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',
57 'import-by-queue' => 0,
58 'auto-overlay-exact' => 0,
59 'auto-overlay-1match' => 0,
60 'auto-overlay-best-match' => 0
63 $OpenILS::Utils::Cronscript::debug=1 if $debug;
64 $Getopt::Long::debug=1 if $debug > 1;
65 my $o = OpenILS::Utils::Cronscript->new(\%defaults);
69 if (grep {$_ eq '--'} @ARGV) {
70 print "Splitting options into groups\n" if $debug;
73 $_ eq '--' and last; # stop at the first --
74 push @script_args, $_;
81 print "Calling MyGetOptions ",
82 (@script_args ? "with options: " . join(' ', @script_args) : 'without options from command line'),
85 my $real_opts = $o->MyGetOptions(\@script_args);
87 # GetOptionsFromArray(\@script_args, \%defaults, %defaults); # similar to
89 $real_opts->{tempdir} ||= tempdir_setting(); # This doesn't go in defaults because it reads config, must come after bootstrap
91 my $bufsize = $real_opts->{buffsize};
92 my $bib_source = $real_opts->{source};
93 my $osrf_config = $real_opts->{'osrf-config'};
94 my $oils_username = $real_opts->{user};
95 my $oils_password = $real_opts->{password};
96 my $help = $real_opts->{help};
97 my $merge_profile = $real_opts->{'merge-profile'};
98 my $queue_id = $real_opts->{queue};
99 my $tempdir = $real_opts->{tempdir};
100 my $import_by_queue = $real_opts->{'import-by-queue'};
101 $debug += $real_opts->{debug};
103 foreach (keys %$real_opts) {
104 print("real_opt->{$_} = ", $real_opts->{$_}, "\n") if $real_opts->{debug} or $debug;
106 my $wait_time = $real_opts->{wait};
109 # DEFAULTS for Net::Server
110 my $filename = fileparse($0, '.pl');
111 my $conf_file = (-r "$filename.conf") ? "$filename.conf" : undef;
112 # $conf_file is the Net::Server config for THIS script (not EG), if it exists and is readable
117 pod2usage(1) if $help;
118 unless ($oils_password) {
119 print STDERR "\nERROR: password option required for session login\n\n";
123 print Dumper($o) if $debug;
126 foreach my $ref (qw/bufsize bib_source osrf_config oils_username oils_password help conf_file debug/) {
128 printf "%16s => %s\n", $ref, (eval("\$$ref") || '');
133 print Dumper($real_opts);
137 sub tempdir_setting {
138 my $ret = $apputils->simplereq( qw# opensrf.settings opensrf.settings.xpath.get
139 /opensrf/default/apps/open-ils.vandelay/app_settings/databases/importer # );
140 return $ret->[0] || '/tmp';
146 WARNING: This script provides no security layer. Any client that has
147 access to the server+port can inject MARC records into the system.
153 return $apputils->simplereq(
155 'open-ils.cat.biblio.record.xml.import',
160 sub old_process_batch_data {
161 my $data = shift or $logger->error("process_batch_data called without any data");
169 open $handle, '<', \$data;
172 my $batch = MARC::Batch->new('USMARC', $handle);
183 eval { $rec = $batch->next; };
186 $logger->error("Failed parsing MARC record $index");
190 last unless $rec; # The only way out
192 my $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source);
194 # has the session timed out?
