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; |
18 |
|
|
19 |
my $debug = 0; |
my $debug = 0; |
20 |
$|=1; |
$|=1; |
43 |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n"; |
44 |
my $user = $Conf{SearchUser} || ''; |
my $user = $Conf{SearchUser} || ''; |
45 |
|
|
46 |
|
my $use_hest = $Conf{HyperEstraierIndex}; |
47 |
|
my ($index_path, $index_node_url) = BackupPC::SearchLib::getHyperEstraier_url($use_hest); |
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:ij", \%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 $debug) |
62 |
|
-i update Hyper Estraier full text index |
63 |
|
-j update full text, don't check existing files |
64 |
|
|
65 |
|
Option -j is variation on -i. It will allow faster initial creation |
66 |
|
of full-text index from existing database. |
67 |
|
|
68 |
EOF |
EOF |
69 |
exit 1; |
exit 1; |
70 |
} |
} |
71 |
|
|
72 |
###################################create tables############################3 |
if ($opt{v}) { |
73 |
|
print "Debug level at $opt{v}\n"; |
74 |
|
$debug = $opt{v}; |
75 |
|
} |
76 |
|
|
77 |
|
#---- subs ---- |
78 |
|
|
79 |
|
sub fmt_time { |
80 |
|
my $t = shift || return; |
81 |
|
my $out = ""; |
82 |
|
my ($ss,$mm,$hh) = gmtime($t); |
83 |
|
$out .= "${hh}h" if ($hh); |
84 |
|
$out .= sprintf("%02d:%02d", $mm,$ss); |
85 |
|
return $out; |
86 |
|
} |
87 |
|
|
88 |
|
sub curr_time { |
89 |
|
return strftime($t_fmt,localtime()); |
90 |
|
} |
91 |
|
|
92 |
|
my $hest_db; |
93 |
|
my $hest_node; |
94 |
|
|
95 |
|
sub signal { |
96 |
|
my($sig) = @_; |
97 |
|
if ($hest_db) { |
98 |
|
print "\nCaught a SIG$sig--syncing database and shutting down\n"; |
99 |
|
$hest_db->sync(); |
100 |
|
$hest_db->close(); |
101 |
|
} |
102 |
|
exit(0); |
103 |
|
} |
104 |
|
|
105 |
|
$SIG{'INT'} = \&signal; |
106 |
|
$SIG{'QUIT'} = \&signal; |
107 |
|
|
108 |
|
sub hest_update { |
109 |
|
|
110 |
|
my ($host_id, $share_id, $num) = @_; |
111 |
|
|
112 |
|
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
113 |
|
|
114 |
|
unless ($use_hest) { |
115 |
|
print STDERR "HyperEstraier support not enabled in configuration\n"; |
116 |
|
return; |
117 |
|
} |
118 |
|
|
119 |
|
print curr_time," updating HyperEstraier:"; |
120 |
|
|
121 |
|
my $t = time(); |
122 |
|
|
123 |
|
my $offset = 0; |
124 |
|
my $added = 0; |
125 |
|
|
126 |
|
print " opening index $use_hest"; |
127 |
|
if ($index_path) { |
128 |
|
$hest_db = HyperEstraier::Database->new(); |
129 |
|
$hest_db->open($TopDir . $index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
130 |
|
print " directly"; |
131 |
|
} elsif ($index_node_url) { |
132 |
|
$hest_node ||= HyperEstraier::Node->new($index_node_url); |
133 |
|
$hest_node->set_auth('admin', 'admin'); |
134 |
|
print " via node URL"; |
135 |
|
} else { |
136 |
|
die "don't know how to use HyperEstraier Index $use_hest"; |
137 |
|
} |
138 |
|
print " increment is " . EST_CHUNK . " files:"; |
139 |
|
|
140 |
|
my $results = 0; |
141 |
|
|
142 |
|
do { |
143 |
|
|
144 |
|
my $where = ''; |
145 |
|
my @data; |
146 |
|
if ($host_id && $share_id && $num) { |
147 |
|
$where = qq{ |
148 |
|
WHERE |
149 |
|
hosts.id = ? AND |
150 |
|
shares.id = ? AND |
151 |
|
files.backupnum = ? |
152 |
|
}; |
153 |
|
@data = ( $host_id, $share_id, $num ); |
154 |
|
} |
155 |
|
|
156 |
|
my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset); |
157 |
|
|
158 |
|
