datadog 2.28.0 → 2.41.0

data/lib/datadog/appsec/contrib/rest_client/request_ssrf_detection_patch.rb

@@ -1,8 +1,11 @@
 # frozen_string_literal: true
 
-require_relative '../../event'
-require_relative '../../trace_keeper'
-require_relative '../../security_event'
+require_relative "../../event"
+require_relative "../../trace_keeper"
+require_relative "../../security_event"
+require_relative "../../utils/http/body"
+require_relative "../../utils/http/body_reader"
+require_relative "../../utils/http/media_type"
 
 module Datadog
   module AppSec
@@ -10,35 +13,149 @@ module Datadog
       module RestClient
         # Module that adds SSRF detection to RestClient::Request#execute
         module RequestSSRFDetectionPatch
+          REDIRECT_STATUS_CODES = (300..399).freeze
+
           def execute(&block)
             context = AppSec.active_context
             return super unless context && AppSec.rasp_enabled?
 
-            timeout = Datadog.configuration.appsec.waf_timeout
+            headers = normalize_request_headers
+            # @type var ephemeral_data: ::Datadog::AppSec::Context::input_data
             ephemeral_data = {
-              'server.io.net.url' => url,
-              'server.io.net.request.method' => method.to_s.upcase,
-              'server.io.net.request.headers' => normalize_request_headers
+              "server.io.net.url" => url,
+              "server.io.net.request.method" => method.to_s.upcase,
+              "server.io.net.request.headers" => headers,
             }
 
+            is_redirect = context.state[:downstream_redirect_url] == url
+            if is_redirect
+              context.state.delete(:downstream_redirect_url)
+              sample_body = true
+            else
+              sample_body = mark_body_sampling!(context)
+            end
+
+            if !is_redirect && sample_body
+              body = parse_request_body(payload, content_type: headers["content-type"])
+              ephemeral_data["server.io.net.request.body"] = body if body
+            end
+
+            timeout = Datadog.configuration.appsec.waf_timeout
             result = context.run_rasp(Ext::RASP_SSRF, {}, ephemeral_data, timeout, phase: Ext::RASP_REQUEST_PHASE)
             handle(result, context: context) if result.match?
 
-            response = super
+            # NOTE: RestClient raises exceptions for non-2xx responses. For POST/PUT/PATCH
+            #       requests with 3xx redirects, RestClient raises instead of auto-following.
+            #       We rescue to process the response before re-raising.
+            begin
+              response = super
+            rescue ::RestClient::Exception => e
+              response = e.response
+              process_response(response, sample_body: sample_body) if response.is_a?(::RestClient::AbstractResponse)
+
+              raise
+            end
+
+            process_response(response, sample_body: sample_body) if response.is_a?(::RestClient::AbstractResponse)
+            response
+          end
+
+          def process_response(response, sample_body:)
+            context = AppSec.active_context
+            return unless context
 
+            headers = normalize_response_headers(response)
+            # @type var ephemeral_data: ::Datadog::AppSec::Context::input_data
             ephemeral_data = {
-              'server.io.net.response.status' => response.code.to_s,
-              'server.io.net.response.headers' => normalize_response_headers(response)
+              "server.io.net.response.status" => response.code.to_s,
+              "server.io.net.response.headers" => headers,
             }
 
+            is_redirect = REDIRECT_STATUS_CODES.cover?(response.code.to_i) && headers.key?("location")
+            context.state[:downstream_redirect_url] = URI.join(url, headers["location"]).to_s if is_redirect && sample_body
+
+            if sample_body && !is_redirect
+              body = parse_response_body(response.body, headers: headers, context: context)
+              ephemeral_data["server.io.net.response.body"] = body if body
+            end
+
+            timeout = Datadog.configuration.appsec.waf_timeout
             result = context.run_rasp(Ext::RASP_SSRF, {}, ephemeral_data, timeout, phase: Ext::RASP_RESPONSE_PHASE)
             handle(result, context: context) if result.match?
-
-            response
           end
 
