Adds support for querying specific indexes

Why are these changes being introduced:

* We need to be able to provide a UI that only queries a specific set
  of data, so we need to enable that in the API

Relevant ticket(s):

* https://mitlibraries.atlassian.net/browse/RDI-208

How does this address that need:

* Adds `index` to both Search and Retrieve GraphQL endpoints and
  documents them as being best left out
* Updates Opensearch and Retrieve models to use any index value provided
  and default to the value provided in ENV if none is provided

Document any side effects to this change:

* The Retrieve model had previously been the same for v1 and v2 but this
  change required a v2 feature flag for Retrieve to ensure the current
  behavior continued to work

app/graphql/types/query_type.rb

@@ -9,9 +9,18 @@ def ping
       'Pong!'
     end
 
-    field :record_id, RecordType, null: false,
-                                  description: 'Retrieve one timdex record' do
-      argument :id, String, required: true
+    if Flipflop.v2?
+      field :record_id, RecordType, null: false,
+                                    description: 'Retrieve one timdex record' do
+        argument :id, String, required: true
+        argument :index, String, required: false, default_value: nil,
+                                 description: 'It is not recommended to provide an index value unless we have provided you with one for your specific use case'
+      end
+    else
+      field :record_id, RecordType, null: false,
+                                    description: 'Retrieve one timdex record' do
+        argument :id, String, required: true
+      end
     end
 
     if Flipflop.v2?
@@ -26,8 +35,8 @@ def info
     end
 
     if Flipflop.v2?
-      def record_id(id:)
-        result = Retrieve.new.fetch(id, Timdex::OSClient)
+      def record_id(id:, index:)
+        result = Retrieve.new.fetch(id, Timdex::OSClient, index)
         result['hits']['hits'].first['_source']
       rescue Elasticsearch::Transport::Transport::Errors::NotFound
         raise GraphQL::ExecutionError, "Record '#{id}' not found"
@@ -44,6 +53,8 @@ def record_id(id:)
         argument :subjects, String, required: false, default_value: nil
         argument :title, String, required: false, default_value: nil
         argument :from, String, required: false, default_value: '0'
+        argument :index, String, required: false, default_value: nil,
+                                 description: 'It is not recommended to provide an index value unless we have provided you with one for your specific use case'
 
         # applied facets
         argument :collection_facet, [String], required: false, default_value: nil
@@ -81,11 +92,11 @@ def record_id(id:)
 
     if Flipflop.v2?
       def search(searchterm:, citation:, contributors:, funding_information:, identifiers:, locations:, subjects:,
-                 title:, from:, **facets)
+                 title:, index:, from:, **facets)
         query = construct_query(searchterm, citation, contributors, funding_information, identifiers, locations,
                                 subjects, title, facets)
 
-        results = Opensearch.new.search(from, query, Timdex::OSClient)
+        results = Opensearch.new.search(from, query, Timdex::OSClient, index)
 
         response = {}
         response[:hits] = results['hits']['total']['value']

app/models/opensearch.rb

@@ -2,12 +2,17 @@ class Opensearch
   SIZE = 20
   MAX_PAGE = 200
 
-  def search(from, params, client)
+  def search(from, params, client, index = nil)
     @params = params
-    client.search(index: ENV.fetch('ELASTICSEARCH_INDEX', nil),
+    index = default_index unless index.present?
+    client.search(index: index,
                   body: build_query(from))
   end
 
+  def default_index
+    ENV.fetch('ELASTICSEARCH_INDEX', nil)
+  end
+
   # Construct the json query to send to elasticsearch
   def build_query(from)
     {

app/models/retrieve.rb

@@ -1,17 +1,24 @@
 class Retrieve
-  def fetch(id, client)
+  def fetch(id, client, index = nil)
     f = to_filter(id)
-    record = client.search(index: ENV['ELASTICSEARCH_INDEX'], body: f)
+
+    index = default_index unless index.present?
+
+    record = client.search(index: index, body: f)
 
     if client.instance_of?(OpenSearch::Client)
       raise OpenSearch::Transport::Transport::Errors::NotFound if record['hits']['total']['value'].zero?
-    else
-      raise Elasticsearch::Transport::Transport::Errors::NotFound if record['hits']['total'].zero?
+    elsif record['hits']['total'].zero?
+      raise Elasticsearch::Transport::Transport::Errors::NotFound
     end
 
     record
   end
 
+  def default_index
+    ENV.fetch('ELASTICSEARCH_INDEX', nil)
+  end
+
   def to_filter(id)
     {
       query: {

test/controllers/graphql_controller_v2_test.rb

@@ -153,7 +153,7 @@ def setup
                                 }' }
       assert_equal(200, response.status)
       json = JSON.parse(response.body)
-      assert json['data']['search']['records'].first['contributors'].any? { |c| c.has_value? 'Moon, Intae' }
+      assert json['data']['search']['records'].first['contributors'].any? { |c| c.value? 'Moon, Intae' }
     end
   end
 
