dm6: bump version to 0.20101024, more comprehensive garbage collector

[dupemerge] / dm6

diff --git a/dm6 b/dm6
index 4cd11b02f43767f63a913cbaa7715c6fd553eecc..ff3a9eb2b57244eed3ac9c8d071ebf58c5afcbf9 100755 (executable)
--- a/dm6
+++ b/dm6
@@ -41,7 +41,7 @@ sub usage {
 Usage: $0 link-dir
 Hashes a NUL-separated list of files on stdin into link-dir.
 
-Version: 0.20100514
+Version: 0.20101024
 USAGE
 }
 
@@ -58,12 +58,15 @@ sub link_files {
        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";
-       }
+       my $saved_bang;
+       $saved_bang = $! unless rename($tmp_to, $to);
+
+       # If $to exists and is a hardlink to $tmp_to (or $from),
+       # rename returns success but $tmp_to still exists.
+       print STDERR "\bU";
+       unlink($tmp_to) or warn "unlink: $tmp_to: $!" if -e $tmp_to;
+
+       die "rename: $tmp_to -> $from: $saved_bang" if $saved_bang;
        print STDERR "\b";
 }
 
@@ -165,7 +168,6 @@ while (<STDIN>) {
                        # Compute digest
                        print STDERR 'd';
                        my $digest = digest($file);
-                       print STDERR "\b";
 
                        # Base64 uses /, we prefer _
                        $digest =~ y:/:_:;
@@ -214,13 +216,44 @@ while (<STDIN>) {
 }
 
 # Garbage collection
-print STDERR "\nGarbage collection in '$link_dir'...";
+print STDERR "\nGarbage collection in '$link_dir'...\n";
 chdir($link_dir) || die "chdir: $link_dir: $!";
-print STDERR "\nRemoving files with link count < 3...";
-system("find . -type f -links -3 -print0 | xargs -0rt rm -f") and die "system: exit status $?";
-print STDERR "\nRemoving empty directories...";
+
+my ($last_inode) = '';
+my @last_links;
+
+sub handle_gc_file {
+       my ($line) = @_;
+       my ($inode, $link) = ($line =~ /^(\S+) (.+)\0$/os);
+       $inode ||= '';
+       if ($inode ne $last_inode) {
+               my ($dev, $ino, $links) = ($last_inode =~ /^(\d+):(\d+):(\d+)$/os);
+               if (defined($links)) {
+                       if ($links && $links == @last_links) {
+                               print STDERR "rm -f @last_links\n";
+                               for my $unlink (@last_links) {
+                                       unlink($unlink) or warn "unlink: $unlink: $!";
+                               }
+                       }
+               } else {
+                       warn "Could not parse '$last_inode' in '$line'" unless $last_inode eq '';
+               }
+               @last_links = ();
+       }
+       $last_inode = $inode;
+       push(@last_links, $link);
+}
+
+print STDERR "Removing files contained entirely in '$link_dir'...\n";
+open(FIND, "find . -type f -printf '%D:%i:%n %p\\0' | sort -z --compress-program=gzip |") or die "open: find: $!";
+while (<FIND>) {
+       handle_gc_file($_);
+}
+handle_gc_file('');
+
+print STDERR "Removing empty directories...\n";
 system("find . -type d -empty -print0 | xargs -0rt rmdir -p --ignore-fail-on-non-empty") and die "system: exit status $?";
-print STDERR "\nDone.\n";
+print STDERR "Done.\n";
 
 exit(0);