datadog 2.14.0 → 2.16.0

data/lib/datadog/error_tracking/component.rb

@@ -0,0 +1,167 @@
+# frozen_string_literal: true
+
+require 'set'
+require_relative 'collector'
+require_relative 'filters'
+
+module Datadog
+  module ErrorTracking
+    # Component for Error Tracking.
+    #
+    # Only one instance of the Component should ever be active.
+    #
+    # The component instance records every handled exceptions from the configured scopes
+    # (user, third_party packages, specified files or everything).
+    class Component
+      LOCK = Mutex.new
+
+      class << self
+        def build(settings, tracer, logger)
+          return if !settings.respond_to?(:error_tracking) || (settings.error_tracking.handled_errors.nil? &&
+            settings.error_tracking.handled_errors_include.empty?)
+
+          return unless environment_supported?(logger)
+
+          new(
+            tracer: tracer,
+            handled_errors: settings.error_tracking.handled_errors,
+            handled_errors_include: settings.error_tracking.handled_errors_include,
+          ).tap(&:start)
+        end
+
+        def environment_supported?(logger)
+          if RUBY_ENGINE != 'ruby'
+            logger.warn("error tracking: cannot enable error tracking: MRI is required, but running on #{RUBY_ENGINE}")
+            false
+          elsif RUBY_VERSION < '2.6'
+            logger.warn(
+              "error tracking: cannot enable error tracking: Ruby 2.6+ is required, but running
+              on #{RUBY_VERSION}"
+            )
+            false
+          else
+            true
+          end
+        end
+      end
+
+      def initialize(tracer:, handled_errors:, handled_errors_include:)
+        @tracer = tracer
+
+        # Hash containing the paths to the instrumented files
+        @instrumented_files = Set.new unless handled_errors_include.empty?
+        # Array containing file paths, file names and gems names to instrument.
+        # This is coming from the DD_ERROR_TRACKING_HANDLED_ERRORS_INCLUDE env variable
+        @handled_errors_include = handled_errors_include
+
+        # Filter function is used to filter out the exception
+        # we do not want to report. For instance exception from gems.
+        @filter_function = Filters.generate_filter(handled_errors, @instrumented_files)
+
+        # :rescue event was added in Ruby 3.3
+        #
+        # Before Ruby3.3 the TracePoint listen for :raise events.
+        # If an error is not handled, we will delete the according
+        # span event in the collector.
+        event = (RUBY_VERSION >= '3.3') ? :rescue : :raise
+
+        # This TracePoint is in charge of capturing the handled exceptions
+        # and of adding the corresponding span events to the collector
+        @handled_exc_tracker = create_exc_tracker_trace_point(event)
+
+        if @instrumented_files
+          # The only thing we know about the handled errors is the path of the file
+          # in which the error was rescued. Therefore, we need to retrieve the path
+          # of the files the user want to instrument. This TracePoint is used for that
+          # purpose
+          @include_path_getter = create_script_compiled_trace_point
+        end
+      end
+
+      def create_exc_tracker_trace_point(event)
+        TracePoint.new(event) do |tp|
+          active_span = @tracer.active_span
+          if active_span
+            raised_exception = tp.raised_exception
+            # Note that in 3.2, this will give the path of where the error was raised
+            # which may cause the handled_error_include env variable to malfunction.
+            rescue_file_path = tp.path
+            if @filter_function.call(rescue_file_path)
+              span_event = generate_span_event(raised_exception)
+              LOCK.synchronize do
+                collector = active_span.get_collector_or_initialize { Collector.new }
+                collector.add_span_event(active_span, span_event, raised_exception)
+              end
+            end
+          end
+        end
+      end
+
+      def create_script_compiled_trace_point
+        TracePoint.new(:script_compiled) do |tp|
+          next if tp.eval_script
+
+          path = tp.instruction_sequence.path
+          next if path.nil?
+
+          @handled_errors_include.each do |file_to_instr|
+            # The user can provide either
+            # - absolute_path starting with '/'. In that case the path of the file
+            #   should begin with file_to_instr
+            # - a relative_path starting with './'. In that case, we extend the path
+            #   and it is the same as above
+            # - otherwise we just check if the name provided is in the path and is
+            #   either the name of a folder or of a ruby file.
+            regex =
+              if file_to_instr.start_with?('/')
+                %r{\A#{Regexp.escape(file_to_instr)}(?:/|\.rb\z|\z)}
+              elsif file_to_instr.start_with?('./')
+                abs_path = File.expand_path(file_to_instr)
+                %r{\A#{Regexp.escape(abs_path)}(?:/|\.rb\z|\z)}
+              else
+                %r{/#{Regexp.escape(file_to_instr)}(?:/|\.rb\z|\z)}
+              end
+
+            add_instrumented_file(path) if path.match?(regex)
+          end
+        end
+      end
+
+      # Starts the TracePoints.
+      #
+      # Enables the script_compiled TracePoint if handled_errors_include is not empty.
+      def start
+        @handled_exc_tracker.enable
+        @include_path_getter&.enable
+      end
+
+      # Shuts down error tracker.
+      #
+      # Disables the TracePoints.
+      def shutdown!
+        @handled_exc_tracker.disable
+        @include_path_getter&.disable
+      end
+
+      private
+
+      # Generates a span event from the exception info.
+      #
+      # The event follows the otel semantics.
+      # https://opentelemetry.io/docs/specs/otel/trace/exceptions/
+      def generate_span_event(exception)
+        formatted_exception = Datadog::Core::Error.build_from(exception)
+        attributes = {
+          'exception.type' => formatted_exception.type,
+          'exception.message' => formatted_exception.message,
+          'exception.stacktrace' => formatted_exception.backtrace
+        }
+        Datadog::Tracing::SpanEvent.new('exception', attributes: attributes)
+      end
+
+      def add_instrumented_file(file_path)
+        @instrumented_files&.add(file_path)
+      end
+    end
+  end
+end

