Update display data format to accommodate multiple linked terms per r…

…ecorded value
mdholloway · Jan 8, 2025 · d95536d · d95536d
1 parent cf7638e
commit d95536d
Show file tree

Hide file tree

Showing 5 changed files with 51 additions and 50 deletions.
diff --git a/lib/digital_scriptorium/claim_transformer.rb b/lib/digital_scriptorium/claim_transformer.rb
@@ -16,13 +16,15 @@ def self.transform(claim, export_hash, config)
       authority_property_id = config['authority']
 
       value = primary_value_from_claim(claim, export_hash)
+      display_data = { 'recorded_value' => value }
+      search_terms = [value]
 
       solr_props['id'] = [value] if requested_fields.include? 'id'
       solr_props["#{prefix}_meta"] = [value] if requested_fields.include? 'meta'
 
       unless authority_property_id && claim.qualifiers_by_property_id?(authority_property_id)
-        solr_props["#{prefix}_display"] = [{ 'PV' => value }.to_json] if requested_fields.include? 'display'
-        solr_props["#{prefix}_search"] = [value] if requested_fields.include? 'search'
+        solr_props["#{prefix}_display"] = [display_data.to_json] if requested_fields.include? 'display'
+        solr_props["#{prefix}_search"] = search_terms if requested_fields.include? 'search'
         solr_props["#{prefix}_facet"] = [value] if requested_fields.include? 'facet'
 
         solr_props['images_facet'] = ['Yes'] if value && claim.property_id == IIIF_MANIFEST
@@ -31,37 +33,37 @@ def self.transform(claim, export_hash, config)
         return solr_props
       end
 
-      display_entries = []
-      search_entries = [value]
       facets = []
+      linked_terms = []
 
       claim.qualifiers_by_property_id(authority_property_id).each do |qualifier|
-        display_props = { 'PV' => value }
-
         authority_id = qualifier.entity_id_value
         authority = export_hash[authority_id]
 
-        if authority
-          label = authority.label('en')
+        next unless authority
+
+        term = {}
+        label = authority.label('en')
 
-          display_props['QL'] = label
-          search_entries << label
-          facets << label
+        term['label'] = label
+        search_terms << label
+        facets << label
 
-          external_uri = authority.claim_by_property_id(EXTERNAL_URI)&.data_value
-          wikidata_id = authority.claim_by_property_id(WIKIDATA_QID)&.data_value
-          wikidata_uri = wikidata_id && "https://www.wikidata.org/wiki/#{wikidata_id}"
+        external_uri = authority.claim_by_property_id(EXTERNAL_URI)&.data_value
+        wikidata_id = authority.claim_by_property_id(WIKIDATA_QID)&.data_value
+        wikidata_uri = wikidata_id && "https://www.wikidata.org/wiki/#{wikidata_id}"
 
-          # Only one or the other of these seem to exist for a given item in practice.
-          display_props['QU'] = external_uri if external_uri
-          display_props['QU'] = wikidata_uri if wikidata_uri
-        end
+        # Only one or the other of these seem to exist for a given item in practice.
+        term['source_url'] = external_uri if external_uri
+        term['source_url'] = wikidata_uri if wikidata_uri
 
-        display_entries << display_props.to_json
+        linked_terms << term
       end
 
-      solr_props["#{prefix}_display"] = display_entries.uniq if requested_fields.include? 'display'
-      solr_props["#{prefix}_search"] = search_entries.uniq if requested_fields.include? 'search'
+      display_data['linked_terms'] = linked_terms.uniq if linked_terms.any?
+
+      solr_props["#{prefix}_display"] = [display_data.to_json] if requested_fields.include? 'display'
+      solr_props["#{prefix}_search"] = search_terms.uniq if requested_fields.include? 'search'
       solr_props["#{prefix}_facet"] = facets.uniq if requested_fields.include? 'facet'
 
       solr_props

diff --git a/lib/digital_scriptorium/name_claim_transformer.rb b/lib/digital_scriptorium/name_claim_transformer.rb
@@ -14,46 +14,47 @@ def self.transform(claim, export_hash)
       prefix = role_label.downcase.split.last
 
