karafka 1.4.0.rc1

data/lib/karafka/routing/proxy.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Karafka
+  module Routing
+    # Proxy is used as a translation layer in between the DSL and raw topic and consumer group
+    # objects.
+    class Proxy
+      attr_reader :target
+
+      # We should proxy only non ? and = methods as we want to have a regular dsl
+      IGNORED_POSTFIXES = %w[
+        ?
+        =
+        !
+      ].freeze
+
+      private_constant :IGNORED_POSTFIXES
+
+      # @param target [Object] target object to which we proxy any DSL call
+      # @param block [Proc] block that we want to evaluate in the proxy context
+      def initialize(target, &block)
+        @target = target
+        instance_eval(&block)
+      end
+
+      # Translates the no "=" DSL of routing into elements assignments on target
+      # @param method_name [Symbol] name of the missing method
+      # @param arguments [Array] array with it's arguments
+      # @param block [Proc] block provided to the method
+      def method_missing(method_name, *arguments, &block)
+        return super unless respond_to_missing?(method_name)
+
+        @target.public_send(:"#{method_name}=", *arguments, &block)
+      end
+
+      # Tells whether or not a given element exists on the target
+      # @param method_name [Symbol] name of the missing method
+      # @param include_private [Boolean] should we include private in the check as well
+      def respond_to_missing?(method_name, include_private = false)
+        return false if IGNORED_POSTFIXES.any? { |postfix| method_name.to_s.end_with?(postfix) }
+
+        @target.respond_to?(:"#{method_name}=", include_private) || super
+      end
+    end
+  end
+end

data/lib/karafka/routing/router.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Karafka
+  # Namespace for all elements related to requests routing
+  module Routing
+    # Karafka framework Router for routing incoming messages to proper consumers
+    # @note Since Kafka does not provide namespaces or modules for topics, they all have "flat"
+    #  structure so all the routes are being stored in a single level array
+    module Router
+      # Find a proper topic based on full topic id
+      # @param topic_id [String] proper topic id (already mapped, etc) for which we want to find
+      #   routing topic
+      # @return [Karafka::Routing::Route] proper route details
+      # @raise [Karafka::Topic::NonMatchingTopicError] raised if topic name does not match
+      #   any route defined by user using routes.draw
+      def find(topic_id)
+        App.consumer_groups.each do |consumer_group|
+          consumer_group.topics.each do |topic|
+            return topic if topic.id == topic_id
+          end
+        end
+
+        raise(Errors::NonMatchingRouteError, topic_id)
+      end
+
+      module_function :find
+    end
+  end
+end

data/lib/karafka/routing/topic.rb

@@ -0,0 +1,62 @@
+# frozen_string_literal: true
+
+module Karafka
+  module Routing
+    # Topic stores all the details on how we should interact with Kafka given topic
+    # It belongs to a consumer group as from 0.6 all the topics can work in the same consumer group
+    # It is a part of Karafka's DSL
+    class Topic
+      extend Helpers::ConfigRetriever
+      extend Forwardable
+
+      attr_reader :id, :consumer_group
+      attr_accessor :consumer
+
+      def_delegator :@consumer_group, :batch_fetching
+
+      # @param [String, Symbol] name of a topic on which we want to listen
+      # @param consumer_group [Karafka::Routing::ConsumerGroup] owning consumer group of this topic
+      def initialize(name, consumer_group)
+        @name = name.to_s
+        @consumer_group = consumer_group
+        @attributes = {}
+        # @note We use identifier related to the consumer group that owns a topic, because from
+        #   Karafka 0.6 we can handle multiple Kafka instances with the same process and we can
+        #   have same topic name across multiple Kafkas
+        @id = "#{consumer_group.id}_#{@name}"
+      end
+
+      # Initializes default values for all the options that support defaults if their values are
+      # not yet specified. This is need to be done (cannot be lazy loaded on first use) because
+      # everywhere except Karafka server command, those would not be initialized on time - for
+      # example for Sidekiq
+      def build
+        Karafka::AttributesMap.topic.each { |attr| send(attr) }
+        self
+      end
+
+      # @return [Class, nil] Class (not an instance) of a responder that should respond from
+      #   consumer back to Kafka (useful for piping data flows)
+      def responder
+        @responder ||= Karafka::Responders::Builder.new(consumer).build
+      end
+
+      Karafka::AttributesMap.topic.each do |attribute|
+        config_retriever_for(attribute)
+      end
+
+      # @return [Hash] hash with all the topic attributes
+      # @note This is being used when we validate the consumer_group and its topics
+      def to_h
+        map = Karafka::AttributesMap.topic.map do |attribute|
+          [attribute, public_send(attribute)]
+        end
+
+        Hash[map].merge!(
+          id: id,
+          consumer: consumer
+        )
+      end
+    end
+  end
+end

