datadog 2.0.0.beta1

data/lib/datadog/appsec/contrib/rack/reactive/response.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rack
+        module Reactive
+          # Dispatch data from a Rack response to the WAF context
+          module Response
+            ADDRESSES = [
+              'response.status',
+              'response.headers',
+            ].freeze
+            private_constant :ADDRESSES
+
+            def self.publish(op, gateway_response)
+              catch(:block) do
+                op.publish('response.status', gateway_response.status)
+                op.publish('response.headers', gateway_response.headers)
+
+                nil
+              end
+            end
+
+            def self.subscribe(op, waf_context)
+              op.subscribe(*ADDRESSES) do |*values|
+                Datadog.logger.debug { "reacted to #{ADDRESSES.inspect}: #{values.inspect}" }
+
+                response_status = values[0]
+                response_headers = values[1]
+                response_headers_no_cookies = response_headers.dup.tap { |h| h.delete('set-cookie') }
+
+                waf_args = {
+                  'server.response.status' => response_status.to_s,
+                  'server.response.headers' => response_headers,
+                  'server.response.headers.no_cookies' => response_headers_no_cookies,
+                }
+
+                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
+end

data/lib/datadog/appsec/contrib/rack/request_body_middleware.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+
+require_relative 'gateway/request'
+require_relative '../../instrumentation/gateway'
+require_relative '../../response'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rack
+        # Rack request body middleware for AppSec
+        # This should be inserted just below Rack::JSONBodyParser or
+        # legacy Rack::PostBodyContentTypeParser from rack-contrib
+        class RequestBodyMiddleware
+          def initialize(app, opt = {})
+            @app = app
+          end
+
+          def call(env)
+            context = env[Datadog::AppSec::Ext::SCOPE_KEY]
+
+            return @app.call(env) unless context
+
+            # TODO: handle exceptions, except for @app.call
+
+            request_return, request_response = Instrumentation.gateway.push(
+              'rack.request.body',
+              Gateway::Request.new(env)
+            ) do
+              @app.call(env)
+            end
+
+            if request_response
+              blocked_event = request_response.find { |action, _event| action == :block }
+              request_return = AppSec::Response.negotiate(env, blocked_event.last[:actions]).to_rack if blocked_event
+            end
+
+            request_return
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rack/request_middleware.rb

@@ -0,0 +1,196 @@
+# frozen_string_literal: true
+
+require 'json'
+
+require_relative 'gateway/request'
+require_relative 'gateway/response'
+require_relative '../../instrumentation/gateway'
+require_relative '../../processor'
+require_relative '../../response'
+
+require_relative '../../../tracing/client_ip'
+require_relative '../../../tracing/contrib/rack/header_collection'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rack
+        # Topmost Rack middleware for AppSec
+        # This should be inserted just below Datadog::Tracing::Contrib::Rack::TraceMiddleware
+        class RequestMiddleware
+          def initialize(app, opt = {})
+            @app = app
+
+            @oneshot_tags_sent = false
+          end
+
+          # rubocop:disable Metrics/AbcSize,Metrics/PerceivedComplexity,Metrics/CyclomaticComplexity,Metrics/MethodLength
+          def call(env)
+            return @app.call(env) unless Datadog::AppSec.enabled?
+
+            boot = Datadog::Core::Remote::Tie.boot
+            Datadog::Core::Remote::Tie::Tracing.tag(boot, active_span)
+
+            processor = nil
+            ready = false
+            scope = nil
+
+            # For a given request, keep using the first Rack stack scope for
+            # nested apps. Don't set `context` local variable so that on popping
+            # out of this nested stack we don't finalize the parent's context
+            return @app.call(env) if active_scope(env)
+
+            Datadog::AppSec.reconfigure_lock do
+              processor = Datadog::AppSec.processor
+
+              if !processor.nil? && processor.ready?
