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/$logger/;
31 use OpenSRF::AppSession;
32 use OpenILS::Utils::Cronscript;
33 require 'oils_header.pl';
34 use vars qw/$apputils/;
42 'merge-profile=i' => 0,
44 # 'osrf-config=s' => '/openils/conf/opensrf_core.xml',
52 'import-by-queue' => 0
55 $OpenILS::Utils::Cronscript::debug=1 if $debug;
56 $Getopt::Long::debug=1 if $debug > 1;
57 my $o = OpenILS::Utils::Cronscript->new(\%defaults);
61 if (grep {$_ eq '--'} @ARGV) {
62 print "Splitting options into groups\n" if $debug;
65 $_ eq '--' and last; # stop at the first --
66 push @script_args, $_;
73 print "Calling MyGetOptions ",
74 (@script_args ? "with options: " . join(' ', @script_args) : 'without options from command line'),
77 my $real_opts = $o->MyGetOptions(\@script_args);
79 # GetOptionsFromArray(\@script_args, \%defaults, %defaults); # similar to
81 $real_opts->{tempdir} ||= tempdir_setting(); # This doesn't go in defaults because it reads config, must come after bootstrap
83 my $bufsize = $real_opts->{buffsize};
84 my $bib_source = $real_opts->{source};
85 my $osrf_config = $real_opts->{'osrf-config'};
86 my $oils_username = $real_opts->{user};
87 my $oils_password = $real_opts->{password};
88 my $help = $real_opts->{help};
89 my $merge_profile = $real_opts->{'merge-profile'};
90 my $queue_id = $real_opts->{queue};
91 my $tempdir = $real_opts->{tempdir};
92 my $import_by_queue = $real_opts->{'import-by-queue'};
93 $debug += $real_opts->{debug};
95 foreach (keys %$real_opts) {
96 print("real_opt->{$_} = ", $real_opts->{$_}, "\n") if $real_opts->{debug} or $debug;
98 my $wait_time = $real_opts->{wait};
101 # DEFAULTS for Net::Server
102 my $filename = fileparse($0, '.pl');
103 my $conf_file = (-r "$filename.conf") ? "$filename.conf" : undef;
104 # $conf_file is the Net::Server config for THIS script (not EG), if it exists and is readable
109 pod2usage(1) if $help;
110 unless ($oils_password) {
111 print STDERR "\nERROR: password option required for session login\n\n";
115 print Dumper($o) if $debug;
118 foreach my $ref (qw/bufsize bib_source osrf_config oils_username oils_password help conf_file debug/) {
120 printf "%16s => %s\n", $ref, (eval("\$$ref") || '');
125 print Dumper($real_opts);
129 sub tempdir_setting {
130 my $ret = $apputils->simplereq( qw# opensrf.settings opensrf.settings.xpath.get
131 /opensrf/default/apps/open-ils.vandelay/app_settings/databases/importer # );
132 return $ret->[0] || '/tmp';
138 WARNING: This script provides no security layer. Any client that has
139 access to the server+port can inject MARC records into the system.
145 return $apputils->simplereq(
147 'open-ils.cat.biblio.record.xml.import',
152 sub old_process_batch_data {
153 my $data = shift or $logger->error("process_batch_data called without any data");
157 open $handle, '<', \$data;
158 my $batch = MARC::Batch->new('USMARC', $handle);
166 eval { $rec = $batch->next; };
169 $logger->error("Failed parsing MARC record $index");
172 last unless $rec; # The only way out
174 my $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source);
176 # has the session timed out?
177 if (oils_event_equals($resp, 'NO_SESSION')) {
179 $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source); # try again w/ new token
181 oils_event_die($resp);
186 sub process_spool { # filename
188 my $marcfile = shift;
191 if($import_by_queue) {
193 # don't collect the record IDs, just spool the queue
195 $apputils->simplereq(
197 'open-ils.vandelay.bib.process_spool',
208 # collect the newly queued record IDs for processing
210 my $req = $vl_ses->request(
211 'open-ils.vandelay.bib.process_spool.stream_results',
213 undef, # cache key not needed
220 while(my $resp = $req->recv) {
223 $logger->error("Error spooling MARC data: $resp");
225 } elsif($resp->content) {
226 push(@rec_ids, $resp->content);
234 sub bib_queue_import {
236 my $extra = {auto_overlay_exact => 1};
237 $extra->{merge_profile} = $merge_profile if $merge_profile;
239 if($import_by_queue) {
241 $apputils->simplereq(
243 'open-ils.vandelay.bib_queue.import',
251 # import explicit record IDs
253 $apputils->simplereq(
255 'open-ils.vandelay.bib_record.list.import',
263 sub process_batch_data {
264 my $data = shift or $logger->error("process_batch_data called without any data");
267 $vl_ses = OpenSRF::AppSession->create('open-ils.vandelay');
269 my ($handle, $tempfile) = File::Temp->tempfile("$0_XXXX", DIR => $tempdir) or die "Cannot write tempfile in $tempdir";
273 $logger->info("Calling process_spool on tempfile $tempfile (queue: $queue_id; source: $bib_source)");
274 my $rec_ids = process_spool($tempfile);
276 if (oils_event_equals($rec_ids, 'NO_SESSION')) { # has the session timed out?
278 $rec_ids = process_spool($tempfile); # try again w/ new token
281 my $resp = bib_queue_import($rec_ids);
283 if (oils_event_equals($resp, 'NO_SESSION')) { # has the session timed out?
285 $resp = bib_queue_import(); # try again w/ new token
287 oils_event_die($resp);
290 sub process_request { # The core Net::Server method
292 my $socket = $self->{server}->{client};
296 # Reading <STDIN> blocks until the client is closed. Instead of waiting
297 # for that, give each inbound record $wait_time seconds to fully arrive
298 # and pull the data directly from the socket
300 local $SIG{ALRM} = sub { die "alarm\n" };
302 $data .= $buf while $socket->sysread($buf, $bufsize);
305 if ($real_opts->{noqueue}) {
306 old_process_batch_data($data);
308 process_batch_data($data);
313 # the authtoken will timeout after the configured inactivity period.
314 # When that happens, get a new one.
316 $authtoken = oils_login($oils_username, $oils_password, 'staff')
317 or die "Unable to login to Evergreen as user $oils_username";
323 osrf_connect($osrf_config);
325 print "Calling Net::Server run ", (@ARGV ? "with command-line options: " . join(' ', @ARGV) : ''), "\n";
326 __PACKAGE__->run(conf_file => $conf_file);
332 marc_stream_importer.pl - Import MARC records via bare socket connection.
336 ./marc_stream_importer.pl [common opts ...] [script opts ...] -- [Net::Server opts ...] &
338 This script uses the EG common options from B<Cronscript>. See --help output for those.
340 Run C<perldoc marc_stream_importer.pl> for full documentation.
342 Note the extra C<--> to separate options for the script wrapper from options for the
343 underlying L<Net::Server> options.
345 Note: this script has to be run in the same directory as B<oils_header.pl>.
347 Typical execution will include a trailing C<&> to run in the background.
351 This script is a L<Net::Server::PreFork> instance for shoving records into Evergreen from a remote system.
355 The only required option is --password
357 --password =<eg_password>
358 --user =<eg_username> default: admin
359 --source =<bib_source> default: 1 Integer
360 --merge-profile =<i> default: 0
361 --tempdir =</temp/dir/> default: from L<opensrf.conf> <open-ils.vandelay/app_settings/databases/importer>
362 --source =<i> default: 1
363 --import-by-queue =<i> default: 0
366 =head2 Old style: --noqueue and associated options
368 To bypass vandelay queue processing and push directly into the database (as the old style)
370 --noqueue default: OFF
371 --buffsize =<i> default: 4096 Buffer size. Only used by --noqueue
372 --wait =<i> default: 5 Seconds to read socket before processing. Only used by --noqueue
374 =head2 Net::Server Options
376 By default, the script will use the Net::Server configuration file B<marc_stream_importer.conf>. You can
377 override this by passing a filepath with the --conf_file option.
379 Other Net::Server options include: --port=<port> --min_servers=<X> --max_servers=<Y> and --log_file=[path/to/file]
381 See L<Net::Server> for a complete list.
385 =head3 OCLC Connexion
387 To use this script with OCLC Connexion, configure the client as follows:
389 Under Tools -> Options -> Export (tab)
390 Create -> Choose Connection -> OK -> Leave translation at "None"
391 -> Create -> Create -> choose TCP/IP (internet)
392 -> Enter hostname and Port, leave 'Use Telnet Protocol' checked
393 -> Create/OK your way out of the dialogs
394 Record Characteristics (button) -> Choose 'UTF-8 Unicode' for the Character Set
397 OCLC and Connexion are trademark/service marks of OCLC Online Computer Library Center, Inc.
401 WARNING: This script provides no inherent security layer. Any client that has
402 access to the server+port can inject MARC records into the system.
403 Use the available options (like allow/deny) in the Net::Server config file
404 or via the command line to restrict access as necessary.
408 ./marc_stream_importer.pl \
409 admin open-ils connexion --port 5555 --min_servers 2 \
410 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
414 L<Net::Server::PreFork>, L<marc_stream_importer.conf>
418 Bill Erickson <erickson@esilibrary.com>
419 Joe Atzberger <jatzberger@esilibrary.com>