data/lib/karafka/routing/topic_mapper.rb

@@ -0,0 +1,53 @@
+# frozen_string_literal: true
+
+module Karafka
+  module Routing
+    # Default topic mapper that does not remap things
+    # Mapper can be used for Kafka providers that require namespaced topic names. Instead of being
+    # provider dependent, we can then define mapper and use internally "pure" topic names in
+    # routes and responders
+    #
+    # @example Mapper for mapping prefixed topics
+    #   class MyMapper
+    #     PREFIX = "my_user_name."
+    #
+    #     def incoming(topic)
+    #       topic.to_s.gsub(PREFIX, '')
+    #     end
+    #
+    #     def outgoing(topic)
+    #       "#{PREFIX}#{topic}"
+    #     end
+    #   end
+    #
+    # @example Mapper for replacing "." with "_" in topic names
+    #   class MyMapper
+    #     PREFIX = "my_user_name."
+    #
+    #     def incoming(topic)
+    #       topic.to_s.gsub('.', '_')
+    #     end
+    #
+    #     def outgoing(topic)
+    #       topic.to_s.gsub('_', '.')
+    #     end
+    #   end
+    class TopicMapper
+      # @param topic [String, Symbol] topic
+      # @return [String, Symbol] same topic as on input
+      # @example
+      #   incoming('topic_name') #=> 'topic_name'
+      def incoming(topic)
+        topic
+      end
+
+      # @param topic [String, Symbol] topic
+      # @return [String, Symbol] same topic as on input
+      # @example
+      #   outgoing('topic_name') #=> 'topic_name'
+      def outgoing(topic)
+        topic
+      end
+    end
+  end
+end

data/lib/karafka/serialization/json/deserializer.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Karafka
+  # Module for all supported by default serialization and deserialization ways
+  module Serialization
+    # Namespace for json ser/der
+    module Json
+      # Default Karafka Json deserializer for loading JSON data
+      class Deserializer
+        # @param params [Karafka::Params::Params] Full params object that we want to deserialize
+        # @return [Hash] hash with deserialized JSON data
+        # @example
+        #   params = {
+        #     'payload' => "{\"a\":1}",
+        #     'topic' => 'my-topic',
+        #     'headers' => { 'message_type' => :test }
+        #   }
+        #   Deserializer.call(params) #=> { 'a' => 1 }
+        def call(params)
+          params.raw_payload.nil? ? nil : ::JSON.parse(params.raw_payload)
+        rescue ::JSON::ParserError => e
+          raise ::Karafka::Errors::DeserializationError, e
+        end
+      end
+    end
+  end
+end

data/lib/karafka/serialization/json/serializer.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Karafka
+  # Module for all supported by default serialization and deserialization ways
+  module Serialization
+    module Json
+      # Default Karafka Json serializer for serializing data
+      class Serializer
+        # @param content [Object] any object that we want to convert to a json string
+        # @return [String] Valid JSON string containing serialized data
+        # @raise [Karafka::Errors::SerializationError] raised when we don't have a way to
+        #   serialize provided data to json
+        # @note When string is passed to this method, we assume that it is already a json
+        #   string and we don't serialize it again. This allows us to serialize data before
+        #   it is being forwarded to this serializer if we want to have a custom (not that simple)
+        #   json serialization
+        #
+        # @example From an ActiveRecord object
+        #   Serializer.call(Repository.first) #=> "{\"repository\":{\"id\":\"04b504e0\"}}"
+        # @example From a string (no changes)
+        #   Serializer.call("{\"a\":1}") #=> "{\"a\":1}"
+        def call(content)
+          return content if content.is_a?(String)
+          return content.to_json if content.respond_to?(:to_json)
+
+          raise Karafka::Errors::SerializationError, content
+        end
+      end
+    end
+  end
+end

