1 |
dpavlin |
14 |
#!/usr/local/bin/perl -w |
2 |
dpavlin |
6 |
|
3 |
|
|
use strict; |
4 |
dpavlin |
48 |
use lib "__INSTALLDIR__/lib"; |
5 |
|
|
|
6 |
dpavlin |
6 |
use DBI; |
7 |
|
|
use BackupPC::Lib; |
8 |
|
|
use BackupPC::View; |
9 |
|
|
use Data::Dumper; |
10 |
|
|
use Getopt::Std; |
11 |
dpavlin |
37 |
use Time::HiRes qw/time/; |
12 |
dpavlin |
38 |
use File::Pid; |
13 |
dpavlin |
37 |
use POSIX qw/strftime/; |
14 |
dpavlin |
118 |
use BackupPC::SearchLib; |
15 |
dpavlin |
48 |
|
16 |
dpavlin |
6 |
use constant BPC_FTYPE_DIR => 5; |
17 |
dpavlin |
99 |
use constant EST_CHUNK => 100000; |
18 |
dpavlin |
6 |
|
19 |
dpavlin |
246 |
# daylight saving time change offset for 1h |
20 |
|
|
my $dst_offset = 60 * 60; |
21 |
|
|
|
22 |
dpavlin |
30 |
my $debug = 0; |
23 |
dpavlin |
14 |
$|=1; |
24 |
dpavlin |
6 |
|
25 |
dpavlin |
51 |
my $start_t = time(); |
26 |
|
|
|
27 |
dpavlin |
38 |
my $pidfile = new File::Pid; |
28 |
|
|
|
29 |
|
|
if (my $pid = $pidfile->running ) { |
30 |
|
|
die "$0 already running: $pid\n"; |
31 |
|
|
} elsif ($pidfile->pid ne $$) { |
32 |
|
|
$pidfile->remove; |
33 |
|
|
$pidfile = new File::Pid; |
34 |
|
|
} |
35 |
dpavlin |
39 |
$pidfile->write; |
36 |
|
|
print STDERR "$0 using pid ",$pidfile->pid," file ",$pidfile->file,"\n"; |
37 |
dpavlin |
38 |
|
38 |
dpavlin |
37 |
my $t_fmt = '%Y-%m-%d %H:%M:%S'; |
39 |
|
|
|
40 |
dpavlin |
6 |
my $hosts; |
41 |
|
|
my $bpc = BackupPC::Lib->new || die; |
42 |
|
|
my %Conf = $bpc->Conf(); |
43 |
|
|
my $TopDir = $bpc->TopDir(); |
44 |
dpavlin |
14 |
my $beenThere = {}; |
45 |
dpavlin |
6 |
|
46 |
dpavlin |
51 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
47 |
|
|
my $user = $Conf{SearchUser} || ''; |
48 |
dpavlin |
116 |
|
49 |
|
|
my $use_hest = $Conf{HyperEstraierIndex}; |
50 |
dpavlin |
118 |
my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest); |
51 |
dpavlin |
6 |
|
52 |
dpavlin |
49 |
my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 }); |
53 |
|
|
|
54 |
dpavlin |
6 |
my %opt; |
55 |
|
|
|
56 |
dpavlin |
246 |
if ( !getopts("cdm:v:ijf", \%opt ) ) { |
57 |
dpavlin |
6 |
print STDERR <<EOF; |
58 |
dpavlin |
246 |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i|-j|-f] |
59 |
dpavlin |
6 |
|
60 |
|
|
Options: |
61 |
dpavlin |
14 |
-c create database on first use |
62 |
|
|
-d delete database before import |
63 |
|
|
-m num import just num increments for one host |
64 |
dpavlin |
44 |
-v num set verbosity (debug) level (default $debug) |
65 |
dpavlin |
131 |
-i update Hyper Estraier full text index |
66 |
|
|
-j update full text, don't check existing files |
67 |
dpavlin |
246 |
-f don't do anything with full text index |
68 |
dpavlin |
131 |
|
69 |
|
|
Option -j is variation on -i. It will allow faster initial creation |
70 |
|
|
of full-text index from existing database. |
71 |
|
|
|
72 |
dpavlin |
246 |
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 |
dpavlin |
6 |
EOF |
76 |
|
|
exit 1; |
77 |
|
|
} |
78 |
|
|
|
79 |
dpavlin |
81 |
if ($opt{v}) { |
80 |
|
|
print "Debug level at $opt{v}\n"; |
81 |
|
|
$debug = $opt{v}; |
82 |
dpavlin |
246 |
} 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 |
dpavlin |
81 |
} |
86 |
|
|
|
87 |
dpavlin |
86 |
#---- subs ---- |
88 |
|
|
|
89 |
|
|
sub fmt_time { |
90 |
|
|
my $t = shift || return; |
91 |
|
|
my $out = ""; |
92 |
|
|
my ($ss,$mm,$hh) = gmtime($t); |
