X-Git-Url: http://git.rot13.org/?p=BackupPC.git;a=blobdiff_plain;f=bin%2FBackupPC_ASA_PostArchive_Update;h=8d69bf633febf5c79a974944c7ccb583ba01bbe3;hp=97c131830be3d76ae59b86ff24e6095341de0c17;hb=d1bfa4d92bec24954f2e7877f6531644bb84e8f3;hpb=f5ef8b2546333a4b7c7cf27ae62542346fbeeef5

diff --git a/bin/BackupPC_ASA_PostArchive_Update b/bin/BackupPC_ASA_PostArchive_Update
index 97c1318..8d69bf6 100755
--- a/bin/BackupPC_ASA_PostArchive_Update
+++ b/bin/BackupPC_ASA_PostArchive_Update
@@ -7,15 +7,14 @@ use DBI;
 use BackupPC::Lib;
 use BackupPC::View;
 use BackupPC::Attrib qw/:all/;
-#use Data::Dumper;
 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 Getopt::Std;
+use Archive::Tar::Streamed;
+use Algorithm::Diff;
 use File::Slurp;
+use Getopt::Long::Descriptive;
 
 =head1 NAME
 
@@ -23,56 +22,28 @@ BackupPC_ASA_PostArchive_Update
 
 =head1 DESCRIPTION
 
-	# /etc/BackupPC/pc/dvd_tar.pl
+	# /etc/BackupPC/pc/_search_archive.pl
 
 =cut
 
-# FIXME
-my $debug = $ENV{DEBUG} || 1;
-my $check = $ENV{CHECK} || 1;
-
 
 my $bpc = BackupPC::Lib->new || die "can't create BackupPC::Lib";
+$bpc->ConfigRead('_search_archive'); # read our configuration
 my %Conf = $bpc->Conf();
-warn "## ARGV=",dump @ARGV;
-
-
-my $args;
-my $name;
-foreach ( @ARGV ) {
-	my $v = $_;
-	if ( m/(\w+)=(.+)/ ) {
-		$name = $1;
-		$v = $2;
-	}
-	if ( $name =~ m/List/ ) {
-		push @{ $args->{$name} }, $v;
-	} else {
-		$args->{$name} = $v;
-	}
-}
-
-warn "args = ",dump($args);
-
-sub check_archive;
-
-foreach ( 0 .. $#{ $args->{HostList} } ) {
-
-	my $host = $args->{'HostList'}->[$_];
-	my $num  = $args->{'BackupList'}->[$_];
-
-	check_archive $host => $num;
-
-}
-
-exit;
-
 
 use BackupPC::Search;
 %BackupPC::Search::Conf = %Conf;
 
-my $path = abs_path($0);
-$path =~ s{/[^/]+$}{/}; # FIXME remove?
+my ($opt,$usage) = describe_options(
+"%c %o",
+[ 'host|h=s@',	"import just host(s)" ],
+[ 'num|n=s@',	"import just backup number(s)" ],
+[ 'check|c',	"check archives on disk and sync", { default => 1 } ],
+[ 'debug|d',	"debug", { default => 1 } ],
+[ 'help',	"show help" ],
+);
+
+print($usage->text), exit if $opt->help;
 
 $|=1;
 
@@ -80,10 +51,7 @@ my $start_t = time();
 
 my $t_fmt = '%Y-%m-%d %H:%M:%S';
 
-warn "## Conf = ",dump( \%Conf );
-
-my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
-my $user = $Conf{SearchUser} || '';
+#warn "## Conf = ",dump( \%Conf );
 
 my $dbh = DBI->connect($Conf{SearchDSN}, $Conf{SearchUser}, "", { RaiseError => 1, AutoCommit => 0 });
 
@@ -105,10 +73,10 @@ sub fmt_time {
 
 my $hsn_cache;
 
-sub get_backup_id($$$) {
-	my ($host, $share, $num) = @_;
+sub get_backup_id($$) {
+	my ($host, $num) = @_;
 
