Register nested sourced_from path segments on the root index

elasticgraph-datastore_core/spec/integration/elastic_graph/datastore_core/index_definition/index_spec.rb

@@ -89,7 +89,7 @@ def index_def_named(name, rollover: nil)
             current_sources: [SELF_RELATIONSHIP_NAME],
             fields_by_path: {},
             has_had_multiple_sources: false,
-            sourced_from_nested_paths_by_relationship: {}
+            sourced_from_nested_paths_by_qualified_relationship: {}
           )
 
           DatastoreCore::IndexDefinition.with(

elasticgraph-datastore_core/spec/integration/elastic_graph/datastore_core/index_definition/rollover_index_template_spec.rb

@@ -558,7 +558,7 @@ def index_def_named(name, rollover: nil)
             current_sources: [SELF_RELATIONSHIP_NAME],
             fields_by_path: {},
             has_had_multiple_sources: false,
-            sourced_from_nested_paths_by_relationship: {}
+            sourced_from_nested_paths_by_qualified_relationship: {}
           )
 
           DatastoreCore::IndexDefinition.with(

elasticgraph-schema_artifacts/lib/elastic_graph/schema_artifacts/runtime_metadata/index_definition.rb

@@ -17,24 +17,24 @@ module RuntimeMetadata
       # Runtime metadata related to a datastore index definition.
       #
       # @private
-      class IndexDefinition < ::Data.define(:route_with, :rollover, :default_sort_fields, :current_sources, :fields_by_path, :has_had_multiple_sources, :sourced_from_nested_paths_by_relationship)
+      class IndexDefinition < ::Data.define(:route_with, :rollover, :default_sort_fields, :current_sources, :fields_by_path, :has_had_multiple_sources, :sourced_from_nested_paths_by_qualified_relationship)
         ROUTE_WITH = "route_with"
         ROLLOVER = "rollover"
         DEFAULT_SORT_FIELDS = "default_sort_fields"
         CURRENT_SOURCES = "current_sources"
         FIELDS_BY_PATH = "fields_by_path"
         HAS_HAD_MULTIPLE_SOURCES = "has_had_multiple_sources"
-        SOURCED_FROM_NESTED_PATHS_BY_RELATIONSHIP = "sourced_from_nested_paths_by_relationship"
+        SOURCED_FROM_NESTED_PATHS_BY_QUALIFIED_RELATIONSHIP = "sourced_from_nested_paths_by_qualified_relationship"
 
-        def initialize(route_with:, rollover:, default_sort_fields:, current_sources:, fields_by_path:, has_had_multiple_sources:, sourced_from_nested_paths_by_relationship:)
+        def initialize(route_with:, rollover:, default_sort_fields:, current_sources:, fields_by_path:, has_had_multiple_sources:, sourced_from_nested_paths_by_qualified_relationship:)
           super(
             route_with: route_with,
             rollover: rollover,
             default_sort_fields: default_sort_fields,
             current_sources: current_sources.to_set,
             fields_by_path: fields_by_path,
             has_had_multiple_sources: has_had_multiple_sources,
-            sourced_from_nested_paths_by_relationship: sourced_from_nested_paths_by_relationship
+            sourced_from_nested_paths_by_qualified_relationship: sourced_from_nested_paths_by_qualified_relationship
           )
         end
 
@@ -46,7 +46,7 @@ def self.from_hash(hash)
             current_sources: hash[CURRENT_SOURCES] || [],
             fields_by_path: (hash[FIELDS_BY_PATH] || {}).transform_values { |h| IndexField.from_hash(h) },
             has_had_multiple_sources: hash[HAS_HAD_MULTIPLE_SOURCES] || false,
-            sourced_from_nested_paths_by_relationship: (hash[SOURCED_FROM_NESTED_PATHS_BY_RELATIONSHIP] || {}).transform_values { |segments| segments.map { |h| SourcedFromNestedPathSegment.from_hash(h) } }
+            sourced_from_nested_paths_by_qualified_relationship: (hash[SOURCED_FROM_NESTED_PATHS_BY_QUALIFIED_RELATIONSHIP] || {}).transform_values { |segments| segments.map { |h| SourcedFromNestedPathSegment.from_hash(h) } }
           )
         end
 
@@ -59,7 +59,7 @@ def to_dumpable_hash
             HAS_HAD_MULTIPLE_SOURCES => (true if has_had_multiple_sources),
             ROLLOVER => rollover&.to_dumpable_hash,
             ROUTE_WITH => route_with,
-            SOURCED_FROM_NESTED_PATHS_BY_RELATIONSHIP => sourced_from_nested_paths_by_relationship.transform_values { |segments| segments.map(&:to_dumpable_hash) }
+            SOURCED_FROM_NESTED_PATHS_BY_QUALIFIED_RELATIONSHIP => sourced_from_nested_paths_by_qualified_relationship.transform_values { |segments| segments.map(&:to_dumpable_hash) }
           }
         end
 

elasticgraph-schema_artifacts/sig/elastic_graph/schema_artifacts/runtime_metadata/index_definition.rbs

