return ParseTextProto(str, kTensorShapePrefix, proto);
}

string MangleTensor(const TensorProto& tensor) {
  return absl::StrCat(kTensorPrefix, PrintShortTextProto(tensor));
}

Status DemangleTensor(absl::string_view str, TensorProto* proto) {
  return ParseTextProto(str, kTensorPrefix, proto);
}

string MangleDataType(const DataType& dtype) {

}

void TF_SetShape(TF_Tensor* t, const int64_t* dims, int num_dims) {
  tensorflow::down_cast<tensorflow::TensorInterface*>(t->tensor)->SetShape(
      dims, num_dims);
}

int TF_NumDims(const TF_Tensor* t) { return t->tensor->NumDims(); }

int64_t TF_Dim(const TF_Tensor* t, int dim_index) {
  return t->tensor->Dim(dim_index);
}

size_t TF_TensorByteSize(const TF_Tensor* t) { return t->tensor->ByteSize(); }

// Return the type of a tensor element.
TF_CAPI_EXPORT extern TF_DataType TF_TensorType(const TF_Tensor*);

// Set a new shape for the Tensor.
TF_CAPI_EXPORT extern void TF_SetShape(TF_Tensor* tensor, const int64_t* dims,
                                       int num_dims);

// Return the number of dimensions that the tensor has.
TF_CAPI_EXPORT extern int TF_NumDims(const TF_Tensor*);

}

PJRT_Buffer* TF_GetPjRtCBuffer(TF_Tensor* c_tensor, TF_Status* status) {
  tensorflow::Tensor tensor;
  auto s = tensorflow::TF_TensorToTensor(c_tensor, &tensor);
  if (!s.ok()) {
    status->status = s;
    return nullptr;
  }
  absl::StatusOr<PJRT_Buffer*> c_buffer =
      tensorflow::GetPjRtCBufferFromTensor(&tensor);
  if (!c_buffer.ok()) {
    status->status = c_buffer.status();

  %0 = "tf.Less"(%index, %size) {device = "/device:CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i1>
  func.return %0 : tensor<i1>
}

    TF_BoolTensor:$condition,
    Arg<TF_Tensor, [{= A `Tensor` which may have the same shape as `condition`.
If `condition` is rank 1, `x` may have higher rank,
but its first dimension must match the size of `condition`.}]>:$then_value,
    Arg<TF_Tensor, [{= A `Tensor` with the same type and shape as `x`.}]>:$else_value
  );

  let results = (outs

  }
  absl::Status s;
  tensorflow::Tensor tensor;
  s = cc_ctx->allocate_temp(static_cast<tensorflow::DataType>(dtype),
                            tensorflow::TensorShape(dimarray), &tensor,
                            allocator_attr);
  if (!s.ok()) {
    ::tensorflow::Set_TF_Status_from_Status(status, s);
    return nullptr;
  }
  TF_Tensor* tf_tensor;
  tf_tensor = TF_TensorFromTensor(tensor, &s);
  if (!s.ok()) {

  let summary = "Set a tensor in a promise";

  let description = [{
    Set a tensor in a promise.

    $promise: A value of type !mlrt.promise. The underlying value will always be a tensorflow tensor.
    $tensor: A tensorflow tensor.
  }];

  let arguments = (ins
    MlrtPromiseType:$promise,
    TF_Tensor:$tensor
  );
}

def TFMapFnOp : TensorflowMlrt_Op<"tf_map_fn", [AttrSizedOperandSegments, Pure]> {

    void (*binary_add_func)(TF_OpKernelContext* ctx, TF_Tensor* a, TF_Tensor* b,
                            TF_Tensor* out));

static Status CCBinaryAddFunc(
    ::tensorflow::OpKernelContext* cc_ctx, const Tensor& cc_a,
    const Tensor& cc_b, Tensor* cc_out,
    void (*binary_add_func)(TF_OpKernelContext* ctx, TF_Tensor* a, TF_Tensor* b,
                            TF_Tensor* out)) {
  if (cc_a.dtype() == ::tensorflow::DT_INVALID) {

// CHECK-SAME: [[arg0:%.*]]: !tfrt_fallback.tf_tensor,
// CHECK-SAME: [[arg1:%.*]]: !tfrt_fallback.tf_tensor)
// CHECK-SAME: -> (!tfrt.chain, !tfrt_fallback.tf_tensor)
func.func @func_basic(
    %arg0: tensor<3x1xf32> {tf_saved_model.index_path = [0]},
    %arg1: tensor<!tf_type.resource<tensor<1x3xf32>>> {tf_saved_model.index_path = [1]})
      -> (tensor<3x3xf32> {tf_saved_model.index_path = []})