datadog 2.8.0 → 2.39.0

data/lib/datadog/appsec/security_engine/engine.rb

@@ -0,0 +1,197 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module SecurityEngine
+      # SecurityEngine::Engine creates WAF builder and manages its configuration.
+      # It also rebuilds WAF handle from the WAF builder when configuration changes.
+      class Engine
+        DEFAULT_RULES_CONFIG_PATH = "ASM_DD/default"
+        TELEMETRY_ACTIONS = %w[init update].freeze
+        DIAGNOSTICS_CONFIG_KEYS = %w[
+          rules
+          custom_rules
+          exclusions
+          actions
+          processors
+          scanners
+          rules_override
+          rules_data
+          exclusion_data
+        ].freeze
+
+        def initialize(appsec_settings:, telemetry:)
+          @default_ruleset = appsec_settings.ruleset
+
+          # NOTE: replace appsec_settings argument with default_ruleset when removing these deprecated settings
+          @default_ip_denylist = appsec_settings.ip_denylist
+          @default_user_id_denylist = appsec_settings.user_id_denylist
+          @default_ip_passlist = appsec_settings.ip_passlist
+
+          @waf_builder = WAF::HandleBuilder.new(
+            obfuscator: {
+              key_regex: appsec_settings.obfuscator_key_regex,
+              value_regex: appsec_settings.obfuscator_value_regex
+            }
+          )
+
+          diagnostics = load_default_config(telemetry: telemetry)
+          report_configuration_diagnostics(diagnostics, action: "init", telemetry: telemetry)
+          @ruleset_version = diagnostics["ruleset_version"]
+
+          @handle_ref = ThreadSafeRef.new(@waf_builder.build_handle)
+
+          metric("init", success: true, ruleset_version: @ruleset_version, telemetry: telemetry)
+        rescue WAF::Error => e
+          error_message = "AppSec security engine failed to initialize"
+
+          Datadog.logger.error("#{error_message}, error #{e.inspect}")
+          telemetry.report(e, description: error_message)
+
+          metric("init", success: false, ruleset_version: @ruleset_version, telemetry: telemetry)
+
+          raise e
+        end
+
+        def new_runner
+          SecurityEngine::Runner.new(@handle_ref, ruleset_version: @ruleset_version)
+        end
+
+        def add_or_update_config(config, path:)
+          is_ruleset_update = path.include?("ASM_DD")
+
+          # default config has to be removed when adding an ASM_DD config
+          remove_config_at_path(DEFAULT_RULES_CONFIG_PATH) if is_ruleset_update
+
+          diagnostics = @waf_builder.add_or_update_config(config, path: path)
+          @reconfigured_ruleset_version = diagnostics["ruleset_version"] if diagnostics.key?("ruleset_version")
+          report_configuration_diagnostics(diagnostics, action: "update", telemetry: AppSec.telemetry)
+
+          # we need to load default config if diagnostics contains top-level error for rules or processors
+          if is_ruleset_update &&
+              (diagnostics.key?("error") ||
+              diagnostics.dig("rules", "error") ||
+              diagnostics.dig("processors", "errors"))
+            diagnostics = load_default_config(telemetry: AppSec.telemetry)
+            @reconfigured_ruleset_version = diagnostics["ruleset_version"]
+            report_configuration_diagnostics(diagnostics, action: "update", telemetry: AppSec.telemetry)
+          end
+
+          diagnostics
+        rescue WAF::Error => e
+          error_message = "libddwaf builder failed to add or update config at path: #{path}"
+
+          Datadog.logger.debug("#{error_message}, error: #{e.inspect}")
+          AppSec.telemetry.report(e, description: error_message)
+        end
+
+        def remove_config_at_path(path)
+          result = @waf_builder.remove_config_at_path(path)
+
+          if result && path != DEFAULT_RULES_CONFIG_PATH && path.include?("ASM_DD")
+            diagnostics = load_default_config(telemetry: AppSec.telemetry)
+            @reconfigured_ruleset_version = diagnostics["ruleset_version"]
+            report_configuration_diagnostics(diagnostics, action: "update", telemetry: AppSec.telemetry)
+          end
+
+          result
+        rescue WAF::Error => e
+          error_message = "libddwaf handle builder failed to remove config at path: #{path}"
+
+          Datadog.logger.error("#{error_message}, error: #{e.inspect}")
+          AppSec.telemetry.report(e, description: error_message)
+        end
+
+        def reconfigure!
+          new_waf_handle = @waf_builder.build_handle
+          @ruleset_version = @reconfigured_ruleset_version
+
+          @handle_ref.current = new_waf_handle
+
+          metric("updates", success: true, ruleset_version: @ruleset_version, telemetry: AppSec.telemetry)
+        rescue WAF::Error => e
+          # WAF::Error can only be raised during new WAF handle creation or when reading known addresses.
+          # This means that the current WAF handle was not yet substituted.
+          error_message = "AppSec security engine failed to reconfigure, reverting to the previous configuration"
+
+          Datadog.logger.error("#{error_message}, error #{e.inspect}")
+          AppSec.telemetry.report(e, description: error_message)
+
+          metric("updates", success: false, ruleset_version: @reconfigured_ruleset_version, telemetry: AppSec.telemetry)
+        end
+
+        private
+
+        def metric(metric_name, success:, ruleset_version:, telemetry:)
+          telemetry.inc(
+            Ext::TELEMETRY_METRICS_NAMESPACE,
+            "waf.#{metric_name}",
+            1,
+            tags: {
+              waf_version: Datadog::AppSec::WAF::VERSION::BASE_STRING,
+              event_rules_version: ruleset_version.to_s,
+              success: success.to_s
+            }
+          )
+        end
+
+        def load_default_config(telemetry:)
+          config = AppSec::Processor::RuleLoader.load_rules(telemetry: telemetry, ruleset: @default_ruleset)
+
+          # deprecated - ip and user id denylists should be configured via RC
+          config["rules_data"] ||= AppSec::Processor::RuleLoader.load_data(
+            ip_denylist: @default_ip_denylist,
+            user_id_denylist: @default_user_id_denylist
+          )
+
+          # deprecated - ip passlist should be configured via RC
+          config["exclusions"] ||= AppSec::Processor::RuleLoader.load_exclusions(ip_passlist: @default_ip_passlist)
+
+          @waf_builder.add_or_update_config(config, path: DEFAULT_RULES_CONFIG_PATH)
+        end
+
+        def report_configuration_diagnostics(diagnostics, action:, telemetry:)
+          raise ArgumentError, "action must be one of TELEMETRY_ACTIONS" unless TELEMETRY_ACTIONS.include?(action)
+
+          common_tags = {
+            waf_version: Datadog::AppSec::WAF::VERSION::BASE_STRING,
+            event_rules_version: diagnostics["ruleset_version"].to_s,
+            action: action
+          }
+
+          if diagnostics["error"]
+            telemetry.inc(
+              Ext::TELEMETRY_METRICS_NAMESPACE, "waf.config_errors", 1,
+              tags: common_tags.merge(scope: "top-level")
+            )
+
+            telemetry.error(diagnostics["error"])
+          end
+
+          diagnostics.each do |config_key, config_diagnostics|
+            next unless DIAGNOSTICS_CONFIG_KEYS.include?(config_key)
+            next if !config_diagnostics.key?("error") && config_diagnostics.fetch("errors", []).empty?
+
+            if config_diagnostics["error"]
+              telemetry.error(config_diagnostics["error"])
+
+              telemetry.inc(
+                Ext::TELEMETRY_METRICS_NAMESPACE, "waf.config_errors", 1,
+                tags: common_tags.merge(config_key: config_key, scope: "top-level")
+              )
+            elsif config_diagnostics["errors"]
+              config_diagnostics["errors"].each do |error, config_ids|
+                telemetry.error("#{error}: [#{config_ids.join(",")}]")
+              end
+
+              telemetry.inc(
+                Ext::TELEMETRY_METRICS_NAMESPACE, "waf.config_errors", config_diagnostics["errors"].count,
+                tags: common_tags.merge(config_key: config_key, scope: "item")
+              )
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/security_engine/result.rb

