Refactor, version bumpv2.1

Do initial size filtering and collection in the wanted sub, and delay
inode checking until filtering each size

1 files changed, 29 insertions, 34 deletions

diff --git a/checkem b/checkem
index 1aa05db..4f8fa84 100755
--- a/checkem
+++ b/checkem
@@ -19,15 +19,15 @@ use Digest::SHA;
 use 5.009003;
 
 # Version number to make Perl::Critic happy
-our $VERSION = 2.0;
+our $VERSION = 2.1;
 
 # If no arguments, work with the current working directory
 croak('Need at least one dir to search') if !@ARGV;
 
-# Start an array of filenames...
-my @fns;
+# Start a hash of filesizes to file names/stats...
+my %sizes;
 
-# ... and fill it up with File::Find.
+# ...and fill it up with File::Find.
 find {
     no_chdir => 1,
     wanted   => sub {
@@ -35,49 +35,44 @@ find {
         ## no critic (ProhibitFiletest_f)
         return if !-f $fn;
         return if -l $fn;
-        push @fns, $fn;
+        my $st = stat $fn or return;
+        return if !$st->size;
+        push @{ $sizes{ $st->size } },
+          {
+            fn => $fn,
+            st => $st,
+          };
         return;
     },
 }, @ARGV;
 
-# Start a table of device and inode to detect hard links, to completely skip
-# any files we've already seen
-my %inodes;
-
-# Build a table of filesizes to list of filenames with that size
-my %sizes;
-FN: for my $fn (@fns) {
-
-    # Get file metadata
-    my $st = stat $fn or next FN;
+# If there's more than one filename of any of the sizes, look for hard links,
+# checksum them if not linked, and push them into a sums table
+my ( %sums, $dig );
+FS: for my $fs ( grep { @{$_} > 1 } values %sizes ) {
 
-    # Skip the file if it's empty
-    next FN if !$st->size;
+    # Keep a temporary table of inodes to catch hard links
+    my %inos;
 
-    # Try to get a device ID and inode, so we can ignore hard-linked files
-    # using the inodes lookup table if we've seen that tuple before
-    if ( $st->dev && $st->ino ) {
-        next FN if exists $inodes{ $st->dev }{ $st->ino };
-        $inodes{ $st->dev }{ $st->ino } = 1;
-    }
+    # Iterate through each file in the list
+  F: for my $f ( @{$fs} ) {
 
-    # Add the filename to its size bucket
-    push @{ $sizes{ $st->size } }, $fn;
-}
+        # Catch hard links on compliant systems by keeping a dev/inode hash
+        my ( $dev, $ino ) = ( $f->{st}->dev, $f->{st}->ino );
+        if ( $dev && $ino ) {
+            next F if exists $inos{$dev}{$ino};
+            $inos{$dev}{$ino} = $f;
+        }
 
-# If there's more than one filename of any of the sizes, checksum them and push
-# them into a sums table
-my ( %sums, $dig );
-SIZE: for my $fns ( grep { @{$_} > 1 } values %sizes ) {
-    for my $fn ( @{$fns} ) {
+        # Files still the same size and not hard linked, group by digest
         $dig //= Digest::SHA->new('sha256');
-        $dig->addfile($fn);
-        push @{ $sums{ $dig->digest() } }, $fn;
+        $dig->addfile( $f->{fn} );
+        push @{ $sums{ $dig->digest() } }, $f;
     }
 }
 
 # Print the groups of matched files (more than one share a checksum in the
 # final table)
 for my $group ( grep { @{$_} > 1 } values %sums ) {
-    printf "%s\n\n", join "\n", @{$group};
+    printf "%s\n\n", join "\n", map { $_->{fn} } @{$group};
 }