195 if (oils_event_equals($resp, 'NO_SESSION')) {
197 $resp = xml_import($authtoken, $rec->as_xml_record, $bib_source); # try again w/ new token
199 oils_event_die($resp);
203 return ($imported, $failed);
206 sub process_spool { # filename
208 my $marcfile = shift;
211 if($import_by_queue) {
213 # don't collect the record IDs, just spool the queue
215 $apputils->simplereq(
217 'open-ils.vandelay.bib.process_spool',
228 # collect the newly queued record IDs for processing
230 my $req = $vl_ses->request(
231 'open-ils.vandelay.bib.process_spool.stream_results',
233 undef, # cache key not needed
240 while(my $resp = $req->recv) {
243 $logger->error("Error spooling MARC data: $resp");
245 } elsif($resp->content) {
246 push(@rec_ids, $resp->content);
254 sub bib_queue_import {
256 my $extra = {import_no_match => 1};
257 $extra->{merge_profile} = $merge_profile if $merge_profile;
258 $extra->{auto_overlay_1match} = 1 if $real_opts->{'auto-overlay-1match'};
259 $extra->{auto_overlay_best_match} = 1 if $real_opts->{'auto-overlay-best-match'};
261 # default to exact match if no strategy is chosen
262 $extra->{auto_overlay_exact} = 1
263 if $real_opts->{'auto-overlay-exact'} or
264 not ($extra->{auto_overlay_1match} or $extra->{auto_overlay_best_match});
269 if($import_by_queue) {
272 $req = $vl_ses->request(
273 'open-ils.vandelay.bib_queue.import',
280 # import explicit record IDs
282 $req = $vl_ses->request(
283 'open-ils.vandelay.bib_record.list.import',
290 # collect the successfully imported vandelay records
292 while(my $resp = $req->recv) {
294 $logger->error("Error importing MARC data: $resp");
296 } elsif(my $data = $resp->content) {
298 if($data->{err_event}) {
300 $logger->error(Dumper($data->{err_event}));
304 push(@cleanup_recs, $data->{imported}) if $data->{imported};
309 # clean up the successfully imported vandelay records to prevent queue bloat
310 my $pcrud = OpenSRF::AppSession->create('open-ils.pcrud');
312 $pcrud->request('open-ils.pcrud.transaction.begin', $authtoken)->recv;
315 foreach (@cleanup_recs) {
319 $pcrud->request('open-ils.pcrud.delete.vqbr', $authtoken, $_)->recv;
323 $logger->error("Error deleteing queued bib record $_: $err");
327 $pcrud->request('open-ils.pcrud.transaction.commit', $authtoken)->recv unless $err;
330 $logger->info("imported queued vandelay records: @cleanup_recs");
331 return (scalar(@cleanup_recs), $failed);
334 sub process_batch_data {
335 my $data = shift or $logger->error("process_batch_data called without any data");
339 $vl_ses = OpenSRF::AppSession->create('open-ils.vandelay');
341 my ($handle, $tempfile);
343 ($handle, $tempfile) = File::Temp->tempfile("$0_XXXX", DIR => $tempdir) or die "Cannot write tempfile in $tempdir";
350 $logger->info("Calling process_spool on tempfile $tempfile (queue: $queue_id; source: $bib_source)");
351 my $rec_ids = process_spool($tempfile);
353 if (oils_event_equals($rec_ids, 'NO_SESSION')) { # has the session timed out?
355 $rec_ids = process_spool($tempfile); # try again w/ new token
358 my ($imported, $failed) = bib_queue_import($rec_ids);
360 if (oils_event_equals($imported, 'NO_SESSION')) { # has the session timed out?
362 ($imported, $failed) = bib_queue_import(); # try again w/ new token
365 oils_event_die($imported);
367 return ($imported, $failed);
370 sub process_request { # The core Net::Server method
372 my $client = $self->{server}->{client};
374 my $sockname = getpeername($client);
375 my ($port, $ip_addr) = unpack_sockaddr_in($sockname);
376 $logger->info("stream parser received contact from ".inet_ntoa($ip_addr));
378 my $ph = OpenSRF::Transport::PeerHandle->retrieve;
379 if(!$ph->flush_socket()) {
380 $logger->error("We received a request, bu we are no longer connected to opensrf. ".
381 "Exiting and dropping request from $client");
387 local $SIG{ALRM} = sub { die "alarm\n" };
388 alarm $wait_time; # prevent accidental tie ups of backend processes
389 local $/ = "\x1D"; # MARC record separator
395 $logger->error("reading from STDIN failed or timed out: $@");
399 $logger->info("stream parser read " . length($data) . " bytes");
401 my ($imported, $failed) = (0, 0);
403 new_auth_token(); # login
405 if ($real_opts->{noqueue}) {
406 ($imported, $failed) = old_process_batch_data($data);
408 ($imported, $failed) = process_batch_data($data);
411 my $profile = (!$merge_profile) ? '' :
412 $apputils->simplereq(
414 'open-ils.pcrud.retrieve.vmp',
416 $merge_profile)->name;
419 $msg .= "Successfully imported $imported records using merge profile '$profile'\n" if $imported;
420 $msg .= "Failed to import $failed records\n" if $failed;
424 clear_auth_token(); # logout
427 sub standalone_process_request { # The command line version
430 $logger->info("stream parser received file processing request for $file");
432 my $ph = OpenSRF::Transport::PeerHandle->retrieve;
433 if(!$ph->flush_socket()) {
434 $logger->error("We received a request, bu we are no longer connected to opensrf. ".
435 "Exiting and dropping request for $file");
439 my ($imported, $failed) = (0, 0);
441 new_auth_token(); # login
443 if ($real_opts->{noqueue}) {
444 ($imported, $failed) = old_process_batch_data($file, 1);
446 ($imported, $failed) = process_batch_data($file, 1);
449 my $profile = (!$merge_profile) ? '' :
450 $apputils->simplereq(
452 'open-ils.pcrud.retrieve.vmp',
454 $merge_profile)->name;
457 $msg .= "Successfully imported $imported records using merge profile '$profile'\n" if $imported;
458 $msg .= "Failed to import $failed records\n" if $failed;
462 clear_auth_token(); # logout
466 # the authtoken will timeout after the configured inactivity period.
467 # When that happens, get a new one.
469 $authtoken = oils_login($oils_username, $oils_password, 'staff')
470 or die "Unable to login to Evergreen as user $oils_username";
474 sub clear_auth_token {
475 $apputils->simplereq(
477 'open-ils.auth.session.delete',
484 osrf_connect($osrf_config);
485 if ($real_opts->{nodaemon}) {
486 if (!$real_opts->{spoolfile}) {
487 print " --nodaemon mode requested, but no --spoolfile supplied!\n";
490 standalone_process_request($real_opts->{spoolfile});
492 print "Calling Net::Server run ", (@ARGV ? "with command-line options: " . join(' ', @ARGV) : ''), "\n";
493 __PACKAGE__->run(conf_file => $conf_file);
500 marc_stream_importer.pl - Import MARC records via bare socket connection.
504 ./marc_stream_importer.pl [common opts ...] [script opts ...] -- [Net::Server opts ...] &
506 This script uses the EG common options from B<Cronscript>. See --help output for those.
508 Run C<perldoc marc_stream_importer.pl> for full documentation.
510 Note the extra C<--> to separate options for the script wrapper from options for the
511 underlying L<Net::Server> options.
513 Note: this script has to be run in the same directory as B<oils_header.pl>.
515 Typical server-style execution will include a trailing C<&> to run in the background.
519 This script is a L<Net::Server::PreFork> instance for shoving records into Evergreen from a remote system.
523 The only required option is --password
525 --password =<eg_password>
526 --user =<eg_username> default: admin
527 --source =<bib_source> default: 1 Integer
528 --merge-profile =<i> default: 0
529 --tempdir =</temp/dir/> default: from L<opensrf.conf> <open-ils.vandelay/app_settings/databases/importer>
530 --source =<i> default: 1
531 --import-by-queue =<i> default: 0
532 --spoolfile =<import_file> default: NONE File to import in --nodaemon mode
533 --nodaemon default: OFF When used with --spoolfile, turns off Net::Server mode and runs this utility in the foreground
536 =head2 Old style: --noqueue and associated options
538 To bypass vandelay queue processing and push directly into the database (as the old style)
540 --noqueue default: OFF
541 --buffsize =<i> default: 4096 Buffer size. Only used by --noqueue
542 --wait =<i> default: 5 Seconds to read socket before processing. Only used by --noqueue
544 =head2 Net::Server Options
546 By default, the script will use the Net::Server configuration file B<marc_stream_importer.conf>. You can
547 override this by passing a filepath with the --conf_file option.
549 Other Net::Server options include: --port=<port> --min_servers=<X> --max_servers=<Y> and --log_file=[path/to/file]
551 See L<Net::Server> for a complete list.
555 =head3 OCLC Connexion
557 To use this script with OCLC Connexion, configure the client as follows:
559 Under Tools -> Options -> Export (tab)
560 Create -> Choose Connection -> OK -> Leave translation at "None"
561 -> Create -> Create -> choose TCP/IP (internet)
562 -> Enter hostname and Port, leave 'Use Telnet Protocol' checked
563 -> Create/OK your way out of the dialogs
564 Record Characteristics (button) -> Choose 'UTF-8 Unicode' for the Character Set
567 OCLC and Connexion are trademark/service marks of OCLC Online Computer Library Center, Inc.
571 WARNING: This script provides no inherent security layer. Any client that has
572 access to the server+port can inject MARC records into the system.
573 Use the available options (like allow/deny) in the Net::Server config file
574 or via the command line to restrict access as necessary.
578 ./marc_stream_importer.pl \
579 admin open-ils connexion --port 5555 --min_servers 2 \
580 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
582 ./marc_stream_importer.pl \
583 admin open-ils connexion --port 5555 --min_servers 2 \
584 --max_servers=20 --log_file=/openils/var/log/marc_net_importer.log &
588 L<Net::Server::PreFork>, L<marc_stream_importer.conf>
592 Bill Erickson <erickson@esilibrary.com>
593 Joe Atzberger <jatzberger@esilibrary.com>
594 Mike Rylander <miker@esilibrary.com> (nodaemon+spoolfile mode)