#!/usr/local/bin/perl -w

use strict;
use lib "/usr/local/BackupPC/lib";

use DBI;
use BackupPC::Lib;
use BackupPC::View;
use BackupPC::Attrib qw/:all/;
use Data::Dump qw(dump);
use Time::HiRes qw/time/;
use POSIX qw/strftime/;
use Cwd qw/abs_path/;
use Archive::Tar::Streamed;
use Algorithm::Diff;
use File::Slurp;
use Getopt::Long::Descriptive;

=head1 NAME

BackupPC_ASA_PostArchive_Update

=head1 DESCRIPTION

	# /etc/BackupPC/pc/_search_archive.pl

=cut


my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
$bpc->ConfigRead('_search_archive'); # read our configuration
my %Conf = $bpc->Conf();

use BackupPC::Search;
%BackupPC::Search::Conf = %Conf;

my ($opt,$usage) = describe_options(
"%c %o",
[ 'host|h=s@',	"import just host(s)" ],
[ 'num|n=s@',	"import just backup number(s)" ],
[ 'ok=n',       "xferOK", { default => 0 } ],
[ 'check|c',	"check archives on disk and sync", { default => 1 } ],
[ 'debug|d',	"debug", { default => 1 } ],
[ 'help',	"show help" ],
);

print($usage->text), exit if $opt->help;

$|=1;

my $start_t = time();

my $t_fmt = '%Y-%m-%d %H:%M:%S';

#warn "## Conf = ",dump( \%Conf );

my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });

#---- subs ----


sub curr_time {
	return strftime($t_fmt,localtime());
}

sub fmt_time {
	my $t = shift || return;
	my $out = "";
	my ($ss,$mm,$hh) = gmtime($t);
	$out .= "${hh}h" if ($hh);
	$out .= sprintf("%02d:%02d", $mm,$ss);
	return $out;
}

my $hsn_cache;

sub get_backup_id($$) {
	my ($host, $num) = @_;

	my $key = "$host $num";
	return $hsn_cache->{$key} if ($hsn_cache->{$key});

	# all backup parts will be attached to first share in backups
	my $sth = $dbh->prepare(qq{
		SELECT 
			min(backups.id)
		FROM backups 
		INNER JOIN shares	ON backups.shareID=shares.ID
		INNER JOIN hosts	ON backups.hostID = hosts.ID
		WHERE hosts.name = ? and backups.num = ?
	});
	$sth->execute($host, $num);
	die "can't find backup $host:$num" unless $sth->rows == 1;
	my ($id) = $sth->fetchrow_array;

	$hsn_cache->{$key} = $id;

	print STDERR "# $key == $id\n" if $opt->debug;

	return $id;
}

sub backup_inc_deleted($) {
	my $backup_id = shift;
	my $sth_inc_deleted = $dbh->prepare(qq{
		update backups set
			inc_deleted = true
		where id = ?
	});
	$sth_inc_deleted->execute($backup_id);
}

sub system_ok {
	warn "## system_ok @_\n";
	system(@_) == 0 || die "system @_:$!";
}

my $sth_inc_size = $dbh->prepare(qq{
	update backups set
		inc_size = ?,
		parts = ?,
		inc_deleted = false
	where id = ?
});

sub read_only {
	my $full = shift;
	my $perm = (stat $full)[2] & 0444;
	warn sprintf("chmod %03o %s\n",$perm,$full);
	chmod $perm, $full || die $!;
}

sub check_archive {
	my ($host,$num) = @_;
	warn "# check_archive $host $num";

	my $t = time();

	my $glob = "$Conf{ArchiveDest}/$host.$num.*";

	my @tar_parts = sort map { s/^\Q$Conf{ArchiveDest}\E\/*//; $_ } glob $glob ;

	if ( ! @tar_parts ) {
		warn "ERROR: no files for $glob";
		return;
	}

	print curr_time, " check $host $num\n";

	my $sth = $dbh->prepare(qq{
		SELECT count(*)
		FROM files
		JOIN shares on shares.id = shareid
		JOIN hosts on hosts.id = shares.hostid
		WHERE hosts.name = ? and backupnum = ?
	});
	$sth->execute($host, $num);
	my ($files) = $sth->fetchrow_array;

	if ( $files == 0 ) {
		warn "EMPTY INCREMENT, cleanup ",dump( @tar_parts );
		foreach my $path ( @tar_parts ) {
			my $full = "$Conf{ArchiveDest}/$path";
			warn "rm $full\n";
			unlink $full || die "can't remove $full: $!";
		}
		return;
	}

