11 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
12 |
use File::Pid; |
use File::Pid; |
13 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
14 |
|
use BackupPC::SearchLib; |
15 |
|
|
16 |
use constant BPC_FTYPE_DIR => 5; |
use constant BPC_FTYPE_DIR => 5; |
17 |
use constant EST_CHUNK => 100000; |
use constant EST_CHUNK => 100000; |
18 |
|
|
19 |
|
# daylight saving time change offset for 1h |
20 |
|
my $dst_offset = 60 * 60; |
21 |
|
|
22 |
my $debug = 0; |
my $debug = 0; |
23 |
$|=1; |
$|=1; |
24 |
|
|
47 |
my $user = $Conf{SearchUser} || ''; |
my $user = $Conf{SearchUser} || ''; |
48 |
|
|
49 |
my $use_hest = $Conf{HyperEstraierIndex}; |
my $use_hest = $Conf{HyperEstraierIndex}; |
50 |
my ($index_path, $index_node_url) = getHyperEstraier_url($use_hest); |
my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest); |
51 |
|
|
52 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
53 |
|
|
54 |
my %opt; |
my %opt; |
55 |
|
|
56 |
if ( !getopts("cdm:v:i", \%opt ) ) { |
if ( !getopts("cdm:v:ijf", \%opt ) ) { |
57 |
print STDERR <<EOF; |
print STDERR <<EOF; |
58 |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i|-j|-f] |
59 |
|
|
60 |
Options: |
Options: |
61 |
-c create database on first use |
-c create database on first use |
62 |
-d delete database before import |
-d delete database before import |
63 |
-m num import just num increments for one host |
-m num import just num increments for one host |
64 |
-v num set verbosity (debug) level (default $debug) |
-v num set verbosity (debug) level (default $debug) |
65 |
-i update HyperEstraier full text index |
-i update Hyper Estraier full text index |
66 |
|
-j update full text, don't check existing files |
67 |
|
-f don't do anything with full text index |
68 |
|
|
69 |
|
Option -j is variation on -i. It will allow faster initial creation |
70 |
|
of full-text index from existing database. |
71 |
|
|
72 |
|
Option -f will create database which is out of sync with full text index. You |
73 |
|
will have to re-run $0 with -i to fix it. |
74 |
|
|
75 |
EOF |
EOF |
76 |
exit 1; |
exit 1; |
77 |
} |
} |
79 |
if ($opt{v}) { |
if ($opt{v}) { |
80 |
print "Debug level at $opt{v}\n"; |
print "Debug level at $opt{v}\n"; |
81 |
$debug = $opt{v}; |
$debug = $opt{v}; |
82 |
|
} elsif ($opt{f}) { |
83 |
|
print "WARNING: disabling full-text index update. You need to re-run $0 -j !\n"; |
84 |
|
($use_hest, $index_path, $index_node_url) = (undef, undef, undef); |
85 |
} |
} |
86 |
|
|
87 |
#---- subs ---- |
#---- subs ---- |
119 |
|
|
120 |
my ($host_id, $share_id, $num) = @_; |
my ($host_id, $share_id, $num) = @_; |
121 |
|
|
122 |
unless ($use_hest) { |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
123 |
|
|
124 |
|
unless (defined($use_hest)) { |
125 |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
126 |
|
$use_hest = 0; |
127 |
return; |
return; |
128 |
} |
} |
129 |
|
|
137 |
print " opening index $use_hest"; |
print " opening index $use_hest"; |
138 |
if ($index_path) { |
if ($index_path) { |
139 |
$hest_db = HyperEstraier::Database->new(); |
$hest_db = HyperEstraier::Database->new(); |
140 |
$hest_db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
141 |
print " directly"; |
print " directly"; |
142 |
} elsif ($index_node_url) { |
} elsif ($index_node_url) { |
143 |
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
154 |
|
|
155 |
my $where = ''; |
my $where = ''; |
156 |
my @data; |
my @data; |
157 |
if ($host_id && $share_id && $num) { |
if (defined($host_id) && defined($share_id) && defined($num)) { |
158 |
$where = qq{ |
$where = qq{ |
159 |
WHERE |
WHERE |
160 |
hosts.id = ? AND |
hosts.id = ? AND |
208 |
my $fid = $row->{'fid'} || die "no fid?"; |