93 |
|
|
$out .= "${hh}h" if ($hh); |
94 |
|
|
$out .= sprintf("%02d:%02d", $mm,$ss); |
95 |
|
|
return $out; |
96 |
|
|
} |
97 |
|
|
|
98 |
|
|
sub curr_time { |
99 |
|
|
return strftime($t_fmt,localtime()); |
100 |
|
|
} |
101 |
|
|
|
102 |
dpavlin |
89 |
my $hest_db; |
103 |
dpavlin |
116 |
my $hest_node; |
104 |
dpavlin |
86 |
|
105 |
dpavlin |
97 |
sub signal { |
106 |
|
|
my($sig) = @_; |
107 |
|
|
if ($hest_db) { |
108 |
|
|
print "\nCaught a SIG$sig--syncing database and shutting down\n"; |
109 |
|
|
$hest_db->sync(); |
110 |
|
|
$hest_db->close(); |
111 |
|
|
} |
112 |
|
|
exit(0); |
113 |
|
|
} |
114 |
|
|
|
115 |
|
|
$SIG{'INT'} = \&signal; |
116 |
|
|
$SIG{'QUIT'} = \&signal; |
117 |
|
|
|
118 |
dpavlin |
89 |
sub hest_update { |
119 |
dpavlin |
81 |
|
120 |
dpavlin |
89 |
my ($host_id, $share_id, $num) = @_; |
121 |
|
|
|
122 |
dpavlin |
131 |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
123 |
|
|
|
124 |
dpavlin |
116 |
unless ($use_hest) { |
125 |
|
|
print STDERR "HyperEstraier support not enabled in configuration\n"; |
126 |
|
|
return; |
127 |
|
|
} |
128 |
|
|
|
129 |
dpavlin |
98 |
print curr_time," updating HyperEstraier:"; |
130 |
dpavlin |
86 |
|
131 |
|
|
my $t = time(); |
132 |
dpavlin |
89 |
|
133 |
dpavlin |
98 |
my $offset = 0; |
134 |
|
|
my $added = 0; |
135 |
dpavlin |
89 |
|
136 |
dpavlin |
130 |
print " opening index $use_hest"; |
137 |
dpavlin |
116 |
if ($index_path) { |
138 |
|
|
$hest_db = HyperEstraier::Database->new(); |
139 |
dpavlin |
118 |
$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
140 |
dpavlin |
130 |
print " directly"; |
141 |
dpavlin |
116 |
} elsif ($index_node_url) { |
142 |
|
|
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
143 |
|
|
$hest_node->set_auth('admin', 'admin'); |
144 |
dpavlin |
130 |
print " via node URL"; |
145 |
dpavlin |
116 |
} else { |
146 |
|
|
die "don't know how to use HyperEstraier Index $use_hest"; |
147 |
|
|
} |
148 |
dpavlin |
107 |
print " increment is " . EST_CHUNK . " files:"; |
149 |
dpavlin |
99 |
|
150 |
dpavlin |
98 |
my $results = 0; |
151 |
dpavlin |
81 |
|
152 |
dpavlin |
98 |
do { |
153 |
dpavlin |
81 |
|
154 |
dpavlin |
98 |
my $where = ''; |
155 |
dpavlin |
104 |
my @data; |
156 |
dpavlin |
190 |
if (defined($host_id) && defined($share_id) && defined($num)) { |
157 |
dpavlin |
98 |
$where = qq{ |
158 |
|
|
WHERE |
159 |
|
|
hosts.id = ? AND |
160 |
|
|
shares.id = ? AND |
161 |
|
|
files.backupnum = ? |
162 |
|
|
}; |
163 |
dpavlin |
104 |
@data = ( $host_id, $share_id, $num ); |
164 |
dpavlin |
98 |
} |
165 |
dpavlin |
89 |
|
166 |
dpavlin |
98 |
my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset); |
167 |
dpavlin |
81 |
|
168 |
dpavlin |
98 |
my $sth = $dbh->prepare(qq{ |
169 |
|
|
SELECT |
170 |
|
|
files.id AS fid, |
171 |
|
|
hosts.name AS hname, |
172 |
|
|
shares.name AS sname, |
173 |
|
|
-- shares.share AS sharename, |
174 |
|
|
files.backupnum AS backupnum, |
175 |
|
|
-- files.name AS filename, |
176 |
|
|
files.path AS filepath, |
177 |
|
|
files.date AS date, |
178 |
|
|
files.type AS type, |
179 |
|
|
files.size AS size, |
180 |
|
|
files.shareid AS shareid, |
181 |
|
|
backups.date AS backup_date |
182 |
|
|
FROM files |
183 |
|
|
INNER JOIN shares ON files.shareID=shares.ID |
184 |
|
|
INNER JOIN hosts ON hosts.ID = shares.hostID |
185 |
|
|
INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID |
186 |
|
|
$where |
187 |
|
|
$limit |
188 |
|
|
}); |
189 |
dpavlin |
81 |
|
190 |
dpavlin |
104 |
$sth->execute(@data); |
191 |
dpavlin |
98 |
$results = $sth->rows; |
192 |
dpavlin |
81 |
|
193 |
dpavlin |
98 |
if ($results == 0) { |
194 |
dpavlin |
107 |
print " - no new files\n"; |
195 |
dpavlin |
98 |
last; |
196 |
|
|
} |
197 |
dpavlin |
89 |
|
198 |
dpavlin |
98 |
sub fmt_date { |
199 |
|
|
my $t = shift || return; |
200 |
|
|
my $iso = BackupPC::Lib::timeStamp($t); |
201 |
|
|
$iso =~ s/\s/T/; |
202 |
|
|
return $iso; |
203 |
|
|
} |
204 |
dpavlin |
89 |
|
205 |
dpavlin |
98 |
while (my $row = $sth->fetchrow_hashref()) { |
206 |
dpavlin |
89 |
|
207 |
dpavlin |
98 |
my $fid = $row->{'fid'} || die "no fid?"; |
208 |
|
|
my $uri = 'file:///' . $fid; |
209 |
dpavlin |
82 |
|
210 |
dpavlin |
131 |
unless ($skip_check) { |
211 |
|
|
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
212 |
|
|
next unless ($id == -1); |
213 |
|
|
} |
214 |
dpavlin |
82 |
|
215 |
dpavlin |
98 |
# create a document object |
216 |
|
|
my $doc = HyperEstraier::Document->new; |
217 |
dpavlin |
81 |
|
218 |
dpavlin |
98 |
# add attributes to the document object |
219 |
|
|
$doc->add_attr('@uri', $uri); |
220 |
dpavlin |
89 |
|
221 |
dpavlin |
98 |
foreach my $c (@{ $sth->{NAME} }) { |
222 |
dpavlin |
241 |
$doc->add_attr($c, $row->{$c}) if (defined($row->{$c})); |
223 |
dpavlin |
98 |
} |
224 |
dpavlin |
89 |
|
225 |
dpavlin |
98 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
226 |
dpavlin |
81 |
|
227 |
dpavlin |
98 |
# add the body text to the document object |
228 |
|
|
my $path = $row->{'filepath'}; |
229 |
|
|
$doc->add_text($path); |
230 |
|
|
$path =~ s/(.)/$1 /g; |
231 |
|
|
$doc->add_hidden_text($path); |
232 |
dpavlin |
81 |
|
233 |
dpavlin |
98 |
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
234 |
|
|
|
235 |
|
|
# register the document object to the database |
236 |
dpavlin |
116 |
if ($hest_db) { |
237 |
|
|
$hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
238 |
|
|
} elsif ($hest_node) { |
239 |
|
|
$hest_node->put_doc($doc); |
240 |
|
|
} else { |
241 |
|
|
die "not supported"; |
242 |
|
|
} |
243 |
dpavlin |
98 |
$added++; |
244 |
dpavlin |
81 |
} |
245 |
|
|
|
246 |
dpavlin |
98 |
print " $added"; |
247 |
dpavlin |
116 |
$hest_db->sync() if ($index_path); |
248 |
dpavlin |
81 |
|
249 |
dpavlin |
98 |
$offset += EST_CHUNK; |
250 |
dpavlin |
81 |
|
251 |
dpavlin |
98 |
} while ($results == EST_CHUNK); |
252 |
dpavlin |
81 |
|
253 |
dpavlin |
116 |
if ($index_path) { |
254 |
|
|
print ", close"; |
255 |
|
|
$hest_db->close(); |
256 |
|
|
} |
257 |
dpavlin |
81 |
|
258 |
dpavlin |
86 |
my $dur = (time() - $t) || 1; |
259 |
dpavlin |
98 |
printf(" [%.2f/s dur: %s]\n", |
260 |
dpavlin |
89 |
( $added / $dur ), |
261 |
dpavlin |
86 |
fmt_time($dur) |
262 |
|
|
); |
263 |
dpavlin |
89 |
} |
264 |
dpavlin |
86 |
|
265 |
dpavlin |
89 |
#---- /subs ---- |
266 |
|
|
|
267 |
|
|
|
268 |
|
|
## update index ## |
269 |
dpavlin |
239 |
if (($opt{i} || $opt{j} || ($index_path && ! -e $TopDir . $index_path)) && !$opt{c}) { |
270 |
dpavlin |
89 |
# update all |
271 |
|
|
print "force update of HyperEstraier index "; |
272 |
dpavlin |
239 |
print "importing existing data" unless (-e $TopDir . $index_path); |
273 |
dpavlin |
89 |
print "by -i flag" if ($opt{i}); |
274 |
dpavlin |
