X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?p=dupemerge;a=blobdiff_plain;f=dm6;h=a88c83e1d17ba04a21b04fcff8762b48301a66da;hp=c6f93d906e94025621b230a6baa07e83dda44a5f;hb=de7efbca76dfc8ca745d8dabc2f9a99d9af0e4b4;hpb=5b34b57b3f1927f907fafbb70913b23fe6d36b12

diff --git a/dm6 b/dm6
index c6f93d9..a88c83e 100755
--- a/dm6
+++ b/dm6
@@ -7,8 +7,9 @@ use File::Compare;
 use File::Path;
 use File::Temp;
 use File::stat;
+use MIME::Base64;
 
-# Copyright (C) 2010 Zygo Blaxell <dm5@mailtoo.hungrycats.org>
+# Copyright (C) 2010 Zygo Blaxell <dupemerge@mailtoo.hungrycats.org>
 
 # 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
@@ -40,7 +41,7 @@ sub usage {
 Usage: $0 link-dir
 Hashes a NUL-separated list of files on stdin into link-dir.
 
-Version: 20100513.0
+Version: 0.20100514
 USAGE
 }
 
@@ -63,11 +64,11 @@ sub link_files {
 		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";
+	print STDERR "\b";
 }
 
 my $link_dir = shift @ARGV;
-(-d $link_dir) or usage;
+usage unless $link_dir;
 
 my $prefix_length = 3;
 
@@ -81,18 +82,48 @@ sub slash_prefix {
 	return ($prefix, $suffix);
 }
 
+sub mkdir_p {
+	my ($dir) = @_;
+	return if -d $dir;
+	$dir =~ s:/+$::os;
+	my $parent;
+	($parent = $dir) =~ s:[^/]+$::os;
+	if ($parent ne $dir) {
+		mkdir_p($parent);
+		print STDERR 'm';
+		mkdir($dir) or die "mkdir: $dir: $!";
+	}
+	die "mkdir: $dir: $!" unless -d $dir;
+}
+
 sub prepare_parents {
 	my ($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 $parent;
+	mkdir_p($parent);
 	return "$parent/$suffix";
 }
 
+sub name_quad {
+	my ($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.
 my $link_count_max = 31990;
 
 $/ = "\0";
@@ -114,7 +145,7 @@ while (<STDIN>) {
 			next if ($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;
@@ -145,7 +176,7 @@ 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) {
@@ -159,7 +190,7 @@ 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 ($digest_st) {
@@ -169,23 +200,33 @@ while (<STDIN>) {
 				# Old, replace input with old file
 				print STDERR '-';
 				link_files($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);
 
 		}
 	};
 	warn "$file: $@" if $@;
 }
 
+# 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 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";
+
 exit(0);
 
 __END__