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/;
21 use MARC::File::XML ( BinaryEncoding => 'UTF-8' );
22 use MARC::File::USMARC;
25 use File::Basename qw/fileparse/;
27 use Getopt::Long qw(:DEFAULT GetOptionsFromArray);
31 use OpenSRF::Utils::Logger qw/$logger/;
32 use OpenSRF::AppSession;
33 use OpenSRF::EX qw/:try/;
34 use OpenILS::Utils::Cronscript;
35 use OpenSRF::Transport::PeerHandle;
36 require 'oils_header.pl';
37 use vars qw/$apputils/;
45 'merge-profile=i' => 0,
47 # 'osrf-config=s' => '/openils/conf/opensrf_core.xml',
55 'import-by-queue' => 0
58 $OpenILS::Utils::Cronscript::debug=1 if $debug;
59 $Getopt::Long::debug=1 if $debug > 1;
60 my $o = OpenILS::Utils::Cronscript->new(\%defaults);
64 if (grep {$_ eq '--'} @ARGV) {
65 print "Splitting options into groups\n" if $debug;
68 $_ eq '--' and last; # stop at the first --
69 push @script_args, $_;
76 print "Calling MyGetOptions ",
77 (@script_args ? "with options: " . join(' ', @script_args) : 'without options from command line'),
80 my $real_opts = $o->MyGetOptions(\@script_args);
82 # GetOptionsFromArray(\@script_args, \%defaults, %defaults); # similar to
84 $real_opts->{tempdir} ||= tempdir_setting(); # This doesn't go in defaults because it reads config, must come after bootstrap
86 my $bufsize = $real_opts->{buffsize};
87 my $bib_source = $real_opts->{source};
88 my $osrf_config = $real_opts->{'osrf-config'};
89 my $oils_username = $real_opts->{user};
90 my $oils_password = $real_opts->{password};
91 my $help = $real_opts->{help};
92 my $merge_profile = $real_opts->{'merge-profile'};
93 my $queue_id = $real_opts->{queue};
94 my $tempdir = $real_opts->{tempdir};
95 my $import_by_queue = $real_opts->{'import-by-queue'};
96 $debug += $real_opts->{debug};
98 foreach (keys %$real_opts) {
99 print("real_opt->{$_} = ", $real_opts->{$_}, "\n") if $real_opts->{debug} or $debug;
101 my $wait_time = $real_opts->{wait};
104 # DEFAULTS for Net::Server
105 my $filename = fileparse($0, '.pl');
106 my $conf_file = (-r "$filename.conf") ? "$filename.conf" : undef;
107 # $conf_file is the Net::Server config for THIS script (not EG), if it exists and is readable
112 pod2usage(1) if $help;
113 unless ($oils_password) {
114 print STDERR "\nERROR: password option required for session login\n\n";
118 print Dumper($o) if $debug;
121 foreach my $ref (qw/bufsize bib_source osrf_config oils_username oils_password help conf_file debug/) {
123 printf "%16s => %s\n", $ref, (eval("\$$ref") || '');
128 print Dumper($real_opts);
132 sub tempdir_setting {
133 my $ret = $apputils->simplereq( qw# opensrf.settings opensrf.settings.xpath.get
134 /opensrf/default/apps/open-ils.vandelay/app_settings/databases/importer # );
135 return $ret->[0] || '/tmp';
141 WARNING: This script provides no security layer. Any client that has
142 access to the server+port can inject MARC records into the system.
148 return $apputils->simplereq(
150 'open-ils.cat.biblio.record.xml.import',
155 sub old_process_batch_data {
156 my $data = shift or $logger->error("process_batch_data called without any data");
160 open $handle, '<', \$data;
161 my $batch = MARC::Batch->new('USMARC', $handle);
172 eval { $rec = $batch->next; };
175 $logger->error("Failed parsing MARC record $index");
179 last unless $rec; # The only way out
181 my $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source);
183 # has the session timed out?
