extensions/tensorflow/TFApplyGraph.h

/**
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#pragma once

#include <atomic>

#include <core/Resource.h>
#include <core/Processor.h>
#include <tensorflow/core/public/session.h>
#include <concurrentqueue.h>
#include "io/InputStream.h"
#include "io/OutputStream.h"

namespace org::apache::nifi::minifi::processors {

class TFApplyGraph : public core::Processor {
 public:
  explicit TFApplyGraph(const std::string &name, const utils::Identifier &uuid = {})
      : Processor(name, uuid),
        logger_(logging::LoggerFactory<TFApplyGraph>::getLogger()) {
  }

  EXTENSIONAPI static constexpr const char* Description = "Applies a TensorFlow graph to the tensor protobuf supplied as input. The tensor is fed into the node specified by the Input Node property. "
    "The output FlowFile is a tensor protobuf extracted from the node specified by the Output Node property. TensorFlow graphs are read dynamically by feeding a graph "
    "protobuf to the processor with the tf.type property set to graph.";

  EXTENSIONAPI static const core::Property InputNode;
  EXTENSIONAPI static const core::Property OutputNode;
  static auto properties() {
    return std::array{
      InputNode,
      OutputNode
    };
  }

  EXTENSIONAPI static const core::Relationship Success;
  EXTENSIONAPI static const core::Relationship Retry;
  EXTENSIONAPI static const core::Relationship Failure;
  static auto relationships() {
    return std::array{
      Success,
      Retry,
      Failure
    };
  }

  EXTENSIONAPI static constexpr bool SupportsDynamicProperties = false;
  EXTENSIONAPI static constexpr bool SupportsDynamicRelationships = false;
  EXTENSIONAPI static constexpr core::annotation::Input InputRequirement = core::annotation::Input::INPUT_ALLOWED;
  EXTENSIONAPI static constexpr bool IsSingleThreaded = false;

  ADD_COMMON_VIRTUAL_FUNCTIONS_FOR_PROCESSORS

  void initialize() override;
  void onSchedule(core::ProcessContext *context, core::ProcessSessionFactory *sessionFactory) override;
  void onTrigger(core::ProcessContext* /*context*/, core::ProcessSession* /*session*/) override {
    logger_->log_error("onTrigger invocation with raw pointers is not implemented");
  }
  void onTrigger(const std::shared_ptr<core::ProcessContext> &context,
                 const std::shared_ptr<core::ProcessSession> &session) override;

  struct TFContext {
    std::shared_ptr<tensorflow::Session> tf_session;
    uint32_t graph_version;
  };

  class GraphReadCallback : public InputStreamCallback {
   public:
    explicit GraphReadCallback(std::shared_ptr<tensorflow::GraphDef> graph_def)
        : graph_def_(std::move(graph_def)) {
    }
    ~GraphReadCallback() override = default;
    int64_t process(const std::shared_ptr<io::InputStream>& stream) override;

   private:
    std::shared_ptr<tensorflow::GraphDef> graph_def_;
  };

  class TensorReadCallback : public InputStreamCallback {
   public:
    explicit TensorReadCallback(std::shared_ptr<tensorflow::TensorProto> tensor_proto)
        : tensor_proto_(std::move(tensor_proto)) {
    }
    ~TensorReadCallback() override = default;
    int64_t process(const std::shared_ptr<io::InputStream>& stream) override;

   private:
    std::shared_ptr<tensorflow::TensorProto> tensor_proto_;
  };

  class TensorWriteCallback : public OutputStreamCallback {
   public:
    explicit TensorWriteCallback(std::shared_ptr<tensorflow::TensorProto> tensor_proto)
        : tensor_proto_(std::move(tensor_proto)) {
    }
    ~TensorWriteCallback() override = default;
    int64_t process(const std::shared_ptr<io::OutputStream>& stream) override;

   private:
    std::shared_ptr<tensorflow::TensorProto> tensor_proto_;
  };

 private:
  std::shared_ptr<logging::Logger> logger_;
  std::string input_node_;
  std::string output_node_;
  std::shared_ptr<tensorflow::GraphDef> graph_def_;
  std::mutex graph_def_mtx_;
  uint32_t graph_version_ = 0;
  moodycamel::ConcurrentQueue<std::shared_ptr<TFContext>> tf_context_q_;
};

}  // namespace org::apache::nifi::minifi::processors