+                scope = Datadog::AppSec::Scope.activate_scope(active_trace, active_span, processor)
+                env[Datadog::AppSec::Ext::SCOPE_KEY] = scope
+                ready = true
+              end
+            end
+
+            # TODO: handle exceptions, except for @app.call
+
+            return @app.call(env) unless ready
+
+            gateway_request = Gateway::Request.new(env)
+
+            add_appsec_tags(processor, scope, env)
+
+            request_return, request_response = catch(::Datadog::AppSec::Ext::INTERRUPT) do
+              Instrumentation.gateway.push('rack.request', gateway_request) do
+                @app.call(env)
+              end
+            end
+
+            if request_response
+              blocked_event = request_response.find { |action, _options| action == :block }
+              request_return = AppSec::Response.negotiate(env, blocked_event.last[:actions]).to_rack if blocked_event
+            end
+
+            gateway_response = Gateway::Response.new(
+              request_return[2],
+              request_return[0],
+              request_return[1],
+              scope: scope,
+            )
+
+            _response_return, response_response = Instrumentation.gateway.push('rack.response', gateway_response)
+
+            result = scope.processor_context.extract_schema
+
+            if result
+              scope.processor_context.events << {
+                trace: scope.trace,
+                span: scope.service_entry_span,
+                waf_result: result,
+              }
+            end
+
+            scope.processor_context.events.each do |e|
+              e[:response] ||= gateway_response
+              e[:request]  ||= gateway_request
+            end
+
+            AppSec::Event.record(scope.service_entry_span, *scope.processor_context.events)
+
+            if response_response
+              blocked_event = response_response.find { |action, _options| action == :block }
+              request_return = AppSec::Response.negotiate(env, blocked_event.last[:actions]).to_rack if blocked_event
+            end
+
+            request_return
+          ensure
+            if scope
+              add_waf_runtime_tags(scope)
+              Datadog::AppSec::Scope.deactivate_scope
+            end
+          end
+          # rubocop:enable Metrics/AbcSize,Metrics/PerceivedComplexity,Metrics/CyclomaticComplexity,Metrics/MethodLength
+
+          private
+
+          def active_scope(env)
+            env[Datadog::AppSec::Ext::SCOPE_KEY]
+          end
+
+          def active_trace
+            # TODO: factor out tracing availability detection
+
+            return unless defined?(Datadog::Tracing)
+
+            Datadog::Tracing.active_trace
+          end
+
+          def active_span
+            # TODO: factor out tracing availability detection
+
+            return unless defined?(Datadog::Tracing)
+
+            Datadog::Tracing.active_span
+          end
+
+          def add_appsec_tags(processor, scope, env)
+            span = scope.service_entry_span
+            trace = scope.trace
+
+            return unless trace && span
+
+            span.set_tag('_dd.appsec.enabled', 1)
+            span.set_tag('_dd.runtime_family', 'ruby')
+            span.set_tag('_dd.appsec.waf.version', Datadog::AppSec::WAF::VERSION::BASE_STRING)
+
+            if span && span.get_tag(Tracing::Metadata::Ext::HTTP::TAG_CLIENT_IP).nil?
+              request_header_collection = Datadog::Tracing::Contrib::Rack::Header::RequestHeaderCollection.new(env)
+
+              # always collect client ip, as this is part of AppSec provided functionality
+              Datadog::Tracing::ClientIp.set_client_ip_tag!(
+                span,
+                headers: request_header_collection,
+                remote_ip: env['REMOTE_ADDR']
+              )
+            end
+
+            if processor.diagnostics
+              diagnostics = processor.diagnostics
+
+              span.set_tag('_dd.appsec.event_rules.version', diagnostics['ruleset_version'])
+
+              unless @oneshot_tags_sent
+                # Small race condition, but it's inoccuous: worst case the tags
+                # are sent a couple of times more than expected
+                @oneshot_tags_sent = true
+
+                span.set_tag('_dd.appsec.event_rules.loaded', diagnostics['rules']['loaded'].size.to_f)
+                span.set_tag('_dd.appsec.event_rules.error_count', diagnostics['rules']['failed'].size.to_f)
+                span.set_tag('_dd.appsec.event_rules.errors', JSON.dump(diagnostics['rules']['errors']))
+                span.set_tag('_dd.appsec.event_rules.addresses', JSON.dump(processor.addresses))
+
+                # Ensure these tags reach the backend
+                trace.keep!
