1 #!/usr/local/bin/perl -w
4 use lib "__INSTALLDIR__/lib";
11 use Time::HiRes qw/time/;
13 use POSIX qw/strftime/;
14 use BackupPC::SearchLib;
16 use constant BPC_FTYPE_DIR => 5;
17 use constant EST_CHUNK => 100000;
24 my $pidfile = new File::Pid;
26 if (my $pid = $pidfile->running ) {
27 die "$0 already running: $pid\n";
28 } elsif ($pidfile->pid ne $$) {
30 $pidfile = new File::Pid;
33 print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n";
35 my $t_fmt = '%Y-%m-%d %H:%M:%S';
38 my $bpc = BackupPC::Lib->new || die;
39 my %Conf = $bpc->Conf();
40 my $TopDir = $bpc->TopDir();
43 my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
44 my $user = $Conf{SearchUser} || '';
46 my $use_hest = $Conf{HyperEstraierIndex};
47 my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest);
49 my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
53 if ( !getopts("cdm:v:ij", \%opt ) ) {
55 usage: $0 [-c|-d] [-m num] [-v|-v level] [-i]
58 -c create database on first use
59 -d delete database before import
60 -m num import just num increments for one host
61 -v num set verbosity (debug) level (default $debug)
62 -i update Hyper Estraier full text index
63 -j update full text, don't check existing files
65 Option -j is variation on -i. It will allow faster initial creation
66 of full-text index from existing database.
73 print "Debug level at $opt{v}\n";
80 my $t = shift || return;
82 my ($ss,$mm,$hh) = gmtime($t);
83 $out .= "${hh}h" if ($hh);
84 $out .= sprintf("%02d:%02d", $mm,$ss);
89 return strftime($t_fmt,localtime());
98 print "\nCaught a SIG$sig--syncing database and shutting down\n";
105 $SIG{'INT'} = \&signal;
106 $SIG{'QUIT'} = \&signal;
110 my ($host_id, $share_id, $num) = @_;
112 my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n";
115 print STDERR "HyperEstraier support not enabled in configuration\n";
119 print curr_time," updating HyperEstraier:";
126 print " opening index $use_hest";
128 $hest_db = HyperEstraier::Database->new();
129 $hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT);
131 } elsif ($index_node_url) {
132 $hest_node ||= HyperEstraier::Node->new($index_node_url);
133 $hest_node->set_auth('admin', 'admin');
134 print " via node URL";
136 die "don't know how to use HyperEstraier Index $use_hest";
138 print " increment is " . EST_CHUNK . " files:";
146 if ($host_id && $share_id && $num) {
153 @data = ( $host_id, $share_id, $num );
156 my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset);
158 my $sth = $dbh->prepare(qq{
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,
170 files.shareid AS shareid,
171 backups.date AS backup_date
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
180 $sth->execute(@data);
181 $results = $sth->rows;
184 print " - no new files\n";
189 my $t = shift || return;
190 my $iso = BackupPC::Lib::timeStamp($t);
195 while (my $row = $sth->fetchrow_hashref()) {
197 my $fid = $row->{'fid'} || die "no fid?";
198 my $uri = 'file:///' . $fid;
200 unless ($skip_check) {
201 my $id = ($hest_db || $hest_node)->uri_to_id($uri);
202 next unless ($id == -1);
205 # create a document object
206 my $doc = HyperEstraier::Document->new;
208 # add attributes to the document object
209 $doc->add_attr('@uri', $uri);
211 foreach my $c (@{ $sth->{NAME} }) {
212 $doc->add_attr($c, $row->{$c}) if ($row->{$c});
215 #$doc->add_attr('@cdate', fmt_date($row->{'date'}));
217 # 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);
223 print STDERR $doc->dump_draft,"\n" if ($debug > 1);
225 # register the document object to the database
227 $hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN);
228 } elsif ($hest_node) {
229 $hest_node->put_doc($doc);
237 $hest_db->sync() if ($index_path);
239 $offset += EST_CHUNK;
241 } while ($results == EST_CHUNK);
248 my $dur = (time() - $t) || 1;
249 printf(" [%.2f/s dur: %s]\n",
259 if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) {
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 print "by -j flag" if ($opt{j});
272 my $index = shift || return;
273 my ($table,$col,$unique) = split(/:/, $index);
276 print "$index on $table($col)" . ( $unique ? "u" : "" ) . " ";
277 $dbh->do(qq{ create $unique index $index on $table($col) });
280 print "creating tables...\n";
284 ID SERIAL PRIMARY KEY,
285 name VARCHAR(30) NOT NULL,
291 create table shares (
292 ID SERIAL PRIMARY KEY,
293 hostID INTEGER NOT NULL references hosts(id),
294 name VARCHAR(30) NOT NULL,
295 share VARCHAR(200) NOT NULL
301 ID SERIAL PRIMARY KEY,
302 num INTEGER NOT NULL,
303 name VARCHAR(255) NOT NULL,
309 create table backups (
311 hostID INTEGER NOT NULL references hosts(id),
312 num INTEGER NOT NULL,
313 date integer NOT NULL,
314 type CHAR(4) not null,
315 shareID integer not null references shares(id),
316 size bigint not null,
317 inc_size bigint not null default -1,
318 inc_deleted boolean default false,
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,
339 create table archive (
342 total_size bigint default -1,
344 username varchar(20) not null,
345 date timestamp default now(),
352 create table archive_backup
354 archive_id int not null references archive(id),
355 backup_id int not null references backups(id),
356 primary key(archive_id, backup_id)
360 print "creating indexes:";
362 foreach my $index (qw(
384 ## delete data before inseting ##
387 foreach my $table (qw(files dvds backups shares hosts)) {
389 $dbh->do(qq{ DELETE FROM $table });
396 ## insert new values ##
399 $hosts = $bpc->HostInfoRead();
