#!/usr/bin/perl  -w


use strict;
use warnings;
use Getopt::Long;

my $version = "1.2";
my $options = get_options($version);


my $fastqInput = $$options{'i'};
my $fastaOutput = $$options{'o'};
my $minLen = $$options{'m'};
my $maxLen = $$options{'l'};

$minLen //= 25;
$maxLen //= 35;

if ($fastqInput =~ /.gz$/) {
    open(CHART, "gzip -dc $fastqInput |") || die "cannot open $fastqInput\n";
} else {
    open(CHART, "<$fastqInput") || die "cannot open $fastqInput\n";
}

if ($fastaOutput =~ /\.gz$/) {
    open OUT, "|gzip >$fastaOutput" ||die $!;
} else {
    open OUT, ">$fastaOutput" ||die $!;
}


my %seqs;
while(<CHART>) {
   chomp;
   my $seqname = $_;
   my $seq=<CHART>;
   my $strand=<CHART>;
   my $qual=<CHART>;
   $seq=~s/\s+//g;
   $seq=~s/\n+//g;
   next if length($seq) < $minLen or length($seq) > $maxLen;
   if(not exists $seqs{$seq}) {
     $seqs{$seq} = 1;
   } else {
     $seqs{$seq}++;
   }
}
close CHART;

my $n=0;
for my $key (reverse sort {$seqs{$a}<=>$seqs{$b}} keys %seqs) {
    $n++;
    print OUT ">seq$n"."_x$seqs{$key}\n$key\n";
}
close OUT;



sub usage_message {
    my $version_num = shift;
    my $usage_message = "\nConvert fastq to collapsed fasta:
    Usage: perl $0 -i <fq> -o <fa> -m <minimum_length> -l <maximum_length>
    <fq> : fastq file without adaptor (.fq .fastq).
    <fa> : output fasta file (.fa .fasta).
    <minimum_length> : discard reads shorter than minimum length (integer, default: 25).
    <maximum_length> : discard reads longer than maximum length (integer, default: 35).
    \n";
    return $usage_message;
}

sub get_options {
    my($v_num) = shift;
    my %options = ();
    GetOptions(\%options,
	       'help',
	       'i=s',
	       'o=s',
	       'm=i',
	       'l=i'
    );
    unless(%options) {
	my $usage_message = usage_message($v_num);
	die "$usage_message\n";
    }
    return \%options;
}