X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=blobdiff_plain;f=dm6;h=c824c0e517832fe418d3fb8fc7cec2278d1102c4;hp=1f76cae0269a8d85a0899a5596aea10264ad2c0f;hb=3a2190b200435c28a3099695ef65d0d983dcf56e;hpb=f7dda48b96404d1abfa75b0290b42890caad4425 diff --git a/dm6 b/dm6 index 1f76cae..c824c0e 100755 --- a/dm6 +++ b/dm6 @@ -9,7 +9,7 @@ use File::Temp; use File::stat; use MIME::Base64; -# Copyright (C) 2010 Zygo Blaxell +# Copyright (C) 2010 Zygo Blaxell # 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 @@ -41,7 +41,7 @@ sub usage { Usage: $0 link-dir Hashes a NUL-separated list of files on stdin into link-dir. -Version: 0.20100514 +Version: 0.20100518 USAGE } @@ -58,12 +58,15 @@ sub link_files { print STDERR "\bL"; link($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 $saved_bang; print STDERR "\b"; } @@ -104,22 +107,18 @@ sub prepare_parents { return "$parent/$suffix"; } -sub name_quad { +sub name_ino { my ($int64) = @_; - my $packed = pack('Q', $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 $base64_packed; } -sub name_ino { - my ($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. @@ -175,7 +174,7 @@ while () { $digest =~ y:/:_:; # Check link to digest - my $digest_link = prepare_parents($link_dir, "${digest}D"); + my $digest_link = prepare_parents($link_dir, $digest); print STDERR 'D'; my $digest_st = lstat($digest_link); if ($digest_st) { @@ -220,8 +219,8 @@ while () { # Garbage collection print STDERR "\nGarbage collection in '$link_dir'..."; 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 files with link count < 3 and temporary links..."; +system('find . -type f \( -links -3 -o -name ".*" \) -print0 | xargs -0rt rm -f') and die "system: exit status $?"; print STDERR "\nRemoving empty directories..."; system("find . -type d -empty -print0 | xargs -0rt rmdir -p --ignore-fail-on-non-empty") and die "system: exit status $?"; print STDERR "\nDone.\n";