405 $sth->{insert_hosts} = $dbh->prepare(qq{
406 INSERT INTO hosts (name, IP) VALUES (?,?)
409 $sth->{hosts_by_name} = $dbh->prepare(qq{
410 SELECT ID FROM hosts WHERE name=?
413 $sth->{backups_count} = $dbh->prepare(qq{
416 WHERE hostID=? AND num=? AND shareid=?
419 $sth->{insert_backups} = $dbh->prepare(qq{
420 INSERT INTO backups (hostID, num, date, type, shareid, size)
424 $sth->{insert_files} = $dbh->prepare(qq{
426 (shareID, backupNum, name, path, date, type, size)
427 VALUES (?,?,?,?,?,?,?)
430 foreach my $host_key (keys %{$hosts}) {
432 my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key";
434 $sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'});
436 unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) {
437 $sth->{insert_hosts}->execute(
438 $hosts->{$host_key}->{'host'},
439 $hosts->{$host_key}->{'ip'}
442 $hostID = $dbh->last_insert_id(undef,undef,'hosts',undef);
445 print "host ".$hosts->{$host_key}->{'host'}.": ";
447 # get backups for a host
448 my @backups = $bpc->BackupInfoRead($hostname);
449 my $incs = scalar @backups;
450 print "$incs increments\n";
455 foreach my $backup (@backups) {
458 last if ($opt{m} && $inc_nr > $opt{m});
460 my $backupNum = $backup->{'num'};
461 my @backupShares = ();
463 printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n",
464 $hosts->{$host_key}->{'host'},
465 $inc_nr, $incs, $backupNum,
466 $backup->{type} || '?',
467 $backup->{nFilesNew} || '?', $backup->{nFiles} || '?',
468 strftime($t_fmt,localtime($backup->{startTime})),
469 fmt_time($backup->{endTime} - $backup->{startTime})
472 my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1);
473 foreach my $share ($files->shareList($backupNum)) {
477 $shareID = getShareID($share, $hostID, $hostname);
479 $sth->{backups_count}->execute($hostID, $backupNum, $shareID);
480 my ($count) = $sth->{backups_count}->fetchrow_array();
481 # skip if allready in database!
482 next if ($count > 0);
485 print curr_time," ", $share;
487 my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID);
489 $sth->{insert_backups}->execute(
492 $backup->{'endTime'},
493 substr($backup->{'type'},0,4),
501 my $dur = (time() - $t) || 1;
502 printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n",
504 ($size / 1024 / 1024),
509 hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0);
518 print "total duration: ",fmt_time(time() - $start_t),"\n";
524 my ($share, $hostID, $hostname) = @_;
526 $sth->{share_id} ||= $dbh->prepare(qq{
527 SELECT ID FROM shares WHERE hostID=? AND name=?
530 $sth->{share_id}->execute($hostID,$share);
532 my ($id) = $sth->{share_id}->fetchrow_array();
534 return $id if (defined($id));
536 $sth->{insert_share} ||= $dbh->prepare(qq{
542 my $drop_down = $hostname . '/' . $share;
543 $drop_down =~ s#//+#/#g;
545 $sth->{insert_share}->execute($hostID,$share, $drop_down);
546 return $dbh->last_insert_id(undef,undef,'shares',undef);
554 my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_;
556 return $beenThere->{$key} if (defined($beenThere->{$key}));
558 $sth->{file_in_db} ||= $dbh->prepare(qq{
560 WHERE shareID = ? and
567 my @param = ($shareID,$path,$date,$size);
568 $sth->{file_in_db}->execute(@param);
569 my $rows = $sth->{file_in_db}->rows;
570 print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3);
572 $beenThere->{$key}++;
574 $sth->{'insert_files'}->execute(@data) unless ($rows);
578 ####################################################
579 # recursing through filesystem structure and #
580 # and returning flattened files list #
581 ####################################################
582 sub recurseDir($$$$$$$$) {
584 my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_;
586 print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1);
588 my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0);
593 print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2);
594 my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir);
596 # first, add all the entries in current directory
597 foreach my $path_key (keys %{$filesInBackup}) {
598 print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3);
603 $filesInBackup->{$path_key}->{'relPath'},
604 $filesInBackup->{$path_key}->{'mtime'},
605 $filesInBackup->{$path_key}->{'type'},
606 $filesInBackup->{$path_key}->{'size'}
609 my $key = join(" ", (
613 $filesInBackup->{$path_key}->{'mtime'},
614 $filesInBackup->{$path_key}->{'size'}
618 if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) {
619 print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2);
621 if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
622 $new_dirs++ unless ($found);
623 print STDERR " dir\n" if ($debug >= 2);
625 $new_files++ unless ($found);
626 print STDERR " file\n" if ($debug >= 2);
628 $size += $filesInBackup->{$path_key}->{'size'} || 0;
631 if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) {
634 my $full_path = $dir . '/' . $path_key;
635 push @stack, $full_path;
636 print STDERR "### store to stack: $full_path\n" if ($debug >= 3);
638 # my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key});
650 print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2);
652 while ( my $dir = shift @stack ) {
653 my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID);
654 print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1);
663 return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size);