// A simple logging device to test custom device registration.
#include <memory>

#include "absl/strings/match.h"
#include "tensorflow/c/c_api.h"
#include "tensorflow/c/eager/c_api.h"
#include "tensorflow/c/eager/c_api_experimental.h"
#include "tensorflow/c/eager/c_api_test_util.h"
#include "tensorflow/c/eager/custom_device_testutil.h"
#include "tensorflow/c/tf_status.h"

limitations under the License.
==============================================================================*/

#ifndef TENSORFLOW_C_EAGER_CUSTOM_DEVICE_TESTUTIL_H_
#define TENSORFLOW_C_EAGER_CUSTOM_DEVICE_TESTUTIL_H_

// A simple logging device to test custom device registration.
#include <memory>

#include "tensorflow/c/c_api.h"
#include "tensorflow/c/eager/c_api.h"
#include "tensorflow/c/eager/c_api_experimental.h"

void AllocateParallelDevice(const char* device_name,
                            const char* const* underlying_devices,
                            int num_underlying_devices,
                            TFE_CustomDevice* device, void** device_info);

}  // namespace parallel_device
}  // namespace tensorflow

  underlying_devices_vector.reserve(num_underlying_devices);
  for (int device_index = 0; device_index < num_underlying_devices;
       ++device_index) {
    underlying_devices_vector.push_back(underlying_devices[device_index]);
  }
  std::unique_ptr<ParallelDevice> parallel_device(
      new ParallelDevice(underlying_devices_vector));
  *device_info =

  const char* first_device = "/job:worker/replica:0/task:1/device:CPU:0";
  const char* second_device = "/job:worker/replica:0/task:2/device:CPU:0";
  const char* device_name = "/job:localhost/replica:0/task:0/device:CUSTOM:0";
  std::array<const char*, 2> underlying_devices{first_device, second_device};
  RegisterParallelDevice(context.get(), device_name, underlying_devices,
                         status.get());

  std::vector<std::string> devices{
      "/job:localhost/replica:0/task:0/device:CPU:0",
      "/job:localhost/replica:0/task:0/device:CPU:1"};
  ParallelDevice parallel_device(std::move(devices));
  {
    std::unique_ptr<ParallelTensor> float_tensors =
        parallel_device.ScalarsFromSequence<float>({10.0, 11.0}, context.get(),

}

template <std::size_t num_devices>
void RegisterParallelDevice(
    TFE_Context* context, const char* device_name,
    const std::array<const char*, num_devices>& underlying_devices,
    TF_Status* status) {
  TFE_CustomDevice device;
  void* device_info;
  tensorflow::parallel_device::AllocateParallelDevice(
      device_name, underlying_devices.data(), underlying_devices.size(),
      &device, &device_info);

  std::unique_ptr<ParallelTensor> DeviceIDs(TFE_Context* context,
                                            TF_Status* status) const;

  // The number of devices operations run on.
  size_t num_underlying_devices() const { return underlying_devices_.size(); }

  // The devices operations run on.
  const std::vector<std::string>& underlying_devices() const {
    return underlying_devices_;

  const char* device_name = "/job:localhost/replica:0/task:0/device:CUSTOM:0";
  const char* first_device_name =
      "/job:localhost/replica:0/task:0/device:CPU:0";
  const char* second_device_name =
      "/job:localhost/replica:0/task:0/device:CPU:1";
  std::array<const char*, 2> underlying_devices{first_device_name,
                                                second_device_name};

ParallelDevice::ParallelDevice(const std::vector<std::string>& devices,
                               bool is_async, int in_flight_nodes_limit)
    : underlying_devices_(devices),
      default_cancellation_manager_(absl::make_unique<CancellationManager>()) {
  device_threads_.reserve(devices.size());
  for (int device_index = 0; device_index < devices.size(); ++device_index) {
    device_threads_.emplace_back(new DeviceThread(