	my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
	unlink $md5_path if -e $md5_path && -s $md5_path == 0; # fix empty

	my $read_protect = 0;

	if ( ! -e $md5_path ) {
		system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
		read_only $md5_path;
		$read_protect = 1;
	} else {
		system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum -c $md5_path" if $opt->check;
	}

	my $md5sum;
	foreach ( split(/\n/, read_file "$Conf{ArchiveDest}/$host.$num.md5" ) ) {
		my ( $md5, $path ) = split(/\s+/,$_);
		$md5sum->{$path} = $md5;
		read_only "$Conf{ArchiveDest}/$path" if $read_protect;
	}

	# depending on expected returned value this is used like:
	# my $uncompress_size = get_gzip_size('/full/path/to.gz');
	# my ($compress_size, $uncompress_size) = get_gzip_size('/path.gz');
	sub get_gzip_size($) {
		my $filename = shift;
		die "file $filename problem: $!" unless (-r $filename);

		if ( $filename !~ m/\.gz$/ ) {
			return -s $filename;
		}

		open(my $gzip, $Conf{GzipPath}." -l $filename |") || die "can't gzip -l $filename: $!";
		local $/ = undef;
		my $line = <$gzip>;
		close($gzip);

		my ($comp, $uncomp) = (0,0);

		if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
			if (wantarray) {
				return [ $1, $2 ];
			} else {
				return $2;
			}
		} else {
			warn "ERROR can't parse: $line";
			return -s $filename;
		}
	}

	sub check_part {
		my ($host, $num, $part_nr, $tar_size, $size, $md5, $items, $filename) = @_;
		my $backup_id = get_backup_id($host, $num);
		my $sth_md5 = $dbh->prepare(qq{
			select
				id, tar_size, size, md5, items, filename
			from backup_parts
			where backup_id = ? and part_nr = ? and filename = ?
		});

		$sth_md5->execute($backup_id, $part_nr, $filename);

		if (my $row = $sth_md5->fetchrow_hashref) {
			return if (
				$row->{tar_size} >= $tar_size &&
				$row->{size} == $size &&
				$row->{md5} eq $md5 &&
				$row->{items} == $items
			);
			print ", deleting invalid backup_parts $row->{id}";
			$dbh->do(qq{ delete from backup_parts where id = $row->{id} });
		}
		print ", inserting new";
		my $sth_insert = $dbh->prepare(qq{
			insert into backup_parts (
				backup_id,
				part_nr,
				tar_size,
				size,
				md5,
				items,
				filename
			) values (?,?,?,?,?,?,?)
		});

		$sth_insert->execute($backup_id, $part_nr, $tar_size, $size, $md5, $items, $filename);
	}

	print " [parts: ",join(", ", @tar_parts),"]" if $opt->debug;

	my @tar_files;

	my $backup_part;

	print " reading" if $opt->debug;

	my $part_nr = 0;
	my $inc_size = 0;

	foreach my $filename (@tar_parts) {

		next if $filename eq "$host.$num.md5";

		print "\n\t- $filename";

		my $path = "$Conf{ArchiveDest}/$filename";
		$path =~ s{//+}{/}g;

		my $size = (stat( $path ))[7] || die "can't stat $path: $!";

		if ($size > $Conf{ArchiveMediaSize}) {
			print ", part bigger than media $size > $Conf{ArchiveMediaSize}\n";
			return 0;
		}

		print ", $size bytes";

=for later

		open(my $fh, "gzip -cd $path |") or die "can't open $path: $!";
		binmode($fh);
		my $tar = Archive::Tar::Streamed->new($fh);

		my $tar_size_inarc = 0;
		my $items = 0;

		while(my $entry = $tar->next) {
			push @tar_files, $entry->name;
			$items++;
			$tar_size_inarc += $entry->size;

			if ($tar_size_inarc > $Conf{ArchiveMediaSize}) {
				print ", part $filename is too big $tar_size_inarc > $Conf{ArchiveMediaSize}\n";
				return 0;
			}

		}

		close($fh);

		print ", $items items";

		if ($tar_size_inarc == 0 && $items == 0) {
			print ", EMPTY tar\n";

			my $backup_id = get_backup_id($host, $share, $num);
			backup_inc_deleted( $backup_id );