131 |
print "by -j flag" if ($opt{j}); |
275 |
dpavlin |
89 |
print "\n"; |
276 |
|
|
hest_update(); |
277 |
dpavlin |
81 |
} |
278 |
|
|
|
279 |
dpavlin |
89 |
## create tables ## |
280 |
dpavlin |
14 |
if ($opt{c}) { |
281 |
dpavlin |
49 |
sub do_index { |
282 |
|
|
my $index = shift || return; |
283 |
dpavlin |
152 |
my ($table,$col,$unique) = split(/:/, $index); |
284 |
dpavlin |
49 |
$unique ||= ''; |
285 |
dpavlin |
152 |
$index =~ s/\W+/_/g; |
286 |
|
|
print "$index on $table($col)" . ( $unique ? "u" : "" ) . " "; |
287 |
dpavlin |
49 |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
288 |
|
|
} |
289 |
|
|
|
290 |
dpavlin |
14 |
print "creating tables...\n"; |
291 |
dpavlin |
183 |
|
292 |
|
|
$dbh->do( qq{ |
293 |
dpavlin |
14 |
create table hosts ( |
294 |
dpavlin |
49 |
ID SERIAL PRIMARY KEY, |
295 |
dpavlin |
14 |
name VARCHAR(30) NOT NULL, |
296 |
|
|
IP VARCHAR(15) |
297 |
dpavlin |
6 |
); |
298 |
dpavlin |
183 |
|
299 |
dpavlin |
14 |
create table shares ( |
300 |
dpavlin |
49 |
ID SERIAL PRIMARY KEY, |
301 |
dpavlin |
14 |
hostID INTEGER NOT NULL references hosts(id), |
302 |
|
|
name VARCHAR(30) NOT NULL, |
303 |
dpavlin |
152 |
share VARCHAR(200) NOT NULL |
304 |
dpavlin |
6 |
); |
305 |
dpavlin |
130 |
|
306 |
|
|
create table dvds ( |
307 |
|
|
ID SERIAL PRIMARY KEY, |
308 |
|
|
num INTEGER NOT NULL, |
309 |
|
|
name VARCHAR(255) NOT NULL, |
310 |
|
|
mjesto VARCHAR(255) |
311 |
|
|
); |
312 |
dpavlin |
183 |
|
313 |
dpavlin |
14 |
create table backups ( |
314 |
dpavlin |
152 |
id serial, |
315 |
dpavlin |
14 |
hostID INTEGER NOT NULL references hosts(id), |
316 |
|
|
num INTEGER NOT NULL, |
317 |
dpavlin |
49 |
date integer NOT NULL, |
318 |
|
|
type CHAR(4) not null, |
319 |
dpavlin |
65 |
shareID integer not null references shares(id), |
320 |
dpavlin |
136 |
size bigint not null, |
321 |
dpavlin |
152 |
inc_size bigint not null default -1, |
322 |
|
|
inc_deleted boolean default false, |
323 |
dpavlin |
196 |
parts integer not null default 1, |
324 |
dpavlin |
152 |
PRIMARY KEY(id) |
325 |
dpavlin |
6 |
); |
326 |
|
|
|
327 |
dpavlin |
14 |
create table files ( |
328 |
dpavlin |
152 |
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 |
dpavlin |
6 |
); |
338 |
|
|
|
339 |
dpavlin |
152 |
create table archive ( |
340 |
|
|
id serial, |
341 |
dpavlin |
130 |
dvd_nr int not null, |
342 |
dpavlin |
152 |
total_size bigint default -1, |
343 |
dpavlin |
130 |
note text, |
344 |
|
|
username varchar(20) not null, |
345 |
dpavlin |
152 |
date timestamp default now(), |
346 |
dpavlin |
130 |
primary key(id) |
347 |
|
|
); |
348 |
|
|
|
349 |
dpavlin |
183 |
create table archive_backup ( |
350 |
dpavlin |
160 |
archive_id int not null references archive(id) on delete cascade, |
351 |
dpavlin |
152 |
backup_id int not null references backups(id), |
352 |
dpavlin |
130 |
primary key(archive_id, backup_id) |
353 |
|
|
); |
354 |
dpavlin |
183 |
|
355 |
|
|
create table archive_burned ( |
356 |
dpavlin |
199 |
archive_id int references archive(id), |
357 |
dpavlin |
202 |
date timestamp default now(), |
358 |
dpavlin |
199 |
part int not null default 1, |
359 |
dpavlin |
202 |
copy int not null default 1, |
360 |
dpavlin |
196 |
iso_size bigint default -1 |
361 |
dpavlin |
183 |
); |
362 |
|
|
|
363 |
dpavlin |
235 |
create table backup_parts ( |
364 |
|
|
id serial, |
365 |
|
|
