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; |
42 |
|
|
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} || ''; |
|
my $index_path = $Conf{HyperEstraierIndex}; |
|
|
$index_path = $TopDir . '/' . $index_path; |
|
|
$index_path =~ s#//#/#g; |
|
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:i", \%opt ) ) { |
if ( !getopts("cdm:v:ij", \%opt ) ) { |
54 |
print STDERR <<EOF; |
print STDERR <<EOF; |
55 |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
usage: $0 [-c|-d] [-m num] [-v|-v level] [-i] |
56 |
|
|
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 |
-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 |
} |
} |
90 |
} |
} |
91 |
|
|
92 |
my $hest_db; |
my $hest_db; |
93 |
|
my $hest_node; |
94 |
|
|
95 |
sub hest_update { |
sub signal { |
96 |
|
my($sig) = @_; |
97 |
my ($host_id, $share_id, $num) = @_; |
if ($hest_db) { |
98 |
|
print "\nCaught a SIG$sig--syncing database and shutting down\n"; |
99 |
print curr_time," updating HyperEstraier: files"; |
$hest_db->sync(); |
100 |
|
$hest_db->close(); |
101 |
|
} |
102 |
|
exit(0); |
103 |
|
} |
104 |
|
|
105 |
my $t = time(); |
$SIG{'INT'} = \&signal; |
106 |
|
$SIG{'QUIT'} = \&signal; |
107 |
|
|
108 |
my $where = ''; |
sub hest_update { |
|
if ($host_id && $share_id && $num) { |
|
|
$where = qq{ |
|
|
WHERE |
|
|
hosts.id = ? AND |
|
|
shares.id = ? AND |
|
|
files.backupnum = ? |
|
|
}; |
|
|
} |
|
109 |
|
|
110 |
my $sth = $dbh->prepare(qq{ |
my ($host_id, $share_id, $num) = @_; |
|
SELECT |
|
|
files.id AS fid, |
|
|
hosts.name AS hname, |
|
|
shares.name AS sname, |
|
|
-- shares.share AS sharename, |
|
|
files.backupnum AS backupnum, |
|
|
-- files.name AS filename, |
|
|
files.path AS filepath, |
|
|
files.date AS date, |
|
|
files.type AS filetype, |
|
|
files.size AS size, |
|
|
files.shareid AS shareid, |
|
|
backups.date AS backup_date |
|
|
FROM files |
|
|
INNER JOIN shares ON files.shareID=shares.ID |
|
|
INNER JOIN hosts ON hosts.ID = shares.hostID |
|
|
INNER JOIN backups ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID |
|
|
$where |
|
|
}); |
|
111 |
|
|
112 |
$sth->execute(@_); |
my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n"; |
|
my $results = $sth->rows; |
|
113 |
|
|
114 |
if ($results == 0) { |
unless ($use_hest) { |
115 |
print " no files\n"; |
print STDERR "HyperEstraier support not enabled in configuration\n"; |
116 |
return; |
return; |
117 |
} |
} |
118 |
|
|
119 |
my $dot = int($results / 15) || 1; |
print curr_time," updating HyperEstraier:"; |
120 |
|
|
121 |
print " $results ($dot/#)"; |
my $t = time(); |
|
|
|
|
sub fmt_date { |
|
|
my $t = shift || return; |
|
|
my $iso = BackupPC::Lib::timeStamp($t); |
|
|
$iso =~ s/\s/T/; |
|
|
return $iso; |
|
|
} |
|
|
|
|
|
my $max = int($results / $dot); |
|
122 |
|
|
123 |
print " index $index_path..."; |
my $offset = 0; |
124 |
use HyperEstraier; |
my $added = 0; |
|
my $db = HyperEstraier::Database->new(); |
|
125 |
|
|
126 |
# unless ($hest_db) { |
print " opening index $use_hest"; |
127 |
# print " open reader"; |
if ($index_path) { |
128 |
# $hest_db = HyperEstraier::Database->new(); |
$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 |
$db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT); |
do { |
143 |
|
|
144 |
my $added = 0; |
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 |
while (my $row = $sth->fetchrow_hashref()) { |
my $limit = sprintf('LIMIT '.EST_CHUNK.' OFFSET %d', $offset); |
