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 |
|
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
123 |
|
|
124 |
unless ($use_hest) { |
unless ($use_hest) { |
125 |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
126 |
return; |
return; |
136 |
print " opening index $use_hest"; |
print " opening index $use_hest"; |
137 |
if ($index_path) { |
if ($index_path) { |
138 |
$hest_db = HyperEstraier::Database->new(); |
$hest_db = HyperEstraier::Database->new(); |
139 |
$hest_db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
140 |
print " directly"; |
print " directly"; |
141 |
} elsif ($index_node_url) { |
} elsif ($index_node_url) { |
142 |
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
153 |
|
|
154 |
my $where = ''; |
my $where = ''; |
155 |
my @data; |
my @data; |
156 |
if ($host_id && $share_id && $num) { |
if (defined($host_id) && defined($share_id) && defined($num)) { |
157 |
$where = qq{ |
$where = qq{ |
158 |
WHERE |
WHERE |
159 |
hosts.id = ? AND |
hosts.id = ? AND |
207 |
my $fid = $row->{'fid'} || die "no fid?"; |
my $fid = $row->{'fid'} || die "no fid?"; |
208 |
my $uri = 'file:///' . $fid; |
my $uri = 'file:///' . $fid; |
209 |
|
|
210 |
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
unless ($skip_check) { |
211 |
next unless ($id == -1); |
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
212 |
|
next unless ($id == -1); |
213 |
|
} |
214 |
|
|
215 |
# create a document object |
# create a document object |
216 |
my $doc = HyperEstraier::Document->new; |
my $doc = HyperEstraier::Document->new; |
219 |
$doc->add_attr('@uri', $uri); |
$doc->add_attr('@uri', $uri); |
220 |
|
|
221 |
foreach my $c (@{ $sth->{NAME} }) { |
foreach my $c (@{ $sth->{NAME} }) { |
222 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
$doc->add_attr($c, $row->{$c}) if (defined($row->{$c})); |
223 |
} |
} |
224 |
|
|
225 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
266 |
|
|
267 |
|
|
268 |
## update index ## |
## update index ## |
269 |
if (($opt{i} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
if (($opt{i} || $opt{j} || ($index_path && ! -e $TopDir . $index_path)) && !$opt{c}) { |
270 |
# update all |
# update all |
271 |
print "force update of HyperEstraier index "; |
print "force update of HyperEstraier index "; |
272 |
print "importing existing data" unless (-e $index_path); |
print "importing existing data" unless (-e $TopDir . $index_path); |
273 |
print "by -i flag" if ($opt{i}); |
print "by -i flag" if ($opt{i}); |
274 |
|
print "by -j flag" if ($opt{j}); |
275 |
print "\n"; |
print "\n"; |
276 |
hest_update(); |
hest_update(); |
277 |
} |
} |
280 |
if ($opt{c}) { |
if ($opt{c}) { |
281 |
sub do_index { |
sub do_index { |
282 |
my $index = shift || return; |
my $index = shift || return; |
283 |
my ($table,$col,$unique) = split(/_/, $index); |
my ($table,$col,$unique) = split(/:/, $index); |
284 |
$unique ||= ''; |
$unique ||= ''; |
285 |
$index =~ s/,/_/g; |
$index =~ s/\W+/_/g; |
286 |
|
print "$index on $table($col)" . ( $unique ? "u" : "" ) . " "; |
287 |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
288 |
} |
} |
289 |
|
|
290 |
print "creating tables...\n"; |
print "creating tables...\n"; |
291 |
|
|
292 |
$dbh->do(qq{ |
$dbh->do( qq{ |
293 |
create table hosts ( |
create table hosts ( |
294 |
ID SERIAL PRIMARY KEY, |
ID SERIAL PRIMARY KEY, |
295 |
name VARCHAR(30) NOT NULL, |
name VARCHAR(30) NOT NULL, |
296 |
IP VARCHAR(15) |
IP VARCHAR(15) |
297 |
); |
); |
298 |
}); |
|
|
|
|
|
$dbh->do(qq{ |
|
299 |
create table shares ( |
create table shares ( |
300 |
ID SERIAL PRIMARY KEY, |
ID SERIAL PRIMARY KEY, |
301 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
302 |
name VARCHAR(30) NOT NULL, |
name VARCHAR(30) NOT NULL, |
303 |
share VARCHAR(200) NOT NULL, |
share VARCHAR(200) NOT NULL |
|
localpath VARCHAR(200) |
