Merge pull request #475 from psu-libraries/176-punctuation-fix

right to left language formatting

.gitignore

@@ -26,5 +26,6 @@ config/environments/*.local.yml
 .bash_history
 .cache
 .local
+.envrc
 data/
 vendor/**/*

.rubocop.yml

@@ -49,6 +49,7 @@ RSpec/NestedGroups:
     - 'spec/lib/psulib_traject/processors/call_number/lc_spec.rb'
     - 'spec/lib/psulib_traject/processors/call_number/dewey_spec.rb'
     - 'spec/lib/psulib_traject/processors/title_display_spec.rb'
+    - 'spec/lib/psulib_traject/processors/pub_display_spec.rb'
 
 RSpec/ExpectActual:
   Exclude:

Gemfile.lock

@@ -251,6 +251,7 @@ PLATFORMS
   universal-java-14
   universal-java-15
   universal-java-17
+  universal-java-19
   x86_64-darwin-20
   x86_64-linux
 

config/traject.rb

@@ -218,13 +218,19 @@
 to_field 'pub_date_itsi', process_publication_date
 
 ## Publication fields for display
-to_field 'publication_display_ssm', extract_marc('260abcefg3:264|*1|abc3') # display in search results
-to_field 'overall_imprint_display_ssm', extract_marc('260abcefg3:264|*0|abc3:264|*1|abc3:264|*2|abc3:264|*3|abc3') # display on single item page
 to_field 'copyright_display_ssm', extract_marc('264|*4|c')
-to_field 'edition_display_ssm', extract_marc('250ab3')
 to_field 'cartographic_mathematical_data_ssm', extract_marc('255abcdefg')
 to_field 'other_edition_ssm', extract_marc('775|0*|iabcdefghkmnor')
 to_field 'collection_facet', extract_marc('793a')
+to_field 'publication_display_ssm', extract_marc('260abcefg3:264|*1|abc3') # display in search results
+to_field 'overall_imprint_display_ssm', extract_marc('260abcefg3:264|*0|abc3:264|*1|abc3:264|*2|abc3:264|*3|abc3') # display on single item page
+to_field 'edition_display_ssm', extract_marc('250ab3')
+# processes display fields to help format vernacular display
+each_record do |_record, context|
+  PsulibTraject::Processors::PubDisplay.new('publication', context).call
+  PsulibTraject::Processors::PubDisplay.new('overall_imprint', context).call
+  PsulibTraject::Processors::PubDisplay.new('edition', context).call
+end
 
 ## Publication fields for Illiad and Aeon
 to_field 'pub_date_illiad_ssm', extract_marc('260c:264|*1|c'), trim_punctuation

lib/psulib_traject.rb

@@ -34,6 +34,7 @@ module PsulibTraject
   require 'psulib_traject/processors/record_type'
   require 'psulib_traject/processors/summary_holdings'
   require 'psulib_traject/processors/title_display'
+  require 'psulib_traject/processors/pub_display'
   require 'psulib_traject/processors/oclc_extract'
   require 'psulib_traject/shelf_key'
   require 'psulib_traject/solr_manager'

lib/psulib_traject/processors/pub_display.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module PsulibTraject::Processors
+  class PubDisplay
+    def initialize(field, context)
+      @context = context
+      @field = field
+    end
+
+    def call
+      return if final.nil?
+
+      final[1] = vern_clean unless vern.nil?
+    end
+
+    private
+
+      attr_accessor :context, :field
+
+      def vern
+        final.length <= 1 ? nil : final[1]
+      end
+
+      def final
+        context.output_hash["#{field}_display_ssm"]
+      end
+
+      def vern_clean
+        return vern unless /[\u0621-\u064A]+\.$/.match?(vern) # regex to check for arabic
+
+        vern_value = vern.gsub('.', '')
+        ".#{vern_value}"
+      end
+  end
+end