@@ -177,9 +177,9 @@ def setup
       assert_equal 'A common table : 80 recipes and stories from my shared cultures /',
                    json['data']['search']['records'].first['title']
       assert json['data']['search']['records'].first['contributors'].any? { |c|
-               c.has_value? 'McTernan, Cynthia Chen, author.'
+               c.value? 'McTernan, Cynthia Chen, author.'
              }
-      assert json['data']['search']['records'].first['identifiers'].any? { |i| i.has_value? '163565002X (hardback)' }
+      assert json['data']['search']['records'].first['identifiers'].any? { |i| i.value? '163565002X (hardback)' }
     end
   end
 
@@ -317,7 +317,6 @@ def setup
 
   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:
         '{
@@ -330,8 +329,7 @@ def setup
               }
             }
           }
-        }'
-      }
+        }' }
 
       json = JSON.parse(response.body)
       initial_source_array = json['data']['search']['aggregations']['source']
@@ -348,8 +346,7 @@ def setup
               }
             }
           }
-        }'
-      }
+        }' }
       assert_equal(200, response.status)
 
       json = JSON.parse(response.body)
@@ -359,14 +356,13 @@ def setup
       assert_equal(2, filtered_source_array.count)
 
       expected_sources = ['zenodo', 'dspace@mit']
-      actual_sources = filtered_source_array.map{|source| source["key"]}
+      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:
         '{
@@ -379,8 +375,7 @@ def setup
               }
             }
           }
-        }'
-      }
+        }' }
 
       json = JSON.parse(response.body)
       initial_source_array = json['data']['search']['aggregations']['source']
@@ -397,8 +392,7 @@ def setup
               }
             }
           }
-        }'
-      }
+        }' }
       assert_equal(200, response.status)
 
       json = JSON.parse(response.body)
@@ -408,8 +402,40 @@ def setup
       assert_equal(1, filtered_source_array.count)
 
       expected_sources = ['dspace@mit']
-      actual_sources = filtered_source_array.map{|source| source["key"]}
+      actual_sources = filtered_source_array.map { |source| source['key'] }
       assert_equal(expected_sources, actual_sources)
     end
   end
+
+  test 'graphqlv2 can retrieve a record from a default index' do
+    # fragile test: specific item expected in default index
+    VCR.use_cassette('graphql v2 retrieve from default index') do
+      post '/graphql', params: { query:
+        '{
+          recordId(id: "dspace:1721.1-44968") {
+            timdexRecordId
+            title
+          }
+        }' }
+
+      json = JSON.parse(response.body)
+      assert_equal('dspace:1721.1-44968', json['data']['recordId']['timdexRecordId'])
+    end
+  end
+
+  test 'graphqlv2 can retrive a record from a specified index' do
+    # fragile test: specific item expected in specified index
+    VCR.use_cassette('graphql v2 retrieve from rdi* index') do
+      post '/graphql', params: { query:
+        '{
+          recordId(id: "zenodo:5728409", index: "rdi*") {
+            timdexRecordId
+            title
+          }
+        }' }
+
+      json = JSON.parse(response.body)
+      assert_equal('zenodo:5728409', json['data']['recordId']['timdexRecordId'])
+    end
+  end
 end

test/models/opensearch_test.rb

@@ -59,6 +59,26 @@ class OpensearchTest < ActiveSupport::TestCase
     assert matches.select { |m| m['subjects.value'] == 'assured' }
   end
 
+  test 'can override index' do
+    # fragile test: assumes opensearch instance with at least one index prefixed with `rdi`
+    VCR.use_cassette('opensearch non-default index') do
+      params = { title: 'data' }
+      results = Opensearch.new.search(0, params, Timdex::OSClient, 'rdi*')
+      assert results['hits']['hits'].map { |hit| hit['_index'] }.uniq.map { |index| index.start_with?('rdi') }.any?
+    end
+  end
+
+  test 'default index' do
+    # fragile test: assumes opensearch instance with at least one index promoted to timdex-prod and no promoted indexes
+    # that start with rdi*
+    VCR.use_cassette('opensearch default index') do
+      params = { title: 'data' }
+      results = Opensearch.new.search(0, params, Timdex::OSClient)
+      refute results['hits']['hits'].map { |hit| hit['_index'] }.uniq.map { |index| index.start_with?('rdi') }.any?
+      assert results['hits']['hits'].map { |hit| hit['_index'] }.uniq.any?
+    end
+  end
+
   test 'searches a single field' do
     VCR.use_cassette('opensearch single field') do
       params = { title: 'spice' }