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 |
dpavlin |
130 |
}); |
375 |
|
|
|
376 |
dpavlin |
153 |
print "creating indexes: "; |
377 |
dpavlin |
6 |
|
378 |
dpavlin |
14 |
foreach my $index (qw( |
379 |
dpavlin |
152 |
hosts:name |
380 |
|
|
backups:hostID |
381 |
|
|
backups:num |
382 |
|
|
backups:shareID |
383 |
|
|
shares:hostID |
384 |
|
|
shares:name |
385 |
|
|
files:shareID |
386 |
|
|
files:path |
387 |
|
|
files:name |
388 |
|
|
files:date |
389 |
|
|
files:size |
390 |
|
|
archive:dvd_nr |
391 |
dpavlin |
183 |
archive_burned:archive_id |
392 |
dpavlin |
235 |
backup_parts:backup_id,part_nr |
393 |
dpavlin |
14 |
)) { |
394 |
dpavlin |
49 |
do_index($index); |
395 |
dpavlin |
14 |
} |
396 |
dpavlin |
153 |
|
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 |
dpavlin |
49 |
print "...\n"; |
405 |
dpavlin |
14 |
|
406 |
dpavlin |
49 |
$dbh->commit; |
407 |
dpavlin |
14 |
|
408 |
|
|
} |
409 |
|
|
|
410 |
dpavlin |
89 |
## delete data before inseting ## |
411 |
dpavlin |
14 |
if ($opt{d}) { |
412 |
|
|
print "deleting "; |
413 |
dpavlin |
49 |
foreach my $table (qw(files dvds backups shares hosts)) { |
414 |
dpavlin |
14 |
print "$table "; |
415 |
|
|
$dbh->do(qq{ DELETE FROM $table }); |
416 |
|
|
} |
417 |
|
|
print " done...\n"; |
418 |
dpavlin |
49 |
|
419 |
dpavlin |
51 |
$dbh->commit; |
420 |
dpavlin |
14 |
} |
421 |
|
|
|
422 |
dpavlin |
89 |
## insert new values ## |
423 |
dpavlin |
6 |
|
424 |
|
|
# get hosts |
425 |
dpavlin |
8 |
$hosts = $bpc->HostInfoRead(); |
426 |
dpavlin |
6 |
my $hostID; |
427 |
|
|
my $shareID; |
428 |
|
|
|
429 |
dpavlin |
14 |
my $sth; |
430 |
dpavlin |
6 |
|
431 |
dpavlin |
14 |
$sth->{insert_hosts} = $dbh->prepare(qq{ |
432 |
|
|
INSERT INTO hosts (name, IP) VALUES (?,?) |
433 |
|
|
}); |
434 |
dpavlin |
6 |
|
435 |
dpavlin |
14 |
$sth->{hosts_by_name} = $dbh->prepare(qq{ |
436 |
|
|
SELECT ID FROM hosts WHERE name=? |
437 |
|
|
}); |
438 |
|
|
|
439 |
dpavlin |
65 |
$sth->{backups_count} = $dbh->prepare(qq{ |
440 |
dpavlin |
14 |
SELECT COUNT(*) |
441 |
|
|
FROM backups |
442 |
dpavlin |
65 |
WHERE hostID=? AND num=? AND shareid=? |
443 |
dpavlin |
14 |
}); |
444 |
|
|
|
445 |
|
|
$sth->{insert_backups} = $dbh->prepare(qq{ |
446 |
dpavlin |
66 |
INSERT INTO backups (hostID, num, date, type, shareid, size) |
447 |
dpavlin |
242 |
VALUES (?,?,?,?,?,-1) |
448 |
dpavlin |
14 |
}); |
449 |
|
|
|
450 |
dpavlin |
242 |
$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 |
dpavlin |
14 |
$sth->{insert_files} = $dbh->prepare(qq{ |
456 |
|
|
INSERT INTO files |
457 |
dpavlin |
62 |
(shareID, backupNum, name, path, date, type, size) |
458 |
|
|
VALUES (?,?,?,?,?,?,?) |
459 |
dpavlin |
14 |
}); |
460 |
|
|
|
461 |
|
|
foreach my $host_key (keys %{$hosts}) { |
462 |
|
|
|
463 |
|
|
my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key"; |
464 |
|
|
|
465 |
|
|
$sth->{hosts_by_name}->execute($hosts->{$host_key}->{'host'}); |
466 |
|
|
|
467 |
|
|
unless (($hostID) = $sth->{hosts_by_name}->fetchrow_array()) { |
468 |
|
|
$sth->{insert_hosts}->execute( |
469 |
|
|
$hosts->{$host_key}->{'host'}, |
470 |
|
|
$hosts->{$host_key}->{'ip'} |
471 |
|
|
); |
472 |
|
|
|
473 |
dpavlin |
49 |
$hostID = $dbh->last_insert_id(undef,undef,'hosts',undef); |
474 |
dpavlin |
14 |
} |
475 |
|
|
|
476 |
dpavlin |
67 |
print "host ".$hosts->{$host_key}->{'host'}.": "; |
477 |
dpavlin |