spec/integration/publication_spec.rb

@@ -0,0 +1,72 @@
+# frozen_string_literal: true
+
+RSpec::Matchers.define_negated_matcher :not_include, :include
+
+RSpec.describe 'Publication' do
+  let(:leader) { '1234567890' }
+
+  describe 'Record with a right to left vernacular edition' do
+    let(:field) { 'edition_display_ssm' }
+    let(:edition_250) do
+      { '250' => { 'subfields' => [{ '6' => '880-03' },
+                                   { 'a' => 'Chāp-i avval.' }] } }
+    end
+    let(:edition_vern_250) do
+      { '880' => { 'subfields' => [{ '6' => '250-03' },
+                                   { 'a' => 'چاپ اول.' }] } }
+    end
+    let(:result) { indexer.map_record(MARC::Record.new_from_hash('fields' => [edition_250, edition_vern_250], 'leader' => leader)) }
+
+    it 'has the vernacular edition with a period on the left in the edition statement' do
+      expect(result[field]).to eq ['Chāp-i avval.', '.چاپ اول']
+      expect(result[field].length).to eq 2
+    end
+
+    it 'has empty vern and latin edition field' do
+      expect(result).not_to include('edition_vern')
+      expect(result).not_to include('edition_latin')
+    end
+  end
+
+  describe 'Record with a left to right vernacular edition' do
+    let(:field) { 'edition_display_ssm' }
+    let(:edition_250) do
+      { '250' => { 'subfields' => [{ '6' => '880-03' },
+                                   { 'a' => 'Tōkyō : Hayakawa Shobō, 1999.' }] } }
+    end
+    let(:edition_vern_250) do
+      { '880' => { 'subfields' => [{ '6' => '250-03' },
+                                   { 'a' => '東京 : 早川書房, 1999.' }] } }
+    end
+    let(:result) { indexer.map_record(MARC::Record.new_from_hash('fields' => [edition_250, edition_vern_250], 'leader' => leader)) }
+
+    it 'has the vernacular edition with a period on the right in the edition statement' do
+      expect(result[field]).to eq ['Tōkyō : Hayakawa Shobō, 1999.', '東京 : 早川書房, 1999.']
+      expect(result[field].length).to eq 2
+    end
+
+    it 'has empty vern and latin edition field' do
+      expect(result).not_to include('edition_vern')
+      expect(result).not_to include('edition_latin')
+    end
+  end
+
+  describe 'Record with no vernacular edition' do
+    let(:field) { 'edition_display_ssm' }
+    let(:edition_250) do
+      { '250' => { 'subfields' => [{ '6' => '880-03' },
+                                   { 'a' => 'First Edition.' }] } }
+    end
+    let(:result) { indexer.map_record(MARC::Record.new_from_hash('fields' => [edition_250], 'leader' => leader)) }
+
+    it 'has the vernacular edition with a period on the right in the edition statement' do
+      expect(result[field]).to eq ['First Edition.']
+      expect(result[field].length).to eq 1
+    end
+
+    it 'has empty vern and latin edition field' do
+      expect(result).not_to include('edition_vern')
+      expect(result).not_to include('edition_latin')
+    end
+  end
+end

spec/lib/psulib_traject/processors/pub_display_spec.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+RSpec.describe PsulibTraject::Processors::PubDisplay do
+  let(:context) { instance_double 'Context' }
+
+  before do
+    allow(context).to receive(:output_hash).and_return(output_hash)
+  end
+
+  describe '::call' do
+    context 'when there is not a vernacular version of the field' do
+      let(:output_hash) do
+        { 'edition_display_ssm' => ['thing'] }
+      end
+
+      it 'sets the field_display_ssm to the field_latin and deletes field_latin' do
+        described_class.new('edition', context).call
+        expect(output_hash['edition_display_ssm']).to eq ['thing']
+      end
+    end
+
+    context 'when there is a vernacular version of the field' do
+      context 'when field_vern contains right to left arabic equals sign' do
+        context 'when there is a period on the right end of field_vern' do
+          let(:output_hash) do
+            { 'edition_display_ssm' => ['Chāp-i avval.', 'چاپ اول.'] }
+          end
+
+          it 'moves the period of the vern to the left, adds field_latin & field_vern to field_display_ssm, and deletes field_latin and field_vern' do
+            described_class.new('edition', context).call
+            expect(output_hash['edition_display_ssm']).to eq ['Chāp-i avval.', '.چاپ اول']
+          end
+        end
+
+        context 'when there is not a period on the right end of field_vern' do
+          let(:output_hash) do
+            { 'edition_display_ssm' => ['Chāp-i avval.', 'چاپ اول'] }
+          end
+
+          it 'does not change the field_vern, adds field_latin & field_vern to field_display_ssm, and deletes field_latin and field_vern' do
+            described_class.new('edition', context).call
+            expect(output_hash['edition_display_ssm']).to eq ['Chāp-i avval.', 'چاپ اول']
+          end
+        end
+      end
+
+      context 'when field_vern does not contain right to left arabic equals sign' do
+        let(:output_hash) do
+          { 'edition_display_ssm' => ['Tōkyō : Hayakawa Shobō, 1999.', '東京 : 早川書房, 1999.'] }
+        end
+
+        it 'adds field_latin & field_vern to field_display_ssm and deletes field_latin and field_vern' do
+          described_class.new('edition', context).call
+          expect(output_hash['edition_display_ssm']).to eq ['Tōkyō : Hayakawa Shobō, 1999.', '東京 : 早川書房, 1999.']
+        end
+      end
+    end
+  end
+end