datadog 2.0.0.beta1

data/lib/datadog/appsec/event.rb

@@ -0,0 +1,171 @@
+# frozen_string_literal: true
+
+require 'json'
+require 'zlib'
+require 'base64'
+
+require_relative 'rate_limiter'
+
+module Datadog
+  module AppSec
+    # AppSec event
+    module Event
+      ALLOWED_REQUEST_HEADERS = %w[
+        X-Forwarded-For
+        X-Client-IP
+        X-Real-IP
+        X-Forwarded
+        X-Cluster-Client-IP
+        Forwarded-For
+        Forwarded
+        Via
+        True-Client-IP
+        Content-Length
+        Content-Type
+        Content-Encoding
+        Content-Language
+        Host
+        User-Agent
+        Accept
+        Accept-Encoding
+        Accept-Language
+      ].map!(&:downcase).freeze
+
+      ALLOWED_RESPONSE_HEADERS = %w[
+        Content-Length
+        Content-Type
+        Content-Encoding
+        Content-Language
+      ].map!(&:downcase).freeze
+
+      MAX_ENCODED_SCHEMA_SIZE = 25000
+      # For more information about this number
+      # please check https://github.com/DataDog/dd-trace-rb/pull/3177#issuecomment-1747221082
+      MIN_SCHEMA_SIZE_FOR_COMPRESSION = 260
+
+      # Record events for a trace
+      #
+      # This is expected to be called only once per trace for the rate limiter
+      # to properly apply
+      class << self
+        def record(span, *events)
+          # ensure rate limiter is called only when there are events to record
+          return if events.empty? || span.nil?
+
+          Datadog::AppSec::RateLimiter.limit(:traces) do
+            record_via_span(span, *events)
+          end
+        end
+
+        def record_via_span(span, *events)
+          events.group_by { |e| e[:trace] }.each do |trace, event_group|
+            unless trace
+              Datadog.logger.debug { "{ error: 'no trace: cannot record', event_group: #{event_group.inspect}}" }
+              next
+            end
+
+            trace.keep!
+            trace.set_tag(
+              Datadog::Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER,
+              Datadog::Tracing::Sampling::Ext::Decision::ASM
+            )
+
+            # prepare and gather tags to apply
+            service_entry_tags = build_service_entry_tags(event_group)
+
+            # apply tags to service entry span
+            service_entry_tags.each do |key, value|
+              span.set_tag(key, value)
+            end
+          end
+        end
+
+        # rubocop:disable Metrics/MethodLength
+        def build_service_entry_tags(event_group)
+          waf_events = []
+          entry_tags = event_group.each_with_object({ '_dd.origin' => 'appsec' }) do |event, tags|
+            # TODO: assume HTTP request context for now
+            if (request = event[:request])
+              request.headers.each do |header, value|
+                tags["http.request.headers.#{header}"] = value if ALLOWED_REQUEST_HEADERS.include?(header.downcase)
+              end
+
+              tags['http.host'] = request.host
+              tags['http.useragent'] = request.user_agent
+              tags['network.client.ip'] = request.remote_addr
+            end
+
+            if (response = event[:response])
+              response.headers.each do |header, value|
+                tags["http.response.headers.#{header}"] = value if ALLOWED_RESPONSE_HEADERS.include?(header.downcase)
+              end
+            end
+
+            waf_result = event[:waf_result]
+            # accumulate triggers
+            waf_events += waf_result.events
+
+            waf_result.derivatives.each do |key, value|
+              parsed_value = json_parse(value)
+              next unless parsed_value
+
+              parsed_value_size = parsed_value.size
+
+              schema_value = if parsed_value_size >= MIN_SCHEMA_SIZE_FOR_COMPRESSION
+                               compressed_and_base64_encoded(parsed_value)
+                             else
+                               parsed_value
+                             end
+              next unless schema_value
+
+              if schema_value.size >= MAX_ENCODED_SCHEMA_SIZE
+                Datadog.logger.debug do
+                  "Schema key: #{key} exceeds the max size value. It will not be included as part of the span tags"
+                end
+                next
+              end
+
+              tags[key] = schema_value
+            end
+
+            tags
+          end
+
+          appsec_events = json_parse({ triggers: waf_events })
+          entry_tags['_dd.appsec.json'] = appsec_events if appsec_events
+          entry_tags
+        end
+        # rubocop:enable Metrics/MethodLength
+
+        private
+
+        def compressed_and_base64_encoded(value)
+          Base64.encode64(gzip(value))
+        rescue TypeError => e
+          Datadog.logger.debug do
+            "Failed to compress and encode value when populating AppSec::Event. Error: #{e.message}"
+          end
+          nil
+        end
+
+        def json_parse(value)
+          JSON.dump(value)
+        rescue ArgumentError => e
+          Datadog.logger.debug do
+            "Failed to parse value to JSON when populating AppSec::Event. Error: #{e.message}"
+          end
+          nil
+        end
+
+        def gzip(value)
+          sio = StringIO.new
+          # For an in depth comparison of Zlib options check https://github.com/DataDog/dd-trace-rb/pull/3177#issuecomment-1747215473
+          gz = Zlib::GzipWriter.new(sio, Zlib::BEST_SPEED, Zlib::DEFAULT_STRATEGY)
+          gz.write(value)
+          gz.close
+          sio.string
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/ext.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Ext
+      INTERRUPT = :datadog_appsec_interrupt
+      SCOPE_KEY = 'datadog.appsec.scope'
+    end
+  end
+end