my $sth = $dbh->prepare(qq{ |
159 |
|
SELECT |
160 |
|
files.id AS fid, |
161 |
|
hosts.name AS hname, |
162 |
|
shares.name AS sname, |
163 |
|
-- shares.share AS sharename, |
164 |
|
files.backupnum AS backupnum, |
165 |
|
-- files.name AS filename, |
166 |
|
files.path AS filepath, |
167 |
|
files.date AS date, |
168 |
|
files.type AS type, |
169 |
|
files.size AS size, |
170 |
|
files.shareid AS shareid, |
171 |
|
backups.date AS backup_date |
172 |
|
FROM files |
173 |
|
INNER JOIN shares ON files.shareID=shares.ID |
174 |
|
INNER JOIN hosts ON hosts.ID = shares.hostID |
175 |
|
INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID |
176 |
|
$where |
177 |
|
$limit |
178 |
|
}); |
179 |
|
|
180 |
|
$sth->execute(@data); |
181 |
|
$results = $sth->rows; |
182 |
|
|
183 |
|
if ($results == 0) { |
184 |
|
print " - no new files\n"; |
185 |
|
last; |
186 |
|
} |
187 |
|
|
188 |
|
sub fmt_date { |
189 |
|
my $t = shift || return; |
190 |
|
my $iso = BackupPC::Lib::timeStamp($t); |
191 |
|
$iso =~ s/\s/T/; |
192 |
|
return $iso; |
193 |
|
} |
194 |
|
|
195 |
|
while (my $row = $sth->fetchrow_hashref()) { |
196 |
|
|
197 |
|
my $fid = $row->{'fid'} || die "no fid?"; |
198 |
|
my $uri = 'file:///' . $fid; |
199 |
|
|
200 |
|
unless ($skip_check) { |
201 |
|
my $id = ($hest_db || $hest_node)->uri_to_id($uri); |
202 |
|
next unless ($id == -1); |
203 |
|
} |
204 |
|
|
205 |
|
# create a document object |
206 |
|
my $doc = HyperEstraier::Document->new; |
207 |
|
|
208 |
|
# add attributes to the document object |
209 |
|
$doc->add_attr('@uri', $uri); |
210 |
|
|
211 |
|
foreach my $c (@{ $sth->{NAME} }) { |
212 |
|
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
213 |
|
} |
214 |
|
|
215 |
|
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
216 |
|
|
217 |
|
# add the body text to the document object |
218 |
|
my $path = $row->{'filepath'}; |
219 |
|
$doc->add_text($path); |
220 |
|
$path =~ s/(.)/$1 /g; |
221 |
|
$doc->add_hidden_text($path); |
222 |
|
|
223 |
|
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
224 |
|
|
225 |
|
# register the document object to the database |
226 |
|
if ($hest_db) { |
227 |
|
$hest_db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
228 |
|
} elsif ($hest_node) { |
229 |
|
$hest_node->put_doc($doc); |
230 |
|
} else { |
231 |
|
die "not supported"; |
232 |
|
} |
233 |
|
$added++; |
234 |
|
} |
235 |
|
|
236 |
|
print " $added"; |
237 |
|
$hest_db->sync() if ($index_path); |
238 |
|
|
239 |
|
$offset += EST_CHUNK; |
240 |
|
|
241 |
|
} while ($results == EST_CHUNK); |
242 |
|
|
243 |
|
if ($index_path) { |
244 |
|
print ", close"; |
245 |
|
$hest_db->close(); |
246 |
|
} |
247 |
|
|
248 |
|
my $dur = (time() - $t) || 1; |
249 |
|
printf(" [%.2f/s dur: %s]\n", |
250 |
|
( $added / $dur ), |
251 |
|
fmt_time($dur) |
252 |
|
); |
253 |
|
} |
254 |
|
|
255 |
|
#---- /subs ---- |
256 |
|
|
257 |
|
|
258 |
|
## update index ## |
259 |
|
if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
260 |
|
# update all |
261 |
|
print "force update of HyperEstraier index "; |
262 |
|
print "importing existing data" unless (-e $index_path); |
263 |
|
print "by -i flag" if ($opt{i}); |
264 |
|
print "by -j flag" if ($opt{j}); |
265 |
|
print "\n"; |
266 |
|
hest_update(); |
267 |
|
} |
268 |
|
|
269 |
|
## create tables ## |
270 |
if ($opt{c}) { |
if ($opt{c}) { |
