1 |
#!/usr/local/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
use lib "__INSTALLDIR__/lib"; |
5 |
|
6 |
use DBI; |
7 |
use BackupPC::Lib; |
8 |
use BackupPC::View; |
9 |
use Data::Dumper; |
10 |
use Getopt::Std; |
11 |
use Time::HiRes qw/time/; |
12 |
use File::Pid; |
13 |
use POSIX qw/strftime/; |
14 |
|
15 |
use constant BPC_FTYPE_DIR => 5; |
16 |
|
17 |
my $debug = 0; |
18 |
$|=1; |
19 |
|
20 |
my $start_t = time(); |
21 |
|
22 |
my $pidfile = new File::Pid; |
23 |
|
24 |
if (my $pid = $pidfile->running ) { |
25 |
die "$0 already running: $pid\n"; |
26 |
} elsif ($pidfile->pid ne $$) { |
27 |
$pidfile->remove; |
28 |
$pidfile = new File::Pid; |
29 |
} |
30 |
$pidfile->write; |
31 |
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; |
32 |
|
33 |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
34 |
|
35 |
my $hosts; |
36 |
my $bpc = BackupPC::Lib->new || die; |
37 |
my %Conf = $bpc->Conf(); |
38 |
my $TopDir = $bpc->TopDir(); |
39 |
my $beenThere = {}; |
40 |
|
41 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
42 |
my $user = $Conf{SearchUser} || ''; |
43 |
my $index_path = $Conf{HyperEstraierIndex}; |
44 |
$index_path = $TopDir . '/' . $index_path; |
45 |
$index_path =~ s#//#/#g; |
46 |
|
47 |
|
48 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
49 |
|
50 |
my %opt; |
51 |
|
52 |
if ( !getopts("cdm:v:i", \%opt ) ) { |
53 |
print STDERR <<EOF; |
54 |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
55 |
|
56 |
Options: |
57 |
-c create database on first use |
58 |
-d delete database before import |
59 |
-m num import just num increments for one host |
60 |
-v num set verbosity (debug) level (default $debug) |
61 |
-i update HyperEstraier full text index |
62 |
EOF |
63 |
exit 1; |
64 |
} |
65 |
|
66 |
if ($opt{v}) { |
67 |
print "Debug level at $opt{v}\n"; |
68 |
$debug = $opt{v}; |
69 |
} |
70 |
|
71 |
#---- subs ---- |
72 |
|
73 |
sub fmt_time { |
74 |
my $t = shift || return; |
75 |
my $out = ""; |
76 |
my ($ss,$mm,$hh) = gmtime($t); |
77 |
$out .= "${hh}h" if ($hh); |
78 |
$out .= sprintf("%02d:%02d", $mm,$ss); |
79 |
return $out; |
80 |
} |
81 |
|
82 |
sub curr_time { |
83 |
return strftime($t_fmt,localtime()); |
84 |
} |
85 |
|
86 |
my $hest_db; |
87 |
|
88 |
sub hest_update { |
89 |
|
90 |
my ($host_id, $share_id, $num) = @_; |
91 |
|
92 |
print curr_time," updating HyperEstraier: select files"; |
93 |
|
94 |
my $t = time(); |
95 |
|
96 |
my $where = ''; |
97 |
if ($host_id && $share_id && $num) { |
98 |
$where = qq{ |
99 |
WHERE |
100 |
hosts.id = ? AND |
101 |
shares.id = ? AND |
102 |
files.backupnum = ? |
103 |
}; |
104 |
} |
105 |
|
106 |
my $sth = $dbh->prepare(qq{ |
107 |
SELECT |
108 |
files.id AS fid, |
109 |
hosts.name AS hname, |
110 |
shares.name AS sname, |
111 |
-- shares.share AS sharename, |
112 |
files.backupnum AS backupnum, |
113 |
-- files.name AS filename, |
114 |
files.path AS filepath, |
115 |
files.date AS date, |
116 |
files.type AS type, |
117 |
files.size AS size, |
118 |
files.shareid AS shareid, |
119 |
backups.date AS backup_date |
120 |
FROM files |
121 |
INNER JOIN shares ON files.shareID=shares.ID |
122 |
INNER JOIN hosts ON hosts.ID = shares.hostID |
123 |
INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID |
124 |
$where |
125 |
}); |
126 |
|
127 |
$sth->execute(@_); |
