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