13 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
14 |
|
|
15 |
use constant BPC_FTYPE_DIR => 5; |
use constant BPC_FTYPE_DIR => 5; |
16 |
|
use constant EST_SYNC_EVERY => 10000; |
17 |
|
|
18 |
my $debug = 0; |
my $debug = 0; |
19 |
$|=1; |
$|=1; |
41 |
|
|
42 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
43 |
my $user = $Conf{SearchUser} || ''; |
my $user = $Conf{SearchUser} || ''; |
44 |
|
my $index_path = $Conf{HyperEstraierIndex}; |
45 |
|
$index_path = $TopDir . '/' . $index_path; |
46 |
|
$index_path =~ s#//#/#g; |
47 |
|
|
48 |
|
|
49 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
50 |
|
|
51 |
my %opt; |
my %opt; |
52 |
|
|
53 |
if ( !getopts("cdm:v:", \%opt ) ) { |
if ( !getopts("cdm:v:i", \%opt ) ) { |
54 |
print STDERR <<EOF; |
print STDERR <<EOF; |
55 |
usage: $0 [-c|-d] [-m num] [-v|-v level] |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
56 |
|
|
57 |
Options: |
Options: |
58 |
-c create database on first use |
-c create database on first use |
59 |
-d delete database before import |
-d delete database before import |
60 |
-m num import just num increments for one host |
-m num import just num increments for one host |
61 |
-v num set verbosity (debug) level (default $debug) |
-v num set verbosity (debug) level (default $debug) |
62 |
|
-i update HyperEstraier full text index |
63 |
EOF |
EOF |
64 |
exit 1; |
exit 1; |
65 |
} |
} |
66 |
|
|
67 |
###################################create tables############################3 |
if ($opt{v}) { |
68 |
|
print "Debug level at $opt{v}\n"; |
69 |
|
$debug = $opt{v}; |
70 |
|
} |
71 |
|
|
72 |
|
#---- subs ---- |
73 |
|
|
74 |
|
sub fmt_time { |
75 |
|
my $t = shift || return; |
76 |
|
my $out = ""; |
77 |
|
my ($ss,$mm,$hh) = gmtime($t); |
78 |
|
$out .= "${hh}h" if ($hh); |
79 |
|
$out .= sprintf("%02d:%02d", $mm,$ss); |
80 |
|
return $out; |
81 |
|
} |
82 |
|
|
83 |
|
sub curr_time { |
84 |
|
return strftime($t_fmt,localtime()); |
85 |
|
} |
86 |
|
|
87 |
|
my $hest_db; |
88 |
|
|
89 |
|
sub signal { |
90 |
|
my($sig) = @_; |
91 |
|
if ($hest_db) { |
92 |
|
print "\nCaught a SIG$sig--syncing database and shutting down\n"; |
93 |
|
$hest_db->sync(); |
94 |
|
$hest_db->close(); |
95 |
|
} |
96 |
|
exit(0); |
97 |
|
} |
98 |
|
|
99 |
|
$SIG{'INT'} = \&signal; |
100 |
|
$SIG{'QUIT'} = \&signal; |
101 |
|
|
102 |
|
sub hest_update { |
103 |
|
|
104 |
|
my ($host_id, $share_id, $num) = @_; |
105 |
|
|
106 |
|
print curr_time," updating HyperEstraier: select files"; |
107 |
|
|
108 |
|
my $t = time(); |
109 |
|
|
110 |
|
my $where = ''; |
111 |
|
if ($host_id && $share_id && $num) { |
112 |
|
$where = qq{ |
113 |
|
WHERE |
114 |
|
hosts.id = ? AND |
115 |
|
shares.id = ? AND |
116 |
|
files.backupnum = ? |
117 |
|
}; |
118 |
|
} |
119 |
|
|
120 |
|
my $sth = $dbh->prepare(qq{ |
121 |
|
SELECT |
122 |
|
files.id AS fid, |
123 |
|
hosts.name AS hname, |
124 |
|
shares.name AS sname, |
125 |
|
-- shares.share AS sharename, |
126 |
|
files.backupnum AS backupnum, |
127 |
|
-- files.name AS filename, |
128 |
|
files.path AS filepath, |
129 |
|
files.date AS date, |
130 |
|
files.type AS type, |
131 |
|
files.size AS size, |
132 |
|
files.shareid AS shareid, |
