elasticgraph-graphql 0.18.0.4 → 0.19.0.0.rc1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b0cb73d5ec2940bb792cbcada0c615dd1415de774bb9111ffe09ae293cf2de38
-  data.tar.gz: fb0f23546ee15fd3c9f434518dde7ccb79df0838795392d6b0a8c9e60df01667
+  metadata.gz: 01af70cea7a26c2897ca64c179e0ed5a9a482a9229a46304b2e59b2f249148c3
+  data.tar.gz: 9a6aca68a2663012e54bf022a7153ec4443b3bc59846c63b91e6084b07b61574
 SHA512:
-  metadata.gz: 4c21c60909ba34c7034b2c5506288f54d4e29aa715e9741e1f88a046b54d6a694082b657c00aa495da7d28eb771bec28c2a23d0c10b3c5dbfa46e20088c9802f
-  data.tar.gz: ecbd8f84362640638e36384875d43c8a9443a413bb407dd2fca5e1d18792bfe618ac5b15c14a9d0c7d2656b7234620d73ce95078bc18ed4cc72eecbec4958ad3
+  metadata.gz: 5ead565225393723335e8dcc0dea15a6858dd005e5f96531ad60aa21b2cf0166760cb5c25e788b0dbb5d367c50ddb1335f9fd5129a2ee173df0d04c436e5ed15
+  data.tar.gz: fc360a136de5620ac53375046dce9e93a90b19f4ea3eb2331f8ca2dbda2f2701b55156e04ac60f71ab4bf45c5eca4e45a2efcf2236cef25f78503ef1beca5cda

data/elasticgraph-graphql.gemspec

@@ -13,7 +13,7 @@ ElasticGraphGemspecHelper.define_elasticgraph_gem(gemspec_file: __FILE__, catego
 
   spec.add_dependency "elasticgraph-datastore_core", eg_version
   spec.add_dependency "elasticgraph-schema_artifacts", eg_version
-  spec.add_dependency "graphql", "~> 2.3.14"
+  spec.add_dependency "graphql", "~> 2.4.5"
 
   spec.add_development_dependency "elasticgraph-admin", eg_version
   spec.add_development_dependency "elasticgraph-elasticsearch", eg_version

data/lib/elastic_graph/graphql/aggregation/field_path_encoder.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 
 module ElasticGraph
   class GraphQL
@@ -38,7 +38,7 @@ module ElasticGraph
 
         private_class_method def self.verify_delimiters(str)
           if str.to_s.include?(DELIMITER)
-            raise InvalidArgumentValueError, %("#{str}" contains delimiter: "#{DELIMITER}")
+            raise Errors::InvalidArgumentValueError, %("#{str}" contains delimiter: "#{DELIMITER}")
           end
         end
       end

data/lib/elastic_graph/graphql/aggregation/key.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/aggregation/field_path_encoder"
 
 module ElasticGraph
@@ -77,7 +77,7 @@ module ElasticGraph
         def self.verify_no_delimiter_in(*parts)
           parts.each do |part|
             if part.to_s.include?(DELIMITER)
-              raise InvalidArgumentValueError, %("#{part}" contains delimiter: "#{DELIMITER}")
+              raise Errors::InvalidArgumentValueError, %("#{part}" contains delimiter: "#{DELIMITER}")
             end
           end
         end

data/lib/elastic_graph/graphql/aggregation/non_composite_grouping_adapter.rb

@@ -114,8 +114,17 @@ module ElasticGraph
           # (by doc count descending, then by the key values ascending), and then take only first `size`.
           def sort_and_truncate_buckets(buckets, size)
             buckets
-              .sort_by { |b| [-b.fetch("doc_count"), b.fetch("key_values")] }
-              .first(size)
+              .sort_by do |b|
+                # We convert key values to strings to ensure they are comparable. Otherwise, we can get an error like:
+                #
+                # > ArgumentError: comparison of Array with Array failed
+                #
+                # Note that this turns it into a lexicographical sort rather than a more type-aware sort
+                # (10 will sort before 2, for example), but that's fine. We only sort by `key_values` as
+                # a time breaker to ensure deterministic results, but don't particularly care which buckets
+                # come first.
+                [-b.fetch("doc_count"), b.fetch("key_values").map(&:to_s)]
+              end.first(size)
           end
 
           def missing_bucket_path_from(buckets_path)

data/lib/elastic_graph/graphql/config.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/client"
 require "elastic_graph/schema_artifacts/runtime_metadata/extension_loader"
 
@@ -33,14 +33,14 @@ module ElasticGraph
         extra_keys = parsed_yaml.keys - EXPECTED_KEYS
 
         unless extra_keys.empty?
-          raise ConfigError, "Unknown `graphql` config settings: #{extra_keys.join(", ")}"
+          raise Errors::ConfigError, "Unknown `graphql` config settings: #{extra_keys.join(", ")}"
         end
 
         extension_loader = SchemaArtifacts::RuntimeMetadata::ExtensionLoader.new(::Module.new)
         extension_mods = parsed_yaml.fetch("extension_modules", []).map do |mod_hash|
           extension_loader.load(mod_hash.fetch("extension_name"), from: mod_hash.fetch("require_path"), config: {}).extension_class.tap do |mod|
             unless mod.instance_of?(::Module)
-              raise ConfigError, "`#{mod_hash.fetch("extension_name")}` is not a module, but all application extension modules must be modules."
+              raise Errors::ConfigError, "`#{mod_hash.fetch("extension_name")}` is not a module, but all application extension modules must be modules."
             end
           end
         end

