#include "tensorflow/c/tf_buffer_internal.h"
#include "tensorflow/core/framework/attr_value_util.h"
#include "tensorflow/core/framework/function.h"
#include "tensorflow/core/framework/function.pb.h"
#include "tensorflow/core/framework/graph_to_functiondef.h"
#include "tensorflow/core/framework/node_def.pb.h"
#include "tensorflow/core/framework/node_def_util.h"
#include "tensorflow/core/framework/tensor.pb.h"  // NOLINT

#include "xla/pjrt/pjrt_c_api_client.h"
#include "xla/pjrt/tfrt_cpu_pjrt_client.h"
#include "xla/shape.h"
#include "xla/shape_util.h"
#include "tensorflow/core/framework/types.h"
#include "tensorflow/core/tfrt/common/async_value_tensor.h"
#include "tensorflow/core/tfrt/common/pjrt_util.h"
#include "tsl/lib/core/status_test_util.h"
#include "tsl/platform/casts.h"
#include "tsl/platform/status_matchers.h"
#include "tsl/protobuf/error_codes.pb.h"

#include "tensorflow/c/eager/tfe_tensorhandle_internal.h"
#include "tensorflow/c/tf_status.h"
#include "tensorflow/c/tf_status_internal.h"
#include "tensorflow/core/lib/gtl/cleanup.h"
#include "tensorflow/core/platform/env.h"
#include "tensorflow/core/platform/mutex.h"
#include "tensorflow/core/util/device_name_utils.h"

namespace tensorflow {
namespace parallel_device {
namespace {

class OpDeleter {
 public:

#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/tf_status.h"
#include "tensorflow/core/lib/gtl/cleanup.h"
#include "tensorflow/core/platform/test.h"

namespace {

struct LoggingDevice {
  tensorflow::string device_name;
  tensorflow::string underlying_device;
  // Set to true whenever a TensorHandle is copied onto the device

#include "xla/pjrt/pjrt_client.h"
#include "tensorflow/core/common_runtime/next_pluggable_device/plugin_resource.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/resource_handle.h"
#include "tensorflow/core/framework/resource_mgr.h"
#include "tensorflow/core/framework/tensor.h"
#include "tensorflow/core/framework/types.h"
#include "tensorflow/core/platform/refcount.h"
#include "tensorflow/core/platform/status.h"

      std::move(maybe_parallel_results.value()));
  std::vector<MaybeParallelTensorOwned> result_content;
  result_content.reserve(parallel_results.size());
  for (std::unique_ptr<ParallelTensor>& parallel_result : parallel_results) {
    result_content.push_back(
        MaybeParallelTensorOwned(std::move(parallel_result)));
  }
  result.emplace(std::move(result_content));
  return result;
}

#include "tensorflow/core/framework/function.h"
#include "tensorflow/core/grappler/clusters/single_machine.h"
#include "tensorflow/core/grappler/costs/graph_properties.h"
#include "tensorflow/core/grappler/grappler_item.h"
#include "tensorflow/core/grappler/inputs/trivial_test_graph_input_yielder.h"
#include "tensorflow/core/lib/core/status_test_util.h"
#include "tensorflow/core/platform/test.h"

namespace tensorflow {

#include "tensorflow/c/experimental/ops/math_ops.h"
#include "tensorflow/c/tf_status_helper.h"
#include "tensorflow/c/tf_tensor.h"
#include "tensorflow/core/lib/llvm_rtti/llvm_rtti.h"
#include "tensorflow/core/platform/errors.h"
#include "tensorflow/core/platform/test.h"

namespace tensorflow {
namespace gradients {
namespace internal {
namespace {
using std::vector;
using tensorflow::TF_StatusPtr;

#include <random>

#include "absl/strings/string_view.h"
#include "tensorflow/c/tf_status_helper.h"
#include "tensorflow/core/platform/path.h"
#include "tensorflow/core/platform/stacktrace_handler.h"
#include "tensorflow/core/platform/test.h"

#define ASSERT_TF_OK(x) ASSERT_EQ(TF_OK, TF_GetCode(x)) << TF_Message(x)
#define EXPECT_TF_OK(x) EXPECT_EQ(TF_OK, TF_GetCode(x)) << TF_Message(x)

#include "tensorflow/c/eager/parallel_device/parallel_device_testlib.h"
#include "tensorflow/c/eager/tfe_tensorhandle_internal.h"
#include "tensorflow/c/tf_status_internal.h"
#include "tensorflow/core/lib/core/status_test_util.h"
#include "tensorflow/core/platform/test.h"

// NOTE(allenl): These tests currently go through TFE_Execute and so are
// integration testing rather than purely testing the parallel device. They