6 |
|
478 |
dpavlin |
14 |
# get backups for a host |
479 |
|
|
my @backups = $bpc->BackupInfoRead($hostname); |
480 |
dpavlin |
67 |
my $incs = scalar @backups; |
481 |
|
|
print "$incs increments\n"; |
482 |
dpavlin |
6 |
|
483 |
dpavlin |
14 |
my $inc_nr = 0; |
484 |
dpavlin |
67 |
$beenThere = {}; |
485 |
dpavlin |
14 |
|
486 |
|
|
foreach my $backup (@backups) { |
487 |
dpavlin |
40 |
|
488 |
dpavlin |
14 |
$inc_nr++; |
489 |
|
|
last if ($opt{m} && $inc_nr > $opt{m}); |
490 |
|
|
|
491 |
|
|
my $backupNum = $backup->{'num'}; |
492 |
|
|
my @backupShares = (); |
493 |
|
|
|
494 |
dpavlin |
67 |
printf("%-10s %2d/%-2d #%-2d %s %5s/%5s files (date: %s dur: %s)\n", |
495 |
|
|
$hosts->{$host_key}->{'host'}, |
496 |
|
|
$inc_nr, $incs, $backupNum, |
497 |
|
|
$backup->{type} || '?', |
498 |
|
|
$backup->{nFilesNew} || '?', $backup->{nFiles} || '?', |
499 |
dpavlin |
57 |
strftime($t_fmt,localtime($backup->{startTime})), |
500 |
dpavlin |
67 |
fmt_time($backup->{endTime} - $backup->{startTime}) |
501 |
|
|
); |
502 |
dpavlin |
14 |
|
503 |
dpavlin |
34 |
my $files = BackupPC::View->new($bpc, $hostname, \@backups, 1); |
504 |
dpavlin |
14 |
foreach my $share ($files->shareList($backupNum)) { |
505 |
|
|
|
506 |
dpavlin |
37 |
my $t = time(); |
507 |
|
|
|
508 |
dpavlin |
14 |
$shareID = getShareID($share, $hostID, $hostname); |
509 |
|
|
|
510 |
dpavlin |
65 |
$sth->{backups_count}->execute($hostID, $backupNum, $shareID); |
511 |
|
|
my ($count) = $sth->{backups_count}->fetchrow_array(); |
512 |
|
|
# skip if allready in database! |
513 |
|
|
next if ($count > 0); |
514 |
|
|
|
515 |
|
|
# dump some log |
516 |
dpavlin |
86 |
print curr_time," ", $share; |
517 |
dpavlin |
65 |
|
518 |
dpavlin |
242 |
$sth->{insert_backups}->execute( |
519 |
|
|
$hostID, |
520 |
|
|
$backupNum, |
521 |
|
|
$backup->{'endTime'}, |
522 |
|
|
substr($backup->{'type'},0,4), |
523 |
|
|
$shareID, |
524 |
|
|
); |
525 |
|
|
|
526 |
dpavlin |
66 |
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
527 |
dpavlin |
65 |
|
528 |
dpavlin |
242 |
$sth->{update_backups_size}->execute( |
529 |
|
|
$size, |
530 |
dpavlin |
65 |
$hostID, |
531 |
|
|
$backupNum, |
532 |
|
|
$backup->{'endTime'}, |
533 |
dpavlin |
130 |
substr($backup->{'type'},0,4), |
534 |
dpavlin |
66 |
$shareID, |
535 |
dpavlin |
65 |
); |
536 |
|
|
|
537 |
|
|
print " commit"; |
538 |
|
|
$dbh->commit(); |
539 |
|
|
|
540 |
dpavlin |
50 |
my $dur = (time() - $t) || 1; |
541 |
dpavlin |
66 |
printf(" %d/%d files %d/%d dirs %0.2f MB [%.2f/s dur: %s]\n", |
542 |
dpavlin |
37 |
$nf, $f, $nd, $d, |
543 |
dpavlin |
66 |
($size / 1024 / 1024), |
544 |
dpavlin |
50 |
( ($f+$d) / $dur ), |
545 |
|
|
fmt_time($dur) |
546 |
dpavlin |
37 |
); |
547 |
dpavlin |
89 |
|
548 |
dpavlin |
107 |
hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0); |
549 |
dpavlin |
14 |
} |
550 |
dpavlin |
29 |
|
551 |
dpavlin |
6 |
} |
552 |
|
|
} |
553 |
dpavlin |
14 |
undef $sth; |
554 |
dpavlin |
6 |
$dbh->commit(); |
555 |
|
|
$dbh->disconnect(); |
556 |
|
|
|
557 |
dpavlin |
51 |
print "total duration: ",fmt_time(time() - $start_t),"\n"; |
558 |
|
|
|
559 |
dpavlin |
38 |
$pidfile->remove; |
560 |
|
|
|
561 |
dpavlin |
14 |
sub getShareID() { |
562 |
dpavlin |
6 |
|
563 |
dpavlin |
14 |
my ($share, $hostID, $hostname) = @_; |
564 |
|
|
|
565 |
|
|
$sth->{share_id} ||= $dbh->prepare(qq{ |
566 |
