datadog 2.21.0 → 2.38.0

data/lib/datadog/open_feature/exposures/event.rb

@@ -0,0 +1,60 @@
+# frozen_string_literal: true
+
+require_relative '../../core/utils/time'
+
+module Datadog
+  module OpenFeature
+    module Exposures
+      # A data model for an exposure event
+      module Event
+        TARGETING_KEY_FIELD = 'targeting_key'
+        ALLOWED_FIELD_TYPES = [String, Integer, Float, TrueClass, FalseClass].freeze
+
+        class << self
+          def cache_key(result, flag_key:, context:)
+            "#{flag_key}:#{context.targeting_key}"
+          end
+
+          def cache_value(result, flag_key:, context:)
+            "#{result.allocation_key}:#{result.variant}"
+          end
+
+          def build(result, flag_key:, context:)
+            {
+              timestamp: current_timestamp_ms,
+              allocation: {
+                key: result.allocation_key
+              },
+              flag: {
+                key: flag_key
+              },
+              variant: {
+                key: result.variant
+              },
+              subject: {
+                id: context.targeting_key,
+                attributes: extract_attributes(context)
+              }
+            }.freeze
+          end
+
+          private
+
+          # NOTE: We take all filds of the context that does not support nesting
+          #       and will ignore targeting key as it will be set as `subject.id`
+          def extract_attributes(context)
+            context.fields.select do |key, value|
+              next false if key == TARGETING_KEY_FIELD
+
+              ALLOWED_FIELD_TYPES.include?(value.class)
+            end
+          end
+
+          def current_timestamp_ms
+            (Datadog::Core::Utils::Time.now.to_f * 1000).to_i
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/open_feature/exposures/reporter.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+require_relative 'event'
+require_relative 'deduplicator'
+
+module Datadog
+  module OpenFeature
+    module Exposures
+      # This class is responsible for reporting exposures to the Agent
+      class Reporter
+        def initialize(worker, telemetry:, logger:)
+          @worker = worker
+          @logger = logger
+          @telemetry = telemetry
+          @deduplicator = Deduplicator.new
+        end
+
+        # NOTE: Reporting expects evaluation context to be always present, but it
+        #       might be missing depending on the customer way of using flags evaluation API.
+        #       In addition to that the evaluation result must be marked for reporting.
+        def report(result, flag_key:, context:)
+          return false if context.nil?
+          return false unless result.log?
+
+          key = Event.cache_key(result, flag_key: flag_key, context: context)
+          value = Event.cache_value(result, flag_key: flag_key, context: context)
+          return false if @deduplicator.duplicate?(key, value)
+
+          event = Event.build(result, flag_key: flag_key, context: context)
+          @worker.enqueue(event)
+        rescue => e
+          @logger.debug { "OpenFeature: Failed to report resolution details: #{e.class}: #{e.message}" }
+          @telemetry.report(e, description: 'OpenFeature: Failed to report resolution details')
+
+          false
+        end
+      end
+    end
+  end
+end

