block · jwils · Jul 1, 2026 · Jun 30, 2026 · Jul 1, 2026
@@ -138,7 +138,7 @@ def fetch_artifact_configuration(schema_artifacts, index_def_name)
             end
           end
 
-          context "when a concrete index has been derived from the template", :factories do
+          context "when a concrete index has been derived from the template", :factories, :ingests_json_data do
             include ConcreteIndexAdapter
 
             # our schema in the tests here is more limited than the main widget schema, so select only some fields.

@@ -38,7 +38,7 @@ def simulate_presence_of_extra_setting(admin, index_definition_name, name, value
         end
       end
 
-      RSpec.shared_examples_for IndexDefinitionConfigurator, :uses_datastore, :builds_indexer do
+      RSpec.shared_examples_for IndexDefinitionConfigurator, :uses_datastore do
         let(:output_io) { StringIO.new }
         let(:clock) { class_double(::Time, now: ::Time.utc(2024, 3, 20, 12, 0, 0)) }
         let(:mapping_removal_note_snippet) { "extra fields listed here will not actually get removed" }

@@ -15,7 +15,7 @@
 module ElasticGraph
   class DatastoreCore
     module IndexDefinition
-      RSpec.describe RolloverIndexTemplate, :uses_datastore, :builds_indexer do
+      RSpec.describe RolloverIndexTemplate, :ingests_json_data do
         # Use different index names than any other tests use, because most tests expect a specific index
         # configuration (based on `config/schema.graphql`) and we do not want to mess with it here.
         let(:index_prefix) { unique_index_name }

@@ -19,7 +19,13 @@ class << self
 
   self.camel_case_cluster_configured = false
 
-  RSpec.shared_context "ElasticGraph GraphQL acceptance support", :factories, :uses_datastore, :capture_logs, :builds_indexer, :builds_admin do
+  # We list the `:ingests_json_data` component tags (`:json_ingestion_schema_definition`,
+  # `:uses_datastore`, `:builds_indexer`) individually rather than using `:ingests_json_data`
+  # itself. Applying the derived `:ingests_json_data` tag to a `shared_context` and then combining
+  # it with `include_examples` crashes RSpec (`undefined method 'key?' for nil`), whereas literal
+  # tags are inherited correctly. Once https://github.com/rspec/rspec/issues/333 is fixed we can
+  # collapse these back to `:ingests_json_data`.
+  RSpec.shared_context "ElasticGraph GraphQL acceptance support", :factories, :json_ingestion_schema_definition, :uses_datastore, :builds_indexer, :capture_logs, :builds_admin do
     include GraphQLSupport
     include PreventSearchesFromUsingWriteRequests
 

@@ -8,12 +8,11 @@
 
 require "elastic_graph/graphql"
 require "elastic_graph/indexer"
-require "elastic_graph/json_ingestion/schema_definition/api_extension"
 require "elastic_graph/schema_definition/rake_tasks"
 require "support/graphql"
 
 module ElasticGraph
-  RSpec.describe "Querying an evolving schema", :uses_datastore, :factories, :capture_logs, :in_temp_dir, :rake_task do
+  RSpec.describe "Querying an evolving schema", :ingests_json_data, :factories, :capture_logs, :in_temp_dir, :rake_task do
     include GraphQLSupport
 
     let(:path_to_schema) { "config/schema.rb" }
@@ -66,7 +65,7 @@ def dump_schema_artifacts(json_schema_version:, team_extras: "")
       run_rake "schema_artifacts:dump" do |output|
         SchemaDefinition::RakeTasks.new(
           schema_element_name_form: :snake_case,
-          extension_modules: [JSONIngestion::SchemaDefinition::APIExtension],
+          extension_modules: json_ingestion_schema_definition_extension_modules,
           index_document_sizes: true,
           path_to_schema: path_to_schema,
           schema_artifacts_directory: "config/schema/artifacts",

@@ -18,7 +18,14 @@ class GraphQL
     # triggered a `null_pointer_exception` in one shard, which caused the datastore to
     # return a partial response that our router handled by logging a warning. The
     # `:capture_logs` here caused the tests to fail until we fixed that issue.
-    RSpec.shared_context "DatastoreQueryIntegrationSupport", :uses_datastore, :factories, :capture_logs do
+    #
+    # We list the `:ingests_json_data` component tags (`:json_ingestion_schema_definition`,
+    # `:uses_datastore`, `:builds_indexer`) individually rather than using `:ingests_json_data`
+    # itself. Applying the derived `:ingests_json_data` tag to a `shared_context` and then combining
+    # it with `include_examples` crashes RSpec (`undefined method 'key?' for nil`), whereas literal
+    # tags are inherited correctly. Once https://github.com/rspec/rspec/issues/333 is fixed we can
+    # collapse these back to `:ingests_json_data`.
+    RSpec.shared_context "DatastoreQueryIntegrationSupport", :json_ingestion_schema_definition, :uses_datastore, :builds_indexer, :factories, :capture_logs do
       include AggregationsHelpers
       include SortSupport
 