           private
 
+          def mark_body_sampling!(context)
+            max = Datadog.configuration.appsec.api_security.downstream_body_analysis.max_requests
+            return false if context.state[:downstream_body_analyzed_count] >= max
+            return false unless context.downstream_body_sampler.sample?
+
+            context.state[:downstream_body_analyzed_count] += 1
+            true
+          end
+
+          def parse_request_body(body, content_type:)
+            media_type = Utils::HTTP::MediaType.parse(content_type)
+            return unless media_type
+
+            # NOTE: Request body analysis is best-effort, non-rewindable payloads are skipped
+            limit = Datadog.configuration.appsec.api_security.downstream_body_analysis.max_downstream_body_bytes
+            content = read_body(body, limit: limit)
+            return if content.nil? || content.bytesize > limit
+
+            Utils::HTTP::Body.parse(content, media_type: media_type, limit: limit)
+          end
+
+          def parse_response_body(body, headers:, context:)
+            return unless readable_body?(body)
+
+            media_type = Utils::HTTP::MediaType.parse(headers["content-type"])
+            if !media_type || media_type.type != "application"
+              context.metrics.record_ignored_downstream_response_body(:content_type_invalid)
+              return
+            end
+
+            subtype = media_type.subtype
+            if subtype != "json" && !subtype.end_with?("+json") && subtype != "x-www-form-urlencoded"
+              context.metrics.record_ignored_downstream_response_body(:content_type_invalid)
+              return
+            end
+
+            content_length_value = headers["content-length"]
+            if !content_length_value.is_a?(String) || !content_length_value.match?(/\A[1-9][0-9]*\z/)
+              context.metrics.record_ignored_downstream_response_body(:content_length_missing)
+              return
+            end
+
+            content_length = content_length_value.to_i
+            max = Datadog.configuration.appsec.api_security.downstream_body_analysis.max_downstream_body_bytes
+            if content_length > max
+              context.metrics.record_ignored_downstream_response_body(:content_length_too_big)
+              return
+            end
+
+            content = read_body(body, limit: content_length)
+            return if content.nil?
+
+            if content.bytesize > content_length
+              context.metrics.record_ignored_downstream_response_body(:content_exceed_content_length)
+              return
+            end
+
+            Utils::HTTP::Body.parse(content, media_type: media_type, limit: content_length)
+          end
+
+          def readable_body?(body)
+            body.is_a?(String) || (body.respond_to?(:read) && body.respond_to?(:rewind))
+          end
+
+          def read_body(body, limit:)
+            Utils::HTTP::BodyReader.read(body, limit: limit, rewind_before_read: true)
+          rescue
+            nil
+          end
+
           # NOTE: Starting version 2.1.0 headers are already normalized via internal
           #       variable `@processed_headers_lowercase`. In case it's available,
           #       we use it to avoid unnecessary transformation.
@@ -55,7 +172,8 @@ module Datadog
           # FIXME: Steep has issues with `transform_values!` modifying the original
           #        type and it failed with "Cannot allow block body" error
           def normalize_response_headers(response) # steep:ignore MethodBodyTypeMismatch
-            response.net_http_res.to_hash.transform_values! { |value| Array(value).join(', ') } # steep:ignore BlockBodyTypeMismatch
+            response.net_http_res.to_hash
+              .transform_values! { |value| Array(value).join(", ") } # steep:ignore BlockBodyTypeMismatch
           end
 
           def handle(result, context:)

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

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative '../../rack/gateway/request'
+require_relative "../../rack/gateway/request"
 
 module Datadog
   module AppSec

data/lib/datadog/appsec/contrib/sinatra/gateway/route_params.rb

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative '../../../instrumentation/gateway/argument'
+require_relative "../../../instrumentation/gateway/argument"
 
 module Datadog
   module AppSec

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

@@ -1,9 +1,9 @@
 # frozen_string_literal: true
 
-require_relative '../../../event'
-require_relative '../../../trace_keeper'
-require_relative '../../../security_event'
-require_relative '../../../instrumentation/gateway'
+require_relative "../../../event"
+require_relative "../../../trace_keeper"
+require_relative "../../../security_event"
+require_relative "../../../instrumentation/gateway"
 
 module Datadog
   module AppSec
@@ -22,12 +22,35 @@ module Datadog
               end
 
               def watch_request_dispatch(gateway = Instrumentation.gateway)
-                gateway.watch('sinatra.request.dispatch', :appsec) do |stack, gateway_request|
+                gateway.watch("sinatra.request.dispatch") do |stack, gateway_request|
                   context = gateway_request.env[AppSec::Ext::CONTEXT_KEY] # : Context
 
