perl-scripts/Finders/fdf

#!/usr/bin/perl

# Daniel "Trizen" Șuteu
# License: GPLv3
# Date: 01 January 2012
# Edit: 24 August 2024
# https://github.com/trizen

# Find and list duplicate files from one or more paths, with options for
# deleting or replacing duplicate files with symbolic links to the main file.

use 5.005;
use strict;
use warnings;

use File::Find     qw(find);
use File::Compare  qw(compare);
use File::Basename qw(basename);
use Getopt::Long   qw(GetOptions);

my %order_callbacks = (
    path => sub { sort @_ },
    name => sub {
        map  { $_->[1] }
        sort { $a->[0] cmp $b->[0] }
        map  { [basename($_), $_] } @_;
    },
    time => sub {
        map  { $_->[1] }
        sort { $a->[0] <=> $b->[0] }
        map  { [-M $_, $_] } @_;
    },
);

my @dirs = grep { (-d) or (-f) } @ARGV;
die <<"HELP" if !@dirs;
usage: $0 [options] /my/path [...]

Options:
  -f, --first      : keep only the first duplicated file
  -l, --last       : keep only the last duplicated file
  -s, --symlink    : replace duplicate files with symbolic links (with -f or -l)
  -o, --order=type : order the results by: path, name or time
  -m, --min-size=i : minimum size in bytes (default: 0)

HELP

my $keep_first;
my $keep_last;
my $create_symlinks;
my $order_by = 'time';
my $min_size = 0;

GetOptions(
           'f|first!'           => \$keep_first,
           'l|last!'            => \$keep_last,
           's|symlink!'         => \$create_symlinks,
           'o|order|order-by=s' => \$order_by,
           'm|min-size=i'       => \$min_size,
          )
  or die("$0: error in command line arguments\n");

if (not exists $order_callbacks{$order_by}) {
    local $" = ", ";
    die "$0: invalid value `$order_by` for `--order`: valid values are: @{[sort keys %order_callbacks]}\n";
}

sub find_duplicated_files (&@) {
    my $callback = shift;

    my %files;
    find {
        no_chdir => 1,
        wanted   => sub {
            lstat;
            (-f _) && (not -l _) && ((-s _) >= $min_size) && push @{$files{-s _}}, $_;
        }
    } => @_;

    foreach my $files (values %files) {

        next if $#{$files} < 1;

        my %dups;
        foreach my $i (0 .. $#{$files} - 1) {
            for (my $j = $i + 1 ; $j <= $#{$files} ; $j++) {
                if (compare($files->[$i], $files->[$j]) == 0) {
                    push @{$dups{$files->[$i]}}, splice @{$files}, $j--, 1;
                }
            }
        }

        while (my ($fparent, $fdups) = each %dups) {
            $callback->($order_callbacks{$order_by}($fparent, @{$fdups}));
        }
    }

    return;
}

{
    local $, = "\n";
    local $\ = "\n";

    find_duplicated_files {
        my (@files) = @_;

        print @files, "-" x 80;

        my $main_file = (
                           $keep_first ? shift(@files)
                         : $keep_last  ? pop(@files)
                         :               return
                        );

        foreach my $file (@files) {

            print ":: Removing: `$file`";

            unlink($file) or do {
                warn "error: can't delete file `$file': $!\n";
                next;
            };

            if ($create_symlinks) {
                print ":: Symlinking: `$main_file` <- `$file`";
                symlink($main_file, $file) or do {
                    warn "error: can't create symbolic link for `$file': $!\n";
                    next;
                };
            }
        }
    } @dirs;
}