Merge 82add09 into 94f720f

app/services/qa/linked_data/deep_sort_service.rb

@@ -91,9 +91,9 @@ def downcase_string(idx = 0)
           to_downcase(literal(idx))
         end
 
-        def language(literal = literals.first)
-          return literal.language if literal.respond_to?(:language)
-          nil
+        def language(lit = literals.first)
+          return nil unless Qa::LinkedData::LanguageService.literal_has_language_marker? lit
+          lit.language
         end
 
         def includes_preferred_language?
@@ -206,8 +206,7 @@ def only_one_has_language_marker?(lit, other_lit)
           end
 
           def language?(lit)
-            lang = lit.language if lit.respond_to?(:language)
-            lang.present?
+            Qa::LinkedData::LanguageService.literal_has_language_marker? lit
           end
 
           def preferred_language?(lang)

app/services/qa/linked_data/graph_service.rb

@@ -14,17 +14,16 @@ def load_graph(url:)
 
         # Create a new graph with statements filtered out
         # @param graph [RDF::Graph] the original graph to be filtered
-        # @param language [String | Symbol | Array<String|Symbol>] will keep any statement whose object's language matches the language filter
-        #          (only applies to statements that respond to language) (e.g. "en" or :en or ["en", "fr"] or [:en, :fr])
+        # @param language [Array<Symbol>] will keep any statement whose object's language matches the language filter
+        #          (only applies to statements that respond to language) (e.g. [:fr] or [:en, :fr])
         # @param remove_blanknode_subjects [Boolean] will remove any statement whose subject is a blanknode, if true
         # @return [RDF::Graph] a new instance of graph with statements not matching the filters removed
         def filter(graph:, language: nil, remove_blanknode_subjects: false)
           return graph unless graph.present?
           return graph unless language.present? || remove_blanknode_subjects
-          filtered_graph = graph
-          language = normalize_language(language)
-          filtered_graph.each do |st|
-            filtered_graph.delete(st) if filter_out_blanknode(remove_blanknode_subjects, st.subject) || filter_out_language(language, st.object)
+          filtered_graph = deep_copy(graph: graph)
+          filtered_graph.statements.each do |st|
+            filtered_graph.delete(st) if filter_out_blanknode(remove_blanknode_subjects, st.subject) || filter_out_language(graph, language, st)
           end
           filtered_graph
         end
@@ -42,17 +41,39 @@ def object_values(graph:, subject:, predicate:)
           values
         end
 
+        def deep_copy(graph:)
+          new_graph = RDF::Graph.new
+          graph.statements.each do |st|
+            new_graph.insert(st.dup)
+          end
+        end
+
         private
 
           def filter_out_blanknode(remove, subj)
             remove && subj.anonymous?
           end
 
-          def filter_out_language(language, obj)
+          # Filter out language based on...
+          # * do not remove if the object literal does not respond to :language
+          # * do not remove if the object literal does not have a language marker
+          # * do not remove if the object has the targeted language marker
+          # * do not remove if none of the other objects with this statement's predicate have the targeted language marker
+          def filter_out_language(graph, language, statement)
             return false if language.blank?
-            return false unless obj.respond_to?(:language)
-            return false if obj.language.blank?
-            !language.include?(obj.language)
+            return false unless Qa::LinkedData::LanguageService.literal_has_language_marker?(statement.object)
+            objects = object_values(graph: graph, subject: statement.subject, predicate: statement.predicate)
+            return false unless at_least_one_object_has_language?(objects, language)
+            !language.include?(statement.object.language)
+          end
+
+          def at_least_one_object_has_language?(objects, language)
+            objects.each do |obj|
+              next unless Qa::LinkedData::LanguageService.literal_has_language_marker?(obj)
+              next unless language.include? obj.language
+              return true
+            end
+            false
           end
 
           def process_error(e, url)
@@ -78,15 +99,6 @@ def ioerror_code(e)
             z = msg.size - 2
             msg[a..z]
           end
-
-          # Normalize language
-          # @param [String | Symbol | Array] language for filtering graph (e.g. "en" OR :en OR ["en", "fr"] OR [:en, :fr])
-          # @return [Array<Symbol>] an array of languages encoded as symbols (e.g. [:en] OR [:en, :fr])
-          def normalize_language(language)
-            return language if language.blank?
-            language = [language] unless language.is_a? Array
-            language.map(&:to_sym)
-          end
       end
     end
   end

