#!/usr/bin/env perl

use strict;
use warnings;
use FileHandle;
use Getopt::Long;
use File::Basename;

my $A_IN      = "";
my $A_OUT     = "";
my $A_MIN     = 0;
my $A_HELP    = 0;

our $PRG_NAME = basename($0);

my $result = &GetOptions("i=s"    => \$A_IN,
                         "o=s"    => \$A_OUT,
                         "min=s"  => \$A_MIN,
                         "h"      => \$A_HELP
                        );

if (!$result || $A_HELP) {
    usage();
}

open(IN, $A_IN) or usage("[Error] Cannot open $A_IN in [r] mode.");
open(OUT, ">$A_OUT") or usage("[Error] Cannot open $A_OUT in [w] mode.");


my ($first, $count) = (1, 0);
my ($seq, $prevname, $name) = ("", "", "");

while (my $line = <IN>) {
    chomp $line; 
    if ($line =~ m/>/) {
	$prevname = $name;
	$name = $line;
	if(!$first){
	    $seq =~ s/N*N/N/g;
	    $count = 0;
	    my @splitline = split(/N/,$seq);
	    foreach (@splitline) {
		$count++;
		if (length($_) >= $A_MIN){
		    print OUT $prevname."_part".$count."\n";
		    print OUT $_."\n";
		}
	    }
	}
	$first = 0;
	$seq = "";
    }
    else{
	$seq = $seq.$line;
    }
}

$seq =~ s/N*N/N/g;
$count = 0;
my @splitline = split(/N/,$seq);
foreach (@splitline) {
    $count++;
    if (length($_) >= $A_MIN){
	print OUT $prevname."_part".$count."\n";
	print OUT $_."\n";
    }
}

sub usage {
    my $usage = "
$PRG_NAME - split input sequence at each stretch of unknown bases (N).
    -i <file>   : input file
    -o <file>   : output file
    -min <int>  : minimum contig size, default 0
    -h          : this help
\n";
    warn $usage;
    exit 1;
}