Annotation of /trunk/bin/BackupPC_updatedb

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

use strict;
use lib "__INSTALLDIR__/lib";

use DBI;
use BackupPC::Lib;
use BackupPC::View;
use Data::Dumper;
use Getopt::Std;
use Time::HiRes qw/time/;
use File::Pid;
use POSIX qw/strftime/;
use BackupPC::SearchLib;

use constant BPC_FTYPE_DIR => 5;
use constant EST_CHUNK => 100000;

my $debug = 0;
$|=1;

my $start_t = time();

my $pidfile = new File::Pid;

if (my $pid = $pidfile->running ) {
        die "$0 already running: $pid\n";
} elsif ($pidfile->pid ne $$) {
        $pidfile->remove;
        $pidfile = new File::Pid;
}
$pidfile->write;
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n";

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

my $hosts;
my $bpc = BackupPC::Lib->new || die;
my %Conf = $bpc->Conf();
my $TopDir = $bpc->TopDir();
my $beenThere = {};

my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
my $user = $Conf{SearchUser} || '';

my $use_hest = $Conf{HyperEstraierIndex};
my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest);

my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });

my %opt;

if ( !getopts("cdm:v:ij", \%opt ) ) {
        print STDERR <<EOF;
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i]

Options:
        -c      create database on first use
        -d      delete database before import
        -m num  import just num increments for one host
        -v num  set verbosity (debug) level (default $debug)
        -i      update Hyper Estraier full text index
        -j      update full text, don't check existing files

Option -j is variation on -i. It will allow faster initial creation
of full-text index from existing database.

EOF
        exit 1;
}

if ($opt{v}) {
        print "Debug level at $opt{v}\n";
        $debug = $opt{v};
}

#---- subs ----

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;
}

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

my $hest_db;
my $hest_node;

sub signal {
        my($sig) = @_;
        if ($hest_db) {
                print "\nCaught a SIG$sig--syncing database and shutting down\n";
                $hest_db->sync();
                $hest_db->close();
        }
        exit(0);
}

$SIG{'INT'}  = \&signal;
$SIG{'QUIT'} = \&signal;

sub hest_update {

        my ($host_id, $share_id, $num) = @_;

        my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n";

        unless ($use_hest) {
                print STDERR "HyperEstraier support not enabled in configuration\n";
                return;
        }

        print curr_time," updating HyperEstraier:";

        my $t = time();

        my $offset = 0;
        my $added = 0;

        print " opening index $use_hest";
        if ($index_path) {
                $hest_db = HyperEstraier::Database->new();
                $hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT);
                print " directly";
        } elsif ($index_node_url) {
                $hest_node ||= HyperEstraier::Node->new($index_node_url);
                $hest_node->set_auth('admin', 'admin');
                print " via node URL";
        } else {
                die "don't know how to use HyperEstraier Index $use_hest";
        }
        print " increment is " . EST_CHUNK . " files:";

        my $results = 0;

        do {

                my $where = '';
                my @data;
                if (defined($host_id) && defined($share_id) && defined($num)) {
                        $where = qq{
                        WHERE
                                hosts.id = ? AND
                                shares.id = ? AND
                                files.backupnum = ?
                        };
                        @data = ( $host_id, $share_id, $num );
                }

                my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset);

                my $sth = $dbh->prepare(qq{
                        SELECT
                                files.id                        AS fid,
                                hosts.name                      AS hname,
                                shares.name                     AS sname,
                                -- shares.share                 AS sharename,
                                files.backupnum                 AS backupnum,
                                -- files.name                   AS filename,
                                files.path                      AS filepath,
                                files.date                      AS date,
                                files.type                      AS type,
                                files.size                      AS size,
                                files.shareid                   AS shareid,
                                backups.date                    AS backup_date
                        FROM files 
                                INNER JOIN shares       ON files.shareID=shares.ID
                                INNER JOIN hosts        ON hosts.ID = shares.hostID
                                INNER JOIN backups      ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID
                        $where
                        $limit
                });

                $sth->execute(@data);
                $results = $sth->rows;

                if ($results == 0) {
                        print " - no new files\n";
                        last;
                }

                sub fmt_date {
                        my $t = shift || return;
                        my $iso = BackupPC::Lib::timeStamp($t);
                        $iso =~ s/\s/T/;
                        return $iso;
                }

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

                        my $fid = $row->{'fid'} || die "no fid?";
                        my $uri = 'file:///' . $fid;

                        unless ($skip_check) {
                                my $id = ($hest_db || $hest_node)->uri_to_id($uri);
                                next unless ($id == -1);
                        }

                        # create a document object 
                        my $doc = HyperEstraier::Document->new;

                        # add attributes to the document object 
                        $doc->add_attr('@uri', $uri);

                        foreach my $c (@{ $sth->{NAME} }) {
                                $doc->add_attr($c, $row->{$c}) if ($row->{$c});
                        }

                        #$doc->add_attr('@cdate', fmt_date($row->{'date'}));

                        # add the body text to the document object 
                        my $path = $row->{'filepath'};
                        $doc->add_text($path);
                        $path =~ s/(.)/$1 /g;
                        $doc->add_hidden_text($path);

                        print STDERR $doc->dump_draft,"\n" if ($debug > 1);