@@ -0,0 +1,103 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module SecurityEngine
+      # A namespace for value-objects representing the result of WAF check.
+      module Result
+        # A generic result without indication of its type.
+        class Base
+          attr_reader :events, :actions, :attributes, :duration_ns, :duration_ext_ns
+
+          def initialize(events:, actions:, attributes:, duration_ns:, duration_ext_ns:, timeout:, keep:, input_truncated:)
+            @events = events
+            @actions = actions
+            @attributes = attributes
+            @duration_ns = duration_ns
+            @duration_ext_ns = duration_ext_ns
+
+            @keep = !!keep
+            @timeout = !!timeout
+            @input_truncated = !!input_truncated
+          end
+
+          def timeout?
+            @timeout
+          end
+
+          def keep?
+            @keep
+          end
+
+          def input_truncated?
+            @input_truncated
+          end
+
+          def match?
+            raise NotImplementedError
+          end
+
+          def error?
+            raise NotImplementedError
+          end
+        end
+
+        # A result that indicates a security rule match
+        class Match < Base
+          def match?
+            true
+          end
+
+          def error?
+            false
+          end
+        end
+
+        # A result that indicates a successful security rules check without a match
+        class Ok < Base
+          def match?
+            false
+          end
+
+          def error?
+            false
+          end
+        end
+
+        # A result that indicates an internal security library error
+        class Error
+          attr_reader :events, :actions, :attributes, :duration_ns, :duration_ext_ns
+
+          def initialize(duration_ext_ns:, input_truncated:)
+            @events = []
+            @actions = {}.freeze
+            @attributes = {}.freeze
+            @duration_ns = 0
+            @duration_ext_ns = duration_ext_ns
+            @input_truncated = !!input_truncated
+          end
+
+          def keep?
+            false
+          end
+
+          def timeout?
+            false
+          end
+
+          def input_truncated?
+            @input_truncated
+          end
+
+          def match?
+            false
+          end
+
+          def error?
+            true
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/security_engine/runner.rb