@@ -11,7 +11,7 @@
 module ElasticGraph
   class GraphQL
     module Resolvers
-      RSpec.describe IndexedTypeRootFieldsResolver, :factories, :uses_datastore, :resolver do
+      RSpec.describe IndexedTypeRootFieldsResolver, :factories, :ingests_json_data, :resolver do
         context "when the field being resolved is a relay connection field" do
           let(:graphql) { build_graphql }
 

@@ -15,7 +15,7 @@
 module ElasticGraph
   class GraphQL
     module Resolvers
-      RSpec.describe NestedRelationshipsSource, :factories, :uses_datastore, :capture_logs do
+      RSpec.describe NestedRelationshipsSource, :factories, :ingests_json_data, :capture_logs do
         include AggregationsHelpers
 
         let(:merged_queries_message_type) { "NestedRelationshipsMergedQueries" }

@@ -11,7 +11,7 @@
 module ElasticGraph
   class GraphQL
     module Resolvers
-      RSpec.describe NestedRelationships, :factories, :uses_datastore, :capture_logs, :resolver do
+      RSpec.describe NestedRelationships, :factories, :ingests_json_data, :capture_logs, :resolver do
         # :expect_search_routing because the relation we use here uses an outbaund foreign key, which
         # is implemented via a filter on `id` (the search routing field)
         context "when the field being resolved is a relay connection field", :expect_search_routing do

@@ -13,7 +13,7 @@
 module ElasticGraph
   class GraphQL
     module Resolvers
-      RSpec.describe QuerySource, :factories, :uses_datastore do
+      RSpec.describe QuerySource, :factories, :ingests_json_data do
         let(:graphql) { build_graphql }
 
         it "batches up multiple queries, sending an msearch request containing them all to the datastore" do

@@ -447,7 +447,6 @@ def resolve(field:, object:, args:, context:)
 
               schema.scalar_type "Operands" do |t|
                 t.mapping type: nil
-                t.json_schema type: "null"
               end
 
               schema.object_type "Widget" do |t|

@@ -19,7 +19,6 @@ module Resolvers
           self.schema_artifacts = generate_schema_artifacts do |schema|
             schema.scalar_type "MyInt" do |t|
               t.mapping type: "integer"
-              t.json_schema type: "integer"
             end
 
             schema.object_type "PersonIdentifiers" do |t|

@@ -14,7 +14,6 @@ module ScalarCoercionAdapters
       RSpec.describe "NoOp" do
         include_context("scalar coercion adapter support", "SomeCustomScalar", schema_definition: ->(schema) do
           schema.scalar_type "SomeCustomScalar" do |t|
-            t.json_schema type: "null"
             t.mapping type: nil
           end
 

@@ -955,7 +955,6 @@ def search_index_definitions_from(type_name: "TheType")
             schema = define_schema do |s|
               s.scalar_type "CustomScalar" do |t|
                 t.mapping type: "keyword"
-                t.json_schema type: "string"
               end
             end
 
@@ -967,7 +966,6 @@ def search_index_definitions_from(type_name: "TheType")
             schema = define_schema do |s|
               s.scalar_type "CustomScalar" do |t|
                 t.mapping type: "keyword"
-                t.json_schema type: "string"
                 t.grouping_missing_value_placeholder "MISSING"
               end
             end
@@ -980,7 +978,6 @@ def search_index_definitions_from(type_name: "TheType")
             schema = define_schema do |s|
               s.scalar_type "CustomScalar" do |t|
                 t.mapping type: "keyword"
-                t.json_schema type: "string"
                 t.grouping_missing_value_placeholder nil
               end
             end
@@ -1007,7 +1004,6 @@ def search_index_definitions_from(type_name: "TheType")
                 t.field "name", "String"
                 t.field "count", "Int"
                 t.field "price", "Float"
-                t.field "big_number", "JsonSafeLong"
                 t.index "things"
               end
             end
@@ -1016,7 +1012,6 @@ def search_index_definitions_from(type_name: "TheType")
             expect(schema.type_named("String").grouping_missing_value_placeholder).to eq MISSING_STRING_PLACEHOLDER_VALUE
             expect(schema.type_named("Int").grouping_missing_value_placeholder).to eq MISSING_NUMERIC_PLACEHOLDER
             expect(schema.type_named("Float").grouping_missing_value_placeholder).to eq MISSING_NUMERIC_PLACEHOLDER
-            expect(schema.type_named("JsonSafeLong").grouping_missing_value_placeholder).to eq MISSING_NUMERIC_PLACEHOLDER
           end
 
