#!/usr/bin/perl
use strict;
use warnings;
use File::Spec;
use File::Path qw(remove_tree);
use File::Find;
use Cwd qw(getcwd abs_path);

my ($files, $depth, $total_commits) = @ARGV;
if (!defined $files || !defined $depth || !defined $total_commits) {
	die "Usage: perl bm_history.pl <file_count> <depth> <total_commits>\n";
}

my $base_dir = getcwd();
my $urn_bin  = abs_path(File::Spec->catfile("..", "urn")); 
my $seed_bin = abs_path("seed.pl");
my $bm_repo  = "bm_repo";

my $sample_rate = int($total_commits / 5) || 1;
my %results;

sub get_size {
	my $dir = shift;
	return "0 KB" unless -d $dir;
	my $size = `du -sk $dir 2>/dev/null`;
	$size =~ /^(\d+)/;
	return ($1 || 0) . " KB";
}

sub count_inodes {
	my $dir = shift;
	return 0 unless -d $dir;
	my $count = `find $dir 2>/dev/null | wc -l`;
	$count =~ s/\s+//g;
	return $count || 0;
}

sub run_history_benchmark {
	my ($tool_name) = @_;
	print ">>> Starting History Benchmark: $tool_name\n";

	remove_tree($bm_repo) if -d $bm_repo;
	system("perl $seed_bin $files $depth > /dev/null 2>&1") == 0 
		or die "FATAL: seed.pl failed.\n";
	
	chdir($bm_repo) or die "FATAL: Could not enter $bm_repo: $!\n";

	my $init_cmd = ($tool_name eq "URN") ? "perl $urn_bin init" : "git init";
	system("$init_cmd > /dev/null 2>&1");

	my @file_list;
	find(sub {
		push @file_list, $File::Find::name if -f $_ && $File::Find::name !~ m{\.(git|urn)};
	}, ".");

	my $count = scalar(@file_list);
	die "FATAL: No files found in $bm_repo after seeding!" if $count == 0;

	# Use explicit add + commit to ensure untracked files are caught
	my $initial_cmd = ($tool_name eq "URN") ? "perl $urn_bin add . && perl $urn_bin commit -m 'initial'" 
	                                        : "git add . && git commit -m 'initial'";
	system("$initial_cmd > /dev/null 2>&1");

	for my $i (1 .. $total_commits) {
		my $to_modify = int($files * 0.02) || 1;
		for (1 .. $to_modify) {
			my $target = $file_list[rand @file_list];
			if (open(my $fh, '>>', $target)) {
				print $fh "Commit $i: Mod for $tool_name test.\n";
				close($fh);
			}
		}

		my $msg = "commit_$i";
		my $run_cmd = ($tool_name eq "URN") ? "perl $urn_bin add . && perl $urn_bin commit -m '$msg'" 
		                                    : "git add . && git commit -m '$msg'";
		
		if ($i % $sample_rate == 0 || $i == $total_commits) {
			# Run add and commit together inside the time wrapper
			my $raw = `(/usr/bin/time -l sh -c "$run_cmd" > /dev/null) 2>&1`;
			
			my ($real) = $raw =~ /(\d+\.\d+)\s+real/;
			my ($rss)  = $raw =~ /(\d+)\s+maximum resident set size/;
			my ($maj)  = $raw =~ /(\d+)\s+page faults caused by physical I\/O/;
			my ($min)  = $raw =~ /(\d+)\s+page reclaims by virtual memory/;

			my $meta = ($tool_name eq "URN") ? ".urn" : ".git";
			$results{$i}{$tool_name} = {
				real   => $real // "0.00",
				rss    => $rss ? sprintf("%.2f MB", $rss / 1024 / 1024) : "0.00 MB",
				faults => sprintf("Maj:%d/Min:%d", $maj // 0, $min // 0),
				inodes => count_inodes($meta),
				size   => get_size($meta),
			};
			print "   [Commit $i] $tool_name sampled.\n";
		} else {
			system("$run_cmd > /dev/null 2>&1");
		}
	}
	chdir($base_dir);
}

run_history_benchmark("URN");
run_history_benchmark("GIT");

my $out_file = "BM_HISTORY_${files}_${total_commits}.txt";
open(my $res, '>', $out_file) or die $!;
print $res "=============================================================\n";
print $res " HISTORY BENCHMARK: $files files ($total_commits commits)\n";
print $res "=============================================================\n\n";

foreach my $i (sort { $a <=> $b } keys %results) {
	my $u = $results{$i}{"URN"};
	my $g = $results{$i}{"GIT"};
	print $res "SNAPSHOT: Commit #$i\n";
	print $res "-------------------------------------------------------------\n";
	printf $res "%-15s | %-20s | %-20s\n", "METRIC", "URN", "GIT";
	print $res "----------------+----------------------+---------------------\n";
	printf $res "%-15s | %20s | %20s\n", "Time", $u->{real}."s", $g->{real}."s";
	printf $res "%-15s | %20s | %20s\n", "Max RSS", $u->{rss}, $g->{rss};
	printf $res "%-15s | %20s | %20s\n", "Page Faults", $u->{faults}, $g->{faults};
	printf $res "%-15s | %20s | %20s\n", "Inodes", $u->{inodes}, $g->{inodes};
	printf $res "%-15s | %20s | %20s\n", "Repo Size", $u->{size}, $g->{size};
	print $res "-------------------------------------------------------------\n\n";
}
close($res);

print "\nRun complete! Opening results...\n";
system($ENV{PAGER} || 'less', $out_file);