#! /usr/bin/perl
# Assuming that you have perl, and that you have changed the permissions on this file so that it can be executable ("chmod u+x diff_col")
# then the usage is: 
#                  diff_col [colnum1] [colnum2] [threshold] [file1] [file2]
#                                               [threshold] is in percent
#
#	written by: Jaideep Singh
#	      date: March 24, 2007
#	     email: singhj AT jlab DOT org
#
#             desc: compares the nth columns of file 1 and file 2
#                   outputs the data when the rel diff is larger than threshold in pct.
#                   skips over lines starting with "#"
#                   columns are assumed to be separated by some whitespace

$colnum1 = $ARGV[0];
$colnum2 = $ARGV[1];
$thresh = $ARGV[2];
$file1  = $ARGV[3];
$file2  = $ARGV[4];

$usage = "Usage: diff_col [colnum1] [colnum2] [threshold] [file1] [file2]\n";
die "$usage" unless ($#ARGV == 4);

open(FILEONE,"<$file1") or die "Having trouble opening $file1\n";
open(FILETWO,"<$file2") or die "Having trouble opening $file2\n";

@data1 = ();
@data2 = ();

$num1 = 0;
$num2 = 0;

while(<FILEONE>) {
    chomp;
    unless (/^#/) {
        $raw = $_;
        $raw =~ s/^\s+//;
        @line_parts = split /\s+/ , $raw;
        $data1[$num1] = $line_parts[$colnum1];
        $num1++;
    }
}

print "column number is out of bounds in $file1 ($#line_parts)\n"  if ($#line_parts < $colnum1);

while(<FILETWO>) {
    chomp;
    unless (/^#/) {
        $raw = $_;
        $raw =~ s/^\s+//;
        @line_parts = split /\s+/ , $raw;
        $data2[$num2] = $line_parts[$colnum2];
        $num2++;
    }
}

print "column number is out of bounds in $file2 ($#line_parts)\n"  if ($#line_parts < $colnum2);

close(FILEONE);
close(FILETWO);

if ($num1 == $num2) {
    print "number of rows match! $num1\n";
    $num = $num1
}
elsif ($num1 > $num2) {
    print "number of rows do not match: $num1 !\ $num2\n";
    $num = $num2;
}
else {
    print "number of rows do not match: $num1 !\ $num2\n";
    $num = $num1;
}

$badcol0 = 0;
$badcolnum = 0;
$badcolstr = 0;

for ($n = 0 ; $n < $num ; $n++) {
    $one = $data1[$n];
    $two = $data2[$n];
#    if ( ($one =~ /[0123456789\.\+\-eE]+/) and ($two =~ /[0123456789\.\+\-eE]+/)  ) {
    if ( ($one =~ /\d+/) and ($two =~ /\d+/)  ) {
        if ( $one + $two == 0.0 && $one == 0.0) {
#                $ratio = $two-$one;
#                if ( abs($ratio) > $thresh ) {
#                    $ratio = sprintf "%.2F" , $ratio;
#                    print "$n \t $one \t $two \t $ratio \t sum is 0.0\n";
#                    $badcol++;
#                }
        }
        else {
                $ratio = 200.0*($two-$one)/($two+$one);
                if ( abs($ratio) >= $thresh ) {
                    $ratio = sprintf "%.2F" , $ratio;
                    print "$n \t $one \t $two \t $ratio \n";
                    $badcolnum++;
                }
        }
     }
     elsif ($one ne $two) {
            print "$n \t $one \t $two \t strings \n";
                    $badcolstr++;
     }
#     else { print "$one \t $two \t oh really...\n"; }
}

print "\"bad\" sum = 0 columns: $badcol0   out of $num\n" unless ($badcol0   == 0);
print "\"bad\" nummber columns: $badcolnum out of $num\n" unless ($badcolnum == 0);
print "\"bad\" string  columns: $badcolstr out of $num\n" unless ($badcolstr == 0);