       recorded_name = claim.data_value
-      display_names = { 'PV' => recorded_name }
-      search_names = [recorded_name]
+      display_data = { 'recorded_value' => recorded_name }
+      search_terms = [recorded_name]
 
       name_in_original_script = claim.qualifier_by_property_id(IN_ORIGINAL_SCRIPT)&.data_value&.value
-      display_names['AGR'] = name_in_original_script if name_in_original_script
-      search_names << name_in_original_script if name_in_original_script
+      display_data['AGR'] = name_in_original_script if name_in_original_script
+      search_terms << name_in_original_script if name_in_original_script
 
       unless claim.qualifiers_by_property_id? NAME_IN_AUTHORITY_FILE
         return {
-          "#{prefix}_display" => [display_names.to_json],
-          "#{prefix}_search" => search_names,
+          "#{prefix}_display" => [display_data.to_json],
+          "#{prefix}_search" => search_terms,
           "#{prefix}_facet" => [recorded_name]
         }
       end
 
-      display_entries = []
       facets = []
+      linked_terms = []
 
       claim.qualifiers_by_property_id(NAME_IN_AUTHORITY_FILE).each do |qualifier|
-        display_names_for_qualifier = { 'PV' => recorded_name }
-        display_names_for_qualifier['AGR'] = name_in_original_script if name_in_original_script
+        term = {}
 
         name_entity_id = qualifier.entity_id_value
         name_item = export_hash[name_entity_id]
         name_label = name_item.label('en')
 
-        display_names_for_qualifier['QL'] = name_label
-        search_names << name_label
+        term['label'] = name_label
+        search_terms << name_label
         facets << name_label
 
         wikidata_id = name_item.claim_by_property_id(WIKIDATA_QID).data_value
         wikidata_url = "https://www.wikidata.org/wiki/#{wikidata_id}"
-        display_names_for_qualifier['QU'] = wikidata_url if wikidata_url
+        term['source_url'] = wikidata_url if wikidata_url
 
-        display_entries << display_names_for_qualifier.to_json
+        linked_terms << term
       end
 
+      display_data['linked_terms'] = linked_terms.uniq if linked_terms.any?
+
       {
-        "#{prefix}_display" => display_entries.uniq,
-        "#{prefix}_search" => search_names.uniq,
+        "#{prefix}_display" => [display_data.to_json],
+        "#{prefix}_search" => search_terms.uniq,
         "#{prefix}_facet" => facets.uniq
       }
     end

diff --git a/spec/digital_scriptorium/claim_transformer_spec.rb b/spec/digital_scriptorium/claim_transformer_spec.rb
@@ -47,7 +47,7 @@ module DigitalScriptorium
       solr_item = described_class.transform(id_claim, export_hash, config[PropertyId::DS_ID])
       expected = {
         'id' => ['DS121'],
-        'id_display' => ['{"PV":"DS121"}'],
+        'id_display' => ['{"recorded_value":"DS121"}'],
         'id_search' => ['DS121']
       }
       expect(solr_item).to eq(expected)
@@ -57,7 +57,7 @@ module DigitalScriptorium
       institution_claim = StatementRepresenter.new(Statement.new).from_json(institution_json)
       solr_item = described_class.transform(institution_claim, export_hash, config[PropertyId::HOLDING_INSTITUTION_AS_RECORDED])
       expected = {
-        'institution_display' => ['{"PV":"University of Pennsylvania","QL":"University of Pennsylvania","QU":"https://www.wikidata.org/wiki/Q49117"}'],
+        'institution_display' => ['{"recorded_value":"University of Pennsylvania","linked_terms":[{"label":"University of Pennsylvania","source_url":"https://www.wikidata.org/wiki/Q49117"}]}'],
         'institution_search' => ['University of Pennsylvania'],
         'institution_facet' => ['University of Pennsylvania']
       }
@@ -67,15 +67,15 @@ module DigitalScriptorium
     it 'transforms a holding status claim' do
       status_claim = StatementRepresenter.new(Statement.new).from_json(status_json)
       solr_item = described_class.transform(status_claim, export_hash, config[PropertyId::HOLDING_STATUS])
-      expected = { 'holding_status_display' => ['{"PV":"Current"}'] }
+      expected = { 'holding_status_display' => ['{"recorded_value":"Current"}'] }
       expect(solr_item).to eq(expected)
     end
 