app/services/qa/linked_data/language_service.rb

@@ -0,0 +1,30 @@
+# Service to determine which language to use for sorting and filtering.
+module Qa
+  module LinkedData
+    class LanguageService
+      class << self
+        def preferred_language(user_language: nil, authority_language: nil)
+          return normalize_language(user_language) if user_language.present?
+          return normalize_language(authority_language) if authority_language.present?
+          normalize_language(Qa.config.default_language)
+        end
+
+        def literal_has_language_marker?(literal)
+          return false unless literal.respond_to?(:language)
+          literal.language.present?
+        end
+
+        private
+
+          # Normalize language
+          # @param [String | Symbol | Array] language for filtering graph (e.g. "en" OR :en OR ["en", "fr"] OR [:en, :fr])
+          # @return [Array<Symbol>] an array of languages encoded as symbols (e.g. [:en] OR [:en, :fr])
+          def normalize_language(language)
+            return language if language.blank?
+            language = [language] unless language.is_a? Array
+            language.map(&:to_sym)
+          end
+      end
+    end
+  end
+end

app/services/qa/linked_data/language_sort_service.rb

@@ -39,8 +39,8 @@ def construct_sorted_literals
         end
 
         def language(literal)
-          language = literal.language if literal.respond_to?(:language)
-          language.present? ? language : LANGUAGE_LOCALE_KEY_FOR_NO_LANGUAGE
+          return LANGUAGE_LOCALE_KEY_FOR_NO_LANGUAGE unless Qa::LinkedData::LanguageService.literal_has_language_marker? literal
+          literal.language
         end
 
         def move_no_language_to_end

lib/generators/qa/install/templates/config/initializers/qa.rb

@@ -10,4 +10,8 @@
   # requiring a restart of rails. By default, reloading through the browser is not allowed
   # when the token is nil or blank.  Change to any string to control who has access to reload.
   # config.authorized_reload_token = 'YOUR_AUTH_TOKEN_DEFINED_HERE'
+
+  # For linked data access, specify default language for sorting and selection.  The default is only used if a language is not
+  # specified in the authority's configuration file and not passed in as a parameter.  (e.g. :en, [:en], or [:en, :fr])
+  # config.default_language = :en
 end

lib/qa/authorities/linked_data/find_term.rb

@@ -10,7 +10,8 @@ def initialize(term_config)
         @term_config = term_config
       end
 
-      attr_reader :term_config, :graph
+      attr_reader :term_config, :full_graph, :filtered_graph, :language
+      private :full_graph, :filtered_graph, :language
 
       delegate :term_subauthority?, to: :term_config
 
@@ -35,24 +36,27 @@ def initialize(term_config)
       #     "http://schema.org/sameAs":["http://id.loc.gov/authorities/names/n79021621","https://viaf.org/viaf/126293486"] } }
       def find(id, language: nil, replacements: {}, subauth: nil, jsonld: false)
         raise Qa::InvalidLinkedDataAuthority, "Unable to initialize linked data term sub-authority #{subauth}" unless subauth.nil? || term_subauthority?(subauth)
-        language ||= term_config.term_language
+        @language = Qa::LinkedData::LanguageService.preferred_language(user_language: language, authority_language: term_config.term_language)
         url = Qa::LinkedData::AuthorityUrlService.build_url(action_config: term_config, action: :term, action_request: id, substitutions: replacements, subauthority: subauth)
         Rails.logger.info "QA Linked Data term url: #{url}"
-        load_graph(url: url, language: language)
-        return "{}" unless graph.size.positive?
-        return graph.dump(:jsonld, standard_prefixes: true) if jsonld
+        load_graph(url: url)
+        return "{}" unless full_graph.size.positive?
+        return full_graph.dump(:jsonld, standard_prefixes: true) if jsonld
         parse_term_authority_response(id)
       end
 
       private
 
-        def load_graph(url:, language:)
-          @graph = Qa::LinkedData::GraphService.load_graph(url: url)
-          @graph = Qa::LinkedData::GraphService.filter(graph: @graph, language: language) unless language.blank?
+        def load_graph(url:)
+          # @graph = Qa::LinkedData::GraphService.load_graph(url: url)
+          @full_graph = Qa::LinkedData::GraphService.load_graph(url: url)
+          return unless @full_graph.size.positive?
+          @filtered_graph = Qa::LinkedData::GraphService.deep_copy(graph: @full_graph)
+          @filtered_graph = Qa::LinkedData::GraphService.filter(graph: @filtered_graph, language: language) unless language.blank?
         end
 
