De-duplicate providers in case where e.g. content provider and host institution are the same org. Hide empty string address fields

Author: fbacall

lib/bioschemas/generator.rb

@@ -96,10 +96,10 @@ def self.address(event)
           'addressRegion' => event.county,
           'addressCountry' => event.country,
           'postalCode' => event.postcode
-        }.compact,
+        }.compact_blank,
         'latitude' => event.latitude,
         'longitude' => event.longitude
-      }.compact
+      }.compact_blank
     end
 
     # Reverse the process used by TeSS_RDF_Extractors to turn a list of values into a markdown list.
@@ -137,15 +137,6 @@ def self.external_resources(resource)
     def self.provider(resource)
       p = []
 
-      if resource.respond_to?(:host_institutions)
-        resource.host_institutions.each do |i|
-          p << {
-            '@type' => 'Organization',
-            'name' => i
-          }
-        end
-      end
-
       if resource.content_provider
         p << {
           '@type' => 'Organization',
@@ -162,6 +153,17 @@ def self.provider(resource)
         }
       end
 
+      if resource.respond_to?(:host_institutions)
+        resource.host_institutions.each do |i|
+          p << {
+            '@type' => 'Organization',
+            'name' => i
+          }
+        end
+      end
+
+      p.uniq! { |o| o['name'].downcase.strip }
+
       p.any? ? p : nil
     end
   end

test/models/event_test.rb

@@ -757,4 +757,22 @@ class EventTest < ActiveSupport::TestCase
     refute event.valid?
     assert event.errors.added?(:keywords, :too_long, count: 20)
   end
+
+  test 'blank address fields are removed from event bioschemas json' do
+    e = Event.new(city: 'Manchester', country: '', event_types: ['workshops_and_courses'])
+    bioschemas = e.to_bioschemas.first.generate
+    address = bioschemas[:hasCourseInstance].first[:location]['address']
+    assert_equal 'Manchester', address['addressLocality']
+    assert_equal ['@type', 'addressLocality'].sort, address.keys.sort
+  end
+
+  test 'de-duplicates provider in event bioschemas json' do
+    e = Event.new(content_provider: content_providers(:goblet), host_institutions: 'Goblet',
+                  event_types: ['workshops_and_courses'])
+
+    bioschemas = e.to_bioschemas.first.generate
+    assert_equal 1, bioschemas[:provider].length
+    assert_equal 'Goblet', bioschemas[:provider].first['name']
+    assert_equal 'http://mygoblet.org', bioschemas[:provider].first['url']
+  end
 end