     it 'transforms a shelfmark claim' do
       shelfmark_claim = StatementRepresenter.new(Statement.new).from_json(shelfmark_json)
       solr_item = described_class.transform(shelfmark_claim, export_hash, config[PropertyId::SHELFMARK])
       expected = {
-        'shelfmark_display' => ['{"PV":"Oversize LJS 110"}'],
+        'shelfmark_display' => ['{"recorded_value":"Oversize LJS 110"}'],
         'shelfmark_search' => ['Oversize LJS 110']
       }
       expect(solr_item).to eq(expected)
@@ -92,7 +92,7 @@ module DigitalScriptorium
       title_claim = StatementRepresenter.new(Statement.new).from_json(title_json)
       solr_item = described_class.transform(title_claim, export_hash, config[PropertyId::TITLE_AS_RECORDED])
       expected = {
-        'title_display' => ['{"PV":"Kitāb al-Majisṭī","QL":"Almagest"}'],
+        'title_display' => ['{"recorded_value":"Kitāb al-Majisṭī","linked_terms":[{"label":"Almagest"}]}'],
         'title_search' => ['Kitāb al-Majisṭī', 'Almagest'],
         'title_facet' => ['Almagest']
       }
@@ -103,7 +103,7 @@ module DigitalScriptorium
       genre_claim = StatementRepresenter.new(Statement.new).from_json(genre_json)
       solr_item = described_class.transform(genre_claim, export_hash, config[PropertyId::GENRE_AS_RECORDED])
       expected = {
-        'term_display' => ['{"PV":"Deeds","QL":"deeds","QU":"http://vocab.getty.edu/aat/300027249"}'],
+        'term_display' => ['{"recorded_value":"Deeds","linked_terms":[{"label":"deeds","source_url":"http://vocab.getty.edu/aat/300027249"}]}'],
         'term_search' => %w[Deeds deeds],
         'term_facet' => ['deeds']
       }
@@ -114,7 +114,7 @@ module DigitalScriptorium
       language_claim = StatementRepresenter.new(Statement.new).from_json(language_json)
       solr_item = described_class.transform(language_claim, export_hash, config[PropertyId::LANGUAGE_AS_RECORDED])
       expected = {
-        'language_display' => ['{"PV":"In Latin","QL":"Latin","QU":"https://www.wikidata.org/wiki/Q397"}'],
+        'language_display' => ['{"recorded_value":"In Latin","linked_terms":[{"label":"Latin","source_url":"https://www.wikidata.org/wiki/Q397"}]}'],
         'language_search' => ['In Latin', 'Latin'],
         'language_facet' => ['Latin']
       }
@@ -125,7 +125,7 @@ module DigitalScriptorium
       physical_description_claim = StatementRepresenter.new(Statement.new).from_json(physical_description_json)
       solr_item = described_class.transform(physical_description_claim, export_hash, config[PropertyId::PHYSICAL_DESCRIPTION])
       expected = {
-        'physical_description_display' => ['{"PV":"Extent: 1 parchment ; 170 x 245 mm."}'],
+        'physical_description_display' => ['{"recorded_value":"Extent: 1 parchment ; 170 x 245 mm."}'],
         'physical_description_search' => ['Extent: 1 parchment ; 170 x 245 mm.']
       }
       expect(solr_item).to eq(expected)
@@ -153,8 +153,7 @@ module DigitalScriptorium
       solr_item = described_class.transform(place_claim, export_hash, config[PropertyId::PRODUCTION_PLACE_AS_RECORDED])
       expected = {
         'place_display' => [
-          '{"PV":"[Provence or Spain],","QL":"Provence","QU":"http://vocab.getty.edu/tgn/7012209"}',
-          '{"PV":"[Provence or Spain],","QL":"Spain","QU":"http://vocab.getty.edu/tgn/1000095"}'
+          '{"recorded_value":"[Provence or Spain],","linked_terms":[{"label":"Provence","source_url":"http://vocab.getty.edu/tgn/7012209"},{"label":"Spain","source_url":"http://vocab.getty.edu/tgn/1000095"}]}'
         ],
         'place_search' => [
           '[Provence or Spain],',