157 |
|
|
158 |
my $fid = $row->{'fid'} || die "no fid?"; |
my $sth = $dbh->prepare(qq{ |
159 |
my $uri = 'file:///' . $fid; |
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 |
next if ($db->uri_to_id($uri)); |
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 |
# create a document object |
while (my $row = $sth->fetchrow_hashref()) { |
|
my $doc = HyperEstraier::Document->new; |
|
196 |
|
|
197 |
# add attributes to the document object |
my $fid = $row->{'fid'} || die "no fid?"; |
198 |
$doc->add_attr('@uri', $uri); |
my $uri = 'file:///' . $fid; |
199 |
|
|
200 |
foreach my $c (@{ $sth->{NAME} }) { |
unless ($skip_check) { |
201 |
$doc->add_attr($c, $row->{$c}) if ($row->{$c}); |
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 |
|
$added++; |
234 |
} |
} |
235 |
|
|
236 |
#$doc->add_attr('@cdate', fmt_date($row->{'date'})); |
print " $added"; |
237 |
|
$hest_db->sync() if ($index_path); |
238 |
|
|
239 |
# add the body text to the document object |
$offset += EST_CHUNK; |
|
my $path = $row->{'filepath'}; |
|
|
$doc->add_text($path); |
|
|
$path =~ s/(.)/$1 /g; |
|
|
$doc->add_hidden_text($path); |
|
|
|
|
|
print STDERR $doc->dump_draft,"\n" if ($debug > 1); |
|
|
|
|
|
# register the document object to the database |
|
|
$db->put_doc($doc, $HyperEstraier::Database::PDCLEAN); |
|
|
|
|
|
$added++; |
|
|
if ($added % $dot == 0) { |
|
|
print "$max "; |
|
|
$max--; |
|
|
} |
|
240 |
|
|
241 |
} |
} while ($results == EST_CHUNK); |
242 |
|
|
243 |
print "sync $added new files"; |
if ($index_path) { |
244 |
$db->sync(); |
print ", close"; |
245 |
print " close"; |
$hest_db->close(); |
246 |
$db->close(); |
} |
247 |
|
|
248 |
my $dur = (time() - $t) || 1; |
my $dur = (time() - $t) || 1; |
249 |
printf(" [%.2f/s new %.2f/s dur: %s]\n", |
printf(" [%.2f/s dur: %s]\n", |
|
( $results / $dur ), |
|
250 |
( $added / $dur ), |
( $added / $dur ), |
251 |
fmt_time($dur) |
fmt_time($dur) |
252 |
); |
); |
256 |
|
|
257 |
|
|
258 |
## update index ## |
## update index ## |
259 |
if ($opt{i} || ($index_path && ! -e $index_path)) { |
if (($opt{i} || $opt{j} || ($index_path && ! -e $index_path)) && !$opt{c}) { |
260 |
# update all |
# update all |
261 |
print "force update of HyperEstraier index "; |
print "force update of HyperEstraier index "; |
262 |
print "importing existing data" unless (-e $index_path); |
print "importing existing data" unless (-e $index_path); |
263 |
print "by -i flag" if ($opt{i}); |
print "by -i flag" if ($opt{i}); |
264 |
|
print "by -j flag" if ($opt{j}); |
265 |
print "\n"; |
print "\n"; |
266 |
hest_update(); |
hest_update(); |
267 |
} |
} |
295 |
localpath VARCHAR(200) |
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 ( |
312 |
date integer NOT NULL, |
date integer NOT NULL, |
313 |
type CHAR(4) not null, |
type CHAR(4) not null, |
314 |
shareID integer not null references shares(id), |
shareID integer not null references shares(id), |
315 |
size integer not null, |
size bigint not null, |
316 |
PRIMARY KEY(hostID, num, shareID) |
PRIMARY KEY(hostID, num, shareID) |
317 |
); |
); |
318 |
}); |
}); |
319 |
|
|
320 |
#do_index('backups_hostid,num_unique'); |
#do_index('backups_hostid,num_unique'); |
321 |
|
|
|
$dbh->do(qq{ |
|
|
create table dvds ( |
|
|
ID SERIAL PRIMARY KEY, |
|
|
num INTEGER NOT NULL, |
|
|
name VARCHAR(255) NOT NULL, |
|
|
mjesto VARCHAR(255) |
|
|
); |
|
|
}); |
|
322 |
|
|
323 |
$dbh->do(qq{ |