data/lib/datadog/appsec/extensions.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+require_relative 'configuration'
+
+module Datadog
+  module AppSec
+    # Extends Datadog tracing with AppSec features
+    module Extensions
+      # Inject AppSec into global objects.
+      def self.activate!
+        Core::Configuration::Settings.extend(Configuration::Settings)
+      end
+    end
+  end
+end

data/lib/datadog/appsec/instrumentation/gateway/argument.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Instrumentation
+      class Gateway
+        # Base class for Gateway Arguments
+        class Argument; end # rubocop:disable Lint/EmptyClass
+
+        # Gateway User argument
+        class User < Argument
+          attr_reader :id
+
+          def initialize(id)
+            super()
+            @id = id
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/instrumentation/gateway.rb

@@ -0,0 +1,64 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # Instrumentation for AppSec
+    module Instrumentation
+      # Instrumentation gateway implementation
+      class Gateway
+        # Instrumentation gateway middleware
+        class Middleware
+          attr_reader :key, :block
+
+          def initialize(key, &block)
+            @key = key
+            @block = block
+          end
+
+          def call(stack, env)
+            @block.call(stack, env)
+          end
+        end
+
+        private_constant :Middleware
+
+        def initialize
+          @middlewares = Hash.new { |h, k| h[k] = [] }
+        end
+
+        def push(name, env, &block)
+          block ||= -> {}
+
+          middlewares_for_name = middlewares[name]
+
+          return [block.call, nil] if middlewares_for_name.empty?
+
+          wrapped = lambda do |_env|
+            [block.call, nil]
+          end
+
+          # TODO: handle exceptions, except for wrapped
+          stack = middlewares_for_name.reverse.reduce(wrapped) do |next_, middleware|
+            lambda do |env_|
+              middleware.call(next_, env_)
+            end
+          end
+
+          stack.call(env)
+        end
+
+        def watch(name, key, &block)
+          @middlewares[name] << Middleware.new(key, &block) unless middlewares[name].any? { |m| m.key == key }
+        end
+
+        private
+
+        attr_reader :middlewares
+      end
+
+      def self.gateway
+        @gateway ||= Gateway.new # TODO: not thread safe
+      end
+    end
+  end
+end

data/lib/datadog/appsec/instrumentation.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # Instrumentation for AppSec
+    module Instrumentation
+    end
+  end
+end