data/lib/karafka/server.rb

@@ -0,0 +1,86 @@
+# frozen_string_literal: true
+
+module Karafka
+  # Karafka consuming server class
+  class Server
+    @consumer_threads = Concurrent::Array.new
+
+    # How long should we sleep between checks on shutting down consumers
+    SUPERVISION_SLEEP = 0.1
+    # What system exit code should we use when we terminated forcefully
+    FORCEFUL_EXIT_CODE = 2
+    # This factor allows us to calculate how many times we have to sleep before
+    # a forceful shutdown
+    SUPERVISION_CHECK_FACTOR = (1 / SUPERVISION_SLEEP)
+
+    private_constant :SUPERVISION_SLEEP, :FORCEFUL_EXIT_CODE, :SUPERVISION_CHECK_FACTOR
+
+    class << self
+      # Set of consuming threads. Each consumer thread contains a single consumer
+      attr_accessor :consumer_threads
+
+      # Writer for list of consumer groups that we want to consume in our current process context
+      attr_writer :consumer_groups
+
+      # Method which runs app
+      def run
+        process.on_sigint { stop_supervised }
+        process.on_sigquit { stop_supervised }
+        process.on_sigterm { stop_supervised }
+        run_supervised
+      end
+
+      # @return [Array<String>] array with names of consumer groups that should be consumed in a
+      #   current server context
+      def consumer_groups
+        # If not specified, a server will listed on all the topics
+        @consumer_groups ||= Karafka::App.consumer_groups.map(&:name).freeze
+      end
+
+      private
+
+      # @return [Karafka::Process] process wrapper instance used to catch system signal calls
+      def process
+        Karafka::App.config.internal.process
+      end
+
+      # Starts Karafka with a supervision
+      # @note We don't need to sleep because Karafka::Fetcher is locking and waiting to
+      # finish loop (and it won't happen until we explicitly want to stop)
+      def run_supervised
+        process.supervise
+        Karafka::App.run!
+        Karafka::App.config.internal.fetcher.call
+      end
+
+      # Stops Karafka with a supervision (as long as there is a shutdown timeout)
+      # If consumers won't stop in a given time frame, it will force them to exit
+      def stop_supervised
+        Karafka::App.stop!
+
+        # Temporary patch until https://github.com/dry-rb/dry-configurable/issues/93 is fixed
+        timeout = Thread.new { Karafka::App.config.shutdown_timeout }.join.value
+
+        # We check from time to time (for the timeout period) if all the threads finished
+        # their work and if so, we can just return and normal shutdown process will take place
+        (timeout * SUPERVISION_CHECK_FACTOR).to_i.times do
+          if consumer_threads.count(&:alive?).zero?
+            Thread.new { Karafka.monitor.instrument('app.stopped') }.join
+            return
+          end
+
+          sleep SUPERVISION_SLEEP
+        end
+
+        raise Errors::ForcefulShutdownError
+      rescue Errors::ForcefulShutdownError => e
+        Thread.new { Karafka.monitor.instrument('app.stopping.error', error: e) }.join
+        # We're done waiting, lets kill them!
+        consumer_threads.each(&:terminate)
+
+        # exit! is not within the instrumentation as it would not trigger due to exit
+        Kernel.exit! FORCEFUL_EXIT_CODE
+      end
+    end
+  end
+end

