2 # ---------------------------------------------------------------
3 # Copyright © 2013 Merrimack Valley Library Consortium
4 # Jason Stephenson <jstephenson@mvlc.org>
6 # This program is part of Evergreen; you can redistribute it and/or
7 # modify it under the terms of the GNU General Public License as
8 # published by the Free Software Foundation; either version 2 of the
9 # License, or (at your option) any later version.
11 # This program is distributed in the hope that it will be useful,
12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 # GNU General Public License for more details.
15 # ---------------------------------------------------------------
18 use OpenILS::Utils::Fieldmapper;
19 use OpenILS::Application::AppUtils;
20 use OpenSRF::Utils::JSON;
23 use MARC::File::XML (BinaryEncoding => 'UTF-8');
24 use Date::Manip::Date;
25 my $U = 'OpenILS::Application::AppUtils';
27 binmode(STDERR, ':utf8');
31 our $config = Marque::Config->new();
32 Fieldmapper->import(IDL => $config->option_value('xml-idl'));
34 # Look for passed in ids:
36 if ($config->need_ids()) {
37 print STDERR "Waiting for input\n";
39 push @ids, $i if ($i =~ /^\s*[0-9]+\s*$/);
44 if ($config->option_value('type') eq 'authority') {
45 $exporter = Marque::Authority->new(\@ids);
47 $exporter = Marque::Biblio->new(\@ids);
50 Marque::Output::output($exporter);
52 # ------------------------------------------------------------------
53 package Marque::Config;
56 use List::MoreUtils qw(none);
58 use OpenSRF::Utils::SettingsClient;
60 use constant FORMATS => qw(USMARC UNIMARC XML BRE ARE);
61 use constant STORES => qw(reporter cstore storage);
62 use constant TYPES => qw(authority biblio);
70 # For command line options.
73 # set some default values
74 $opts{'format'} = 'USMARC';
75 $opts{'encoding'} = 'MARC8';
76 $opts{'type'} = 'biblio';
79 $opts{'config'} = '@sysconfdir@/opensrf_core.xml';
80 $opts{'store'} = 'reporter';
103 This script exports MARC authority, bibliographic, and serial holdings
104 records from an Evergreen database.
106 Input to this script can consist of a list of record IDs, with one record ID
107 per line, corresponding to the record ID in the Evergreen database table of
108 your requested record type.
110 Alternately, passing the --all option will attempt to export all records of
111 the specified type from the Evergreen database. The --all option starts at
112 record ID 1 and increments the ID by 1 until the largest ID in the database
113 is retrieved. This may not be very efficient for databases with large gaps
114 in their ID sequences.
117 --help or -h This screen.
118 --config or -c Configuration file [@sysconfdir@/opensrf_core.xml]
119 --format or -f Output format (USMARC, UNIMARC, XML, BRE, ARE) [USMARC]
120 --encoding or -e Output encoding (UTF-8, ISO-8859-?, MARC8) [MARC8]
121 --xml-idl or -x Location of the IDL XML
122 --timeout Remains for backward compatibility. No longer used.
123 --type or -t Record type (BIBLIO, AUTHORITY) [BIBLIO]
124 --all or -a Export all records; ignores input list
125 --replace_001 Replace the 001 field value with the record ID
126 --store Use the given storage backend to connect to the database.
127 Choices are (reporter, cstore, storage) [reporter]
128 --since Export records modified since a certain date and time.
130 Additional options for type = 'BIBLIO':
131 --items or -i Include items (holdings) in the output
132 --money Currency symbol to use in item price field [\$]
133 --mfhd Export serial MFHD records for associated bib records
134 Not compatible with --format=BRE
135 --location or -l MARC Location Code for holdings from
136 http://www.loc.gov/marc/organizations/orgshome.html
137 --library Export the bibliographic records that have attached
138 holdings for the listed library or libraries as
139 identified by shortname
143 To export a set of USMARC records in a file named "output_file" based on the
144 IDs contained in a file named "list_of_ids":
145 cat list_of_ids | $0 > output_file
147 To export a set of MARC21XML authority records in a file named "output.xml"
148 for all authority records in the database:
149 $0 --format XML --type AUTHORITY --all > output.xml
151 To export a set of USMARC bibliographic records encoded in UTF-8 in a file
152 named "sys1_bibs.mrc" based on records which have attached callnumbers for the
153 libraries with the short names "BR1" and "BR2":
155 $0 --library BR1 --library BR2 --encoding UTF-8 > sys1_bibs.mrc
161 OpenSRF::System->bootstrap_client( config_file => $opts{config} );
162 my $sclient = OpenSRF::Utils::SettingsClient->new();
163 unless ($opts{'xml-idl'}) {
164 $opts{'xml-idl'} = $sclient->config_value('IDL');
167 # Validate some of the settings.