128 |
my $results = $sth->rows; |
129 |
|
130 |
if ($results == 0) { |
131 |
print " - no files, skipping\n"; |
132 |
return; |
133 |
} |
134 |
|
135 |
my $dot = int($results / 15) || 1; |
136 |
|
137 |
print " $results ($dot/#)"; |
138 |
|
139 |
sub fmt_date { |
140 |
my $t = shift || return; |
141 |
my $iso = BackupPC::Lib::timeStamp($t); |
142 |
$iso =~ s/\s/T/; |
143 |
return $iso; |
144 |
} |
145 |
|
146 |
my $max = int($results / $dot); |
147 |
|
148 |
print ", opening index $index_path..."; |
149 |
use HyperEstraier; |
150 |
my $db = HyperEstraier::Database->new(); |
151 |
|
152 |
# unless ($hest_db) { |
153 |
# print " open reader"; |
154 |
# $hest_db = HyperEstraier::Database->new(); |
155 |
# |
156 |
# } |
157 |
|
158 |
|
159 |
$db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
160 |
|
161 |
my $added = 0; |
162 |
|
163 |
while (my $row = $sth->fetchrow_hashref()) { |
164 |
|
165 |
my $fid = $row->{'fid'} || die "no fid?"; |
166 |
my $uri = 'file:///' . $fid; |
167 |
|
168 |
my $id = $db->uri_to_id($uri); |
169 |
next unless ($id == -1); |
170 |
|
171 |
# create a document object |
172 |
my $doc = HyperEstraier::Document->new; |
173 |
|
174 |
# add attributes to the document object |
175 |
$doc->add_attr('@uri', $uri); |
176 |
|
177 |
foreach my $c (@{ $sth->{NAME} }) { |
178 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
179 |
} |
180 |
|
181 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
182 |
|
183 |
# add the body text to the document object |
184 |
my $path = $row->{'filepath'}; |
185 |
$doc->add_text($path); |
186 |
$path =~ s/(.)/$1 /g; |
187 |
$doc->add_hidden_text($path); |
188 |
|
189 |
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
190 |
|
191 |
# register the document object to the database |
192 |
$db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
193 |
|
194 |
$added++; |
195 |
if ($added % $dot == 0) { |
196 |
print "$max "; |
197 |
$max--; |
198 |
} |
199 |
|
200 |
} |
201 |
|
202 |
print "sync $added new files"; |
203 |
$db->sync(); |
204 |
print ", close"; |
205 |
$db->close(); |
206 |
|
207 |
my $dur = (time() - $t) || 1; |
208 |
printf(" [%.2f/s new %.2f/s dur: %s]\n", |
209 |
( $results / $dur ), |
210 |
( $added / $dur ), |
211 |
fmt_time($dur) |
212 |
); |
213 |
} |
214 |
|
215 |
#---- /subs ---- |
216 |
|
217 |
|
218 |
## update index ## |
219 |
if (($opt{i} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
220 |
# update all |
221 |
print "force update of HyperEstraier index "; |
222 |
print "importing existing data" unless (-e $index_path); |
223 |
print "by -i flag" if ($opt{i}); |
224 |
print "\n"; |
225 |
hest_update(); |
226 |
} |
227 |
|
228 |
## create tables ## |
229 |
if ($opt{c}) { |
230 |
sub do_index { |
231 |
my $index = shift || return; |
232 |
my ($table,$col,$unique) = split(/_/, $index); |
233 |
$unique ||= ''; |
234 |
$index =~ s/,/_/g; |
235 |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
236 |
} |
237 |
|
238 |
print "creating tables...\n"; |
239 |
|
240 |
$dbh->do(qq{ |
241 |
create table hosts ( |
242 |
ID SERIAL PRIMARY KEY, |
243 |
name VARCHAR(30) NOT NULL, |
244 |
IP VARCHAR(15) |
245 |
); |
246 |
}); |
247 |
|
248 |
$dbh->do(qq{ |
249 |
create table shares ( |
250 |
ID SERIAL PRIMARY KEY, |
251 |
hostID INTEGER NOT NULL references hosts(id), |
252 |