my $fid = $row->{'fid'} || die "no fid?"; |
209 |
my $uri = 'file:///' . $fid; |
my $uri = 'file:///' . $fid; |
210 |
|
|
211 |
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
unless ($skip_check) { |
212 |
next unless ($id == -1); |
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
213 |
|
next unless ($id == -1); |
214 |
|
} |
215 |
|
|
216 |
# create a document object |
# create a document object |
217 |
my $doc = HyperEstraier::Document->new; |
my $doc = HyperEstraier::Document->new; |
220 |
$doc->add_attr('@uri', $uri); |
$doc->add_attr('@uri', $uri); |
221 |
|
|
222 |
foreach my $c (@{ $sth->{NAME} }) { |
foreach my $c (@{ $sth->{NAME} }) { |
223 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
$doc->add_attr($c, $row->{$c}) if (defined($row->{$c})); |
224 |
} |
} |
225 |
|
|
226 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
267 |
|
|
268 |
|
|
269 |
## update index ## |
## update index ## |
270 |
if (($opt{i} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
if (($opt{i} || $opt{j} || ($index_path && ! -e $TopDir . $index_path)) && !$opt{c}) { |
271 |
# update all |
# update all |
272 |
print "force update of HyperEstraier index "; |
print "force update of HyperEstraier index "; |
273 |
print "importing existing data" unless (-e $index_path); |
print "importing existing data" unless (-e $TopDir . $index_path); |
274 |
print "by -i flag" if ($opt{i}); |
print "by -i flag" if ($opt{i}); |
275 |
|
print "by -j flag" if ($opt{j}); |
276 |
print "\n"; |
print "\n"; |
277 |
hest_update(); |
hest_update(); |
278 |
} |
} |
281 |
if ($opt{c}) { |
if ($opt{c}) { |
282 |
sub do_index { |
sub do_index { |
283 |
my $index = shift || return; |
my $index = shift || return; |
284 |
my ($table,$col,$unique) = split(/_/, $index); |
my ($table,$col,$unique) = split(/:/, $index); |
285 |
$unique ||= ''; |
$unique ||= ''; |
286 |
$index =~ s/,/_/g; |
$index =~ s/\W+/_/g; |
287 |
|
print "$index on $table($col)" . ( $unique ? "u" : "" ) . " "; |
288 |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
289 |
} |
} |
290 |
|
|
291 |
print "creating tables...\n"; |
print "creating tables...\n"; |
292 |
|
|
293 |
$dbh->do(qq{ |
$dbh->do( qq{ |
294 |
create table hosts ( |
create table hosts ( |
295 |
ID SERIAL PRIMARY KEY, |
ID SERIAL PRIMARY KEY, |
296 |
name VARCHAR(30) NOT NULL, |
name VARCHAR(30) NOT NULL, |
297 |
IP VARCHAR(15) |
IP VARCHAR(15) |
298 |
); |
); |
299 |
}); |
|
|
|
|
|
$dbh->do(qq{ |
|
300 |
create table shares ( |
create table shares ( |
301 |
ID SERIAL PRIMARY KEY, |
ID SERIAL PRIMARY KEY, |
302 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
303 |
name VARCHAR(30) NOT NULL, |
name VARCHAR(30) NOT NULL, |
304 |
share VARCHAR(200) NOT NULL, |
share VARCHAR(200) NOT NULL |
|
localpath VARCHAR(200) |
|
305 |
); |
); |
306 |
}); |
|
307 |
|
create table dvds ( |
308 |
$dbh->do(qq{ |
ID SERIAL PRIMARY KEY, |
309 |
|
num INTEGER NOT NULL, |
310 |
|
name VARCHAR(255) NOT NULL, |
311 |
|
mjesto VARCHAR(255) |
312 |
|
); |
313 |
|
|
314 |
create table backups ( |
create table backups ( |
315 |
|
id serial, |
316 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
317 |
num INTEGER NOT NULL, |
num INTEGER NOT NULL, |
318 |
date integer NOT NULL, |
date integer NOT NULL, |
319 |
type CHAR(4) not null, |
type CHAR(4) not null, |
320 |
shareID integer not null references shares(id), |
shareID integer not null references shares(id), |
321 |
size integer not null, |
size bigint not null, |
322 |
PRIMARY KEY(hostID, num, shareID) |
inc_size bigint not null default -1, |
323 |
|
inc_deleted boolean default false, |
324 |
|
parts integer not null default 1, |
325 |
|
PRIMARY KEY(id) |
326 |
); |
); |
|
}); |
|
327 |
|
|
328 |
#do_index('backups_hostid,num_unique'); |