@@ -0,0 +1,111 @@
+# frozen_string_literal: true
+
+require_relative "result"
+
+module Datadog
+  module AppSec
+    module SecurityEngine
+      # A class that check input via security engine (WAF) and respond with result.
+      class Runner
+        SUCCESSFUL_EXECUTION_CODES = [:ok, :match].freeze
+
+        attr_reader :ruleset_version
+
+        def initialize(handle_ref, ruleset_version:)
+          @mutex = Mutex.new
+          @handle_ref = handle_ref
+          @waf_handle = handle_ref.acquire
+          @ruleset_version = ruleset_version
+
+          @debug_tag = "libddwaf:#{WAF::VERSION::STRING} method:ddwaf_run"
+        end
+
+        def run(persistent_data, ephemeral_data, timeout = WAF::LibDDWAF::DDWAF_RUN_TIMEOUT)
+          @mutex.lock
+
+          start_ns = Core::Utils::Time.get_time(:nanosecond)
+          persistent_data.reject! do |_, v|
+            next false if v.is_a?(TrueClass) || v.is_a?(FalseClass)
+
+            v.nil? || (v.respond_to?(:empty?) && v.empty?)
+          end
+
+          ephemeral_data.reject! do |_, v|
+            next false if v.is_a?(TrueClass) || v.is_a?(FalseClass)
+
+            v.nil? || (v.respond_to?(:empty?) && v.empty?)
+          end
+
+          result = try_run(persistent_data, ephemeral_data, timeout)
+          stop_ns = Core::Utils::Time.get_time(:nanosecond)
+
+          report_execution(result)
+
+          unless SUCCESSFUL_EXECUTION_CODES.include?(result.status)
+            return Result::Error.new(duration_ext_ns: stop_ns - start_ns, input_truncated: result.input_truncated?)
+          end
+
+          klass = (result.status == :match) ? Result::Match : Result::Ok
+          klass.new(
+            events: result.events,
+            actions: result.actions,
+            attributes: result.attributes,
+            keep: result.keep?,
+            timeout: result.timeout?,
+            duration_ns: result.duration,
+            duration_ext_ns: (stop_ns - start_ns),
+            input_truncated: result.input_truncated?
+          )
+        ensure
+          @mutex.unlock
+        end
+
+        def waf_context
+          @waf_context ||= @waf_handle.build_context
+        end
+
+        def waf_addresses
+          @waf_handle.known_addresses
+        end
+
+        def finalize!
+          @waf_context&.finalize!
+        ensure
+          @handle_ref.release(@waf_handle)
+        end
+
+        private
+
+        def try_run(persistent_data, ephemeral_data, timeout)
+          waf_context.run(persistent_data, ephemeral_data, timeout)
+        rescue WAF::LibDDWAFError => e
+          Datadog.logger.debug { "#{@debug_tag} execution error: #{e.class}: #{e.message} backtrace: #{e.backtrace&.first(3)}" }
+          AppSec.telemetry.report(e, description: "libddwaf-rb internal low-level error")
+
+          WAF::Result.new(
+            status: :err_internal,
+            events: [],
+            actions: {},
+            attributes: {},
+            duration: 0,
+            keep: false,
+            timeout: false
+          )
+        end
+
+        def report_execution(result)
+          Datadog.logger.debug { "#{@debug_tag} execution timed out: #{result.inspect}" } if result.timeout?
+
+          if SUCCESSFUL_EXECUTION_CODES.include?(result.status)
+            Datadog.logger.debug { "#{@debug_tag} execution result: #{result.inspect}" }
+          else
+            message = "#{@debug_tag} execution error: #{result.status.inspect}"
+
+            Datadog.logger.debug { message }
+            AppSec.telemetry.error(message)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/security_engine.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # A namespace for secutiry library we use to detect and prevent threats.
+    module SecurityEngine
+    end
+  end
+end