|
304 |
); |
); |
305 |
}); |
|
306 |
|
create table dvds ( |
307 |
$dbh->do(qq{ |
ID SERIAL PRIMARY KEY, |
308 |
|
num INTEGER NOT NULL, |
309 |
|
name VARCHAR(255) NOT NULL, |
310 |
|
mjesto VARCHAR(255) |
311 |
|
); |
312 |
|
|
313 |
create table backups ( |
create table backups ( |
314 |
|
id serial, |
315 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
316 |
num INTEGER NOT NULL, |
num INTEGER NOT NULL, |
317 |
date integer NOT NULL, |
date integer NOT NULL, |
318 |
type CHAR(4) not null, |
type CHAR(4) not null, |
319 |
shareID integer not null references shares(id), |
shareID integer not null references shares(id), |
320 |
size integer not null, |
size bigint not null, |
321 |
PRIMARY KEY(hostID, num, shareID) |
inc_size bigint not null default -1, |
322 |
|
inc_deleted boolean default false, |
323 |
|
parts integer not null default 1, |
324 |
|
PRIMARY KEY(id) |
325 |
); |
); |
|
}); |
|
326 |
|
|
327 |
#do_index('backups_hostid,num_unique'); |
create table files ( |
328 |
|
ID SERIAL, |
329 |
|
shareID INTEGER NOT NULL references shares(id), |
330 |
|
backupNum INTEGER NOT NULL, |
331 |
|
name VARCHAR(255) NOT NULL, |
332 |
|
path VARCHAR(255) NOT NULL, |
333 |
|
date integer NOT NULL, |
334 |
|
type INTEGER NOT NULL, |
335 |
|
size bigint NOT NULL, |
336 |
|
primary key(id) |
337 |
|
); |
338 |
|
|
339 |
$dbh->do(qq{ |
create table archive ( |
340 |
create table dvds ( |
id serial, |
341 |
ID SERIAL PRIMARY KEY, |
dvd_nr int not null, |
342 |
num INTEGER NOT NULL, |
total_size bigint default -1, |
343 |
name VARCHAR(255) NOT NULL, |
note text, |
344 |
mjesto VARCHAR(255) |
username varchar(20) not null, |
345 |
|
date timestamp default now(), |
346 |
|
primary key(id) |
347 |
|
); |
348 |
|
|
349 |
|
create table archive_backup ( |
350 |
|
archive_id int not null references archive(id) on delete cascade, |
351 |
|
backup_id int not null references backups(id), |
352 |
|
primary key(archive_id, backup_id) |
353 |
); |
); |
|
}); |
|
354 |
|
|
355 |
$dbh->do(qq{ |
create table archive_burned ( |
356 |
create table files ( |
archive_id int references archive(id), |
357 |
ID SERIAL PRIMARY KEY, |
date timestamp default now(), |
358 |
shareID INTEGER NOT NULL references shares(id), |
part int not null default 1, |
359 |
backupNum INTEGER NOT NULL, |
copy int not null default 1, |
360 |
name VARCHAR(255) NOT NULL, |
iso_size bigint default -1 |
361 |
path VARCHAR(255) NOT NULL, |
); |
362 |
date integer NOT NULL, |
|
363 |
type INTEGER NOT NULL, |
create table backup_parts ( |
364 |
size INTEGER NOT NULL, |
id serial, |
365 |
dvdid INTEGER references dvds(id) |
backup_id int references backups(id), |
366 |
|
part_nr int not null check (part_nr > 0), |
367 |
|
tar_size bigint not null check (tar_size > 0), |
368 |
|
size bigint not null check (size > 0), |
369 |
|
md5 text not null, |
370 |
|
items int not null check (items > 0), |
371 |
|
date timestamp default now(), |
372 |
|
primary key(id) |
373 |
); |
); |
374 |
}); |
}); |
375 |
|
|
376 |
print "creating indexes:"; |
print "creating indexes: "; |
377 |
|
|
378 |
foreach my $index (qw( |
foreach my $index (qw( |
379 |
hosts_name |
hosts:name |
380 |
backups_hostID |
backups:hostID |
381 |
backups_num |
backups:num |
382 |
shares_hostID |
backups:shareID |
383 |
shares_name |
shares:hostID |
384 |
files_shareID |
shares:name |
385 |
files_path |
files:shareID |
386 |
files_name |
files:path |
387 |
files_date |
files:name |
388 |
files_size |
files:date |
389 |
|
files:size |
390 |
|
archive:dvd_nr |
391 |
|
archive_burned:archive_id |
392 |
|
backup_parts:backup_id,part_nr |
393 |
)) { |
)) { |
|
print " $index"; |
|
394 |
do_index($index); |
do_index($index); |
395 |
} |
} |
396 |
|
|
397 |
|
print " creating sequence: "; |
398 |
|
foreach my $seq (qw/dvd_nr/) { |
399 |
|
print "$seq "; |
400 |
|