data/lib/datadog/error_tracking/configuration/settings.rb

@@ -0,0 +1,63 @@
+# frozen_string_literal: true
+
+require_relative '../ext'
+
+module Datadog
+  module ErrorTracking
+    module Configuration
+      # Settings
+      module Settings
+        def self.extended(base)
+          base = base.singleton_class unless base.is_a?(Class)
+          add_settings!(base)
+        end
+
+        def self.add_settings!(base)
+          base.class_eval do
+            # Error Tracking specific configurations.
+            # @public_api
+            settings :error_tracking do
+              # Enable automatic reporting of handled errors and defines the scope
+              # for which to report errors: user code, gems, or both. Possible
+              # values are: all | user | third_party.
+              #
+              # @default 'DD_ERROR_TRACKING_HANDLED_ERRORS' environment variable, otherwise `nil`
+              # @return [String, nil]
+              option :handled_errors do |o|
+                o.type :string, nilable: true
+                o.default Ext::DEFAULT_HANDLED_ERRORS
+                o.env Ext::ENV_HANDLED_ERRORS
+                o.setter do |value|
+                  next value if Ext::VALID_HANDLED_ERRORS.include?(value)
+
+                  unless value.nil?
+                    Datadog.logger.warn(
+                      "Invalid handled errors scope: #{value}. " \
+                      "Supported values are: #{Ext::VALID_HANDLED_ERRORS.join(" | ")}. " \
+                      'Deactivating the feature.'
+                    )
+                  end
+
+                  Ext::DEFAULT_HANDLED_ERRORS
+                end
+              end
+
+              # Enable automatic reporting of handled errors and specify what files should be
+              # instrumented. The value is a list of comma separated paths, filenames or gem names.
+              # The paths can be absolute, starting with '/' or relative to directory in which the program
+              # is launched, starting with './'.
+              #
+              # @default 'DD_ERROR_TRACKING_HANDLED_ERRORS_MODULES' environment variable, otherwise `nil`
+              # @return [String, nil]
+              option :handled_errors_include do |o|
+                o.type :array
+                o.default []
+                o.env Ext::ENV_HANDLED_ERRORS_INCLUDE
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/error_tracking/configuration.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+module Datadog
+  module ErrorTracking
+    # Configuration for ErrorTracking
+    module Configuration
+    end
+  end
+end
+
+require_relative 'configuration/settings'