$dbh->do(qq{ |
324 |
create table files ( |
create table files ( |
329 |
path VARCHAR(255) NOT NULL, |
path VARCHAR(255) NOT NULL, |
330 |
date integer NOT NULL, |
date integer NOT NULL, |
331 |
type INTEGER NOT NULL, |
type INTEGER NOT NULL, |
332 |
size INTEGER NOT NULL, |
size INTEGER NOT NULL |
|
dvdid INTEGER references dvds(id) |
|
333 |
); |
); |
334 |
}); |
}); |
335 |
|
|
336 |
|
|
337 |
|
$dbh->do( qq{ |
338 |
|
create table archive |
339 |
|
( |
340 |
|
id int not null, |
341 |
|
dvd_nr int not null, |
342 |
|
note text, |
343 |
|
username varchar(20) not null, |
344 |
|
date timestamp, |
345 |
|
primary key(id) |
346 |
|
); |
347 |
|
} |
348 |
|
); |
349 |
|
|
350 |
|
$dbh->do( qq{ |
351 |
|
create table archive_backup |
352 |
|
( |
353 |
|
archive_id int not null, |
354 |
|
backup_id int not null, |
355 |
|
status text, |
356 |
|
primary key(archive_id, backup_id) |
357 |
|
); |
358 |
|
}); |
359 |
|
|
360 |
|
$dbh->do( qq{ |
361 |
|
create table workflows( |
362 |
|
id int not null, |
363 |
|
step_id int not null, |
364 |
|
start timestamp, |
365 |
|
stop timestamp, |
366 |
|
username varchar(20), |
367 |
|
archive_id int not null, |
368 |
|
running boolean default true, |
369 |
|
primary key(id) |
370 |
|
); |
371 |
|
}); |
372 |
|
|
373 |
|
$dbh->do( qq{ |
374 |
|
create table workflow_step |
375 |
|
( |
376 |
|
step_id int not null, |
377 |
|
code text, |
378 |
|
next_step int, |
379 |
|
stop boolean default false, |
380 |
|
primary key(step_id) |
381 |
|
); |
382 |
|
}); |
383 |
|
|
384 |
|
$dbh->do( qq{ |
385 |
|
alter table workflow_step |
386 |
|
add constraint fk_workflow_next_step |
387 |
|
foreign key(next_step) |
388 |
|
references workflow_step(step_id); |
389 |
|
}); |
390 |
|
|
391 |
|
$dbh->do( qq{ |
392 |
|
alter table workflows |
393 |
|
add constraint fk_workflows_step_id |
394 |
|
foreign key(step_id) |
395 |
|
references workflow_step(step_id); |
396 |
|
}); |
397 |
|
|
398 |
|
$dbh->do( qq{ |
399 |
|
alter table workflows |
400 |
|
add constraint fk_workflows_archive_id |
401 |
|
foreign key(archive_id) |
402 |
|
references archive(id); |
403 |
|
}); |
404 |
|
|
405 |
|
$dbh->do( qq{ |
406 |
|
create table workflow_log |
407 |
|
( |
408 |
|
workflow_id int not null, |
409 |
|
step_id int not null, |
410 |
|
date timestamp not null, |
411 |
|
status text, |
412 |
|
primary key(workflow_id, step_id) |
413 |
|
); |
414 |
|
}); |
415 |
|
|
416 |
|
$dbh->do( qq{ |
417 |
|
alter table workflow_log |
418 |
|
add constraint fk_workflow_log_workflow_id |
419 |
|
foreign key (workflow_id) |
420 |
|
references workflows(id); |
421 |
|
}); |
422 |
|
|
423 |
|
$dbh->do( qq{ |
424 |
|
alter table workflow_log |
425 |
|
add constraint fk_workflow_log_step_id |
426 |
|
foreign key (step_id) |
427 |
|
references workflow_step(step_id); |
428 |
|
}); |
429 |
|
|
430 |
print "creating indexes:"; |
print "creating indexes:"; |
431 |
|
|
432 |
foreach my $index (qw( |
foreach my $index (qw( |
559 |
$hostID, |
$hostID, |
560 |
$backupNum, |
$backupNum, |
561 |
$backup->{'endTime'}, |
$backup->{'endTime'}, |
562 |
$backup->{'type'}, |
substr($backup->{'type'},0,4), |
563 |
$shareID, |
$shareID, |
564 |
$size, |
$size, |
565 |
); |
); |
575 |
fmt_time($dur) |
fmt_time($dur) |
576 |
); |
); |
577 |
|
|
578 |
hest_update($hostID, $shareID, $backupNum); |
hest_update($hostID, $shareID, $backupNum) if ($nf + $nd > 0); |
579 |
} |
} |
580 |
|
|
581 |
} |
} |