data/lib/karafka/setup/config.rb

@@ -0,0 +1,223 @@
+# frozen_string_literal: true
+
+module Karafka
+  # Module containing all Karafka setup related elements like configuration settings,
+  # config validations and configurators for external gems integration
+  module Setup
+    # Configurator for setting up all the framework details that are required to make it work
+    # @note If you want to do some configurations after all of this is done, please add to
+    #   karafka/config a proper file (needs to inherit from Karafka::Setup::Configurators::Base
+    #   and implement setup method) after that everything will happen automatically
+    # @note This config object allows to create a 1 level nesting (nodes) only. This should be
+    #   enough and will still keep the code simple
+    # @see Karafka::Setup::Configurators::Base for more details about configurators api
+    class Config
+      extend Dry::Configurable
+
+      # Contract for checking the config provided by the user
+      CONTRACT = Karafka::Contracts::Config.new.freeze
+
+      private_constant :CONTRACT
+
+      # Available settings
+      # option client_id [String] kafka client_id - used to provide
+      #   default Kafka groups namespaces and identify that app in kafka
+      setting :client_id
+      # What backend do we want to use to process messages
+      setting :backend, :inline
+      # option logger [Instance] logger that we want to use
+      setting :logger, ::Karafka::Instrumentation::Logger.new
+      # option monitor [Instance] monitor that we will to use (defaults to Karafka::Monitor)
+      setting :monitor, ::Karafka::Instrumentation::Monitor.new
+      # Mapper used to remap consumer groups ids, so in case users migrate from other tools
+      # or they need to maintain their own internal consumer group naming conventions, they
+      # can easily do it, replacing the default client_id + consumer name pattern concept
+      setting :consumer_mapper, Routing::ConsumerMapper.new
+      # Mapper used to remap names of topics, so we can have a clean internal topic naming
+      # despite using any Kafka provider that uses namespacing, etc
+      # It needs to implement two methods:
+      #   - #incoming - for remapping from the incoming message to our internal format
+      #   - #outgoing - for remapping from internal topic name into outgoing message
+      setting :topic_mapper, Routing::TopicMapper.new
+      # Default serializer for converting whatever we want to send to kafka to json
+      setting :serializer, Karafka::Serialization::Json::Serializer.new
+      # Default deserializer for converting incoming data into ruby objects
+      setting :deserializer, Karafka::Serialization::Json::Deserializer.new
+      # If batch_fetching is true, we will fetch kafka messages in batches instead of 1 by 1
+      # @note Fetching does not equal consuming, see batch_consuming description for details
+      setting :batch_fetching, true
+      # If batch_consuming is true, we will have access to #params_batch instead of #params.
+      # #params_batch will contain params received from Kafka (may be more than 1) so we can
+      # process them in batches
+      setting :batch_consuming, false
+      # option shutdown_timeout [Integer, nil] the number of seconds after which Karafka no
+      #   longer wait for the consumers to stop gracefully but instead we force terminate
+      #   everything.
+      setting :shutdown_timeout, 60
+
+      # option kafka [Hash] - optional - kafka configuration options
+      setting :kafka do
+        # Array with at least one host
+        setting :seed_brokers, %w[kafka://127.0.0.1:9092]
+        # option session_timeout [Integer] the number of seconds after which, if a client
+        #   hasn't contacted the Kafka cluster, it will be kicked out of the group.
+        setting :session_timeout, 30
+        # Time that a given partition will be paused from fetching messages, when message
+        # consumption fails. It allows us to process other partitions, while the error is being
+        # resolved and also "slows" things down, so it prevents from "eating" up all messages and
+        # consuming them with failed code. Use `nil` if you want to pause forever and never retry.
+        setting :pause_timeout, 10
+        # option pause_max_timeout [Integer, nil] the maximum number of seconds to pause for,
+        #   or `nil` if no maximum should be enforced.
+        setting :pause_max_timeout, nil
+        # option pause_exponential_backoff [Boolean] whether to enable exponential backoff
+        setting :pause_exponential_backoff, false
+        # option offset_commit_interval [Integer] the interval between offset commits,
+        #   in seconds.
+        setting :offset_commit_interval, 10
+        # option offset_commit_threshold [Integer] the number of messages that can be
+        #   processed before their offsets are committed. If zero, offset commits are
+        #   not triggered by message consumption.
+        setting :offset_commit_threshold, 0
+        # option heartbeat_interval [Integer] the interval between heartbeats; must be less
+        #   than the session window.
+        setting :heartbeat_interval, 10
+        # option offset_retention_time [Integer] The length of the retention window, known as
+        #   offset retention time
+        setting :offset_retention_time, nil
+        # option fetcher_max_queue_size [Integer] max number of items in the fetch queue that
+        #   are stored for further processing. Note, that each item in the queue represents a
+        #   response from a single broker
+        setting :fetcher_max_queue_size, 10
+        # option max_bytes_per_partition [Integer] the maximum amount of data fetched
+        #   from a single partition at a time.
+        setting :max_bytes_per_partition, 1_048_576
+        #  whether to consume messages starting at the beginning or to just consume new messages
+        setting :start_from_beginning, true
+        # option min_bytes [Integer] the minimum number of bytes to read before
+        #   returning messages from the server; if `max_wait_time` is reached, this
+        #   is ignored.
+        setting :min_bytes, 1
+        # option max_bytes [Integer] the maximum number of bytes to read before returning messages
+        #   from each broker.
+        setting :max_bytes, 10_485_760
+        # option max_wait_time [Integer, Float] max_wait_time is the maximum number of seconds to