name VARCHAR(30) NOT NULL, |
253 |
share VARCHAR(200) NOT NULL, |
254 |
localpath VARCHAR(200) |
255 |
); |
256 |
}); |
257 |
|
258 |
$dbh->do(qq{ |
259 |
create table backups ( |
260 |
hostID INTEGER NOT NULL references hosts(id), |
261 |
num INTEGER NOT NULL, |
262 |
date integer NOT NULL, |
263 |
type CHAR(4) not null, |
264 |
shareID integer not null references shares(id), |
265 |
size integer not null, |
266 |
PRIMARY KEY(hostID, num, shareID) |
267 |
); |
268 |
}); |
269 |
|
270 |
#do_index('backups_hostid,num_unique'); |
271 |
|
272 |
$dbh->do(qq{ |
273 |
create table dvds ( |
274 |
ID SERIAL PRIMARY KEY, |
275 |
num INTEGER NOT NULL, |
276 |
name VARCHAR(255) NOT NULL, |
277 |
mjesto VARCHAR(255) |
278 |
); |
279 |
}); |
280 |
|
281 |
$dbh->do(qq{ |
282 |
create table files ( |
283 |
ID SERIAL PRIMARY KEY, |
284 |
shareID INTEGER NOT NULL references shares(id), |
285 |
backupNum INTEGER NOT NULL, |
286 |
name VARCHAR(255) NOT NULL, |
287 |
path VARCHAR(255) NOT NULL, |
288 |
date integer NOT NULL, |
289 |
type INTEGER NOT NULL, |
290 |
size INTEGER NOT NULL, |
291 |
dvdid INTEGER references dvds(id) |
292 |
); |
293 |
}); |
294 |
|
295 |
print "creating indexes:"; |
296 |
|
297 |
foreach my $index (qw( |
298 |
hosts_name |
299 |
backups_hostID |
300 |
backups_num |
301 |
shares_hostID |
302 |
shares_name |
303 |
files_shareID |
304 |
files_path |
305 |
files_name |
306 |
files_date |
307 |
files_size |
308 |
)) { |
309 |
print " $index"; |
310 |
do_index($index); |
311 |
} |
312 |
print "...\n"; |
313 |
|
314 |
$dbh->commit; |
315 |
|
316 |
} |
317 |
|
318 |
## delete data before inseting ## |
319 |
if ($opt{d}) { |
320 |
print "deleting "; |
321 |
foreach my $table (qw(files dvds backups shares hosts)) { |
322 |
print "$table "; |
323 |
$dbh->do(qq{ DELETE FROM $table }); |
324 |
} |
325 |
print " done...\n"; |
326 |
|
327 |
$dbh->commit; |
328 |
} |
329 |
|
330 |
## insert new values ## |
331 |
|
332 |
# get hosts |
333 |
$hosts = $bpc->HostInfoRead(); |
334 |
my $hostID; |
335 |
my $shareID; |
336 |
|
337 |
my $sth; |
338 |
|
339 |
$sth->{insert_hosts} = $dbh->prepare(qq{ |
340 |
INSERT INTO hosts (name, IP) VALUES (?,?) |
341 |
}); |
342 |
|
343 |
$sth->{hosts_by_name} = $dbh->prepare(qq{ |
344 |
SELECT ID FROM hosts WHERE name=? |
345 |
}); |
346 |
|
347 |
$sth->{backups_count} = $dbh->prepare(qq{ |
348 |
SELECT COUNT(*) |
349 |
FROM backups |
350 |
WHERE hostID=? AND num=? AND shareid=? |
351 |
}); |
352 |
|
353 |
$sth->{insert_backups} = $dbh->prepare(qq{ |
354 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
355 |
VALUES (?,?,?,?,?,?) |
356 |
}); |
357 |
|
358 |
$sth->{insert_files} = $dbh->prepare(qq{ |
359 |
INSERT INTO files |
360 |
(shareID, backupNum, name, path, date, type, size) |
361 |
VALUES (?,?,?,?,?,?,?) |
362 |
}); |
363 |
|
364 |
foreach my $host_key (keys %{$hosts}) { |
365 |
|
366 |
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
367 |
|
368 |
$sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'}); |
369 |
|
370 |
unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) { |
371 |
$sth->{insert_hosts}->execute( |
372 |
$hosts->{$host_key}->{'host'}, |
373 |
$hosts->{$host_key}->{'ip'} |
374 |
); |
375 |
|
376 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
377 |
} |
378 |
|
379 |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