                        # register the document object to the database
                        if ($hest_db) {
                                $hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN);
                        } elsif ($hest_node) {
                                $hest_node->put_doc($doc);
                        } else {
                                die "not supported";
                        }
                        $added++;
                }

                print " $added";
                $hest_db->sync() if ($index_path);

                $offset += EST_CHUNK;

        } while ($results == EST_CHUNK);

        if ($index_path) {
                print ", close";
                $hest_db->close();
        }

        my $dur = (time() - $t) || 1;
        printf(" [%.2f/s dur: %s]\n",
                ( $added / $dur ),
                fmt_time($dur)
        );
}

#---- /subs ----


## update index ##
if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) {
        # update all
        print "force update of HyperEstraier index ";
        print "importing existing data" unless (-e $index_path);
        print "by -i flag" if ($opt{i});
        print "by -j flag" if ($opt{j});
        print "\n";
        hest_update();
}

## create tables ##
if ($opt{c}) {
        sub do_index {
                my $index = shift || return;
                my ($table,$col,$unique) = split(/:/, $index);
                $unique ||= '';
                $index =~ s/\W+/_/g;
                print "$index on $table($col)" . ( $unique ? "u" : "" ) . " ";
                $dbh->do(qq{ create $unique index $index on $table($col) });
        }

        print "creating tables...\n";

        $dbh->do( qq{
                create table hosts (
                        ID      SERIAL          PRIMARY KEY,
                        name    VARCHAR(30)     NOT NULL,
                        IP      VARCHAR(15)
                );            

                create table shares (
                        ID      SERIAL          PRIMARY KEY,
                        hostID  INTEGER         NOT NULL references hosts(id),
                        name    VARCHAR(30)     NOT NULL,
                        share   VARCHAR(200)    NOT NULL
                );            

                create table dvds (
                        ID      SERIAL          PRIMARY KEY, 
                        num     INTEGER         NOT NULL,
                        name    VARCHAR(255)    NOT NULL,
                        mjesto  VARCHAR(255)
                );

                create table backups (
                        id      serial,
                        hostID  INTEGER         NOT NULL references hosts(id),
                        num     INTEGER         NOT NULL,
                        date    integer         NOT NULL, 
                        type    CHAR(4)         not null,
                        shareID integer         not null references shares(id),
                        size    bigint          not null,
                        inc_size bigint         not null default -1,
                        inc_deleted boolean     default false,
                        parts   integer         not null default 1,
                        PRIMARY KEY(id)
                );            

                create table files (
                        ID              SERIAL,
                        shareID         INTEGER NOT NULL references shares(id),
                        backupNum       INTEGER NOT NULL,
                        name            VARCHAR(255) NOT NULL,
                        path            VARCHAR(255) NOT NULL,
                        date            integer NOT NULL,
                        type            INTEGER NOT NULL,
                        size            bigint  NOT NULL,
                        primary key(id)
                );

                create table archive (
                        id              serial,
                        dvd_nr          int not null,
                        total_size      bigint default -1,
                        note            text,
                        username        varchar(20) not null,
                        date            timestamp default now(),
                        primary key(id)
                );      

                create table archive_backup (
                        archive_id      int not null references archive(id) on delete cascade,
                        backup_id       int not null references backups(id),
                        primary key(archive_id, backup_id)
                );

                create table archive_burned (
                        archive_id      int references archive(id),
                        date            date default now(),
                        part            int not null default 1,
                        iso_size bigint default -1
                );

        });

        print "creating indexes: ";

        foreach my $index (qw(
                hosts:name
                backups:hostID
                backups:num
                backups:shareID
                shares:hostID
                shares:name
                files:shareID
                files:path
                files:name
                files:date
                files:size
                archive:dvd_nr
                archive_burned:archive_id
        )) {
                do_index($index);
        }

        print " creating sequence: ";
        foreach my $seq (qw/dvd_nr/) {
                print "$seq ";
                $dbh->do( qq{ CREATE SEQUENCE $seq } );
        }


        print "...\n";

        $dbh->commit;

}

## delete data before inseting ##
if ($opt{d}) {
        print "deleting ";
        foreach my $table (qw(files dvds backups shares hosts)) {
                print "$table ";
                $dbh->do(qq{ DELETE FROM $table });
        }
        print " done...\n";

        $dbh->commit;
}

## insert new values ##

# get hosts
$hosts = $bpc->HostInfoRead();
my $hostID;
my $shareID;

my $sth;

$sth->{insert_hosts} = $dbh->prepare(qq{
INSERT INTO hosts (name, IP) VALUES (?,?)
});

$sth->{hosts_by_name} = $dbh->prepare(qq{
SELECT ID FROM hosts WHERE name=?
});

$sth->{backups_count} = $dbh->prepare(qq{
SELECT COUNT(*)
FROM backups
WHERE hostID=? AND num=? AND shareid=?
});

$sth->{insert_backups} = $dbh->prepare(qq{
INSERT INTO backups (hostID, num, date, type, shareid, size)
VALUES (?,?,?,?,?,?)
});

