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/$logger/;
31 use OpenSRF::AppSession;
32 use OpenSRF::EX qw/:try/;
33 use OpenILS::Utils::Cronscript;
34 use OpenSRF::Transport::PeerHandle;
35 require 'oils_header.pl';
36 use vars qw/$apputils/;
44 'merge-profile=i' => 0,
46 # 'osrf-config=s' => '/openils/conf/opensrf_core.xml',
54 'import-by-queue' => 0
57 $OpenILS::Utils::Cronscript::debug=1 if $debug;
58 $Getopt::Long::debug=1 if $debug > 1;
59 my $o = OpenILS::Utils::Cronscript->new(\%defaults);
63 if (grep {$_ eq '--'} @ARGV) {
64 print "Splitting options into groups\n" if $debug;
67 $_ eq '--' and last; # stop at the first --
68 push @script_args, $_;
75 print "Calling MyGetOptions ",
76 (@script_args ? "with options: " . join(' ', @script_args) : 'without options from command line'),
79 my $real_opts = $o->MyGetOptions(\@script_args);
81 # GetOptionsFromArray(\@script_args, \%defaults, %defaults); # similar to
83 $real_opts->{tempdir} ||= tempdir_setting(); # This doesn't go in defaults because it reads config, must come after bootstrap
85 my $bufsize = $real_opts->{buffsize};
86 my $bib_source = $real_opts->{source};
87 my $osrf_config = $real_opts->{'osrf-config'};
88 my $oils_username = $real_opts->{user};
89 my $oils_password = $real_opts->{password};
90 my $help = $real_opts->{help};
91 my $merge_profile = $real_opts->{'merge-profile'};
92 my $queue_id = $real_opts->{queue};
93 my $tempdir = $real_opts->{tempdir};
94 my $import_by_queue = $real_opts->{'import-by-queue'};
95 $debug += $real_opts->{debug};
97 foreach (keys %$real_opts) {
98 print("real_opt->{$_} = ", $real_opts->{$_}, "\n") if $real_opts->{debug} or $debug;
100 my $wait_time = $real_opts->{wait};
103 # DEFAULTS for Net::Server
104 my $filename = fileparse($0, '.pl');
105 my $conf_file = (-r "$filename.conf") ? "$filename.conf" : undef;
106 # $conf_file is the Net::Server config for THIS script (not EG), if it exists and is readable
111 pod2usage(1) if $help;
112 unless ($oils_password) {
113 print STDERR "\nERROR: password option required for session login\n\n";
117 print Dumper($o) if $debug;
120 foreach my $ref (qw/bufsize bib_source osrf_config oils_username oils_password help conf_file debug/) {
122 printf "%16s => %s\n", $ref, (eval("\$$ref") || '');
127 print Dumper($real_opts);
131 sub tempdir_setting {
132 my $ret = $apputils->simplereq( qw# opensrf.settings opensrf.settings.xpath.get
133 /opensrf/default/apps/open-ils.vandelay/app_settings/databases/importer # );
134 return $ret->[0] || '/tmp';
140 WARNING: This script provides no security layer. Any client that has
141 access to the server+port can inject MARC records into the system.
147 return $apputils->simplereq(
149 'open-ils.cat.biblio.record.xml.import',
154 sub old_process_batch_data {
155 my $data = shift or $logger->error("process_batch_data called without any data");
159 open $handle, '<', \$data;
160 my $batch = MARC::Batch->new('USMARC', $handle);
171 eval { $rec = $batch->next; };
174 $logger->error("Failed parsing MARC record $index");
178 last unless $rec; # The only way out
180 my $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source);
182 # has the session timed out?
183 if (oils_event_equals($resp, 'NO_SESSION')) {
185 $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source); # try again w/ new token
187 oils_event_die($resp);
191 return ($imported, $failed);
194 sub process_spool { # filename
196 my $marcfile = shift;
199 if($import_by_queue) {
201 # don't collect the record IDs, just spool the queue
203 $apputils->simplereq(
205 'open-ils.vandelay.bib.process_spool',
216 # collect the newly queued record IDs for processing
218 my $req = $vl_ses->request(
219 'open-ils.vandelay.bib.process_spool.stream_results',
221 undef, # cache key not needed
228 while(my $resp = $req->recv) {
231 $logger->error("Error spooling MARC data: $resp");
233 } elsif($resp->content) {
234 push(@rec_ids, $resp->content);
242 sub bib_queue_import {
244 my $extra = {auto_overlay_exact => 1};
245 $extra->{merge_profile} = $merge_profile if $merge_profile;
250 if($import_by_queue) {
253 $req = $vl_ses->request(
254 'open-ils.vandelay.bib_queue.import',
261 # import explicit record IDs
263 $req = $vl_ses->request(
264 'open-ils.vandelay.bib_record.list.import',
271 # collect the successfully imported vandelay records
273 while(my $resp = $req->recv) {
275 $logger->error("Error importing MARC data: $resp");
277 } elsif(my $data = $resp->content) {
279 if($data->{err_event}) {
281 $logger->error(Dumper($data->{err_event}));
285 push(@cleanup_recs, $data->{imported}) if $data->{imported};
290 # clean up the successfully imported vandelay records to prevent queue bloat
291 my $pcrud = OpenSRF::AppSession->create('open-ils.pcrud');
293 $pcrud->request('open-ils.pcrud.transaction.begin', $authtoken)->recv;
296 foreach (@cleanup_recs) {
300 $pcrud->request('open-ils.pcrud.delete.vqbr', $authtoken, $_)->recv;
304 $logger->error("Error deleteing queued bib record $_: $err");
308 $pcrud->request('open-ils.pcrud.transaction.commit', $authtoken)->recv unless $err;
311 $logger->info("imported queued vandelay records: @cleanup_recs");
312 return (scalar(@cleanup_recs), $failed);
315 sub process_batch_data {
316 my $data = shift or $logger->error("process_batch_data called without any data");
319 $vl_ses = OpenSRF::AppSession->create('open-ils.vandelay');
321 my ($handle, $tempfile) = File::Temp->tempfile("$0_XXXX", DIR => $tempdir) or die "Cannot write tempfile in $tempdir";
325 $logger->info("Calling process_spool on tempfile $tempfile (queue: $queue_id; source: $bib_source)");
326 my $rec_ids = process_spool($tempfile);
328 if (oils_event_equals($rec_ids, 'NO_SESSION')) { # has the session timed out?
330 $rec_ids = process_spool($tempfile); # try again w/ new token
333 my ($imported, $failed) = bib_queue_import($rec_ids);
335 if (oils_event_equals($imported, 'NO_SESSION')) { # has the session timed out?
337 ($imported, $failed) = bib_queue_import(); # try again w/ new token
340 oils_event_die($imported);
342 return ($imported, $failed);
345 sub process_request { # The core Net::Server method
347 my $client = $self->{server}->{client};
349 $logger->info("stream parser received contact from $client");
351 my $ph = OpenSRF::Transport::PeerHandle->retrieve;
352 if(!$ph->flush_socket()) {
353 $logger->error("We received a request, bu we are no longer connected to opensrf. ".
354 "Exiting and dropping request from $client");
360 local $SIG{ALRM} = sub { die "alarm\n" };
361 alarm $wait_time; # prevent accidental tie ups of backend processes
362 local $/ = "\x1D"; # MARC record separator
368 $logger->error("reading from STDIN failed or timed out: $@");
372 $logger->info("stream parser read " . length($data) . " bytes");
374 my ($imported, $failed) = (0, 0);
376 new_auth_token(); # login
378 if ($real_opts->{noqueue}) {
379 ($imported, $failed) = old_process_batch_data($data);
381 ($imported, $failed) = process_batch_data($data);
384 my $profile = (!$merge_profile) ? '' :
385 $apputils->simplereq(
387 'open-ils.pcrud.retrieve.vmp',
389 $merge_profile)->name;
392 $msg .= "Successfully imported $imported records using merge profile '$profile'\n" if $imported;
393 $msg .= "Failed to import $failed records\n" if $failed;
397 clear_auth_token(); # logout
401 # the authtoken will timeout after the configured inactivity period.