-	my $key = "$host $share $num";
+	my $key = "$host $num";
 	return $hsn_cache->{$key} if ($hsn_cache->{$key});
 
 	my $sth = $dbh->prepare(qq{
@@ -117,14 +85,14 @@ sub get_backup_id($$$) {
 		FROM backups 
 		INNER JOIN shares	ON backups.shareID=shares.ID
 		INNER JOIN hosts	ON backups.hostID = hosts.ID
-		WHERE hosts.name = ? and shares.name = ? and backups.num = ?
+		WHERE hosts.name = ? and backups.num = ?
 	});
-	$sth->execute($host, $share, $num);
+	$sth->execute($host, $num);
 	my ($id) = $sth->fetchrow_array;
 
-	$hsn_cache->{"$host $share $num"} = $id;
+	$hsn_cache->{"$host $num"} = $id;
 
-	print STDERR "# $host $share $num == $id\n" if $debug;
+	print STDERR "# $host $num == $id\n" if $opt->debug;
 
 	return $id;
 }
@@ -144,24 +112,43 @@ sub system_ok {
 	system(@_) == 0 || die "system @_:$!";
 }
 
+my $sth_inc_size = $dbh->prepare(qq{
+	update backups set
+		inc_size = ?,
+		parts = ?,
+		inc_deleted = false
+	where id = ?
+});
+
 sub check_archive {
 	my ($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";
 
 	my $md5_path = "$Conf{ArchiveDest}/$host.$num.md5";
 	unlink $md5_path if -s $md5_path == 0; # fix empty
 
 	if ( ! -e $md5_path ) {
-		system_ok "md5sum $Conf{ArchiveDest}/$host.$num.* > $md5_path";
+		system_ok "cd $Conf{ArchiveDest} && /usr/bin/md5sum $host.$num.* > $md5_path";
 	} else {
-		system_ok "md5sum -c $md5_path" if $check;
+		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 ( $path, $md5 ) = split(/\s+/,$_);
+		my ( $md5, $path ) = split(/\s+/,$_);
 		$md5sum->{$path} = $md5;
 	}
 
@@ -171,27 +158,33 @@ sub check_archive {
 	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>;
-		chomp($line);
-		$line = <$gzip> if ($line =~ /^\s+compressed/);
+		close($gzip);
 
 		my ($comp, $uncomp) = (0,0);
 
-		if ($line =~ m/^\s+(\d+)\s+(\d+)\s+\d+\.\d+/) {
+		if ($line =~ m/\s+(\d+)\s+(\d+)\s+\d+\.\d+/s) {
 			if (wantarray) {
 				return [ $1, $2 ];
 			} else {
 				return $2;
 			}
 		} else {
-			die "can't find size in line: $line";
+			warn "ERROR can't parse: $line";
+			return -s $filename;
 		}
 	}
 
 	sub check_part {
-		my ($host, $share, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
-		my $backup_id = get_backup_id($host, $share, $num);
+		my ($host, $num, $part_nr, $tar_size, $size, $md5, $items) = @_;
+		my $backup_id = get_backup_id($host, $num);
 		my $sth_md5 = $dbh->prepare(qq{
 			select
 				id, tar_size, size, md5, items
@@ -227,31 +220,30 @@ sub check_archive {
 		$dbh->commit;
 	}
 
-	my @tar_parts =
-		sort map { s/^\Q$Conf{ArchiveDest}\E//; $_ }
-		glob "$Conf{ArchiveDest}/$host.$num.*"
-		;
-	print " [parts: ",join(", ", @tar_parts),"]" if $debug;
+	print " [parts: ",join(", ", @tar_parts),"]" if $opt->debug;
 
-	my $same = 1;
 	my @tar_files;
 
 	my $backup_part;
 
-	print " reading" if $debug;
+	print " reading" if $opt->debug;
+
+	my $part_nr = 0;
+	my $inc_size = 0;
 
-	my $part_nr = 1;
+	foreach my $filename (@tar_parts) {
 
-	foreach my $tarfilename (@tar_parts) {
+		next if $filename eq "$host.$num.md5";
 
