Merge pull request #749 from sul-dlss/647-refactor-name-extraction

Refactor name extraction to introduce role mapping

.rubocop_todo.yml

@@ -113,6 +113,7 @@ RSpec/DescribedClass:
     - 'spec/lib/traject/macros/harvard_scw_spec.rb'
     - 'spec/lib/traject/macros/laussane_spec.rb'
     - 'spec/lib/traject/macros/manchester_spec.rb'
+    - 'spec/lib/traject/macros/mods_spec.rb'
     - 'spec/lib/traject/macros/string_helper_spec.rb'
     - 'spec/lib/traject/macros/title_extraction_spec.rb'
 

lib/macros/language_extraction.rb

@@ -32,6 +32,7 @@ def arabic_script_lang_or_default(arabic_script_lang, default)
           ar_values = []
           default_values = []
           accumulator.each do |val|
+            val = translate_role_to_ar(val)
             lang_code = val.match?(/[ضصثقفغعهخحمنتالبيسشظطذدزرو]/) ? arabic_script_lang : default
             ar_values << val if lang_code == arabic_script_lang
             default_values << val if lang_code == default
@@ -63,5 +64,17 @@ def tei_lower_resource_language
         accumulator.replace([{ language: TO_BCP47[:"#{language}"], values: [extracted_string] }]) if extracted_string
       end
     end
+
+    # Returns the provided value with the english role translated to arabic if found
+    def translate_role_to_ar(val)
+      role_name = val.match(/(\()(\w+)(\))/)&.to_a&.at(2)
+      return val unless role_name
+
+      role_ar_map = Traject::TranslationMap.new('role_ar_from_en')
+      translated_role = role_ar_map[role_name]
+      return val unless translated_role
+
+      val.gsub(role_name, translated_role)
+    end
   end
 end

lib/macros/mods.rb

@@ -21,7 +21,10 @@ def extract_name(xpath, role: nil, exclude: nil)
         name_nodes.each do |val|
           name_parts << val&.content&.strip
         end
-        accumulator.replace(["#{name_parts.join(', ')} (#{role})"]) if name_parts.present?
+
+        role_map = Traject::TranslationMap.new('role_from_contributor')
+        role = " (#{role_map[role]})" if role
+        accumulator.replace(["#{name_parts.join(', ')}#{role}"]) if name_parts.present?
       end
     end
 

lib/translation_maps/role_ar_from_en.yaml

@@ -0,0 +1,12 @@
+Artist: فنان
+Author: مؤلف
+Calligrapher: خطاط
+Copyist: الناسخ
+Creator: المنشئ
+Former owner: مالك سابق
+Former repository: المستودع السابق
+Illuminator: المنور
+Illustrator: المصور
+Painter: دهان
+Patron: نمط
+Scribe: كاتب

lib/translation_maps/role_from_contributor.yaml

@@ -0,0 +1,14 @@
+artist: Artist
+author: Author
+calligrapher: Calligrapher
+copyist: Copyist
+copyist.: Copyist
+cre: Creator
+creator: Creator
+former owner.: Former owner
+former repository: Former repository
+illuminator: Illuminator
+illustrator: Illustrator
+painter (artist): Painter
+patron: Patron
+scribe.: Scribe

spec/lib/traject/macros/language_extraction_spec.rb

@@ -19,5 +19,14 @@
         expect(callable.call(nil, [extracted_string])).to eq([{ language: 'ar-Arab', values: [extracted_string] }])
       end
     end
+
+    context 'when extracted string contains Arabic characters and an english role' do
+      it 'returns the correct script value arabic_script_lang' do
+        extracted_string = 'الولايات المتحدة الامريكيه (Copyist)'
+        translated_string = 'الولايات المتحدة الامريكيه (الناسخ)'
+        callable = instance.arabic_script_lang_or_default('ar-Arab', 'en')
+        expect(callable.call(nil, [extracted_string])).to eq([{ language: 'ar-Arab', values: [translated_string] }])
+      end
+    end
   end
 end

spec/lib/traject/macros/mods_spec.rb

@@ -0,0 +1,73 @@
+# frozen_string_literal: true
+
+require 'macros/mods'
+require 'macros/language_extraction'
+
+RSpec.describe Macros::Mods do
+  subject(:indexer) do
+    Traject::Indexer.new.tap do |indexer|
+      indexer.instance_eval do
+        extend Macros::Mods
+        extend Macros::LanguageExtraction
+      end
+    end
+  end
+
+  describe '#extract_name' do
+    let(:record) do
+      <<~XML
+        <mods:mods xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+          xmlns:mods="http://www.loc.gov/mods/v3"
+          xmlns:sets="http://hul.harvard.edu/ois/xml/ns/sets"
+          xmlns:xlink="http://www.w3.org/1999/xlink"
+          xmlns:marc="http://www.loc.gov/MARC21/slim"
+          xmlns:HarvardDRS="http://hul.harvard.edu/ois/xml/ns/HarvardDRS"
+          xmlns:librarycloud="http://hul.harvard.edu/ois/xml/ns/librarycloud" xsi:schemaLocation="http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-6.xsd" version="3.6">
+          <mods:name>
+            <mods:namePart>Doe, John</mods:namePart>
+            <mods:role>
+              <mods:roleTerm type="text">copyist.</mods:roleTerm>
+            </mods:role>
+          </mods:name>
+        </mods>
+      XML
+    end
+    let(:ng_rec) { Nokogiri::XML.parse(record) }
+
+    context 'when contributor role matches the search' do
+      before do
+        indexer.instance_eval do
+          to_field 'cho_contributor', extract_name('//*/mods:name[1][mods:role/mods:roleTerm/', role: 'copyist.')
+        end
+      end
+
+      it 'returns values with a mapped role' do
+        expect(indexer.map_record(ng_rec)).to eq({ 'cho_contributor' => ['Doe, John (Copyist)'] })
+      end
+    end
+
+    context 'when contributor role does not match the search' do
+      before do
+        indexer.instance_eval do
+          to_field 'cho_contributor', extract_name('//*/mods:name[1][mods:role/mods:roleTerm/', role: 'scribe.')
+        end
+      end
+
+      it 'returns an empty hash' do
+        expect(indexer.map_record(ng_rec)).to eq({})
+      end
+    end
+
+    context 'when role is excluded from the search' do
+      before do
+        indexer.instance_eval do
+          to_field 'cho_contributor', extract_name('//*/mods:name[1][mods:role/mods:roleTerm/', exclude: true)
+        end
+      end
+
+      it 'returns values without role included' do
+        expect(indexer.map_record(ng_rec)).to eq({ 'cho_contributor' => ['Doe, John'] })
+      end
+    end
+  end
+end