1: #!/usr/bin/perl
   2: 
   3: ##########################################################################
   4: #                                                                        #
   5: # How diffy works:                                                       #
   6: # - it simply compares files and prints out the names of identical files #
   7: # - if you give it 2 paths it compares the files in those paths          #
   8: # - if you give it 1 path, it compares just the files in this path       #
   9: # - if you give it no path, the current directory is taken               #
  10: # - if you give the option -h, hardlings will be shown too               #
  11: #                                                                        #
  12: # TODO:                                                                  #
  13: #  make it faster, smaller and more robust                               #
  14: #  print usage with -h and get better options                            #
  15: #                                                                        #
  16: ##########################################################################
  17: 
  18: $hardlinks = 0;
  19: 
  20: 
  21: if ( "$ENV{'DIFFY_SEARCH_SIZE'}" ) {
  22:     $min_size = $ENV{'DIFFY_SEARCH_SIZE'};
  23: } else {
  24:     $min_size = "20k";
  25: }
  26: 
  27: $numDirs = 0;
  28: foreach (@ARGV) {
  29:     if($_ eq "?") {
  30: 	die "*** Usage: diffy [directoy_1 [directory_2]] [-h]\n\n";
  31:     }
  32:     
  33:     if($_ eq "-h") {
  34: 	$hardlinks = 1;
  35:     } elsif($numDirs < 2) {
  36: 	if($numDirs == 0) { $dir1 = $_ }
  37: 	if($numDirs == 1) { $dir2 = $_ }
  38: 		
  39: 	$numDirs++;
  40:     }
  41: }
  42: 
  43: if($numDirs == 0) {
  44:     $dir1 = ".";
  45:     
  46:     $numDirs = 1;
  47: }
  48: 
  49: if($numDirs == 1) {
  50:     $dir2 = $dir1;
  51: }
  52: 
  53: 
  54: %files1 = &getFiles($dir1);
  55: %files2 = &getFiles($dir2);
  56: 
  57: 
  58: @keys1 = sort { $files1{$a} <=> $files1{$b} } keys(%files1);
  59: @keys2 = sort { $files2{$a} <=> $files2{$b} } keys(%files2);
  60: 
  61: 
  62: $numKeys1 = scalar(@keys1) + 1 ;
  63: $numKeys2 = scalar(@keys2) + 1 ;
  64: 
  65: 
  66: $value       = $files1{$keys1[0]};
  67: $i2          = 0;
  68: $index       = 0;
  69: %savedFiles1 = %empty; # $savedFiles1{index} = filename
  70: %savedFiles2 = %empty;
  71: 
  72: for($i1 = 0; $i1 < $numKeys1; $i1++) {
  73:     if($value != $files1{$keys1[$i1]}) {
  74: 	if($index > 0) {
  75: 	    ### now delete double pairs and print the remaining pairs
  76: 	    
  77: 	    @savedKeys = keys(%savedFiles1);
  78: 	    $numSaved  = scalar(@savedKeys);
  79: 			
  80: 	    while($numSaved > 0) {
  81: 		for($iSave = 1; $iSave < $numSaved; $iSave++) {
  82: 		    if($savedFiles1{$savedKeys[0]} eq $savedFiles2{$savedKeys[$iSave]} &&
  83: 		       $savedFiles2{$savedKeys[0]} eq $savedFiles1{$savedKeys[$iSave]}) {
  84: 			
  85: 			delete $savedFiles1{$savedKeys[$iSave]};
  86: 			delete $savedFiles2{$savedKeys[$iSave]};
  87: 			last;
  88: 		    }
  89: 		}
  90: 		print "ln -f \"".$savedFiles1{$savedKeys[0]}."\""." "."\"".$savedFiles2{$savedKeys[0]}."\""."\n";
  91: 		
  92: 		delete $savedFiles1{$savedKeys[0]};
  93: 		delete $savedFiles2{$savedKeys[0]};
  94: 		
  95: 		@savedKeys = keys(%savedFiles1);
  96: 		$numSaved  = scalar(@savedKeys);
  97: 	    }
  98: 	}
  99: 	$index       = 0;
 100: 	%savedFiles1 = %empty; # just to be sure
 101: 	%savedFiles2 = %empty;
 102: 	
 103: 	$value = $files1{$keys1[$i1]};
 104:     }
 105:     
 106:     while($files2{$keys2[$i2]} < $files1{$keys1[$i1]}) {
 107: 	$i2++;
 108: 	if($i2 > $numKeys2) { exit; }
 109:     }
 110:     
 111:     if($files2{$keys2[$i2]} == $files1{$keys1[$i1]}) {
 112: 	$i2Old = $2;
 113: 	
 114: 	while($files2{$keys2[$i2]} == $files1{$keys1[$i1]}) {
 115: 	    if($keys2[$i2] ne $keys1[$i1]) {
 116: 		open(FILE, $keys1[$i1]);
 117: 		$inode1 = (stat(FILE))[1];
 118: 		close(FILE);
 119: 		
 120: 		open(FILE, $keys2[$i2]);
 121: 		$inode2 = (stat(FILE))[1];
 122: 		close(FILE);
 123: 		
 124: 		if($inode1 == $inode2) {
 125: 		    if($hardlinks == 1) {
 126: 			$index++;
 127: 			$savedFiles1{$index} = $keys1[$i1];
 128: 			$savedFiles2{$index} = $keys2[$i2];
 129: 		    }
 130: 		} else {
 131: 		    if(system("cmp -s \"$keys1[$i1]\" \"$keys2[$i2]\"") == 0) {
 132: 			$index++;
 133: 			$savedFiles1{$index} = $keys1[$i1];
 134: 			$savedFiles2{$index} = $keys2[$i2];
 135: 		    }
 136: 		}
 137: 	    }
 138: 	    $i2++;
 139: 	    if($i2 > $numKeys2) { exit; }
 140: 	}
 141: 	$i2 = $i2Old;
 142: 
 143:         # $i2 gets a reset, because every file in %files1 has to be compared with
 144:         # every file in %files2 of equal size. Otherwise some files would be skipped.	
 145:     }
 146: }
 147: 
 148: sub getFiles {
 149:     local($dir, %list, $file, $size);
 150:     
 151:     $dir = $_[0];
 152:     unless(-d $dir) { die "*** ERROR: $dir is not a directory.\n"; }
 153:     
 154:     open(DIR, "find $dir -size +$min_size -type f -printf \"%p %s\\n\" |") || 
 155: 	die "*** ERROR: Cannot access $dir.\n";
 156:     while(<DIR>) {
 157: 	chop;
 158: 	($file, $size) = split ( /\s/, $_, 2 );
 159: 	if(-f $file) {
 160: 	    $list{$file} = $size;
 161: 	}
 162:     }
 163:     close(DIR);
 164:     
 165:     unless(keys(%list)) { die "*** No files in $dir\n"; }    
 166:     %list;
 167: }