Program Listing for File kafka_source.hpp

            

            
#pragma once

#include "morpheus/messages/meta.hpp"

#include <cudf/io/types.hpp>
#include <librdkafka/rdkafkacpp.h>
#include <mrc/channel/status.hpp>// for Status
#include <mrc/node/source_properties.hpp>// for SourceProperties<>::source_type_t
#include <mrc/segment/builder.hpp>
#include <mrc/segment/object.hpp>// for Object
#include <pymrc/node.hpp>
#include <rxcpp/rx.hpp>// for apply, make_subscriber, observable_member, is_on_error<>::not_void, is_on_next_of<>::not_void, trace_activity

#include <cstddef>// for size_t
#include <cstdint>// for int32_t, uint32_t
#include <map>
#include <memory>
#include <string>
#include <vector>

namespace morpheus {

/****** Component public implementations *******************/
/****** KafkaSourceStage************************************/

#pragma GCC visibility push(default)
class KafkaSourceStage : public mrc::pymrc::PythonSource<std::shared_ptr<MessageMeta>>
{
  public:
    using base_t = mrc::pymrc::PythonSource<std::shared_ptr<MessageMeta>>;
    using typename base_t::source_type_t;
    using typename base_t::subscriber_fn_t;

    KafkaSourceStage(size_t max_batch_size,
                     std::string topic,
                     int32_t batch_timeout_ms,
                     std::map<std::string, std::string> config,
                     bool disable_commit        = false,
                     bool disable_pre_filtering = false,
                     size_t stop_after          = 0,
                     bool async_commits         = true);

    ~KafkaSourceStage() override = default;

    std::size_t max_batch_size();

    int32_t batch_timeout_ms();

  private:
    subscriber_fn_t build();

    std::unique_ptr<RdKafka::Conf> build_kafka_conf(const std::map<std::string, std::string> &config_in);

    std::unique_ptr<RdKafka::KafkaConsumer> create_consumer(RdKafka::RebalanceCb &rebalancer);

    cudf::io::table_with_metadata load_table(const std::string &buffer);

    std::shared_ptr<morpheus::MessageMeta> process_batch(
        std::vector<std::unique_ptr<RdKafka::Message>> &&message_batch);

    size_t m_max_batch_size{128};
    uint32_t m_batch_timeout_ms{100};

    std::string m_topic;
    std::map<std::string, std::string> m_config;

    bool m_disable_commit{false};
    bool m_disable_pre_filtering{false};
    bool m_requires_commit{false};  // Whether or not manual committing is required
    bool m_async_commits{true};
    size_t m_stop_after{0};

    void *m_rebalancer;
};

/****** KafkaSourceStageInferenceProxy**********************/
struct KafkaSourceStageInterfaceProxy
{
    static std::shared_ptr<mrc::segment::Object<KafkaSourceStage>> init(mrc::segment::Builder& builder,
                                                                        const std::string& name,
                                                                        size_t max_batch_size,
                                                                        std::string topic,
                                                                        int32_t batch_timeout_ms,
                                                                        std::map<std::string, std::string> config,
                                                                        bool disable_commit,
                                                                        bool disable_pre_filtering,
                                                                        size_t stop_after  = 0,
                                                                        bool async_commits = true);
};
#pragma GCC visibility pop// end of group
}  // namespace morpheus