# prom3.pl (GetContigInfo)
#
# Daniel Westreich, July 2002
# Mullins Lab, Microbiology Dept.
# University of Washington
#
# Usage:	prom3.pl filename
#

use IO::Socket;

$host = "www.ncbi.nlm.nih.gov";

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

while(<>) {
    if (/^\#/) #comments
    {
	next;
    }
    chomp($_);
    @words= split(/\|/);
    if (@words == 0)
    {
	next;
    }
    $gene = @words[0];
    $contigurl = @words[1];

    print "For gene ".$gene."\n";

    $temp = GetContigInfo($gene, $contigurl);
    if ($temp =~ /Error/)
    {
	$error = true;
	print OUT $temp."\n";
    }
    else
    {
	print OUT $gene."|".$temp."\n";
    }
}

if ($error eq true)
{
    print "** ERROR.  Check log file contiginfo.out for details.\n\n";
    print "When errors are corrected, invoke next step with 'prom4.pl contiginfo.out'\n";
    exit();
}
else
{
    print "\n\nNo errors.  Continue with step 4 (command line 'prom4.pl contiginfo.out')? [Y/n]"; 

    if (<STDIN> =~/^[nN]/)
    {
	print "You can invoke the next step of this process at any time with 'prom4.pl contiginfo.out'";
    }
    else
    {
	system("prom4.pl contiginfo.out");
    }
}

#  ABANDONED GETTING CONTIGID below; can always get it later from the url itself
sub GetContigInfo {
    $gene = $_[0];
    $url = $_[1];
    
    $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 $url HTTP/1.0\n\n";

    $contigid = "#$gene:  Error - failed to get contigid from $url";
    
    while (<$remote>)
    {
	if ($_ =~ /checkbox.+[1-9].+\[gi\:(.+)\]/)
	{
	    $gi = $1;
	    last;
	}
    }
    while (<$remote>)
    {
	if ($_ =~ /plus strand/)
	{
	    $strand = "Crick";
	    last;
	}
	if ($_ =~ /minus strand/)
	{
	    $strand = "Watson";
	    last;
	}
    }
    -close $remote;
    return $gi."|".$strand;
}