271 |
sub do_index { |
sub do_index { |
272 |
my $index = shift || return; |
my $index = shift || return; |
273 |
my ($table,$col,$unique) = split(/_/, $index); |
my ($table,$col,$unique) = split(/:/, $index); |
274 |
$unique ||= ''; |
$unique ||= ''; |
275 |
$index =~ s/,/_/g; |
$index =~ s/\W+/_/g; |
276 |
|
print "$index on $table($col)" . ( $unique ? "u" : "" ) . " "; |
277 |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
$dbh->do(qq{ create $unique index $index on $table($col) }); |
278 |
} |
} |
279 |
|
|
292 |
ID SERIAL PRIMARY KEY, |
ID SERIAL PRIMARY KEY, |
293 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
294 |
name VARCHAR(30) NOT NULL, |
name VARCHAR(30) NOT NULL, |
295 |
share VARCHAR(200) NOT NULL, |
share VARCHAR(200) NOT NULL |
|
localpath VARCHAR(200) |
|
296 |
); |
); |
297 |
}); |
}); |
298 |
|
|
299 |
|
$dbh->do(qq{ |
300 |
|
create table dvds ( |
301 |
|
ID SERIAL PRIMARY KEY, |
302 |
|
num INTEGER NOT NULL, |
303 |
|
name VARCHAR(255) NOT NULL, |
304 |
|
mjesto VARCHAR(255) |
305 |
|
); |
306 |
|
}); |
307 |
|
|
308 |
$dbh->do(qq{ |
$dbh->do(qq{ |
309 |
create table backups ( |
create table backups ( |
310 |
|
id serial, |
311 |
hostID INTEGER NOT NULL references hosts(id), |
hostID INTEGER NOT NULL references hosts(id), |
312 |
num INTEGER NOT NULL, |
num INTEGER NOT NULL, |
313 |
date integer NOT NULL, |
date integer NOT NULL, |
314 |
type CHAR(4) not null, |
type CHAR(4) not null, |
315 |
shareID integer not null references shares(id), |
shareID integer not null references shares(id), |
316 |
size integer not null, |
size bigint not null, |
317 |
PRIMARY KEY(hostID, num, shareID) |
inc_size bigint not null default -1, |
318 |
|
inc_deleted boolean default false, |
319 |
|
PRIMARY KEY(id) |
320 |
); |
); |
321 |
}); |
}); |
322 |
|
|
323 |
#do_index('backups_hostid,num_unique'); |
$dbh->do(qq{ |
324 |
|
create table files ( |
325 |
$dbh->do(qq{ |
ID SERIAL, |
326 |
create table dvds ( |
shareID INTEGER NOT NULL references shares(id), |
327 |
ID SERIAL PRIMARY KEY, |
backupNum INTEGER NOT NULL, |
328 |
num INTEGER NOT NULL, |
name VARCHAR(255) NOT NULL, |
329 |
name VARCHAR(255) NOT NULL, |
path VARCHAR(255) NOT NULL, |
330 |
mjesto VARCHAR(255) |
date integer NOT NULL, |
331 |
|
type INTEGER NOT NULL, |
332 |
|
size bigint NOT NULL, |
333 |
|
primary key(id) |
334 |
); |
); |
335 |
}); |
}); |
336 |
|
|
337 |
$dbh->do(qq{ |
|
338 |
create table files ( |
$dbh->do( qq{ |
339 |
ID SERIAL PRIMARY KEY, |
create table archive ( |
340 |
shareID INTEGER NOT NULL references shares(id), |
id serial, |
341 |
backupNum INTEGER NOT NULL, |
dvd_nr int not null, |
342 |
name VARCHAR(255) NOT NULL, |
total_size bigint default -1, |
343 |
path VARCHAR(255) NOT NULL, |
note text, |
344 |
date integer NOT NULL, |
username varchar(20) not null, |
345 |
type INTEGER NOT NULL, |
date timestamp default now(), |
346 |
size INTEGER NOT NULL, |
primary key(id) |
347 |
dvdid INTEGER references dvds(id) |
); |
348 |
|
} |
349 |
|
); |
350 |
|
|
351 |
|
$dbh->do( qq{ |
352 |
|
create table archive_backup |
353 |
|
( |
354 |
|
archive_id int not null references archive(id) on delete cascade, |
355 |
|
backup_id int not null references backups(id), |
356 |
|
primary key(archive_id, backup_id) |
357 |
); |
); |
358 |
}); |
}); |
359 |
|
|
360 |
print "creating indexes:"; |
print "creating indexes: "; |