168 if ($opts{all} && $opts{library}) {
169 die('Incompatible arguments: you cannot combine a request for all ' .
170 'records with a request for records by library');
172 if ($opts{all} && $opts{since}) {
173 die('Incompatible arguments: you cannot combine a request for all ' .
174 'records with a request for records added or changed since a certain date');
176 $opts{type} = lc($opts{type});
177 if (none {$_ eq $opts{type}} (TYPES)) {
178 die "Please select a supported type. ".
179 "Right now that means one of [".
180 join('|',(FORMATS)). "]\n";
182 $opts{format} = uc($opts{format});
183 if (none {$_ eq $opts{format}} (FORMATS)) {
184 die "Please select a supported format. ".
185 "Right now that means one of [".
186 join('|',(FORMATS)). "]\n";
189 if ($opts{format} eq 'ARE' && $opts{type} ne 'authority') {
190 die "Format ARE is not compatible with type " . $opts{type};
192 if ($opts{format} eq 'BRE' && $opts{type} ne 'biblio') {
193 die "Format BRE is not compatible with type " . $opts{type};
195 if ($opts{format} eq 'BRE' && $opts{items}) {
196 die "Format BRE is not compatible with exporting holdings."
200 if ($opts{type} ne 'biblio') {
201 die "MFHD export only works with bibliographic records.";
202 } elsif ($opts{format} eq 'BRE') {
203 die "MFHD export incompatible with format BRE.";
207 $opts{store} = lc($opts{store});
208 if (none {$_ eq $opts{store}} (STORES)) {
209 die "Please select a supported store. ".
210 "Right now that means one of [".
211 join('|',(STORES)). "]\n";
214 if ($opts{store} eq 'reporter') {
215 $app = 'open-ils.reporter-store';
217 $app = 'open-ils.' . $opts{store};
219 if ($app eq 'open-ils.storage') {
220 $self->{dbsettings} = $sclient->config_value(
221 apps => $app => app_settings => databases => 'database');
223 $self->{dbsettings} = $sclient->config_value(
224 apps => $app => app_settings => 'database');
227 $opts{encoding} = uc($opts{encoding});
229 $self->{'options'} = \%opts;
235 my ($self, $option) = @_;
236 return $self->{options}->{$option};
239 sub database_settings {
241 return $self->{dbsettings};
248 $rv = 0 if ($self->{options}->{all});
249 $rv = 0 if ($self->{options}->{since});
250 $rv = 0 if ($self->{options}->{library});
255 # ------------------------------------------------------------------
256 # This package exists to get a connection to the database. Since
257 # we'll need one for both biblio records and authorities, we've made a
258 # single subpackage with a function so that we don't have to duplicate
260 package Marque::Connector;
264 # Pass a Marque::Config object's database_settings return value into
265 # this to get a DBI connection.
267 # my $db = Marque::Connector::connect($config->database_settings);
271 # Build a connect string from the args.
272 my $connect_string = 'DBI:Pg:';
273 $connect_string .= 'dbname=' . $args->{db} . ';';
274 $connect_string .= 'host=' . $args->{host} . ';';
275 $connect_string .= 'port=' . $args->{port};
277 my $db_handle = DBI->connect($connect_string,
278 $args->{user}, $args->{pw});
282 # A function to get the date into a format better for PostgreSQL.
286 if (ref($input) eq 'Date::Manip::Date') {
289 $date = Date::Manip::Date->new();
290 if ($date->parse($input)) {
291 die "Can't parse date $input";
294 return $date->printf("%Y-%m-%dT%H:%M:%S%z");
297 # ------------------------------------------------------------------
298 # You would typically have the next two packages inherit from a common
299 # superclass, but ineritance doesn't seem to work when all packages
300 # are in single file, so we have some duplicated code between these
303 # Get bibliographic records from the database.
304 package Marque::Biblio;
309 my $self = {idlist => $idlist};
310 $self->{handle} = Marque::Connector::connect(
311 $Marque::config->database_settings);
312 $self->{since_date} = Date::Manip::Date->new;
313 $self->{since_date}->parse($Marque::config->option_value('since'));
315 # We need multiple fieldmapper classes depending on our
316 # options. We'll just get the information that we'll need for them
317 # all right here instead of only fetching the information when
319 $self->{breClass} = Fieldmapper::class_for_hint('bre');
320 $self->{acnClass} = Fieldmapper::class_for_hint('acn');
321 $self->{acpClass} = Fieldmapper::class_for_hint('acp');
322 $self->{sreClass} = Fieldmapper::class_for_hint('sre');
324 # Make an arrayref of shortname ids if the library option was
327 if ($Marque::config->option_value('library')) {
328 # This is done not only for speed, but to prevent SQL injection.
329 my $sth = $self->{handle}->prepare('select id from actor.org_unit where shortname=any(?::text[])');
330 if ($sth->execute($Marque::config->option_value('library'))) {
331 my $r = $sth->fetchall_arrayref();
332 my @ids = map {$_->[0]} @{$r};
333 $self->{libs} = \@ids;
345 # Get the field names and tables for our classes. We add the fully
346 # qualified table names to the fields so that the joins will work.
347 my $breTable = $self->{breClass}->Table();
348 my @breFields = map {$breTable . '.' . $_} $self->{breClass}->real_fields();
349 my $acnTable = $self->{acnClass}->Table();
350 my $acpTable = $self->{acpClass}->Table();
352 # Now we build the query in pieces:
354 # We always select the bre fields:
355 my $select = 'select distinct ' . join(',', @breFields);
356 # We always use the bre table.
357 my $from = "from $breTable";
359 # If have the libraries or items options, we need to join the
360 # asset.call_number table. If we have both, this variable checks
361 # that it has already been joined so we don't create an invalid
364 # Join to the acn table as needed for the library option.
365 if (@{$self->{libs}}) {
369 join $acnTable on $acnTable.record = $breTable.id
370 and $acnTable.owning_lib in (
372 $from .= join(',', @{$self->{libs}}) . ")";
373 $from .= "\nand $acnTable.deleted = 'f'" unless ($Marque::config->option_value('since'));
376 if ($Marque::config->option_value('items')) {
377 unless ($acn_joined) {
378 $from .= "\njoin $acnTable on $acnTable.record = $breTable.id";
379 $from .= "\nand $acnTable.deleted = 'f'" unless ($Marque::config->option_value('since'));
381 $from .= "\njoin $acpTable on $acpTable.call_number = $acnTable.id";
382 $from .= "\nand $acpTable.deleted = 'f'" unless ($Marque::config->option_value('since'));
385 # The where really depends on a few options:
386 my $where = "where $breTable.id > 0 and ";
387 # We fill in the where as necessary.
388 if ($self->{idlist} && @{$self->{idlist}}) {
389 $where .= "$breTable.id in (" . join(',', @{$self->{idlist}}) . ')';
390 } elsif ($Marque::config->option_value('since')) {
391 my $since_str = Marque::Connector::db_date($self->{since_date});
392 $where .= "($breTable.edit_date > '$since_str'";
393 $where .= " or $breTable.create_date > '$since_str')";
395 # We want all non-deleted records.
396 $where .= "$breTable.deleted = 'f'";
399 $self->{query} = $select . "\n" . $from . "\n" . $where;
404 $self->build_query() unless ($self->{query});
405 $self->{sth} = $self->{handle}->prepare($self->{query});
406 return $self->{sth}->execute;
413 # $r holds the record object, either sre or bre. $marc holds the
414 # current record's MARC, either sre.marc or bre.marc
416 # If we have the mfhd option and we've previously retrieved some
417 # sres, then we output one of the retrieved sres for each call
418 # until we run out. These sres "go with" the previous bib record.
419 if ($Marque::config->option_value('mfhd') && $self->{mfhds} && @{$self->{mfhds}}) {
420 $r = shift(@{$self->{mfhds}});
422 $marc = MARC::Record->new_from_xml($r->marc(),
423 $Marque::config->option_value('encoding'),
424 $Marque::config->option_value('format'));
427 print STDERR "Error in serial record " . $r->id() . "\n";
429 import MARC::File::XML; # Reset SAX Parser.
430 return $self->next();
433 my $data = $self->{sth}->fetchrow_hashref;
435 $r = $self->{breClass}->from_bare_hash($data);
436 if ($Marque::config->option_value('format') eq 'BRE') {
437 $output = OpenSRF::Utils::JSON->perl2JSON($r);
440 $marc = MARC::Record->new_from_xml($r->marc(),
441 $Marque::config->option_value('encoding'),
442 $Marque::config->option_value('format'));
445 print STDERR "Error in bibliograpic record " . $r->id() . "\n";
447 import MARC::File::XML; # Reset SAX Parser.
448 return $self->next();
450 if ($Marque::config->option_value('replace_001')) {
451 my $tcn = $marc->field('001');
453 $tcn->update($r->id());
455 $tcn = MARC::Field->new('001', $r->id());
456 $marc->insert_fields_ordered($tcn);
459 if ($Marque::config->option_value('items')) {
460 my @acps = $self->acps_for_bre($r);
461 foreach my $acp (@acps) {
463 my $location = $Marque::config->option_value('location');
464 my $price = ($acp->price() ? $Marque::config->option_value('money').$acp->price() : '');
465 $marc->insert_grouped_field(
468 ($location ? ('a' => $location) : ()),
469 b => $acp->call_number()->owning_lib()->shortname(),
470 b => $acp->circ_lib()->shortname(),
471 c => $acp->location()->name(),
472 j => $acp->call_number()->label(),
473 ($acp->circ_modifier() ? (g => $acp->circ_modifier()) : ()),
474 p => $acp->barcode(),
475 ($price ? (y => $price) : ()),
476 ($acp->copy_number() ? (t => $acp->copy_number()) : ()),
477 ($U->is_true($acp->ref()) ? (x => 'reference') : ()),
478 (!$U->is_true($acp->holdable()) ? (x => 'unholdable') : ()),
479 (!$U->is_true($acp->circulate()) ? (x => 'noncirculating') : ()),
480 (!$U->is_true($acp->opac_visible()) ? (x => 'hidden') : ())
484 if ($Marque::config->option_value('mfhd')) {
485 $self->{mfhds} = [$self->sres_for_bre($r)];
490 # Common stuff that doesn't depend on record type.
492 if ($Marque::config->option_value('since')) {
493 my $leader = $marc->leader();
494 if ($U->is_true($r->deleted())) {
495 substr($leader, 5, 1) = 'd';
496 $marc->leader($leader);
498 my $create_date = Date::Manip::Date->new;
499 $create_date->parse($r->create_date());
500 my $edit_date = Date::Manip::Date->new;
501 $edit_date->parse($r->edit_date());
502 if ($self->{since_date}->cmp($create_date) < 0) {
503 substr($leader, 5, 1) = 'n';
504 $marc->leader($leader);
505 } elsif ($self->{since_date}->cmp($edit_date) < 0) {
506 substr($leader, 5, 1) = 'c';
507 $marc->leader($leader);
511 if ($Marque::config->option_value('format') eq 'XML') {
513 $output = $marc->as_xml_record;
514 $output =~ s/^<\?.+?\?>$//mo;
517 print STDERR "Error in bibliograpic record " . $r->id() . "\n";
519 return $self->next();
523 $output = $marc->as_usmarc;
526 print STDERR "Error in bibliograpic record " . $r->id() . "\n";
528 return $self->next();
535 # Returns a list of aou objects in an array.
538 unless ($self->{orgs} && @{$self->{orgs}}) {
539 my $fmClass = Fieldmapper::class_for_hint('aou');
540 my @classFields = $fmClass->real_fields();
541 my $classTable = $fmClass->Table();
542 my $query = 'select ' . join(',', @classFields);
543 $query .= "\nfrom $classTable";
544 my $sth = $self->{handle}->prepare($query);
545 if ($sth->execute()) {
546 my $result = $sth->fetchall_arrayref({});
547 my @orgs = map {$fmClass->from_bare_hash($_)} @{$result};
548 $self->{orgs} = \@orgs;
553 return @{$self->{orgs}};
556 # Returns an array of acpl objects.
560 unless ($self->{shelves} && @{$self->{shelves}}) {
561 my $fmClass = Fieldmapper::class_for_hint('acpl');
562 my @classFields = $fmClass->real_fields();
563 my $classTable = $fmClass->Table();
564 my $query = 'select ' . join(',', @classFields);
565 $query .= "\nfrom $classTable";
566 my $result = $self->{handle}->selectall_arrayref($query, {Slice=>{}});
567 my @shelves = map {$fmClass->from_bare_hash($_)} @{$result};
568 $self->{shelves} = \@shelves;
571 return @{$self->{shelves}};
574 # Returns an array of acn objects for a given bre object or id.
578 $bre = $bre->id() if (ref($bre));
580 unless ($self->{acnHandle}) {
581 my $query = "select " . join(',', $self->{acnClass}->real_fields());
582 $query .= "\nfrom " . $self->{acnClass}->Table();
583 $query .= "\nwhere record = ?";
584 if (@{$self->{libs}}) {
585 $query .= "\nand owning_lib in (";
586 $query .= join(',', @{$self->{libs}}) . ")";
588 $query .= "\nand deleted = 'f'" unless($Marque::config->option_value('since'));
589 $self->{acnHandle} = $self->{handle}->prepare($query);
592 if ($self->{acnHandle}->execute($bre)) {
593 my $result = $self->{acnHandle}->fetchall_arrayref({});
594 return map {$self->{acnClass}->from_bare_hash($_)} @{$result};
597 # If for some reason, we don't find anything.
601 # Returns an array of acp objects for a given bre object or id.
605 $bre = $bre->id() if (ref($bre));
607 my @orgs = $self->orgs();
608 my @locations = $self->shelves();
610 my @acns = $self->acns_for_bre($bre);
612 my $query = 'select ' . join(',', $self->{acpClass}->real_fields());
613 $query .= "\nfrom " . $self->{acpClass}->Table();
614 $query .= "\nwhere call_number in (";
615 $query .= join(',', map {$_->id()} @acns);
617 $query .= "\nand deleted = 'f'" unless ($Marque::config->option_value('since'));
618 my $result = $self->{handle}->selectall_arrayref($query, {Slice=>{}});
619 if ($result && @{$result}) {
620 my @acps = map {$self->{acpClass}->from_bare_hash($_)} @{$result};
622 my $cn = $_->call_number();
623 my $clib = $_->circ_lib();
624 my $loc = $_->location();
625 my ($org) = grep {$_->id() == $clib} @orgs;
626 my ($acn) = grep {$_->id() == $cn} @acns;
627 my ($location) = grep {$_->id() == $loc} @locations;
628 my $olib = $acn->owning_lib();
629 my ($owner) = grep {$_->id() == $olib} @orgs;
630 $acn->owning_lib($owner);
631 $_->call_number($acn);
633 $_->location($location);
639 # If for some reason, we don't find anything.
643 # Retreive an array for sre objects when the --mfhd option is used.
647 $bre = $bre->id() if (ref($bre));
649 # Build a query to retrieve SREs when the MFHD option is passed.
650 if ($Marque::config->option_value('mfhd')) {
651 # Create a persistent handle as needed.
652 unless ($self->{sreSth}) {
653 my $query = "select " . join(',', $self->{sreClass}->real_fields());
654 $query .= "\nfrom " . $self->{sreClass}->Table();
655 $query .= "\nwhere record = ?";
656 $query .= "\nand deleted = 'f'" unless ($Marque::config->option_value('since'));
657 $self->{sreSth} = $self->{handle}->prepare($query);
659 if ($self->{sreSth}->execute($bre)) {
660 while (my $data = $self->{sreSth}->fetchrow_hashref) {
661 push @sres, $self->{sreClass}->from_bare_hash($data);
663 $self->{sreSth}->finish; # Sometimes DBI complains.
670 # Get authority records from the database.
671 package Marque::Authority;
676 my $self = {idlist => $idlist};
677 $self->{handle} = Marque::Connector::connect(
678 $Marque::config->database_settings);
679 $self->{fmClass} = Fieldmapper::class_for_hint('are');
680 $self->{since_date} = Date::Manip::Date->new;
681 $self->{since_date}->parse($Marque::config->option_value('since'));
689 # Get the information for are object from the Fieldmapper:
690 my @fields = $self->{fmClass}->real_fields();
691 my $table = $self->{fmClass}->Table();
693 # Build the actual query.
694 my $select = "select " . join(',', @fields);
695 my $from = "from $table";
696 my $where = 'where ';
698 # If we have an idlist, we pretty much ignore anything else.
699 if ($self->{idlist} && @{$self->{idlist}}) {
700 $where .= 'id in (' . join(',', @{$self->{idlist}}) . ')';
701 } elsif ($Marque::config->option_value('since')) {
702 my $since_str = Marque::Connector::db_date($self->{since_date});
703 $where .= "edit_date > '$since_str'";
704 $where .= " or create_date > '$since_str'";
706 # We want all non-deleted records.
707 $where .= "deleted = 'f'";
710 $self->{query} = $select . "\n" . $from . "\n" . $where;
715 $self->build_query() unless ($self->{query});
716 $self->{sth} = $self->{handle}->prepare($self->{query});
717 return $self->{sth}->execute;
723 my $data = $self->{sth}->fetchrow_hashref;
726 my $format = $Marque::config->option_value('format');
727 my $r = $self->{fmClass}->from_bare_hash($data);
728 if ($format eq 'ARE') {
729 $output = OpenSRF::Utils::JSON->perl2JSON($r);
733 $marc = MARC::Record->new_from_xml($r->marc(),
734 $Marque::config->option_value('encoding'),
735 $Marque::config->option_value('format'));
738 print STDERR "Error in authority record " . $r->id() . "\n";
740 import MARC::File::XML; # Reset SAX Parser.
741 return $self->next();
743 if ($Marque::config->option_value('replace_001')) {
744 my $tcn = $marc->field('001');
746 $tcn->update($r->id());
748 $tcn = MARC::Field->new('001', $r->id());
749 $marc->insert_fields_ordered($tcn);
752 if ($Marque::config->option_value('since')) {
753 my $leader = $marc->leader();
754 if ($U->is_true($r->deleted())) {
755 substr($leader, 5, 1) = 'd';
756 $marc->leader($leader);
758 my $create_date = Date::Manip::Date->new;
759 $create_date->parse($r->create_date());
760 my $edit_date = Date::Manip::Date->new;
761 $edit_date->parse($r->edit_date());
762 if ($self->{since_date}->cmp($create_date) < 0) {
763 substr($leader, 5, 1) = 'n';
764 $marc->leader($leader);
765 } elsif ($self->{since_date}->cmp($edit_date) < 0) {
766 substr($leader, 5, 1) = 'c';
767 $marc->leader($leader);
771 if ($Marque::config->option_value('format') eq 'XML') {
773 $output = $marc->as_xml_record;
774 $output =~ s/^<\?.+?\?>$//mo;
777 print STDERR "Error in authority record " . $r->id() . "\n";
779 return $self->next();
783 $output = $marc->as_usmarc;
786 print STDERR "Error in authority record " . $r->id() . "\n";
788 return $self->next();
797 # ------------------------------------------------------------------
798 # Since the ultimate output is largely independent of the type of the
799 # records, we use a single subpackage to group our output routines.
800 package Marque::Output;
803 my $extractor = shift;
804 if ($extractor->execute_query) {
805 if ($Marque::config->option_value('encoding') eq 'UTF-8') {
806 binmode(STDOUT, ':utf8');
808 binmode(STDOUT, ':raw');
812 while (my $output = $extractor->next()) {
817 print STDERR $extractor->{query} if ($Marque::config->option_value('debug'));
818 die "Database query failed!";
823 if ($Marque::config->option_value('format') eq 'XML') {
824 my $encoding = $Marque::config->option_value('encoding');
826 <?xml version="1.0" encoding="$encoding"?>
827 <collection xmlns='http://www.loc.gov/MARC21/slim'>
833 if ($Marque::config->option_value('format') eq 'XML') {
834 print "</collection>\n";