ddtrace 0.53.0 → 1.9.0

data/lib/datadog/appsec/processor.rb

@@ -0,0 +1,194 @@
+# typed: ignore
+
+require_relative 'assets'
+
+module Datadog
+  module AppSec
+    # Processor integrates libddwaf into datadog/appsec
+    class Processor
+      # Context manages a sequence of runs
+      class Context
+        attr_reader :time_ns, :time_ext_ns, :timeouts, :events
+
+        def initialize(processor)
+          @context = Datadog::AppSec::WAF::Context.new(processor.send(:handle))
+          @time_ns = 0.0
+          @time_ext_ns = 0.0
+          @timeouts = 0
+          @events = []
+        end
+
+        def run(input, timeout = WAF::LibDDWAF::DDWAF_RUN_TIMEOUT)
+          start_ns = Core::Utils::Time.get_time(:nanosecond)
+
+          # TODO: remove multiple assignment
+          _code, res = _ = @context.run(input, timeout)
+          # @type var res: WAF::Result
+
+          stop_ns = Core::Utils::Time.get_time(:nanosecond)
+
+          @time_ns += res.total_runtime
+          @time_ext_ns += (stop_ns - start_ns)
+          @timeouts += 1 if res.timeout
+
+          res
+        end
+
+        def finalize
+          @context.finalize
+        end
+      end
+
+      attr_reader :ruleset_info, :addresses
+
+      def initialize
+        @ruleset_info = nil
+        @addresses = []
+
+        unless load_libddwaf && load_ruleset && create_waf_handle
+          Datadog.logger.warn { 'AppSec is disabled, see logged errors above' }
+
+          return
+        end
+
+        update_ip_denylist
+      end
+
+      def ready?
+        !@ruleset.nil? && !@handle.nil?
+      end
+
+      def new_context
+        Context.new(self)
+      end
+
+      def update_rule_data(data)
+        @handle.update_rule_data(data)
+      end
+
+      def toggle_rules(map)
+        @handle.toggle_rules(map)
+      end
+
+      def update_ip_denylist(denylist = Datadog::AppSec.settings.ip_denylist, id: 'blocked_ips')
+        denylist ||= []
+
+        ruledata_setting = [
+          {
+            'id' => id,
+            'type' => 'data_with_expiration',
+            'data' => denylist.map { |ip| { 'value' => ip.to_s, 'expiration' => 2**63 } }
+          }
+        ]
+
+        update_rule_data(ruledata_setting)
+      end
+
+      def finalize
+        @handle.finalize
+      end
+
+      protected
+
+      attr_reader :handle
+
+      private
+
+      def load_libddwaf
+        Processor.require_libddwaf && Processor.libddwaf_provides_waf?
+      end
+
+      def load_ruleset
+        ruleset_setting = Datadog::AppSec.settings.ruleset
+
+        begin
+          @ruleset = case ruleset_setting
+                     when :recommended, :risky, :strict
+                       JSON.parse(Datadog::AppSec::Assets.waf_rules(ruleset_setting))
+                     when String
+                       JSON.parse(File.read(ruleset_setting))
+                     when File, StringIO
+                       JSON.parse(ruleset_setting.read || '').tap { ruleset_setting.rewind }
+                     when Hash
+                       ruleset_setting
+                     else
+                       raise ArgumentError, "unsupported value for ruleset setting: #{ruleset_setting.inspect}"
+                     end
+
+          true
+        rescue StandardError => e
+          Datadog.logger.error do
+            "libddwaf ruleset failed to load, ruleset: #{ruleset_setting.inspect} error: #{e.inspect}"
+          end
+
+          false
+        end
+      end
+
+      def create_waf_handle
+        # TODO: this may need to be reset if the main Datadog logging level changes after initialization
+        Datadog::AppSec::WAF.logger = Datadog.logger if Datadog.logger.debug? && Datadog::AppSec.settings.waf_debug
+
+        obfuscator_config = {
+          key_regex: Datadog::AppSec.settings.obfuscator_key_regex,
+          value_regex: Datadog::AppSec.settings.obfuscator_value_regex,
+        }
+        @handle = Datadog::AppSec::WAF::Handle.new(@ruleset, obfuscator: obfuscator_config)
+        @ruleset_info = @handle.ruleset_info
+        @addresses = @handle.required_addresses
+
+        true
+      rescue WAF::LibDDWAF::Error => e
+        Datadog.logger.error do
+          "libddwaf failed to initialize, error: #{e.inspect}"
+        end
+
+        @ruleset_info = e.ruleset_info if e.ruleset_info
+
+        false
+      rescue StandardError => e
+        Datadog.logger.error do
+          "libddwaf failed to initialize, error: #{e.inspect}"
+        end
+
+        false
+      end
+
+      class << self
+        # check whether libddwaf is required *and* able to provide the needed feature
+        def libddwaf_provides_waf?
+          defined?(Datadog::AppSec::WAF) ? true : false
+        end
+
+        # libddwaf raises a LoadError on unsupported platforms; it may at some
+        # point succeed in being required yet not provide a specific needed feature.
+        def require_libddwaf
+          Datadog.logger.debug { "libddwaf platform: #{libddwaf_platform}" }
+
+          require 'libddwaf'
+
+          true
+        rescue LoadError => e
+          Datadog.logger.error do
+            'libddwaf failed to load,' \
+              "installed platform: #{libddwaf_platform} ruby platforms: #{ruby_platforms} error: #{e.inspect}"
+          end
+
+          false
+        end
+
+        def libddwaf_spec
+          Gem.loaded_specs['libddwaf']
+        end
+
+        def libddwaf_platform
+          libddwaf_spec ? libddwaf_spec.platform.to_s : 'unknown'
+        end
+
+        def ruby_platforms
+          Gem.platforms.map(&:to_s)
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/rate_limiter.rb