data/lib/datadog/open_feature/exposures/worker.rb

@@ -0,0 +1,116 @@
+# frozen_string_literal: true
+
+require_relative '../../core/utils/time'
+require_relative '../../core/workers/queue'
+require_relative '../../core/workers/polling'
+
+require_relative 'buffer'
+require_relative 'batch_builder'
+
+module Datadog
+  module OpenFeature
+    module Exposures
+      # This class is responsible for sending exposures to the Agent
+      class Worker
+        include Core::Workers::Queue
+        include Core::Workers::Polling
+
+        GRACEFUL_SHUTDOWN_EXTRA_SECONDS = 5
+        GRACEFUL_SHUTDOWN_WAIT_INTERVAL_SECONDS = 0.5
+
+        DEFAULT_FLUSH_INTERVAL_SECONDS = 30
+        DEFAULT_BUFFER_LIMIT = Buffer::DEFAULT_LIMIT
+
+        def initialize(
+          settings:,
+          transport:,
+          telemetry:,
+          logger:,
+          flush_interval_seconds: DEFAULT_FLUSH_INTERVAL_SECONDS,
+          buffer_limit: DEFAULT_BUFFER_LIMIT
+        )
+          @logger = logger
+          @transport = transport
+          @telemetry = telemetry
+          @batch_builder = BatchBuilder.new(settings)
+          @buffer_limit = buffer_limit
+
+          self.buffer = Buffer.new(buffer_limit)
+          self.fork_policy = Core::Workers::Async::Thread::FORK_POLICY_RESTART
+          self.loop_base_interval = flush_interval_seconds
+          self.enabled = true
+        end
+
+        def start
+          return if !enabled? || running?
+
+          perform
+        end
+
+        def stop(force_stop = false, timeout = Core::Workers::Polling::DEFAULT_SHUTDOWN_TIMEOUT)
+          buffer.close if running?
+
+          super
+        end
+
+        def enqueue(event)
+          buffer.push(event)
+          start unless running?
+
+          true
+        end
+
+        def dequeue
+          [buffer.pop, buffer.dropped_count]
+        end
+
+        def perform(*args)
+          events, dropped = args
+          send_events(Array(events), dropped.to_i)
+        end
+
+        def graceful_shutdown
+          return false unless enabled? || !run_loop?
+
+          self.enabled = false
+
+          started = Core::Utils::Time.get_time
+          wait_time = loop_base_interval + GRACEFUL_SHUTDOWN_EXTRA_SECONDS
+
+          loop do
+            break if buffer.empty? && !in_iteration?
+
+            sleep(GRACEFUL_SHUTDOWN_WAIT_INTERVAL_SECONDS)
+            break if Core::Utils::Time.get_time - started > wait_time
+          end
+
+          stop(true)
+        end
+
+        private
+
+        def send_events(events, dropped)
+          return if events.empty?
+
+          if dropped.positive?
+            @logger.debug { "OpenFeature: Resolution details worker dropped #{dropped} event(s) due to full buffer" }
+          end
+
+          payload = @batch_builder.payload_for(events)
+          response = @transport.send_exposures(payload)
+
+          unless response&.ok?
+            @logger.debug { "OpenFeature: Resolution details upload response was not OK: #{response.inspect}" }
+          end
+
+          response
+        rescue => e
+          @logger.debug { "OpenFeature: Failed to flush resolution details events: #{e.class}: #{e.message}" }
+          @telemetry.report(e, description: 'OpenFeature: Failed to flush resolution details events')
+
+          nil
+        end
+      end
+    end
+  end
+end

data/lib/datadog/open_feature/ext.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Datadog
+  module OpenFeature
+    module Ext
+      ERROR = 'ERROR'
+      DEFAULT = 'DEFAULT'
+      INITIALIZING = 'INITIALIZING'
+      UNKNOWN_TYPE = 'UNKNOWN_TYPE'
+      PARSE_ERROR = 'PARSE_ERROR'
+      GENERAL = 'GENERAL'
+      PROVIDER_FATAL = 'PROVIDER_FATAL'
+      PROVIDER_NOT_READY = 'PROVIDER_NOT_READY'
+    end
+  end
+end