			$dbh->commit;

			return 1;
		}

=cut

		# FIXME
		my $tar_size = get_gzip_size( $path );

		#
		# finally, check if backup_parts table in database is valid
		#

		my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
		my $items = 1;
		$part_nr++;

		check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items, $filename);

		# round increment size to 2k block size
		$inc_size += int((($size + 2048) / 2048 ) * 2048);
	}

	$sth_inc_size->execute(
		$inc_size,
		$part_nr,
		get_backup_id($host, $num),
	);

	warn "## commit\n";
	$dbh->commit;

	return;

=for removed

	@tar_files = sort @tar_files;
	print "\n\t",($#tar_files + 1), " tar files";

	my $sth = $dbh->prepare(qq{
		SELECT path,type
		FROM files
		JOIN shares on shares.id = shareid
		JOIN hosts on hosts.id = shares.hostid
		WHERE hosts.name = ? and backupnum = ?
	});
	$sth->execute($host, $num);
	my @db_files;
	while( my $row = $sth->fetchrow_hashref ) {

		my $path = $row->{'path'} || die "no path?";
		$path =~ s#^/#./#;
		$path .= '/' if ($row->{'type'} == BPC_FTYPE_DIR);
		push @db_files, $path;
	}

	print " ",($#db_files + 1), " database files, diff";

	@db_files = sort @db_files;

	my $same = 1;

	if ($#tar_files != $#db_files) {
		$same = 0;
		print " NUMBER";
	} else {
		my $diff = Algorithm::Diff->new(\@tar_files, \@db_files);
		while ( $diff->Next() ) {
			next if $diff->Same();
			$same = 0;
			print "< $_\n" for $diff->Items(1);
			print "> $_\n" for $diff->Items(2);
		}
	}

	print " ",($same ? 'ok' : 'DIFFERENT'),
		", dur: ",fmt_time(time() - $t), "\n";

	$dbh->commit;

	return $same;
=cut
}


#----- main

exit unless $opt->host;

foreach ( 0 .. $#{ $opt->host } ) {

	my $host = lc $opt->host->[$_];
	my $num  = $opt->num->[$_];

	if ( ! $opt->ok ) {
		warn "ERROR $host $num running cleanup";
		foreach my $path ( glob "$Conf{ArchiveDest}/$host.$num.*" ) {
			warn "# rm $path";
			unlink $path || die $!;
		}
	} else {
		check_archive $host => $num;
	}

}

exit; # FIXME

my $sth = $dbh->prepare( qq{
	
select
	backups.id as backup_id,
	hosts.name as host,
	shares.name as share,
	backups.num as num,
	backups.date,
	inc_size,
	parts,
	count(backup_parts.backup_id) as backup_parts
from backups
	join shares on backups.hostid = shares.hostid
		and shares.id = backups.shareid
	join hosts on shares.hostid = hosts.id
	full outer join backup_parts on backups.id = backup_parts.backup_id
where not inc_deleted and backups.size > 0
group by backups.id, hosts.name, shares.name, backups.num, backups.date, inc_size, parts, backup_parts.backup_id
order by backups.date

} );

$sth->execute();
my $num_backups = $sth->rows;
my $curr_backup = 1;

while (my $row = $sth->fetchrow_hashref) {

	$curr_backup++;

	my $tar_file = BackupPC::Search::getGzipName($row->{'host'}, $row->{'share'}, $row->{'num'});

	# this will return -1 if file doesn't exist
	my $size = BackupPC::Search::get_tgz_size_by_name($tar_file);

	print "# host: ".$row->{host}.", share: ".$row->{'share'}.", backup_num:".$row->{num}." size: $size backup.size: ", $row->{inc_size},"\n" if $opt->debug;

	if ( $row->{'inc_size'} != -1 && $size != -1 && $row->{'inc_size'} >= $size && $row->{parts} == $row->{backup_parts}) {
		if ($opt->check) {
			tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
		} else {
			next;
		}
	}

	print curr_time, " creating $curr_backup/$num_backups ", $row->{host}, ":", $row->{share}, " #", $row->{num},
		" ", strftime('%Y-%m-%d', localtime($row->{date})), " -> $tar_file";

	my $t = time();

=for later
	# re-create archive?
	my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
	print STDERR "## $cmd\n" if ($opt->debug);

	if (system($cmd) != 0) {
		print STDERR " FAILED, marking this backup deleted";
		backup_inc_deleted( $row->{backup_id} );
	}
=cut

	print ", dur: ",fmt_time(time() - $t), "\n";

}

undef $sth;
$dbh->disconnect;