|
|
SELECT ID FROM shares WHERE hostID=? AND name=? |
567 |
|
|
}); |
568 |
|
|
|
569 |
|
|
$sth->{share_id}->execute($hostID,$share); |
570 |
|
|
|
571 |
|
|
my ($id) = $sth->{share_id}->fetchrow_array(); |
572 |
|
|
|
573 |
|
|
return $id if (defined($id)); |
574 |
|
|
|
575 |
|
|
$sth->{insert_share} ||= $dbh->prepare(qq{ |
576 |
|
|
INSERT INTO shares |
577 |
dpavlin |
152 |
(hostID,name,share) |
578 |
|
|
VALUES (?,?,?) |
579 |
dpavlin |
14 |
}); |
580 |
|
|
|
581 |
dpavlin |
25 |
my $drop_down = $hostname . '/' . $share; |
582 |
|
|
$drop_down =~ s#//+#/#g; |
583 |
|
|
|
584 |
dpavlin |
152 |
$sth->{insert_share}->execute($hostID,$share, $drop_down); |
585 |
dpavlin |
49 |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
586 |
dpavlin |
6 |
} |
587 |
|
|
|
588 |
dpavlin |
14 |
sub found_in_db { |
589 |
|
|
|
590 |
dpavlin |
48 |
my @data = @_; |
591 |
|
|
shift @data; |
592 |
dpavlin |
14 |
|
593 |
dpavlin |
74 |
my ($key, $shareID,undef,$name,$path,$date,undef,$size) = @_; |
594 |
dpavlin |
48 |
|
595 |
|
|
return $beenThere->{$key} if (defined($beenThere->{$key})); |
596 |
|
|
|
597 |
dpavlin |
14 |
$sth->{file_in_db} ||= $dbh->prepare(qq{ |
598 |
dpavlin |
48 |
SELECT 1 FROM files |
599 |
dpavlin |
14 |
WHERE shareID = ? and |
600 |
|
|
path = ? and |
601 |
dpavlin |
246 |
size = ? and |
602 |
|
|
( date = ? or date = ? or date = ? ) |
603 |
dpavlin |
74 |
LIMIT 1 |
604 |
dpavlin |
14 |
}); |
605 |
|
|
|
606 |
dpavlin |
246 |
my @param = ($shareID,$path,$size,$date, $date-$dst_offset, $date+$dst_offset); |
607 |
dpavlin |
14 |
$sth->{file_in_db}->execute(@param); |
608 |
dpavlin |
48 |
my $rows = $sth->{file_in_db}->rows; |
609 |
dpavlin |
74 |
print STDERR "## found_in_db($shareID,$path,$date,$size) ",( $rows ? '+' : '-' ), join(" ",@param), "\n" if ($debug >= 3); |
610 |
dpavlin |
48 |
|
611 |
|
|
$beenThere->{$key}++; |
612 |
|
|
|
613 |
|
|
$sth->{'insert_files'}->execute(@data) unless ($rows); |
614 |
dpavlin |
14 |
return $rows; |
615 |
dpavlin |
6 |
} |
616 |
|
|
|
617 |
|
|
#################################################### |
618 |
|
|
# recursing through filesystem structure and # |
619 |
|
|
# and returning flattened files list # |
620 |
|
|
#################################################### |
621 |
dpavlin |
14 |
sub recurseDir($$$$$$$$) { |
622 |
dpavlin |
6 |
|
623 |
dpavlin |
35 |
my ($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID) = @_; |
624 |
dpavlin |
14 |
|
625 |
dpavlin |
44 |
print STDERR "\nrecurse($hostname,$backupNum,$share,$dir,$shareID)\n" if ($debug >= 1); |
626 |
dpavlin |
29 |
|
627 |
dpavlin |
66 |
my ($nr_files, $new_files, $nr_dirs, $new_dirs, $size) = (0,0,0,0,0); |
628 |
dpavlin |
14 |
|
629 |
dpavlin |
27 |
{ # scope |
630 |
dpavlin |
29 |
my @stack; |
631 |
dpavlin |
14 |
|
632 |
dpavlin |
45 |
print STDERR "# dirAttrib($backupNum, $share, $dir)\n" if ($debug >= 2); |
633 |
dpavlin |
27 |
my $filesInBackup = $files->dirAttrib($backupNum, $share, $dir); |
634 |
dpavlin |
14 |
|
635 |
dpavlin |
27 |
# first, add all the entries in current directory |
636 |
|
|
foreach my $path_key (keys %{$filesInBackup}) { |
637 |
dpavlin |
66 |
print STDERR "# file ",Dumper($filesInBackup->{$path_key}),"\n" if ($debug >= 3); |
638 |
dpavlin |
27 |
my @data = ( |
639 |
|
|
$shareID, |
640 |
|
|
$backupNum, |
641 |
|
|
$path_key, |
642 |
|
|
$filesInBackup->{$path_key}->{'relPath'}, |
