#include "tensorflow/core/platform/status.h"

namespace tensorflow {

// Extracts the attributes of a MLIR operation and populates the converted
// attributes in a proto map<string, AttrValue>.
Status GetAttrValuesFromOperation(
    mlir::Operation* inst, llvm::StringRef name,
    const tensorflow::OpRegistrationData* op_reg_data,
    bool ignore_unregistered_attrs, AttrValueMap* attributes);

          mlir::dyn_cast<mlir::TF::ResourceType>(arg_type.getElementType())) {
    llvm::ArrayRef<mlir::TensorType> subtypes = resource_type.getSubtypes();
    if (!subtypes.empty()) {
      AttrValue handle_dtypes_attr;
      AttrValue handle_shapes_attr;
      for (mlir::TensorType subtype : subtypes) {
        DataType dtype;
        TF_RETURN_IF_ERROR(ConvertToDataType(subtype.getElementType(), &dtype));

          mlir::dyn_cast<mlir::TF::ResourceType>(arg_type.getElementType())) {
    llvm::ArrayRef<mlir::TensorType> subtypes = resource_type.getSubtypes();
    if (!subtypes.empty()) {
      AttrValue handle_dtypes_attr;
      AttrValue handle_shapes_attr;
      for (mlir::TensorType subtype : subtypes) {
        DataType dtype;
        TF_RETURN_IF_ERROR(ConvertToDataType(subtype.getElementType(), &dtype));

      const std::string& func_name);

  // Converts the given non-function-call AttrValue to an MLIR Attribute.
  absl::StatusOr<mlir::Attribute> ConvertAttributeValue(const AttrValue& value);

  // Converts the given function-call AttrValue to MLIR Attributes and pushes
  // them to the given attributes list. For example, if there is a kFunc
  // AttrValue {name : foo, attrs : {k1 : bar, k2 : rfc}}, it will convert it to

                                  const void* proto, size_t proto_len,
                                  TF_Status* status) {
  tensorflow::AttrValue attr_value;
  if (!attr_value.ParseFromArray(proto, proto_len)) {
    status->status = InvalidArgument(
        "Unparseable AttrValue proto passed to "
        "TF_FunctionSetAttrValueProto");
    return;
  }

  auto fdef_or = func->record->mutable_fdef();

    .SetShapeFn([](::tensorflow::shape_inference::InferenceContext* c) {
      c->set_output(0, c->input(0));
      c->set_output(1, c->Scalar());
      c->set_output(2, c->Scalar());

      const tensorflow::AttrValue* num_bins_attr;
      TF_RETURN_IF_ERROR(c->GetAttr("num_bins", &num_bins_attr));
      c->set_output(3, c->MakeShape({num_bins_attr->i()}));

      return absl::OkStatus();
    });

        for (const auto& attr : node->attrs()) {
          auto attr_key = attr.first;
          auto attr_value = attr.second;
          if (attr_key == attrKey &&
              attr_value.value_case() == AttrValue::kType &&
              attr_value.type() == DT_RESOURCE) {
            return true;
            break;
          }
        }
      }
    }
  }
  return false;
}

    }

    OpBuilder builder(loc.getContext());
    for (const auto& name_and_value : node_def.attr()) {
      const std::string& attr_name = name_and_value.first;
      const tensorflow::AttrValue& attr_value = name_and_value.second;
      absl::StatusOr<Attribute> mlir_attr =
          tensorflow::ConvertAttributeValue(attr_value, &builder);
      if (!mlir_attr.ok()) {

        break;
      }
      case ::tensorflow::AttrValue::kI:
        flex_builder->Int(key, attr.i());
        break;
      case ::tensorflow::AttrValue::kF:
        flex_builder->Float(key, attr.f());
        break;
      case ::tensorflow::AttrValue::kB:
        flex_builder->Bool(key, attr.b());
        break;
      case tensorflow::AttrValue::kList:
        if (attr.list().s_size() > 0) {

  EXPECT_EQ(0, TF_OperationOutputNumConsumers(TF_Output{feed, 0}));
  EXPECT_EQ(0, TF_OperationNumControlInputs(feed));
  EXPECT_EQ(0, TF_OperationNumControlOutputs(feed));

  tensorflow::AttrValue attr_value;
  ASSERT_TRUE(GetAttrValue(feed, "dtype", &attr_value, s)) << TF_Message(s);
  EXPECT_EQ(attr_value.type(), tensorflow::DT_INT32);

  // Test not found errors in TF_Operation*() query functions.