#!/usr/bin/perl

##########################################################################
#                                                                        #
# How diffy works:                                                       #
# - it simply compares files and prints out the names of identical files #
# - if you give it 2 paths it compares the files in those paths          #
# - if you give it 1 path, it compares just the files in this path       #
# - if you give it no path, the current directory is taken               #
# - if you give the option -h, hardlings will be shown too               #
#                                                                        #
# TODO:                                                                  #
#  make it faster, smaller and more robust                               #
#  print usage with -h and get better options                            #
#                                                                        #
##########################################################################

$hardlinks = 0;


if ( "$ENV{'DIFFY_SEARCH_SIZE'}" ) {
    $min_size = $ENV{'DIFFY_SEARCH_SIZE'};
} else {
    $min_size = "20k";
}

$numDirs = 0;
foreach (@ARGV) {
    if($_ eq "?") {
	die "*** Usage: diffy [directoy_1 [directory_2]] [-h]\n\n";
    }
    
    if($_ eq "-h") {
	$hardlinks = 1;
    } elsif($numDirs < 2) {
	if($numDirs == 0) { $dir1 = $_ }
	if($numDirs == 1) { $dir2 = $_ }
		
	$numDirs++;
    }
}

if($numDirs == 0) {
    $dir1 = ".";
    
    $numDirs = 1;
}

if($numDirs == 1) {
    $dir2 = $dir1;
}


%files1 = &getFiles($dir1);
%files2 = &getFiles($dir2);


@keys1 = sort { $files1{$a} <=> $files1{$b} } keys(%files1);
@keys2 = sort { $files2{$a} <=> $files2{$b} } keys(%files2);


$numKeys1 = scalar(@keys1) + 1 ;
$numKeys2 = scalar(@keys2) + 1 ;


$value       = $files1{$keys1[0]};
$i2          = 0;
$index       = 0;
%savedFiles1 = %empty; # $savedFiles1{index} = filename
%savedFiles2 = %empty;

for($i1 = 0; $i1 < $numKeys1; $i1++) {
    if($value != $files1{$keys1[$i1]}) {
	if($index > 0) {
	    ### now delete double pairs and print the remaining pairs
	    
	    @savedKeys = keys(%savedFiles1);
	    $numSaved  = scalar(@savedKeys);
			
	    while($numSaved > 0) {
		for($iSave = 1; $iSave < $numSaved; $iSave++) {
		    if($savedFiles1{$savedKeys[0]} eq $savedFiles2{$savedKeys[$iSave]} &&
		       $savedFiles2{$savedKeys[0]} eq $savedFiles1{$savedKeys[$iSave]}) {
			
			delete $savedFiles1{$savedKeys[$iSave]};
			delete $savedFiles2{$savedKeys[$iSave]};
			last;
		    }
		}
		print "ln -f \"".$savedFiles1{$savedKeys[0]}."\""." "."\"".$savedFiles2{$savedKeys[0]}."\""."\n";
		
		delete $savedFiles1{$savedKeys[0]};
		delete $savedFiles2{$savedKeys[0]};
		
		@savedKeys = keys(%savedFiles1);
		$numSaved  = scalar(@savedKeys);
	    }
	}
	$index       = 0;
	%savedFiles1 = %empty; # just to be sure
	%savedFiles2 = %empty;
	
	$value = $files1{$keys1[$i1]};
    }
    
    while($files2{$keys2[$i2]} < $files1{$keys1[$i1]}) {
	$i2++;
	if($i2 > $numKeys2) { exit; }
    }
    
    if($files2{$keys2[$i2]} == $files1{$keys1[$i1]}) {
	$i2Old = $2;
	
	while($files2{$keys2[$i2]} == $files1{$keys1[$i1]}) {
	    if($keys2[$i2] ne $keys1[$i1]) {
		open(FILE, $keys1[$i1]);
		$inode1 = (stat(FILE))[1];
		close(FILE);
		
		open(FILE, $keys2[$i2]);
		$inode2 = (stat(FILE))[1];
		close(FILE);
		
		if($inode1 == $inode2) {
		    if($hardlinks == 1) {
			$index++;
			$savedFiles1{$index} = $keys1[$i1];
			$savedFiles2{$index} = $keys2[$i2];
		    }
		} else {
		    if(system("cmp -s \"$keys1[$i1]\" \"$keys2[$i2]\"") == 0) {
			$index++;
			$savedFiles1{$index} = $keys1[$i1];
			$savedFiles2{$index} = $keys2[$i2];
		    }
		}
	    }
	    $i2++;
	    if($i2 > $numKeys2) { exit; }
	}
	$i2 = $i2Old;

        # $i2 gets a reset, because every file in %files1 has to be compared with
        # every file in %files2 of equal size. Otherwise some files would be skipped.	
    }
}

sub getFiles {
    local($dir, %list, $file, $size);
    
    $dir = $_[0];
    unless(-d $dir) { die "*** ERROR: $dir is not a directory.\n"; }
    
    open(DIR, "find $dir -size +$min_size -type f -printf \"%p %s\\n\" |") || 
	die "*** ERROR: Cannot access $dir.\n";
    while(<DIR>) {
	chop;
	($file, $size) = split ( /\s/, $_, 2 );
	if(-f $file) {
	    $list{$file} = $size;
	}
    }
    close(DIR);
    
    unless(keys(%list)) { die "*** No files in $dir\n"; }    
    %list;
}