create table files ( |
329 |
|
ID SERIAL, |
330 |
|
shareID INTEGER NOT NULL references shares(id), |
331 |
|
backupNum INTEGER NOT NULL, |
332 |
|
name VARCHAR(255) NOT NULL, |
333 |
|
path VARCHAR(255) NOT NULL, |
334 |
|
date integer NOT NULL, |
335 |
|
type INTEGER NOT NULL, |
336 |
|
size bigint NOT NULL, |
337 |
|
primary key(id) |
338 |
|
); |
339 |
|
|
340 |
$dbh->do(qq{ |
create table archive ( |
341 |
create table dvds ( |
id serial, |
342 |
ID SERIAL PRIMARY KEY, |
dvd_nr int not null, |
343 |
num INTEGER NOT NULL, |
total_size bigint default -1, |
344 |
name VARCHAR(255) NOT NULL, |
note text, |
345 |
mjesto VARCHAR(255) |
username varchar(20) not null, |
346 |
|
date timestamp default now(), |
347 |
|
primary key(id) |
348 |
|
); |
349 |
|
|
350 |
|
create table archive_backup ( |
351 |
|
archive_id int not null references archive(id) on delete cascade, |
352 |
|
backup_id int not null references backups(id), |
353 |
|
primary key(archive_id, backup_id) |
354 |
); |
); |
|
}); |
|
355 |
|
|
356 |
$dbh->do(qq{ |
create table archive_burned ( |
357 |
create table files ( |
archive_id int references archive(id), |
358 |
ID SERIAL PRIMARY KEY, |
date timestamp default now(), |
359 |
shareID INTEGER NOT NULL references shares(id), |
part int not null default 1, |
360 |
backupNum INTEGER NOT NULL, |
copy int not null default 1, |
361 |
name VARCHAR(255) NOT NULL, |
iso_size bigint default -1 |
362 |
path VARCHAR(255) NOT NULL, |
); |
363 |
date integer NOT NULL, |
|
364 |
type INTEGER NOT NULL, |
create table backup_parts ( |
365 |
size INTEGER NOT NULL, |
id serial, |
366 |
dvdid INTEGER references dvds(id) |
backup_id int references backups(id), |
367 |
|
part_nr int not null check (part_nr > 0), |
368 |
|
tar_size bigint not null check (tar_size > 0), |
369 |
|
size bigint not null check (size > 0), |
370 |
|
md5 text not null, |
371 |
|
items int not null check (items > 0), |
372 |
|
date timestamp default now(), |
373 |
|
primary key(id) |
374 |
); |
); |
375 |
}); |
}); |
376 |
|
|
377 |
print "creating indexes:"; |
print "creating indexes: "; |
378 |
|
|
379 |
foreach my $index (qw( |
foreach my $index (qw( |
380 |
hosts_name |
hosts:name |
381 |
backups_hostID |
backups:hostID |
382 |
backups_num |
backups:num |
383 |
shares_hostID |
backups:shareID |
384 |
shares_name |
shares:hostID |
385 |
files_shareID |
shares:name |
386 |
files_path |
files:shareID |
387 |
files_name |
files:path |
388 |
files_date |
files:name |
389 |
files_size |
files:date |
390 |
|
files:size |
391 |
|
archive:dvd_nr |
392 |
|
archive_burned:archive_id |
393 |
|
backup_parts:backup_id,part_nr |
394 |
)) { |
)) { |
|
print " $index"; |
|
395 |
do_index($index); |
do_index($index); |
396 |
} |
} |
397 |
|
|
398 |
|
print " creating sequence: "; |
399 |
|
foreach my $seq (qw/dvd_nr/) { |
400 |
|
print "$seq "; |
401 |
|
$dbh->do( qq{ CREATE SEQUENCE $seq } ); |
402 |
|
} |
403 |
|
|
404 |
|
|
405 |
print "...\n"; |
print "...\n"; |
406 |
|
|
407 |
$dbh->commit; |
$dbh->commit; |
445 |
|
|
446 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
$sth->{insert_backups} = $dbh->prepare(qq{ |
447 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
448 |
VALUES (?,?,?,?,?,?) |
VALUES (?,?,?,?,?,-1) |
449 |
|
}); |
450 |
|
|
451 |
|
$sth->{update_backups_size} = $dbh->prepare(qq{ |
452 |
|
UPDATE backups SET size = ? |
453 |
|
WHERE hostID = ? and num = ? and date = ? and type =? and shareid = ? |
454 |
}); |
}); |
455 |
|
|
456 |
$sth->{insert_files} = $dbh->prepare(qq{ |
$sth->{insert_files} = $dbh->prepare(qq{ |
459 |
VALUES (?,?,?,?,?,?,?) |
VALUES (?,?,?,?,?,?,?) |
460 |
}); |
}); |
461 |
|
|
462 |
foreach my $host_key (keys %{$hosts}) { |