data/lib/datadog/error_tracking/ext.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Datadog
+  module ErrorTracking
+    # Error Tracking constants
+    module Ext
+      ENV_HANDLED_ERRORS = 'DD_ERROR_TRACKING_HANDLED_ERRORS'
+      ENV_HANDLED_ERRORS_INCLUDE = 'DD_ERROR_TRACKING_HANDLED_ERRORS_INCLUDE'
+      HANDLED_ERRORS_ALL = 'all'
+      HANDLED_ERRORS_USER = 'user'
+      HANDLED_ERRORS_THIRD_PARTY = 'third_party'
+      DEFAULT_HANDLED_ERRORS = nil
+      VALID_HANDLED_ERRORS = [HANDLED_ERRORS_ALL, HANDLED_ERRORS_USER, HANDLED_ERRORS_THIRD_PARTY].freeze
+      SPAN_EVENTS_HAS_EXCEPTION = '_dd.span_events.has_exception'
+      RUBY_VERSION_WITH_RESCUE_EVENT = '3.3'
+    end
+  end
+end

data/lib/datadog/error_tracking/extensions.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require_relative 'configuration'
+require_relative '../core/configuration'
+
+module Datadog
+  module ErrorTracking
+    # Extends Datadog tracing with ErrorTracking features
+    module Extensions
+      # Inject Error Tracking into global objects.
+      def self.activate!
+        Core::Configuration::Settings.extend(Configuration::Settings)
+      end
+    end
+  end
+end

data/lib/datadog/error_tracking/filters.rb

@@ -0,0 +1,77 @@
+# frozen_string_literal: true
+
+module Datadog
+  module ErrorTracking
+    # Based on configuration, the TracePoint listening to :rescue or :raise
+    # may report more handled errors than we want to report. Therefore we need
+    # a function to filter the events. As the filter function both depends
+    # on configuration and is called numerous time, we generate it during
+    # during the initialization of the feature to have the best performance
+    # possible.
+    #
+    # @api private
+    module Filters
+      module_function
+
+      def get_gem_name(file_path)
+        regex = %r{gems/([^/]+)-\d}
+        regex_match = regex.match(file_path)
+        return unless regex_match
+
+        gem_name = regex_match[1]
+
+        begin
+          Gem::Specification.find_by_name(gem_name) # steep:ignore
+        rescue Gem::MissingSpecError
+          nil
+        end
+      end
+
+      def user_code?(file_path)
+        !get_gem_name(file_path)
+      end
+
+      def datadog_code?(file_path)
+        file_path.include?('lib/datadog/')
+      end
+
+      def third_party_code?(file_path)
+        gem_name = get_gem_name(file_path)
+        gem_name && gem_name != "datadog"
+      end
+
+      def file_included?(file_path, instrumented_files)
+        instrumented_files.include?(file_path)
+      end
+
+      # Generate the proc used in the TracePoint
+      def generate_filter(to_instrument_scope, handled_errors_include = nil)
+        case to_instrument_scope
+        # If DD_ERROR_TRACKING_HANDLED_ERRORS is set
+        when 'all'
+          proc { |file_path| !datadog_code?(file_path) }
+        when 'user'
+          # If DD_ERROR_TRACKING_HANDLED_ERRORS_INCLUDE is set
+          if handled_errors_include
+            proc { |file_path|
+              user_code?(file_path) || file_included?(file_path, handled_errors_include)
+            }
+          else
+            proc { |file_path| user_code?(file_path) }
+          end
+        when 'third_party'
+          if handled_errors_include
+            proc { |file_path|
+              third_party_code?(file_path) || file_included?(file_path, handled_errors_include)
+            }
+          else
+            proc { |file_path| third_party_code?(file_path) }
+          end
+        else
+          # If only DD_ERROR_TRACKING_HANDLED_ERRORS_INCLUDE is set
+          proc { |file_path| file_included?(file_path, handled_errors_include) }
+        end
+      end
+    end
+  end
+end

data/lib/datadog/error_tracking.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+require_relative 'error_tracking/collector'
+require_relative 'error_tracking/component'
+require_relative 'error_tracking/configuration'
+require_relative 'error_tracking/ext'
+require_relative 'error_tracking/extensions'
+require_relative 'error_tracking/filters'
+
+module Datadog
+  # Namespace for Datadog ErrorTracking.
+  #
+  # @api private
+  module ErrorTracking
+    # Expose ErrorTracking to global shared objects
+    Extensions.activate!
+  end
+end

data/lib/datadog/kit/identity.rb

@@ -70,7 +70,7 @@ module Datadog
             if Datadog::AppSec.active_context
               active_span.set_tag('_dd.appsec.user.collection_mode', 'sdk')
 
-              user = ::Datadog::AppSec::Instrumentation::Gateway::User.new(id, others[:login])
+              user = ::Datadog::AppSec::Instrumentation::Gateway::User.new(id, others[:login], session_id)
               ::Datadog::AppSec::Instrumentation.gateway.push('identity.set_user', user)
             end
           end

data/lib/datadog/profiling/collectors/info.rb

@@ -31,6 +31,9 @@ module Datadog
         # Instead of trying to figure out real process start time by checking
         # /proc or some other complex/non-portable way, approximate start time
         # by time of requirement of this file.
+        #
+        # Note: this does not use Core::Utils::Time.now because this constant
+        # gets initialized before a user has a chance to configure the library.
         START_TIME = Time.now.utc.freeze
 
         def collect_platform_info

data/lib/datadog/profiling/encoded_profile.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Profiling
+    # This class exists to wrap a ddog_prof_EncodedProfile into a Ruby object
+    #
+    # This class is not empty; all of this class is implemented as native code.
+    class EncodedProfile # rubocop:disable Lint/EmptyClass
+    end
+  end
+end

data/lib/datadog/profiling/exporter.rb

@@ -49,7 +49,7 @@ module Datadog
         @internal_metadata = internal_metadata
         # NOTE: At the time of this comment collected info does not change over time so we'll hardcode
         #       it on startup to prevent serializing the same info on every flush.
-        @info_json = JSON.fast_generate(info_collector.info).freeze
+        @info_json = JSON.generate(info_collector.info).freeze
       end
 
       def flush
@@ -57,7 +57,7 @@ module Datadog
         serialization_result = pprof_recorder.serialize
         return if serialization_result.nil?
 
-        start, finish, compressed_pprof, profile_stats = serialization_result
+        start, finish, encoded_profile, profile_stats = serialization_result
         @last_flush_finish_at = finish
 
         if duration_below_threshold?(start, finish)
@@ -70,8 +70,7 @@ module Datadog
         Flush.new(
           start: start,
           finish: finish,
-          pprof_file_name: Datadog::Profiling::Ext::Transport::HTTP::PPROF_DEFAULT_FILENAME,
-          pprof_data: compressed_pprof.to_s,
+          encoded_profile: encoded_profile,
           code_provenance_file_name: Datadog::Profiling::Ext::Transport::HTTP::CODE_PROVENANCE_FILENAME,
           code_provenance_data: uncompressed_code_provenance,
           tags_as_array: Datadog::Profiling::TagBuilder.call(settings: Datadog.configuration).to_a,

data/lib/datadog/profiling/ext.rb

@@ -26,7 +26,6 @@ module Datadog
           TAG_GIT_REPOSITORY_URL = "git.repository_url"
           TAG_GIT_COMMIT_SHA = "git.commit.sha"
 
-          PPROF_DEFAULT_FILENAME = "rubyprofile.pprof"
           CODE_PROVENANCE_FILENAME = "code-provenance.json"
         end
       end

data/lib/datadog/profiling/flush.rb

@@ -9,10 +9,9 @@ module Datadog
       attr_reader \
         :start,
         :finish,
-        :pprof_file_name,
-        :pprof_data, # gzipped pprof bytes
+        :encoded_profile,
         :code_provenance_file_name,
-        :code_provenance_data, # gzipped json bytes
+        :code_provenance_data,
         :tags_as_array,
         :internal_metadata_json,
         :info_json
@@ -20,8 +19,7 @@ module Datadog
       def initialize(
         start:,
         finish:,
-        pprof_file_name:,
-        pprof_data:,
+        encoded_profile:,
         code_provenance_file_name:,
         code_provenance_data:,
         tags_as_array:,
@@ -30,8 +28,7 @@ module Datadog
       )
         @start = start
         @finish = finish
-        @pprof_file_name = pprof_file_name
-        @pprof_data = pprof_data
+        @encoded_profile = encoded_profile
         @code_provenance_file_name = code_provenance_file_name
         @code_provenance_data = code_provenance_data
         @tags_as_array = tags_as_array

data/lib/datadog/profiling/http_transport.rb

@@ -20,34 +20,21 @@ module Datadog
             [:agent, agent_settings.url].freeze
           end
 