-                  persistent_data = {
-                    'server.request.body' => gateway_request.form_hash
-                  }
+                  context.state[:web_framework] = "sinatra"
+
+                  request = gateway_request.request
+                  next stack.call(request) unless gateway_request.collectable_body?
+
+                  # NOTE: A limit of 0 disables request body collection entirely.
+                  limit = Datadog.configuration.appsec.body_parsing_size_limit
+                  next stack.call(request) if limit.zero?
+
+                  persistent_data = {}
+                  byte_length = gateway_request.body_bytesize(limit)
+
+                  if byte_length
+                    persistent_data["server.request.body.byte_length"] = byte_length
+
+                    if byte_length <= limit
+                      body = gateway_request.form_hash
+                      persistent_data["server.request.body"] = body if body
+                    end
+                  # NOTE: Body was parsed before measurement, keep byte_length unset
+                  elsif gateway_request.env.key?("rack.request.form_hash")
+                    body = gateway_request.env["rack.request.form_hash"]
+                    persistent_data["server.request.body"] = body if body
+                  end
+
+                  next stack.call(request) if persistent_data.empty?
 
                   result = context.run_waf(persistent_data, {}, Datadog.configuration.appsec.waf_timeout)
 
@@ -49,12 +72,12 @@ module Datadog
               end
 
               def watch_request_routed(gateway = Instrumentation.gateway)
-                gateway.watch('sinatra.request.routed', :appsec) do |stack, args|
+                gateway.watch("sinatra.request.routed") do |stack, args|
                   gateway_request, gateway_route_params = args # : [Gateway::Request, Gateway::RouteParams]
                   context = gateway_request.env[AppSec::Ext::CONTEXT_KEY] # : Context
 
                   persistent_data = {
-                    'server.request.path_params' => gateway_route_params.params
+                    "server.request.path_params" => gateway_route_params.params,
                   }
 
                   result = context.run_waf(persistent_data, {}, Datadog.configuration.appsec.waf_timeout)
@@ -75,11 +98,11 @@ module Datadog
               end
 
               def watch_response_body_json(gateway = Instrumentation.gateway)
-                gateway.watch('sinatra.response.body.json', :appsec) do |stack, container|
+                gateway.watch("sinatra.response.body.json") do |stack, container|
                   context = container.context # : Context
 
                   persistent_data = {
-                    'server.response.body' => container.data
+                    "server.response.body" => container.data,
                   }
                   result = context.run_waf(persistent_data, {}, Datadog.configuration.appsec.waf_timeout)
 

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

@@ -1,9 +1,9 @@
 # frozen_string_literal: true
 
-require_relative '../integration'
+require_relative "../integration"
 
-require_relative 'patcher'
-require_relative 'request_middleware'
+require_relative "patcher"
+require_relative "request_middleware"
 
 module Datadog
   module AppSec
@@ -13,12 +13,12 @@ module Datadog
         class Integration
           include Datadog::AppSec::Contrib::Integration
 
-          MINIMUM_VERSION = Gem::Version.new('1.4.0')
+          MINIMUM_VERSION = Gem::Version.new("1.4.0")
 
           register_as :sinatra
 
           def self.version
-            Gem.loaded_specs['sinatra']&.version
+            Gem.loaded_specs["sinatra"]&.version
           end
 
           def self.loaded?

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

@@ -1,15 +1,15 @@
 # frozen_string_literal: true
 
-require_relative '../../../tracing/contrib'
+require_relative "../../../tracing/contrib"
 
-require_relative '../../response'
-require_relative '../rack/request_middleware'
-require_relative 'framework'
-require_relative 'gateway/watcher'
-require_relative 'gateway/route_params'
-require_relative 'gateway/request'
-require_relative 'patches/json_patch'
-require_relative '../../../tracing/contrib/sinatra/framework'
+require_relative "../../response"
+require_relative "../rack/request_middleware"
+require_relative "framework"
+require_relative "gateway/watcher"
+require_relative "gateway/route_params"
+require_relative "gateway/request"
+require_relative "patches/json_patch"
+require_relative "../../../tracing/contrib/sinatra/framework"
 
 module Datadog
   module AppSec
@@ -59,7 +59,7 @@ module Datadog
             # TODO: handle exceptions, except for super
 
             gateway_request = Gateway::Request.new(env)
-            request_return, _gateway_request = Instrumentation.gateway.push('sinatra.request.dispatch', gateway_request) do
+            request_return, _gateway_request = Instrumentation.gateway.push("sinatra.request.dispatch", gateway_request) do
               super
             end
 
@@ -91,7 +91,7 @@ module Datadog
               gateway_request = Gateway::Request.new(env)
               gateway_route_params = Gateway::RouteParams.new(route_params)
 
-              Instrumentation.gateway.push('sinatra.request.routed', [gateway_request, gateway_route_params])
+              Instrumentation.gateway.push("sinatra.request.routed", [gateway_request, gateway_route_params])
 
               yield(*super_args)
             end

data/lib/datadog/appsec/contrib/sinatra/patches/json_patch.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative '../../../utils/hash_coercion'
-require_relative '../../../instrumentation/gateway/argument'
+require_relative "../../../utils/hash_coercion"
+require_relative "../../../instrumentation/gateway/argument"
 
 module Datadog
   module AppSec
@@ -19,7 +19,7 @@ module Datadog
               return super unless data
 
               container = Instrumentation::Gateway::DataContainer.new(data, context: context)
-              Instrumentation.gateway.push('sinatra.response.body.json', container)
+              Instrumentation.gateway.push("sinatra.response.body.json", container)
 
               super
             end

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

@@ -6,7 +6,8 @@ module Datadog
       module Sinatra
         # Rack middleware for AppSec on Sinatra
         class RequestMiddleware
-          def initialize(app, opt = {})
+          # TODO: opt is never used, it can probably be safely removed
+          def initialize(app, opt = {}) # steep:ignore DifferentMethodParameterKind
             @app = app
           end
 

data/lib/datadog/appsec/counter_sampler.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require_relative "../core/knuth_sampler"
+
+module Datadog
+  module AppSec
+    # Sampler that uses an internal counter to make deterministic sampling decisions.
+    #
+    # Each call to {#sample?} increments the counter and uses it as input to
+    # the underlying Knuth multiplicative hash algorithm.
+    #
+    # @api private
+    class CounterSampler
+      def initialize(rate = 1.0)
+        @sampler = Core::KnuthSampler.new(rate)
+        @counter = 0
+      end
+
+      def sample?
+        @counter += 1
+        @sampler.sample?(@counter)
+      end
+    end
+  end
+end

data/lib/datadog/appsec/default_header_tags.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+module Datadog
+  module AppSec
+    # Sets the always-on span tags built from a fixed allowlist of request and
+    # response headers.
+    #
+    # NOTE: Unlike {Event#record}, which reports request headers only when
+    #       a security event fires, these are tagged on every request/response
+    #
+    # @api private
+    module DefaultHeaderTags
+      # NOTE: Contains additional WAF vendor headers
+      REQUEST_HEADERS_TAGS = %w[
+        accept
+        content-type
+        user-agent
+        akamai-user-risk
+        x-amzn-trace-id
+        x-cloud-trace-context
+        x-appgw-trace-id
+        x-sigsci-requestid
+        x-sigsci-tags
+        cf-ray
+        cloudfront-viewer-ja3-fingerprint
+      ].freeze
+
+      RESPONSE_HEADERS_TAGS = %w[
+        content-type
+        content-length
+        content-encoding
+        content-language
+      ].freeze
+
+      module_function
+
+      def tag_request(span, headers)
+        REQUEST_HEADERS_TAGS.each do |name|
+          value = headers.get(name)
+          span.set_tag("http.request.headers.#{name}", value) if value
+        end
+      end
+
+      def tag_response(span, headers)
+        RESPONSE_HEADERS_TAGS.each do |name|
+          value = headers.get(name)
+          span.set_tag("http.response.headers.#{name}", value.to_s) if value
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/event.rb

@@ -1,15 +1,15 @@
 # frozen_string_literal: true
 
-require 'json'
-require_relative 'rate_limiter'
-require_relative 'trace_keeper'
-require_relative 'compressed_json'
+require "json"
+require_relative "rate_limiter"
+require_relative "trace_keeper"
+require_relative "compressed_json"
 
 module Datadog
   module AppSec
     # AppSec event
     module Event
-      ATTRIBUTES_SCHEMA_KEY_PREFIX = '_dd.appsec.s.'
+      ATTRIBUTES_SCHEMA_KEY_PREFIX = "_dd.appsec.s."
       ATTRIBUTES_SCHEMA_MAX_COMPRESSED_SIZE = 25000
       ALLOWED_REQUEST_HEADERS = %w[
         x-forwarded-for
@@ -32,25 +32,18 @@ module Datadog
         accept-language
       ].freeze
 
-      ALLOWED_RESPONSE_HEADERS = %w[
-        content-length
-        content-type
-        content-encoding
-        content-language
-      ].freeze
-
       class << self
         def tag(context, waf_result)
           return if context.span.nil?
 
-          if waf_result.actions.key?('block_request') || waf_result.actions.key?('redirect_request')
-            context.span.set_tag('appsec.blocked', 'true')
+          if waf_result.actions.key?("block_request") || waf_result.actions.key?("redirect_request")
+            context.span.set_tag("appsec.blocked", "true")
           end
 
-          context.span.set_tag('appsec.event', 'true')
+          context.span.set_tag("appsec.event", "true")
         end
 
-        def record(context, request: nil, response: nil)
+        def record(context, request: nil)
           return if context.events.empty? || context.span.nil?
 