$sth->{insert_files} = $dbh->prepare(qq{
INSERT INTO files
        (shareID, backupNum, name, path, date, type, size)
        VALUES (?,?,?,?,?,?,?)
});

foreach my $host_key (keys %{$hosts}) {

        my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key";

        $sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'});

        unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) {
                $sth->{insert_hosts}->execute(
                        $hosts->{$host_key}->{'host'},
                        $hosts->{$host_key}->{'ip'}
                );

                $hostID = $dbh->last_insert_id(undef,undef,'hosts',undef);
        }

        print "host ".$hosts->{$host_key}->{'host'}.": ";
 
        # get backups for a host
        my @backups = $bpc->BackupInfoRead($hostname);
        my $incs = scalar @backups;
        print  "$incs increments\n";

        my $inc_nr = 0;
        $beenThere = {};

        foreach my $backup (@backups) {

                $inc_nr++;
                last if ($opt{m} && $inc_nr > $opt{m});

                my $backupNum = $backup->{'num'};
                my @backupShares = ();

                printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", 
                        $hosts->{$host_key}->{'host'},
                        $inc_nr, $incs, $backupNum, 
                        $backup->{type} || '?',
                        $backup->{nFilesNew} || '?', $backup->{nFiles} || '?',
                        strftime($t_fmt,localtime($backup->{startTime})),
                        fmt_time($backup->{endTime} - $backup->{startTime})
                );

                my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1);
                foreach my $share ($files->shareList($backupNum)) {

                        my $t = time();

                        $shareID = getShareID($share, $hostID, $hostname);
                
                        $sth->{backups_count}->execute($hostID, $backupNum, $shareID);
                        my ($count) = $sth->{backups_count}->fetchrow_array();
                        # skip if allready in database!
                        next if ($count > 0);

                        # dump some log
                        print curr_time," ", $share;

                        my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID);

                        $sth->{insert_backups}->execute(
                                $hostID,
                                $backupNum,
                                $backup->{'endTime'},
                                substr($backup->{'type'},0,4),
                                $shareID,
                                $size,
                        );

                        print " commit";
                        $dbh->commit();

                        my $dur = (time() - $t) || 1;
                        printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n",
                                $nf, $f, $nd, $d,
                                ($size / 1024 / 1024),
                                ( ($f+$d) / $dur ),
                                fmt_time($dur)
                        );

                        hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0);
                }

        }
}
undef $sth;
$dbh->commit();
$dbh->disconnect();

print "total duration: ",fmt_time(time() - $start_t),"\n";

$pidfile->remove;

sub getShareID() {

        my ($share, $hostID, $hostname) = @_;

        $sth->{share_id} ||= $dbh->prepare(qq{
                SELECT ID FROM shares WHERE hostID=? AND name=?
        });

        $sth->{share_id}->execute($hostID,$share);

        my ($id) = $sth->{share_id}->fetchrow_array();

        return $id if (defined($id));

        $sth->{insert_share} ||= $dbh->prepare(qq{
                INSERT INTO shares 
                        (hostID,name,share) 
                VALUES (?,?,?)
        });

        my $drop_down = $hostname . '/' . $share;
        $drop_down =~ s#//+#/#g;

        $sth->{insert_share}->execute($hostID,$share, $drop_down);
        return $dbh->last_insert_id(undef,undef,'shares',undef);
}

sub found_in_db {

        my @data = @_;
        shift @data;

        my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_;

        return $beenThere->{$key} if (defined($beenThere->{$key}));

        $sth->{file_in_db} ||= $dbh->prepare(qq{
                SELECT 1 FROM files
                WHERE shareID = ? and
                        path = ? and 
                        date = ? and
                        size = ?
                LIMIT 1
        });

        my @param = ($shareID,$path,$date,$size);
        $sth->{file_in_db}->execute(@param);
        my $rows = $sth->{file_in_db}->rows;
        print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3);

        $beenThere->{$key}++;

        $sth->{'insert_files'}->execute(@data) unless ($rows);
        return $rows;
}

####################################################
# recursing through filesystem structure and       #
# and returning flattened files list               #
####################################################
sub recurseDir($$$$$$$$) {

        my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_;

        print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1);

        my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0);

        { # scope
                my @stack;

                print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2);
                my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir);

                # first, add all the entries in current directory
                foreach my $path_key (keys %{$filesInBackup}) {
                        print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3);
                        my @data = (
                                $shareID,
                                $backupNum,
                                $path_key,
                                $filesInBackup->{$path_key}->{'relPath'},
                                $filesInBackup->{$path_key}->{'mtime'},
                                $filesInBackup->{$path_key}->{'type'},
                                $filesInBackup->{$path_key}->{'size'}
                        );

                        my $key = join(" ", (
                                $shareID,
                                $dir,
                                $path_key,
                                $filesInBackup->{$path_key}->{'mtime'},
                                $filesInBackup->{$path_key}->{'size'}
                        ));

                        my $found;
                        if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) {
                                print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2);

                                if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
                                        $new_dirs++ unless ($found);
                                        print STDERR " dir\n" if ($debug >= 2);
                                } else {
                                        $new_files++ unless ($found);
                                        print STDERR " file\n" if ($debug >= 2);
                                }
                                $size += $filesInBackup->{$path_key}->{'size'} || 0;
                        }

                        if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
                                $nr_dirs++;