data/lib/datadog/appsec/security_event.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # A class that represents a security event of any kind. It could be an event
+    # representing an attack or fingerprinting results as attributes or an API
+    # security check with extracted schema.
+    class SecurityEvent
+      SCHEMA_KEY_PREFIX = "_dd.appsec.s."
+      FINGERPRINT_KEY_PREFIX = "_dd.appsec.fp."
+
+      attr_reader :waf_result, :trace, :span
+
+      def initialize(waf_result, trace:, span:)
+        @waf_result = waf_result
+        @trace = trace
+        @span = span
+      end
+
+      def keep?
+        @waf_result.keep?
+      end
+
+      def schema?
+        return @has_schema if defined?(@has_schema)
+
+        @has_schema = @waf_result.attributes.any? { |name, _| name.start_with?(SCHEMA_KEY_PREFIX) }
+      end
+
+      def fingerprint?
+        return @has_fingerprint if defined?(@has_fingerprint)
+
+        @has_fingerprint = @waf_result.attributes.any? { |name, _| name.start_with?(FINGERPRINT_KEY_PREFIX) }
+      end
+    end
+  end
+end

data/lib/datadog/appsec/thread_safe_ref.rb

@@ -0,0 +1,61 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # This class is used for referencing an object that might be marked
+    # for finalization in another thread.
+    #
+    # References to the object are counted, and objects marked for finalization
+    # can be safely finalized when their reference count reaches zero.
+    class ThreadSafeRef
+      def initialize(initial_obj, finalizer: :finalize!)
+        @current = initial_obj
+        @finalizer = finalizer
+
+        @counters = Hash.new(0)
+        @outdated = []
+        @mutex = Mutex.new
+      end
+
+      def acquire
+        @mutex.synchronize do
+          @counters[@current] += 1
+
+          @current
+        end
+      end
+
+      def release(obj)
+        @mutex.synchronize do
+          @counters[obj] -= 1
+
+          @outdated.reject! do |outdated_obj|
+            next unless @counters[outdated_obj].zero?
+
+            finalize(outdated_obj)
+          end
+        end
+      end
+
+      def current=(obj)
+        @mutex.synchronize do
+          @outdated << @current
+
+          @current = obj
+        end
+      end
+
+      private
+
+      def finalize(obj)
+        obj.public_send(@finalizer)
+
+        true
+      rescue => e
+        Datadog.logger.debug("Couldn't finalize #{obj.class.name} object, error: #{e.inspect}")
+
+        true
+      end
+    end
+  end
+end

