ddtrace 1.14.0 → 1.16.0

data/lib/datadog/appsec/event.rb

@@ -1,4 +1,6 @@
 require 'json'
+require 'zlib'
+require 'base64'
 
 require_relative 'rate_limiter'
 
@@ -34,78 +36,132 @@ module Datadog
         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
-      def self.record(span, *events)
-        # ensure rate limiter is called only when there are events to record
-        return if events.empty? || span.nil?
+      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)
+          Datadog::AppSec::RateLimiter.limit(:traces) do
+            record_via_span(span, *events)
+          end
         end
-      end
 
-      def self.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
+        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
+            )
 
-          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)
-          # complex types are unsupported, we need to serialize to a string
-          triggers = service_entry_tags.delete('_dd.appsec.triggers')
-          span.set_tag('_dd.appsec.json', JSON.dump({ triggers: triggers }))
-
-          # apply tags to service entry span
-          service_entry_tags.each do |key, value|
-            span.set_tag(key, value)
+            # 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
-      end
 
-      def self.build_service_entry_tags(event_group)
-        event_group.each_with_object({}) do |event, tags|
-          # TODO: assume HTTP request context for now
+        # 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 (request = event[:request])
-            request_headers = request.headers.select do |k, _|
-              ALLOWED_REQUEST_HEADERS.include?(k.downcase)
+            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
 
-            request_headers.each do |header, value|
-              tags["http.request.headers.#{header}"] = value
+            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['http.host'] = request.host
-            tags['http.useragent'] = request.user_agent
-            tags['network.client.ip'] = request.remote_addr
+            tags
           end
 
-          if (response = event[:response])
-            response_headers = response.headers.select do |k, _|
-              ALLOWED_RESPONSE_HEADERS.include?(k.downcase)
-            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
 
-            response_headers.each do |header, value|
-              tags["http.response.headers.#{header}"] = value
-            end
+        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
 
-          tags['_dd.origin'] = 'appsec'
+        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
 
-          # accumulate triggers
-          tags['_dd.appsec.triggers'] ||= []
-          tags['_dd.appsec.triggers'] += event[:waf_result].data
+        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

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

@@ -24,7 +24,7 @@ module Datadog
                 scope = Datadog::AppSec.active_scope
 
                 AppSec::Reactive::Operation.new('identity.set_user') do |op|
-                  Monitor::Reactive::SetUser.subscribe(op, scope.processor_context) do |result, _block|
+                  Monitor::Reactive::SetUser.subscribe(op, scope.processor_context) do |result|
                     if result.status == :match
                       # TODO: should this hash be an Event instance instead?
                       event = {
@@ -44,10 +44,10 @@ module Datadog
                     end
                   end
 
-                  _result, block = Monitor::Reactive::SetUser.publish(op, user)
+                  block = Monitor::Reactive::SetUser.publish(op, user)
                 end
 
-                throw(Datadog::AppSec::Ext::INTERRUPT, [nil, [:block, event]]) if block
+                throw(Datadog::AppSec::Ext::INTERRUPT, [nil, [[:block, event]]]) if block
 
                 ret, res = stack.call(user)
 

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

@@ -38,11 +38,8 @@ module Datadog
               when :match
                 Datadog.logger.debug { "WAF: #{result.inspect}" }
 
-                block = result.actions.include?('block')
-
-                yield [result, block]
-
-                throw(:block, [result, true]) if block
+                yield result
+                throw(:block, true) unless result.actions.empty?
               when :ok
                 Datadog.logger.debug { "WAF OK: #{result.inspect}" }
               when :invalid_call

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_merger.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require_relative '../assets'
+
 module Datadog
   module AppSec
     class Processor
@@ -16,8 +18,25 @@ module Datadog
           end
         end
 
