Merge pull request #602 from scientist-softserv/uri-to-human-readable…

…-string

🎁 Convert URI to human readable strings

app/indexers/allinson_flex/dynamic_indexer_behavior.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+# OVERRIDE AllinsonFlex v0.1.0 to turn URI's into human readable strings
+#
+# For whatever reason our decorator pattern was not overriding #generate_solr_document
+# If you check AllinsonFlex::DynamicIndexerBehavior.ancestors, you get what you expect
+# with a decorator.  If you checked the #source_location of :generate_solr_document
+# you also get what you expect.  However, when running a reindex, I was not going
+# through the decorator.  I decided since this is only one method, overriding the entire
+# mixin was acceptable for now.
+
+module AllinsonFlex
+  module DynamicIndexerBehavior
+    include UriToStringBehavior
+    extend ActiveSupport::Concern
+
+    RANGE = "http://www.w3.org/2001/XMLSchema#anyURI"
+
+    included do
+      class_attribute :model_class
+    end
+
+    def generate_solr_document
+      dynamic_schema_service = object.dynamic_schema_service
+      uri_properties = uri_properties_from(dynamic_schema_service)
+
+      super.tap do |solr_doc|
+        dynamic_schema_service.indexing_properties.each_pair do |prop_name, index_field_names|
+          value = if uri_properties.include?(prop_name.to_s)
+                    uri_to_value_for(object.send(prop_name))
+                  else
+                    object.send(prop_name)
+                  end
+
+          index_field_names.each { |index_field_name| solr_doc[index_field_name] = value }
+        end
+      end
+    end
+
+    private
+
+      def uri_properties_from(dynamic_schema_service)
+        schema = HashWithIndifferentAccess.new(dynamic_schema_service.dynamic_schema.schema)
+        props_hash = schema[:properties]
+        # remove rdf_type since they are also URIs but not for our purposes
+        props_hash.keys.select { |k, _v| props_hash[k][:range] == RANGE } - local_authorities - ['rdf_type']
+      end
+
+      def local_authorities
+        # Hyku has these pluralized while m3 has these singularized
+        Qa::Authorities::Local.names.map(&:singularize)
+      end
+  end
+end

app/indexers/app_indexer.rb

@@ -9,6 +9,8 @@ class AppIndexer < Hyrax::WorkIndexer
   # Utk does not include based_near and does not need deep indexing.
   # include Hyrax::IndexesLinkedMetadata
 
+  include AllinsonFlex::DynamicIndexerBehavior
+
   # Uncomment this block if you want to add custom indexing behavior:
   def generate_solr_document
     super.tap do |solr_doc|
@@ -24,7 +26,6 @@ def generate_solr_document
   private
 
     def all_creators
-      props = SolrDocument.creator_fields
-      props.map { |prop| Array(object.try(prop)) }.flatten.compact
+      SolrDocument.creator_fields.map { |prop| uri_to_value_for(object.try(prop)) }.flatten.compact
     end
 end

app/indexers/attachment_indexer.rb

@@ -9,6 +9,5 @@ class AttachmentIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Attachment
 end

app/indexers/audio_indexer.rb

@@ -9,6 +9,5 @@ class AudioIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Audio
 end

app/indexers/book_indexer.rb

@@ -9,6 +9,5 @@ class BookIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Book
 end

app/indexers/compound_object_indexer.rb

@@ -9,6 +9,5 @@ class CompoundObjectIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::CompoundObject
 end

app/indexers/generic_work_indexer.rb

@@ -9,6 +9,5 @@ class GenericWorkIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::GenericWork
 end

app/indexers/image_indexer.rb

@@ -9,6 +9,5 @@ class ImageIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Image
 end

app/indexers/newspaper_indexer.rb

@@ -9,6 +9,5 @@ class NewspaperIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Newspaper
 end

app/indexers/pdf_indexer.rb

@@ -9,6 +9,5 @@ class PdfIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Pdf
 end

app/indexers/uri_to_string_behavior.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module UriToStringBehavior
+  extend ActiveSupport::Concern
+
+  # UTK uses this label to house the value that needs to be rendered
+  LABEL = "http://www.w3.org/2004/02/skos/core#prefLabel"
+
+  # Retrieves a value for a given URI.
+  #
+  # @param value [String] the value to retrieve. If this value starts with 'http', it is treated as a URI.
+  # @return [String]
+  #
+  # @example
+  #   uri_to_value_for('http://example.com') #=> "Failed to load RDF data: ..."
+  #   uri_to_value_for('http://id.loc.gov/authorities/names/n2017180154') #=> "University of Tennessee"
+  #   uri_to_value_for('Doe, John') #=> "Doe, John"
+  def uri_to_value_for(value)
+    return value.map { |v| uri_to_value_for(v) } if value.is_a?(Enumerable)
+    return if value.blank?
+    return value unless value.is_a?(String)
+    return value unless value.start_with?('http')
+
+    uri = value
+
+    begin
+      graph = RDF::Graph.load(uri)
+    rescue StandardError => e
+      Rails.logger.error("Failed to load RDF data: #{e.message}")
+      return "#{uri} (Failed to load URI)"
+    end
+
+    subject = RDF::URI.new(uri)
+    predicate = RDF::URI.new(LABEL)
+    object = graph.query([subject, predicate, nil]).objects.first
+    return "#{uri} (No label found)" if object.blank?
+
+    object.to_s
+  end
+end

app/indexers/video_indexer.rb

@@ -9,6 +9,5 @@ class VideoIndexer < AppIndexer
   #    solr_doc['my_custom_field_ssim'] = object.my_custom_property
   #  end
   # end
-  include AllinsonFlex::DynamicIndexerBehavior
   self.model_class = ::Video
 end