380 |
|
381 |
# get backups for a host |
382 |
my @backups = $bpc->BackupInfoRead($hostname); |
383 |
my $incs = scalar @backups; |
384 |
print "$incs increments\n"; |
385 |
|
386 |
my $inc_nr = 0; |
387 |
$beenThere = {}; |
388 |
|
389 |
foreach my $backup (@backups) { |
390 |
|
391 |
$inc_nr++; |
392 |
last if ($opt{m} && $inc_nr > $opt{m}); |
393 |
|
394 |
my $backupNum = $backup->{'num'}; |
395 |
my @backupShares = (); |
396 |
|
397 |
printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", |
398 |
$hosts->{$host_key}->{'host'}, |
399 |
$inc_nr, $incs, $backupNum, |
400 |
$backup->{type} || '?', |
401 |
$backup->{nFilesNew} || '?', $backup->{nFiles} || '?', |
402 |
strftime($t_fmt,localtime($backup->{startTime})), |
403 |
fmt_time($backup->{endTime} - $backup->{startTime}) |
404 |
); |
405 |
|
406 |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
407 |
foreach my $share ($files->shareList($backupNum)) { |
408 |
|
409 |
my $t = time(); |
410 |
|
411 |
$shareID = getShareID($share, $hostID, $hostname); |
412 |
|
413 |
$sth->{backups_count}->execute($hostID, $backupNum, $shareID); |
414 |
my ($count) = $sth->{backups_count}->fetchrow_array(); |
415 |
# skip if allready in database! |
416 |
next if ($count > 0); |
417 |
|
418 |
# dump some log |
419 |
print curr_time," ", $share; |
420 |
|
421 |
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
422 |
|
423 |
$sth->{insert_backups}->execute( |
424 |
$hostID, |
425 |
$backupNum, |
426 |
$backup->{'endTime'}, |
427 |
$backup->{'type'}, |
428 |
$shareID, |
429 |
$size, |
430 |
); |
431 |
|
432 |
print " commit"; |
433 |
$dbh->commit(); |
434 |
|
435 |
my $dur = (time() - $t) || 1; |
436 |
printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n", |
437 |
$nf, $f, $nd, $d, |
438 |
($size / 1024 / 1024), |
439 |
( ($f+$d) / $dur ), |
440 |
fmt_time($dur) |
441 |
); |
442 |
|
443 |
hest_update($hostID, $shareID, $backupNum); |
444 |
} |
445 |
|
446 |
} |
447 |
} |
448 |
undef $sth; |
449 |
$dbh->commit(); |
450 |
$dbh->disconnect(); |
451 |
|
452 |
print "total duration: ",fmt_time(time() - $start_t),"\n"; |
453 |
|
454 |
$pidfile->remove; |
455 |
|
456 |
sub getShareID() { |
457 |
|
458 |
my ($share, $hostID, $hostname) = @_; |
459 |
|
460 |
$sth->{share_id} ||= $dbh->prepare(qq{ |
461 |
SELECT ID FROM shares WHERE hostID=? AND name=? |
462 |
}); |
463 |
|
464 |
$sth->{share_id}->execute($hostID,$share); |
465 |
|
466 |
my ($id) = $sth->{share_id}->fetchrow_array(); |
467 |
|
468 |
return $id if (defined($id)); |
469 |
|
470 |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
471 |
INSERT INTO shares |
472 |
(hostID,name,share,localpath) |
473 |
VALUES (?,?,?,?) |
474 |
}); |
475 |
|
476 |
my $drop_down = $hostname . '/' . $share; |
477 |
$drop_down =~ s#//+#/#g; |
478 |
|
479 |
$sth->{insert_share}->execute($hostID,$share, $drop_down ,undef); |
480 |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
481 |
} |
482 |
|
483 |
sub found_in_db { |
484 |
|
485 |
my @data = @_; |
486 |
shift @data; |
487 |
|
488 |
my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_; |
489 |
|
490 |
return $beenThere->{$key} if (defined($beenThere->{$key})); |
491 |
|
492 |
$sth->{file_in_db} ||= $dbh->prepare(qq{ |
493 |
SELECT 1 FROM files |
494 |
WHERE shareID = ? and |
495 |
