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;

# daylight saving time change offset for 1h
my $dst_offset = 60 * 60;

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:ijf", \%opt ) ) {
        print STDERR <<EOF;
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i|-j|-f]

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
        -f      don't do anything with full text index

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

Option -f will create database which is out of sync with full text index. You
will have to re-run $0 with -i to fix it.

EOF
        exit 1;
}

if ($opt{v}) {
        print "Debug level at $opt{v}\n";
        $debug = $opt{v};
} elsif ($opt{f}) {
        print "WARNING: disabling full-text index update. You need to re-run $0 -j !\n";
        ($use_hest, $index_path, $index_node_url) = (undef, undef, undef);
}

#---- 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 (defined($use_hest)) {
                print STDERR "HyperEstraier support not enabled in configuration\n";
                $use_hest = 0;
                return;
        }

        return unless($use_hest);

        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 (defined($row->{$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 $TopDir . $index_path)) && !$opt{c}) {
        # update all
        print "force update of HyperEstraier index ";
        print "importing existing data" unless (-e $TopDir . $index_path);
        print "by -i flag" if ($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            timestamp default now(),
                        part            int not null default 1,
                        copy            int not null default 1,
                        iso_size bigint default -1
                );

                create table backup_parts (
                        id serial,
                        backup_id int references backups(id),
                        part_nr int not null check (part_nr > 0),
                        tar_size bigint not null check (tar_size > 0),
                        size bigint not null check (size > 0),
                        md5 text not null,
                        items int not null check (items > 0),
                        date timestamp default now(),
                        primary key(id)
                );
        });

        print "creating indexes: ";

        foreach my $index (qw(
                hosts:name
                backups:hostID
                backups:num
                backups:shareID
                shares:hostID
                shares:name
                files:shareID
                files:path
                files:name
                files:date
                files:size
                archive:dvd_nr
                archive_burned:archive_id
                backup_parts:backup_id,part_nr
        )) {
                do_index($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 (?,?,?,?,?,-1)
});

$sth->{update_backups_size} = $dbh->prepare(qq{
UPDATE backups SET size = ?
WHERE hostID = ? and num = ? and date = ? and type =? and shareid = ?
});

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

my @hosts = keys %{$hosts};
my $host_nr = 0;

foreach my $host_key (@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);
        }

        $host_nr++;
        print "host ", $hosts->{$host_key}->{'host'}, " [", 
                $host_nr, "/", ($#hosts + 1), "]: ";
 
        # 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;

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

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

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

                        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 
                        size = ? and
                        ( date = ? or date = ? or date = ? )
                LIMIT 1
        });

        my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset);
        $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 $key_dst_prev = join(" ", (
                                $shareID,
                                $dir,
                                $path_key,
                                $filesInBackup->{$path_key}->{'mtime'} - $dst_offset,
                                $filesInBackup->{$path_key}->{'size'}
                        ));

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

                        my $found;
                        if (
                                ! defined($beenThere->{$key}) &&
                                ! defined($beenThere->{$key_dst_prev}) &&
                                ! defined($beenThere->{$key_dst_next}) &&
                                ! ($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);
}
