# prom4.pl (GetPromoterInfo)
#
# Daniel Westreich, July 2002
# Mullins Lab, Microbiology Dept.
# University of Washington
#
# Usage:	prom4.pl filename
# Alt usage:	prom4.pl filename > out.htm
#

use IO::Socket;

$host = "www.ncbi.nlm.nih.gov";
$frontcontigdoc = "/entrez/query.fcgi?cmd=Retrieve&db=nucleotide&list_uids=";
$endcontigdoc = "&dopt=GenBank";

$error = false;
open(OUT, ">promoterinfo.out");

while(<>) {
    if (/^\#/) #comments
    {
	next;
    }
    chomp($_);
    @words= split(/\|/);
    if (@words == 0)
    {
	next;
    }
    $gene = @words[0];
    $gi = @words[1];
    $strand = @words[2];
    
    print "For gene ".$gene."\n";

    $address = GetAddressFromContigURL($gene,$gi,$strand);
    if ($address =~ /Error/)
    {
	$error = true;
    }
    print OUT $address."\n";

}

if ($error eq true)
{
    print "** ERROR.  Check log file promoterinfo.out for details.\n\n";
    print "When errors are corrected, invoke next step with 'prom5.pl promoterinfo.out'";
    exit();
}
else
{
    print "\n\nNo errors.  Continue with step 5 (command line 'prom5.pl promoterinfo.out')? [Y/n]"; #

    if (<STDIN> =~/^[nN]/)
    {
	print "You can invoke the next step of this process at any time with 'prom5.pl promoterinfo.out'";
    }
    else
    {
	system("prom5.pl promoterinfo.out");
    }
}

######################
sub GetAddress {
    if ($_[0] eq "Crick")
    {
	$_[2] =~ /\,[0-9]+\.\.([0-9]+)\)/;
	if ($1 eq "")	# then there is only one segment to the CDS
	{
		$_[2] =~ /[0-9]+\.\.([0-9]+)/;
		return $1;
	}
	return $1;
    }
    else # ($_[0] eq "Watson")
    {
	$_[1] =~ /join\(([0-9]+)\.\.[0-9]+/;

	if ($1 eq "") 	# then there's no "join" in the line
	{
		$_[1] =~ /([0-9]+)\.\.[0-9]+/;
		return $1;
	}
	return $1;
    }
}

sub GetAddressFromContigURL {
    $gene = $_[0];
    $gi = $_[1];
    $strand = $_[2];
    $document = $frontcontigdoc.$gi.$endcontigdoc;

    print("**".$host.$document." **\n");

    $remote = IO::Socket::INET->new( Proto => "tcp", PeerAddr => $host, PeerPort => "http(80)", );
    unless($remote)  { die "cannot connect to http daemon on $host" }
    $remote->autoflush(1);
    print $remote "GET $document HTTP/1.0\n\n";

    $address = "#Error - could not find address in contig for $gene\n#Check host http://".$host.$document."\n\n";

    $continue = 1;
    while (<$remote>)
    {
	if ($_ =~ /gene/)
	{
	    $next = <$remote>;
	    if ($next =~ /$gene/)
	    {
		while(<$remote>)
		{
		    if ($_ =~ /mRNA<\/a>\s+complement/ || $_ =~ /mRNA<\/a>\s+join/)
		    {
			$firstRNA = $_;
			while(<$remote>)
			{
			    if ($_ =~ /gene/) { last; }
			    $lastRNA = $_;
			}
			$addressRNA = GetAddress($strand, $firstRNA, $lastRNA);
			print "address of RNA is ".$addressRNA."\n";
			$address = $gene."|".$gi."|".$strand."|".$addressRNA;
			$continue = 0;
			last;
		    }
		}
	    }
	}
	if ($continue == 0) { last; }
    }

-close $remote;

if ($address =~ /Error/)
{
    return $address;
}

return $address;
}