# prom2.pl (GetContigURLs)
#
# Daniel Westreich, July 2002
# Mullins Lab, Microbiology Dept.
# University of Washington
#
# Usage:	prom2.pl filename
#

use IO::Socket;

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

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

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

    print "For gene ".$gene."\n";
    $contigurl = GetContigURLFromMap($gene, $mapurl);
    if ($contigurl =~ /Error/)
    {
	$error = true;
	print OUT $contigurl."\n";
    }
    else
    {
	#UGLY HACK
	$contigurl =~ s/cgi/fcgi/;

	print OUT $gene."|".$contigurl."\n";
    }

}

if ($error eq true)
{
    print "** ERROR.  Check log file contigurls.out for details.\n\n";
    print "When errors are corrected, invoke next step with 'prom3.pl contigurls.out'\n";
    exit();
}
else
{
    print "\n\nNo errors.  Continue with step 3 (command line 'prom3.pl contigurls.out')? [Y/n]"; 

    if (<STDIN> =~/^[nN]/)
    {
	print "You can invoke the next step of this process at any time with 'prom3.pl contigurls.out'";
    }
    else
    {
	system("prom3.pl contigurls.out");
    }
}
    
######################
# looking for url associated with "sv" after $gene in $url
sub GetContigURLFromMap {
    $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";

    $contigurl = "#Error:  no 'sv' tag found after $gene in $url";

    while (<$remote>)
    {
	if ($_ =~ />$gene<.+>sv</)
	{
	    @words = split(/\"/);
	    chomp(@words);
	    $length = @words;
	    for ($i = 0; $i < $length; $i++)
	    {
		if (@words[$i] =~ />sv</)
		{
		    $contigurl = @words[($i-1)];
		    last;
		}
	    }
	}
    }
    -close $remote;
    return $contigurl;
}