@@ -8,7 +8,7 @@ module ElasticGraph
         attr_reader current_sources: ::Set[::String]
         attr_reader fields_by_path: ::Hash[::String, IndexField]
         attr_reader has_had_multiple_sources: bool
-        attr_reader sourced_from_nested_paths_by_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
+        attr_reader sourced_from_nested_paths_by_qualified_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
 
         def initialize: (
           route_with: ::String,
@@ -17,7 +17,7 @@ module ElasticGraph
           current_sources: ::Set[::String],
           fields_by_path: ::Hash[::String, IndexField],
           has_had_multiple_sources: bool,
-          sourced_from_nested_paths_by_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
+          sourced_from_nested_paths_by_qualified_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
         ) -> void
 
         def with: (
@@ -27,7 +27,7 @@ module ElasticGraph
           ?current_sources: ::Enumerable[::String],
           ?fields_by_path: ::Hash[::String, IndexField],
           ?has_had_multiple_sources: bool,
-          ?sourced_from_nested_paths_by_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
+          ?sourced_from_nested_paths_by_qualified_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
         ) -> IndexDefinition
       end
 
@@ -38,7 +38,7 @@ module ElasticGraph
         CURRENT_SOURCES: "current_sources"
         FIELDS_BY_PATH: "fields_by_path"
         HAS_HAD_MULTIPLE_SOURCES: "has_had_multiple_sources"
-        SOURCED_FROM_NESTED_PATHS_BY_RELATIONSHIP: "sourced_from_nested_paths_by_relationship"
+        SOURCED_FROM_NESTED_PATHS_BY_QUALIFIED_RELATIONSHIP: "sourced_from_nested_paths_by_qualified_relationship"
 
         def initialize: (
           route_with: ::String,
@@ -47,7 +47,7 @@ module ElasticGraph
           current_sources: ::Enumerable[::String],
           fields_by_path: ::Hash[::String, IndexField],
           has_had_multiple_sources: bool,
-          sourced_from_nested_paths_by_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
+          sourced_from_nested_paths_by_qualified_relationship: ::Hash[::String, ::Array[sourcedFromNestedPathSegment]]
         ) -> void
 
         def self.from_hash: (::Hash[::String, untyped]) -> IndexDefinition

elasticgraph-schema_artifacts/spec/unit/elastic_graph/schema_artifacts/runtime_metadata/index_definition_spec.rb

@@ -25,7 +25,7 @@ module RuntimeMetadata
             current_sources: Set.new,
             fields_by_path: {},
             has_had_multiple_sources: false,
-            sourced_from_nested_paths_by_relationship: {}
+            sourced_from_nested_paths_by_qualified_relationship: {}
           )
         end
 

elasticgraph-schema_artifacts/spec/unit/elastic_graph/schema_artifacts/runtime_metadata/schema_spec.rb

@@ -128,7 +128,7 @@ module RuntimeMetadata
                   "foo.bar" => IndexField.new(source: "other")
                 },
                 has_had_multiple_sources: false,
-                sourced_from_nested_paths_by_relationship: {
+                sourced_from_nested_paths_by_qualified_relationship: {
                   "currency" => [
                     ListPathSegment.new(field: "costs", source_field: "cost_id"),
                     ObjectPathSegment.new(field: "details")
@@ -142,7 +142,7 @@ module RuntimeMetadata
                 current_sources: [SELF_RELATIONSHIP_NAME],
                 fields_by_path: {},
                 has_had_multiple_sources: false,
-                sourced_from_nested_paths_by_relationship: {}
+                sourced_from_nested_paths_by_qualified_relationship: {}
               ),
               "components" => IndexDefinition.new(
                 route_with: "group_id",
@@ -151,7 +151,7 @@ module RuntimeMetadata
                 current_sources: [SELF_RELATIONSHIP_NAME],
                 fields_by_path: {},
                 has_had_multiple_sources: true,
-                sourced_from_nested_paths_by_relationship: {}
+                sourced_from_nested_paths_by_qualified_relationship: {}
               )
             },
             schema_element_names: SchemaElementNames.new(
@@ -279,7 +279,7 @@ module RuntimeMetadata
                     "source" => "other"
                   }
                 },
-                "sourced_from_nested_paths_by_relationship" => {
+                "sourced_from_nested_paths_by_qualified_relationship" => {
                   "currency" => [
                     {"field" => "costs", "source_field" => "cost_id"},
                     {"field" => "details"}

elasticgraph-schema_definition/lib/elastic_graph/schema_definition/indexing/index.rb