data/lib/elastic_graph/graphql/datastore_query/document_paginator.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "graphql"
 
 module ElasticGraph

data/lib/elastic_graph/graphql/datastore_query/index_expression_builder.rb

@@ -15,7 +15,7 @@ module ElasticGraph
       # Responsible for building a search index expression for a specific query based on the filters.
       class IndexExpressionBuilder
         def initialize(schema_names:)
-          @filter_value_set_extractor = Filtering::FilterValueSetExtractor.new(schema_names, Support::TimeSet::ALL) do |operator, filter_value|
+          @filter_value_set_extractor = Filtering::FilterValueSetExtractor.new(schema_names, Support::TimeSet::ALL, Support::TimeSet::EMPTY) do |operator, filter_value|
             case operator
             when :gt, :gte, :lt, :lte
               if date_string?(filter_value)

data/lib/elastic_graph/graphql/datastore_query/paginator.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/support/memoizable_data"
 
 module ElasticGraph

data/lib/elastic_graph/graphql/datastore_query/routing_picker.rb

@@ -16,8 +16,9 @@ module ElasticGraph
         def initialize(schema_names:)
           # @type var all_values_set: _RoutingValueSet
           all_values_set = RoutingValueSet::ALL
+          empty_set = RoutingValueSet::EMPTY
 
-          @filter_value_set_extractor = Filtering::FilterValueSetExtractor.new(schema_names, all_values_set) do |operator, filter_value|
+          @filter_value_set_extractor = Filtering::FilterValueSetExtractor.new(schema_names, all_values_set, empty_set) do |operator, filter_value|
             if operator == :equal_to_any_of
               # This calls `.compact` to remove `nil` filter_value values
               RoutingValueSet.of(filter_value.compact)
@@ -70,6 +71,7 @@ module ElasticGraph
         end
 
         ALL = of_all_except([])
+        EMPTY = of([])
 
         def intersection(other_set)
           # Here we return `self` to preserve the commutative property of `intersection`. Returning `self`

data/lib/elastic_graph/graphql/datastore_query.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/aggregation/query"
 require "elastic_graph/graphql/aggregation/query_optimizer"
 require "elastic_graph/graphql/decoded_cursor"
@@ -65,7 +65,7 @@ module ElasticGraph
         )
 
         if search_index_definitions.empty?
-          raise SearchFailedError, "Query is invalid, since it contains no `search_index_definitions`."
+          raise Errors::SearchFailedError, "Query is invalid, since it contains no `search_index_definitions`."
         end
       end
     }
@@ -121,7 +121,7 @@ module ElasticGraph
             missing_queries = expected_queries - actual_queries
             extra_queries = actual_queries - expected_queries
 
-            raise SearchFailedError, "The `responses_hash` does not have the expected set of queries as keys. " \
+            raise Errors::SearchFailedError, "The `responses_hash` does not have the expected set of queries as keys. " \
               "This can cause problems for the `GraphQL::Dataloader` and suggests a bug in the logic that should be fixed.\n\n" \
               "Missing queries (#{missing_queries.size}):\n#{missing_queries.map(&:inspect).join("\n")}.\n\n" \
               "Extra queries (#{extra_queries.size}): #{extra_queries.map(&:inspect).join("\n")}"
@@ -133,7 +133,7 @@ module ElasticGraph
       # Both query objects are left unchanged.
       def merge(other_query)
         if search_index_definitions != other_query.search_index_definitions
-          raise ElasticGraph::InvalidMergeError, "`search_index_definitions` conflict while merging between " \
+          raise ElasticGraph::Errors::InvalidMergeError, "`search_index_definitions` conflict while merging between " \
             "#{search_index_definitions} and #{other_query.search_index_definitions}"
         end
 
@@ -177,11 +177,11 @@ module ElasticGraph
       end
 
       # Returns the name of the datastore cluster as a String where this query should be setn.
-      # Unless exactly 1 cluster name is found, this method raises a ConfigError.
+      # Unless exactly 1 cluster name is found, this method raises a Errors::ConfigError.
       def cluster_name
         cluster_name = search_index_definitions.map(&:cluster_to_query).uniq
         return cluster_name.first if cluster_name.size == 1
-        raise ConfigError, "Found different datastore clusters (#{cluster_name}) to query " \
+        raise Errors::ConfigError, "Found different datastore clusters (#{cluster_name}) to query " \
           "for query targeting indices: #{search_index_definitions}"
       end
 

data/lib/elastic_graph/graphql/datastore_response/search_response.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/decoded_cursor"
 require "elastic_graph/graphql/datastore_response/document"
 require "forwardable"
@@ -66,7 +66,7 @@ module ElasticGraph
         end
 
         def total_document_count
-          super || raise(CountUnavailableError, "#{__method__} is unavailable; set `query.total_document_count_needed = true` to make it available")
+          super || raise(Errors::CountUnavailableError, "#{__method__} is unavailable; set `query.total_document_count_needed = true` to make it available")
         end
 