133 |
|
backups.date AS backup_date |
134 |
|
FROM files |
135 |
|
INNER JOIN shares ON files.shareID=shares.ID |
136 |
|
INNER JOIN hosts ON hosts.ID = shares.hostID |
137 |
|
INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID |
138 |
|
$where |
139 |
|
}); |
140 |
|
|
141 |
|
$sth->execute(@_); |
142 |
|
my $results = $sth->rows; |
143 |
|
|
144 |
|
if ($results == 0) { |
145 |
|
print " - no files, skipping\n"; |
146 |
|
return; |
147 |
|
} |
148 |
|
|
149 |
|
my $dot = int($results / 15) || 1; |
150 |
|
|
151 |
|
print " $results ($dot/#)"; |
152 |
|
|
153 |
|
sub fmt_date { |
154 |
|
my $t = shift || return; |
155 |
|
my $iso = BackupPC::Lib::timeStamp($t); |
156 |
|
$iso =~ s/\s/T/; |
157 |
|
return $iso; |
158 |
|
} |
159 |
|
|
160 |
|
my $max = int($results / $dot); |
161 |
|
|
162 |
|
print ", opening index $index_path..."; |
163 |
|
use HyperEstraier; |
164 |
|
my $db = HyperEstraier::Database->new(); |
165 |
|
|
166 |
|
# unless ($hest_db) { |
167 |
|
# print " open reader"; |
168 |
|
# $hest_db = HyperEstraier::Database->new(); |
169 |
|
# |
170 |
|
# } |
171 |
|
|
172 |
|
|
173 |
|
$db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
174 |
|
|
175 |
|
my $added = 0; |
176 |
|
|
177 |
|
while (my $row = $sth->fetchrow_hashref()) { |
178 |
|
|
179 |
|
my $fid = $row->{'fid'} || die "no fid?"; |
180 |
|
my $uri = 'file:///' . $fid; |
181 |
|
|
182 |
|
my $id = $db->uri_to_id($uri); |
183 |
|
next unless ($id == -1); |
184 |
|
|
185 |
|
# create a document object |
186 |
|
my $doc = HyperEstraier::Document->new; |
187 |
|
|
188 |
|
# add attributes to the document object |
189 |
|
$doc->add_attr('@uri', $uri); |
190 |
|
|
191 |
|
foreach my $c (@{ $sth->{NAME} }) { |
192 |
|
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
193 |
|
} |
194 |
|
|
195 |
|
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
196 |
|
|
197 |
|
# add the body text to the document object |
198 |
|
my $path = $row->{'filepath'}; |
199 |
|
$doc->add_text($path); |
200 |
|
$path =~ s/(.)/$1 /g; |
201 |
|
$doc->add_hidden_text($path); |
202 |
|
|
203 |
|
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
204 |
|
|
205 |
|
# register the document object to the database |
206 |
|
$db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
207 |
|
|
208 |
|
$added++; |
209 |
|
if ($added % $dot == 0) { |
210 |
|
print "$max "; |
211 |
|
$max--; |
212 |
|
} |
213 |
|
|
214 |
|
if ($added % EST_SYNC_EVERY == 0) { |
215 |
|
print "sync "; |
216 |
|
$db->sync(); |
217 |
|
} |
218 |
|
|
219 |
|
} |
220 |
|
|
221 |
|
print "sync $added new files"; |
222 |
|
$db->sync(); |
223 |
|
print ", close"; |
224 |
|
$db->close(); |
225 |
|
|
226 |
|
my $dur = (time() - $t) || 1; |
227 |
|
printf(" [%.2f/s new %.2f/s dur: %s]\n", |
228 |
|
( $results / $dur ), |
229 |
|
( $added / $dur ), |
230 |
|
fmt_time($dur) |
231 |
|
); |
232 |
|
} |
233 |
|
|
234 |
|
#---- /subs ---- |
235 |
|
|
236 |
|
|
237 |
|
## update index ## |
238 |
|
if (($opt{i} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
239 |
|
# update all |
240 |
|
print "force update of HyperEstraier index "; |
241 |
|
print "importing existing data" unless (-e $index_path); |
242 |
|
