#!/usr/bin/perl

use warnings;
use strict;
use Bio::SeqIO;
use Getopt::Long;
use constant USAGE =><<END;

SYNOPSIS:

 seqfilehead.pl [OPTIONS] <int> [seqfile]

DESCRIPTION:

Masks out Fasta entries from a sequence file based on GFF lines.

OPTIONS:

       --informat
              Specifies the input format. Defualt is Fasta.
       --outformat
              Specifies the output format. Default is informat.
       --skip
              Skips first <int> entries instead of printing the <int>
              first.
       --help
             Prints a help.

EXAMPLES:

 # Get the first 10 entries in an Fasta file
 cat file.fa | seqfilehead.pl 5 > newfile.fa

 # Get the first 5 entries in an EMBL file
 seqfilehead.pl --informat EMBL 5 file.embl > newfile.embl

AUTHOR:

Kasper Munch

COPYRIGHT:

This program is free software. You may copy and redistribute it under
the same terms as Perl itself.

END

my $help = 0;
my $informat = 'Fasta';
my $outformat = '';
my $skip = 0;

GetOptions("help" => \$help,
           "skip" => \$skip,
           "informat=s" => \$informat,
           "outformat=s" => \$outformat) or die USAGE;

$outformat ||= $informat;
@ARGV or die USAGE;
my $nr = shift @ARGV;
defined $nr or die "Specify the number of seqs you want.\n";
my $seqfile = shift @ARGV if @ARGV;

my $fh;
my $in;
if ($seqfile) {
  if ($informat eq 'Fasta') {
    open $fh, "$seqfile";
  } else {
    $in = Bio::SeqIO->newFh(-file => "$seqfile", '-format' => $informat) or die "$!\n";
  }
} else {
  if ($informat eq 'Fasta') {
    open $fh, ">&STDIN";
  } else {
    $in = Bio::SeqIO->newFh(-fh => \*STDIN, '-format' => $informat) or die "$!\n";
  }
}

my $i = 0;
if ($informat eq 'Fasta') {
  local $/ = ">";
  while(<$fh>) {
    chomp;

    next unless $_;
    unless (/\n$/) {
      # Seems we only got part of an entry, so we probably encountered a
      # '>' in the description. So let's add another line:
      my $l = $_;
      $l = "$l>" . <$fh>;
      $_ = $l;
      chomp;
    }

    next unless $_;

    if ($skip) {
      if ($nr > $i++) {
        next;
      } else {
        print ">$_";
      }
    } else {
      if ($nr > $i++) {
        print ">$_";
      } else {
        exit;
      }
    }
  }
} else {
  my $out = Bio::SeqIO->newFh(-fh => \*STDOUT, '-format' => $outformat) or die "$!\n";
  while (my $seq = <$in>) {
    if ($skip) {
      if ($nr > $i++) {
        next;
      } else {
        print $out $seq;
      }
    } else {
      if ($nr > $i++) {
        print $out $seq;
      } else {
        exit 1;
      }
    }
  }
}

=head1 SYNOPSIS:

 seqfilehead.pl [OPTIONS] <int> [seqfile]

=head1 DESCRIPTION:

Masks out Fasta entries from a sequence file based on GFF lines.

=head1 OPTIONS:

=over 4

=item --informat

Specifies the input format. Defualt is Fasta.

=item --outformat

Specifies the output format. Default is informat.

=item --skip

Skips first <int> entries instead of printing the <int>
first.

=item --help

Prints a help.

=back

=head1 EXAMPLES:

 # Get the first 10 entries in an Fasta file
 cat file.fa | seqfilehead.pl 5 > newfile.fa

 # Get the first 5 entries in an EMBL file
 seqfilehead.pl --informat EMBL 5 file.embl > newfile.embl

=head1 AUTHOR:

Kasper Munch

=head1 COPYRIGHT:

This program is free software. You may copy and redistribute it under
the same terms as Perl itself.


=cut