path = ? and |
496 |
date = ? and |
497 |
size = ? |
498 |
LIMIT 1 |
499 |
}); |
500 |
|
501 |
my @param = ($shareID,$path,$date,$size); |
502 |
$sth->{file_in_db}->execute(@param); |
503 |
my $rows = $sth->{file_in_db}->rows; |
504 |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
505 |
|
506 |
$beenThere->{$key}++; |
507 |
|
508 |
$sth->{'insert_files'}->execute(@data) unless ($rows); |
509 |
return $rows; |
510 |
} |
511 |
|
512 |
#################################################### |
513 |
# recursing through filesystem structure and # |
514 |
# and returning flattened files list # |
515 |
#################################################### |
516 |
sub recurseDir($$$$$$$$) { |
517 |
|
518 |
my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_; |
519 |
|
520 |
print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1); |
521 |
|
522 |
my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0); |
523 |
|
524 |
{ # scope |
525 |
my @stack; |
526 |
|
527 |
print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2); |
528 |
my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir); |
529 |
|
530 |
# first, add all the entries in current directory |
531 |
foreach my $path_key (keys %{$filesInBackup}) { |
532 |
print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3); |
533 |
my @data = ( |
534 |
$shareID, |
535 |
$backupNum, |
536 |
$path_key, |
537 |
$filesInBackup->{$path_key}->{'relPath'}, |
538 |
$filesInBackup->{$path_key}->{'mtime'}, |
539 |
$filesInBackup->{$path_key}->{'type'}, |
540 |
$filesInBackup->{$path_key}->{'size'} |
541 |
); |
542 |
|
543 |
my $key = join(" ", ( |
544 |
$shareID, |
545 |
$dir, |
546 |
$path_key, |
547 |
$filesInBackup->{$path_key}->{'mtime'}, |
548 |
$filesInBackup->{$path_key}->{'size'} |
549 |
)); |
550 |
|
551 |
my $found; |
552 |
if (! defined($beenThere->{$key}) && ! ($found = found_in_db($key, @data)) ) { |
553 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
554 |
|
555 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
556 |
$new_dirs++ unless ($found); |
557 |
print STDERR " dir\n" if ($debug >= 2); |
558 |
} else { |
559 |
$new_files++ unless ($found); |
560 |
print STDERR " file\n" if ($debug >= 2); |
561 |
} |
562 |
$size += $filesInBackup->{$path_key}->{'size'} || 0; |
563 |
} |
564 |
|
565 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
566 |
$nr_dirs++; |
567 |
|
568 |
my $full_path = $dir . '/' . $path_key; |
569 |
push @stack, $full_path; |
570 |
print STDERR "### store to stack: $full_path\n" if ($debug >= 3); |
571 |
|
572 |
# my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key}); |
573 |
# |
574 |
# $nr_files += $f; |
575 |
# $new_files += $nf; |
576 |
# $nr_dirs += $d; |
577 |
# $new_dirs += $nd; |
578 |
|
579 |
} else { |
580 |
$nr_files++; |
581 |
} |
582 |
} |
583 |
|
584 |
print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2); |
585 |
|
586 |
while ( my $dir = shift @stack ) { |
587 |
my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID); |
588 |
print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1); |
589 |
$nr_files += $f; |
590 |
$new_files += $nf; |
591 |
$nr_dirs += $d; |
592 |
$new_dirs += $nd; |
593 |
$size += $s; |
594 |
} |
595 |
} |
596 |
|
597 |
return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size); |
598 |
} |
599 |
|