#!/usr/bin/perl -w
use strict;

# This script will download the nr databases from NCBI to $DATA_DIR

use File::stat;
use Time::localtime;
use Net::FTP;

my($NCBI_URL) = "ftp.ncbi.nlm.nih.gov";
my($DB_DIR)   = "/blast/db";
my(@DB_FILES) = ("nr.tar.gz", "nt.00.tar.gz", "nt.01.tar.gz", "nt.02.tar.gz", "nt.03.tar.gz");

my($USER);
my($DATA_DIR);

my(@NEWFILES) = ();

my($WGET);
my($WGET_PARMS) = "-Nq";

my($GZIP);
my($TAR);

main();
exit;

sub main
{
	getarguments();
	download();
}

sub getarguments 
{
	my($parm);

	# determine arguments
	if (scalar(@ARGV) > 0) {
		foreach $parm(@ARGV) {
			if ($parm eq "-h") {
				print "Usage: $0 <options>\n";
				print "          -v     Verbose\n";
				print "          -h     This help message\n";
				die "\n";
			}
			if ($parm eq "-v") {
				$WGET_PARMS = "-N";
			}
		}
	}
}

sub download
{
	my($ncbi);
	my($dbfile);

	# set some parameters
	$USER = "root";
	$DATA_DIR = "/usr/local/ncbi/db";
	$WGET = "/usr/bin/wget";
	$GZIP = "/bin/gzip";
	$TAR = "/bin/tar";

	# determine which files to download by date
	$ncbi = Net::FTP->new("$NCBI_URL");
	$ncbi->login("anonymous", "seqadmin\@umdnj.edu");

	my($remote_filedate, $local_filedate);
	chdir "$DATA_DIR";
	foreach $dbfile(@DB_FILES)
	{
		$remote_filedate = 0;
		$local_filedate = 0;

		$remote_filedate = $ncbi->mdtm("$DB_DIR/$dbfile");

		if (-e $dbfile) {
			$local_filedate = stat("$dbfile")->mtime;
		}

		if ($remote_filedate > $local_filedate) {
			push @NEWFILES, $dbfile;
		}
	}

	# download and uncompress the newer files
	foreach $dbfile(@NEWFILES)
	{
		`$WGET $WGET_PARMS ftp://$NCBI_URL$DB_DIR/$dbfile`;
		
		`$GZIP -dc $dbfile | $TAR xvf -`;
		`chown root:root *`;
	}	
}