+        DEFAULT_WAF_PROCESSORS = begin
+          JSON.parse(Datadog::AppSec::Assets.waf_processors)
+        rescue StandardError => e
+          Datadog.logger.error { "libddwaf rulemerger failed to parse default waf processors. Error: #{e.inspect}" }
+          []
+        end
+
+        DEFAULT_WAF_SCANNERS = begin
+          JSON.parse(Datadog::AppSec::Assets.waf_scanners)
+        rescue StandardError => e
+          Datadog.logger.error { "libddwaf rulemerger failed to parse default waf scanners. Error: #{e.inspect}" }
+          []
+        end
+
         class << self
-          def merge(rules:, data: [], overrides: [], exclusions: [], custom_rules: [])
+          def merge(
+            rules:, data: [], overrides: [], exclusions: [], custom_rules: [],
+            processors: DEFAULT_WAF_PROCESSORS, scanners: DEFAULT_WAF_SCANNERS
+          )
             combined_rules = combine_rules(rules)
 
             combined_data = combine_data(data) if data.any?
@@ -29,7 +48,8 @@ module Datadog
             combined_rules['rules_override'] = combined_overrides if combined_overrides
             combined_rules['exclusions'] = combined_exclusions if combined_exclusions
             combined_rules['custom_rules'] = combined_custom_rules if combined_custom_rules
-
+            combined_rules['processors'] = processors
+            combined_rules['scanners'] = scanners
             combined_rules
           end
 

data/lib/datadog/appsec/processor.rb

@@ -22,8 +22,15 @@ module Datadog
 
           start_ns = Core::Utils::Time.get_time(:nanosecond)
 
-          # this WAF::Context#run call is not thread safe as it mutates the context
-          # TODO: remove multiple assignment
+          input.reject! do |_, v|
+            case v
+            when TrueClass, FalseClass
+              false
+            else
+              v.nil? ? true : v.empty?
+            end
+          end
+
           _code, res = @context.run(input, timeout)
 
           stop_ns = Core::Utils::Time.get_time(:nanosecond)
@@ -38,15 +45,36 @@ module Datadog
           @run_mutex.unlock
         end
 
+        def extract_schema
+          return unless extract_schema?
+
+          input = {
+            'waf.context.processor' => {
+              'extract-schema' => true
+            }
+          }
+
+          _code, res = @context.run(input, WAF::LibDDWAF::DDWAF_RUN_TIMEOUT)
+
+          res
+        end
+
         def finalize
           @context.finalize
         end
+
+        private
+
+        def extract_schema?
+          Datadog.configuration.appsec.api_security.enabled &&
+            Datadog.configuration.appsec.api_security.sample_rate.sample?
+        end
       end
 
-      attr_reader :ruleset_info, :addresses
+      attr_reader :diagnostics, :addresses
 
       def initialize(ruleset:)
-        @ruleset_info = nil
+        @diagnostics = nil
         @addresses = []
         settings = Datadog.configuration.appsec
 
@@ -83,7 +111,7 @@ module Datadog
         }
 
         @handle = Datadog::AppSec::WAF::Handle.new(ruleset, obfuscator: obfuscator_config)
-        @ruleset_info = @handle.ruleset_info
+        @diagnostics = @handle.diagnostics
         @addresses = @handle.required_addresses
 
         true
@@ -92,7 +120,7 @@ module Datadog
           "libddwaf failed to initialize, error: #{e.inspect}"
         end
 
-        @ruleset_info = e.ruleset_info if e.ruleset_info
+        @diagnostics = e.diagnostics if e.diagnostics
 
         false
       rescue StandardError => e

data/lib/datadog/appsec/remote.rb

@@ -31,6 +31,7 @@ module Datadog
           CAP_ASM_RESPONSE_BLOCKING,
           CAP_ASM_DD_RULES,
           CAP_ASM_CUSTOM_RULES,
+          CAP_ASM_CUSTOM_BLOCKING_RESPONSE,
         ].freeze
 
         ASM_PRODUCTS = [
@@ -63,6 +64,7 @@ module Datadog
             data = []
             overrides = []
             exclusions = []
+            actions = []
 
             repository.contents.each do |content|
               parsed_content = parse_content(content)
@@ -76,6 +78,7 @@ module Datadog
                 overrides << parsed_content['rules_override'] if parsed_content['rules_override']
                 exclusions << parsed_content['exclusions'] if parsed_content['exclusions']
                 custom_rules << parsed_content['custom_rules'] if parsed_content['custom_rules']
+                actions.concat(parsed_content['actions']) if parsed_content['actions']
               end
             end
 
@@ -95,7 +98,7 @@ module Datadog
               custom_rules: custom_rules,
             )
 