                                my $full_path = $dir . '/' . $path_key;
                                push @stack, $full_path;
                                print STDERR "### store to stack: $full_path\n" if ($debug >= 3);

#                               my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key});
#
#                               $nr_files += $f;
#                               $new_files += $nf;
#                               $nr_dirs += $d;
#                               $new_dirs += $nd;

                        } else {
                                $nr_files++;
                        }
                }

                print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2);

                while ( my $dir = shift @stack ) {
                        my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID);
                        print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1);
                        $nr_files += $f;
                        $new_files += $nf;
                        $nr_dirs += $d;
                        $new_dirs += $nd;
                        $size += $s;
                }
        }

        return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size);
}

1	dpavlin	14	#!/usr/local/bin/perl -w
2	dpavlin	6
3			use strict;
4	dpavlin	48	use lib "__INSTALLDIR__/lib";
5
6	dpavlin	6	use DBI;
7			use BackupPC::Lib;
8			use BackupPC::View;
9			use Data::Dumper;
10			use Getopt::Std;
11	dpavlin	37	use Time::HiRes qw/time/;
12	dpavlin	38	use File::Pid;
13	dpavlin	37	use POSIX qw/strftime/;
14	dpavlin	118	use BackupPC::SearchLib;
15	dpavlin	48
16	dpavlin	6	use constant BPC_FTYPE_DIR => 5;
17	dpavlin	99	use constant EST_CHUNK => 100000;
18	dpavlin	6
19	dpavlin	30	my $debug = 0;
20	dpavlin	14	$\|=1;
21	dpavlin	6
22	dpavlin	51	my $start_t = time();
23
24	dpavlin	38	my $pidfile = new File::Pid;
25
26			if (my $pid = $pidfile->running ) {
27			die "$0 already running: $pid\n";
28			} elsif ($pidfile->pid ne $$) {
29			$pidfile->remove;
30			$pidfile = new File::Pid;
31			}
32	dpavlin	39	$pidfile->write;
33			print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n";
34	dpavlin	38
35	dpavlin	37	my $t_fmt = '%Y-%m-%d %H:%M:%S';
36
37	dpavlin	6	my $hosts;
38			my $bpc = BackupPC::Lib->new \|\| die;
39			my %Conf = $bpc->Conf();
40			my $TopDir = $bpc->TopDir();
41	dpavlin	14	my $beenThere = {};
42	dpavlin	6
43	dpavlin	51	my $dsn = $Conf{SearchDSN} \|\| die "Need SearchDSN in config.pl\n";
44			my $user = $Conf{SearchUser} \|\| '';
45	dpavlin	116
46			my $use_hest = $Conf{HyperEstraierIndex};
47	dpavlin	118	my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest);
48	dpavlin	6
49	dpavlin	49	my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
50
51	dpavlin	6	my %opt;
52
53	dpavlin	131	if ( !getopts("cdm:v:ij", \%opt ) ) {
54	dpavlin	6	print STDERR <<EOF;
55	dpavlin	81	usage: $0 [-c\|-d] [-m num] [-v\|-v level] [-i]
56	dpavlin	6
57			Options:
58	dpavlin	14	-c create database on first use
59			-d delete database before import
60			-m num import just num increments for one host
61	dpavlin	44	-v num set verbosity (debug) level (default $debug)
62	dpavlin	131	-i update Hyper Estraier full text index
63			-j update full text, don't check existing files
64
65			Option -j is variation on -i. It will allow faster initial creation
66			of full-text index from existing database.
67
68	dpavlin	6	EOF
69			exit 1;
70			}
71
72	dpavlin	81	if ($opt{v}) {
73			print "Debug level at $opt{v}\n";
74			$debug = $opt{v};
75			}
76
77	dpavlin	86	#---- subs ----
78
79			sub fmt_time {
80			my $t = shift \|\| return;
81			my $out = "";
82			my ($ss,$mm,$hh) = gmtime($t);
83			$out .= "${hh}h" if ($hh);
84			$out .= sprintf("%02d:%02d", $mm,$ss);
85			return $out;
86			}
87
88			sub curr_time {
89			return strftime($t_fmt,localtime());
90			}
91
92	dpavlin	89	my $hest_db;
93	dpavlin	116	my $hest_node;
94	dpavlin	86
95	dpavlin	97	sub signal {
96			my($sig) = @_;
97			if ($hest_db) {
98			print "\nCaught a SIG$sig--syncing database and shutting down\n";
99			$hest_db->sync();
100			$hest_db->close();
101			}
102			exit(0);
103			}
104
105			$SIG{'INT'} = \&signal;
106			$SIG{'QUIT'} = \&signal;
107
108	dpavlin	89	sub hest_update {
109	dpavlin	81
110	dpavlin	89	my ($host_id, $share_id, $num) = @_;
111
112	dpavlin	131	my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n";
113
114	dpavlin	116	unless ($use_hest) {
115			print STDERR "HyperEstraier support not enabled in configuration\n";
116			return;
117			}
118
119	dpavlin	98	print curr_time," updating HyperEstraier:";
120	dpavlin	86
121			my $t = time();
122	dpavlin	89
123	dpavlin	98	my $offset = 0;
124			my $added = 0;
125	dpavlin	89
126	dpavlin	130	print " opening index $use_hest";
127	dpavlin	116	if ($index_path) {
128			$hest_db = HyperEstraier::Database->new();
129	dpavlin	118	$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER \| $HyperEstraier::Database::DBCREAT);
130	dpavlin	130	print " directly";
131	dpavlin	116	} elsif ($index_node_url) {
132			$hest_node \|\|= HyperEstraier::Node->new($index_node_url);
133			$hest_node->set_auth('admin', 'admin');
134	dpavlin	130	print " via node URL";
135	dpavlin	116	} else {
136			die "don't know how to use HyperEstraier Index $use_hest";
137			}
138	dpavlin	107	print " increment is " . EST_CHUNK . " files:";
139	dpavlin	99
140	dpavlin	98	my $results = 0;
141	dpavlin	81
142	dpavlin	98	do {
143	dpavlin	81
144	dpavlin	98	my $where = '';
145	dpavlin	104	my @data;
146	dpavlin	190	if (defined($host_id) && defined($share_id) && defined($num)) {
147	dpavlin	98	$where = qq{
148			WHERE
149			hosts.id = ? AND
150			shares.id = ? AND
151			files.backupnum = ?
152			};
153	dpavlin	104	@data = ( $host_id, $share_id, $num );
154	dpavlin	98	}
155	dpavlin	89
156	dpavlin	98	my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset);
157	dpavlin	81
158	dpavlin	98	my $sth = $dbh->prepare(qq{
159			SELECT
160			files.id AS fid,
161			hosts.name AS hname,
162			shares.name AS sname,
163			-- shares.share AS sharename,
164			files.backupnum AS backupnum,
165			-- files.name AS filename,
166			files.path AS filepath,
167			files.date AS date,
168			files.type AS type,
169			files.size AS size,
170			files.shareid AS shareid,
171			backups.date AS backup_date
172			FROM files
173			INNER JOIN shares ON files.shareID=shares.ID
174			INNER JOIN hosts ON hosts.ID = shares.hostID
175			INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID
176			$where
177			$limit
178			});
179	dpavlin	81
180	dpavlin	104	$sth->execute(@data);
181	dpavlin	98	$results = $sth->rows;
182	dpavlin	81
183	dpavlin	98	if ($results == 0) {
184	dpavlin	107	print " - no new files\n";
185	dpavlin	98	last;
186			}
187	dpavlin	89
188	dpavlin	98	sub fmt_date {
189			my $t = shift \|\| return;
190			my $iso = BackupPC::Lib::timeStamp($t);
191			$iso =~ s/\s/T/;
192			return $iso;
193			}
194	dpavlin	89
195	dpavlin	98	while (my $row = $sth->fetchrow_hashref()) {
196	dpavlin	89
197	dpavlin	98	my $fid = $row->{'fid'} \|\| die "no fid?";
198			my $uri = 'file:///' . $fid;
199	dpavlin	82
200	dpavlin	131	unless ($skip_check) {
201			my $id = ($hest_db \|\| $hest_node)->uri_to_id($uri);
202			next unless ($id == -1);
203			}
204	dpavlin	82
205	dpavlin	98	# create a document object
206			my $doc = HyperEstraier::Document->new;
207	dpavlin	81
208	dpavlin	98	# add attributes to the document object
209			$doc->add_attr('@uri', $uri);
210	dpavlin	89
211	dpavlin	98	foreach my $c (@{ $sth->{NAME} }) {
212			$doc->add_attr($c, $row->{$c}) if ($row->{$c});
213			}
214	dpavlin	89
215	dpavlin	98	#$doc->add_attr('@cdate', fmt_date($row->{'date'}));
216	dpavlin	81
217	dpavlin	98	# add the body text to the document object
218			my $path = $row->{'filepath'};
219			$doc->add_text($path);
220			$path =~ s/(.)/$1 /g;
221			$doc->add_hidden_text($path);
222	dpavlin	81
223	dpavlin	98	print STDERR $doc->dump_draft,"\n" if ($debug > 1);
224
225			# register the document object to the database
226	dpavlin	116	if ($hest_db) {
227			$hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN);
228			} elsif ($hest_node) {
229			$hest_node->put_doc($doc);
230			} else {
231			die "not supported";
232			}
233	dpavlin	98	$added++;
234	dpavlin	81	}
235
236	dpavlin	98	print " $added";
237	dpavlin	116	$hest_db->sync() if ($index_path);
238	dpavlin	81
239	dpavlin	98	$offset += EST_CHUNK;
240	dpavlin	81
241	dpavlin	98	} while ($results == EST_CHUNK);
242	dpavlin	81
243	dpavlin	116	if ($index_path) {
244			print ", close";
245			$hest_db->close();
246			}
247	dpavlin	81
248	dpavlin	86	my $dur = (time() - $t) \|\| 1;
249	dpavlin	98	printf(" [%.2f/s dur: %s]\n",
250	dpavlin	89	( $added / $dur ),
251	dpavlin	86	fmt_time($dur)
252			);
253	dpavlin	89	}
254	dpavlin	86
255	dpavlin	89	#---- /subs ----
256
257
258			## update index ##
259	dpavlin	131	if (($opt{i} \|\| $opt{j} \|\| ($index_path && ! -e $index_path)) && !$opt{c}) {
260	dpavlin	89	# update all
261			print "force update of HyperEstraier index ";
262			print "importing existing data" unless (-e $index_path);
263			print "by -i flag" if ($opt{i});
264	dpavlin	131	print "by -j flag" if ($opt{j});
265	dpavlin	89	print "\n";
266			hest_update();
267	dpavlin	81	}
268
269	dpavlin	89	## create tables ##
270	dpavlin	14	if ($opt{c}) {
271	dpavlin	49	sub do_index {
272			my $index = shift \|\| return;
273	dpavlin	152	my ($table,$col,$unique) = split(/:/, $index);
274	dpavlin	49	$unique \|\|= '';
275	dpavlin	152	$index =~ s/\W+/_/g;
276			print "$index on $table($col)" . ( $unique ? "u" : "" ) . " ";
277	dpavlin	49	$dbh->do(qq{ create $unique index $index on $table($col) });
278			}
279
280	dpavlin	14	print "creating tables...\n";
281	dpavlin	183
282			$dbh->do( qq{
283	dpavlin	14	create table hosts (
284	dpavlin	49	ID SERIAL PRIMARY KEY,
285	dpavlin	14	name VARCHAR(30) NOT NULL,
286			IP VARCHAR(15)
287	dpavlin	6	);
288	dpavlin	183
289	dpavlin	14	create table shares (
290	dpavlin	49	ID SERIAL PRIMARY KEY,
291	dpavlin	14	hostID INTEGER NOT NULL references hosts(id),
292			name VARCHAR(30) NOT NULL,
293	dpavlin	152	share VARCHAR(200) NOT NULL
294	dpavlin	6	);
295	dpavlin	130
296			create table dvds (
297			ID SERIAL PRIMARY KEY,
298			num INTEGER NOT NULL,
299			name VARCHAR(255) NOT NULL,
300			mjesto VARCHAR(255)
301			);
302	dpavlin	183
303	dpavlin	14	create table backups (
304	dpavlin	152	id serial,
305	dpavlin	14	hostID INTEGER NOT NULL references hosts(id),
306			num INTEGER NOT NULL,
307	dpavlin	49	date integer NOT NULL,
308			type CHAR(4) not null,
309	dpavlin	65	shareID integer not null references shares(id),
310	dpavlin	136	size bigint not null,
311	dpavlin	152	inc_size bigint not null default -1,
312			inc_deleted boolean default false,
313	dpavlin	196	parts integer not null default 1,
314	dpavlin	152	PRIMARY KEY(id)
315	dpavlin	6	);
316
317	dpavlin	14	create table files (
318	dpavlin	152	ID SERIAL,
319			shareID INTEGER NOT NULL references shares(id),
320			backupNum INTEGER NOT NULL,
321			name VARCHAR(255) NOT NULL,
322			path VARCHAR(255) NOT NULL,
323			date integer NOT NULL,
324			type INTEGER NOT NULL,
325			size bigint NOT NULL,
326			primary key(id)
327	dpavlin	6	);
328
329	dpavlin	152	create table archive (
330			id serial,
331	dpavlin	130	dvd_nr int not null,
332	dpavlin	152	total_size bigint default -1,
333	dpavlin	130	note text,
334			username varchar(20) not null,
335	dpavlin	152	date timestamp default now(),
336	dpavlin	130	primary key(id)
337			);
338
339	dpavlin	183	create table archive_backup (
340	dpavlin	160	archive_id int not null references archive(id) on delete cascade,
341	dpavlin	152	backup_id int not null references backups(id),
342	dpavlin	130	primary key(archive_id, backup_id)
343			);
344	dpavlin	183
345			create table archive_burned (
346	dpavlin	199	archive_id int references archive(id),
347			date date default now(),
348			part int not null default 1,
349	dpavlin	196	iso_size bigint default -1
350	dpavlin	183	);
351
352	dpavlin	130	});
353
354	dpavlin	153	print "creating indexes: ";
355	dpavlin	6
356	dpavlin	14	foreach my $index (qw(
357	dpavlin	152	hosts:name
358			backups:hostID
359			backups:num
360			backups:shareID
361			shares:hostID
362			shares:name
363			files:shareID
364			files:path
365			files:name
366			files:date
367			files:size
368			archive:dvd_nr
369	dpavlin	183	archive_burned:archive_id
370	dpavlin	14	)) {
371	dpavlin	49	do_index($index);
372	dpavlin	14	}
373	dpavlin	153
374			print " creating sequence: ";
375			foreach my $seq (qw/dvd_nr/) {
376			print "$seq ";
377			$dbh->do( qq{ CREATE SEQUENCE $seq } );
378			}
379
380
381	dpavlin	49	print "...\n";
382	dpavlin	14
383	dpavlin	49	$dbh->commit;
384	dpavlin	14
385			}
386
387	dpavlin	89	## delete data before inseting ##
388	dpavlin	14	if ($opt{d}) {
389			print "deleting ";
390	dpavlin	49	foreach my $table (qw(files dvds backups shares hosts)) {
391	dpavlin	14	print "$table ";
392			$dbh->do(qq{ DELETE FROM $table });
393			}
394			print " done...\n";
395	dpavlin	49
396	dpavlin	51	$dbh->commit;
397	dpavlin	14	}
398
