
    print "# power_of_$base\tflow_pct_by_bytes\tflow_pct_by_pkts\tpct_bytes\tpct_pkts\n";
    print "#\n";
    foreach (@range) {
       $pct += 100*((0+$lt{$_})/$flows);
       $ppct += 100*((0+$plt{$_})/$flows);
       $B = 100*((0+$bytes{$_})/$Bytes);
       $p = 100*((0+$pkts{$_})/$Pkts);
       print "$exp{$_}\t", $pct, "\t", $ppct, "\t", $B, "\t", $p, "\n";
    }
    print "_EOF_\n"' \
\
"$@"