@@ -38,7 +38,21 @@ module Indexing
       #   @return [RolloverConfig, nil] rollover configuration for the index
       # @!attribute [r] has_had_multiple_sources_flag
       #   @return [Boolean] whether this index has ever had multiple sources
-      class Index < Struct.new(:name, :default_sort_pairs, :settings, :schema_def_state, :indexed_type, :routing_field_path, :rollover_config, :has_had_multiple_sources_flag)
+      # @!attribute [r] sourced_from_nested_paths_by_qualified_relationship
+      #   @return [Hash<String, Array<SchemaArtifacts::RuntimeMetadata::ListPathSegment, SchemaArtifacts::RuntimeMetadata::ObjectPathSegment>>]
+      #     map from a qualified (leaf) relationship to the path segments the painless script uses to navigate from this
+      #     root index's documents down to the nested elements that receive `sourced_from` data
+      class Index < Struct.new(
+        :name,
+        :default_sort_pairs,
+        :settings,
+        :schema_def_state,
+        :indexed_type,
+        :routing_field_path,
+        :rollover_config,
+        :has_had_multiple_sources_flag,
+        :sourced_from_nested_paths_by_qualified_relationship
+      )
         include Mixins::HasReadableToSAndInspect.new { |i| i.name }
 
         # @param name [String] name of the index
@@ -55,7 +69,7 @@ def initialize(name, settings, schema_def_state, indexed_type)
 
           settings = DEFAULT_SETTINGS.merge(Support::HashUtil.flatten_and_stringify_keys(settings, prefix: "index"))
 
-          super(name, [], settings, schema_def_state, indexed_type, nil, nil, false)
+          super(name, [], settings, schema_def_state, indexed_type, nil, nil, false, {})
 
           schema_def_state.after_user_definition_complete do
             # `id` is the field Elasticsearch/OpenSearch use for routing by default:
@@ -265,10 +279,16 @@ def runtime_metadata
               )
             end,
             has_had_multiple_sources: has_had_multiple_sources_flag,
-            sourced_from_nested_paths_by_relationship: {} # TODO: Populate with real data once paths are registered on the index
+            sourced_from_nested_paths_by_qualified_relationship: sourced_from_nested_paths_by_qualified_relationship
           )
         end
 
+        # Registers a resolved `parent_relationship` chain on this index.
+        # @api private
+        def register_resolved_relationship_chain(resolved_chain)
+          sourced_from_nested_paths_by_qualified_relationship[resolved_chain.qualified_relationship] = resolved_chain.sourced_from_nested_paths
+        end
+
         private
 
         # A regex that requires at least one non-whitespace character.

elasticgraph-schema_definition/lib/elastic_graph/schema_definition/indexing/relationship_chain_resolver.rb

@@ -7,17 +7,42 @@
 # frozen_string_literal: true
 
 require "elastic_graph/errors"
+require "elastic_graph/schema_artifacts/runtime_metadata/sourced_from_nested_path_segment"
 
 module ElasticGraph
   module SchemaDefinition
     module Indexing
       # The result of resolving a relationship chain.
       #
       # @private
-      ResolvedRelationshipChain = ::Data.define(
-        :root_relationship,  # Relationship - the root relationship (no parent_relationship)
-        :path_segments       # Array<PathSegment> - ordered root-to-leaf
+      class ResolvedRelationshipChain < ::Data.define(
+        :root_relationship,  # Relationship the chain terminated at on the root indexed type
+        :leaf_relationship,  # Relationship the chain was resolved from — backs `sourced_from` field(s)
+        :path_segments       # Array<PathSegment> - the embedding fields to descend, ordered root-to-leaf
       )
+        # The leaf relationship name qualified by its embedding-field path (hence unique per resolved chain)
+        def qualified_relationship
+          (path_segments.map { |segment| segment.field.name_in_index } + [leaf_relationship.name_in_index]).join(".")
+        end
+
+        # The runtime-metadata segments the painless script uses to navigate this chain: a `ListPathSegment` for
+        # each list embedding field (carrying the source field that matches the element) and an `ObjectPathSegment`
+        # for each object embedding field.
+        def sourced_from_nested_paths
+          path_segments.map do |segment|
+            if (source_field = segment.source_field_name)
+              SchemaArtifacts::RuntimeMetadata::ListPathSegment.new(
+                field: segment.field.name_in_index,
+                source_field: source_field
+              )
+            else
+              SchemaArtifacts::RuntimeMetadata::ObjectPathSegment.new(
+                field: segment.field.name_in_index
+              )
+            end
+          end
+        end
+      end
 