-        status, result = validate_exporter(exporter_configuration)
+        status, result = self.class._native_validate_exporter(exporter_configuration)
 
         raise(ArgumentError, "Failed to initialize transport: #{result}") if status == :error
       end
 
       def export(flush)
-        status, result = do_export(
-          exporter_configuration: exporter_configuration,
-          upload_timeout_milliseconds: @upload_timeout_milliseconds,
-
-          # why "timespec"?
-          # libdatadog represents time using POSIX's struct timespec, see
-          # https://www.gnu.org/software/libc/manual/html_node/Time-Types.html
-          # aka it represents the seconds part separate from the nanoseconds part
-          start_timespec_seconds: flush.start.tv_sec,
-          start_timespec_nanoseconds: flush.start.tv_nsec,
-          finish_timespec_seconds: flush.finish.tv_sec,
-          finish_timespec_nanoseconds: flush.finish.tv_nsec,
-
-          pprof_file_name: flush.pprof_file_name,
-          pprof_data: flush.pprof_data,
-          code_provenance_file_name: flush.code_provenance_file_name,
-          code_provenance_data: flush.code_provenance_data,
-
-          tags_as_array: flush.tags_as_array,
-          internal_metadata_json: flush.internal_metadata_json,
-
-          info_json: flush.info_json
+        status, result = self.class._native_do_export(
+          exporter_configuration,
+          @upload_timeout_milliseconds,
+          flush,
+          # TODO: This is going to be removed once we move to libdatadog 17
+          flush.start.tv_sec,
+          flush.start.tv_nsec,
+          flush.finish.tv_sec,
+          flush.finish.tv_nsec,
         )
 
         if status == :ok
@@ -77,42 +64,6 @@ module Datadog
         site && api_key && Core::Environment::VariableHelpers.env_to_bool(Profiling::Ext::ENV_AGENTLESS, false)
       end
 
-      def validate_exporter(exporter_configuration)
-        self.class._native_validate_exporter(exporter_configuration)
-      end
-
-      def do_export(
-        exporter_configuration:,
-        upload_timeout_milliseconds:,
-        start_timespec_seconds:,
-        start_timespec_nanoseconds:,
-        finish_timespec_seconds:,
-        finish_timespec_nanoseconds:,
-        pprof_file_name:,
-        pprof_data:,
-        code_provenance_file_name:,
-        code_provenance_data:,
-        tags_as_array:,
-        internal_metadata_json:,
-        info_json:
-      )
-        self.class._native_do_export(
-          exporter_configuration,
-          upload_timeout_milliseconds,
-          start_timespec_seconds,
-          start_timespec_nanoseconds,
-          finish_timespec_seconds,
-          finish_timespec_nanoseconds,
-          pprof_file_name,
-          pprof_data,
-          code_provenance_file_name,
-          code_provenance_data,
-          tags_as_array,
-          internal_metadata_json,
-          info_json,
-        )
-      end
-
       def config_without_api_key
         "#{exporter_configuration[0]}: #{exporter_configuration[1]}"
       end

data/lib/datadog/profiling/stack_recorder.rb

@@ -63,11 +63,11 @@ module Datadog
         status, result = @no_concurrent_synchronize_mutex.synchronize { self.class._native_serialize(self) }
 
         if status == :ok
-          start, finish, encoded_pprof, profile_stats = result
+          start, finish, encoded_profile, profile_stats = result
 
           Datadog.logger.debug { "Encoded profile covering #{start.iso8601} to #{finish.iso8601}" }
 
-          [start, finish, encoded_pprof, profile_stats]
+          [start, finish, encoded_profile, profile_stats]
         else
           error_message = result
 