@@ -0,0 +1,60 @@
+# typed: ignore
+
+module Datadog
+  module AppSec
+    # Simple per-thread rate limiter
+    # Since AppSec marks sampling to keep on a security event, this limits the flood of egress traces involving AppSec
+    class RateLimiter
+      def initialize(rate)
+        @rate = rate
+        @timestamps = []
+      end
+
+      def limit
+        now = Time.now.to_f
+
+        loop do
+          oldest = @timestamps.first
+
+          break if oldest.nil? || now - oldest < 1
+
+          @timestamps.shift
+        end
+
+        @timestamps << now
+
+        if (count = @timestamps.count) <= @rate
+          yield
+        else
+          Datadog.logger.debug { "Rate limit hit: #{count}/#{@rate} AppSec traces/second" }
+        end
+      end
+
+      class << self
+        def limit(name, &block)
+          rate_limiter(name).limit(&block)
+        end
+
+        # reset a rate limiter: used for testing
+        def reset!(name)
+          Thread.current[:datadog_security_trace_rate_limiter] = nil
+        end
+
+        protected
+
+        def rate_limiter(name)
+          case name
+          when :traces
+            Thread.current[:datadog_security_trace_rate_limiter] ||= RateLimiter.new(trace_rate_limit)
+          else
+            raise "unsupported rate limiter: #{name.inspect}"
+          end
+        end
+
+        def trace_rate_limit
+          Datadog::AppSec.settings.trace_rate_limit
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/reactive/address_hash.rb

@@ -0,0 +1,18 @@
+# typed: ignore
+
+module Datadog
+  module AppSec
+    module Reactive
+      # Address for Reactive Engine
+      class AddressHash < Hash
+        def addresses
+          keys.flatten
+        end
+
+        def with(address)
+          keys.select { |k| k.include?(address) }
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,44 @@
+# typed: ignore
+
+require_relative 'address_hash'
+require_relative 'subscriber'
+
+module Datadog
+  module AppSec
+    module Reactive
+      # Reactive Engine
+      class Engine
+        def initialize
+          @data = {}
+          @subscribers = AddressHash.new { |h, k| h[k] = [] } # TODO: move to AddressHash initializer
+          @children = []
+        end
+
+        def subscribe(*addresses, &block)
+          @subscribers[addresses.freeze] << Subscriber.new(&block).freeze # TODO: move freeze to Subscriber
+        end
+
+        def publish(address, data)
+          # check if someone has address subscribed
+          if @subscribers.addresses.include?(address)
+
+            # someone will be interested, set data
+            @data[address] = data
+
+            # find candidates i.e address groups that contain the just posted address
+            @subscribers.with(address).each do |addresses|
+              # find targets to the address group containing the posted address
+              subscribers = @subscribers[addresses]
+
+              # is all data for the targets available?
+              if (addresses - @data.keys).empty?
+                hash = addresses.each_with_object({}) { |a, h| h[a] = @data[a] }
+                subscribers.each { |s| s.call(*hash.values) }
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/reactive/operation.rb

