Attachment #82272 for bug #20244

View | Details | Raw Unified | Return to bug 20244
Collapse All | Expand All




use MARC::File::XML;
use MIME::Base64;
use Encode qw(encode);
use Business::ISBN;

__PACKAGE__->mk_ro_accessors(qw( index ));
__PACKAGE__->mk_accessors(qw( sort_fields ));

                $record_document->{$field} = sum0(grep { !ref($_) && m/\d+(\.\d+)?/} @{$record_document->{$field}});
            }
        }
        # Index all applicable ISBN forms (ISBN-10 and ISBN-13 with and without dashes)
        foreach my $field (@{$rules->{isbn}}) {
            if (defined $record_document->{$field}) {
                my @isbns = ();
                foreach my $input_isbn (@{$record_document->{$field}}) {
                    my $isbn = Business::ISBN->new($input_isbn);
                    if (defined $isbn && $isbn->is_valid) {
                        my $isbn13 = $isbn->as_isbn13->as_string;
                        push @isbns, $isbn13;
                        $isbn13 =~ s/\-//g;
                        push @isbns, $isbn13;

                        my $isbn10 = $isbn->as_isbn10;
                        if ($isbn10) {
                            $isbn10 = $isbn10->as_string;
                            push @isbns, $isbn10;
                            $isbn10 =~ s/\-//g;
                            push @isbns, $isbn10;
                        }
                    } else {
                        push @isbns, $input_isbn;
                    }
                }
                $record_document->{$field} = \@isbns;
            }
        }

        # TODO: Perhaps should check if $records_document non empty, but really should never be the case
        $record->encoding('UTF-8');
        my @warnings;

        'control_fields' => {},
        'data_fields' => {},
        'sum' => [],
        'isbn' => [],
        'defaults' => {}
    };


        if ($type eq 'sum') {
            push @{$rules->{sum}}, $name;
        }
        elsif ($type eq 'isbn') {
            push @{$rules->{isbn}}, $name;
        }
        elsif ($type eq 'boolean') {
            # boolean gets special handling, if value doesn't exist for a field,
            # it is set to false

Lines 117-123 subtest 'get_elasticsearch_mappings() tests' => sub { Link Here

(-)a/t/Koha/SearchEngine/Elasticsearch.t (-3 / +14 lines)
117		117
118	subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' => sub {	118	subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' => sub {
119		119
120	plan tests => 45;	120	plan tests => 32;
121		121
122	t::lib::Mocks::mock_preference('marcflavour', 'MARC21');	122	t::lib::Mocks::mock_preference('marcflavour', 'MARC21');
123		123
Lines 131-136 subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' Link Here
131	marc_type => 'marc21',	131	marc_type => 'marc21',
132	marc_field => '001',	132	marc_field => '001',
133	},	133	},
		134	{
		135	name => 'isbn',
		136	type => 'isbn',
		137	facet => 0,
		138	suggestible => 0,
		139	sort => 0,
		140	marc_type => 'marc21',
		141	marc_field => '020a',
		142	},
134	{	143	{
135	name => 'author',	144	name => 'author',
136	type => 'string',	145	type => 'string',
Lines 246-251 subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' Link Here
246	$marc_record_1->leader(' cam 22 a 4500');	255	$marc_record_1->leader(' cam 22 a 4500');
247	$marc_record_1->append_fields(	256	$marc_record_1->append_fields(
248	MARC::Field->new('001', '123'),	257	MARC::Field->new('001', '123'),
		258	MARC::Field->new('020', '', '', a => '1-56619-909-3'),
249	MARC::Field->new('100', '', '', a => 'Author 1'),	259	MARC::Field->new('100', '', '', a => 'Author 1'),
250	MARC::Field->new('110', '', '', a => 'Corp Author'),	260	MARC::Field->new('110', '', '', a => 'Corp Author'),
251	MARC::Field->new('210', '', '', a => 'Title 1'),	261	MARC::Field->new('210', '', '', a => 'Title 1'),
Lines 271-277 subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' Link Here
271	my $docs = $see->marc_records_to_documents($records);	281	my $docs = $see->marc_records_to_documents($records);
272		282
273	# First record:	283	# First record:
274
275	is(scalar @{$docs}, 2, 'Two records converted to documents');	284	is(scalar @{$docs}, 2, 'Two records converted to documents');
276		285
277	is($docs->[0][0], '1234567', 'First document biblionumber should be set as first element in document touple');	286	is($docs->[0][0], '1234567', 'First document biblionumber should be set as first element in document touple');
Lines 360-365 subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' Link Here
360	'First document type_of_record_and_bib_level field should be set correctly'	369	'First document type_of_record_and_bib_level field should be set correctly'
361	);	370	);
362		371
		372	is(scalar @{$docs->[0][1]->{isbn}}, 4, 'First document isbn field should contain four values');
		373	is_deeply($docs->[0][1]->{isbn}, ['978-1-56619-909-4', '9781566199094', '1-56619-909-3', '1566199093'], 'First document isbn field should be set correctly');
		374
363	# Second record:	375	# Second record:
364		376
365	is(scalar @{$docs->[1][1]->{author}}, 1, 'Second document author field should contain one value');	377	is(scalar @{$docs->[1][1]->{author}}, 1, 'Second document author field should contain one value');
366	-

Return to bug 20244

Lines 38-43 use List::Util qw( sum0 reduce ); Link Here

(-)a/Koha/SearchEngine/Elasticsearch.pm (+32 lines)
38	use MARC::File::XML;	38	use MARC::File::XML;
39	use MIME::Base64;	39	use MIME::Base64;
40	use Encode qw(encode);	40	use Encode qw(encode);
		41	use Business::ISBN;
41		42
42	__PACKAGE__->mk_ro_accessors(qw( index ));	43	__PACKAGE__->mk_ro_accessors(qw( index ));
43	__PACKAGE__->mk_accessors(qw( sort_fields ));	44	__PACKAGE__->mk_accessors(qw( sort_fields ));
Lines 456-461 sub marc_records_to_documents { Link Here
456	$record_document->{$field} = sum0(grep { !ref($_) && m/\d+(\.\d+)?/} @{$record_document->{$field}});	457	$record_document->{$field} = sum0(grep { !ref($_) && m/\d+(\.\d+)?/} @{$record_document->{$field}});
457	}	458	}
458	}	459	}
		460	# Index all applicable ISBN forms (ISBN-10 and ISBN-13 with and without dashes)
		461	foreach my $field (@{$rules->{isbn}}) {
		462	if (defined $record_document->{$field}) {
		463	my @isbns = ();
		464	foreach my $input_isbn (@{$record_document->{$field}}) {
		465	my $isbn = Business::ISBN->new($input_isbn);
		466	if (defined $isbn && $isbn->is_valid) {
		467	my $isbn13 = $isbn->as_isbn13->as_string;
		468	push @isbns, $isbn13;
		469	$isbn13 =~ s/\-//g;
		470	push @isbns, $isbn13;
		471
		472	my $isbn10 = $isbn->as_isbn10;
		473	if ($isbn10) {
		474	$isbn10 = $isbn10->as_string;
		475	push @isbns, $isbn10;
		476	$isbn10 =~ s/\-//g;
		477	push @isbns, $isbn10;
		478	}
		479	} else {
		480	push @isbns, $input_isbn;
		481	}
		482	}
		483	$record_document->{$field} = \@isbns;
		484	}
		485	}
		486
459	# TODO: Perhaps should check if $records_document non empty, but really should never be the case	487	# TODO: Perhaps should check if $records_document non empty, but really should never be the case
460	$record->encoding('UTF-8');	488	$record->encoding('UTF-8');
461	my @warnings;	489	my @warnings;
Lines 622-627 sub _get_marc_mapping_rules { Link Here
622	'control_fields' => {},	650	'control_fields' => {},
623	'data_fields' => {},	651	'data_fields' => {},
624	'sum' => [],	652	'sum' => [],
		653	'isbn' => [],
625	'defaults' => {}	654	'defaults' => {}
626	};	655	};
627		656
Lines 632-637 sub _get_marc_mapping_rules { Link Here
632	if ($type eq 'sum') {	661	if ($type eq 'sum') {
633	push @{$rules->{sum}}, $name;	662	push @{$rules->{sum}}, $name;
634	}	663	}
		664	elsif ($type eq 'isbn') {
		665	push @{$rules->{isbn}}, $name;
		666	}
635	elsif ($type eq 'boolean') {	667	elsif ($type eq 'boolean') {
636	# boolean gets special handling, if value doesn't exist for a field,	668	# boolean gets special handling, if value doesn't exist for a field,
637	# it is set to false	669	# it is set to false