print "by -i flag" if ($opt{i}); |
243 |
|
print "\n"; |
244 |
|
hest_update(); |
245 |
|
} |
246 |
|
|
247 |
|
## create tables ## |
248 |
if ($opt{c}) { |
if ($opt{c}) { |
249 |
sub do_index { |
sub do_index { |
250 |
my $index = shift || return; |
my $index = shift || return; |
280 |
num INTEGER NOT NULL, |
num INTEGER NOT NULL, |
281 |
date integer NOT NULL, |
date integer NOT NULL, |
282 |
type CHAR(4) not null, |
type CHAR(4) not null, |
283 |
PRIMARY KEY(hostID, num) |
shareID integer not null references shares(id), |
284 |
|
size integer not null, |
285 |
|
PRIMARY KEY(hostID, num, shareID) |
286 |
); |
); |
287 |
}); |
}); |
288 |
|
|
289 |
do_index('backups_hostid,num_unique'); |
#do_index('backups_hostid,num_unique'); |
290 |
|
|
291 |
$dbh->do(qq{ |
$dbh->do(qq{ |
292 |
create table dvds ( |
create table dvds ( |
334 |
|
|
335 |
} |
} |
336 |
|
|
337 |
|
## delete data before inseting ## |
338 |
if ($opt{d}) { |
if ($opt{d}) { |
339 |
print "deleting "; |
print "deleting "; |
340 |
foreach my $table (qw(files dvds backups shares hosts)) { |
foreach my $table (qw(files dvds backups shares hosts)) { |
346 |
$dbh->commit; |
$dbh->commit; |
347 |
} |
} |
348 |
|
|
349 |
if ($opt{v}) { |
## insert new values ## |
|
print "Debug level at $opt{v}\n"; |
|
|
$debug = $opt{v}; |
|
|
} |
|
|
|
|
|
#################################INSERT VALUES############################# |
|
350 |
|
|
351 |
# get hosts |
# get hosts |
352 |
$hosts = $bpc->HostInfoRead(); |
$hosts = $bpc->HostInfoRead(); |
363 |
SELECT ID FROM hosts WHERE name=? |
SELECT ID FROM hosts WHERE name=? |
364 |
}); |
}); |
365 |
|
|
366 |
$sth->{backups_broj} = $dbh->prepare(qq{ |
$sth->{backups_count} = $dbh->prepare(qq{ |
367 |
SELECT COUNT(*) |
SELECT COUNT(*) |
368 |
FROM backups |
FROM backups |
369 |
WHERE hostID=? AND num=? |
WHERE hostID=? AND num=? AND shareid=? |
370 |
}); |
}); |
371 |
|
|
372 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
$sth->{insert_backups} = $dbh->prepare(qq{ |
373 |
INSERT INTO backups (hostID, num, date, type) |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
374 |
VALUES (?,?,?,?) |
VALUES (?,?,?,?,?,?) |
375 |
}); |
}); |
376 |
|
|
377 |
$sth->{insert_files} = $dbh->prepare(qq{ |
$sth->{insert_files} = $dbh->prepare(qq{ |
380 |
VALUES (?,?,?,?,?,?,?) |
VALUES (?,?,?,?,?,?,?) |
381 |
}); |
}); |
382 |
|
|
|
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; |
|
|
} |
|
|
|
|
383 |
foreach my $host_key (keys %{$hosts}) { |
foreach my $host_key (keys %{$hosts}) { |
384 |
|
|
385 |
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"; |
395 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
396 |
} |
} |
397 |
|
|
398 |
print("host ".$hosts->{$host_key}->{'host'}.": "); |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
399 |
|
|
400 |
# get backups for a host |
# get backups for a host |
401 |
my @backups = $bpc->BackupInfoRead($hostname); |
my @backups = $bpc->BackupInfoRead($hostname); |
402 |
print scalar @backups, " increments\n"; |
my $incs = scalar @backups; |
403 |
|
print "$incs increments\n"; |
404 |
|
|
405 |
my $inc_nr = 0; |
my $inc_nr = 0; |
406 |
|
$beenThere = {}; |
407 |
|
|
408 |
foreach my $backup (@backups) { |
