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 |
|
|
53 |
|
|
54 |
my %opt; |
my %opt; |
55 |
|
|
56 |
if ( !getopts("cdm:v:ij", \%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 |
64 |
-v num set verbosity (debug) level (default $debug) |
-v num set verbosity (debug) level (default $debug) |
65 |
-i update Hyper Estraier full text index |
-i update Hyper Estraier full text index |
66 |
-j update full text, don't check existing files |
-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 |
Option -j is variation on -i. It will allow faster initial creation |
70 |
of full-text index from existing database. |
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 ---- |
121 |
|
|
122 |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
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 (defined($use_hest)) { |
125 |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
126 |
|
$use_hest = 0; |
127 |
return; |
return; |
128 |
} |
} |
129 |
|
|
130 |
|
return unless($use_hest); |
131 |
|
|
132 |
print curr_time," updating HyperEstraier:"; |
print curr_time," updating HyperEstraier:"; |
133 |
|
|
134 |
my $t = time(); |
my $t = time(); |
222 |
$doc->add_attr('@uri', $uri); |
$doc->add_attr('@uri', $uri); |
223 |
|
|
224 |
foreach my $c (@{ $sth->{NAME} }) { |
foreach my $c (@{ $sth->{NAME} }) { |
225 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
$doc->add_attr($c, $row->{$c}) if (defined($row->{$c})); |
226 |
} |
} |
227 |
|
|
228 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
269 |
|
|
270 |
|
|
271 |
## update index ## |
## update index ## |
272 |
if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
if (($opt{i} || $opt{j} || ($index_path && ! -e $TopDir . $index_path)) && !$opt{c}) { |
273 |
# update all |
# update all |
274 |
print "force update of HyperEstraier index "; |
print "force update of HyperEstraier index "; |
275 |
print "importing existing data" unless (-e $index_path); |
print "importing existing data" unless (-e $TopDir . $index_path); |
276 |
print "by -i flag" if ($opt{i}); |
print "by -i flag" if ($opt{i}); |
277 |
print "by -j flag" if ($opt{j}); |
print "by -j flag" if ($opt{j}); |
278 |
print "\n"; |
print "\n"; |
447 |
|
|
448 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
$sth->{insert_backups} = $dbh->prepare(qq{ |
449 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
450 |
VALUES (?,?,?,?,?,?) |
VALUES (?,?,?,?,?,-1) |
451 |
|
}); |
452 |
|
|
453 |
|
$sth->{update_backups_size} = $dbh->prepare(qq{ |
454 |
|
UPDATE backups SET size = ? |
455 |
|
WHERE hostID = ? and num = ? and date = ? and type =? and shareid = ? |
456 |
}); |
}); |
457 |
|
|
458 |
$sth->{insert_files} = $dbh->prepare(qq{ |
$sth->{insert_files} = $dbh->prepare(qq{ |
461 |
VALUES (?,?,?,?,?,?,?) |
VALUES (?,?,?,?,?,?,?) |
462 |
}); |
}); |
463 |
|
|
464 |
foreach my $host_key (keys %{$hosts}) { |
my @hosts = keys %{$hosts}; |
465 |
|
my $host_nr = 0; |
466 |
|
|
467 |
|
foreach my $host_key (@hosts) { |
468 |
|
|
469 |
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"; |
470 |
|
|
479 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
480 |
} |
} |
481 |
|
|
482 |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
$host_nr++; |
483 |
|
print "host ", $hosts->{$host_key}->{'host'}, " [", |
484 |
|
$host_nr, "/", ($#hosts + 1), "]: "; |
485 |
|
|
486 |
# get backups for a host |
# get backups for a host |
487 |
my @backups = $bpc->BackupInfoRead($hostname); |
my @backups = $bpc->BackupInfoRead($hostname); |
523 |
# dump some log |
# dump some log |
524 |
print curr_time," ", $share; |
print curr_time," ", $share; |
525 |
|
|
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
|
|
|
|
526 |
$sth->{insert_backups}->execute( |
$sth->{insert_backups}->execute( |
527 |
$hostID, |
$hostID, |
528 |
$backupNum, |
$backupNum, |
529 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
530 |
substr($backup->{'type'},0,4), |
substr($backup->{'type'},0,4), |
531 |
$shareID, |
$shareID, |
532 |
|
); |
533 |
|
|
534 |
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
535 |
|
|
536 |
|
$sth->{update_backups_size}->execute( |
537 |
$size, |
$size, |
538 |
|
$hostID, |
539 |
|
$backupNum, |
540 |
|
$backup->{'endTime'}, |
541 |
|
substr($backup->{'type'},0,4), |
542 |
|
$shareID, |
543 |
); |
); |
544 |
|
|
545 |
print " commit"; |
print " commit"; |
606 |
SELECT 1 FROM files |
SELECT 1 FROM files |
607 |
WHERE shareID = ? and |
WHERE shareID = ? and |
608 |
path = ? and |
path = ? and |
609 |
date = ? and |
size = ? and |
610 |
size = ? |
( date = ? or date = ? or date = ? ) |
611 |
LIMIT 1 |
LIMIT 1 |
612 |
}); |
}); |
613 |
|
|
614 |
my @param = ($shareID,$path,$date,$size); |
my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset); |
615 |
$sth->{file_in_db}->execute(@param); |
$sth->{file_in_db}->execute(@param); |
616 |
my $rows = $sth->{file_in_db}->rows; |
my $rows = $sth->{file_in_db}->rows; |
617 |
print STDERR "## found_in_db($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); |
661 |
$filesInBackup->{$path_key}->{'size'} |
$filesInBackup->{$path_key}->{'size'} |
662 |
)); |
)); |
663 |
|
|
664 |
|
my $key_dst_prev = join(" ", ( |
665 |
|
$shareID, |
666 |
|
$dir, |
667 |
|
$path_key, |
668 |
|
$filesInBackup->{$path_key}->{'mtime'} - $dst_offset, |
669 |
|
$filesInBackup->{$path_key}->{'size'} |
670 |
|
)); |
671 |
|
|
672 |
|
my $key_dst_next = join(" ", ( |
673 |
|
$shareID, |
674 |
|
$dir, |
675 |
|
$path_key, |
676 |
|
$filesInBackup->{$path_key}->{'mtime'} + $dst_offset, |
677 |
|
$filesInBackup->{$path_key}->{'size'} |
678 |
|
)); |
679 |
|
|
680 |
my $found; |
my $found; |
681 |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
if ( |
682 |
|
! defined($beenThere->{$key}) && |
683 |
|
! defined($beenThere->{$key_dst_prev}) && |
684 |
|
! defined($beenThere->{$key_dst_next}) && |
685 |
|
! ($found = found_in_db($key, @data)) |
686 |
|
) { |
687 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
688 |
|
|
689 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |