2 # Copyright (C) 2008-2010 Equinox Software, Inc.
3 # Author: Bill Erickson <erickson@esilibrary.com>
5 # This program is free software; you can redistribute it and/or
6 # modify it under the terms of the GNU General Public License
7 # as published by the Free Software Foundation; either version 2
8 # of the License, or (at your option) any later version.
10 # This program is distributed in the hope that it will be useful,
11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 # GNU General Public License for more details.
16 use strict; use warnings;
17 use Net::Server::PreFork;
18 use base qw/Net::Server::PreFork/;
22 use MARC::File::USMARC;
25 use File::Basename qw/fileparse/;
27 use Getopt::Long qw(:DEFAULT GetOptionsFromArray);
30 use OpenSRF::Utils::Logger qw/$logger/;
31 use OpenSRF::AppSession;
32 use OpenSRF::EX qw/:try/;
33 use OpenILS::Utils::Cronscript;
34 require 'oils_header.pl';
35 use vars qw/$apputils/;
43 'merge-profile=i' => 0,
45 # 'osrf-config=s' => '/openils/conf/opensrf_core.xml',
53 'import-by-queue' => 0
56 $OpenILS::Utils::Cronscript::debug=1 if $debug;
57 $Getopt::Long::debug=1 if $debug > 1;
58 my $o = OpenILS::Utils::Cronscript->new(\%defaults);
62 if (grep {$_ eq '--'} @ARGV) {
63 print "Splitting options into groups\n" if $debug;
66 $_ eq '--' and last; # stop at the first --
67 push @script_args, $_;
74 print "Calling MyGetOptions ",
75 (@script_args ? "with options: " . join(' ', @script_args) : 'without options from command line'),
78 my $real_opts = $o->MyGetOptions(\@script_args);
80 # GetOptionsFromArray(\@script_args, \%defaults, %defaults); # similar to
82 $real_opts->{tempdir} ||= tempdir_setting(); # This doesn't go in defaults because it reads config, must come after bootstrap
84 my $bufsize = $real_opts->{buffsize};
85 my $bib_source = $real_opts->{source};
86 my $osrf_config = $real_opts->{'osrf-config'};
87 my $oils_username = $real_opts->{user};
88 my $oils_password = $real_opts->{password};
89 my $help = $real_opts->{help};
90 my $merge_profile = $real_opts->{'merge-profile'};
91 my $queue_id = $real_opts->{queue};
92 my $tempdir = $real_opts->{tempdir};
93 my $import_by_queue = $real_opts->{'import-by-queue'};
94 $debug += $real_opts->{debug};
96 foreach (keys %$real_opts) {
97 print("real_opt->{$_} = ", $real_opts->{$_}, "\n") if $real_opts->{debug} or $debug;
99 my $wait_time = $real_opts->{wait};
102 # DEFAULTS for Net::Server
103 my $filename = fileparse($0, '.pl');
104 my $conf_file = (-r "$filename.conf") ? "$filename.conf" : undef;
105 # $conf_file is the Net::Server config for THIS script (not EG), if it exists and is readable
110 pod2usage(1) if $help;
111 unless ($oils_password) {
112 print STDERR "\nERROR: password option required for session login\n\n";
116 print Dumper($o) if $debug;
119 foreach my $ref (qw/bufsize bib_source osrf_config oils_username oils_password help conf_file debug/) {
121 printf "%16s => %s\n", $ref, (eval("\$$ref") || '');
126 print Dumper($real_opts);
130 sub tempdir_setting {
131 my $ret = $apputils->simplereq( qw# opensrf.settings opensrf.settings.xpath.get
132 /opensrf/default/apps/open-ils.vandelay/app_settings/databases/importer # );
133 return $ret->[0] || '/tmp';
139 WARNING: This script provides no security layer. Any client that has
140 access to the server+port can inject MARC records into the system.
146 return $apputils->simplereq(
148 'open-ils.cat.biblio.record.xml.import',
153 sub old_process_batch_data {
154 my $data = shift or $logger->error("process_batch_data called without any data");
158 open $handle, '<', \$data;
159 my $batch = MARC::Batch->new('USMARC', $handle);
167 eval { $rec = $batch->next; };
170 $logger->error("Failed parsing MARC record $index");
173 last unless $rec; # The only way out
175 my $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source);
177 # has the session timed out?
178 if (oils_event_equals($resp, 'NO_SESSION')) {
180 $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source); # try again w/ new token
182 oils_event_die($resp);
187 sub process_spool { # filename
189 my $marcfile = shift;
192 if($import_by_queue) {
194 # don't collect the record IDs, just spool the queue
196 $apputils->simplereq(
198 'open-ils.vandelay.bib.process_spool',
209 # collect the newly queued record IDs for processing
211 my $req = $vl_ses->request(
212 'open-ils.vandelay.bib.process_spool.stream_results',
214 undef, # cache key not needed
221 while(my $resp = $req->recv) {
224 $logger->error("Error spooling MARC data: $resp");
226 } elsif($resp->content) {
227 push(@rec_ids, $resp->content);
235 sub bib_queue_import {
237 my $extra = {auto_overlay_exact => 1};
238 $extra->{merge_profile} = $merge_profile if $merge_profile;
243 if($import_by_queue) {
246 $req = $vl_ses->request(
247 'open-ils.vandelay.bib_queue.import',
254 # import explicit record IDs
256 $req = $vl_ses->request(
257 'open-ils.vandelay.bib_record.list.import',
264 # collect the successfully imported vandelay records
265 while(my $resp = $req->recv) {
267 $logger->error("Error importing MARC data: $resp");
269 } elsif(my $data = $resp->content) {
270 push(@cleanup_recs, $data->{imported}) unless $data->{err_event};
274 # clean up the successfully imported vandelay records to prevent queue bloat
275 my $pcrud = OpenSRF::AppSession->create('open-ils.pcrud');
277 $pcrud->request('open-ils.pcrud.transaction.begin', $authtoken)->recv;
280 foreach (@cleanup_recs) {
284 $pcrud->request('open-ils.pcrud.delete.vqbr', $authtoken, $_)->recv;
288 $logger->error("Error deleteing queued bib record $_: $err");
292 $pcrud->request('open-ils.pcrud.transaction.commit', $authtoken)->recv unless $err;
296 sub process_batch_data {
297 my $data = shift or $logger->error("process_batch_data called without any data");
300 $vl_ses = OpenSRF::AppSession->create('open-ils.vandelay');
302 my ($handle, $tempfile) = File::Temp->tempfile("$0_XXXX", DIR => $tempdir) or die "Cannot write tempfile in $tempdir";
306 $logger->info("Calling process_spool on tempfile $tempfile (queue: $queue_id; source: $bib_source)");
307 my $rec_ids = process_spool($tempfile);
309 if (oils_event_equals($rec_ids, 'NO_SESSION')) { # has the session timed out?
311 $rec_ids = process_spool($tempfile); # try again w/ new token
314 my $resp = bib_queue_import($rec_ids);
316 if (oils_event_equals($resp, 'NO_SESSION')) { # has the session timed out?
318 $resp = bib_queue_import(); # try again w/ new token
320 oils_event_die($resp);
323 sub process_request { # The core Net::Server method
325 my $socket = $self->{server}->{client};
329 # Reading <STDIN> blocks until the client is closed. Instead of waiting
330 # for that, give each inbound record $wait_time seconds to fully arrive
331 # and pull the data directly from the socket
333 local $SIG{ALRM} = sub { die "alarm\n" };
335 $data .= $buf while $socket->sysread($buf, $bufsize);
338 if ($real_opts->{noqueue}) {
339 old_process_batch_data($data);
341 process_batch_data($data);
346 # the authtoken will timeout after the configured inactivity period.
347 # When that happens, get a new one.
349 $authtoken = oils_login($oils_username, $oils_password, 'staff')
350 or die "Unable to login to Evergreen as user $oils_username";
356 osrf_connect($osrf_config);
358 print "Calling Net::Server run ", (@ARGV ? "with command-line options: " . join(' ', @ARGV) : ''), "\n";
359 __PACKAGE__->run(conf_file => $conf_file);
365 marc_stream_importer.pl - Import MARC records via bare socket connection.
369 ./marc_stream_importer.pl [common opts ...] [script opts ...] -- [Net::Server opts ...] &
371 This script uses the EG common options from B<Cronscript>. See --help output for those.
373 Run C<perldoc marc_stream_importer.pl> for full documentation.
375 Note the extra C<--> to separate options for the script wrapper from options for the
376 underlying L<Net::Server> options.
378 Note: this script has to be run in the same directory as B<oils_header.pl>.
380 Typical execution will include a trailing C<&> to run in the background.
384 This script is a L<Net::Server::PreFork> instance for shoving records into Evergreen from a remote system.
388 The only required option is --password
390 --password =<eg_password>
391 --user =<eg_username> default: admin
392 --source =<bib_source> default: 1 Integer
393 --merge-profile =<i> default: 0
394 --tempdir =</temp/dir/> default: from L<opensrf.conf> <open-ils.vandelay/app_settings/databases/importer>
395 --source =<i> default: 1
396 --import-by-queue =<i> default: 0
399 =head2 Old style: --noqueue and associated options
401 To bypass vandelay queue processing and push directly into the database (as the old style)
403 --noqueue default: OFF
404 --buffsize =<i> default: 4096 Buffer size. Only used by --noqueue
405 --wait =<i> default: 5 Seconds to read socket before processing. Only used by --noqueue
407 =head2 Net::Server Options
409 By default, the script will use the Net::Server configuration file B<marc_stream_importer.conf>. You can
410 override this by passing a filepath with the --conf_file option.
412 Other Net::Server options include: --port=<port> --min_servers=<X> --max_servers=<Y> and --log_file=[path/to/file]
414 See L<Net::Server> for a complete list.
418 =head3 OCLC Connexion
420 To use this script with OCLC Connexion, configure the client as follows:
422 Under Tools -> Options -> Export (tab)
423 Create -> Choose Connection -> OK -> Leave translation at "None"
424 -> Create -> Create -> choose TCP/IP (internet)
425 -> Enter hostname and Port, leave 'Use Telnet Protocol' checked
426 -> Create/OK your way out of the dialogs
427 Record Characteristics (button) -> Choose 'UTF-8 Unicode' for the Character Set
430 OCLC and Connexion are trademark/service marks of OCLC Online Computer Library Center, Inc.
434 WARNING: This script provides no inherent security layer. Any client that has
435 access to the server+port can inject MARC records into the system.
436 Use the available options (like allow/deny) in the Net::Server config file
437 or via the command line to restrict access as necessary.
441 ./marc_stream_importer.pl \
442 admin open-ils connexion --port 5555 --min_servers 2 \
443 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
447 L<Net::Server::PreFork>, L<marc_stream_importer.conf>
451 Bill Erickson <erickson@esilibrary.com>
452 Joe Atzberger <jatzberger@esilibrary.com>