           Datadog::AppSec::RateLimiter.thread_local.limit do
@@ -64,9 +57,8 @@ module Datadog
               if event_group.any? { |event| event.keep? || event.schema? }
                 TraceKeeper.keep!(trace)
 
-                context.span['_dd.origin'] = 'appsec'
+                context.span["_dd.origin"] = "appsec"
                 context.span.set_tags(request_tags(request)) if request
-                context.span.set_tags(response_tags(response)) if response
               end
 
               context.span.set_tags(waf_tags(event_group))
@@ -79,9 +71,9 @@ module Datadog
         def request_tags(request)
           tags = {}
 
-          tags['http.host'] = request.host if request.host
-          tags['http.useragent'] = request.user_agent if request.user_agent
-          tags['network.client.ip'] = request.remote_addr if request.remote_addr
+          tags["http.host"] = request.host if request.host
+          tags["http.useragent"] = request.user_agent if request.user_agent
+          tags["network.client.ip"] = request.remote_addr if request.remote_addr
 
           request.headers.each_with_object(tags) do |(name, value), memo|
             next unless ALLOWED_REQUEST_HEADERS.include?(name)
@@ -90,14 +82,6 @@ module Datadog
           end
         end
 
-        def response_tags(response)
-          response.headers.each_with_object({}) do |(name, value), memo|
-            next unless ALLOWED_RESPONSE_HEADERS.include?(name)
-
-            memo["http.response.headers.#{name}"] = value
-          end
-        end
-
         def waf_tags(security_events)
           triggers = []
 
@@ -119,7 +103,7 @@ module Datadog
             end
           end
 
-          tags['_dd.appsec.json'] = json_parse({triggers: triggers}) unless triggers.empty?
+          tags["_dd.appsec.json"] = json_parse({triggers: triggers}) unless triggers.empty?
           tags
         end
 
@@ -129,7 +113,7 @@ module Datadog
         def json_parse(value)
           JSON.dump(value)
         rescue ArgumentError, Encoding::UndefinedConversionError, JSON::JSONError => e
-          AppSec.telemetry.report(e, description: 'AppSec: Failed to convert value into JSON')
+          AppSec.telemetry.report(e, description: "AppSec: Failed to convert value into JSON")
 
           nil
         end

data/lib/datadog/appsec/ext.rb

@@ -3,23 +3,24 @@
 module Datadog
   module AppSec
     module Ext
-      RASP_SQLI = 'sql_injection'
-      RASP_LFI = 'lfi'
-      RASP_SSRF = 'ssrf'
-      RASP_REQUEST_PHASE = 'request'
-      RASP_RESPONSE_PHASE = 'response'
+      RASP_SQLI = "sql_injection"
+      RASP_LFI = "lfi"
+      RASP_SSRF = "ssrf"
+      RASP_REQUEST_PHASE = "request"
+      RASP_RESPONSE_PHASE = "response"
 
       PRODUCT_BIT = 0b00000010
 
       INTERRUPT = :datadog_appsec_interrupt
-      CONTEXT_KEY = 'datadog.appsec.context'
+      CONTEXT_KEY = "datadog.appsec.context"
       ACTIVE_CONTEXT_KEY = :datadog_appsec_active_context
-      EXPLOIT_PREVENTION_EVENT_CATEGORY = 'exploit'
+      EXPLOIT_PREVENTION_EVENT_CATEGORY = "exploit"
 
-      TAG_APPSEC_ENABLED = '_dd.appsec.enabled'
-      TAG_METASTRUCT_STACK_TRACE = '_dd.stack'
+      TAG_APPSEC_ENABLED = "_dd.appsec.enabled"
+      TAG_NORMALIZED_ROUTE = "_dd.appsec.normalized_route"
+      TAG_METASTRUCT_STACK_TRACE = "_dd.stack"
 
-      TELEMETRY_METRICS_NAMESPACE = 'appsec'
+      TELEMETRY_METRICS_NAMESPACE = "appsec"
     end
   end
 end

data/lib/datadog/appsec/extensions.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative 'configuration'
-require_relative '../core/configuration'
+require_relative "configuration"
+require_relative "../core/configuration"
 
 module Datadog
   module AppSec

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

@@ -7,10 +7,9 @@ module Datadog
         # NOTE: This class extracted as-is and will be deprecated
         # Instrumentation gateway middleware
         class Middleware
-          attr_reader :key, :block
+          attr_reader :block
 
-          def initialize(key, &block)
-            @key = key
+          def initialize(&block)
             @block = block
           end