X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=blobdiff_plain;f=dm6;h=34bb654f41f9bf77c33e933557414e51e99c2a68;hp=8b45f507058e652a4093d6c51fe8fb7c57ae0091;hb=941242a8d5dfcd2fbbbd306d00ef38115f841600;hpb=b2ef2da7bbf765cff8910935cfbdedca3f7296cf

diff --git a/dm6 b/dm6
index 8b45f50..34bb654 100755
--- a/dm6
+++ b/dm6
@@ -7,6 +7,7 @@ use File::Compare;
 use File::Path;
 use File::Temp;
 use File::stat;
+use MIME::Base64;
 
 # Copyright (C) 2010 Zygo Blaxell <dm5@mailtoo.hungrycats.org>
 
@@ -67,7 +68,7 @@ sub link_files {
 }
 
 my $link_dir = shift @ARGV;
-(-d $link_dir) or usage;
+usage unless $link_dir;
 
 my $prefix_length = 3;
 
@@ -103,7 +104,25 @@ sub prepare_parents {
 	return "$parent/$suffix";
 }
 
+sub name_quad {
+	my ($int64) = @_;
+	my $packed = pack('Q', $int64);
+	$packed =~ s/^\0+//os;
+	my $base64_packed = encode_base64($packed, '');
+	# 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 $base64_packed;
+}
+
+sub name_ino {
+	my ($ino) = @_;
+	return name_quad($ino) . 'I';
+}
+
 # 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;
 
 $/ = "\0";
@@ -125,7 +144,7 @@ while (<STDIN>) {
 			next if ($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;
@@ -156,7 +175,7 @@ while (<STDIN>) {
 			$digest =~ y:/:_:;
 
 			# Check link to digest
-			my $digest_link = prepare_parents("$link_dir/digest", $digest);
+			my $digest_link = prepare_parents($link_dir, "${digest}D");
 			print STDERR 'D';
 			my $digest_st = lstat($digest_link);
 			if ($digest_st) {
@@ -170,7 +189,7 @@ while (<STDIN>) {
 			print STDERR "\b";
 
 			# Which file are we keeping?
-			my $keep_file;
+			my $keep_ino;
 
 			# If digest link exists, link it to file
 			if ($digest_st) {
@@ -180,17 +199,18 @@ while (<STDIN>) {
 				# Old, replace input with old file
 				print STDERR '-';
 				link_files($digest_link, $file);
-				$keep_file = $digest_link;
+				$keep_ino = $digest_st->ino;
 			} else {
 				# New, add input to digest
 				print STDERR '+';
 				link_files($file, $digest_link);
-				$keep_file = $file;
+				$keep_ino = $st->ino;
 			}
 
 			# A link to the inode indicates we are done, so do it last
-			print STDERR '_';
-			link_files($keep_file, $inode_link);
+			$inode_link = prepare_parents($link_dir, name_ino($keep_ino));
+			print STDERR ' ';
+			link_files($digest_link, $inode_link);
 
 		}
 	};
@@ -201,9 +221,9 @@ while (<STDIN>) {
 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);