           it "returns nil for object types" do

@@ -16,7 +16,6 @@ class GraphQL
         define_schema do |schema|
           schema.scalar_type "_FieldSet" do |t|
             t.mapping type: "keyword"
-            t.json_schema type: "string"
           end
 
           schema.object_type "Widget" do |t|

@@ -48,6 +48,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "elasticgraph-admin", ElasticGraph::VERSION
   spec.add_development_dependency "elasticgraph-elasticsearch", ElasticGraph::VERSION
   spec.add_development_dependency "elasticgraph-indexer", ElasticGraph::VERSION
+  spec.add_development_dependency "elasticgraph-json_ingestion", ElasticGraph::VERSION
   spec.add_development_dependency "elasticgraph-opensearch", ElasticGraph::VERSION
   spec.add_development_dependency "elasticgraph-schema_definition", ElasticGraph::VERSION
 end
@@ -13,7 +13,7 @@
 
 module ElasticGraph
   module HealthCheck
-    RSpec.describe "HealthChecker", :uses_datastore, :factories, :builds_graphql, :capture_logs do
+    RSpec.describe "HealthChecker", :ingests_json_data, :factories, :builds_graphql, :capture_logs do
       let(:now) { ::Time.iso8601("2022-02-14T12:30:00Z") }
       let(:graphql) { build_graphql }
       let(:health_checker) { HealthChecker.build_from(graphql) }

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 module ElasticGraph
-  RSpec.describe "A derived indexing type", :uses_datastore, :factories, :capture_logs do
+  RSpec.describe "A derived indexing type", :ingests_json_data, :factories, :capture_logs do
     let(:indexer) { build_indexer }
 
     it "maintains derived fields, handling nested source and destination fields as needed" do

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 module ElasticGraph
-  RSpec.describe "Indexing into list fields", :uses_datastore, :factories, :capture_logs do
+  RSpec.describe "Indexing into list fields", :ingests_json_data, :factories, :capture_logs do
     let(:indexer) { build_indexer }
 
     it "indexes counts of any list fields so we can later use it for filtering" do

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 module ElasticGraph
-  RSpec.describe "Nested multi-source indexing", :uses_datastore, :factories, :capture_logs do
+  RSpec.describe "Nested multi-source indexing", :ingests_json_data, :factories, :capture_logs do
     let(:indexer) { build_indexer }
     let(:league) { "NBA" }
     let(:formed_on) { "2019-04-23" }

@@ -7,11 +7,10 @@
 # frozen_string_literal: true
 
 require "elastic_graph/indexer"
-require "elastic_graph/json_ingestion/schema_definition/api_extension"
 require "elastic_graph/schema_definition/rake_tasks"
 
 module ElasticGraph
-  RSpec.describe "Indexing schema evolution", :uses_datastore, :factories, :capture_logs, :in_temp_dir, :rake_task do
+  RSpec.describe "Indexing schema evolution", :ingests_json_data, :factories, :capture_logs, :in_temp_dir, :rake_task do
     let(:path_to_schema) { "config/schema.rb" }
 
     before do
@@ -402,7 +401,7 @@ def dump_artifacts
       run_rake "schema_artifacts:dump" do |output|
         SchemaDefinition::RakeTasks.new(
           schema_element_name_form: :snake_case,
-          extension_modules: [JSONIngestion::SchemaDefinition::APIExtension],
+          extension_modules: json_ingestion_schema_definition_extension_modules,
           index_document_sizes: true,
           path_to_schema: path_to_schema,
           schema_artifacts_directory: "config/schema/artifacts",

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 module ElasticGraph
-  RSpec.describe "Top-level multi-source indexing", :uses_datastore, :factories, :capture_logs do
+  RSpec.describe "Top-level multi-source indexing", :ingests_json_data, :factories, :capture_logs do
     let(:indexer) { build_indexer }
 
     it "ingests data from multiple source types into a single document, regardless of the ingestion order" do

@@ -11,7 +11,7 @@
 
 module ElasticGraph
   class Indexer
-    RSpec.describe DatastoreIndexingRouter, :uses_datastore, :capture_logs do
+    RSpec.describe DatastoreIndexingRouter, :ingests_json_data, :capture_logs do
       describe "#source_event_versions_in_index", :factories do
         let(:indexer) { build_indexer }
         let(:router) { indexer.datastore_router }

