X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=blobdiff_plain;f=dm6;h=a5d56a6545a59258d2742e7dcfcedd0a4be5d843;hp=c5ce5d45ee6ec6e0760df40589bef258d3ca1e54;hb=790ad0c146598175d7cca5f0c711ec217e196369;hpb=8104c13303bb6b502c35ace2d956919367e51760

diff --git a/dm6 b/dm6
index c5ce5d4..a5d56a6 100755
--- a/dm6
+++ b/dm6
@@ -6,6 +6,7 @@ use Fcntl qw(:DEFAULT :flock);
 use File::Compare;
 use File::Path;
 use File::Temp;
+use File::stat;
 
 # Copyright (C) 2010 Zygo Blaxell <dm5@mailtoo.hungrycats.org>
 
@@ -38,6 +39,8 @@ sub usage {
 	die <<USAGE;
 Usage: $0 link-dir
 Hashes a NUL-separated list of files on stdin into link-dir.
+
+Version: 20100513.0
 USAGE
 }
 
@@ -45,37 +48,130 @@ USAGE
 sub link_files {
 	my ($from, $to) = (@_);
 
+	print STDERR 'T';
 	my $inode_dir = $to;
 	my $inode_base = $to;
 	$inode_dir =~ s:[^/]*$::o;
 	$inode_base =~ s:^.*/::os;
 	my $tmp_to = File::Temp::tempnam($inode_dir, ".$inode_base.");
+	print STDERR "\bL";
 	link($from, $tmp_to) or die "link: $from -> $tmp_to: $!";
+	print STDERR "\bR";
 	unless (rename($tmp_to, $to)) {
 		my $saved_bang = $!;
+		print STDERR "\bU";
 		unlink($tmp_to) or warn "unlink: $tmp_to: $!";	# Try, possibly in vain, to clean up
 		die "rename: $tmp_to -> $from: $saved_bang";
 	}
+	print STDERR "\b \b";
 }
 
 my $link_dir = shift @ARGV;
 (-d $link_dir) or usage;
 
+sub slash_prefix {
+	my ($file) = @_;
+	my $prefix = substr($file, 0, 3);
+	my $suffix = substr($file, 3);
+	$prefix =~ s:(.):$1/:osg;
+	chop($prefix);
+	return ($prefix, $suffix);
+}
+
+sub prepare_parents {
+	my ($link_dir, $file) = @_;
+	my ($prefix, $suffix) = slash_prefix($file);
+	my $parent = "$link_dir/$prefix";
+	print STDERR 'm';
+	mkpath($parent, { verbose => 0 });
+	print STDERR "\b";
+	die "mkpath: $parent: $!" unless -d $parent;
+	return "$parent/$suffix";
+}
+
+# ext3 cannot handle more than 32000 links to a file.  Leave some headroom.
+my $link_count_max = 31990;
+
 $/ = "\0";
 while (<STDIN>) {
 	my $file = $_;
 	eval {
-		chomp $file;
-		print STDERR "digest($file) = ";
-		my $digest = digest($file);
-		$digest =~ y:/:_:;
-		print STDERR "$digest\n";
-		$digest =~ s:^(.)(.)(.):$1/$2/$3:osg;
-		my ($parent) = ($digest =~ m:^(.*/):osg);
-		$parent = "$link_dir/$parent";
-		mkpath($parent, { verbose => 1 });
-		die "mkpath: $parent: $!" unless -d $parent;
-		link_files($file, "$link_dir/$digest");
+		for (1) {
+			chomp $file;
+
+			# Get file stat data
+			print STDERR '.';
+			my $st = lstat($file);
+			die "lstat: $file: $!" unless $st;
+
+			# Oops?
+			next unless -f _;
+
+			# Skip the file if it has far too many links already
+			next if ($st->nlink > $link_count_max);
+
+			# Check link to inode
+			my $inode_link = prepare_parents("$link_dir/inode", $st->ino);
+			print STDERR 'I';
+			my $inode_st = lstat($inode_link);
+			my $update_links;
+			if ($inode_st) {
+				my $inode_dev = $inode_st->dev;
+				my $inode_ino = $inode_st->ino;
+				my $file_dev = $st->dev;
+				my $file_ino = $st->ino;
+				if ($inode_ino != $file_ino || $inode_dev != $file_dev) {
+					warn "inode link '$inode_link' is wrong (inode $inode_ino should be $file_ino)" if $inode_ino != $file_ino;
+					warn "inode link '$inode_link' is wrong (dev $inode_dev should be $file_dev)" if $inode_dev != $file_dev;
+					$update_links = 1;
+				}
+			} else {
+				$update_links = 1;
+			}
+			print STDERR "\b";
+
+			# If neither criteria for updating link is met, leave it as-is
+			next unless $update_links;
+
+			# Compute digest
+			print STDERR 'd';
+			my $digest = digest($file);
+			print STDERR "\b";
+
+			# Base64 uses /, we prefer _
+			$digest =~ y:/:_:;
+
+			# Check link to digest
+			my $digest_link = prepare_parents("$link_dir/digest", $digest);
+			print STDERR 'D';
+			my $digest_st = lstat($digest_link);
+			if ($digest_st) {
+				my $digest_nlink = $digest_st->nlink;
+				if ($digest_nlink > 31990) {
+					print STDERR 'u';
+					unlink($digest_link) or die "unlink: $digest_link: $!";
+					undef $digest_st;
+				}
+			}
+
+			# If digest link exists, link it to file
+			if ($digest_st) {
+				print STDERR 'c';
+				die "NOT identical!" if compare($digest_link, $file);
+
+				# Old, replace input with old file
+				print STDERR '-';
+				link_files($digest_link, $file);
+			} else {
+				# New, add input to digest
+				print STDERR '+';
+				link_files($file, $digest_link);
+			}
+
+			# A link to the inode indicates we are done, so do it last
+			link_files($file, $inode_link);
+
+		}
 	};
 	warn "$file: $@" if $@;
 }