+                trace.set_tag(
+                  Datadog::Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER,
+                  Datadog::Tracing::Sampling::Ext::Decision::ASM
+                )
+              end
+            end
+          end
+
+          def add_waf_runtime_tags(scope)
+            span = scope.service_entry_span
+            context = scope.processor_context
+
+            return unless span && context
+
+            span.set_tag('_dd.appsec.waf.timeouts', context.timeouts)
+
+            # these tags expect time in us
+            span.set_tag('_dd.appsec.waf.duration', context.time_ns / 1000.0)
+            span.set_tag('_dd.appsec.waf.duration_ext', context.time_ext_ns / 1000.0)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/ext.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        # Rails integration constants
+        module Ext
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/framework.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        # Rails specific framework tie
+        module Framework
+          def self.setup
+            Datadog.configuration.appsec.instrument(:rack)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/gateway/request.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+require_relative '../../../instrumentation/gateway/argument'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        module Gateway
+          # Gateway Request argument. Normalized extration of data from ActionDispatch::Request
+          class Request < Instrumentation::Gateway::Argument
+            attr_reader :request
+
+            def initialize(request)
+              super()
+              @request = request
+            end
+
+            def env
+              request.env
+            end
+
+            def headers
+              request.headers
+            end
+
+            def host
+              request.host
+            end
+
+            def user_agent
+              request.user_agent
+            end
+
+            def remote_addr
+              request.remote_addr
+            end
+
+            def parsed_body
+              # force body parameter parsing, which is done lazily by Rails
+              request.parameters
+
+              # usually Hash<String,String> but can be a more complex
+              # Hash<String,String||Array||Hash> when e.g coming from JSON or
+              # with Rails advanced param square bracket parsing
+              body = request.env['action_dispatch.request.request_parameters']
+
+              return if body.nil?
+
+              body.reject do |k, _v|
+                request.env['action_dispatch.request.path_parameters'].key?(k)
+              end
+            end
+
+            def route_params
+              excluded = [:controller, :action]
+
+              request.env['action_dispatch.request.path_parameters'].reject do |k, _v|
+                excluded.include?(k)
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/gateway/watcher.rb

@@ -0,0 +1,71 @@
+# frozen_string_literal: true
+
+require_relative '../../../instrumentation/gateway'
+require_relative '../../../reactive/operation'
+require_relative '../reactive/action'
+require_relative '../../../event'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        module Gateway
+          # Watcher for Rails gateway events
+          module Watcher
+            class << self
+              def watch
+                gateway = Instrumentation.gateway
+
+                watch_request_action(gateway)
+              end
+
+              def watch_request_action(gateway = Instrumentation.gateway)
+                gateway.watch('rails.request.action', :appsec) do |stack, gateway_request|
+                  block = false
+
+                  event = nil
+                  scope = gateway_request.env[Datadog::AppSec::Ext::SCOPE_KEY]
+
+                  AppSec::Reactive::Operation.new('rails.request.action') do |op|
+                    Rails::Reactive::Action.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,
+                          request: gateway_request,
+                          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 = Rails::Reactive::Action.publish(op, gateway_request)
+                  end
+
+                  next [nil, [[:block, event]]] if block
+
+                  ret, res = stack.call(gateway_request.request)
+
+                  if event
+                    res ||= []
+                    res << [:monitor, event]
+                  end
+
+                  [ret, res]
+                end
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/integration.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+require_relative '../integration'
+
+require_relative 'patcher'
+require_relative 'request_middleware'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        # Description of Rails integration
+        class Integration
+          include Datadog::AppSec::Contrib::Integration
+
+          MINIMUM_VERSION = Gem::Version.new('4')
+
+          register_as :rails, auto_patch: false
+
+          def self.version
+            Gem.loaded_specs['railties'] && Gem.loaded_specs['railties'].version
+          end
+
+          def self.loaded?
+            !defined?(::Rails).nil?
+          end
+
+          def self.compatible?
+            super && version >= MINIMUM_VERSION
+          end
+
+          def self.auto_instrument?
+            true
+          end
+
+          def patcher
+            Patcher
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/contrib/rails/patcher.rb

@@ -0,0 +1,166 @@
+# frozen_string_literal: true
+
+require_relative '../../../core/utils/only_once'
+
+require_relative '../patcher'
+require_relative 'framework'
+require_relative '../../response'
+require_relative '../rack/request_middleware'
+require_relative '../rack/request_body_middleware'
+require_relative 'gateway/watcher'
+require_relative 'gateway/request'
+
+require_relative '../../../tracing/contrib/rack/middlewares'
+
+module Datadog
+  module AppSec
+    module Contrib
+      module Rails
+        # Patcher for AppSec on Rails
+        module Patcher
+          include Datadog::AppSec::Contrib::Patcher
+
+          BEFORE_INITIALIZE_ONLY_ONCE_PER_APP = Hash.new { |h, key| h[key] = Datadog::Core::Utils::OnlyOnce.new }
+          AFTER_INITIALIZE_ONLY_ONCE_PER_APP = Hash.new { |h, key| h[key] = Datadog::Core::Utils::OnlyOnce.new }
+
+          module_function
+
+          def patched?