         def to_s

data/lib/elastic_graph/graphql/datastore_search_router.rb

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 require "elastic_graph/constants"
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/datastore_response/search_response"
 require "elastic_graph/graphql/query_details_tracker"
 require "elastic_graph/support/threading"
@@ -61,7 +61,7 @@ module ElasticGraph
           # Unfortunately, the Elasticsearch/OpenSearch clients don't support setting a per-request client-side timeout,
           # even though Faraday (the underlying HTTP client) does. To work around this, we pass our desired
           # timeout in a specific header that the `SupportTimeouts` Faraday middleware will use.
-          headers = {TIMEOUT_MS_HEADER => msearch_request_timeout_from(queries)}.compact
+          headers = {TIMEOUT_MS_HEADER => msearch_request_timeout_from(queries)&.to_s}.compact
 
           queries_and_header_body_tuples_by_datastore_client = header_body_tuples_by_query.group_by do |(query, header_body_tuples)|
             @datastore_clients_by_name.fetch(query.cluster_name)
@@ -108,7 +108,7 @@ module ElasticGraph
 
         (min_query_deadline - @monotonic_clock.now_in_ms).tap do |timeout|
           if timeout <= 0
-            raise RequestExceededDeadlineError, "It is already #{timeout.abs} ms past the search deadline."
+            raise Errors::RequestExceededDeadlineError, "It is already #{timeout.abs} ms past the search deadline."
           end
         end
       end
@@ -118,6 +118,8 @@ module ElasticGraph
         return if failures.empty?
 
         formatted_failures = failures.map do |(query, response), index|
+          raise_execution_exception_for_known_public_error(response)
+
           # Note: we intentionally omit the body of the request here, because it could contain PII
           # or other sensitive values that we don't want logged.
           <<~ERROR
@@ -127,7 +129,22 @@ module ElasticGraph
           ERROR
         end.join("\n\n")
 
-        raise SearchFailedError, "Got #{failures.size} search failure(s):\n\n#{formatted_failures}"
+        raise Errors::SearchFailedError, "Got #{failures.size} search failure(s):\n\n#{formatted_failures}"
+      end
+
+      # In general, when we receive a datastore response indicating a search failed, we raise
+      # `Errors::SearchFailedError` which translates into a `500 Internal Server Error`. That's
+      # appropriate for transient errors (particularly when there's nothing the client can do about
+      # it) but for non-transient errors that the client can do something about, we'd like to provide
+      # a friendlier error. This method handles those cases.
+      #
+      # GraphQL::ExecutionError is automatically translated into a nice error response.
+      def raise_execution_exception_for_known_public_error(response)
+        if response.dig("error", "caused_by", "type") == "too_many_buckets_exception"
+          max_buckets = response.dig("error", "caused_by", "max_buckets")
+          raise ::GraphQL::ExecutionError, "Aggregation query produces too many groupings. " \
+            "Reduce the grouping cardinality to less than #{max_buckets} and try again."
+        end
       end
 
       # Examine successful query responses and log any shard failure they encounter

data/lib/elastic_graph/graphql/decoded_cursor.rb

@@ -8,7 +8,7 @@
 
 require "base64"
 require "elastic_graph/constants"
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/support/memoizable_data"
 require "json"
 
@@ -39,17 +39,17 @@ module ElasticGraph
       # Tries to decode the given string cursor, returning `nil` if it is invalid.
       def self.try_decode(string)
         decode!(string)
-      rescue InvalidCursorError
+      rescue Errors::InvalidCursorError
         nil
       end
 
-      # Tries to decode the given string cursor, raising an `InvalidCursorError` if it's invalid.
+      # Tries to decode the given string cursor, raising an `Errors::InvalidCursorError` if it's invalid.
       def self.decode!(string)
         return SINGLETON if string == SINGLETON_CURSOR
         json = ::Base64.urlsafe_decode64(string)
         new(::JSON.parse(json))
       rescue ::ArgumentError, ::JSON::ParserError
-        raise InvalidCursorError, "`#{string}` is an invalid cursor."
+        raise Errors::InvalidCursorError, "`#{string}` is an invalid cursor."
       end
 
       # Encodes the cursor to a string using JSON and Base64 encoding.
@@ -79,7 +79,7 @@ module ElasticGraph
         def self.from_sort_list(sort_list)
           sort_fields = sort_list.map do |hash|
             if hash.values.any? { |v| !v.is_a?(::Hash) } || hash.values.flat_map(&:keys) != ["order"]
-              raise InvalidSortFieldsError,
+              raise Errors::InvalidSortFieldsError,
                 "Given `sort_list` contained an invalid entry. Each must be a flat hash with one entry. Got: #{sort_list.inspect}"
             end
 
@@ -89,7 +89,7 @@ module ElasticGraph
           end
 
           if sort_fields.uniq.size < sort_fields.size
-            raise InvalidSortFieldsError,
+            raise Errors::InvalidSortFieldsError,
               "Given `sort_list` contains a duplicate field, which the CursorEncoder cannot handler. " \
               "The caller is responsible for de-duplicating the sort list fist. Got: #{sort_list.inspect}"
           end
@@ -99,7 +99,7 @@ module ElasticGraph
 
         def build(sort_values)
           unless sort_values.size == sort_fields.size
-            raise CursorEncodingError,
+            raise Errors::CursorEncodingError,
               "size of sort values (#{sort_values.inspect}) does not match the " \
               "size of sort fields (#{sort_fields.inspect})"
           end

data/lib/elastic_graph/graphql/filtering/boolean_query.rb

@@ -33,12 +33,7 @@ module ElasticGraph
           bool_node[occurrence].concat(clauses)
         end
 