184 if (oils_event_equals($resp, 'NO_SESSION')) {
186 $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source); # try again w/ new token
188 oils_event_die($resp);
192 return ($imported, $failed);
195 sub process_spool { # filename
197 my $marcfile = shift;
200 if($import_by_queue) {
202 # don't collect the record IDs, just spool the queue
204 $apputils->simplereq(
206 'open-ils.vandelay.bib.process_spool',
217 # collect the newly queued record IDs for processing
219 my $req = $vl_ses->request(
220 'open-ils.vandelay.bib.process_spool.stream_results',
222 undef, # cache key not needed
229 while(my $resp = $req->recv) {
232 $logger->error("Error spooling MARC data: $resp");
234 } elsif($resp->content) {
235 push(@rec_ids, $resp->content);
243 sub bib_queue_import {
245 my $extra = {auto_overlay_exact => 1};
246 $extra->{merge_profile} = $merge_profile if $merge_profile;
251 if($import_by_queue) {
254 $req = $vl_ses->request(
255 'open-ils.vandelay.bib_queue.import',
262 # import explicit record IDs
264 $req = $vl_ses->request(
265 'open-ils.vandelay.bib_record.list.import',
272 # collect the successfully imported vandelay records
274 while(my $resp = $req->recv) {
276 $logger->error("Error importing MARC data: $resp");
278 } elsif(my $data = $resp->content) {
280 if($data->{err_event}) {
282 $logger->error(Dumper($data->{err_event}));
286 push(@cleanup_recs, $data->{imported}) if $data->{imported};
291 # clean up the successfully imported vandelay records to prevent queue bloat
292 my $pcrud = OpenSRF::AppSession->create('open-ils.pcrud');
294 $pcrud->request('open-ils.pcrud.transaction.begin', $authtoken)->recv;
297 foreach (@cleanup_recs) {
301 $pcrud->request('open-ils.pcrud.delete.vqbr', $authtoken, $_)->recv;
305 $logger->error("Error deleteing queued bib record $_: $err");
309 $pcrud->request('open-ils.pcrud.transaction.commit', $authtoken)->recv unless $err;
312 $logger->info("imported queued vandelay records: @cleanup_recs");
313 return (scalar(@cleanup_recs), $failed);
316 sub process_batch_data {
317 my $data = shift or $logger->error("process_batch_data called without any data");
320 $vl_ses = OpenSRF::AppSession->create('open-ils.vandelay');
322 my ($handle, $tempfile) = File::Temp->tempfile("$0_XXXX", DIR => $tempdir) or die "Cannot write tempfile in $tempdir";
326 $logger->info("Calling process_spool on tempfile $tempfile (queue: $queue_id; source: $bib_source)");
327 my $rec_ids = process_spool($tempfile);
329 if (oils_event_equals($rec_ids, 'NO_SESSION')) { # has the session timed out?
331 $rec_ids = process_spool($tempfile); # try again w/ new token
334 my ($imported, $failed) = bib_queue_import($rec_ids);
336 if (oils_event_equals($imported, 'NO_SESSION')) { # has the session timed out?
338 ($imported, $failed) = bib_queue_import(); # try again w/ new token
341 oils_event_die($imported);
343 return ($imported, $failed);
346 sub process_request { # The core Net::Server method
348 my $client = $self->{server}->{client};
350 my $sockname = getpeername($client);
351 my ($port, $ip_addr) = unpack_sockaddr_in($sockname);
352 $logger->info("stream parser received contact from ".inet_ntoa($ip_addr));
354 my $ph = OpenSRF::Transport::PeerHandle->retrieve;
355 if(!$ph->flush_socket()) {
356 $logger->error("We received a request, bu we are no longer connected to opensrf. ".
357 "Exiting and dropping request from $client");
363 local $SIG{ALRM} = sub { die "alarm\n" };
364 alarm $wait_time; # prevent accidental tie ups of backend processes
365 local $/ = "\x1D"; # MARC record separator
371 $logger->error("reading from STDIN failed or timed out: $@");
375 $logger->info("stream parser read " . length($data) . " bytes");
377 my ($imported, $failed) = (0, 0);
379 new_auth_token(); # login
381 if ($real_opts->{noqueue}) {
382 ($imported, $failed) = old_process_batch_data($data);
384 ($imported, $failed) = process_batch_data($data);
387 my $profile = (!$merge_profile) ? '' :
388 $apputils->simplereq(
390 'open-ils.pcrud.retrieve.vmp',
392 $merge_profile)->name;
395 $msg .= "Successfully imported $imported records using merge profile '$profile'\n" if $imported;
396 $msg .= "Failed to import $failed records\n" if $failed;
400 clear_auth_token(); # logout
404 # the authtoken will timeout after the configured inactivity period.
405 # When that happens, get a new one.
407 $authtoken = oils_login($oils_username, $oils_password, 'staff')
408 or die "Unable to login to Evergreen as user $oils_username";
412 sub clear_auth_token {
413 $apputils->simplereq(
415 'open-ils.auth.session.delete',
422 osrf_connect($osrf_config);
423 print "Calling Net::Server run ", (@ARGV ? "with command-line options: " . join(' ', @ARGV) : ''), "\n";
424 __PACKAGE__->run(conf_file => $conf_file);
430 marc_stream_importer.pl - Import MARC records via bare socket connection.
434 ./marc_stream_importer.pl [common opts ...] [script opts ...] -- [Net::Server opts ...] &
436 This script uses the EG common options from B<Cronscript>. See --help output for those.
438 Run C<perldoc marc_stream_importer.pl> for full documentation.
440 Note the extra C<--> to separate options for the script wrapper from options for the
441 underlying L<Net::Server> options.
443 Note: this script has to be run in the same directory as B<oils_header.pl>.
445 Typical execution will include a trailing C<&> to run in the background.
449 This script is a L<Net::Server::PreFork> instance for shoving records into Evergreen from a remote system.
453 The only required option is --password
455 --password =<eg_password>
456 --user =<eg_username> default: admin
457 --source =<bib_source> default: 1 Integer
458 --merge-profile =<i> default: 0
459 --tempdir =</temp/dir/> default: from L<opensrf.conf> <open-ils.vandelay/app_settings/databases/importer>
460 --source =<i> default: 1
461 --import-by-queue =<i> default: 0
464 =head2 Old style: --noqueue and associated options
466 To bypass vandelay queue processing and push directly into the database (as the old style)
468 --noqueue default: OFF
469 --buffsize =<i> default: 4096 Buffer size. Only used by --noqueue
470 --wait =<i> default: 5 Seconds to read socket before processing. Only used by --noqueue
472 =head2 Net::Server Options
474 By default, the script will use the Net::Server configuration file B<marc_stream_importer.conf>. You can
475 override this by passing a filepath with the --conf_file option.
477 Other Net::Server options include: --port=<port> --min_servers=<X> --max_servers=<Y> and --log_file=[path/to/file]
479 See L<Net::Server> for a complete list.
483 =head3 OCLC Connexion
485 To use this script with OCLC Connexion, configure the client as follows:
487 Under Tools -> Options -> Export (tab)
488 Create -> Choose Connection -> OK -> Leave translation at "None"
489 -> Create -> Create -> choose TCP/IP (internet)
490 -> Enter hostname and Port, leave 'Use Telnet Protocol' checked
491 -> Create/OK your way out of the dialogs
492 Record Characteristics (button) -> Choose 'UTF-8 Unicode' for the Character Set
495 OCLC and Connexion are trademark/service marks of OCLC Online Computer Library Center, Inc.
499 WARNING: This script provides no inherent security layer. Any client that has
500 access to the server+port can inject MARC records into the system.
501 Use the available options (like allow/deny) in the Net::Server config file
502 or via the command line to restrict access as necessary.
506 ./marc_stream_importer.pl \
507 admin open-ils connexion --port 5555 --min_servers 2 \
508 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
512 L<Net::Server::PreFork>, L<marc_stream_importer.conf>
516 Bill Erickson <erickson@esilibrary.com>
517 Joe Atzberger <jatzberger@esilibrary.com>