data/lib/datadog/appsec/monitor/gateway/watcher.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+require_relative '../../instrumentation/gateway'
+require_relative '../../reactive/operation'
+require_relative '../reactive/set_user'
+
+module Datadog
+  module AppSec
+    module Monitor
+      module Gateway
+        # Watcher for Apssec internal events
+        module Watcher
+          class << self
+            def watch
+              gateway = Instrumentation.gateway
+
+              watch_user_id(gateway)
+            end
+
+            def watch_user_id(gateway = Instrumentation.gateway)
+              gateway.watch('identity.set_user', :appsec) do |stack, user|
+                block = false
+                event = nil
+                scope = Datadog::AppSec.active_scope
+
+                AppSec::Reactive::Operation.new('identity.set_user') do |op|
+                  Monitor::Reactive::SetUser.subscribe(op, scope.processor_context) do |result|
+                    if result.status == :match
+                      # TODO: should this hash be an Event instance instead?
+                      event = {
+                        waf_result: result,
+                        trace: scope.trace,
+                        span: scope.service_entry_span,
+                        user: user,
+                        actions: result.actions
+                      }
+
+                      if scope.service_entry_span
+                        scope.service_entry_span.set_tag('appsec.blocked', 'true') if result.actions.include?('block')
+                        scope.service_entry_span.set_tag('appsec.event', 'true')
+                      end
+
+                      scope.processor_context.events << event
+                    end
+                  end
+
+                  block = Monitor::Reactive::SetUser.publish(op, user)
+                end
+
+                throw(Datadog::AppSec::Ext::INTERRUPT, [nil, [[:block, event]]]) if block
+
+                ret, res = stack.call(user)
+
+                if event
+                  res ||= []
+                  res << [:monitor, event]
+                end
+
+                [ret, res]
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/monitor/reactive/set_user.rb

@@ -0,0 +1,58 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Monitor
+      module Reactive
+        # Dispatch data from Datadog::Kit::Identity.set_user to the WAF context
+        module SetUser
+          ADDRESSES = [
+            'usr.id',
+          ].freeze
+          private_constant :ADDRESSES
+
+          def self.publish(op, user)
+            catch(:block) do
+              op.publish('usr.id', user.id)
+
+              nil
+            end
+          end
+
+          def self.subscribe(op, waf_context)
+            op.subscribe(*ADDRESSES) do |*values|
+              Datadog.logger.debug { "reacted to #{ADDRESSES.inspect}: #{values.inspect}" }
+
+              user_id = values[0]
+
+              waf_args = {
+                'usr.id' => user_id,
+              }
+
+              waf_timeout = Datadog.configuration.appsec.waf_timeout
+              result = waf_context.run(waf_args, waf_timeout)
+
+              Datadog.logger.debug { "WAF TIMEOUT: #{result.inspect}" } if result.timeout
+
+              case result.status
+              when :match
+                Datadog.logger.debug { "WAF: #{result.inspect}" }
+
+                yield result
+                throw(:block, true) unless result.actions.empty?
+              when :ok
+                Datadog.logger.debug { "WAF OK: #{result.inspect}" }
+              when :invalid_call
+                Datadog.logger.debug { "WAF CALL ERROR: #{result.inspect}" }
+              when :invalid_rule, :invalid_flow, :no_rule
+                Datadog.logger.debug { "WAF RULE ERROR: #{result.inspect}" }
+              else
+                Datadog.logger.debug { "WAF UNKNOWN: #{result.status.inspect} #{result.inspect}" }
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/monitor.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+require_relative 'monitor/gateway/watcher'
+
+module Datadog
+  module AppSec
+    # Monitor for internal AppSec Events
+    module Monitor
+    end
+  end
+end

data/lib/datadog/appsec/processor/actions.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    class Processor
+      # Actions store the actions information in memory
+      # Also, takes care of merging when RC send new information
+      module Actions
+        class << self
+          def actions
+            @actions ||= []
+          end
+
+          def fecth_configuration(action)
+            actions.find { |action_configuration| action_configuration['id'] == action }
+          end
+
+          def merge(actions_to_merge)
+            return if actions_to_merge.empty?
+
+            if actions.empty?
+              @actions = actions_to_merge
+            else
+              merged_actions = []
+              actions_dup = actions.dup
+
+              actions_to_merge.each do |new_action|
+                existing_action = actions_dup.find { |action| new_action['id'] == action['id'] }
+
+                # the old action is discard and the new kept
+                actions_dup.delete(existing_action) if existing_action
+                merged_actions << new_action
+              end
+
+              @actions = merged_actions.concat(actions_dup)
+            end
+          end
+
+          private
+
+          # Used in tests
+          def reset
+            @actions = []
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/processor/rule_loader.rb