-        # For `any_of: []` we need a way to force the datastore to match no documents, but
-        # I haven't found any sort of literal `false` we can pass in the compound expression
-        # or even a literal `1 = 0` as is sometimes used in SQL. Instead, we use this for that
-        # case.
-        empty_array = [] # : ::Array[untyped]
-        ALWAYS_FALSE_FILTER = filter({ids: {values: empty_array}})
+        ALWAYS_FALSE_FILTER = filter({match_none: {}})
       end
     end
   end

data/lib/elastic_graph/graphql/filtering/filter_interpreter.rb

@@ -59,14 +59,9 @@ module ElasticGraph
 
         def process_filter_hash(bool_node, filter_hash, field_path)
           filter_hash.each do |field_or_op, expression|
-            # `nil` filter predicates should be ignored, so we can safely `compact` them out.
-            # It also is simpler to handle them once here instead of the different branches
-            # below having to be aware of possible `nil` predicates.
-            expression = expression.compact if expression.is_a?(::Hash)
-
             case filter_node_interpreter.identify_node_type(field_or_op, expression)
             when :empty
-              # This is an "empty" filter predicate and we can ignore it.
+              # This is an "empty" filter predicate and can be treated as `true`.
             when :not
               process_not_expression(bool_node, expression, field_path)
             when :list_any_filter
@@ -109,10 +104,14 @@ module ElasticGraph
 
         def process_not_expression(bool_node, expression, field_path)
           sub_filter = build_bool_hash do |inner_node|
-            process_filter_hash(inner_node, expression, field_path)
+            process_filter_hash(inner_node, expression || {}, field_path)
           end
 
-          return unless sub_filter
+          unless sub_filter
+            # Since an empty expression is treated as `true`, convert to `false` when negating.
+            BooleanQuery::ALWAYS_FALSE_FILTER.merge_into(bool_node)
+            return
+          end
 
           # Prevent any negated filters from being unnecessarily double-negated by
           # converting them to a positive filter (i.e., !!A == A).
@@ -138,6 +137,8 @@ module ElasticGraph
         # this because we do not generate `any_satisfy` filters on `object` list fields (instead,
         # they get generated on their leaf fields).
         def process_list_any_filter_expression(bool_node, filter, field_path)
+          return if filter.nil? || filter == {}
+
           if filters_on_sub_fields?(filter)
             process_any_satisfy_filter_expression_on_nested_object_list(bool_node, filter, field_path)
           else
@@ -188,22 +189,37 @@ module ElasticGraph
           end
         end
 
+        # We want to provide the following semantics for `any_of`:
+        #
+        # * `filter: {anyOf: []}` -> return no results
+        # * `filter: {anyOf: [{field: null}]}` -> return all results
+        # * `filter: {anyOf: [{field: null}, {field: ...}]}` -> return all results
         def process_any_of_expression(bool_node, expressions, field_path)
+          return if expressions.nil? || expressions == {}
+
+          if expressions.empty?
+            # When our `expressions` array is empty, we want to match no documents. However, that's
+            # not the behavior the datastore will give us if we have an empty array in the query under
+            # `should`. To get the behavior we want, we need to pass the datastore some filter criteria
+            # that will evaluate to false for every document.
+            BooleanQuery::ALWAYS_FALSE_FILTER.merge_into(bool_node)
+            return
+          end
+
           shoulds = expressions.filter_map do |expression|
             build_bool_hash do |inner_bool_node|
               process_filter_hash(inner_bool_node, expression, field_path)
             end
           end
 
-          # When our `shoulds` array is empty, the filtering semantics we want is to match no documents.
-          # However, that's not the behavior the datastore will give us if we have an empty array in the
-          # query under `should`. To get the behavior we want, we need to pass the datastore some filter
-          # criteria that will evaluate to false for every document.
-          bool_query = shoulds.empty? ? BooleanQuery::ALWAYS_FALSE_FILTER : BooleanQuery.should(*shoulds)
-          bool_query.merge_into(bool_node)
+          return if shoulds.size < expressions.size
+
+          BooleanQuery.should(*shoulds).merge_into(bool_node)
         end
 
         def process_all_of_expression(bool_node, expressions, field_path)
+          return if expressions.nil? || expressions == {}
+
           # `all_of` represents an AND. AND is the default way that `process_filter_hash` combines
           # filters so we just have to call it for each sub-expression.
           expressions.each do |sub_expression|
@@ -212,6 +228,8 @@ module ElasticGraph
         end
 
         def process_operator_expression(bool_node, operator, expression, field_path)
+          return if expression.nil? || expression == {}
+
           # `operator` is a filtering operator, and `expression` is the value the filtering
           # operator should be applied to. The `op_applicator` lambda, when called, will
           # return a Clause instance (defined in this module).
@@ -299,6 +317,8 @@ module ElasticGraph
         end
 
         def process_list_count_expression(bool_node, expression, field_path)
+          return if expression.nil? || expression == {}
+
           # Normally, we don't have to do anything special for list count expressions.
           # That's the case, for example, for an expression like:
           #
@@ -313,7 +333,7 @@ module ElasticGraph
           # While we index an explicit count of 0, the count field will be missing from documents indexed before
           # the list field was defined on the ElasticGraph schema. To properly match those documents, we need to
           # convert this into an OR (using `any_of`) to also match documents that lack the field entirely.
-          unless excludes_zero?(expression)
+          if filters_to_range_including_zero?(expression)
             expression = {schema_names.any_of => [
               expression,
               {schema_names.equal_to_any_of => [nil]}
@@ -326,7 +346,7 @@ module ElasticGraph
         def build_bool_hash(&block)
           bool_node = Hash.new { |h, k| h[k] = [] }.tap(&block)
 