foreach my $backup (@backups) { |
409 |
|
|
413 |
my $backupNum = $backup->{'num'}; |
my $backupNum = $backup->{'num'}; |
414 |
my @backupShares = (); |
my @backupShares = (); |
415 |
|
|
416 |
print $hosts->{$host_key}->{'host'}, |
printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", |
417 |
"\t#$backupNum\t", $backup->{type} || '?', " ", |
$hosts->{$host_key}->{'host'}, |
418 |
$backup->{nFilesNew} || '?', "/", $backup->{nFiles} || '?', |
$inc_nr, $incs, $backupNum, |
419 |
" files (date: ", |
$backup->{type} || '?', |
420 |
|
$backup->{nFilesNew} || '?', $backup->{nFiles} || '?', |
421 |
strftime($t_fmt,localtime($backup->{startTime})), |
strftime($t_fmt,localtime($backup->{startTime})), |
422 |
" dur: ", |
fmt_time($backup->{endTime} - $backup->{startTime}) |
|
fmt_time($backup->{endTime} - $backup->{startTime}), |
|
|
")\n"; |
|
|
|
|
|
$sth->{backups_broj}->execute($hostID, $backupNum); |
|
|
my ($broj) = $sth->{backups_broj}->fetchrow_array(); |
|
|
next if ($broj > 0); |
|
|
|
|
|
$sth->{insert_backups}->execute( |
|
|
$hostID, |
|
|
$backupNum, |
|
|
$backup->{'endTime'}, |
|
|
$backup->{'type'} |
|
423 |
); |
); |
|
$dbh->commit(); |
|
424 |
|
|
425 |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
426 |
foreach my $share ($files->shareList($backupNum)) { |
foreach my $share ($files->shareList($backupNum)) { |
427 |
|
|
428 |
my $t = time(); |
my $t = time(); |
429 |
|
|
|
print strftime($t_fmt,localtime())," ", $share; |
|
430 |
$shareID = getShareID($share, $hostID, $hostname); |
$shareID = getShareID($share, $hostID, $hostname); |
431 |
|
|
432 |
my ($f, $nf, $d, $nd) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
$sth->{backups_count}->execute($hostID, $backupNum, $shareID); |
433 |
|
my ($count) = $sth->{backups_count}->fetchrow_array(); |
434 |
|
# skip if allready in database! |
435 |
|
next if ($count > 0); |
436 |
|
|
437 |
|
# dump some log |
438 |
|
print curr_time," ", $share; |
439 |
|
|
440 |
|
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
441 |
|
|
442 |
|
$sth->{insert_backups}->execute( |
443 |
|
$hostID, |
444 |
|
$backupNum, |
445 |
|
$backup->{'endTime'}, |
446 |
|
$backup->{'type'}, |
447 |
|
$shareID, |
448 |
|
$size, |
449 |
|
); |
450 |
|
|
451 |
|
print " commit"; |
452 |
|
$dbh->commit(); |
453 |
|
|
454 |
my $dur = (time() - $t) || 1; |
my $dur = (time() - $t) || 1; |
455 |
printf(" %d/%d files %d/%d dirs [%.2f/s dur: %s]\n", |
printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n", |
456 |
$nf, $f, $nd, $d, |
$nf, $f, $nd, $d, |
457 |
|
($size / 1024 / 1024), |
458 |
( ($f+$d) / $dur ), |
( ($f+$d) / $dur ), |
459 |
fmt_time($dur) |
fmt_time($dur) |
460 |
); |
); |
461 |
$dbh->commit(); |
|
462 |
|
hest_update($hostID, $shareID, $backupNum); |
463 |
} |
} |
464 |
|
|
465 |
} |
} |
504 |
my @data = @_; |
my @data = @_; |
505 |
shift @data; |
shift @data; |
506 |
|
|
507 |
my ($key, $shareID,undef,$name,$path,undef,$date,undef,$size) = @_; |
my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_; |
508 |
|
|
509 |
return $beenThere->{$key} if (defined($beenThere->{$key})); |
return $beenThere->{$key} if (defined($beenThere->{$key})); |
510 |
|
|
512 |