@@ -0,0 +1,123 @@
+# frozen_string_literal: true
+
+require_relative '../assets'
+
+module Datadog
+  module AppSec
+    class Processor
+      # RuleLoader utility modules
+      # that load appsec rules and data from  settings
+      module RuleLoader
+        class << self
+          def load_rules(ruleset:)
+            begin
+              case ruleset
+              when :recommended, :strict
+                JSON.parse(Datadog::AppSec::Assets.waf_rules(ruleset))
+              when :risky
+                Datadog.logger.warn(
+                  'The :risky Application Security Management ruleset has been deprecated and no longer available.'\
+                  'The `:recommended` ruleset will be used instead.'\
+                  'Please remove the `appsec.ruleset = :risky` setting from your Datadog.configure block.'
+                )
+                JSON.parse(Datadog::AppSec::Assets.waf_rules(:recommended))
+              when String
+                JSON.parse(File.read(File.expand_path(ruleset)))
+              when File, StringIO
+                JSON.parse(ruleset.read || '').tap { ruleset.rewind }
+              when Hash
+                ruleset
+              else
+                raise ArgumentError, "unsupported value for ruleset setting: #{ruleset.inspect}"
+              end
+            rescue StandardError => e
+              Datadog.logger.error do
+                "libddwaf ruleset failed to load, ruleset: #{ruleset.inspect} error: #{e.inspect}"
+              end
+
+              nil
+            end
+          end
+
+          def load_data(ip_denylist: [], user_id_denylist: [])
+            data = []
+            data << [denylist_data('blocked_ips', ip_denylist)] if ip_denylist.any?
+            data << [denylist_data('blocked_users', user_id_denylist)] if user_id_denylist.any?
+
+            data
+          end
+
+          def load_exclusions(ip_passlist: [])
+            exclusions = []
+            exclusions << [passlist_exclusions(ip_passlist)] if ip_passlist.any?
+
+            exclusions
+          end
+
+          private
+
+          def denylist_data(id, denylist)
+            {
+              'id' => id,
+              'type' => 'data_with_expiration',
+              'data' => denylist.map { |v| { 'value' => v.to_s, 'expiration' => 2**63 } }
+            }
+          end
+
+          def passlist_exclusions(ip_passlist) # rubocop:disable Metrics/MethodLength
+            case ip_passlist
+            when Array
+              pass = ip_passlist
+              monitor = []
+            when Hash
+              pass = ip_passlist[:pass]
+              monitor = ip_passlist[:monitor]
+            else
+              pass = []
+              monitor = []
+            end
+
+            exclusions = []
+
+            exclusions << {
+              'conditions' => [
+                {
+                  'operator' => 'ip_match',
+                  'parameters' => {
+                    'inputs' => [
+                      {
+                        'address' => 'http.client_ip'
+                      }
+                    ],
+                    'list' => pass
+                  }
+                }
+              ],
+              'id' => SecureRandom.uuid,
+            }
+
+            exclusions << {
+              'conditions' => [
+                {
+                  'operator' => 'ip_match',
+                  'parameters' => {
+                    'inputs' => [
+                      {
+                        'address' => 'http.client_ip'
+                      }
+                    ],
+                    'list' => monitor
+                  }
+                }
+              ],
+              'id' => SecureRandom.uuid,
+              'on_match' => 'monitor'
+            }
+
+            exclusions
+          end
+        end
+      end
+    end
+  end
+end