361 |
|
|
362 |
foreach my $index (qw( |
foreach my $index (qw( |
363 |
hosts_name |
hosts:name |
364 |
backups_hostID |
backups:hostID |
365 |
backups_num |
backups:num |
366 |
shares_hostID |
backups:shareID |
367 |
shares_name |
shares:hostID |
368 |
files_shareID |
shares:name |
369 |
files_path |
files:shareID |
370 |
files_name |
files:path |
371 |
files_date |
files:name |
372 |
files_size |
files:date |
373 |
|
files:size |
374 |
|
archive:dvd_nr |
375 |
)) { |
)) { |
|
print " $index"; |
|
376 |
do_index($index); |
do_index($index); |
377 |
} |
} |
378 |
|
|
379 |
|
print " creating sequence: "; |
380 |
|
foreach my $seq (qw/dvd_nr/) { |
381 |
|
print "$seq "; |
382 |
|
$dbh->do( qq{ CREATE SEQUENCE $seq } ); |
383 |
|
} |
384 |
|
|
385 |
|
|
386 |
print "...\n"; |
print "...\n"; |
387 |
|
|
388 |
$dbh->commit; |
$dbh->commit; |
389 |
|
|
390 |
} |
} |
391 |
|
|
392 |
|
## delete data before inseting ## |
393 |
if ($opt{d}) { |
if ($opt{d}) { |
394 |
print "deleting "; |
print "deleting "; |
395 |
foreach my $table (qw(files dvds backups shares hosts)) { |
foreach my $table (qw(files dvds backups shares hosts)) { |
401 |
$dbh->commit; |
$dbh->commit; |
402 |
} |
} |
403 |
|
|
404 |
if ($opt{v}) { |
## insert new values ## |
|
print "Debug level at $opt{v}\n"; |
|
|
$debug = $opt{v}; |
|
|
} |
|
|
|
|
|
#################################INSERT VALUES############################# |
|
405 |
|
|
406 |
# get hosts |
# get hosts |
407 |
$hosts = $bpc->HostInfoRead(); |
$hosts = $bpc->HostInfoRead(); |
435 |
VALUES (?,?,?,?,?,?,?) |
VALUES (?,?,?,?,?,?,?) |
436 |
}); |
}); |
437 |
|
|
|
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; |
|
|
} |
|
|
|
|
438 |
foreach my $host_key (keys %{$hosts}) { |
foreach my $host_key (keys %{$hosts}) { |
439 |
|
|
440 |
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 $host_key"; |
490 |
next if ($count > 0); |
next if ($count > 0); |
491 |
|
|
492 |
# dump some log |
# dump some log |
493 |
print strftime($t_fmt,localtime())," ", $share; |
print curr_time," ", $share; |
494 |
|
|
495 |
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID); |
496 |
|
|
498 |
$hostID, |
$hostID, |
499 |
$backupNum, |
$backupNum, |
500 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
501 |
$backup->{'type'}, |
substr($backup->{'type'},0,4), |
502 |
$shareID, |
$shareID, |
503 |
$size, |
$size, |
504 |
); |
); |
513 |
( ($f+$d) / $dur ), |
( ($f+$d) / $dur ), |
514 |
fmt_time($dur) |
fmt_time($dur) |
515 |
); |
); |
516 |
|
|
517 |
|
hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0); |
518 |
} |
} |
519 |
|
|
520 |
} |
} |
543 |
|
|
544 |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
$sth->{insert_share} ||= $dbh->prepare(qq{ |
545 |
INSERT INTO shares |
INSERT INTO shares |
546 |
(hostID,name,share,localpath) |
(hostID,name,share) |
547 |
VALUES (?,?,?,?) |
VALUES (?,?,?) |
548 |
}); |
}); |
549 |
|
|
550 |
my $drop_down = $hostname . '/' . $share; |
my $drop_down = $hostname . '/' . $share; |
551 |
$drop_down =~ s#//+#/#g; |
$drop_down =~ s#//+#/#g; |
552 |
|
|
553 |
$sth->{insert_share}->execute($hostID,$share, $drop_down ,undef); |
$sth->{insert_share}->execute($hostID,$share, $drop_down); |
554 |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
return $dbh->last_insert_id(undef,undef,'shares',undef); |
555 |
} |
} |
556 |
|
|