Merge pull request #534 from MITLibraries/rdi-170-searching-multiple-…

…sources

OpenSearch and GraphQL for multiple source filters

app/graphql/types/query_type.rb

@@ -52,7 +52,7 @@ def record_id(id:)
         argument :format_facet, [String], required: false, default_value: nil
         argument :languages_facet, [String], required: false, default_value: nil
         argument :literary_form_facet, String, required: false, default_value: nil
-        argument :source_facet, String, required: false, default_value: 'All'
+        argument :source_facet, [String], required: false, default_value: nil
         argument :subjects_facet, [String], required: false, default_value: nil
       end
     else

app/models/opensearch.rb

@@ -4,7 +4,7 @@ class Opensearch
 
   def search(from, params, client)
     @params = params
-    client.search(index: ENV['ELASTICSEARCH_INDEX'],
+    client.search(index: ENV.fetch('ELASTICSEARCH_INDEX', nil),
                   body: build_query(from))
   end
 
@@ -114,7 +114,8 @@ def filters
 
     f.push filter_single(@params[:literary_form_facet], 'literary_form') if @params[:literary_form_facet]
 
-    f.push filter_single(@params[:source_facet], 'source') if @params[:source_facet]
+    f.push filter_sources(@params[:source_facet]) if @params[:source_facet]
+
     f.push filter(@params[:subjects_facet], 'subjects') if @params[:subjects_facet]
     f
   end
@@ -155,6 +156,26 @@ def filter_single(param, field)
     }
   end
 
+  def filter_sources(param)
+    {
+      bool: {
+        should: source_array(param)
+      }
+    }
+  end
+
+  def source_array(param)
+    sources = []
+    param.each do |source|
+      sources << {
+        term: {
+          source: source
+        }
+      }
+    end
+    sources
+  end
+
   # https://www.elastic.co/guide/en/elasticsearch/reference/current/search-aggregations-bucket-terms-aggregation.html
   def aggregations
     {

test/controllers/graphql_controller_v2_test.rb

@@ -314,4 +314,102 @@ def setup
     json = JSON.parse(response.body)
     assert(json['errors'].first['message'].present?)
   end
+
+  test 'graphqlv2 filter multiple sources' do
+    VCR.use_cassette('graphql v2 filter multiple sources') do
+
+      # no filters to return all sources. used later to test filters return less than the total.
+      post '/graphql', params: { query:
+        '{
+          search(searchterm: "data") {
+            hits
+            aggregations {
+              source {
+                key
+                docCount
+              }
+            }
+          }
+        }'
+      }
+
+      json = JSON.parse(response.body)
+      initial_source_array = json['data']['search']['aggregations']['source']
+
+      # filtering to 2 sources returns 2 sources
+      post '/graphql', params: { query:
+        '{
+          search(searchterm: "data", sourceFacet: ["Zenodo", "DSpace@MIT"]) {
+            hits
+            aggregations {
+              source {
+                key
+                docCount
+              }
+            }
+          }
+        }'
+      }
+      assert_equal(200, response.status)
+
+      json = JSON.parse(response.body)
+      filtered_source_array = json['data']['search']['aggregations']['source']
+
+      assert(initial_source_array.count > filtered_source_array.count)
+      assert_equal(2, filtered_source_array.count)
+
+      expected_sources = ['zenodo', 'dspace@mit']
+      actual_sources = filtered_source_array.map{|source| source["key"]}
+      assert_equal(expected_sources, actual_sources)
+    end
+  end
+
+  test 'graphqlv2 filter single source' do
+    VCR.use_cassette('graphql v2 filter single source') do
+
+      # no filters to return all sources. used later to test filters return less than the total.
+      post '/graphql', params: { query:
+        '{
+          search(searchterm: "data") {
+            hits
+            aggregations {
+              source {
+                key
+                docCount
+              }
+            }
+          }
+        }'
+      }
+
+      json = JSON.parse(response.body)
+      initial_source_array = json['data']['search']['aggregations']['source']
+
+      # filtering to 1 sources returns 1 source
+      post '/graphql', params: { query:
+        '{
+          search(searchterm: "data", sourceFacet: ["DSpace@MIT"]) {
+            hits
+            aggregations {
+              source {
+                key
+                docCount
+              }
+            }
+          }
+        }'
+      }
+      assert_equal(200, response.status)
+
+      json = JSON.parse(response.body)
+      filtered_source_array = json['data']['search']['aggregations']['source']
+
+      assert(initial_source_array.count > filtered_source_array.count)
+      assert_equal(1, filtered_source_array.count)
+
+      expected_sources = ['dspace@mit']
+      actual_sources = filtered_source_array.map{|source| source["key"]}
+      assert_equal(expected_sources, actual_sources)
+    end
+  end
 end

test/models/opensearch_test.rb

@@ -72,17 +72,32 @@ class OpensearchTest < ActiveSupport::TestCase
     VCR.use_cassette('opensearch single field nested') do
       params = { contributors: 'mcternan' }
       results = Opensearch.new.search(0, params, Timdex::OSClient)
-      assert_equal "A common table : 80 recipes and stories from my shared cultures /",
+      assert_equal 'A common table : 80 recipes and stories from my shared cultures /',
                    results['hits']['hits'].first['_source']['title']
     end
   end
 
   test 'searches multiple fields' do
     VCR.use_cassette('opensearch multiple fields') do
-      params = { q: 'chinese', title: 'common', contributors: 'mcternan'}
+      params = { q: 'chinese', title: 'common', contributors: 'mcternan' }
       results = Opensearch.new.search(0, params, Timdex::OSClient)
-      assert_equal "A common table : 80 recipes and stories from my shared cultures /",
+      assert_equal 'A common table : 80 recipes and stories from my shared cultures /',
                    results['hits']['hits'].first['_source']['title']
     end
   end
+
+  test 'source_array creates correct query structure' do
+    sources = ['Zenodo', 'DSpace@MIT']
+    expected = [{ term: { source: 'Zenodo' } }, { term: { source: 'DSpace@MIT' } }]
+
+    assert_equal(expected, Opensearch.new.source_array(sources))
+  end
+
+  test 'filter_sources creates correct query structure' do
+    sources = ['Zenodo', 'DSpace@MIT']
+    expected = { bool: { should: [{ term: { source: 'Zenodo' } },
+                                  { term: { source: 'DSpace@MIT' } }] } }
+
+    assert_equal(expected, Opensearch.new.filter_sources(sources))
+  end
 end