Merge branch 'performance'

[dupemerge] / faster-dupemerge

diff --git a/faster-dupemerge b/faster-dupemerge
index 6223ab0adc2d7cacb4dee7e7c4f4e71a2aa3a1d4..d35f591a442a9c7bb6c52ea160bed284825089b0 100755 (executable)
--- a/faster-dupemerge
+++ b/faster-dupemerge
@@ -4,7 +4,7 @@ use Fcntl qw(:DEFAULT :flock);
 use File::Compare;
 use File::Temp;
 
-# Copyright (C) 2003-2006 Zygo Blaxell <zblaxell@feedme.hungrycats.org>
+# Copyright (C) 2003-2009 Zygo Blaxell <faster-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
@@ -333,8 +333,9 @@ hash_file:
                if ($ok) {
                        print STDERR "\t\tDigest is $digest\n" if $debug;
 
-                       my $incumbent = $hash_to_inode{$digest};
-                       if (defined($incumbent)) {
+                       my $incumbent_list = ($hash_to_inode{$digest} ||= []);
+                       my $incumbent_matched = 0;
+                       for my $incumbent (@$incumbent_list) {
                                print STDERR "\t\tInodes $incumbent and $candidate have same hash\n" if $debug;
 
                                my $finished = 0;
@@ -344,7 +345,7 @@ link_start:
                                until ($finished) {
                                        my @incumbent_names = sort keys(%{$inode_to_file_name{$incumbent}});
                                        my @candidate_names = sort keys(%{$inode_to_file_name{$candidate}});
-                                       print STDERR "\t\tLinks to $incumbent:",   join("\n\t\t\t", '', @incumbent_names),   "\n" if $debug;
+                                       print STDERR "\t\tLinks to $incumbent:", join("\n\t\t\t", '', @incumbent_names), "\n" if $debug;
                                        print STDERR "\t\tLinks to $candidate:", join("\n\t\t\t", '', @candidate_names), "\n" if $debug;
 
 incumbent_file:
@@ -411,6 +412,7 @@ candidate_file:
                                                                                print STDERR "$quoted_incumbent_file and $quoted_candidate_file have same hash but do not compare equal!\n" unless $skip_hashes;
                                                                        } else {
                                                                                $identical = 1;
+                                                                               $incumbent_matched = 1;
                                                                        }
                                                                        $compare_count++;
                                                                        $compare_bytes += $incumbent_size;
@@ -493,13 +495,9 @@ candidate_file:
                                                                        # My random number generator chooses the incumbent's size.
 
                                                                        if ($link_done) {
-                                                                               # Since we're in a dry run, the filesystem doesn't change.
-                                                                               # Our notion of what the filesystem should look like should not change either.
                                                                                delete $inode_to_file_name{$to_inode}->{$to_file};
-                                                                               unless ($dry_run) {
-                                                                                       $inode_to_file_name{$from_inode}->{$to_file} = undef;
-                                                                                       $hash_to_inode{$digest} = $from_inode;
-                                                                               }
+                                                                               $inode_to_file_name{$from_inode}->{$to_file} = undef;
+                                                                               $hash_to_inode{$digest} = [ $from_inode ];
 
                                                                                $hard_links++;
                                                                                if ($to_nlink == 1) {
@@ -527,9 +525,10 @@ candidate_file:
                                        }
                                        $finished = 1;
                                }
-                       } else {
+                       }
+                       unless ($incumbent_matched) {
                                print STDERR "\t\tNew hash entered\n" if $debug;
-                               $hash_to_inode{$digest} = $candidate;
+                               push(@$incumbent_list, $candidate);
                        }
                } else {
                        warn "No digests found for inode $candidate\n";
@@ -544,7 +543,7 @@ end_merge:
 }
 
 while (<FIND>) {
-       my ($weak_key, $inode, $name) = m/^(\d+ \d+ \d+ \d+ -?\d+) (\d+:\d+) (.+)\0$/so;
+       my ($weak_key, $inode, $name) = m/^(\d+ \d+ \d+ \d+ -?[\d.]+) (\d+:\d+) (.+)\0$/so;
        die "read error: $!\nLast input line was '$_'" unless defined($name);
 
        print STDERR "weak_key=$weak_key inode=$inode name=$name\n" if $debug;