#!/usr/bin/perl

if ($ARGV[0] eq "-help" or $#ARGV < 1) { die "
USAGE:   $0 [d/t] sander_output_files

Where 'd' is used to create a list of distance restraint 
violations and 't' is used to create a list of torsion 
restraint violations. The violations are taken from a list
of sander output files, which may contain wildcards.
$0 creates a list of the violations including 
the number of structures in which each violation occurs,
the mean violation, the distance bin or the violated side 
of the torsion restraint, the restraint, and the crosspeak 
number (if distance restraints are being examined). Violations
are listed from most frequent to least frequent. Equally 
frequent violations are listed from largest mean violation
to smallest.
\n";
}

$flag = lc(shift(@ARGV));

foreach $file (@ARGV) {
	if ($flag eq "d") { $input = "grep ' d ' ".$file." | "; }
	elsif ($flag eq "t") { $input = "grep ' t\$' ".$file." | "; }
	open(VIOLS, $input) or die "can't open $file";
	while (<VIOLS>) {
		$_ =~ s/^\s+//;
		@field = split(/\s+/, $_);
		$field[6] = substr($field[6], 0, -1);
		$restraint = sprintf "%8.3f  %5s %3s %-3s -- %5s %3s %-3s",
   $field[8], $field[0], $field[1], $field[2], $field[4], $field[5], $field[6];
		$xpkno{$restraint} = $field[12];
		++$number{$restraint};
		unless ( $violation{$restraint} ) {
			$violation{$restraint} = $field[9];
			$sum_violations{$restraint} = $field[9];
		}
		else {
			$sum_violations{$restraint} += $field[9];
		}
	}
}
foreach $restraint (keys %violation) {
	$mean = $sum_violations{$restraint}/$number{$restraint};
	push @output, sprintf "%5d   %6.3f  %s   %4s", $number{$restraint}, $mean, $restraint, $xpkno{$restraint};
}
@sorted_output = sort biggest_first @output;
print "---------------------------------------------------------------\n";
print " #viols   mean      bin     first atom      second atom   xpkno\n";
print "---------------------------------------------------------------\n";
foreach $violation (@sorted_output) { print "$violation\n"; }

sub biggest_first { $b cmp $a; }
end;