#!/usr/bin/perl
# The first line might be #!/usr/local/bin/perl etc 
# depending on the system.
# gsort.pl: 
# Do not delete this comment.#####################

if ( $#ARGV < 0 ) { 
	print "usage: $0 summary_file hom_file switch > outfile.\n";
	print "Homology matrix is extracted according to the summary information.\n";
	print "switch: 0 - default; 1 - output also related sequences.\n";
	exit 1;
}

$infile = $ARGV[0];
$homfile = $ARGV[1];
$switch = $ARGV[2];
$startRelated = "  Related groups:";
$endRelated = "  END Related groups.";
$domainlist = "Domain list";

print STDERR "Summary file is $infile.\n";
print STDERR "Homolog file is $homfile.\n";
if ( $switch == 1) {
	print "Output includes related groups.\n";
}
open(INFILE, $infile) || die "Unable to open summary file: $infile.\n";
open(HOMFILE, $homfile) || die "Unable to open homology file: $homfile.\n";
$exitflag = 0; 
while(<INFILE>) {
	if ( $_ =~ /^\ / || $_ =~ /^Number/ || $_ eq "\n" ) {
		next;
	}
	@list = split( /\s+/, $_ );
	$grpno = $list[0];
	$grpstring = "Group " . $grpno . ":";
	$outputflag = 0;
	while(<HOMFILE>) {
		if ( $outputflag == 1 ){
			print $_;
			if ( $_ eq "\n" ) {
				$counter += 1;
				if ( $counter >= 2 ) {
					if ( $switch == 1 ) {
						$outputflag = 2;
						next;
					} else {
						$outputflag = 0;
						last;
					}
				}
			}
		} elsif ( $outputflag == 2 ) {
			if ( $_ =~ /^$startRelated/ ) {
				print $_;
				$counter = 0;
			} elsif ( $_ =~ /^$endRelated/ ) {
				print $_;
				print "\n";
				$outputflag = 0;
				$counter = 0;
				last;
			} elsif ( $_ =~/$domainlist/ ) {
				$counter = -1;
				print $_;
			} elsif ( $_ eq "\n" ) {
				$counter += 1;
				if ( $counter >= 2 ) {
					$counter = 0;
					$outputflag = 0;
					last;
				} else {
					print $_;
				}
			} else {
				print $_;
			}
				
		} else {
			$line = $_;
			if ( $line eq "\n" ) {
				next;
			}
			if ( $line =~ /^$grpstring/ ) {
				$outputflag = 1;
				$counter = 0;
				print $line;
				next;
			}
		}
	}
}
close(INFILE); 
close(HOMFILE); 
#end