         def parse_term_authority_response(id)
-          results = extract_preds(graph, preds_for_term)
+          results = extract_preds(filtered_graph, preds_for_term)
           consolidated_results = consolidate_term_results(results)
           json_results = convert_term_to_json(consolidated_results)
           termhash = select_json_result_for_id(json_results, id)
@@ -130,7 +134,7 @@ def select_json_result_for_id(json_results, id)
 
         def predicates_with_subject_uri(expected_uri) # rubocop:disable Metrics/MethodLength
           predicates_hash = {}
-          graph.statements.each do |st|
+          @full_graph.statements.each do |st|
             subj = st.subject.to_s
             next unless subj == expected_uri
             pred = st.predicate.to_s

lib/qa/authorities/linked_data/search_query.rb

@@ -11,7 +11,7 @@ def initialize(search_config)
       end
 
       attr_reader :search_config, :graph, :language
-      private :language
+      private :graph, :language
 
       delegate :subauthority?, :supports_sort?, to: :search_config
 
@@ -27,8 +27,7 @@ def initialize(search_config)
       #     {"uri":"http://id.worldcat.org/fast/409667","id":"409667","label":"Cornell, Ezra, 1807-1874"} ]
       def search(query, language: nil, replacements: {}, subauth: nil)
         raise Qa::InvalidLinkedDataAuthority, "Unable to initialize linked data search sub-authority #{subauth}" unless subauth.nil? || subauthority?(subauth)
-        language ||= search_config.language
-        @language = language
+        @language = Qa::LinkedData::LanguageService.preferred_language(user_language: language, authority_language: search_config.language)
         url = Qa::LinkedData::AuthorityUrlService.build_url(action_config: search_config, action: :search, action_request: query, substitutions: replacements, subauthority: subauth)
         Rails.logger.info "QA Linked Data search url: #{url}"
         load_graph(url: url)

lib/qa/configuration.rb

@@ -30,5 +30,12 @@ def valid_authority_reload_token?(token)
 
     # Hold linked data authority configs
     attr_accessor :linked_data_authority_configs
+
+    # For linked data access, specify default language for sorting and selection.  The default is only used if a language is not
+    # specified in the authority's configuration file and not passed in as a parameter.  (e.g. :en, [:en], or [:en, :fr])
+    attr_writer :default_language
+    def default_language
+      @default_language ||= :en
+    end
   end
 end

spec/fixtures/authorities/linked_data/lod_lang_defaults.json

@@ -18,7 +18,7 @@
       "term_id": "term_id"
     },
     "term_id": "ID",
-    "language": [ "en" ],
+    "language": [ "fr" ],
     "results": {
       "id_predicate":       "http://id.loc.gov/vocabulary/identifiers/lccn",
       "label_predicate":    "http://www.w3.org/2004/02/skos/core#prefLabel",
@@ -43,7 +43,7 @@
     "qa_replacement_patterns": {
       "query": "query"
     },
-    "language": [ "en" ],
+    "language": [ "fr" ],
     "results": {
       "id_predicate":       "http://purl.org/dc/terms/identifier",
       "label_predicate":    "http://www.w3.org/2004/02/skos/core#prefLabel",

spec/fixtures/lod_lang_search_filtering.nt

@@ -0,0 +1,11 @@
+<http://id.worldcat.org/fast/530369> <http://purl.org/dc/terms/identifier> "530369" .
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#prefLabel> "buttermilk"@en
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#prefLabel> "Babeurre"@fr
+<http://id.worldcat.org/fast/530369> <http://www.w3.org/2004/02/skos/core#prefLabel> "Buttermilch"
+<http://id.worldcat.org/fast/5140> <http://purl.org/dc/terms/identifier> "5140" .
+<http://id.worldcat.org/fast/5140> <http://www.w3.org/2004/02/skos/core#prefLabel> "dried milk"@en
+<http://id.worldcat.org/fast/5140> <http://www.w3.org/2004/02/skos/core#prefLabel> "lait en poudre"@fr
+<http://id.worldcat.org/fast/5140> <http://www.w3.org/2004/02/skos/core#prefLabel> "getrocknete Milch"@de
+<http://id.worldcat.org/fast/557490> <http://purl.org/dc/terms/identifier> "557490" .
+<http://id.worldcat.org/fast/557490> <http://www.w3.org/2004/02/skos/core#prefLabel> "condensed milk"@en
+<http://id.worldcat.org/fast/557490> <http://www.w3.org/2004/02/skos/core#prefLabel> "Kondensmilch"@de