use strict;
use warnings;

my $file = $ARGV[0];
open (DATAFILE, "<", $file) or die("error:$!");
$file =~ /^([\w\-\_]+)./;
my $file_name = $1;

my (@name, @data, $samples, $check, $population, @pop_name, $ind, @indiv_num);
for (<DATAFILE>){
	chomp($_);
	if ($_ =~ /^Pop/i){
		$population++;
		$_ =~ /([\w\-\_\.]+)\s*$/;
		push (@pop_name, $1);
		unless ($population == 1){
			push (@indiv_num, $ind);
			$ind = 0;
		}
	}elsif ($_ =~ /^>\w/){
		$_ =~ s/>//;
		push (@name, $_);
		$samples++;
		$ind++;
		$check = 1;
	}elsif ($_ =~ /\w/){
		if ($check == 1){
			push (@data, $_);
		}else{
			$data[-1] = $data[-1].$_;
		}
		$check = 0;
	}
}
push (@indiv_num, $ind);

my $length_check = length ($data[0]);
foreach (@data){
	my $length_check2 = length($_);
	unless ($length_check == $length_check2){
		print "Error!! Sequence length differ in some samples\n";
		exit;
	}
	$length_check = $length_check2;
}
open (OUTPUT, ">", "${file_name}.arp") or die("error:$!");

#Arpfileì¬
print OUTPUT <<"EOF";
[Profile]
Title="DNA sequence data"
NbSamples=$population
GenotypicData=0
DataType=DNA
LocusSeparator=NONE
MissingData='?'
[Data]
[[Samples]]
EOF
my $total_indiv = 0;
for (my $n = 0; $n < $population; $n++){
	unless ($pop_name[$n] =~ /Pop/i){
		print OUTPUT "SampleName=\"$pop_name[$n]\"\n";
	}else{
		print OUTPUT "SampleName=\"$name[$total_indiv]\"\n";
	}
	print OUTPUT "SampleSize=$indiv_num[$n]\n";
	print OUTPUT "SampleData= {\n";
	for (my $m = 0; $m < $indiv_num[$n]; $m++){
		print OUTPUT "$name[$m+$total_indiv]\t1\t$data[$m+$total_indiv]\n";
	}
	$total_indiv += $indiv_num[$n];
	print OUTPUT "}\n";
}
print <<"EOF";
The total number of populations = $population
The total number of individuals = $samples
${file_name}.arp was created
EOF
close (DATAFILE);
close (OUTPUT);