-          # To ignore "empty" filter predicates we need to return `nil` here.
+          # To treat "empty" filter predicates as `true` we need to return `nil` here.
           return nil if bool_node.empty?
 
           # According to https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-bool-query.html#bool-min-should-match,
@@ -337,20 +357,28 @@ module ElasticGraph
           {bool: bool_node}
         end
 
-        # Determines if the given filter expression excludes the value `0`.
-        def excludes_zero?(expression)
-          expression.any? do |operator, operand|
-            case operator
-            when schema_names.equal_to_any_of then !operand.include?(0)
-            when schema_names.lt then operand <= 0
-            when schema_names.lte then operand < 0
-            when schema_names.gt then operand >= 0
-            when schema_names.gte then operand > 0
-            else
-              # :nocov: -- all operators are covered above. But simplecov complains about an implicit `else` branch being uncovered, so here we've defined it to wrap it with `:nocov:`.
-              false
-              # :nocov:
-            end
+        # Determines if the given expression filters to a range that includes `0`.
+        # If it does not do any filtering (e.g. an empty expression) it will return `false`.
+        def filters_to_range_including_zero?(expression)
+          expression = expression.compact
+
+          expression.size > 0 && expression.none? do |operator, operand|
+            operator_excludes_zero?(operator, operand)
+          end
+        end
+
+        # Determines if the given operator and operand exclude 0 as a matched value.
+        def operator_excludes_zero?(operator, operand)
+          case operator
+          when schema_names.equal_to_any_of then !operand.include?(0)
+          when schema_names.lt then operand <= 0
+          when schema_names.lte then operand < 0
+          when schema_names.gt then operand >= 0
+          when schema_names.gte then operand > 0
+          else
+            # :nocov: -- all operators are covered above. But simplecov complains about an implicit `else` branch being uncovered, so here we've defined it to wrap it with `:nocov:`.
+            false
+            # :nocov:
           end
         end
 

data/lib/elastic_graph/graphql/filtering/filter_node_interpreter.rb

@@ -25,22 +25,32 @@ module ElasticGraph
         end
 
         def identify_node_type(field_or_op, sub_expression)
-          return :empty if sub_expression.nil? || sub_expression == {}
+          identify_by_field_or_op(field_or_op) || identify_by_sub_expr(sub_expression) || :unknown
+        end
+
+        def filter_operators
+          @filter_operators ||= build_filter_operators(runtime_metadata)
+        end
+
+        private
+
+        def identify_by_field_or_op(field_or_op)
           return :not if field_or_op == schema_names.not
           return :list_any_filter if field_or_op == schema_names.any_satisfy
           return :all_of if field_or_op == schema_names.all_of
           return :any_of if field_or_op == schema_names.any_of
           return :operator if filter_operators.key?(field_or_op)
           return :list_count if field_or_op == LIST_COUNTS_FIELD
-          return :sub_field if sub_expression.is_a?(::Hash)
-          :unknown
-        end
 
-        def filter_operators
-          @filter_operators ||= build_filter_operators(runtime_metadata)
+          nil
         end
 
-        private
+        def identify_by_sub_expr(sub_expression)
+          return :empty if sub_expression.nil? || sub_expression == {}
+          return :sub_field if sub_expression.is_a?(::Hash)
+
+          nil
+        end
 
         def build_filter_operators(runtime_metadata)
           filter_by_time_of_day_script_id = runtime_metadata

data/lib/elastic_graph/graphql/filtering/filter_value_set_extractor.rb

@@ -12,9 +12,10 @@ module ElasticGraph
       # Responsible for extracting a set of values from query filters, based on a using a custom
       # set type that is able to efficiently model the "all values" case.
       class FilterValueSetExtractor
-        def initialize(schema_names, all_values_set, &build_set_for_filter)
+        def initialize(schema_names, all_values_set, empty_set, &build_set_for_filter)
           @schema_names = schema_names
           @all_values_set = all_values_set
+          @empty_set = empty_set
           @build_set_for_filter = build_set_for_filter
         end
 
@@ -55,7 +56,7 @@ module ElasticGraph
           # outside the `map_reduce_sets` block below so we only do it once instead of N times.
           target_field_path_parts = target_field_path.split(".")
 
