#!/usr/bin/perl
# This script is written to find file copies
# It generates report and suggest to choose which copy to delete						
# Jes's scripts, copyleft			
##################################################################
use strict;
use Time::HiRes qw(gettimeofday tv_interval);
use File::Compare;
use File::Find;
use Term::ANSIColor;
my $VERSION = '1.0';
use Getopt::Long qw(:config no_ignore_case);

	my @bb = gettimeofday; 	# script's processing begin
	my (@files, @tfiles);
	my ($prefix, $help, $greps, $mode, $report);
	$| = 1;
	GetOptions("d=s" => \$prefix, "h" => \$help, "g=s"=>\$greps, "s"=>\$mode, "f=s"=>\$report);
		unless($prefix or $help) {
			print "enter path(s) to start (comma separated if several):";
			while(<STDIN>){
				chomp;		$prefix = $_;		last;
			}
		}
		unless($report or $help) {
			print "enter path to report file or I'll just write the compare results to STDOUT:";
			while(<STDIN>){
				chomp;		$report = $_;		last;
			}
		}
	print "$report" and print color("red"), " file is a directory!\n", color("reset") and exit if (-d $report);
	my @paths = split (/\,/, $prefix) if ($prefix);
	if ($prefix) {
		foreach my $p (@paths) {
			print "$p" and print color("red"), " is not a directory!\n", color("reset") and exit unless (-d $p);
		}
	}
	print "the directory to start: $prefix\n" unless ($help);
	print "gathering from directory $prefix ...\n" unless ($help);
	print color("red"), "\nyou should enter directory to start...\n", color("reset") unless($prefix);
	&usage unless($prefix);

	@ARGV = grep {"$_"} @paths unless @ARGV;
	my @files;
	find sub {push @files, $File::Find::name}, @ARGV;			# files gathering
	foreach my $f (@files) {									# grep files
			if ($greps) {
				my @greps = split(/\,/,$greps);
				push (@tfiles,$f) unless (grep {($f =~/\.$_/)} @greps or -d $f);
			} else {
				push (@tfiles,$f) unless (-d $f);
			}
	}
	my $num = scalar @tfiles;
	print color("blue"), "\n$num files are gathered...\nstarting to compare them...\n", color("reset");
	my @pairs;
	my $ask;
	unless ($mode) {
		print "should I use interactive mode (I'll ask about files deleting) [Y/N]?\n";
		while(<STDIN>){
			chomp;		$ask = $_;		last;
		}
	}

	my $size = 0;			# size of identical files
	my $freesize = 0;		# cleared space on the disk
	open (REP, ">$report") or print color("red"), "path to report file $report is incorrect!\n", color("reset") and exit if ($report);
	foreach my $f1 (@tfiles) {
		my @arr = grep {compare($f1,$_) == 0} @tfiles;			# compare files
		my @checks = grep {$_ eq @arr[0].".".@arr[1]} @pairs if (scalar @arr ==2);
		push (@pairs, @arr[0].".".@arr[1]) if (scalar @arr ==2);
		my $fs;
		if (scalar @arr ==2 and !@checks) {
			print REP "@arr are identical\n" if ($report);
			$fs = -s "@arr[0]";
			$fs = $fs/1000;
			print "@arr" and print color("cyan")," are identical, \ntheir size is: ",color("reset") and print "2 * $fs Kb\n";
		}
		unless($mode) {
			if (scalar @arr ==2 and !@checks and $ask =~/(y|Y)/) {
				print color("green"), "should I delete first file [1], or second [2], or should I do nothing [enter]?\n", color("reset");
				my $action;
				while(<STDIN>){
					chomp;		$action = $_;		last;
				}
				my $del = ($action == 2)?@arr[1]:($action == 1)?@arr[0]:"";
				if ($del) {
					print color("blue"), "deleting $del...", color("reset");
					unlink $del;
					print color("blue"), "done!\n", color("reset");
					print REP "!!! $del is deleted!\n";
					$freesize += $fs;
				}
			}
		}
		$size += $fs;
	}
	my @ee = gettimeofday;	# script's processing end
	my $s = sprintf ("%.2f", tv_interval (\@bb,\@ee));
	print color("cyan"), "found no clones!\n", color("reset") unless @pairs;
	print REP "in $prefix found no clones!\n" if (!@pairs and $report);
	print color("blue"), "overall size of clones is $size Kb\n", color("reset") if ($size != 0);
	print REP "\noverall size of clones is $size Kb\n" if ($report and $size != 0);
	print color("blue"), "released disk space by clone's deleting is $freesize Kb\n", color("reset") if ($size != 0);
	print REP "released disk space by clone's deleting is $freesize Kb\n" if ($report and $size != 0);
	print color("blue"), "done for $s sec...\n", color("reset");
	print REP "\ndone for $s sec...\n" if ($report);
	close REP if ($report);
	&usage if($help);
####	OPTIONS		####
sub usage() {
	print <<EOF;

=========================
compare.pl version $VERSION
=========================

This script can be executed with options:
-d [comma separated paths for checking]
-g [comma separated grep patterns for file types = do not compare these file types]
-s silent mode (just printing a report to STDOUT or see -f option)
-f [path to report file (file name)]
-h this help
Or, you can use this script without options at all: just answer the STDOUT's questions.
Good luck!
EOF
    exit;
}