diff --git a/spec/digital_scriptorium/date_claim_transformer_spec.rb b/spec/digital_scriptorium/date_claim_transformer_spec.rb
@@ -16,7 +16,7 @@ module DigitalScriptorium
       solr_item = described_class.transform(date_claim, export_hash, config[PropertyId::PRODUCTION_DATE_AS_RECORDED])
       expected = {
         'date_meta' => ['1358.'],
-        'date_display' => ['{"PV":"1358.","QL":"fourteenth century (dates CE)","QU":"http://vocab.getty.edu/aat/300404506"}'],
+        'date_display' => ['{"recorded_value":"1358.","linked_terms":[{"label":"fourteenth century (dates CE)","source_url":"http://vocab.getty.edu/aat/300404506"}]}'],
         'date_search' => ['1358.', 'fourteenth century (dates CE)'],
         'date_facet' => ['fourteenth century (dates CE)'],
         'century_int' => [1301],

diff --git a/spec/digital_scriptorium/name_claim_transformer_spec.rb b/spec/digital_scriptorium/name_claim_transformer_spec.rb
@@ -28,7 +28,7 @@ module DigitalScriptorium
     it 'transforms a name claim' do
       solr_item = described_class.transform(name_claim, export_hash)
       expected = {
-        'owner_display' => ['{"PV":"Schoenberg, Lawrence J","QL":"Lawrence J. Schoenberg","QU":"https://www.wikidata.org/wiki/Q107542788"}'],
+        'owner_display' => ['{"recorded_value":"Schoenberg, Lawrence J","linked_terms":[{"label":"Lawrence J. Schoenberg","source_url":"https://www.wikidata.org/wiki/Q107542788"}]}'],
         'owner_search' => ['Schoenberg, Lawrence J', 'Lawrence J. Schoenberg'],
         'owner_facet' => ['Lawrence J. Schoenberg']
       }
@@ -39,8 +39,7 @@ module DigitalScriptorium
       solr_item = described_class.transform(name_multiple_qualifier_values_claim, export_hash)
       expected = {
         'owner_display' => [
-          '{"PV":"From the codex made for Leonello d\'Este. Brought to Wales as war booty by 1813, already in a damaged state, by the Rolls family, later enobled as Barons Llangattock, of The Hendre, Monmouth (Llangattock sale, London, Christie\'s, 8 December 1958, lot 190);","QL":"Leonello d\'Este, Marquis of Ferrara","QU":"https://www.wikidata.org/wiki/Q1379797"}',
-          '{"PV":"From the codex made for Leonello d\'Este. Brought to Wales as war booty by 1813, already in a damaged state, by the Rolls family, later enobled as Barons Llangattock, of The Hendre, Monmouth (Llangattock sale, London, Christie\'s, 8 December 1958, lot 190);","QL":"Baron Llangattock","QU":"https://www.wikidata.org/wiki/Q4862572"}'
+          '{"recorded_value":"From the codex made for Leonello d\'Este. Brought to Wales as war booty by 1813, already in a damaged state, by the Rolls family, later enobled as Barons Llangattock, of The Hendre, Monmouth (Llangattock sale, London, Christie\'s, 8 December 1958, lot 190);","linked_terms":[{"label":"Leonello d\'Este, Marquis of Ferrara","source_url":"https://www.wikidata.org/wiki/Q1379797"},{"label":"Baron Llangattock","source_url":"https://www.wikidata.org/wiki/Q4862572"}]}'
         ],
         'owner_search' => [
           'From the codex made for Leonello d\'Este. Brought to Wales as war booty by 1813, already in a damaged state, by the Rolls family, later enobled as Barons Llangattock, of The Hendre, Monmouth (Llangattock sale, London, Christie\'s, 8 December 1958, lot 190);',