-          # Here we intersect the filter value setbecause when we have multiple `filter_hashes`,
+          # Here we intersect the filter value set, because when we have multiple `filter_hashes`,
           # the filters are ANDed together. Only documents that match ALL the filters will be
           # returned. Therefore, we want the intersection of filter value sets.
           map_reduce_sets(filter_hashes, :intersection, negate: false) do |filter_hash|
@@ -83,7 +84,7 @@ module ElasticGraph
         # to a particular field.
         def filter_value_set_for_filter_hash_entry(field_or_op, filter_value, target_field_path_parts, traversed_field_path_parts, negate:)
           if filter_value.nil?
-            # Any filter with a `nil` value is effectively ignored by our filtering logic, so we need
+            # Any filter with a `nil` value is effectively treated as `true` by our filtering logic, so we need
             # to return our `@all_values_set` to indicate this filter matches all documents.
             @all_values_set
           elsif field_or_op == @schema_names.not
@@ -105,6 +106,11 @@ module ElasticGraph
 
         # Determines the set of filter values for an `any_of` clause, which is used for ORing multiple filters together.
         def filter_value_set_for_any_of(filter_hashes, target_field_path_parts, traversed_field_path_parts, negate:)
+          # Here we treat `any_of: []` as matching no values.
+          if filter_hashes.empty?
+            return negate ? @all_values_set : @empty_set
+          end
+
           # Here we union the filter value sets because `any_of` represents an OR. If we can determine specific
           # filter values for all `any_of` clauses, we will OR them together. Alternately, if we cannot
           # determine specific filter values for any clauses, we will union `@all_values_set`,

data/lib/elastic_graph/graphql/http_endpoint.rb

@@ -62,7 +62,7 @@ module ElasticGraph
 
           HTTPResponse.json(200, result.to_h)
         end
-      rescue RequestExceededDeadlineError
+      rescue Errors::RequestExceededDeadlineError
         HTTPResponse.error(504, "Search exceeded requested timeout.")
       end
 
@@ -130,6 +130,10 @@ module ElasticGraph
         # Ignore an empty string operationName.
         params = params.merge("operationName" => nil) if params["operationName"] && params["operationName"].empty?
 
+        if (variables = params["variables"]) && !variables.is_a?(::Hash)
+          return HTTPResponse.error(400, "`variables` must be a JSON object but was not.")
+        end
+
         yield params
       end
 

data/lib/elastic_graph/graphql/query_adapter/filters.rb

@@ -101,7 +101,7 @@ module ElasticGraph
 
         def filter_value_set_extractor
           @filter_value_set_extractor ||=
-            Filtering::FilterValueSetExtractor.new(schema_element_names, IncludesNilSet) do |operator, filter_value|
+            Filtering::FilterValueSetExtractor.new(schema_element_names, IncludesNilSet, ExcludesNilSet) do |operator, filter_value|
               if operator == :equal_to_any_of && filter_value.include?(nil)
                 IncludesNilSet
               else

data/lib/elastic_graph/graphql/query_executor.rb

@@ -29,7 +29,7 @@ module ElasticGraph
       # Executes the given `query_string` using the provided `variables`.
       #
       # `timeout_in_ms` can be provided to limit how long the query runs for. If the timeout
-      # is exceeded, `RequestExceededDeadlineError` will be raised. Note that `timeout_in_ms`
+      # is exceeded, `Errors::RequestExceededDeadlineError` will be raised. Note that `timeout_in_ms`
       # does not provide an absolute guarantee that the query will take no longer than the
       # provided value; it is only used to halt datastore queries. In process computation
       # can make the total query time exceeded the specified timeout.

data/lib/elastic_graph/graphql/resolvers/query_source.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "graphql/dataloader/source"
+require "graphql"
 
 module ElasticGraph
   class GraphQL

data/lib/elastic_graph/graphql/resolvers/resolvable_value.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/support/memoizable_data"
 
 module ElasticGraph
@@ -23,7 +23,6 @@ module ElasticGraph
           Support::MemoizableData.define(:schema_element_names, *fields) do
             # @implements ResolvableValueClass
             include ResolvableValue
-            # @type var block: (^() -> void)?
             class_exec(&block) if block
           end
         end
@@ -48,7 +47,7 @@ module ElasticGraph
 
         def canonical_name_for(name, element_type)
           schema_element_names.canonical_name_for(name) ||
-            raise(SchemaError, "#{element_type} `#{name}` is not a defined schema element")
+            raise(Errors::SchemaError, "#{element_type} `#{name}` is not a defined schema element")
         end
       end
     end

data/lib/elastic_graph/graphql/schema/arguments.rb

@@ -58,7 +58,7 @@ module ElasticGraph
                 # extra memory allocation and GC for the hash.
                 arg_defn = arg_defns.find do |a|
                   a.keyword == key
-                end || raise(SchemaError, "Cannot find an argument definition for #{key.inspect} on `#{args_owner.name}`")
+                end || raise(Errors::SchemaError, "Cannot find an argument definition for #{key.inspect} on `#{args_owner.name}`")
 
                 next_owner = arg_defn.type.unwrap
                 accumulator[arg_defn.name] = to_schema_form(value, next_owner)

data/lib/elastic_graph/graphql/schema/enum_value.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 
 module ElasticGraph
   class GraphQL
@@ -15,7 +15,7 @@ module ElasticGraph
       class EnumValue < ::Data.define(:name, :type, :runtime_metadata)
         def sort_clauses
           sort_clause = runtime_metadata&.sort_field&.then { |sf| {sf.field_path => {"order" => sf.direction.to_s}} } ||
-            raise(SchemaError, "Runtime metadata provides no `sort_field` for #{type.name}.#{name} enum value.")
+            raise(Errors::SchemaError, "Runtime metadata provides no `sort_field` for #{type.name}.#{name} enum value.")
 
