tools/dataDirectoryDigestBuilder.pl

   1 #!/usr/bin/perl
   2
   3 use Digest::MD5 qw(md5_hex);
   4
   5 die "usage: perl buildTreeFileDigest.pl <config_file> <digest_name> [<old_digest_name>]\n" if (scalar(@ARGV) < 2 || scalar(@ARGV) > 3 || $ARGV[0] eq "-h" || $ARGV[0] eq "-?");
   6
   7 $configFile = shift;
   8 $digest = shift;
   9
  10 sub numerically
  11 {
  12         return ($a <=> $b);
  13 }
  14
  15 sub do_directory
  16 {
  17         local $_;
  18         my $base = $_[0];
  19         my $source = $_[1];
  20         my $name = $_[2];
  21
  22         opendir(DIR, $source) || die "could not open source directory $source\n";
  23         my @filenames = readdir(DIR);
  24         closedir(DIR);
  25
  26         for (@filenames)
  27         {
  28                 next if ($_ eq "." || $_ eq "..");
  29
  30                 if (-d "$source/$_")
  31                 {
  32                         if ($name eq "")
  33                         {
  34                                 &do_directory($base, "$source/$_", "$_");
  35                         }
  36                         else
  37                         {
  38                                 &do_directory($base, "$source/$_", "$name/$_");
  39                         }
  40                 }
  41                 else
  42                 {
  43                         $count += 1;
  44                         print "." if ($count % 500 == 0);
  45
  46                         ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,$atime,$mtime,$ctime,$blksize,$blocks) = stat("$source/$_");
  47
  48                         my $md5sum = "";
  49
  50                         if (defined $old{"$name/$_"})
  51                         {
  52                                 ($oldName, $oldPath, $oldSize, $oldTime, $oldDigest) = split(/\s+/, $old{"$name/$_"}, 5);
  53                                 if ($size == $oldSize && $mtime == $oldTime)
  54                                 {
  55                                         $md5sum = $oldDigest;
  56                                         # print "cached  $name/$_ '$md5sum'\n";
  57                                 }
  58                                 else
  59                                 {
  60                                         # print "changed $name/$_ !($size==$oldSize && $mtime==$oldTime)\n";
  61                                 }
  62                         }
  63                         if ($md5sum eq "")
  64                         {
  65                                 my $buffer;
  66                                 open(FILE, "$source/$_") || die "could not spawn md5sum on file $source/$_";
  67                                         binmode FILE;
  68                                         read(FILE, $buffer, -s "$source/$_");
  69                                 close(FILE);
  70                                 $md5sum = md5_hex($buffer);
  71                                 # print "compute $name/$_ $md5sum\n";
  72                         }
  73
  74                         $files{"$name/$_"} = "$name/$_ $base $size $mtime $md5sum";
  75                 }
  76         }
  77 }
  78
  79 if (@ARGV)
  80 {
  81         open(DIGEST, shift) || die "could not open old digest file\n";
  82         while (<DIGEST>)
  83         {
  84                 chomp;
  85                 ($oldName, $oldPath, $oldSize, $oldTime, $oldDigest) = split(/\s+/, $_);
  86                 # print "caching $oldName: $_\n";
  87                 $old{$oldName} = $_;
  88         }
  89         close(DIGEST);
  90 }
  91
  92
  93 open(CONFIG, $configFile) || die "could not open config file $configFile\n";
  94         while (<CONFIG>)
  95         {
  96                 chomp;
  97
  98                 if (/searchPath/)
  99                 {
 100                         s/^.*searchPath//;
 101                         s/\s+//;
 102                         push(@paths, $_);
 103                 }
 104         }
 105 close(CONFIG);
 106
 107 foreach (sort numerically @paths)
 108 {
 109         $count = 0;
 110         ($priority, $path) = split(/=/);
 111         print "processing $path($priority)";
 112         do_directory($path, $path, "");
 113         print "\n";
 114 }
 115
 116 open(DIGEST, ">$digest");
 117         foreach (sort keys %files)
 118         {
 119                 print DIGEST $files{$_}, "\n";
 120         }
 121 close(DIGEST);