$dbh->do( qq{ CREATE SEQUENCE $seq } ); |
401 |
|
} |
402 |
|
|
403 |
|
|
404 |
print "...\n"; |
print "...\n"; |
405 |
|
|
406 |
$dbh->commit; |
$dbh->commit; |
444 |
|
|
445 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
$sth->{insert_backups} = $dbh->prepare(qq{ |
446 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
447 |
VALUES (?,?,?,?,?,?) |
VALUES (?,?,?,?,?,-1) |
448 |
|
}); |
449 |
|
|
450 |
|
$sth->{update_backups_size} = $dbh->prepare(qq{ |
451 |
|
UPDATE backups SET size = ? |
452 |
|
WHERE hostID = ? and num = ? and date = ? and type =? and shareid = ? |
453 |
}); |
}); |
454 |
|
|
455 |
$sth->{insert_files} = $dbh->prepare(qq{ |
$sth->{insert_files} = $dbh->prepare(qq{ |
515 |
# dump some log |
# dump some log |
516 |
print curr_time," ", $share; |
print curr_time," ", $share; |
517 |
|
|
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
|
|
|
|
518 |
$sth->{insert_backups}->execute( |
$sth->{insert_backups}->execute( |
519 |
$hostID, |
$hostID, |
520 |
$backupNum, |
$backupNum, |
521 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
522 |
$backup->{'type'}, |
substr($backup->{'type'},0,4), |
523 |
$shareID, |
$shareID, |
524 |
|
); |
525 |
|
|
526 |
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
527 |
|
|
528 |
|
$sth->{update_backups_size}->execute( |
529 |
$size, |
$size, |
530 |
|
$hostID, |
531 |
|
$backupNum, |
532 |
|
$backup->{'endTime'}, |
533 |
|
substr($backup->{'type'},0,4), |
534 |
|
$shareID, |
535 |
); |
); |
536 |
|
|
537 |
print " commit"; |
print " commit"; |
574 |
|
|
575 |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
576 |
INSERT INTO shares |
INSERT INTO shares |
577 |
(hostID,name,share,localpath) |
(hostID,name,share) |
578 |
VALUES (?,?,?,?) |
VALUES (?,?,?) |
579 |
}); |
}); |
580 |
|
|
581 |
my $drop_down = $hostname . '/' . $share; |
my $drop_down = $hostname . '/' . $share; |
582 |
$drop_down =~ s#//+#/#g; |
$drop_down =~ s#//+#/#g; |
583 |
|
|
584 |
$sth->{insert_share}->execute($hostID,$share, $drop_down ,undef); |
$sth->{insert_share}->execute($hostID,$share, $drop_down); |
585 |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
586 |
} |
} |
587 |
|
|
598 |
SELECT 1 FROM files |
SELECT 1 FROM files |
599 |
WHERE shareID = ? and |
WHERE shareID = ? and |
600 |
path = ? and |
path = ? and |
601 |
date = ? and |
size = ? and |
602 |
size = ? |
( date = ? or date = ? or date = ? ) |
603 |
LIMIT 1 |
LIMIT 1 |
604 |
}); |
}); |
605 |
|
|
606 |
my @param = ($shareID,$path,$date,$size); |
my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset); |
607 |
$sth->{file_in_db}->execute(@param); |
$sth->{file_in_db}->execute(@param); |
608 |
my $rows = $sth->{file_in_db}->rows; |
my $rows = $sth->{file_in_db}->rows; |
609 |
print STDERR "## found_in_db($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); |
653 |
$filesInBackup->{$path_key}->{'size'} |
$filesInBackup->{$path_key}->{'size'} |
654 |
)); |
)); |
655 |
|
|
656 |
|
my $key_dst_prev = join(" ", ( |
657 |
|
$shareID, |
658 |
|
$dir, |
659 |
|
$path_key, |
660 |
|
$filesInBackup->{$path_key}->{'mtime'} - $dst_offset, |
661 |
|
$filesInBackup->{$path_key}->{'size'} |
662 |
|
)); |
663 |
|
|
664 |
|
my $key_dst_next = join(" ", ( |
665 |
|
$shareID, |
666 |
|
$dir, |
667 |
|
$path_key, |
668 |
|
$filesInBackup->{$path_key}->{'mtime'} + $dst_offset, |
669 |
|
$filesInBackup->{$path_key}->{'size'} |
670 |
|
)); |
671 |
|
|
672 |
my $found; |
my $found; |
673 |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
if ( |
674 |
|
! defined($beenThere->{$key}) && |
675 |
|
! defined($beenThere->{$key_dst_prev}) && |
676 |
|
! defined($beenThere->{$key_dst_next}) && |
677 |
|
! ($found = found_in_db($key, @data)) |
678 |
|
) { |
679 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
680 |
|
|
681 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |