#
# This file is part of Koha.
#
-# Koha is free software; you can redistribute it and/or modify it under the
-# terms of the GNU General Public License as published by the Free Software
-# Foundation; either version 3 of the License, or (at your option) any later
-# version.
+# Koha is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
#
-# Koha is distributed in the hope that it will be useful, but WITHOUT ANY
-# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
-# A PARTICULAR PURPOSE. See the GNU General Public License for more details.
+# Koha is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
#
-# You should have received a copy of the GNU General Public License along
-# with Koha; if not, write to the Free Software Foundation, Inc.,
-# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+# You should have received a copy of the GNU General Public License
+# along with Koha; if not, see <http://www.gnu.org/licenses>.
use base qw(Class::Accessor);
use Koha::Database;
use Koha::Exceptions::Config;
+use Koha::Exceptions::Elasticsearch;
use Koha::SearchFields;
use Koha::SearchMarcMaps;
+use C4::Heading;
use Carp;
+use Clone qw(clone);
use JSON;
use Modern::Perl;
use Readonly;
use Try::Tiny;
use YAML::Syck;
-use Data::Dumper; # TODO remove
+use List::Util qw( sum0 reduce );
+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 ));
my $class = shift @_;
my $self = $class->SUPER::new(@_);
# Check for a valid index
- croak('No index name provided') unless $self->index;
+ Koha::Exceptions::MissingParameter->throw('No index name provided') unless $self->index;
return $self;
}
+=head2 get_elasticsearch
+
+ my $elasticsearch_client = $self->get_elasticsearch();
+
+Returns a C<Search::Elasticsearch> client. The client is cached on a C<Koha::SearchEngine::ElasticSearch>
+instance level and will be reused if method is called multiple times.
+
+=cut
+
+sub get_elasticsearch {
+ my $self = shift @_;
+ unless (defined $self->{elasticsearch}) {
+ my $conf = $self->get_elasticsearch_params();
+ $self->{elasticsearch} = Search::Elasticsearch->new($conf);
+ }
+ return $self->{elasticsearch};
+}
+
=head2 get_elasticsearch_params
my $params = $self->get_elasticsearch_params();
else {
die "No elasticsearch servers were specified in koha-conf.xml.\n";
}
- die "No elasticserver index_name was specified in koha-conf.xml.\n"
+ die "No elasticsearch index_name was specified in koha-conf.xml.\n"
if ( !$es->{index_name} );
# Append the name of this particular index to our namespace
$es->{index_name} .= '_' . $self->index;
$es->{key_prefix} = 'es_';
+ $es->{client} //= '5_0::Direct';
+ $es->{cxn_pool} //= 'Static';
+ $es->{request_timeout} //= 60;
+
return $es;
}
my $settings = $self->get_elasticsearch_settings();
-This provides the settings provided to elasticsearch when an index is created.
-These can do things like define tokenisation methods.
+This provides the settings provided to Elasticsearch when an index is created.
+These can do things like define tokenization methods.
A hashref containing the settings is returned.
sub get_elasticsearch_settings {
my ($self) = @_;
- # Ultimately this should come from a file or something, and not be
- # hardcoded.
- my $settings = {
- index => {
- analysis => {
- analyzer => {
- analyser_phrase => {
- tokenizer => 'icu_tokenizer',
- filter => ['icu_folding'],
- },
- analyser_standard => {
- tokenizer => 'icu_tokenizer',
- filter => ['icu_folding'],
- },
- },
- }
- }
- };
+ # Use state to speed up repeated calls
+ state $settings = undef;
+ if (!defined $settings) {
+ my $config_file = C4::Context->config('elasticsearch_index_config');
+ $config_file ||= C4::Context->config('intranetdir') . '/admin/searchengine/elasticsearch/index_config.yaml';
+ $settings = LoadFile( $config_file );
+ }
+
return $settings;
}
my $mappings = $self->get_elasticsearch_mappings();
-This provides the mappings that get passed to elasticsearch when an index is
+This provides the mappings that get passed to Elasticsearch when an index is
created.
=cut
sub get_elasticsearch_mappings {
my ($self) = @_;
- # TODO cache in the object?
- my $mappings = {
- data => {
- _all => {type => "string", analyzer => "analyser_standard"},
- properties => {
- record => {
- store => "true",
- include_in_all => JSON::false,
- type => "text",
- },
- }
- }
- };
- my %sort_fields;
- my $marcflavour = lc C4::Context->preference('marcflavour');
- $self->_foreach_mapping(
- sub {
- my ( $name, $type, $facet, $suggestible, $sort, $marc_type ) = @_;
- return if $marc_type ne $marcflavour;
- # TODO if this gets any sort of complexity to it, it should
- # be broken out into its own function.
-
- # TODO be aware of date formats, but this requires pre-parsing
- # as ES will simply reject anything with an invalid date.
- my $es_type =
- $type eq 'boolean'
- ? 'boolean'
- : 'text';
-
- if ($es_type eq 'boolean') {
- $mappings->{data}{properties}{$name} = _elasticsearch_mapping_for_boolean( $name, $es_type, $facet, $suggestible, $sort, $marc_type );
- return; #Boolean cannot have facets nor sorting nor suggestions
- } else {
- $mappings->{data}{properties}{$name} = _elasticsearch_mapping_for_default( $name, $es_type, $facet, $suggestible, $sort, $marc_type );
- }
+ # Use state to speed up repeated calls
+ state %all_mappings;
+ state %sort_fields;
+
+ if (!defined $all_mappings{$self->index}) {
+ $sort_fields{$self->index} = {};
+ # Clone the general mapping to break ties with the original hash
+ my $mappings = {
+ data => clone(_get_elasticsearch_field_config('general', ''))
+ };
+ my $marcflavour = lc C4::Context->preference('marcflavour');
+ $self->_foreach_mapping(
+ sub {
+ my ( $name, $type, $facet, $suggestible, $sort, $search, $marc_type ) = @_;
+ return if $marc_type ne $marcflavour;
+ # TODO if this gets any sort of complexity to it, it should
+ # be broken out into its own function.
+
+ # TODO be aware of date formats, but this requires pre-parsing
+ # as ES will simply reject anything with an invalid date.
+ my $es_type = 'text';
+ if ($type eq 'boolean') {
+ $es_type = 'boolean';
+ } elsif ($type eq 'number' || $type eq 'sum') {
+ $es_type = 'integer';
+ } elsif ($type eq 'isbn' || $type eq 'stdno') {
+ $es_type = 'stdno';
+ }
- if ($facet) {
- $mappings->{data}{properties}{ $name . '__facet' } = {
- type => "keyword",
- };
- }
- if ($suggestible) {
- $mappings->{data}{properties}{ $name . '__suggestion' } = {
- type => 'completion',
- analyzer => 'simple',
- search_analyzer => 'simple',
- };
- }
- # Sort is a bit special as it can be true, false, undef.
- # We care about "true" or "undef",
- # "undef" means to do the default thing, which is make it sortable.
- if ($sort || !defined $sort) {
- $mappings->{data}{properties}{ $name . '__sort' } = {
- search_analyzer => "analyser_phrase",
- analyzer => "analyser_phrase",
- type => "text",
- include_in_all => JSON::false,
- fields => {
- phrase => {
- type => "keyword",
- },
- },
- };
- $sort_fields{$name} = 1;
+ if ($search) {
+ $mappings->{data}{properties}{$name} = _get_elasticsearch_field_config('search', $es_type);
+ }
+
+ if ($facet) {
+ $mappings->{data}{properties}{ $name . '__facet' } = _get_elasticsearch_field_config('facet', $es_type);
+ }
+ if ($suggestible) {
+ $mappings->{data}{properties}{ $name . '__suggestion' } = _get_elasticsearch_field_config('suggestible', $es_type);
+ }
+ # Sort is a bit special as it can be true, false, undef.
+ # We care about "true" or "undef",
+ # "undef" means to do the default thing, which is make it sortable.
+ if (!defined $sort || $sort) {
+ $mappings->{data}{properties}{ $name . '__sort' } = _get_elasticsearch_field_config('sort', $es_type);
+ $sort_fields{$self->index}{$name} = 1;
+ }
}
- }
- );
- $self->sort_fields(\%sort_fields);
- return $mappings;
+ );
+ $all_mappings{$self->index} = $mappings;
+ }
+ $self->sort_fields(\%{$sort_fields{$self->index}});
+
+ return $all_mappings{$self->index};
}
-=head2 _elasticsearch_mapping_for_*
+=head2 _get_elasticsearch_field_config
-Get the ES mappings for the given data type or a special mapping case
+Get the Elasticsearch field config for the given purpose and data type.
-Receives the same parameters from the $self->_foreach_mapping() dispatcher
+$mapping = _get_elasticsearch_field_config('search', 'text');
=cut
-sub _elasticsearch_mapping_for_boolean {
- my ( $name, $type, $facet, $suggestible, $sort, $marc_type ) = @_;
+sub _get_elasticsearch_field_config {
- return {
- type => $type,
- null_value => 0,
- };
+ my ( $purpose, $type ) = @_;
+
+ # Use state to speed up repeated calls
+ state $settings = undef;
+ if (!defined $settings) {
+ my $config_file = C4::Context->config('elasticsearch_field_config');
+ $config_file ||= C4::Context->config('intranetdir') . '/admin/searchengine/elasticsearch/field_config.yaml';
+ $settings = LoadFile( $config_file );
+ }
+
+ if (!defined $settings->{$purpose}) {
+ die "Field purpose $purpose not defined in field config";
+ }
+ if ($type eq '') {
+ return $settings->{$purpose};
+ }
+ if (defined $settings->{$purpose}{$type}) {
+ return $settings->{$purpose}{$type};
+ }
+ if (defined $settings->{$purpose}{'default'}) {
+ return $settings->{$purpose}{'default'};
+ }
+ return;
}
-sub _elasticsearch_mapping_for_default {
- my ( $name, $type, $facet, $suggestible, $sort, $marc_type ) = @_;
-
- return {
- search_analyzer => "analyser_standard",
- analyzer => "analyser_standard",
- type => $type,
- fields => {
- phrase => {
- search_analyzer => "analyser_phrase",
- analyzer => "analyser_phrase",
- type => "text",
- },
- raw => {
- type => "keyword",
- }
- },
- };
+=head2 _load_elasticsearch_mappings
+
+Load Elasticsearch mappings in the format of mappings.yaml.
+
+$indexes = _load_elasticsearch_mappings();
+
+=cut
+
+sub _load_elasticsearch_mappings {
+ my $mappings_yaml = C4::Context->config('elasticsearch_index_mappings');
+ $mappings_yaml ||= C4::Context->config('intranetdir') . '/admin/searchengine/elasticsearch/mappings.yaml';
+ return LoadFile( $mappings_yaml );
}
sub reset_elasticsearch_mappings {
- my $mappings_yaml = C4::Context->config('intranetdir') . '/admin/searchengine/elasticsearch/mappings.yaml';
- my $indexes = LoadFile( $mappings_yaml );
+ my ( $self ) = @_;
+ my $indexes = $self->_load_elasticsearch_mappings();
+
+ Koha::SearchMarcMaps->delete;
+ Koha::SearchFields->delete;
while ( my ( $index_name, $fields ) = each %$indexes ) {
while ( my ( $field_name, $data ) = each %$fields ) {
- my $field_type = $data->{type};
- my $field_label = $data->{label};
+
+ my %sf_params = map { $_ => $data->{$_} } grep { exists $data->{$_} } qw/ type label weight staff_client opac facet_order /;
+
+ # Set default values
+ $sf_params{staff_client} //= 1;
+ $sf_params{opac} //= 1;
+
+ $sf_params{name} = $field_name;
+
+ my $search_field = Koha::SearchFields->find_or_create( \%sf_params, { key => 'name' } );
+
my $mappings = $data->{mappings};
- my $search_field = Koha::SearchFields->find_or_create({ name => $field_name, label => $field_label, type => $field_type }, { key => 'name' });
for my $mapping ( @$mappings ) {
- my $marc_field = Koha::SearchMarcMaps->find_or_create({ index_name => $index_name, marc_type => $mapping->{marc_type}, marc_field => $mapping->{marc_field} });
- $search_field->add_to_search_marc_maps($marc_field, { facet => $mapping->{facet} || 0, suggestible => $mapping->{suggestible} || 0, sort => $mapping->{sort} } );
+ my $marc_field = Koha::SearchMarcMaps->find_or_create({
+ index_name => $index_name,
+ marc_type => $mapping->{marc_type},
+ marc_field => $mapping->{marc_field}
+ });
+ $search_field->add_to_search_marc_maps($marc_field, {
+ facet => $mapping->{facet} || 0,
+ suggestible => $mapping->{suggestible} || 0,
+ sort => $mapping->{sort},
+ search => $mapping->{search} // 1
+ });
}
}
}
return $self->_sort_fields_accessor();
}
-# Provides the rules for data conversion.
-sub get_fixer_rules {
- my ($self) = @_;
+=head2 _process_mappings($mappings, $data, $record_document, $altscript)
+ $self->_process_mappings($mappings, $marc_field_data, $record_document, 0)
+
+Process all C<$mappings> targets operating on a specific MARC field C<$data>.
+Since we group all mappings by MARC field targets C<$mappings> will contain
+all targets for C<$data> and thus we need to fetch the MARC field only once.
+C<$mappings> will be applied to C<$record_document> and new field values added.
+The method has no return value.
+
+=over 4
+
+=item C<$mappings>
+
+Arrayref of mappings containing arrayrefs in the format
+[C<$target>, C<$options>] where C<$target> is the name of the target field and
+C<$options> is a hashref containing processing directives for this particular
+mapping.
+
+=item C<$data>
+
+The source data from a MARC record field.
+
+=item C<$record_document>
+
+Hashref representing the Elasticsearch document on which mappings should be
+applied.
+
+=item C<$altscript>
+
+A boolean value indicating whether an alternate script presentation is being
+processed.
+
+=back
+
+=cut
+
+sub _process_mappings {
+ my ($_self, $mappings, $data, $record_document, $altscript) = @_;
+ foreach my $mapping (@{$mappings}) {
+ my ($target, $options) = @{$mapping};
+
+ # Don't process sort fields for alternate scripts
+ my $sort = $target =~ /__sort$/;
+ if ($sort && $altscript) {
+ next;
+ }
+
+ # Copy (scalar) data since can have multiple targets
+ # with differing options for (possibly) mutating data
+ # so need a different copy for each
+ my $_data = $data;
+ $record_document->{$target} //= [];
+ if (defined $options->{substr}) {
+ my ($start, $length) = @{$options->{substr}};
+ $_data = length($data) > $start ? substr $data, $start, $length : '';
+ }
+ if (defined $options->{value_callbacks}) {
+ $_data = reduce { $b->($a) } ($_data, @{$options->{value_callbacks}});
+ }
+ if (defined $options->{property}) {
+ $_data = {
+ $options->{property} => $_data
+ }
+ }
+ push @{$record_document->{$target}}, $_data;
+ }
+}
+
+=head2 marc_records_to_documents($marc_records)
+
+ my $record_documents = $self->marc_records_to_documents($marc_records);
+
+Using mappings stored in database convert C<$marc_records> to Elasticsearch documents.
+
+Returns array of hash references, representing Elasticsearch documents,
+acceptable as body payload in C<Search::Elasticsearch> requests.
+
+=over 4
+
+=item C<$marc_documents>
+
+Reference to array of C<MARC::Record> objects to be converted to Elasticsearch documents.
+
+=back
+
+=cut
+
+sub marc_records_to_documents {
+ my ($self, $records) = @_;
+ my $rules = $self->_get_marc_mapping_rules();
+ my $control_fields_rules = $rules->{control_fields};
+ my $data_fields_rules = $rules->{data_fields};
my $marcflavour = lc C4::Context->preference('marcflavour');
- my @rules;
+ my $use_array = C4::Context->preference('ElasticsearchMARCFormat') eq 'ARRAY';
- $self->_foreach_mapping(
- sub {
- my ( $name, $type, $facet, $suggestible, $sort, $marc_type, $marc_field ) = @_;
- return if $marc_type ne $marcflavour;
- my $options = '';
-
- # There's a bug when using 'split' with something that
- # selects a range
- # The split makes everything into nested arrays, but that's not
- # really a big deal, ES doesn't mind.
- $options = '' unless $marc_field =~ m|_/| || $type eq 'sum';
- push @rules, "marc_map('$marc_field','${name}.\$append', $options)";
- if ($facet) {
- push @rules, "marc_map('$marc_field','${name}__facet.\$append', $options)";
+ my @record_documents;
+
+ foreach my $record (@{$records}) {
+ my $record_document = {};
+ my $mappings = $rules->{leader};
+ if ($mappings) {
+ $self->_process_mappings($mappings, $record->leader(), $record_document, 0);
+ }
+ foreach my $field ($record->fields()) {
+ if ($field->is_control_field()) {
+ my $mappings = $control_fields_rules->{$field->tag()};
+ if ($mappings) {
+ $self->_process_mappings($mappings, $field->data(), $record_document, 0);
+ }
+ }
+ else {
+ my $tag = $field->tag();
+ # Handle alternate scripts in MARC 21
+ my $altscript = 0;
+ if ($marcflavour eq 'marc21' && $tag eq '880') {
+ my $sub6 = $field->subfield('6');
+ if ($sub6 =~ /^(...)-\d+/) {
+ $tag = $1;
+ $altscript = 1;
+ }
+ }
+
+ my $data_field_rules = $data_fields_rules->{$tag};
+ if ($data_field_rules) {
+ my $subfields_mappings = $data_field_rules->{subfields};
+ my $wildcard_mappings = $subfields_mappings->{'*'};
+ foreach my $subfield ($field->subfields()) {
+ my ($code, $data) = @{$subfield};
+ my $mappings = $subfields_mappings->{$code} // [];
+ if ($wildcard_mappings) {
+ $mappings = [@{$mappings}, @{$wildcard_mappings}];
+ }
+ if (@{$mappings}) {
+ $self->_process_mappings($mappings, $data, $record_document, $altscript);
+ }
+ if ( defined @{$mappings}[0] && grep /match-heading/, @{@{$mappings}[0]} ){
+ # Used by the authority linker the match-heading field requires a specific syntax
+ # that is specified in C4/Heading
+ my $heading = C4::Heading->new_from_field( $field, undef, 1 ); #new auth heading
+ next unless $heading;
+ push @{$record_document->{'match-heading'}}, $heading->search_form;
+ }
+ }
+
+ my $subfields_join_mappings = $data_field_rules->{subfields_join};
+ if ($subfields_join_mappings) {
+ foreach my $subfields_group (keys %{$subfields_join_mappings}) {
+ # Map each subfield to values, remove empty values, join with space
+ my $data = join(
+ ' ',
+ grep(
+ $_,
+ map { join(' ', $field->subfield($_)) } split(//, $subfields_group)
+ )
+ );
+ if ($data) {
+ $self->_process_mappings($subfields_join_mappings->{$subfields_group}, $data, $record_document, $altscript);
+ }
+ if ( grep { $_->[0] eq 'match-heading' } @{$subfields_join_mappings->{$subfields_group}} ){
+ # Used by the authority linker the match-heading field requires a specific syntax
+ # that is specified in C4/Heading
+ my $heading = C4::Heading->new_from_field( $field, undef, 1 ); #new auth heading
+ next unless $heading;
+ push @{$record_document->{'match-heading'}}, $heading->search_form;
+ }
+ }
+ }
+ }
+ }
+ }
+ foreach my $field (keys %{$rules->{defaults}}) {
+ unless (defined $record_document->{$field}) {
+ $record_document->{$field} = $rules->{defaults}->{$field};
}
- if ($suggestible) {
- push @rules,
- #"marc_map('$marc_field','${name}__suggestion.input.\$append', $options)"; #must not have nested data structures in .input
- "marc_map('$marc_field','${name}__suggestion.input.\$append')";
+ }
+ foreach my $field (@{$rules->{sum}}) {
+ if (defined $record_document->{$field}) {
+ # TODO: validate numeric? filter?
+ # TODO: Or should only accept fields without nested values?
+ # TODO: Quick and dirty, improve if needed
+ $record_document->{$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;
}
- if ( $type eq 'boolean' ) {
+ }
- # boolean gets special handling, basically if it doesn't exist,
- # it's added and set to false. Otherwise we can't query it.
- push @rules,
- "unless exists('$name') add_field('$name', 0) end";
+ # Remove duplicate values and collapse sort fields
+ foreach my $field (keys %{$record_document}) {
+ if (ref($record_document->{$field}) eq 'ARRAY') {
+ @{$record_document->{$field}} = do {
+ my %seen;
+ grep { !$seen{ref($_) eq 'HASH' && defined $_->{input} ? $_->{input} : $_}++ } @{$record_document->{$field}};
+ };
+ if ($field =~ /__sort$/) {
+ # Make sure to keep the sort field length sensible. 255 was chosen as a nice round value.
+ $record_document->{$field} = [substr(join(' ', @{$record_document->{$field}}), 0, 255)];
+ }
}
- if ($type eq 'sum' ) {
- push @rules, "sum('$name')";
+ }
+
+ # TODO: Perhaps should check if $records_document non empty, but really should never be the case
+ $record->encoding('UTF-8');
+ if ($use_array) {
+ $record_document->{'marc_data_array'} = $self->_marc_to_array($record);
+ $record_document->{'marc_format'} = 'ARRAY';
+ } else {
+ my @warnings;
+ {
+ # Temporarily intercept all warn signals (MARC::Record carps when record length > 99999)
+ local $SIG{__WARN__} = sub {
+ push @warnings, $_[0];
+ };
+ $record_document->{'marc_data'} = encode_base64(encode('UTF-8', $record->as_usmarc()));
}
- if ($self->sort_fields()->{$name}) {
- if ($sort || !defined $sort) {
- push @rules, "marc_map('$marc_field','${name}__sort.\$append', $options)";
+ if (@warnings) {
+ # Suppress warnings if record length exceeded
+ unless (substr($record->leader(), 0, 5) eq '99999') {
+ foreach my $warning (@warnings) {
+ carp $warning;
+ }
}
+ $record_document->{'marc_data'} = $record->as_xml_record($marcflavour);
+ $record_document->{'marc_format'} = 'MARCXML';
+ }
+ else {
+ $record_document->{'marc_format'} = 'base64ISO2709';
}
}
- );
+ push @record_documents, $record_document;
+ }
+ return \@record_documents;
+}
+
+=head2 _marc_to_array($record)
+
+ my @fields = _marc_to_array($record)
+
+Convert a MARC::Record to an array modeled after MARC-in-JSON
+(see https://github.com/marc4j/marc4j/wiki/MARC-in-JSON-Description)
+
+=over 4
+
+=item C<$record>
+
+A MARC::Record object
+
+=back
+
+=cut
+
+sub _marc_to_array {
+ my ($self, $record) = @_;
+
+ my $data = {
+ leader => $record->leader(),
+ fields => []
+ };
+ for my $field ($record->fields()) {
+ my $tag = $field->tag();
+ if ($field->is_control_field()) {
+ push @{$data->{fields}}, {$tag => $field->data()};
+ } else {
+ my $subfields = ();
+ foreach my $subfield ($field->subfields()) {
+ my ($code, $contents) = @{$subfield};
+ push @{$subfields}, {$code => $contents};
+ }
+ push @{$data->{fields}}, {
+ $tag => {
+ ind1 => $field->indicator(1),
+ ind2 => $field->indicator(2),
+ subfields => $subfields
+ }
+ };
+ }
+ }
+ return $data;
+}
+
+=head2 _array_to_marc($data)
+
+ my $record = _array_to_marc($data)
+
+Convert an array modeled after MARC-in-JSON to a MARC::Record
+
+=over 4
+
+=item C<$data>
+
+An array modeled after MARC-in-JSON
+(see https://github.com/marc4j/marc4j/wiki/MARC-in-JSON-Description)
+
+=back
+
+=cut
- push @rules, "move_field(_id,es_id)"; #Also you must set the Catmandu::Store::ElasticSearch->new(key_prefix: 'es_');
- return \@rules;
+sub _array_to_marc {
+ my ($self, $data) = @_;
+
+ my $record = MARC::Record->new();
+
+ $record->leader($data->{leader});
+ for my $field (@{$data->{fields}}) {
+ my $tag = (keys %{$field})[0];
+ $field = $field->{$tag};
+ my $marc_field;
+ if (ref($field) eq 'HASH') {
+ my @subfields;
+ foreach my $subfield (@{$field->{subfields}}) {
+ my $code = (keys %{$subfield})[0];
+ push @subfields, $code;
+ push @subfields, $subfield->{$code};
+ }
+ $marc_field = MARC::Field->new($tag, $field->{ind1}, $field->{ind2}, @subfields);
+ } else {
+ $marc_field = MARC::Field->new($tag, $field)
+ }
+ $record->append_fields($marc_field);
+ }
+;
+ return $record;
+}
+
+=head2 _field_mappings($facet, $suggestible, $sort, $search, $target_name, $target_type, $range)
+
+ my @mappings = _field_mappings($facet, $suggestible, $sort, $search, $target_name, $target_type, $range)
+
+Get mappings, an internal data structure later used by
+L<_process_mappings($mappings, $data, $record_document, $altscript)> to process MARC target
+data for a MARC mapping.
+
+The returned C<$mappings> is not to to be confused with mappings provided by
+C<_foreach_mapping>, rather this sub accepts properties from a mapping as
+provided by C<_foreach_mapping> and expands it to this internal data structure.
+In the caller context (C<_get_marc_mapping_rules>) the returned C<@mappings>
+is then applied to each MARC target (leader, control field data, subfield or
+joined subfields) and integrated into the mapping rules data structure used in
+C<marc_records_to_documents> to transform MARC records into Elasticsearch
+documents.
+
+=over 4
+
+=item C<$facet>
+
+Boolean indicating whether to create a facet field for this mapping.
+
+=item C<$suggestible>
+
+Boolean indicating whether to create a suggestion field for this mapping.
+
+=item C<$sort>
+
+Boolean indicating whether to create a sort field for this mapping.
+
+=item C<$search>
+
+Boolean indicating whether to create a search field for this mapping.
+
+=item C<$target_name>
+
+Elasticsearch document target field name.
+
+=item C<$target_type>
+
+Elasticsearch document target field type.
+
+=item C<$range>
+
+An optional range as a string in the format "<START>-<END>" or "<START>",
+where "<START>" and "<END>" are integers specifying a range that will be used
+for extracting a substring from MARC data as Elasticsearch field target value.
+
+The first character position is "0", and the range is inclusive,
+so "0-2" means the first three characters of MARC data.
+
+If only "<START>" is provided only one character at position "<START>" will
+be extracted.
+
+=back
+
+=cut
+
+sub _field_mappings {
+ my ($_self, $facet, $suggestible, $sort, $search, $target_name, $target_type, $range) = @_;
+ my %mapping_defaults = ();
+ my @mappings;
+
+ my $substr_args = undef;
+ if (defined $range) {
+ # TODO: use value_callback instead?
+ my ($start, $end) = map(int, split /-/, $range, 2);
+ $substr_args = [$start];
+ push @{$substr_args}, (defined $end ? $end - $start + 1 : 1);
+ }
+ my $default_options = {};
+ if ($substr_args) {
+ $default_options->{substr} = $substr_args;
+ }
+
+ # TODO: Should probably have per type value callback/hook
+ # but hard code for now
+ if ($target_type eq 'boolean') {
+ $default_options->{value_callbacks} //= [];
+ push @{$default_options->{value_callbacks}}, sub {
+ my ($value) = @_;
+ # Trim whitespace at both ends
+ $value =~ s/^\s+|\s+$//g;
+ return $value ? 'true' : 'false';
+ };
+ }
+
+ if ($search) {
+ my $mapping = [$target_name, $default_options];
+ push @mappings, $mapping;
+ }
+
+ my @suffixes = ();
+ push @suffixes, 'facet' if $facet;
+ push @suffixes, 'suggestion' if $suggestible;
+ push @suffixes, 'sort' if !defined $sort || $sort;
+
+ foreach my $suffix (@suffixes) {
+ my $mapping = ["${target_name}__$suffix"];
+ # TODO: Hack, fix later in less hideous manner
+ if ($suffix eq 'suggestion') {
+ push @{$mapping}, {%{$default_options}, property => 'input'};
+ }
+ else {
+ push @{$mapping}, $default_options;
+ }
+ push @mappings, $mapping;
+ }
+ return @mappings;
+};
+
+=head2 _get_marc_mapping_rules
+
+ my $mapping_rules = $self->_get_marc_mapping_rules()
+
+Generates rules from mappings stored in database for MARC records to Elasticsearch JSON document conversion.
+
+Since field retrieval is slow in C<MARC::Records> (all fields are itereted through for
+each call to C<MARC::Record>->field) we create an optimized structure of mapping
+rules keyed by MARC field tags holding all the mapping rules for that particular tag.
+
+We can then iterate through all MARC fields for each record and apply all relevant
+rules once per fields instead of retreiving fields multiple times for each mapping rule
+which is terribly slow.
+
+=cut
+
+# TODO: This structure can be used for processing multiple MARC::Records so is currently
+# rebuilt for each batch. Since it is cacheable it could also be stored in an in
+# memory cache which it is currently not. The performance gain of caching
+# would probably be marginal, but to do this could be a further improvement.
+
+sub _get_marc_mapping_rules {
+ my ($self) = @_;
+ my $marcflavour = lc C4::Context->preference('marcflavour');
+ my $field_spec_regexp = qr/^([0-9]{3})([()0-9a-zA-Z]+)?(?:_\/(\d+(?:-\d+)?))?$/;
+ my $leader_regexp = qr/^leader(?:_\/(\d+(?:-\d+)?))?$/;
+ my $rules = {
+ 'leader' => [],
+ 'control_fields' => {},
+ 'data_fields' => {},
+ 'sum' => [],
+ 'isbn' => [],
+ 'defaults' => {}
+ };
+
+ $self->_foreach_mapping(sub {
+ my ($name, $type, $facet, $suggestible, $sort, $search, $marc_type, $marc_field) = @_;
+ return if $marc_type ne $marcflavour;
+
+ if ($type eq 'sum') {
+ push @{$rules->{sum}}, $name;
+ push @{$rules->{sum}}, $name."__sort" if $sort;
+ }
+ elsif ($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
+ $rules->{defaults}->{$name} = 'false';
+ }
+
+ if ($marc_field =~ $field_spec_regexp) {
+ my $field_tag = $1;
+
+ my @subfields;
+ my @subfield_groups;
+ # Parse and separate subfields form subfield groups
+ if (defined $2) {
+ my $subfield_group = '';
+ my $open_group = 0;
+
+ foreach my $token (split //, $2) {
+ if ($token eq "(") {
+ if ($open_group) {
+ Koha::Exceptions::Elasticsearch::MARCFieldExprParseError->throw(
+ "Unmatched opening parenthesis for $marc_field"
+ );
+ }
+ else {
+ $open_group = 1;
+ }
+ }
+ elsif ($token eq ")") {
+ if ($open_group) {
+ if ($subfield_group) {
+ push @subfield_groups, $subfield_group;
+ $subfield_group = '';
+ }
+ $open_group = 0;
+ }
+ else {
+ Koha::Exceptions::Elasticsearch::MARCFieldExprParseError->throw(
+ "Unmatched closing parenthesis for $marc_field"
+ );
+ }
+ }
+ elsif ($open_group) {
+ $subfield_group .= $token;
+ }
+ else {
+ push @subfields, $token;
+ }
+ }
+ }
+ else {
+ push @subfields, '*';
+ }
+
+ my $range = defined $3 ? $3 : undef;
+ my @mappings = $self->_field_mappings($facet, $suggestible, $sort, $search, $name, $type, $range);
+ if ($field_tag < 10) {
+ $rules->{control_fields}->{$field_tag} //= [];
+ push @{$rules->{control_fields}->{$field_tag}}, @mappings;
+ }
+ else {
+ $rules->{data_fields}->{$field_tag} //= {};
+ foreach my $subfield (@subfields) {
+ $rules->{data_fields}->{$field_tag}->{subfields}->{$subfield} //= [];
+ push @{$rules->{data_fields}->{$field_tag}->{subfields}->{$subfield}}, @mappings;
+ }
+ foreach my $subfield_group (@subfield_groups) {
+ $rules->{data_fields}->{$field_tag}->{subfields_join}->{$subfield_group} //= [];
+ push @{$rules->{data_fields}->{$field_tag}->{subfields_join}->{$subfield_group}}, @mappings;
+ }
+ }
+ }
+ elsif ($marc_field =~ $leader_regexp) {
+ my $range = defined $1 ? $1 : undef;
+ my @mappings = $self->_field_mappings($facet, $suggestible, $sort, $search, $name, $type, $range);
+ push @{$rules->{leader}}, @mappings;
+ }
+ else {
+ Koha::Exceptions::Elasticsearch::MARCFieldExprParseError->throw(
+ "Invalid MARC field expression: $marc_field"
+ );
+ }
+ });
+ return $rules;
}
=head2 _foreach_mapping
'search_marc_to_fields.facet',
'search_marc_to_fields.suggestible',
'search_marc_to_fields.sort',
+ 'search_marc_to_fields.search',
'search_marc_map.marc_type',
'search_marc_map.marc_field',
],
'facet',
'suggestible',
'sort',
+ 'search',
'marc_type',
'marc_field',
],
while ( my $search_field = $search_fields->next ) {
$sub->(
- $search_field->name,
+ # Force lower case on indexed field names for case insensitive
+ # field name searches
+ lc($search_field->name),
$search_field->type,
$search_field->get_column('facet'),
$search_field->get_column('suggestible'),
$search_field->get_column('sort'),
+ $search_field->get_column('search'),
$search_field->get_column('marc_type'),
$search_field->get_column('marc_field'),
);
return $configuration;
}
+=head2 get_facetable_fields
+
+my @facetable_fields = Koha::SearchEngine::Elasticsearch->get_facetable_fields();
+
+Returns the list of Koha::SearchFields marked to be faceted in the ES configuration
+
+=cut
+
+sub get_facetable_fields {
+ my ($self) = @_;
+
+ # These should correspond to the ES field names, as opposed to the CCL
+ # things that zebra uses.
+ my @search_field_names = qw( author itype location su-geo title-series subject ccode holdingbranch homebranch ln );
+ my @faceted_fields = Koha::SearchFields->search(
+ { name => { -in => \@search_field_names }, facet_order => { '!=' => undef } }, { order_by => ['facet_order'] }
+ );
+ my @not_faceted_fields = Koha::SearchFields->search(
+ { name => { -in => \@search_field_names }, facet_order => undef }, { order_by => ['facet_order'] }
+ );
+ # This could certainly be improved
+ return ( @faceted_fields, @not_faceted_fields );
+}
+
1;
__END__
projects / srvgit / blobdiff