Merge pull request #832 from datacite/facetable-authors-error

Addresses error where some author aggs with multiple nameIdentifiers result in array index error

app/controllers/concerns/facetable.rb

@@ -408,8 +408,12 @@ def facet_by_fos(arr)
     end
 
     def facet_by_authors(arr)
-      arr.map do |hsh|
-        orcid_id = hsh["key"]
+      arr.map { |hsh|
+        orcid_id = %r{\A(?:(http|https)://(orcid.org)/)(.+)\z}.match?(hsh["key"]) && hsh["key"]
+
+        if orcid_id.nil?
+          next
+        end
 
         # The aggregation query should only return 1 hit, so hence the index
         # into first element
@@ -418,20 +422,20 @@ def facet_by_authors(arr)
         # Filter through creators to find creator that matches the key
         matched_creator = creators.select do |creator|
           if creator.key?("nameIdentifiers")
-            creator["nameIdentifiers"].each do |ni|
-              break ni["nameIdentifier"] == orcid_id
-            end
+            creator["nameIdentifiers"].any? { |ni| ni["nameIdentifier"] == orcid_id }
           end
         end
 
-        title = matched_creator[0]["name"]
+        if matched_creator.any?
+          title = matched_creator[0]["name"]
 
-        {
-          "id" => orcid_id,
-          "title" => title,
-          "count" => hsh["doc_count"],
-        }
-      end
+          {
+            "id" => orcid_id,
+            "title" => title,
+            "count" => hsh["doc_count"],
+          }
+        end
+      }.compact
     end
   end
 end

app/graphql/types/base_connection.rb

@@ -198,8 +198,12 @@ def facet_by_language(arr)
   end
 
   def facet_by_authors(arr)
-    arr.map do |hsh|
-      orcid_id = hsh["key"]
+    arr.map { |hsh|
+      orcid_id = %r{\A(?:(http|https)://(orcid.org)/)(.+)\z}.match?(hsh["key"]) && hsh["key"]
+
+      if orcid_id.nil?
+        next
+      end
 
       # The aggregation query should only return 1 hit, so hence the index
       # into first element
@@ -208,19 +212,19 @@ def facet_by_authors(arr)
       # Filter through creators to find creator that matches the key
       matched_creator = creators.select do |creator|
         if creator.key?("nameIdentifiers")
-          creator["nameIdentifiers"].each do |ni|
-            break ni["nameIdentifier"] == orcid_id
-          end
+          creator["nameIdentifiers"].any? { |ni| ni["nameIdentifier"] == orcid_id }
         end
       end
 
-      title = matched_creator[0]["name"]
+      if matched_creator.any?
+        title = matched_creator[0]["name"]
 
-      {
-        "id" => orcid_id,
-        "title" => title,
-        "count" => hsh["doc_count"],
-      }
-    end
+        {
+          "id" => orcid_id,
+          "title" => title,
+          "count" => hsh["doc_count"],
+        }
+      end
+    }.compact
   end
 end

spec/concerns/facetable_spec.rb

@@ -4,13 +4,62 @@
 
 describe "Facetable", type: :controller do
     let(:author_aggs) { JSON.parse(file_fixture("authors_aggs.json").read) }
+    let(:author_aggs_with_multiple_name_identifiers) { JSON.parse(file_fixture("authors_aggs_with_multiple_name_identifiers.json").read) }
     let(:model) { DataciteDoisController.new }
     it "facet by author" do
       authors = model.facet_by_authors(author_aggs)
 
       expected_result = [{ "id" => "https://orcid.org/0000-0003-1419-2405", "title" => "Fenner, Martin", "count" => 244 }, { "id" => "https://orcid.org/0000-0001-9570-8121", "title" => "Lambert, Simon", "count" => 23 }]
       expect(authors).to eq (expected_result)
     end
+
+    it "facet by author where author may have multiple nameIdentifiers" do
+      authors = model.facet_by_authors(author_aggs_with_multiple_name_identifiers)
+
+      expected_result = [
+        {
+            "id" => "https://orcid.org/0000-0002-0429-5446",
+            "title" => "Nam, Hyung-song",
+            "count" => 28,
+        },
+        {
+            "id" => "https://orcid.org/0000-0003-4973-3128",
+            "title" => "Casares, Ramón",
+            "count" => 12,
+        },
+        {
+            "id" => "https://orcid.org/0000-0002-3776-4755",
+            "title" => "Gomeseria, Ronald",
+            "count" => 4,
+        },
+        {
+            "id" => "https://orcid.org/0000-0002-6014-2161",
+            "title" => "Kartha, Sivan",
+            "count" => 4,
+        },
+        {
+            "id" => "https://orcid.org/0000-0003-1026-5865",
+            "title" => "Willemen, Louise",
+            "count" => 4,
+        },
+        {
+            "id" => "https://orcid.org/0000-0003-4624-488X",
+            "title" => "Schwarz, Nina",
+            "count" => 4,
+        },
+        {
+            "id" => "https://orcid.org/0000-0002-2149-9897",
+            "title" => "A, Subaveerapandiyan",
+            "count" => 3,
+        },
+        {
+            "id" => "https://orcid.org/0000-0002-4541-7294",
+            "title" => "Puntiroli, Michael",
+            "count" => 3,
+        },
+        ]
+      expect(authors).to eq (expected_result)
+    end
   end