#include "tensorflow/core/framework/op.h"
#include "tensorflow/core/framework/op_def.pb.h"
#include "tensorflow/core/grappler/costs/graph_properties.h"
#include "tensorflow/core/grappler/costs/op_performance_data.pb.h"
#include "tensorflow/core/grappler/grappler_item.h"
#include "tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.h"
#include "tensorflow/core/platform/logging.h"
#include "tensorflow/core/platform/status.h"

  return absl::OkStatus();
}

Status SetAttrString(AbstractOperation* op_, const char* attr_name,
                     const char* data, size_t length,
                     ForwardOperation* forward_op_) {
  forward_op_->attrs.Set(attr_name, StringPiece(data, length));
  return op_->SetAttrString(attr_name, data, length);
}
Status SetAttrInt(AbstractOperation* op_, const char* attr_name, int64_t value,

#   variables in between script executions. To forcibly remove the
#   container and start fresh, run "docker rm -f tf". Removing the container
#   destroys some temporary bazel data and causes longer builds.
#
#   You will need the NVIDIA Container Toolkit for GPU testing:
#   https://github.com/NVIDIA/nvidia-container-toolkit
#
#   Note: if you interrupt a bazel command on docker (ctrl-c), you

    unar "libstdc++6_4.8.1-10ubuntu8_arm64.deb" && \
    tar -C "${TARGET}" -xvzf "libstdc++6_4.8.1-10ubuntu8_arm64/data.tar.gz" "./usr/lib/aarch64-linux-gnu/libstdc++.so.6.0.18"  && \
    rm -rf "libstdc++6_4.8.1-10ubuntu8_arm64.deb" "libstdc++6_4.8.1-10ubuntu8_arm64"

mkdir -p "${TARGET}-src"
cd "${TARGET}-src"

    ] + if_not_mobile([
        "//tensorflow/core/kernels:tensor_list",
        "//tensorflow/core/kernels:tensor_list_util",
        "//tensorflow/core/kernels:variant_ops_util",
        "//tensorflow/core/kernels/data:optional_ops_util",
        "//tensorflow/core/platform:abi",
    ]),
    alwayslink = 1,
)

tf_cuda_library(
    name = "ops",
    srcs = [
        "ops.cc",
    ],
    hdrs = [

// optimization pass. Nodes modified during optimization pass have to be
// invalidated, to prevent further incorrect optimizations based on wrong shape
// and data type properties.
typedef struct TF_GraphProperties TF_GraphProperties;

// Create GraphProperties. The item must outlive the properties.
TF_CAPI_EXPORT extern TF_GraphProperties* TF_NewGraphProperties(

  tensorflow::string value = cell->cell.value();
  void* data = tensorflow::port::Malloc(value.length());
  value.copy(static_cast<char*>(data), value.length(), 0);
  buf->data = data;
  buf->length = value.length();
  buf->data_deallocator = [](void* data, size_t length) {
    tensorflow::port::Free(data);
  };
}

TFE_MonitoringStringGauge0* TFE_MonitoringNewStringGauge0(

  tstring* data = static_cast<tstring*>(TF_TensorData(t));
  *data = value;
  TFE_TensorHandle* th = TFE_NewTensorHandleFromTensor(ctx, t, status);
  CHECK_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  TF_DeleteTensor(t);
  TF_DeleteStatus(status);
  return th;
}

TFE_TensorHandle* TestScalarTensorHandle(TFE_Context* ctx, int value) {
  int data[] = {value};
  TF_Status* status = TF_NewStatus();

when dealing with untrusted data. Based on the security history of these
libraries we consider that it is safe to work with untrusted inputs for PNG,
BMP, GIF, WAV, RAW, RAW\_PADDED, CSV and PROTO formats. All other input formats,
including tensorflow-io should be sandboxed if used to process untrusted data.

For example, if an attacker were to upload a malicious video file, they could

                               tensorflow::DataType dtype, void* data,
                               TFE_CustomDeviceTensorHandleMethods methods)
      : tensorflow::CustomDeviceTensorHandle(context, device, dtype),
        data_(data),
        methods_(methods) {}

  ~CAPICustomDeviceTensorHandle() override { methods_.deallocator(data_); }
  void* DevicePointer() const override { return data_; }
  Status NumDims(int* num_dims) const override {