X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dm6;h=b0ff96b84b633c717bf0516038f74e6ae2be494d;hb=a888bfab4a7685b300dbf8da06069dd3d2d75bf0;hp=b7a9cdf57f9beb1318bd2aab794e21ead1310362;hpb=fd37519dbb05cea1e1f8d20fbc944f9380f71f70;p=dupemerge

diff --git a/dm6 b/dm6
index b7a9cdf..b0ff96b 100755
--- a/dm6
+++ b/dm6
@@ -39,6 +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
 }
 
@@ -46,28 +48,34 @@ USAGE
 sub link_files {
 	my ($from, $to) = (@_);
 
-	print STDERR "link '$from' '$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 "\n";
+	print STDERR "\b \b";
 }
 
 my $link_dir = shift @ARGV;
 (-d $link_dir) or usage;
 
+my $prefix_length = 3;
+
 sub slash_prefix {
 	my ($file) = @_;
-	my $prefix = substr($file, 0, 3);
-	my $suffix = substr($file, 3);
+	$file .= '_' x (length($file) + 1 - $prefix_length) if length($file) + 1 < $prefix_length;
+	my $prefix = substr($file, 0, $prefix_length);
+	my $suffix = substr($file, $prefix_length);
 	$prefix =~ s:(.):$1/:osg;
 	chop($prefix);
 	return ($prefix, $suffix);
@@ -77,9 +85,11 @@ sub prepare_parents {
 	my ($link_dir, $file) = @_;
 	my ($prefix, $suffix) = slash_prefix($file);
 	my $parent = "$link_dir/$prefix";
-	mkpath($parent, { verbose => 1 });
+	print STDERR 'm';
+	mkpath($parent, { verbose => 0 });
+	print STDERR "\b";
 	die "mkpath: $parent: $!" unless -d $parent;
-	return "$parent/$prefix/$suffix";
+	return "$parent/$suffix";
 }
 
 # ext3 cannot handle more than 32000 links to a file.  Leave some headroom.
@@ -93,6 +103,7 @@ while (<STDIN>) {
 			chomp $file;
 
 			# Get file stat data
+			print STDERR '.';
 			my $st = lstat($file);
 			die "lstat: $file: $!" unless $st;
 
@@ -104,6 +115,7 @@ while (<STDIN>) {
 
 			# 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) {
@@ -119,26 +131,27 @@ while (<STDIN>) {
 			} 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 "digest($file) = ";
+			print STDERR 'd';
 			my $digest = digest($file);
+			print STDERR "\b";
 
 			# Base64 uses /, we prefer _
 			$digest =~ y:/:_:;
 
-			print STDERR "$digest\n";
-
 			# 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 "Removing '$digest_link' with $digest_nlink links\n";
+					print STDERR 'u';
 					unlink($digest_link) or die "unlink: $digest_link: $!";
 					undef $digest_st;
 				}
@@ -146,11 +159,15 @@ while (<STDIN>) {
 
 			# If digest link exists, link it to file
 			if ($digest_st) {
-				print STDERR "cmp '$digest_link' '$file' ...";
+				print STDERR 'c';
 				die "NOT identical!" if compare($digest_link, $file);
-				print STDERR "\n";
+
+				# Old, replace input with old file
+				print STDERR '-';
 				link_files($digest_link, $file);
 			} else {
+				# New, add input to digest
+				print STDERR '+';
 				link_files($file, $digest_link);
 			}