           [sort_clause]
         end

data/lib/elastic_graph/graphql/schema/field.rb

@@ -6,7 +6,7 @@
 #
 # frozen_string_literal: true
 
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/schema/relation_join"
 require "elastic_graph/graphql/schema/arguments"
 
@@ -126,7 +126,7 @@ module ElasticGraph
         def sort_argument_type
           @sort_argument_type ||= begin
             graphql_argument = @graphql_field.arguments.fetch(schema_element_names.order_by) do
-              raise SchemaError, "`#{schema_element_names.order_by}` argument not defined for field `#{parent_type.name}.#{name}`."
+              raise Errors::SchemaError, "`#{schema_element_names.order_by}` argument not defined for field `#{parent_type.name}.#{name}`."
             end
             @schema.type_from(graphql_argument.type.unwrap)
           end

data/lib/elastic_graph/graphql/schema/type.rb

@@ -7,7 +7,7 @@
 # frozen_string_literal: true
 
 require "elastic_graph/datastore_core/index_definition"
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/schema/field"
 require "elastic_graph/graphql/schema/enum_value"
 require "forwardable"
@@ -118,7 +118,7 @@ module ElasticGraph
         rescue KeyError => e
           msg = "No field named #{field_name} (on type #{name}) could be found"
           msg += "; Possible alternatives: [#{e.corrections.join(", ").delete('"')}]." if e.corrections.any?
-          raise NotFoundError, msg
+          raise Errors::NotFoundError, msg
         end
 
         def enum_value_named(enum_value_name)
@@ -233,7 +233,7 @@ module ElasticGraph
         rescue KeyError => e
           msg = "No enum value named #{enum_value_name} (on type #{name}) could be found"
           msg += "; Possible alternatives: [#{e.corrections.join(", ").delete('"')}]." if e.corrections.any?
-          raise NotFoundError, msg
+          raise Errors::NotFoundError, msg
         end
 
         def build_fields_by_name_hash(schema, graphql_type)

data/lib/elastic_graph/graphql/schema.rb

@@ -10,7 +10,7 @@ require "digest/md5"
 require "forwardable"
 require "graphql"
 require "elastic_graph/constants"
-require "elastic_graph/error"
+require "elastic_graph/errors"
 require "elastic_graph/graphql/monkey_patches/schema_field"
 require "elastic_graph/graphql/monkey_patches/schema_object"
 require "elastic_graph/graphql/schema/field"
@@ -29,7 +29,7 @@ module ElasticGraph
         scalar_types.to_set.union(introspection_types)
       )
 
-      attr_reader :element_names, :defined_types, :config, :graphql_schema, :runtime_metadata
+      attr_reader :element_names, :config, :graphql_schema, :runtime_metadata
 
       def initialize(
         graphql_schema_string:,
@@ -53,7 +53,6 @@ module ElasticGraph
           )
         end
 
-        @types_by_name = Hash.new { |hash, key| hash[key] = lookup_type_by_name(key) }
         @build_resolver = build_resolver
 
         # Note: as part of loading the schema, the GraphQL gem may use the resolver (such
@@ -68,7 +67,7 @@ module ElasticGraph
 
         # Pre-load all defined types so that all field extras can get configured as part
         # of loading the schema, before we execute the first query.
-        @defined_types = build_defined_types_array(@graphql_schema)
+        @types_by_name = build_types_by_name
       end
 
       def type_from(graphql_type)
@@ -80,7 +79,11 @@ module ElasticGraph
       # get type objects for wrapped types, but you need to get it from a field object of that
       # type.
       def type_named(type_name)
-        @types_by_name[type_name.to_s]
+        @types_by_name.fetch(type_name.to_s)
+      rescue KeyError => e
+        msg = "No type named #{type_name} could be found"
+        msg += "; Possible alternatives: [#{e.corrections.join(", ").delete('"')}]." if e.corrections.any?
+        raise Errors::NotFoundError, msg
       end
 
       def document_type_stored_in(index_definition_name)
@@ -88,7 +91,7 @@ module ElasticGraph
           if index_definition_name.include?(ROLLOVER_INDEX_INFIX_MARKER)
             raise ArgumentError, "`#{index_definition_name}` is the name of a rollover index; pass the name of the parent index definition instead."
           else
-            raise NotFoundError, "The index definition `#{index_definition_name}` does not appear to exist. Is it misspelled?"
+            raise Errors::NotFoundError, "The index definition `#{index_definition_name}` does not appear to exist. Is it misspelled?"
           end
         end
       end
@@ -106,6 +109,10 @@ module ElasticGraph
         @indexed_document_types ||= defined_types.select(&:indexed_document?)
       end
 
+      def defined_types
+        @defined_types ||= @types_by_name.except(*BUILT_IN_TYPE_NAMES).values
+      end
+
       def to_s
         "#<#{self.class.name} 0x#{__id__.to_s(16)} indexed_document_types=#{indexed_document_types.map(&:name).sort.to_s.delete(":")}>"
       end
@@ -128,31 +135,21 @@ module ElasticGraph
         def_delegators :resolver, :call, :resolve_type, :coerce_input, :coerce_result
       end
 