+        #   wait before returning data from a single message fetch. By setting this high you also
+        #   increase the fetching throughput - and by setting it low you set a bound on latency.
+        #   This configuration overrides `min_bytes`, so you'll _always_ get data back within the
+        #   time specified. The default value is one second. If you want to have at most five
+        #   seconds of latency, set `max_wait_time` to 5. You should make sure
+        #   max_wait_time * num brokers + heartbeat_interval is less than session_timeout.
+        setting :max_wait_time, 1
+        # option automatically_mark_as_consumed [Boolean] should we automatically mark received
+        # messages as consumed (processed) after non-error consumption
+        setting :automatically_mark_as_consumed, true
+        # option reconnect_timeout [Integer] How long should we wait before trying to reconnect to
+        # Kafka cluster that went down (in seconds)
+        setting :reconnect_timeout, 5
+        # option connect_timeout [Integer] Sets the number of seconds to wait while connecting to
+        # a broker for the first time. When ruby-kafka initializes, it needs to connect to at
+        # least one host.
+        setting :connect_timeout, 10
+        # option socket_timeout [Integer] Sets the number of seconds to wait when reading from or
+        # writing to a socket connection to a broker. After this timeout expires the connection
+        # will be killed. Note that some Kafka operations are by definition long-running, such as
+        # waiting for new messages to arrive in a partition, so don't set this value too low
+        setting :socket_timeout, 30
+        # option partitioner [Object, nil] the partitioner that should be used by the client
+        setting :partitioner, nil
+
+        # SSL authentication related settings
+        # option ca_cert [String, nil] SSL CA certificate
+        setting :ssl_ca_cert, nil
+        # option ssl_ca_cert_file_path [String, nil] SSL CA certificate file path
+        setting :ssl_ca_cert_file_path, nil
+        # option ssl_ca_certs_from_system [Boolean] Use the CA certs from your system's default
+        #   certificate store
+        setting :ssl_ca_certs_from_system, false
+        # option ssl_verify_hostname [Boolean] Verify the hostname for client certs
+        setting :ssl_verify_hostname, true
+        # option ssl_client_cert [String, nil] SSL client certificate
+        setting :ssl_client_cert, nil
+        # option ssl_client_cert_key [String, nil] SSL client certificate password
+        setting :ssl_client_cert_key, nil
+        # option sasl_gssapi_principal [String, nil] sasl principal
+        setting :sasl_gssapi_principal, nil
+        # option sasl_gssapi_keytab [String, nil] sasl keytab
+        setting :sasl_gssapi_keytab, nil
+        # option sasl_plain_authzid [String] The authorization identity to use
+        setting :sasl_plain_authzid, ''
+        # option sasl_plain_username [String, nil] The username used to authenticate
+        setting :sasl_plain_username, nil
+        # option sasl_plain_password [String, nil] The password used to authenticate
+        setting :sasl_plain_password, nil
+        # option sasl_scram_username [String, nil] The username used to authenticate
+        setting :sasl_scram_username, nil
+        # option sasl_scram_password [String, nil] The password used to authenticate
+        setting :sasl_scram_password, nil
+        # option sasl_scram_mechanism [String, nil] Scram mechanism, either 'sha256' or 'sha512'
+        setting :sasl_scram_mechanism, nil
+        # option sasl_over_ssl [Boolean] whether to enforce SSL with SASL
+        setting :sasl_over_ssl, true
+        # option ssl_client_cert_chain [String, nil] client cert chain or nil if not used
+        setting :ssl_client_cert_chain, nil
+        # option ssl_client_cert_key_password [String, nil] the password required to read
+        #   the ssl_client_cert_key
+        setting :ssl_client_cert_key_password, nil
+        # @param sasl_oauth_token_provider [Object, nil] OAuthBearer Token Provider instance that
+        #   implements method token.
+        setting :sasl_oauth_token_provider, nil
+      end
+
+      # Namespace for internal settings that should not be modified
+      # It's a temporary step to "declassify" several things internally before we move to a
+      # non global state
+      setting :internal do
+        # option routing_builder [Karafka::Routing::Builder] builder instance
+        setting :routing_builder, Routing::Builder.new
+        # option status [Karafka::Status] app status
+        setting :status, Status.new
+        # option process [Karafka::Process] process status
+        # @note In the future, we need to have a single process representation for all the karafka
+        #   instances
+        setting :process, Process.new
+        # option fetcher [Karafka::Fetcher] fetcher instance
+        setting :fetcher, Fetcher.new
+        # option configurators [Array<Object>] all configurators that we want to run after
+        #   the setup
+        setting :configurators, [Configurators::WaterDrop.new]
+      end
+
+      class << self
+        # Configuring method
+        # @yield Runs a block of code providing a config singleton instance to it
+        # @yieldparam [Karafka::Setup::Config] Karafka config instance
+        def setup
+          configure { |config| yield(config) }
+        end
+
+        # Everything that should be initialized after the setup
+        # Components are in karafka/config directory and are all loaded one by one
+        # If you want to configure a next component, please add a proper file to config dir
+        def setup_components
+          config
+            .internal
+            .configurators
+            .each { |configurator| configurator.call(config) }
+        end
+
+        # Validate config based on the config contract
+        # @return [Boolean] true if configuration is valid
+        # @raise [Karafka::Errors::InvalidConfigurationError] raised when configuration
+        #   doesn't match with the config contract
+        def validate!
+          validation_result = CONTRACT.call(config.to_h)
+
+          return true if validation_result.success?
+
+          raise Errors::InvalidConfigurationError, validation_result.errors.to_h
+        end
+      end
+    end
+  end
+end