From e5922e46abc2ef23ea0354a38fdeff1041aedd6f Mon Sep 17 00:00:00 2001
From: Guillermo Ramos
Date: Tue, 22 Aug 2023 20:14:16 +0200
Subject: dups: WIP rework

---
 dups | 101 +++++++++++++++++++++++++++++++++----------------------------------
 1 file changed, 49 insertions(+), 52 deletions(-)

diff --git a/dups b/dups
index bf59ee9..4f38482 100755
--- a/dups
+++ b/dups
@@ -3,8 +3,11 @@
 use strict;
 use warnings;
 
+use v5.36;
+
 use List::Util qw<min>;
 use File::Find qw<find>;
+
 use IPC::System::Simple qw<capture>;
 
 @ARGV ge 1 || die "Usage: $0 [opts] <dir1> [<dir2> ...]
@@ -19,45 +22,40 @@ my $yes = 0;
 my $debug = 0;
 my @preferences;
 
+# Map from size to list of filenames
 my %sizes;
+
+# Map from md5 to list of filenames
 my %md5s;
 
-sub fill_sizes {
+sub register_file_size {
     return if -d "$_"; # Skip directories
 
+    my $filename = $File::Find::name;
     my $size = capture("stat", "--printf=%s", $_);
     $sizes{$size} ||= [];
-    push @{$sizes{$size}}, $File::Find::name;
+    push @{$sizes{$size}}, $filename;
 }
 
-sub closeness {
-    my ($file, $dir) = @_;
+sub indir($file, $dir) {
     my ($diffpath) = Cwd::abs_path($file) =~ /^$dir(.*)$/;
-    if ($diffpath) {
-        return () = $diffpath =~ /\//g;
-    } else {
-        return -1;
-    }
+    return !!$diffpath;
 }
 
-sub guess_best_choices {
-    my @best_guesses = @_;
+sub guess_best_choices(@files) {
+    my @best_guesses = ();
 
+    # Remove some choice based on the 'dislike' preference
     foreach my $pref (@preferences) {
         if ($pref->{"type"} eq "dislike") {
             print STDERR "PREF: DISLIKE $pref->{dir}\n" if $debug;
-            my $best_index = 0;
-            my @files = @best_guesses;
             foreach my $file (@files) {
-                my $dislikability = closeness($file, $pref->{"dir"});
-                print STDERR "CLOSENESS($file, $pref->{dir}) = $dislikability\n" if $debug;
-                if ($best_index != -1 && ($dislikability == -1 || $dislikability > $best_index)) {
-                    @best_guesses = ($file);
-                    $best_index = $dislikability;
-                } elsif ($dislikability == $best_index) {
+                if (indir($file, $pref->{"dir"})) {
+                    print STDERR "Discarding $file...\n" if $debug;
+                } else {
                     push @best_guesses, $file;
                 }
-                print STDERR "BEST: [$best_index] @best_guesses\n" if $debug;
+                print STDERR "BEST: @best_guesses\n" if $debug;
             }
         }
     }
@@ -65,8 +63,7 @@ sub guess_best_choices {
     return @best_guesses;
 }
 
-sub keep {
-    my ($choice, @rest) = @_;
+sub keep($choice, @rest) {
     my @delete = grep { $_ ne $choice } @rest;
     print STDERR "\tKEEPING $choice, DELETING @delete\n" if $debug;
     unlink foreach @delete;
@@ -81,7 +78,7 @@ while (my $arg = shift) {
     } elsif ($arg eq "-d" || $arg eq "--dislike") {
         push @preferences, {"type" => "dislike", "dir" => Cwd::abs_path(shift) . "/"};
     } else {
-        find(\&fill_sizes, $arg);
+        find(\&register_file_size, $arg);
     }
 }
 
@@ -115,35 +112,35 @@ foreach my $md5 (keys(%md5s)) {
         print "\t$file\n";
     }
 
-    if (my @best_choices = guess_best_choices(@same_md5_files)) {
-        if (@best_choices == 1) {
-            my $best_choice = $best_choices[0];
-            my $gogo = $yes;
-            unless ($gogo) {
-                print "\n\tBest choice is '$best_choice', do you want to delete the others? (yes/y) ";
-                my $resp = <STDIN>;
-                chomp $resp;
-                $gogo = $resp eq "yes" || $resp eq "y";
-            }
-            if ($gogo) {
-                keep($best_choice, @same_md5_files);
-            }
-        } else {
-            print "\n\tPlease choose one to keep (or press 'enter' to skip):\n";
-            for (my $i = 0; $i < @best_choices; $i++) {
-                my $choice = $best_choices[$i];
-                print "\t  [$i] $choice\n";
-            }
-            print "\t> ";
-            my $index = <STDIN>;
-            chomp $index;
-            if ($index ne "") {
-                $index = int($index);
-                if ($index >= 0 && $index < @best_choices) {
-                    keep($best_choices[$index], @same_md5_files);
-                } else {
-                    print "\n!!\t Index outside of range, ignoring\n";
-                }
+    my @best_choices = guess_best_choices(@same_md5_files);
+    @best_choices = @same_md5_files unless @best_choices;
+    if (@best_choices == 1) {
+        my $best_choice = $best_choices[0];
+        my $gogo = $yes;
+        unless ($gogo) {
+            print "\n\tBest choice is '$best_choice', do you want to delete the others? (yes/y) ";
+            my $resp = <STDIN>;
+            chomp $resp;
+            $gogo = $resp eq "yes" || $resp eq "y";
+        }
+        if ($gogo) {
+            keep($best_choice, @same_md5_files);
+        }
+    } else {
+        print "\n\tPlease choose one to keep (or press 'enter' to skip):\n";
+        for (my $i = 0; $i < @best_choices; $i++) {
+            my $choice = $best_choices[$i];
+            print "\t  [$i] $choice\n";
+        }
+        print "\t> ";
+        my $index = <STDIN>;
+        chomp $index;
+        if ($index ne "") {
+            $index = int($index);
+            if ($index >= 0 && $index < @best_choices) {
+                keep($best_choices[$index], @same_md5_files);
+            } else {
+                print "\n!!\t Index outside of range, ignoring\n";
             }
         }
     }
-- 
cgit v1.2.3