#! /usr/bin/env perl
use strict;
use warnings FATAL => 'all';
use Data::Dumper::Concise;
use Log::Log4perl ':easy';

Log::Log4perl->easy_init;

# try stuff here ...
use File::Spec::Functions qw( catfile );
use IO qw( Dir File );
use Text::CSV_XS;

#
# get the maid_ids ...
my $dir     = IO::Dir->new( $ENV{regeneron_designs} );
my %maid_id = ();
while ( defined( my $file = $dir->read ) ) {
    $maid_id{$1} = $file if $file =~ m/^(\d+)\.gb$/;
}
$dir->close;
INFO "maid id count: ", scalar keys %maid_id;

#
# get the coordinates with maid ids ...
my $csv        = Text::CSV_XS->new( { sep_char => "\t", eol => $/ } );
my $io         = IO::File->new( $ENV{ko_coordinates} );
my %project_id = ();
$csv->column_names(qw( project_id chromosome start end bac_id ));
while ( my $row = $csv->getline_hr($io) ) {
    $project_id{ $row->{project_id} } = $row
        if exists $maid_id{ $row->{project_id} };
}
$io->close;
INFO "projects with maid ids: ", scalar keys %project_id;

#
# print some file names ...
INFO Dumper [
    map {
        $project_id{$_}{file}
            = catfile( $ENV{regeneron_designs}, $_ . ".gb" );
        $project_id{$_};
        } ( keys %project_id )[ 0 .. 5 ]
];

exit 0;