@@ -82,9 +82,9 @@ module Datadog
         status, result = @no_concurrent_synchronize_mutex.synchronize { self.class._native_serialize(self) }
 
         if status == :ok
-          _start, _finish, encoded_pprof = result
+          _start, _finish, encoded_profile = result
 
-          encoded_pprof
+          encoded_profile
         else
           error_message = result
 

data/lib/datadog/profiling.rb

@@ -149,6 +149,7 @@ module Datadog
       require_relative 'profiling/collectors/thread_context'
       require_relative 'profiling/stack_recorder'
       require_relative 'profiling/exporter'
+      require_relative 'profiling/encoded_profile'
       require_relative 'profiling/flush'
       require_relative 'profiling/scheduler'
       require_relative 'profiling/tasks/setup'

data/lib/datadog/tracing/analytics.rb

@@ -16,7 +16,7 @@ module Datadog
         def set_measured(span_op, value = true)
           return if span_op.nil?
 
-          value = value == true || value == 1 ? 1 : 0 # rubocop:disable Style/MultipleComparison
+          value = value == true || value == 1 ? 1 : 0
           span_op.set_metric(Metadata::Ext::Analytics::TAG_MEASURED, value)
         end
       end

data/lib/datadog/tracing/contrib/active_record/integration.rb

@@ -57,7 +57,7 @@ module Datadog
           end
 
           def reset_resolver_cache
-            @resolver&.reset_cache
+            @resolver&.reset_cache if defined?(@resolver)
           end
 
           Contrib::Component.register('activerecord') do |_config|

data/lib/datadog/tracing/contrib/karafka/distributed/propagation.rb

@@ -31,6 +31,8 @@ module Datadog
                     Tracing::Distributed::Datadog.new(fetcher: Tracing::Distributed::Fetcher),
                   Tracing::Configuration::Ext::Distributed::PROPAGATION_STYLE_TRACE_CONTEXT =>
                     Tracing::Distributed::TraceContext.new(fetcher: Tracing::Distributed::Fetcher),
+                  Tracing::Configuration::Ext::Distributed::PROPAGATION_STYLE_BAGGAGE =>
+                    Tracing::Distributed::Baggage.new(fetcher: Tracing::Distributed::Fetcher),
                   Tracing::Configuration::Ext::Distributed::PROPAGATION_STYLE_NONE => Tracing::Distributed::None.new
                 },
                 propagation_style_inject: propagation_style_inject,

data/lib/datadog/tracing/contrib/karafka/monitor.rb

@@ -15,7 +15,7 @@ module Datadog
             worker.processed
           ].freeze
 
-          def instrument(event_id, payload = EMPTY_HASH, &block)
+          def instrument(event_id, payload = {}, &block)
             return super unless TRACEABLE_EVENTS.include?(event_id)
 
             Datadog::Tracing.trace(Ext::SPAN_WORKER_PROCESS) do |span|

data/lib/datadog/tracing/contrib/mongodb/configuration/settings.rb

@@ -48,6 +48,14 @@ module Datadog
               o.type :string, nilable: true
               o.env Ext::ENV_PEER_SERVICE
             end
+
+            # Serializes the command to JSON format, which is the desired format for the agent and Datadog UI.
+            # Setting this to false is deprecated and does not have any advantages.
+            option :json_command do |o|
+              o.type :bool
+              o.env Ext::ENV_JSON_COMMAND
+              o.default false
+            end
           end
         end
       end

data/lib/datadog/tracing/contrib/mongodb/ext.rb

@@ -13,6 +13,7 @@ module Datadog
           # @!visibility private
           ENV_ANALYTICS_ENABLED = 'DD_TRACE_MONGO_ANALYTICS_ENABLED'
           ENV_ANALYTICS_SAMPLE_RATE = 'DD_TRACE_MONGO_ANALYTICS_SAMPLE_RATE'
+          ENV_JSON_COMMAND = 'DD_TRACE_MONGO_JSON_COMMAND'
           DEFAULT_PEER_SERVICE_NAME = 'mongodb'
           SPAN_COMMAND = 'mongo.cmd'
           SPAN_TYPE_COMMAND = 'mongodb'