+            Patcher.instance_variable_get(:@patched)
+          end
+
+          def target_version
+            Integration.version
+          end
+
+          def patch
+            Gateway::Watcher.watch
+            patch_before_initialize
+            patch_after_initialize
+
+            Patcher.instance_variable_set(:@patched, true)
+          end
+
+          def patch_before_initialize
+            ::ActiveSupport.on_load(:before_initialize) do
+              Datadog::AppSec::Contrib::Rails::Patcher.before_initialize(self)
+            end
+          end
+
+          def before_initialize(app)
+            BEFORE_INITIALIZE_ONLY_ONCE_PER_APP[app].run do
+              # Middleware must be added before the application is initialized.
+              # Otherwise the middleware stack will be frozen.
+              add_middleware(app) if Datadog.configuration.tracing[:rails][:middleware]
+              patch_process_action
+            end
+          end
+
+          def add_middleware(app)
+            # Add trace middleware
+            if include_middleware?(Datadog::Tracing::Contrib::Rack::TraceMiddleware, app)
+              app.middleware.insert_after(
+                Datadog::Tracing::Contrib::Rack::TraceMiddleware,
+                Datadog::AppSec::Contrib::Rack::RequestMiddleware
+              )
+            else
+              app.middleware.insert_before(0, Datadog::AppSec::Contrib::Rack::RequestMiddleware)
+            end
+          end
+
+          # Hook into ActionController::Instrumentation#process_action, which encompasses action filters
+          module ProcessActionPatch
+            def process_action(*args)
+              env = request.env
+
+              context = env[Datadog::AppSec::Ext::SCOPE_KEY]
+
+              return super unless context
+
+              # TODO: handle exceptions, except for super
+
+              gateway_request = Gateway::Request.new(request)
+              request_return, request_response = Instrumentation.gateway.push('rails.request.action', gateway_request) do
+                super
+              end
+
+              if request_response
+                blocked_event = request_response.find { |action, _options| action == :block }
+                if blocked_event
+                  @_response = AppSec::Response.negotiate(
+                    env,
+                    blocked_event.last[:actions]
+                  ).to_action_dispatch_response
+                  request_return = @_response.body
+                end
+              end
+
+              request_return
+            end
+          end
+
+          def patch_process_action
+            ::ActionController::Metal.prepend(ProcessActionPatch)
+          end
+
+          def include_middleware?(middleware, app)
+            found = false
+
+            # find tracer middleware reference in Rails::Configuration::MiddlewareStackProxy
+            app.middleware.instance_variable_get(:@operations).each do |operation|
+              args = case operation
+                     when Array
+                       # rails 5.2
+                       _op, args = operation
+                       args
+                     when Proc
+                       if operation.binding.local_variables.include?(:args)
+                         # rails 6.0, 6.1
+                         operation.binding.local_variable_get(:args)
+                       else
+                         # rails 7.0 uses ... to pass args
+                         args_getter = Class.new do
+                           def method_missing(_op, *args) # rubocop:disable Style/MissingRespondToMissing
+                             args
+                           end
+                         end.new
+                         operation.call(args_getter)
+                       end
+                     else
+                       # unknown, pass through
+                       []
+                     end
+
+              found = true if args.include?(middleware)
+            end
+
+            found
+          end
+
+          def inspect_middlewares(app)
+            Datadog.logger.debug { +'Rails middlewares: ' << app.middleware.map(&:inspect).inspect }
+          end
+
+          def patch_after_initialize
+            ::ActiveSupport.on_load(:after_initialize) do
+              Datadog::AppSec::Contrib::Rails::Patcher.after_initialize(self)
+            end
+          end
+
+          def after_initialize(app)
+            AFTER_INITIALIZE_ONLY_ONCE_PER_APP[app].run do
+              # Finish configuring the tracer after the application is initialized.
+              # We need to wait for some things, like application name, middleware stack, etc.
+              setup_security
+              inspect_middlewares(app)
+            end
+          end
+
+          def setup_security
+            Datadog::AppSec::Contrib::Rails::Framework.setup
+          end
+        end
+      end
+    end
+  end
+end