SELECT 1 FROM files |
SELECT 1 FROM files |
513 |
WHERE shareID = ? and |
WHERE shareID = ? and |
514 |
path = ? and |
path = ? and |
|
name = ? and |
|
515 |
date = ? and |
date = ? and |
516 |
size = ? |
size = ? |
517 |
|
LIMIT 1 |
518 |
}); |
}); |
519 |
|
|
520 |
my @param = ($shareID,$path,$name,$date,$size); |
my @param = ($shareID,$path,$date,$size); |
521 |
$sth->{file_in_db}->execute(@param); |
$sth->{file_in_db}->execute(@param); |
522 |
my $rows = $sth->{file_in_db}->rows; |
my $rows = $sth->{file_in_db}->rows; |
523 |
print STDERR "## found_in_db ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
524 |
|
|
525 |
$beenThere->{$key}++; |
$beenThere->{$key}++; |
526 |
|
|
538 |
|
|
539 |
print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1); |
print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1); |
540 |
|
|
541 |
my ($nr_files, $new_files, $nr_dirs, $new_dirs) = (0,0,0,0); |
my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0); |
542 |
|
|
543 |
{ # scope |
{ # scope |
544 |
my @stack; |
my @stack; |
548 |
|
|
549 |
# first, add all the entries in current directory |
# first, add all the entries in current directory |
550 |
foreach my $path_key (keys %{$filesInBackup}) { |
foreach my $path_key (keys %{$filesInBackup}) { |
551 |
|
print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3); |
552 |
my @data = ( |
my @data = ( |
553 |
$shareID, |
$shareID, |
554 |
$backupNum, |
$backupNum, |
567 |
$filesInBackup->{$path_key}->{'size'} |
$filesInBackup->{$path_key}->{'size'} |
568 |
)); |
)); |
569 |
|
|
570 |
|
my $found; |
571 |
if (! defined($beenThere->{$key}) && ! found_in_db($key, @data)) { |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
572 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
573 |
|
|
574 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
575 |
$new_dirs++; |
$new_dirs++ unless ($found); |
576 |
print STDERR " dir\n" if ($debug >= 2); |
print STDERR " dir\n" if ($debug >= 2); |
577 |
} else { |
} else { |
578 |
$new_files++; |
$new_files++ unless ($found); |
579 |
print STDERR " file\n" if ($debug >= 2); |
print STDERR " file\n" if ($debug >= 2); |
580 |
} |
} |
581 |
|
$size += $filesInBackup->{$path_key}->{'size'} || 0; |
582 |
} |
} |
583 |
|
|
584 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
603 |
print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2); |
print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2); |
604 |
|
|
605 |
while ( my $dir = shift @stack ) { |
while ( my $dir = shift @stack ) { |
606 |
my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID); |
my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID); |
607 |
print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1); |
print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1); |
608 |
$nr_files += $f; |
$nr_files += $f; |
609 |
$new_files += $nf; |
$new_files += $nf; |
610 |
$nr_dirs += $d; |
$nr_dirs += $d; |
611 |
$new_dirs += $nd; |
$new_dirs += $nd; |
612 |
|
$size += $s; |
613 |
} |
} |
614 |
} |
} |
615 |
|
|
616 |
return ($nr_files, $new_files, $nr_dirs, $new_dirs); |
return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size); |
617 |
} |
} |
618 |
|
|