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 $where = '';
                my @data;
                if ($host_id && $share_id && $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)
                );            
        });
              
        $dbh->do(qq{
                create table shares (
                        ID      SERIAL          PRIMARY KEY,
                        hostID  INTEGER         NOT NULL references hosts(id),
                        name    VARCHAR(30)     NOT NULL,
                        share   VARCHAR(200)    NOT NULL
                );            
        });

        $dbh->do(qq{
                create table dvds (
                        ID      SERIAL          PRIMARY KEY, 
                        num     INTEGER         NOT NULL,
                        name    VARCHAR(255)    NOT NULL,
                        mjesto  VARCHAR(255)
                );
        });
        
        $dbh->do(qq{
                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,
                        PRIMARY KEY(id)
                );            
        });

        $dbh->do(qq{     
                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)
                );
        });


        $dbh->do( qq{
                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)
                );      
        }
        );

        $dbh->do( qq{
                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)
                );
        });

        print "creating indexes: ";

        foreach my $index (qw(
                hosts:name
                backups:hostID
                backups:num
                backups:shareID
                shares:hostID
                shares:name
                files:shareID
                files:path
                files:name
                files:date
                files:size
                archive:dvd_nr
        )) {
                do_index($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	98	if ($host_id && $share_id && $num) {
147			$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	6
282	dpavlin	14	$dbh->do(qq{
283			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	14	});
289	dpavlin	6
290	dpavlin	14	$dbh->do(qq{
291			create table shares (
292	dpavlin	49	ID SERIAL PRIMARY KEY,
293	dpavlin	14	hostID INTEGER NOT NULL references hosts(id),
294			name VARCHAR(30) NOT NULL,
295	dpavlin	152	share VARCHAR(200) NOT NULL
296	dpavlin	6	);
297	dpavlin	14	});
298	dpavlin	130
299			$dbh->do(qq{
300			create table dvds (
301			ID SERIAL PRIMARY KEY,
302			num INTEGER NOT NULL,
303			name VARCHAR(255) NOT NULL,
304			mjesto VARCHAR(255)
305			);
306			});
307	dpavlin	6
308	dpavlin	14	$dbh->do(qq{
309			create table backups (
310	dpavlin	152	id serial,
311	dpavlin	14	hostID INTEGER NOT NULL references hosts(id),
312			num INTEGER NOT NULL,
313	dpavlin	49	date integer NOT NULL,
314			type CHAR(4) not null,
315	dpavlin	65	shareID integer not null references shares(id),
316	dpavlin	136	size bigint not null,
317	dpavlin	152	inc_size bigint not null default -1,
318			inc_deleted boolean default false,
319			PRIMARY KEY(id)
320	dpavlin	6	);
321	dpavlin	14	});
322	dpavlin	6
323	dpavlin	14	$dbh->do(qq{
324			create table files (
325	dpavlin	152	ID SERIAL,
326			shareID INTEGER NOT NULL references shares(id),
327			backupNum INTEGER NOT NULL,
328			name VARCHAR(255) NOT NULL,
329			path VARCHAR(255) NOT NULL,
330			date integer NOT NULL,
331			type INTEGER NOT NULL,
332			size bigint NOT NULL,
333			primary key(id)
334	dpavlin	6	);
335	dpavlin	14	});
336	dpavlin	6
337	dpavlin	130
338			$dbh->do( qq{
339	dpavlin	152	create table archive (
340			id serial,
341	dpavlin	130	dvd_nr int not null,
342	dpavlin	152	total_size bigint default -1,
343	dpavlin	130	note text,
344			username varchar(20) not null,
345	dpavlin	152	date timestamp default now(),
346	dpavlin	130	primary key(id)
347			);
348			}
349			);
350
351			$dbh->do( qq{
352			create table archive_backup
353			(
354	dpavlin	160	archive_id int not null references archive(id) on delete cascade,
355	dpavlin	152	backup_id int not null references backups(id),
356	dpavlin	130	primary key(archive_id, backup_id)
357			);
358			});
359
360	dpavlin	153	print "creating indexes: ";
361	dpavlin	6
362	dpavlin	14	foreach my $index (qw(
363	dpavlin	152	hosts:name
364			backups:hostID
365			backups:num
366			backups:shareID
367			shares:hostID
368			shares:name
369			files:shareID
370			files:path
371			files:name
372			files:date
373			files:size
374			archive:dvd_nr
375	dpavlin	14	)) {
376	dpavlin	49	do_index($index);
377	dpavlin	14	}
378	dpavlin	153
379			print " creating sequence: ";
380			foreach my $seq (qw/dvd_nr/) {
381			print "$seq ";
382			$dbh->do( qq{ CREATE SEQUENCE $seq } );
383			}
384
385
386	dpavlin	49	print "...\n";
387	dpavlin	14
388	dpavlin	49	$dbh->commit;
389	dpavlin	14
390			}
391
392	dpavlin	89	## delete data before inseting ##
393	dpavlin	14	if ($opt{d}) {
394			print "deleting ";
395	dpavlin	49	foreach my $table (qw(files dvds backups shares hosts)) {
396	dpavlin	14	print "$table ";
397			$dbh->do(qq{ DELETE FROM $table });
398			}
399			print " done...\n";
400	dpavlin	49
401	dpavlin	51	$dbh->commit;
402	dpavlin	14	}
403
404	dpavlin	89	## insert new values ##
405	dpavlin	6
406			# get hosts
407	dpavlin	8	$hosts = $bpc->HostInfoRead();
408	dpavlin	6	my $hostID;
409			my $shareID;
410
411	dpavlin	14	my $sth;
412	dpavlin	6
413	dpavlin	14	$sth->{insert_hosts} = $dbh->prepare(qq{
414			INSERT INTO hosts (name, IP) VALUES (?,?)
415			});
416	dpavlin	6
417	dpavlin	14	$sth->{hosts_by_name} = $dbh->prepare(qq{
418			SELECT ID FROM hosts WHERE name=?
419			});
420
421	dpavlin	65	$sth->{backups_count} = $dbh->prepare(qq{
422	dpavlin	14	SELECT COUNT(*)
423			FROM backups
424	dpavlin	65	WHERE hostID=? AND num=? AND shareid=?
425	dpavlin	14	});
426
427			$sth->{insert_backups} = $dbh->prepare(qq{
428	dpavlin	66	INSERT INTO backups (hostID, num, date, type, shareid, size)
429			VALUES (?,?,?,?,?,?)
430	dpavlin	14	});
431
432			$sth->{insert_files} = $dbh->prepare(qq{
433			INSERT INTO files
434	dpavlin	62	(shareID, backupNum, name, path, date, type, size)
435			VALUES (?,?,?,?,?,?,?)
436	dpavlin	14	});
437
438			foreach my $host_key (keys %{$hosts}) {
439
440			my $hostname = $hosts->{$host_key}->{'host'} \|\| die "can't find host for $host_key";
441
442			$sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'});
443
444			unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) {
445			$sth->{insert_hosts}->execute(
446			$hosts->{$host_key}->{'host'},
447			$hosts->{$host_key}->{'ip'}
448			);
449
450	dpavlin	49	$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef);
451	dpavlin	14	}
452
453	dpavlin	67	print "host ".$hosts->{$host_key}->{'host'}.": ";
454	dpavlin	6
455	dpavlin	14	# get backups for a host
456			my @backups = $bpc->BackupInfoRead($hostname);
457	dpavlin	67	my $incs = scalar @backups;
458			print "$incs increments\n";
459	dpavlin	6
460	dpavlin	14	my $inc_nr = 0;
461	dpavlin	67	$beenThere = {};
462	dpavlin	14
463			foreach my $backup (@backups) {
464	dpavlin	40
465	dpavlin	14	$inc_nr++;
466			last if ($opt{m} && $inc_nr > $opt{m});
467
468			my $backupNum = $backup->{'num'};
469			my @backupShares = ();
470
471	dpavlin	67	printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n",
472			$hosts->{$host_key}->{'host'},
473			$inc_nr, $incs, $backupNum,
474			$backup->{type} \|\| '?',
475			$backup->{nFilesNew} \|\| '?', $backup->{nFiles} \|\| '?',
476	dpavlin	57	strftime($t_fmt,localtime($backup->{startTime})),
477	dpavlin	67	fmt_time($backup->{endTime} - $backup->{startTime})
478			);
479	dpavlin	14
480	dpavlin	34	my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1);
481	dpavlin	14	foreach my $share ($files->shareList($backupNum)) {
482
483	dpavlin	37	my $t = time();
484
485	dpavlin	14	$shareID = getShareID($share, $hostID, $hostname);
486
487	dpavlin	65	$sth->{backups_count}->execute($hostID, $backupNum, $shareID);
488			my ($count) = $sth->{backups_count}->fetchrow_array();
489			# skip if allready in database!
490			next if ($count > 0);
491
492			# dump some log
493	dpavlin	86	print curr_time," ", $share;
494	dpavlin	65
495	dpavlin	66	my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID);
496	dpavlin	65
497			$sth->{insert_backups}->execute(
498			$hostID,
499			$backupNum,
500			$backup->{'endTime'},
501	dpavlin	130	substr($backup->{'type'},0,4),
502	dpavlin	66	$shareID,
503			$size,
504	dpavlin	65	);
505
506			print " commit";
507			$dbh->commit();
508
509	dpavlin	50	my $dur = (time() - $t) \|\| 1;
510	dpavlin	66	printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n",
511	dpavlin	37	$nf, $f, $nd, $d,
512	dpavlin	66	($size / 1024 / 1024),
513	dpavlin	50	( ($f+$d) / $dur ),
514			fmt_time($dur)
515	dpavlin	37	);
516	dpavlin	89
517	dpavlin	107	hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0);
518	dpavlin	14	}
519	dpavlin	29
520	dpavlin	6	}
521			}
522	dpavlin	14	undef $sth;
523	dpavlin	6	$dbh->commit();
524			$dbh->disconnect();
525
526	dpavlin	51	print "total duration: ",fmt_time(time() - $start_t),"\n";
527
528	dpavlin	38	$pidfile->remove;
529
530	dpavlin	14	sub getShareID() {
531	dpavlin	6
532	dpavlin	14	my ($share, $hostID, $hostname) = @_;
533
534			$sth->{share_id} \|\|= $dbh->prepare(qq{
535			SELECT ID FROM shares WHERE hostID=? AND name=?
536			});
537
538			$sth->{share_id}->execute($hostID,$share);
539
540			my ($id) = $sth->{share_id}->fetchrow_array();
541
542			return $id if (defined($id));
543
544			$sth->{insert_share} \|\|= $dbh->prepare(qq{
545			INSERT INTO shares
546	dpavlin	152	(hostID,name,share)
547			VALUES (?,?,?)
548	dpavlin	14	});
549
550	dpavlin	25	my $drop_down = $hostname . '/' . $share;
551			$drop_down =~ s#//+#/#g;
552
553	dpavlin	152	$sth->{insert_share}->execute($hostID,$share, $drop_down);
554	dpavlin	49	return $dbh->last_insert_id(undef,undef,'shares',undef);
555	dpavlin	6	}
556
557	dpavlin	14	sub found_in_db {
558
559	dpavlin	48	my @data = @_;
560			shift @data;
561	dpavlin	14
562	dpavlin	74	my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_;
563	dpavlin	48
564			return $beenThere->{$key} if (defined($beenThere->{$key}));
565
566	dpavlin	14	$sth->{file_in_db} \|\|= $dbh->prepare(qq{
567	dpavlin	48	SELECT 1 FROM files
568	dpavlin	14	WHERE shareID = ? and
569			path = ? and
570			date = ? and
571			size = ?
572	dpavlin	74	LIMIT 1
573	dpavlin	14	});
574
575	dpavlin	67	my @param = ($shareID,$path,$date,$size);
576	dpavlin	14	$sth->{file_in_db}->execute(@param);
577	dpavlin	48	my $rows = $sth->{file_in_db}->rows;
578	dpavlin	74	print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3);
579	dpavlin	48
580			$beenThere->{$key}++;
581
582			$sth->{'insert_files'}->execute(@data) unless ($rows);
583	dpavlin	14	return $rows;
584	dpavlin	6	}
585
586			####################################################
587			# recursing through filesystem structure and #
588			# and returning flattened files list #
589			####################################################
590	dpavlin	14	sub recurseDir($$$$$$$$) {
591	dpavlin	6
592	dpavlin	35	my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_;
593	dpavlin	14
594	dpavlin	44	print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1);
595	dpavlin	29
596	dpavlin	66	my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0);
597	dpavlin	14
598	dpavlin	27	{ # scope
599	dpavlin	29	my @stack;
600	dpavlin	14
601	dpavlin	45	print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2);
602	dpavlin	27	my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir);
603	dpavlin	14
604	dpavlin	27	# first, add all the entries in current directory
605			foreach my $path_key (keys %{$filesInBackup}) {
606	dpavlin	66	print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3);
607	dpavlin	27	my @data = (
608			$shareID,
609			$backupNum,
610			$path_key,
611			$filesInBackup->{$path_key}->{'relPath'},
612			$filesInBackup->{$path_key}->{'mtime'},
613			$filesInBackup->{$path_key}->{'type'},
614			$filesInBackup->{$path_key}->{'size'}
615			);
616
617			my $key = join(" ", (
618			$shareID,
619			$dir,
620			$path_key,
621			$filesInBackup->{$path_key}->{'mtime'},
622			$filesInBackup->{$path_key}->{'size'}
623			));
624
625	dpavlin	70	my $found;
626			if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) {
627	dpavlin	30	print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2);
628	dpavlin	48
629	dpavlin	27	if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
630	dpavlin	70	$new_dirs++ unless ($found);
631	dpavlin	30	print STDERR " dir\n" if ($debug >= 2);
632	dpavlin	27	} else {
633	dpavlin	70	$new_files++ unless ($found);
634	dpavlin	30	print STDERR " file\n" if ($debug >= 2);
635	dpavlin	27	}
636	dpavlin	66	$size += $filesInBackup->{$path_key}->{'size'} \|\| 0;
637	dpavlin	27	}
638
639	dpavlin	14	if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
640	dpavlin	27	$nr_dirs++;
641
642	dpavlin	29	my $full_path = $dir . '/' . $path_key;
643			push @stack, $full_path;
644	dpavlin	30	print STDERR "### store to stack: $full_path\n" if ($debug >= 3);
645	dpavlin	29
646	dpavlin	27	# my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key});
647			#
648			# $nr_files += $f;
649			# $new_files += $nf;
650			# $nr_dirs += $d;
651			# $new_dirs += $nd;
652
653	dpavlin	14	} else {
654	dpavlin	27	$nr_files++;
655	dpavlin	14	}
656			}
657
658	dpavlin	30	print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2);
659	dpavlin	14
660	dpavlin	29	while ( my $dir = shift @stack ) {
661	dpavlin	66	my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID);
662	dpavlin	30	print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1);
663	dpavlin	29	$nr_files += $f;
664			$new_files += $nf;
665			$nr_dirs += $d;
666			$new_dirs += $nd;
667	dpavlin	66	$size += $s;
668	dpavlin	29	}
669	dpavlin	14	}
670
671	dpavlin	66	return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size);
672	dpavlin	6	}
673