my @hosts = keys %{$hosts}; |
463 |
|
my $host_nr = 0; |
464 |
|
|
465 |
|
foreach my $host_key (@hosts) { |
466 |
|
|
467 |
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
468 |
|
|
477 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
478 |
} |
} |
479 |
|
|
480 |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
$host_nr++; |
481 |
|
print "host ", $hosts->{$host_key}->{'host'}, " [", |
482 |
|
$host_nr, "/", ($#hosts + 1), "]: "; |
483 |
|
|
484 |
# get backups for a host |
# get backups for a host |
485 |
my @backups = $bpc->BackupInfoRead($hostname); |
my @backups = $bpc->BackupInfoRead($hostname); |
521 |
# dump some log |
# dump some log |
522 |
print curr_time," ", $share; |
print curr_time," ", $share; |
523 |
|
|
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
|
|
|
|
524 |
$sth->{insert_backups}->execute( |
$sth->{insert_backups}->execute( |
525 |
$hostID, |
$hostID, |
526 |
$backupNum, |
$backupNum, |
527 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
528 |
$backup->{'type'}, |
substr($backup->{'type'},0,4), |
529 |
$shareID, |
$shareID, |
530 |
|
); |
531 |
|
|
532 |
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
533 |
|
|
534 |
|
$sth->{update_backups_size}->execute( |
535 |
$size, |
$size, |
536 |
|
$hostID, |
537 |
|
$backupNum, |
538 |
|
$backup->{'endTime'}, |
539 |
|
substr($backup->{'type'},0,4), |
540 |
|
$shareID, |
541 |
); |
); |
542 |
|
|
543 |
print " commit"; |
print " commit"; |
580 |
|
|
581 |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
582 |
INSERT INTO shares |
INSERT INTO shares |
583 |
(hostID,name,share,localpath) |
(hostID,name,share) |
584 |
VALUES (?,?,?,?) |
VALUES (?,?,?) |
585 |
}); |
}); |
586 |
|
|
587 |
my $drop_down = $hostname . '/' . $share; |
my $drop_down = $hostname . '/' . $share; |
588 |
$drop_down =~ s#//+#/#g; |
$drop_down =~ s#//+#/#g; |
589 |
|
|
590 |
$sth->{insert_share}->execute($hostID,$share, $drop_down ,undef); |
$sth->{insert_share}->execute($hostID,$share, $drop_down); |
591 |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
592 |
} |
} |
593 |
|
|
604 |
SELECT 1 FROM files |
SELECT 1 FROM files |
605 |
WHERE shareID = ? and |
WHERE shareID = ? and |
606 |
path = ? and |
path = ? and |
607 |
date = ? and |
size = ? and |
608 |
size = ? |
( date = ? or date = ? or date = ? ) |
609 |
LIMIT 1 |
LIMIT 1 |
610 |
}); |
}); |
611 |
|
|
612 |
my @param = ($shareID,$path,$date,$size); |
my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset); |
613 |
$sth->{file_in_db}->execute(@param); |
$sth->{file_in_db}->execute(@param); |
614 |
my $rows = $sth->{file_in_db}->rows; |
my $rows = $sth->{file_in_db}->rows; |
615 |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
659 |
$filesInBackup->{$path_key}->{'size'} |
$filesInBackup->{$path_key}->{'size'} |
660 |
)); |
)); |
661 |
|
|
662 |
|
my $key_dst_prev = join(" ", ( |
663 |
|
$shareID, |
664 |
|
$dir, |
665 |
|
$path_key, |
666 |
|
$filesInBackup->{$path_key}->{'mtime'} - $dst_offset, |
667 |
|
$filesInBackup->{$path_key}->{'size'} |
668 |
|
)); |
669 |
|
|
670 |
|
my $key_dst_next = join(" ", ( |
671 |
|
$shareID, |
672 |
|
$dir, |
673 |
|
$path_key, |
674 |
|
$filesInBackup->{$path_key}->{'mtime'} + $dst_offset, |
675 |
|
$filesInBackup->{$path_key}->{'size'} |
676 |
|
)); |
677 |
|
|
678 |
my $found; |
my $found; |
679 |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
if ( |
680 |
|
! defined($beenThere->{$key}) && |
681 |
|
! defined($beenThere->{$key_dst_prev}) && |
682 |
|
! defined($beenThere->{$key_dst_next}) && |
683 |
|
! ($found = found_in_db($key, @data)) |
684 |
|
) { |
685 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
686 |
|
|
687 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |