irc-rel-plot.pl

#!/usr/bin/env perl
use strict;
use warnings;

use Data::Dumper;
use constant THRESHOLD => 5;
my $count;

while ($_ = <STDIN>)
{
    next unless (my ($ts, $user, $msg) = m/(\d{2}:\d{2}) <?([^>]*)> (.*)/);
   
    if ($msg =~ m/^([^ ]*): .*/)
    {  
        my $to = $1;
        $count->{$user}->{msg}->{$to}->{count}++;

        $count->{$user}->{msg}->{$to}->{up}++
            if ($msg =~ /:-?\)/);

        $count->{$user}->{msg}->{$to}->{down}++
            if ($msg =~ /:-?\(/);
    }
    $count->{$user}->{msgCount}++;
}

for my $user (keys %$count)
{
    for my $to (keys %{ $count->{$user}->{msg} })
    {  
        next unless
            $count->{$user}->{msg}->{$to}->{count} > THRESHOLD;

        my $data = $count->{$user}->{msg}->{$to};
        $data->{up} ||= 0;
        $data->{down} ||= 0;

        my $data_rel_pa;
        my $data_rel_pp;
        for (keys %$data)
        {  
            # how many % of all messages were send to a person (w/ smiley)
            $data_rel_pa->{$_} = sprintf '%3.4f', $data->{$_}* 100 / $count->{$user}->{msgCount};
           
            # how many % of messages to that person are send w/ smiley
            $data_rel_pp->{$_} = sprintf '%3.4f', $data->{$_}* 100 / $count->{$user}->{msg}->{$to}->{count};
        }
        print join '; ',
            $user, $to,
            $data_rel_pp->{up}, $data_rel_pp->{down},
            $data_rel_pa->{up}, $data_rel_pa->{down}, $data_rel_pa->{count},
            $data->{count}, $data->{up}, $data->{down};
        print "\n";
   }
}