399	dpavlin	89	## insert new values ##
400	dpavlin	6
401			# get hosts
402	dpavlin	8	$hosts = $bpc->HostInfoRead();
403	dpavlin	6	my $hostID;
404			my $shareID;
405
406	dpavlin	14	my $sth;
407	dpavlin	6
408	dpavlin	14	$sth->{insert_hosts} = $dbh->prepare(qq{
409			INSERT INTO hosts (name, IP) VALUES (?,?)
410			});
411	dpavlin	6
412	dpavlin	14	$sth->{hosts_by_name} = $dbh->prepare(qq{
413			SELECT ID FROM hosts WHERE name=?
414			});
415
416	dpavlin	65	$sth->{backups_count} = $dbh->prepare(qq{
417	dpavlin	14	SELECT COUNT(*)
418			FROM backups
419	dpavlin	65	WHERE hostID=? AND num=? AND shareid=?
420	dpavlin	14	});
421
422			$sth->{insert_backups} = $dbh->prepare(qq{
423	dpavlin	66	INSERT INTO backups (hostID, num, date, type, shareid, size)
424			VALUES (?,?,?,?,?,?)
425	dpavlin	14	});
426
427			$sth->{insert_files} = $dbh->prepare(qq{
428			INSERT INTO files
429	dpavlin	62	(shareID, backupNum, name, path, date, type, size)
430			VALUES (?,?,?,?,?,?,?)
431	dpavlin	14	});
432
433			foreach my $host_key (keys %{$hosts}) {
434
435			my $hostname = $hosts->{$host_key}->{'host'} \|\| die "can't find host for $host_key";
436
437			$sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'});
438
439			unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) {
440			$sth->{insert_hosts}->execute(
441			$hosts->{$host_key}->{'host'},
442			$hosts->{$host_key}->{'ip'}
443			);
444
445	dpavlin	49	$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef);
446	dpavlin	14	}
447
448	dpavlin	67	print "host ".$hosts->{$host_key}->{'host'}.": ";
449	dpavlin	6
450	dpavlin	14	# get backups for a host
451			my @backups = $bpc->BackupInfoRead($hostname);
452	dpavlin	67	my $incs = scalar @backups;
453			print "$incs increments\n";
454	dpavlin	6
455	dpavlin	14	my $inc_nr = 0;
456	dpavlin	67	$beenThere = {};
457	dpavlin	14
458			foreach my $backup (@backups) {
459	dpavlin	40
460	dpavlin	14	$inc_nr++;
461			last if ($opt{m} && $inc_nr > $opt{m});
462
463			my $backupNum = $backup->{'num'};
464			my @backupShares = ();
465
466	dpavlin	67	printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n",
467			$hosts->{$host_key}->{'host'},
468			$inc_nr, $incs, $backupNum,
469			$backup->{type} \|\| '?',
470			$backup->{nFilesNew} \|\| '?', $backup->{nFiles} \|\| '?',
471	dpavlin	57	strftime($t_fmt,localtime($backup->{startTime})),
472	dpavlin	67	fmt_time($backup->{endTime} - $backup->{startTime})
473			);
474	dpavlin	14
475	dpavlin	34	my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1);
476	dpavlin	14	foreach my $share ($files->shareList($backupNum)) {
477
478	dpavlin	37	my $t = time();
479
480	dpavlin	14	$shareID = getShareID($share, $hostID, $hostname);
481
482	dpavlin	65	$sth->{backups_count}->execute($hostID, $backupNum, $shareID);
483			my ($count) = $sth->{backups_count}->fetchrow_array();
484			# skip if allready in database!
485			next if ($count > 0);
486
487			# dump some log
488	dpavlin	86	print curr_time," ", $share;
489	dpavlin	65
490	dpavlin	66	my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID);
491	dpavlin	65
492			$sth->{insert_backups}->execute(
493			$hostID,
494			$backupNum,
495			$backup->{'endTime'},
496	dpavlin	130	substr($backup->{'type'},0,4),
497	dpavlin	66	$shareID,
498			$size,
499	dpavlin	65	);
500
501			print " commit";
502			$dbh->commit();
503
504	dpavlin	50	my $dur = (time() - $t) \|\| 1;
505	dpavlin	66	printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n",
506	dpavlin	37	$nf, $f, $nd, $d,
507	dpavlin	66	($size / 1024 / 1024),
508	dpavlin	50	( ($f+$d) / $dur ),
509			fmt_time($dur)
510	dpavlin	37	);
511	dpavlin	89
512	dpavlin	107	hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0);
513	dpavlin	14	}
514	dpavlin	29
515	dpavlin	6	}
516			}
517	dpavlin	14	undef $sth;
518	dpavlin	6	$dbh->commit();
519			$dbh->disconnect();
520
521	dpavlin	51	print "total duration: ",fmt_time(time() - $start_t),"\n";
522
523	dpavlin	38	$pidfile->remove;
524
525	dpavlin	14	sub getShareID() {
526	dpavlin	6
527	dpavlin	14	my ($share, $hostID, $hostname) = @_;
528
529			$sth->{share_id} \|\|= $dbh->prepare(qq{
530			SELECT ID FROM shares WHERE hostID=? AND name=?
531			});
532
533			$sth->{share_id}->execute($hostID,$share);
534
535			my ($id) = $sth->{share_id}->fetchrow_array();
536
537			return $id if (defined($id));
538
539			$sth->{insert_share} \|\|= $dbh->prepare(qq{
540			INSERT INTO shares