data/lib/datadog/open_feature/flag_evaluation/aggregator.rb

@@ -0,0 +1,286 @@
+# frozen_string_literal: true
+
+module Datadog
+  module OpenFeature
+    module FlagEvaluation
+      # Two-tier aggregation for EVP flagevaluation events.
+      #
+      # Two-tier design:
+      # - full-tier  key: (flag_key, variant, allocation_key, runtime_default, error_message, targeting_key, canonical_context_key)
+      # - degraded-tier key: (flag_key, variant, allocation_key, runtime_default, error_message)
+      # - Drop-and-count when degraded tier is full
+      # - canonical_context_key: sorted type-tagged length-delimited encoding (no hash digest)
+      # - Caps: global_cap=131_072 / per_flag_cap=10_000 / degraded_cap=32_768
+      # - Context pruning: 256 fields / 256 chars (matches flageval-worker backend limits)
+      class Aggregator
+        MAX_CONTEXT_FIELDS = 256
+        MAX_FIELD_LENGTH = 256
+        MAX_CONTEXT_DEPTH = 32
+
+        # Type tags so values of different Ruby types never collide in the canonical key.
+        CTX_TAG_STRING = 's'
+        CTX_TAG_BOOL = 'b'
+        CTX_TAG_INTEGER = 'i'
+        CTX_TAG_FLOAT = 'f'
+        CTX_TAG_OTHER = 'o'
+
+        EVAL_SCALE_TARGET_FLAGS = 2_500
+        EVAL_SCALE_FULL_BUCKETS_PER_FLAG = 50
+        EVAL_SCALE_USERS_PER_FLAG = 1_000
+        EVAL_SCALE_PER_FLAG_HEADROOM_MULTIPLIER = 10
+        EVAL_SCALE_DEGRADED_BUCKETS_PER_FLAG = 10
+        EVAL_SCALE_FULL_BUCKET_TARGET = EVAL_SCALE_TARGET_FLAGS * EVAL_SCALE_FULL_BUCKETS_PER_FLAG
+        EVAL_SCALE_PER_FLAG_BUCKET_TARGET = EVAL_SCALE_PER_FLAG_HEADROOM_MULTIPLIER * EVAL_SCALE_USERS_PER_FLAG
+        EVAL_SCALE_DEGRADED_BUCKET_TARGET = EVAL_SCALE_TARGET_FLAGS * EVAL_SCALE_DEGRADED_BUCKETS_PER_FLAG
+
+        DEFAULT_GLOBAL_CAP = 131_072
+        DEFAULT_PER_FLAG_CAP = EVAL_SCALE_PER_FLAG_BUCKET_TARGET
+        DEFAULT_DEGRADED_CAP = 32_768
+
+        attr_reader :dropped_degraded_overflow
+
+        def initialize(
+          global_cap: DEFAULT_GLOBAL_CAP,
+          per_flag_cap: DEFAULT_PER_FLAG_CAP,
+          degraded_cap: DEFAULT_DEGRADED_CAP
+        )
+          @global_cap = global_cap
+          @per_flag_cap = per_flag_cap
+          @degraded_cap = degraded_cap
+
+          @mutex = Mutex.new
+          # full-tier: Array key -> Hash entry
+          @full = {}
+          # degraded-tier: Array key -> Hash entry
+          @degraded = {}
+          # per-flag full-bucket count for per_flag_cap enforcement
+          @per_flag_full = Hash.new(0)
+          @global_count = 0
+          @dropped_degraded_overflow = 0
+        end
+
+        # Record one evaluation event. Thread-safe. Called from the background writer.
+        def record(
+          flag_key:, variant:, allocation_key:, targeting_key:, eval_time_ms:, attrs:, error_message: nil,
+          runtime_default: nil
+        )
+          runtime_default = variant.nil? if runtime_default.nil?
+          runtime_default = !!runtime_default
+
+          # Normalize nil/empty strings
+          variant = variant.to_s
+          allocation_key = allocation_key.to_s
+          error_message = error_message.to_s
+          targeting_key = targeting_key.to_s
+
+          # Context pruning + canonical key (see prune_context and canonical_context_key).
+          # Runs in the background writer so caller eval threads do not pay the flatten/prune cost.
+          pruned_context = prune_context(attrs)
+          context_key = canonical_context_key(pruned_context)
+
+          full_key = [flag_key, variant, allocation_key, runtime_default, error_message, targeting_key, context_key]
+          evaluation_time_ms = eval_time_ms.to_i
+
+          @mutex.synchronize do
+            # --- Full tier ---
+            if (entry = @full[full_key])
+              observe(entry, evaluation_time_ms)
+              return
+            end
+
+            per_flag_count = @per_flag_full[flag_key]
+            if per_flag_count >= @per_flag_cap
+              add_to_degraded(flag_key, variant, allocation_key, runtime_default, error_message, evaluation_time_ms)
+              return
+            end
+
+            # Count the full-tier attempt before checking the global cap so per-flag overflow stays
+            # active even when the global full-tier cap is already saturated.
+            @per_flag_full[flag_key] = per_flag_count + 1
+
+            if @global_count < @global_cap
+              entry = new_entry(
+                evaluation_time_ms,
+                runtime_default: runtime_default,
+                error_message: error_message,
+                targeting_key: targeting_key,
+                context_attrs: pruned_context
+              )
+              @full[full_key] = entry
+              @global_count += 1
+            else
+              # Route to degraded tier
+              add_to_degraded(flag_key, variant, allocation_key, runtime_default, error_message, evaluation_time_ms)
+            end
+          end
+        end
+
+        # Flush aggregation maps, reset state, return snapshot.
+        # Returns { full: Hash, degraded: Hash, dropped_degraded_overflow: Integer }.
+        # The overflow count is included in the snapshot so the caller can EMIT it before it is
+        # reset — the count is never reset-without-emit (backpressure stays observable).
+        def flush_and_reset
+          @mutex.synchronize do
+            full_snapshot = @full
+            degraded_snapshot = @degraded
+            dropped_snapshot = @dropped_degraded_overflow
+
+            @full = {}
+            @degraded = {}
+            @per_flag_full = Hash.new(0)
+            @global_count = 0
+            @dropped_degraded_overflow = 0
+