-      def lookup_type_by_name(type_name)
-        type_from(@graphql_schema.types.fetch(type_name))
-      rescue KeyError => e
-        msg = "No type named #{type_name} could be found"
-        msg += "; Possible alternatives: [#{e.corrections.join(", ").delete('"')}]." if e.corrections.any?
-        raise NotFoundError, msg
-      end
-
       def resolver
         @resolver ||= @build_resolver.call(self)
       end
 
-      def build_defined_types_array(graphql_schema)
-        graphql_schema
-          .types
-          .values
-          .reject { |t| BUILT_IN_TYPE_NAMES.include?(t.graphql_name) }
-          .map { |t| type_named(t.graphql_name) }
+      def build_types_by_name
+        graphql_schema.types.transform_values do |graphql_type|
+          @types_by_graphql_type[graphql_type]
+        end
       end
 
       def indexed_document_types_by_index_definition_name
         @indexed_document_types_by_index_definition_name ||= indexed_document_types.each_with_object({}) do |type, hash|
           type.index_definitions.each do |index_def|
             if hash.key?(index_def.name)
-              raise SchemaError, "DatastoreCore::IndexDefinition #{index_def.name} is used multiple times: #{type} vs #{hash[index_def.name]}"
+              raise Errors::SchemaError, "DatastoreCore::IndexDefinition #{index_def.name} is used multiple times: #{type} vs #{hash[index_def.name]}"
             end
 
             hash[index_def.name] = type

data/lib/elastic_graph/graphql.rb

@@ -142,7 +142,13 @@ module ElasticGraph
     def graphql_gem_plugins
       @graphql_gem_plugins ||= begin
         require "graphql"
-        {::GraphQL::Dataloader => {}}
+        {
+          # We depend on this to avoid N+1 calls to the datastore.
+          ::GraphQL::Dataloader => {},
+          # This is new in the graphql-ruby 2.4 release, and will be required in the future.
+          # We pass `preload: true` because the way we handle the schema depends on it being preloaded.
+          ::GraphQL::Schema::Visibility => {preload: true}
+        }
       end
     end
 
@@ -244,6 +250,9 @@ module ElasticGraph
     # at boot time instead of deferring dependency loading until we handle the first query. In other environments (such as tests),
     # it's nice to load dependencies when needed.
     def load_dependencies_eagerly
+      require "graphql"
+      ::GraphQL.eager_load!
+
       # run a simple GraphQL query to force load any dependencies needed to handle GraphQL queries
       graphql_query_executor.execute(EAGER_LOAD_QUERY, client: Client::ELASTICGRAPH_INTERNAL)
       graphql_http_endpoint # force load this too.

metadata

@@ -1,14 +1,16 @@
 --- !ruby/object:Gem::Specification
 name: elasticgraph-graphql
 version: !ruby/object:Gem::Version
-  version: 0.18.0.4
+  version: 0.19.0.0.rc1
 platform: ruby
 authors:
 - Myron Marston
+- Ben VandenBos
+- Block Engineering
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-09-06 00:00:00.000000000 Z
+date: 2024-12-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rubocop-factory_bot
@@ -44,42 +46,42 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '3.0'
+        version: '3.1'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '3.0'
+        version: '3.1'
 - !ruby/object:Gem::Dependency
   name: standard
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.40.0
+        version: 1.41.0
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.40.0
+        version: 1.41.0
 - !ruby/object:Gem::Dependency
   name: steep
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.7'
+        version: '1.8'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.7'
+        version: '1.8'
 - !ruby/object:Gem::Dependency
   name: coderay
   requirement: !ruby/object:Gem::Requirement
@@ -134,14 +136,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.12'
+        version: '0.13'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.12'
+        version: '0.13'
 - !ruby/object:Gem::Dependency
   name: simplecov
   requirement: !ruby/object:Gem::Requirement
@@ -252,126 +254,126 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '3.4'
+        version: '3.5'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '3.4'
+        version: '3.5'
 - !ruby/object:Gem::Dependency
   name: elasticgraph-datastore_core
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: elasticgraph-schema_artifacts
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: graphql
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.3.14
+        version: 2.4.5
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.3.14
+        version: 2.4.5
 - !ruby/object:Gem::Dependency
   name: elasticgraph-admin
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: elasticgraph-elasticsearch
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: elasticgraph-opensearch
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: elasticgraph-indexer
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 - !ruby/object:Gem::Dependency
   name: elasticgraph-schema_definition
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 0.18.0.4
+        version: 0.19.0.0.rc1
 description: 
 email:
 - myron@squareup.com
@@ -459,10 +461,15 @@ files:
 - lib/elastic_graph/graphql/schema/type.rb
 - script/dump_time_zones
 - script/dump_time_zones.java
-homepage: 
+homepage: https://block.github.io/elasticgraph/
 licenses:
 - MIT
 metadata:
+  bug_tracker_uri: https://github.com/block/elasticgraph/issues
+  changelog_uri: https://github.com/block/elasticgraph/releases/tag/v0.19.0.0.rc1
+  documentation_uri: https://block.github.io/elasticgraph/docs/main/
+  homepage_uri: https://block.github.io/elasticgraph/
+  source_code_uri: https://github.com/block/elasticgraph/tree/v0.19.0.0.rc1/elasticgraph-graphql
   gem_category: core
 post_install_message: 
 rdoc_options: []