#filiecopare compares two lists, outputs files that match both,in 1 missing in 2, and in 2 missing in 1
#4/4/04 (love that date!) 
#Charles Oppenheimer - distributed under perl artistic license.  go crazy.



$file1 = $ARGV[0] || "ourvalues.txt";
$file2 = $ARGV[1] || "theirvalues.txt";



########### output files
$matchesboth = "filematchesboth.txt";
$missingin2 = "in1missingin2.txt";
$missingin1 = "in2missingin1.txt";


print "Start up - comparing files $file1 and $file2\n";
print "opening files...\n";

open(MATCHESBOTH, "> $matchesboth") || die "Could not open file $matchesboth : $!";
open(MISSINGING2, "> $missingin2") || die "Could not open file $missingin2 : $!";
open(MISSINGING1, "> $missingin1") || die "Could not open file $missingin1 : $!";


print "creating hash1\n";
%filehash1 = &filetohash($file1);

print "creating hash2\n";
%filehash2 = &filetohash($file2);

### main part, compare those in 1 that are in 2 and aren't in 2
#check for matches, and those in file 1 but not 2 

$missing2;
$matchboth;
$missing1;

print "checking for matches in both files and values in file 1 and missing in file2\n";

		for my $key (keys %filehash1) {


			#print "checking key $key for filehash1\n";

			if ($filehash2{$key} == 1) {
				#print "a hash2 of this value equals on i guess...";
				print MATCHESBOTH "$key\n";
				$matchboth++;

			} else {

				print MISSINGING2 "$key\n";
				#print "didn't match hash2 = $key\n";
				$missing2++;

			}

		}

#do a second pass for those in file 2 but not 1
print "checking for values missing in 2 and missing in file1\n";

		for my $key (keys %filehash2) {

			unless ($filehash1{$key} == 1) {

				print MISSINGING1 "$key\n";
				$missing1++;

			} 

		}

print "found $matchboth records matching both files, $missing2 records in file 1 missing from file 2, and $missing1 records in file 2 missing from file 1\n";
	

sub filetohash {

	my $counter;
	my $dir = shift;
	my %filehash;
	open(FILE, "<$dir") || die $!;
	
		while (<FILE>) {

			#print "chomping file $dir, line = $line"; 
			chomp $_; #remove newline
			s/^ *//; #remove leading spaces
			s/ *$//; #remove trailing spaces 

			$filehash{$_} = 1;  #set the currentline to 1, for comparison later
			$counter++;
		} 

		print "counted $counter lines for file $dir\n";

	return %filehash;

}