541	dpavlin	152	(hostID,name,share)
542			VALUES (?,?,?)
543	dpavlin	14	});
544
545	dpavlin	25	my $drop_down = $hostname . '/' . $share;
546			$drop_down =~ s#//+#/#g;
547
548	dpavlin	152	$sth->{insert_share}->execute($hostID,$share, $drop_down);
549	dpavlin	49	return $dbh->last_insert_id(undef,undef,'shares',undef);
550	dpavlin	6	}
551
552	dpavlin	14	sub found_in_db {
553
554	dpavlin	48	my @data = @_;
555			shift @data;
556	dpavlin	14
557	dpavlin	74	my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_;
558	dpavlin	48
559			return $beenThere->{$key} if (defined($beenThere->{$key}));
560
561	dpavlin	14	$sth->{file_in_db} \|\|= $dbh->prepare(qq{
562	dpavlin	48	SELECT 1 FROM files
563	dpavlin	14	WHERE shareID = ? and
564			path = ? and
565			date = ? and
566			size = ?
567	dpavlin	74	LIMIT 1
568	dpavlin	14	});
569
570	dpavlin	67	my @param = ($shareID,$path,$date,$size);
571	dpavlin	14	$sth->{file_in_db}->execute(@param);
572	dpavlin	48	my $rows = $sth->{file_in_db}->rows;
573	dpavlin	74	print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3);
574	dpavlin	48
575			$beenThere->{$key}++;
576
577			$sth->{'insert_files'}->execute(@data) unless ($rows);
578	dpavlin	14	return $rows;
579	dpavlin	6	}
580
581			####################################################
582			# recursing through filesystem structure and #
583			# and returning flattened files list #
584			####################################################
585	dpavlin	14	sub recurseDir($$$$$$$$) {
586	dpavlin	6
587	dpavlin	35	my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_;
588	dpavlin	14
589	dpavlin	44	print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1);
590	dpavlin	29
591	dpavlin	66	my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0);
592	dpavlin	14
593	dpavlin	27	{ # scope
594	dpavlin	29	my @stack;
595	dpavlin	14
596	dpavlin	45	print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2);
597	dpavlin	27	my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir);
598	dpavlin	14
599	dpavlin	27	# first, add all the entries in current directory
600			foreach my $path_key (keys %{$filesInBackup}) {
601	dpavlin	66	print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3);
602	dpavlin	27	my @data = (
603			$shareID,
604			$backupNum,
605			$path_key,
606			$filesInBackup->{$path_key}->{'relPath'},
607			$filesInBackup->{$path_key}->{'mtime'},
608			$filesInBackup->{$path_key}->{'type'},
609			$filesInBackup->{$path_key}->{'size'}
610			);
611
612			my $key = join(" ", (
613			$shareID,
614			$dir,
615			$path_key,
616			$filesInBackup->{$path_key}->{'mtime'},
617			$filesInBackup->{$path_key}->{'size'}
618			));
619
620	dpavlin	70	my $found;
621			if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) {
622	dpavlin	30	print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2);
623	dpavlin	48
624	dpavlin	27	if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
625	dpavlin	70	$new_dirs++ unless ($found);
626	dpavlin	30	print STDERR " dir\n" if ($debug >= 2);
627	dpavlin	27	} else {
628	dpavlin	70	$new_files++ unless ($found);
629	dpavlin	30	print STDERR " file\n" if ($debug >= 2);
630	dpavlin	27	}
631	dpavlin	66	$size += $filesInBackup->{$path_key}->{'size'} \|\| 0;
632	dpavlin	27	}
633
634	dpavlin	14	if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
635	dpavlin	27	$nr_dirs++;
636
637	dpavlin	29	my $full_path = $dir . '/' . $path_key;
638			push @stack, $full_path;
639	dpavlin	30	print STDERR "### store to stack: $full_path\n" if ($debug >= 3);
640	dpavlin	29
641	dpavlin	27	# my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key});
642			#
643			# $nr_files += $f;
644			# $new_files += $nf;
645			# $nr_dirs += $d;
646			# $new_dirs += $nd;
647
648	dpavlin	14	} else {
649	dpavlin	27	$nr_files++;
650	dpavlin	14	}
651			}
652
653	dpavlin	30	print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2);
654	dpavlin	14
655	dpavlin	29	while ( my $dir = shift @stack ) {
656	dpavlin	66	my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID);
657	dpavlin	30	print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1);
658	dpavlin	29	$nr_files += $f;
659			$new_files += $nf;
660			$nr_dirs += $d;
661			$new_dirs += $nd;
662	dpavlin	66	$size += $s;
663	dpavlin	29	}
664	dpavlin	14	}
665
666	dpavlin	66	return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size);
667	dpavlin	6	}
668