-            Datadog::AppSec.reconfigure(ruleset: ruleset)
+            Datadog::AppSec.reconfigure(ruleset: ruleset, actions: actions)
           end
 
           [receiver]

data/lib/datadog/appsec/response.rb

@@ -28,19 +28,80 @@ module Datadog
       end
 
       class << self
-        def negotiate(env)
+        def negotiate(env, actions)
+          # @type var configured_response: Response?
+          configured_response = nil
+          actions.each do |action|
+            # Need to use next to make steep happy :(
+            # I rather use break to stop the execution
+            next if configured_response
+
+            action_configuration = AppSec::Processor::Actions.fecth_configuration(action)
+            next unless action_configuration
+
+            configured_response = case action_configuration['type']
+                                  when 'block_request'
+                                    block_response(env, action_configuration['parameters'])
+                                  when 'redirect_request'
+                                    redirect_response(env, action_configuration['parameters'])
+                                  end
+          end
+
+          configured_response || default_response(env)
+        end
+
+        private
+
+        def default_response(env)
           content_type = content_type(env)
 
-          Datadog.logger.debug { "negotiated response content type: #{content_type}" }
+          body = []
+          body << content(content_type)
 
           Response.new(
             status: 403,
             headers: { 'Content-Type' => content_type },
-            body: [Datadog::AppSec::Assets.blocked(format: CONTENT_TYPE_TO_FORMAT[content_type])]
+            body: body,
           )
         end
 
-        private
+        def block_response(env, options)
+          content_type = if options['type'] == 'auto'
+                           content_type(env)
+                         else
+                           FORMAT_TO_CONTENT_TYPE[options['type']]
+                         end
+
+          body = []
+          body << content(content_type)
+
+          Response.new(
+            status: options['status_code'] || 403,
+            headers: { 'Content-Type' => content_type },
+            body: body,
+          )
+        end
+
+        def redirect_response(env, options)
+          if options['location'] && !options['location'].empty?
+            content_type = content_type(env)
+
+            status = options['status_code'] >= 300 && options['status_code'] < 400 ? options['status_code'] : 303
+
+            headers = {
+              'Content-Type' => content_type,
+              'Location' => options['location']
+            }
+
+            Response.new(
+              status: status,
+              headers: headers,
+              body: [],
+            )
+          else
+            default_response(env)
+          end
+        end
 
         CONTENT_TYPE_TO_FORMAT = {
           'application/json' => :json,
@@ -48,6 +109,11 @@ module Datadog
           'text/plain' => :text,
         }.freeze
 
+        FORMAT_TO_CONTENT_TYPE = {
+          'json' => 'application/json',
+          'html' => 'text/html',
+        }.freeze
+
         DEFAULT_CONTENT_TYPE = 'application/json'
 
         def content_type(env)
@@ -67,6 +133,18 @@ module Datadog
         rescue Datadog::AppSec::Utils::HTTP::MediaRange::ParseError
           DEFAULT_CONTENT_TYPE
         end
+
+        def content(content_type)
+          content_format = CONTENT_TYPE_TO_FORMAT[content_type]
+
+          using_default = Datadog.configuration.appsec.block.templates.using_default?(content_format)
+
+          if using_default
+            Datadog::AppSec::Assets.blocked(format: content_format)
+          else
+            Datadog.configuration.appsec.block.templates.send(content_format)
+          end
+        end
       end
     end
   end

data/lib/datadog/appsec/sample_rate.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # SampleRate basic sample rate
+    class SampleRate
+      attr_reader :rate
+
+      def initialize(rate)
+        @rate = rate
+      end
+
+      def sample?
+        return false if rate <= 0
+        return true if rate >= 1
+
+        Kernel.rand < rate
+      end
+    end
+  end
+end

data/lib/datadog/appsec.rb

@@ -23,12 +23,12 @@ module Datadog
         appsec_component.processor if appsec_component
       end
 
-      def reconfigure(ruleset:)
+      def reconfigure(ruleset:, actions:)
         appsec_component = components.appsec
 
         return unless appsec_component
 
-        appsec_component.reconfigure(ruleset: ruleset)
+        appsec_component.reconfigure(ruleset: ruleset, actions: actions)
       end
 
       def reconfigure_lock(&block)