@@ -10,7 +10,7 @@
 
 module ElasticGraph
   class Indexer
-    RSpec.describe Processor, :uses_datastore, :factories, :capture_logs do
+    RSpec.describe Processor, :ingests_json_data, :factories, :capture_logs do
       let(:indexer) { build_indexer }
 
       context "process non-rollover upsert events" do

@@ -12,7 +12,7 @@
 # For example, to see your indexing preparer being used, this requires that it is registered
 # on your scalar type correctly, whereas if this directly called your indexing preparer
 # it wouldn't require it to be correctly registered.
-RSpec.shared_context "indexing preparer support" do |scalar_type|
+RSpec.shared_context "indexing preparer support", :json_ingestion_schema_definition do |scalar_type|
   before(:context) do
     @record_preparer = build_indexer(clients_by_name: {}, schema_definition: lambda do |schema|
       schema.object_type "MyType" do |t|

@@ -10,7 +10,7 @@
 
 module ElasticGraph
   class Indexer
-    ::RSpec.shared_context "MultipleVersionSupport" do
+    ::RSpec.shared_context "MultipleVersionSupport", :json_ingestion_schema_definition do
       include_context "SchemaDefinitionHelpers"
 
       def build_indexer_with_multiple_schema_versions(schema_versions:)

@@ -15,7 +15,7 @@
 
 module ElasticGraph
   class Indexer
-    RSpec.describe DatastoreIndexingRouter, :capture_logs do
+    RSpec.describe DatastoreIndexingRouter, :json_ingestion_schema_definition, :capture_logs do
       include SpecSupport::BuildsIndexerOperation
 
       let(:main_datastore_client) { instance_spy(Elasticsearch::Client, cluster_name: "main") }

@@ -15,7 +15,7 @@
 module ElasticGraph
   class Indexer
     module Operation
-      RSpec.describe Update do
+      RSpec.describe Update, :json_ingestion_schema_definition do
         include SchemaArtifacts::RuntimeMetadata::RuntimeMetadataSupport
 
         let(:indexer) { build_indexer }

@@ -12,7 +12,7 @@
 
 module ElasticGraph
   class Indexer
-    RSpec.describe RecordPreparer::Factory do
+    RSpec.describe RecordPreparer::Factory, :json_ingestion_schema_definition do
       include_context "MultipleVersionSupport"
 
       let(:factory_with_multiple_versions) do
@@ -54,7 +54,7 @@ class Indexer
       end
     end
 
-    RSpec.describe RecordPreparer do
+    RSpec.describe RecordPreparer, :json_ingestion_schema_definition do
       describe "#prepare_for_index" do
         it "tolerates a `nil` value where an object would usually be" do
           preparer = build_preparer do |s|
@@ -494,7 +494,7 @@ class Indexer
         end
       end
 
-      context "when working with events for an old JSON schema version" do
+      context "when working with events for an old JSON schema version", :json_ingestion_schema_definition do
         include_context "SchemaDefinitionHelpers"
 
         it "handles events for old versions before a field was deleted" do
@@ -578,7 +578,7 @@ def build_preparer_for_old_json_schema_version(v1_def:, v2_def:)
         def define_schema(&schema_definition)
           super(
             schema_element_name_form: "snake_case",
-            extension_modules: [JSONIngestion::SchemaDefinition::APIExtension],
+            extension_modules: json_ingestion_schema_definition_extension_modules,
             &schema_definition
           )
         end

@@ -7,7 +7,6 @@
 # frozen_string_literal: true
 
 require "elastic_graph/constants"
-require "elastic_graph/json_ingestion/schema_definition/api_extension"
 require "elastic_graph/schema_definition/rake_tasks"
 require "yaml"
 
@@ -620,7 +619,7 @@ def as_active_instance
               index_document_sizes: true,
               path_to_schema: "schema.rb",
               schema_artifacts_directory: "config/schema/artifacts",
-              extension_modules: [APIExtension, extension_module],
+              extension_modules: json_ingestion_schema_definition_extension_modules([extension_module]),
               output: output
             )
           end

@@ -8,3 +8,14 @@
 
 # This file contains RSpec configuration for `elasticgraph-json_ingestion`.
 # It is loaded by the shared spec helper at `spec_support/spec_helper.rb`.
+
+require "elastic_graph/json_ingestion/schema_definition/api_extension"
+require "elastic_graph/spec_support/schema_definition_helpers"
+
+RSpec.configure do |config|
+  config.define_derived_metadata(absolute_file_path: %r{/elasticgraph-json_ingestion/}) do |meta|
+    meta[:json_ingestion_schema_definition] = true
+  end
+
+  config.include_context "SchemaDefinitionHelpers", :json_ingestion_schema_definition
+end