@@ -0,0 +1,51 @@
+# typed: ignore
+
+require_relative 'engine'
+
+module Datadog
+  module AppSec
+    module Reactive
+      # Reactive Engine nested operation tracking
+      class Operation
+        attr_reader :reactive,
+          :parent,
+          :name
+
+        def initialize(name, parent = nil, reactive_engine = nil)
+          Datadog.logger.debug { "operation: #{name} initialize" }
+          @name = name
+          @parent = parent
+          @reactive = reactive_engine || (parent.reactive unless parent.nil?) || Reactive::Engine.new
+
+          # TODO: concurrent store
+          # TODO: constant
+          Thread.current[:datadog_security_active_operation] = self
+
+          yield self if block_given?
+        ensure
+          finalize
+        end
+
+        # TODO: use structs instead of an arg splat
+        def subscribe(*addresses, &block)
+          reactive.subscribe(*addresses, &block)
+        end
+
+        def publish(address, data)
+          reactive.publish(address, data)
+        end
+
+        def finalize
+          Datadog.logger.debug { "operation: #{name} finalize" }
+          Thread.current[:datadog_security_active_operation] = parent
+        end
+
+        class << self
+          def active
+            Thread.current[:datadog_security_active_operation]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/reactive/subscriber.rb

@@ -0,0 +1,18 @@
+# typed: ignore
+
+module Datadog
+  module AppSec
+    module Reactive
+      # Reactive Engine subscriber
+      class Subscriber
+        def initialize(&block)
+          @block = block
+        end
+
+        def call(*args)
+          @block.call(*args)
+        end
+      end
+    end
+  end
+end

data/lib/datadog/appsec/response.rb

@@ -0,0 +1,63 @@
+# typed: ignore
+
+require_relative 'assets'
+require_relative 'utils/http/media_range'
+
+module Datadog
+  module AppSec
+    # AppSec response
+    class Response
+      attr_reader :status, :headers, :body
+
+      def initialize(status:, headers: {}, body: [])
+        @status = status
+        @headers = headers
+        @body = body
+      end
+
+      def to_rack
+        [status, headers, body]
+      end
+
+      def to_sinatra_response
+        ::Sinatra::Response.new(body, status, headers)
+      end
+
+      def to_action_dispatch_response
+        ::ActionDispatch::Response.new(status, headers, body)
+      end
+
+      class << self
+        def negotiate(env)
+          Response.new(
+            status: 403,
+            headers: { 'Content-Type' => 'text/html' },
+            body: [Datadog::AppSec::Assets.blocked(format: format(env))]
+          )
+        end
+
+        private
+
+        FORMAT_MAP = {
+          'text/html' => :html,
+          'application/json' => :json,
+          'text/plain' => :text,
+        }.freeze
+
+        DEFAULT_FORMAT = :text
+
+        def format(env)
+          return DEFAULT_FORMAT unless env.key?('HTTP_ACCEPT')
+
+          accepted = env['HTTP_ACCEPT'].split(',').map { |m| Utils::HTTP::MediaRange.new(m) }.sort
+
+          accepted.each_with_object(DEFAULT_FORMAT) do |_default, range|
+            format = FORMAT_MAP.keys.find { |type, _format| range === type }
+
+            return FORMAT_MAP[format] if format
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,201 @@
+# typed: ignore
+
+require_relative 'media_type'
+
+module Datadog
+  module AppSec
+    module Utils
+      module HTTP
+        # Implementation of media range for content negotiation
+        class MediaRange
+          class ParseError < ::StandardError
+          end
+
+          WILDCARD = '*'.freeze
+          WILDCARD_RE = ::Regexp.escape(WILDCARD)
+
+          # See: https://www.rfc-editor.org/rfc/rfc7230#section-3.2.6
+          TOKEN_RE = /[-#$%&'*+.^_`|~A-Za-z0-9]+/.freeze
+
+          # See: https://www.rfc-editor.org/rfc/rfc7231#section-3.1.1.1
+          PARAMETER_RE = %r{ # rubocop:disable Style/RegexpLiteral
+            (?:
+              (?<parameter_name>#{TOKEN_RE})
+              =
+              (?:
+                (?<parameter_value>#{TOKEN_RE})
+                |
+                "(?<parameter_value>[^"]+)"
+              )
+            )
+          }ix.freeze
+
+          # See: https://www.rfc-editor.org/rfc/rfc7231#section-5.3.2
+          ACCEPT_EXT_RE = %r{ # rubocop:disable Style/RegexpLiteral
+            (?:
+              (?<ext_name>#{TOKEN_RE})
+              =
+              (?:
+                (?<ext_value>#{TOKEN_RE})
+                |
+                "(?<ext_value>[^"]+)"
+              )
+            )
+          }ix.freeze
+
+          # See: https://www.rfc-editor.org/rfc/rfc7231#section-5.3.1
+          QVALUE_RE = %r{ # rubocop:disable Style/RegexpLiteral
+            0(?:\.\d{1,3})?
+            |
+            1(?:\.0{1,3})?
+          }ix.freeze
+
+          # See: https://www.rfc-editor.org/rfc/rfc7231#section-5.3.2
+          MEDIA_RANGE_RE = %r{
+            \A
+            (?:
+              (?<type>#{WILDCARD_RE})/(?<subtype>#{WILDCARD_RE})
+              |
+              (?<type>#{TOKEN_RE})/(?<subtype>#{WILDCARD_RE})
+              |
+              (?<type>#{TOKEN_RE})/(?<subtype>#{TOKEN_RE})
+            )
+            (?<parameters>
+              (?:
+                \s*;\s*
+                (?!q=)
+                #{PARAMETER_RE}
+              )*
+            )
+            (?<accept_params>
+              (?<weight>
+                \s*;\s*
+                (?:q=
+                  (?<quality>
+                    #{QVALUE_RE}
+                  )
+                )
+              )
+              (?<accept_exts>
+                (?<accept_ext>
+                  (?:
+                    \s*;\s*
+                    (?!q=)
+                    #{ACCEPT_EXT_RE}
+                  )*
+                )
+              )
+            )?
+            \Z
+          }ix.freeze
+
+          attr_reader :type, :subtype, :quality, :parameters, :accept_ext
+
+          def initialize(media_range) # rubocop:disable Metrics/PerceivedComplexity, Metrics/CyclomaticComplexity
+            media_range_match = MEDIA_RANGE_RE.match(media_range)
+
+            raise ParseError, media_range.inspect if media_range_match.nil?
+
+            @type = (media_range_match['type'] || WILDCARD).downcase
+            @subtype = (media_range_match['subtype'] || WILDCARD).downcase
+            @quality = (media_range_match['quality'] || 1.0).to_f
+            @parameters = {}
+            @accept_ext = {}
+
+            parameters = media_range_match['parameters']
+
+            return if parameters.nil?
+
+            parameters.split(';').map(&:strip).each do |parameter|
+              parameter_match = PARAMETER_RE.match(parameter)
+
+              next if parameter_match.nil?
+
+              parameter_name = parameter_match['parameter_name']
+              parameter_value = parameter_match['parameter_value']
+
+              next if parameter_name.nil? || parameter_value.nil?
+
+              @parameters[parameter_name.downcase] = parameter_value.downcase
+            end
+
+            accept_exts = media_range_match['accept_exts']
+
+            return if accept_exts.nil?
+
+            accept_exts.split(';').map(&:strip).each do |ext|
+              ext_match = ACCEPT_EXT_RE.match(ext)
+
+              next if ext_match.nil?
+
+              ext_name = ext_match['ext_name']
+              ext_value = ext_match['ext_value']
+
+              next if ext_name.nil? || ext_value.nil?
+
+              @accept_ext[ext_name.downcase] = ext_value.downcase
+            end
+          end
+
+          # Compare two MediaRange for ordering
+          def <=>(other)
+            unless (q = quality <=> other.quality) == 0 || q.nil?
+              return q
+            end
+
+            if (s = specificity <=> other.specificity) != 0
+              return s
+            end
+
+            unless wildcard?(:type)
+              if wildcard?(:subtype) && !other.wildcard?(:subtype)
+                return -1
+              elsif !wildcard?(:subtype) && other.wildcard?(:subtype)
+                return 1
+              end
+            end
+
+            if wildcard?(:type) && !other.wildcard?(:type)
+              return -1
+            elsif !wildcard?(:type) && other.wildcard?(:type)
+              return 1
+            end
+
+            0
+          end
+
+          # Compare with a MediaType for match
+          #
+          # returns true if the MediaType is accepted by this MediaRange
+          def ===(other)
+            return self === MediaType.new(other) if other.is_a?(::String)
+
+            type == other.type && subtype == other.subtype && other.parameters.all? { |k, v| parameters[k] == v } ||
+              type == other.type && wildcard?(:subtype) ||
+              wildcard?(:type) && wildcard?(:subtype)
+          end
+
+          def specificity
+            @parameters.count
+          end
+
+          def wildcard?(field = nil)
+            return wildcard?(:type) || wildcard?(:subtype) if field.nil?
+
+            instance_variable_get(:"@#{field}") == WILDCARD
+          end
+
+          def to_s
+            s = "#{@type}/#{@subtype}"
+
+            s << ';' << @parameters.map { |k, v| "#{k}=#{v}" }.join(';') if @parameters.count > 0
+            s << ";q=#{@quality}" if @quality < 1.0
+            s << ';' << @accept_ext.map { |k, v| "#{k}=#{v}" }.join(';') if @accept_ext.count > 0
+
+            s
+          end
+        end
+      end
+    end
+  end
+end