#!/usr/bin/perl -w
#
# Author:  Grant Thiltgen
#
# This script is to collect output data from the multiple neural nets
# run to test our training protocol.

# CHANGE LOG:
# Thu Nov 18 09:20:34 PST 2010 Kevin Karplus
#	Added --net_list option to get list of nets, instead of
#	generating numerically.
# Thu Nov 18 09:30:00 PST 2010 Kevin Karplus
#	Added --best_n option (to replace hard-wired 10)

use strict;
use English;
use File::Basename;
use Getopt::Long;
use Pod::Usage;

my $num_nets;   # The number of nets that have data.
my $txxalign;   # The alignment, for example t05-thin90
my $dataset;    # The dataset, for example dunbrack-30pc-1763
my $arch;       # The architecture of the neural net.
my $alphaname;  # The name of the alphabet used in neural net training
my $ctrain=1;   # Value for declaring if crosstraining data is available
my $data;       # Data used, for example tr12 or t1c2
my $net_list= "../training-data/fullset.nets";	# where are the net names listed
my $best_n=10;	# number of results to print

GetOptions(
	      "number=i" => \$num_nets
	    , "txxalign=s" => \$txxalign
	    , "dataset=s" => \$dataset
	    , "arch=s" => \$arch
            , "alphaname=s" => \$alphaname
	    , "data=s" => \$data
	    , "net_list=s" => \$net_list
	    , "ctrain=i" => \$ctrain
	    , "best_n=i" => \$best_n
	    , "help|?"	=>	sub {pod2usage("verbose"=>1);}
	    , "man"		=> 	sub {pod2usage("verbose"=>2);}
	    )
	   or pod2usage("verbose" => 0);

# Building loops to create the data file.

my $line;       # Information from the files brought into this.
my $value;      # Value of hash for foreach loop.
my %trainbits;  # storing training data in bits
my @temp;       # temporary array for storing data in the line.
my $key;        # value for key in hash.
my @nets;       # Value for nets

my $sum=0;  # Value to calculate average of nets
my $count=0;

open(NETSET, $net_list) or die("Can't open list of nets in $net_list\n");

print STDOUT "Net\tValue\n";
print STDOUT "6S\t9N\n";
while ( my $name = <NETSET>)
{   chomp $name;
    if ((length($name)==0) || ($name =~ /#/)) { next; }
    {
	my $file = "$name/quality-reports/$dataset-$txxalign-$arch-$data-$alphaname-mult50-from-empty.train";
	open (TRAIN, $file) or print STDERR "Can't open $data training file $file for $name\n";
	while ($line = <TRAIN>)
	{
	    chomp $line;
	    if ($line =~ /^#/) {next;}
	    @temp = split /\s+/, $line;
	    $trainbits{$name} = $temp[3];
	}	    
	close TRAIN;
	$sum += $trainbits{$name};
	$count += 1;
    }
}
close NETSET;


my $average = $sum / $count;
print STDOUT "avg\t$average\n";

my $i=0;        # temp value for sorting.
open (NETS, ">networks/$dataset-$txxalign-$arch-$data-$alphaname-mult50.nets");
foreach $key (sort highlow keys %trainbits)
{
    if ($i++ < $best_n)
    {
	print NETS "$key\n";
    }
    print STDOUT "$key\t$trainbits{$key}\n";
}
close (NETS);

sub highlow ($$)
{
    my ($a,$b) = @_;
    $trainbits{$b} <=> $trainbits{$a};
}



__END__