-		print "\n\t- $tarfilename";
+		print "\n\t- $filename";
 
-		my $path = "$Conf{ArchiveDest}/$tarfilename";
+		my $path = "$Conf{ArchiveDest}/$filename";
+		$path =~ s{//+}{/}g;
 
 		my $size = (stat( $path ))[7] || die "can't stat $path: $!";
 
-		if ($size > $Conf{MaxArchiveSize}) {
-			print ", part bigger than media $size > $Conf{MaxArchiveSize}\n";
+		if ($size > $Conf{ArchiveMediaSize}) {
+			print ", part bigger than media $size > $Conf{ArchiveMediaSize}\n";
 			return 0;
 		}
 
@@ -271,8 +263,8 @@ sub check_archive {
 			$items++;
 			$tar_size_inarc += $entry->size;
 
-			if ($tar_size_inarc > $Conf{MaxArchiveFileSize}) {
-				print ", part $tarfilename is too big $tar_size_inarc > $Conf{MaxArchiveFileSize}\n";
+			if ($tar_size_inarc > $Conf{ArchiveChunkSize}) {
+				print ", part $filename is too big $tar_size_inarc > $Conf{ArchiveChunkSize}\n";
 				return 0;
 			}
 
@@ -296,22 +288,28 @@ sub check_archive {
 =cut
 
 		# FIXME
-		my $tar_size = $path =~ m/.gz/ ? get_gzip_size( $path ) : -s $path;
+		my $tar_size = get_gzip_size( $path );
 
 		#
 		# finally, check if backup_parts table in database is valid
 		#
 
-		my $md5 = $md5sum->{$path} || die "no md5sum for $path";
+		my $md5 = $md5sum->{$filename} || die "no md5sum for $filename in ",dump($md5sum);
 		my $items = 1;
+		$part_nr++;
 
-		check_part($host, undef, $num, $part_nr, $tar_size, $size, $md5, $items);
+		check_part($host, $num, $part_nr, $tar_size, $size, $md5, $items);
 
-		$part_nr++;
+		# round increment size to 2k block size
+		$inc_size += int(($size + 2048) / 2048);
 	}
 
-	# short-cut and exit;
-	return $same unless($same);
+	$sth_inc_size->execute(
+		$inc_size,
+		$part_nr,
+		get_backup_id($host, $num),
+	);
+	$dbh->commit;
 
 	@tar_files = sort @tar_files;
 	print "\n\t",($#tar_files + 1), " tar files";
@@ -337,6 +335,8 @@ sub check_archive {
 
 	@db_files = sort @db_files;
 
+	my $same = 1;
+
 	if ($#tar_files != $#db_files) {
 		$same = 0;
 		print " NUMBER";
@@ -359,6 +359,17 @@ sub check_archive {
 
 #----- main
 
+foreach ( 0 .. $#{ $opt->host } ) {
+
+	my $host = $opt->host->[$_];
+	my $num  = $opt->num->[$_];
+
+	check_archive $host => $num;
+
+}
+
+exit; # FIXME
+
 my $sth = $dbh->prepare( qq{
 	
 select
@@ -394,10 +405,10 @@ while (my $row = $sth->fetchrow_hashref) {
 	# 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 $debug;
+	print "# host: ".$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} == $row->{backup_parts}) {
-		if ($check) {
+		if ($opt->check) {
 			tar_check($row->{'host'}, $row->{'share'}, $row->{'num'}, $tar_file) && next;
 		} else {
 			next;
@@ -412,7 +423,7 @@ while (my $row = $sth->fetchrow_hashref) {
 =for later
 	# re-create archive?
 	my $cmd = qq[ $tarIncCreate -h "$row->{host}" -s "$row->{share}" -n $row->{num} -f ];
-	print STDERR "## $cmd\n" if ($debug);
+	print STDERR "## $cmd\n" if ($opt->debug);
 
 	if (system($cmd) != 0) {
 		print STDERR " FAILED, marking this backup deleted";