data/lib/datadog/appsec/trace_keeper.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # This class is used to mark trace as manual keep and tag it as ASM product.
+    module TraceKeeper
+      def self.keep!(trace)
+        return unless trace
+
+        previous_dm = trace.get_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER)
+
+        # NOTE: This action will not set correct decision maker value, so the
+        #       trace keeping must be done with additional steps below
+        trace.keep!
+
+        # NOTE: Preserve decision maker if already set by another product.
+        #       As `trace.keep!` resets `_dd.p.dm` to `MANUAL`, we restore the
+        #       previous value when another product has already claimed the
+        #       decision maker.
+        if previous_dm.nil? || previous_dm == Tracing::Sampling::Ext::Decision::MANUAL
+          trace.set_tag(
+            Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER,
+            Tracing::Sampling::Ext::Decision::ASM
+          )
+        else
+          trace.set_tag(
+            Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER,
+            previous_dm,
+          )
+        end
+
+        trace.set_distributed_source(Ext::PRODUCT_BIT)
+      end
+    end
+  end
+end

data/lib/datadog/appsec/utils/hash_coercion.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Utils
+      # A module for coercing arbitrary objects into hashes.
+      module HashCoercion
+        # A best effort to coerce an object to a hash with methods known to various
+        # frameworks with a fallback to standard library.
+        #
+        # @param object [Object] The object to coerce.
+        # @return [Hash, nil] The coerced `Hash` or `nil` if the object is not coercible.
+        def self.coerce(object)
+          return object.as_json if object.respond_to?(:as_json)
+          return object.to_hash if object.respond_to?(:to_hash)
+          return object.to_h if object.respond_to?(:to_h)
+
+          nil
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/utils/http/body.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+require "json"
+require "cgi"
+
+require_relative "url_encoded"
+
+module Datadog
+  module AppSec
+    module Utils
+      module HTTP
+        # Module for handling HTTP body parsing
+        module Body
+          # Matches Rack's default query bytesize limit
+          DEFAULT_BYTESIZE_LIMIT = 4 * 1024 * 1024
+
+          def self.parse(body, media_type:, limit: DEFAULT_BYTESIZE_LIMIT)
+            return if body.nil?
+
+            body.rewind if body.respond_to?(:rewind) # steep:ignore NoMethod
+            # @type var content: ::String?
+            content = body.respond_to?(:read) ? body.read : body # steep:ignore NoMethod, IncompatibleAssignment
+            body.rewind if body.respond_to?(:rewind) # steep:ignore NoMethod
+
+            return if content.nil? || content.empty?
+
+            if media_type.subtype == "json" || media_type.subtype.end_with?("+json")
+              JSON.parse(content)
+            elsif media_type.subtype == "x-www-form-urlencoded"
+              URLEncoded.parse(content, limit: limit)
+            end
+          rescue => e
+            AppSec.telemetry.report(e, description: "AppSec: Failed to parse body")
+
+            nil
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/utils/http/body_reader.rb

@@ -0,0 +1,61 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Utils
+      module HTTP
+        module BodyReader
+          module_function
+
+          def read(body, limit:, rewind_before_read: false)
+            return body.byteslice(0, limit + 1) if body.is_a?(String)
+            return if body.nil? || !body.respond_to?(:read)
+            return if rewind_before_read && !body.respond_to?(:rewind)
+
+            read_stream(body, limit: limit, rewind_before_read: rewind_before_read)
+          end
+
+          def read_stream(body, limit:, rewind_before_read: false)
+            # @type var rewound: bool
+            rewound = false
+
+            if rewind_before_read
+              return unless rewind(body)
+
+              rewound = true
+            end
+
+            buffer = +"".b
+            max = limit + 1
+
+            while buffer.bytesize <= limit
+              chunk = body.read(max - buffer.bytesize)
+              break if chunk.nil? || chunk.empty?
+
+              buffer << chunk
+            end
+
+            buffer
+          rescue => e
+            Datadog.logger.debug { "AppSec: Failed to read body: #{e.class}: #{e.message}" }
+
+            raise
+          ensure
+            rewind(body) if rewound
+          end
+
+          private_class_method def rewind(body)
+            return false unless body.respond_to?(:rewind)
+
+            # NOTE: Steep can't narrow interfaces via `respond_to?` guard-clause
+            body.rewind # steep:ignore
+            true
+          rescue => e
+            Datadog.logger.debug { "AppSec: Failed to rewind body: #{e.class}: #{e.message}" }
+            false
+          end
+        end
+      end
+    end
+  end
+end