#! /usr/local/bin/perl -w 

#assumes input files are sorted

use strict;

my %lats;
my %IPs;

my $ignore = "ignore";
open IGNORE, $ignore;

my %ignore;
while (<IGNORE>) {
    chomp;
    $ignore{$_} = "bad!";
}

my $lines = 0;

$| = 1; # perform flush after each write to STDOUT

my $cur_key = "init";
my @cur_lats;

while (<>) {
  next:
    if (/^([0-9\.]*) --> ([0-9\.]+) ([0-9]+) ([0-9]+) ([0-9]+)/) {
	#ignore initial measurements. These have caching problems
	goto next if ($3 == 1);
	
	my $lat = ($4 - $5)/1000;
	goto next if ($lat < 0); 
        # latency can be less than zero if 
	# the first query has to be retried (e.g.)
	my $k = "$1-$2";

#	print "$k $cur_key: $_";
	if ($cur_key eq $k) {
	    #processing a stream of measurements
	    push (@cur_lats, $lat);
	} else {
	    # we're starting a new pair
	    # write down the value from the last pair
	    my @tmp = sort {$a <=> $b} @cur_lats;
	    my @alat_f;
	    foreach my $m (@tmp) {
		push (@alat_f, $m) if ($m < 1000);
	    }
	    if (defined $lats{$cur_key}) {
		# this must be a bug or the input wasn't sorted
		die "$k already processed?\n";
	    } elsif ($cur_key eq "init") {
#		print "ignoring initial key\n";
	    } else {
		if ($#alat_f >= 0) {
		    my $key = $#alat_f/4 if ($#alat_f >= 4);
		    $key = $#alat_f/2 if ($#alat_f < 4);
		    $lats{$cur_key} = $alat_f[$key];
#		    $lats{$cur_key} = avg(\@alat_f);
		} else {
		    $lats{$cur_key} = -1;
		}
#		print "$cur_key | $lats{$cur_key} | @tmp\n";
	    }

	    #don't ignore the new measurement
	    @cur_lats = ($lat);
	}
	$cur_key = $k;

	#note that we've seen these IPs
	$IPs{$1} = "p";
	$IPs{$2} = "p";

	#progress meter
	$lines++;
	print STDERR "." if ($lines % 10000 == 0);
	print STDERR "($lines)" if ($lines % 1000000 == 0);
    }
}

my $size = scalar keys %IPs;
print "outputting a $size  x $size matrix\n";

foreach my $from (sort keys %IPs) {
    next if (defined $ignore{$from});
    foreach my $to (sort keys %IPs) {
	next if (defined $ignore{$to});
	if ($from eq $to) {
	    print "0.0 ";
	} else {
	    my $lat = -1;
	    my $key = "$from-$to";

	    if (defined $lats{$key} && $lats{$key} > 1 && $lats{$key} < 800) {
		$lat = $lats{$key};
	    }
	    print "$lat ";
	}
    }
    print "\n";
}

print "row to IP mappings:\n";

my $i = 0;
foreach my $ip (sort keys %IPs) {
    next if (defined $ignore{$ip});
    print "$i: $ip\n";
    $i++;
}

sub avg () {
    my $aref = $_[0];
    my $sum = 0;
    foreach $i (@$aref) {
	$sum += $i;
    }
    return $sum/(@$aref + 1);
}