       # Describes how to navigate from a parent type into a nested child element.
       # For list fields, `source_field_name` identifies which element to update: the element
@@ -75,6 +100,7 @@ def resolve(starting_relationship)
 
           resolved_chain = ResolvedRelationshipChain.new(
             root_relationship: root_relationship,
+            leaf_relationship: starting_relationship,
             path_segments: path_segments.reverse # reverse so root-to-leaf order
           )
 

elasticgraph-schema_definition/lib/elastic_graph/schema_definition/indexing/sourced_from_update_targets_resolver.rb

@@ -67,8 +67,8 @@ def resolve_for_type(object_type, &error_reporter)
             end
           end
 
-          # Resolve any `parent_relationship` chains on this type. For now this only surfaces
-          # configuration errors; later PRs will use the resolved chains to build nested update targets.
+          # Resolve any `parent_relationship` chains on this type, surfacing configuration errors and
+          # registering the resolved path segments on the root index.
           resolve_relationship_chains(object_type, &error_reporter)
 
           results
@@ -102,12 +102,28 @@ def resolve_relationship_chains(object_type)
           relationships_with_parent_ref = object_type.relationships_by_name.each_value.select(&:parent_ref)
           return if relationships_with_parent_ref.empty?
 
+          # The set of relationship names this type's `sourced_from` fields draw their data through. We use
+          # `fields_with_sources` (rather than `sourced_fields_by_relationship_name`) because it works for
+          # non-indexed (embedded) types — which is exactly where nested `sourced_from` fields live.
+          relationship_names_with_sourced_fields = object_type.fields_with_sources.map { |f| (_ = f.source).relationship_name }.to_set
+
           chain_resolver = RelationshipChainResolver.new(schema_def_state: @schema_def_state)
 
           relationships_with_parent_ref.each do |relationship|
-            # TODO: use resolved_chain to build nested update targets once that logic is implemented.
-            _resolved_chain, chain_errors = chain_resolver.resolve(relationship)
+            # Every `parent_relationship` chain is resolved so its configuration errors are surfaced, even for
+            # relationships that don't themselves back any `sourced_from` field.
+            resolved_chain, chain_errors = chain_resolver.resolve(relationship)
             chain_errors.each { |error| yield :sourced_field, error }
+            next unless resolved_chain
+
+            # Only register paths for relationships that back `sourced_from` fields. A pure link in a longer
+            # chain has no nested data of its own; its navigation lives in the leaf relationship's chain.
+            next unless relationship_names_with_sourced_fields.include?(relationship.name)
+
+            # Register the resolved chain on its root index, which records the path segments keyed by the
+            # chain's qualified relationship.
+            root_index = resolved_chain.root_relationship.parent_type.index_def # : Index
+            root_index.register_resolved_relationship_chain(resolved_chain)
           end
         end
 

elasticgraph-schema_definition/sig/elastic_graph/schema_definition/indexing/index.rbs

@@ -8,11 +8,13 @@ module ElasticGraph
         attr_reader rollover_config: RolloverConfig?
         attr_reader has_had_multiple_sources_flag: bool
         attr_reader indexed_type: indexableType
+        attr_reader sourced_from_nested_paths_by_qualified_relationship: ::Hash[::String, ::Array[SchemaArtifacts::RuntimeMetadata::sourcedFromNestedPathSegment]]
 
         def uses_custom_routing?: () -> bool
         def to_index_config: () -> ::Hash[::String, untyped]
         def to_index_template_config: () -> ::Hash[::String, untyped]
         def runtime_metadata: () -> SchemaArtifacts::RuntimeMetadata::IndexDefinition
+        def register_resolved_relationship_chain: (ResolvedRelationshipChain) -> void
 
         private
 

elasticgraph-schema_definition/sig/elastic_graph/schema_definition/indexing/relationship_chain_resolver.rbs

@@ -1,16 +1,23 @@
 module ElasticGraph
   module SchemaDefinition
     module Indexing
-      class ResolvedRelationshipChain
+      class ResolvedRelationshipChainSuperType
         attr_reader root_relationship: SchemaElements::Relationship
+        attr_reader leaf_relationship: SchemaElements::Relationship
         attr_reader path_segments: ::Array[PathSegment]
 
         def initialize: (
           root_relationship: SchemaElements::Relationship,
+          leaf_relationship: SchemaElements::Relationship,
           path_segments: ::Array[PathSegment]
         ) -> void
       end
 
+      class ResolvedRelationshipChain < ResolvedRelationshipChainSuperType
+        def qualified_relationship: () -> ::String
+        def sourced_from_nested_paths: () -> ::Array[SchemaArtifacts::RuntimeMetadata::sourcedFromNestedPathSegment]
+      end
+
       class PathSegment
         attr_reader field: SchemaElements::Field
         attr_reader source_field_name: ::String?