use File::Path;
use File::Temp;
use File::stat;
+use MIME::Base64;
-# Copyright (C) 2010 Zygo Blaxell <dm5@mailtoo.hungrycats.org>
+# Copyright (C) 2010 Zygo Blaxell <dupemerge@mailtoo.hungrycats.org>
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
Usage: $0 link-dir
Hashes a NUL-separated list of files on stdin into link-dir.
-Version: 20100513.0
+Version: 0.20100514
USAGE
}
}
my $link_dir = shift @ARGV;
-(-d $link_dir) or usage;
+usage unless $link_dir;
my $prefix_length = 3;
return "$parent/$suffix";
}
+sub name_ino {
+ my ($int64) = @_;
+ my $packed = pack('Q>', $int64);
+ $packed =~ s/^\0+//os;
+ my $base64_packed = encode_base64($packed, '');
+ $base64_packed =~ y:/:_:;
+ # Don't strip off the trailing padding since it makes the string
+ # so short we end up just putting it back on again.
+ # $base64_packed =~ s/=+$//os;
+ return $base64_packed;
+}
+
# ext3 cannot handle more than 32000 links to a file. Leave some headroom.
+# Arguably this should be configurable, but the losses are miniscule and
+# the coding for option support is not.
my $link_count_max = 31990;
$/ = "\0";
next if ($st->nlink > $link_count_max);
# Check link to inode
- my $inode_link = prepare_parents("$link_dir/inode", $st->ino);
+ my $inode_link = prepare_parents($link_dir, name_ino($st->ino));
print STDERR 'I';
my $inode_st = lstat($inode_link);
my $update_links;
$digest =~ y:/:_:;
# Check link to digest
- my $digest_link = prepare_parents("$link_dir/digest", $digest);
+ my $digest_link = prepare_parents($link_dir, $digest);
print STDERR 'D';
my $digest_st = lstat($digest_link);
if ($digest_st) {
}
# A link to the inode indicates we are done, so do it last
- $inode_link = prepare_parents("$link_dir/inode", $keep_ino);
+ $inode_link = prepare_parents($link_dir, name_ino($keep_ino));
print STDERR ' ';
link_files($digest_link, $inode_link);
print STDERR "\nGarbage collection in '$link_dir'...";
chdir($link_dir) || die "chdir: $link_dir: $!";
print STDERR "\nRemoving files with link count < 3...";
-system("find digest inode -type f -links -3 -print0 | xargs -0 rm -f") and die "system: exit status $?";
+system("find . -type f -links -3 -print0 | xargs -0rt rm -f") and die "system: exit status $?";
print STDERR "\nRemoving empty directories...";
-system("find digest inode -type d -empty -print0 | xargs -0r rmdir -p --ignore-fail-on-non-empty") and die "system: exit status $?";
+system("find . -type d -empty -print0 | xargs -0rt rmdir -p --ignore-fail-on-non-empty") and die "system: exit status $?";
print STDERR "\nDone.\n";
exit(0);