643 |
|
|
$filesInBackup->{$path_key}->{'mtime'}, |
644 |
|
|
$filesInBackup->{$path_key}->{'type'}, |
645 |
|
|
$filesInBackup->{$path_key}->{'size'} |
646 |
|
|
); |
647 |
|
|
|
648 |
|
|
my $key = join(" ", ( |
649 |
|
|
$shareID, |
650 |
|
|
$dir, |
651 |
|
|
$path_key, |
652 |
|
|
$filesInBackup->{$path_key}->{'mtime'}, |
653 |
|
|
$filesInBackup->{$path_key}->{'size'} |
654 |
|
|
)); |
655 |
|
|
|
656 |
dpavlin |
242 |
my $key_dst_prev = join(" ", ( |
657 |
dpavlin |
240 |
$shareID, |
658 |
|
|
$dir, |
659 |
|
|
$path_key, |
660 |
|
|
$filesInBackup->{$path_key}->{'mtime'} - $dst_offset, |
661 |
|
|
$filesInBackup->{$path_key}->{'size'} |
662 |
|
|
)); |
663 |
|
|
|
664 |
dpavlin |
242 |
my $key_dst_next = join(" ", ( |
665 |
dpavlin |
240 |
$shareID, |
666 |
|
|
$dir, |
667 |
|
|
$path_key, |
668 |
|
|
$filesInBackup->{$path_key}->{'mtime'} + $dst_offset, |
669 |
|
|
$filesInBackup->{$path_key}->{'size'} |
670 |
|
|
)); |
671 |
|
|
|
672 |
dpavlin |
70 |
my $found; |
673 |
dpavlin |
240 |
if ( |
674 |
|
|
! defined($beenThere->{$key}) && |
675 |
dpavlin |
242 |
! defined($beenThere->{$key_dst_prev}) && |
676 |
|
|
! defined($beenThere->{$key_dst_next}) && |
677 |
dpavlin |
240 |
! ($found = found_in_db($key, @data)) |
678 |
|
|
) { |
679 |
dpavlin |
30 |
print STDERR "# key: $key [", $beenThere->{$key},"]" if ($debug >= 2); |
680 |
dpavlin |
48 |
|
681 |
dpavlin |
27 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
682 |
dpavlin |
70 |
$new_dirs++ unless ($found); |
683 |
dpavlin |
30 |
print STDERR " dir\n" if ($debug >= 2); |
684 |
dpavlin |
27 |
} else { |
685 |
dpavlin |
70 |
$new_files++ unless ($found); |
686 |
dpavlin |
30 |
print STDERR " file\n" if ($debug >= 2); |
687 |
dpavlin |
27 |
} |
688 |
dpavlin |
66 |
$size += $filesInBackup->{$path_key}->{'size'} || 0; |
689 |
dpavlin |
27 |
} |
690 |
|
|
|
691 |
dpavlin |
14 |
if ($filesInBackup->{$path_key}->{'type'} == BPC_FTYPE_DIR) { |
692 |
dpavlin |
27 |
$nr_dirs++; |
693 |
|
|
|
694 |
dpavlin |
29 |
my $full_path = $dir . '/' . $path_key; |
695 |
|
|
push @stack, $full_path; |
696 |
dpavlin |
30 |
print STDERR "### store to stack: $full_path\n" if ($debug >= 3); |
697 |
dpavlin |
29 |
|
698 |
dpavlin |
27 |
# my ($f,$nf,$d,$nd) = recurseDir($bpc, $hostname, $backups, $backupNum, $share, $path_key, $shareID) unless ($beenThere->{$key}); |
699 |
|
|
# |
700 |
|
|
# $nr_files += $f; |
701 |
|
|
# $new_files += $nf; |
702 |
|
|
# $nr_dirs += $d; |
703 |
|
|
# $new_dirs += $nd; |
704 |
|
|
|
705 |
dpavlin |
14 |
} else { |
706 |
dpavlin |
27 |
$nr_files++; |
707 |
dpavlin |
14 |
} |
708 |
|
|
} |
709 |
|
|
|
710 |
dpavlin |
30 |
print STDERR "## STACK ",join(", ", @stack),"\n" if ($debug >= 2); |
711 |
dpavlin |
14 |
|
712 |
dpavlin |
29 |
while ( my $dir = shift @stack ) { |
713 |
dpavlin |
66 |
my ($f,$nf,$d,$nd, $s) = recurseDir($bpc, $hostname, $files, $backupNum, $share, $dir, $shareID); |
714 |
dpavlin |
30 |
print STDERR "# $dir f: $f nf: $nf d: $d nd: $nd\n" if ($debug >= 1); |
715 |
dpavlin |
29 |
$nr_files += $f; |
716 |
|
|
$new_files += $nf; |
717 |
|
|
$nr_dirs += $d; |
718 |
|
|
$new_dirs += $nd; |
719 |
dpavlin |
66 |
$size += $s; |
720 |
dpavlin |
29 |
} |
721 |
dpavlin |
14 |
} |
722 |
|
|
|
723 |
dpavlin |
66 |
return ($nr_files, $new_files, $nr_dirs, $new_dirs, $size); |
724 |
dpavlin |
6 |
} |
725 |
|
|
|