402 # When that happens, get a new one.
404 $authtoken = oils_login($oils_username, $oils_password, 'staff')
405 or die "Unable to login to Evergreen as user $oils_username";
409 sub clear_auth_token {
410 $apputils->simplereq(
412 'open-ils.auth.session.delete',
419 osrf_connect($osrf_config);
420 print "Calling Net::Server run ", (@ARGV ? "with command-line options: " . join(' ', @ARGV) : ''), "\n";
421 __PACKAGE__->run(conf_file => $conf_file);
427 marc_stream_importer.pl - Import MARC records via bare socket connection.
431 ./marc_stream_importer.pl [common opts ...] [script opts ...] -- [Net::Server opts ...] &
433 This script uses the EG common options from B<Cronscript>. See --help output for those.
435 Run C<perldoc marc_stream_importer.pl> for full documentation.
437 Note the extra C<--> to separate options for the script wrapper from options for the
438 underlying L<Net::Server> options.
440 Note: this script has to be run in the same directory as B<oils_header.pl>.
442 Typical execution will include a trailing C<&> to run in the background.
446 This script is a L<Net::Server::PreFork> instance for shoving records into Evergreen from a remote system.
450 The only required option is --password
452 --password =<eg_password>
453 --user =<eg_username> default: admin
454 --source =<bib_source> default: 1 Integer
455 --merge-profile =<i> default: 0
456 --tempdir =</temp/dir/> default: from L<opensrf.conf> <open-ils.vandelay/app_settings/databases/importer>
457 --source =<i> default: 1
458 --import-by-queue =<i> default: 0
461 =head2 Old style: --noqueue and associated options
463 To bypass vandelay queue processing and push directly into the database (as the old style)
465 --noqueue default: OFF
466 --buffsize =<i> default: 4096 Buffer size. Only used by --noqueue
467 --wait =<i> default: 5 Seconds to read socket before processing. Only used by --noqueue
469 =head2 Net::Server Options
471 By default, the script will use the Net::Server configuration file B<marc_stream_importer.conf>. You can
472 override this by passing a filepath with the --conf_file option.
474 Other Net::Server options include: --port=<port> --min_servers=<X> --max_servers=<Y> and --log_file=[path/to/file]
476 See L<Net::Server> for a complete list.
480 =head3 OCLC Connexion
482 To use this script with OCLC Connexion, configure the client as follows:
484 Under Tools -> Options -> Export (tab)
485 Create -> Choose Connection -> OK -> Leave translation at "None"
486 -> Create -> Create -> choose TCP/IP (internet)
487 -> Enter hostname and Port, leave 'Use Telnet Protocol' checked
488 -> Create/OK your way out of the dialogs
489 Record Characteristics (button) -> Choose 'UTF-8 Unicode' for the Character Set
492 OCLC and Connexion are trademark/service marks of OCLC Online Computer Library Center, Inc.
496 WARNING: This script provides no inherent security layer. Any client that has
497 access to the server+port can inject MARC records into the system.
498 Use the available options (like allow/deny) in the Net::Server config file
499 or via the command line to restrict access as necessary.
503 ./marc_stream_importer.pl \
504 admin open-ils connexion --port 5555 --min_servers 2 \
505 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
509 L<Net::Server::PreFork>, L<marc_stream_importer.conf>
513 Bill Erickson <erickson@esilibrary.com>
514 Joe Atzberger <jatzberger@esilibrary.com>