+            {full: full_snapshot, degraded: degraded_snapshot, dropped_degraded_overflow: dropped_snapshot}
+          end
+        end
+
+        # Prune context: keep first MAX_CONTEXT_FIELDS fields (sorted), skip string values >256 chars.
+        # Keys are sorted before pruning to ensure deterministic subset selection.
+        def prune_context(attrs)
+          self.class.prune_context(attrs)
+        end
+
+        def self.prune_context(attrs)
+          flattened_context = flatten_context(attrs)
+          return {} if flattened_context.empty?
+
+          pruned_context = {}
+          count = 0
+          flattened_context.keys.sort.each do |key|
+            break if count >= MAX_CONTEXT_FIELDS
+
+            value = flattened_context[key]
+            # Skip oversized string values (mirrors flageval-worker pruning behavior).
+            next if value.is_a?(String) && value.length > MAX_FIELD_LENGTH
+
+            pruned_context[key] = value
+            count += 1
+          end
+          pruned_context
+        end
+
+        def self.flatten_context(attrs)
+          return {} unless attrs.is_a?(Hash) && !attrs.empty?
+
+          flattened_context = {}
+          seen = {attrs.object_id => true}
+          attrs.each do |key, value|
+            flatten_value(key.to_s, value, flattened_context, seen, 0)
+          end
+          flattened_context
+        end
+
+        # Canonical context key: sorted type-tagged length-delimited encoding.
+        # Each field is: 8-byte big-endian key length + key bytes + type-tag byte +
+        #                8-byte big-endian value length + value bytes.
+        # No hash digest — the key IS the full encoding (collision-free, no FNV).
+        def canonical_context_key(attrs)
+          return '' if attrs.nil? || attrs.empty?
+
+          buffer = String.new('', encoding: Encoding::BINARY)
+          attrs.keys.sort.each do |key|
+            value = attrs[key]
+            buffer << length_delimited(key.to_s)
+            buffer << context_value_bytes(value)
+          end
+          buffer
+        end
+
+        def self.flatten_value(prefix, value, output, seen, depth)
+          return if depth > MAX_CONTEXT_DEPTH
+
+          case value
+          when Hash
+            object_id = value.object_id
+            return if seen[object_id]
+
+            seen[object_id] = true
+            value.each do |key, child_value|
+              flatten_value("#{prefix}.#{key}", child_value, output, seen, depth + 1)
+            end
+            seen.delete(object_id)
+          when Array
+            object_id = value.object_id
+            return if seen[object_id]
+
+            seen[object_id] = true
+            value.each_with_index do |child_value, index|
+              flatten_value("#{prefix}.#{index}", child_value, output, seen, depth + 1)
+            end
+            seen.delete(object_id)
+          else
+            output[prefix] = value unless value.nil?
+          end
+        end
+        private_class_method :flatten_value
+
+        private
+
+        def context_value_bytes(value)
+          tag, encoded = case value
+          when String then [CTX_TAG_STRING, value.to_s]
+          when TrueClass, FalseClass then [CTX_TAG_BOOL, value.to_s]
+          when Integer then [CTX_TAG_INTEGER, value.to_s]
+          when Float then [CTX_TAG_FLOAT, value.to_s]
+          else [CTX_TAG_OTHER, value.to_s]
+          end
+          String.new(tag, encoding: Encoding::BINARY) + length_delimited(encoded)
+        end
+
+        # 8-byte big-endian length prefix + raw bytes. Unambiguous field boundary.
+        def length_delimited(string)
+          bytes = string.encode(Encoding::BINARY, invalid: :replace, undef: :replace)
+          byte_length = bytes.bytesize
+          # Build 8-byte big-endian length
+          length_bytes = String.new('', encoding: Encoding::BINARY)
+          8.times do |index|
+            length_bytes.prepend(((byte_length >> (8 * index)) & 0xFF).chr(Encoding::BINARY))
+          end
+          length_bytes + bytes
+        end
+
+        def new_entry(evaluation_time_ms, runtime_default:, error_message: nil, targeting_key: nil, context_attrs: nil)
+          {
+            count: 1,
+            first_evaluation: evaluation_time_ms,
+            last_evaluation: evaluation_time_ms,
+            runtime_default: runtime_default,
+            error_message: error_message,
+            targeting_key: targeting_key,
+            context_attrs: context_attrs,
+          }
+        end
+
+        def observe(entry, evaluation_time_ms)
+          entry[:count] += 1
+          entry[:first_evaluation] = evaluation_time_ms if evaluation_time_ms < entry[:first_evaluation]
+          entry[:last_evaluation] = evaluation_time_ms if evaluation_time_ms > entry[:last_evaluation]
+        end
+
+        def add_to_degraded(flag_key, variant, allocation_key, runtime_default, error_message, evaluation_time_ms)
+          degraded_key = [flag_key, variant, allocation_key, runtime_default, error_message]
+
+          if (entry = @degraded[degraded_key])
+            observe(entry, evaluation_time_ms)
+            return
+          end
+
+          # New degraded bucket — check degraded_cap (terminal tier)
+          if @degraded.size >= @degraded_cap
+            # Terminal tier full — drop and count (explicit overflow counter)
+            @dropped_degraded_overflow += 1
+            return
+          end
+
+          # Degraded entry omits targeting_key + context_attrs (schema omitempty fields)
+          @degraded[degraded_key] = new_entry(
+            evaluation_time_ms,
+            runtime_default: runtime_default,
+            error_message: error_message
+          )
+        end
+      end
+    end
+  end
+end