fun x() {
    val xProperty = 10
    fun xFunction() = 11
    typealias xTypealias = 10
    class XClass<XT> {}
    enum class XEnum {
        X_ENUM_ENTRY;

        fun xEnumMember(){}
    }

    class Y <YT> {
        val yProperty = 10
        fun yFunction() = 11
        typealias yTypealias = 10
        class YClass<YTT> {}
        enum class YEnum {
            Y_ENUM_ENTRY;

            fun yEnumMember(){}

 * java.util.function.Function java.util.function.Function}.
 *
 * <p>The {@link Functions} class provides common functions and related utilities.
 *
 * <p>See the Guava User Guide article on <a
 * href="https://github.com/google/guava/wiki/FunctionalExplained">the use of {@code Function}</a>.
 *
 * <h3>For Java 8+ users</h3>
 *

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

namespace tf {
namespace libtf {

class Function {
 public:
  tensorflow::Status RegisterTrace(tensorflow::AbstractFunctionPtr,
                                   TaggedValue input_signature,
                                   TaggedValue output_signature);

  // Executes this function under the execution context.
  //
  // Raises an error is no matching signature is found for TaggedValue.

/**
 * Legacy version of {@link java.util.function.Function java.util.function.Function}.
 *
 * <p>The {@link Functions} class provides common functions and related utilities.
 *
 * <p>As this interface extends {@code java.util.function.Function}, an instance of this type can be
 * used as a {@code java.util.function.Function} directly. To use a {@code
 * java.util.function.Function} in a context where a {@code com.google.common.base.Function} is

    tf_executor.fetch
  }
  func.return
}

// CHECK:      library {
// CHECK:        function {
// CHECK-NEXT:     signature {
// CHECK-NEXT:       name: "bar"
// CHECK-NEXT:     }
// CHECK:          node_def {
// CHECK-NEXT:       name: "tf.Const"
// CHECK-NEXT:       op: "Const"
// CHECK-NEXT:       attr {

}
node {
  name: "custom_embedding_matmul_func_call"
  op: "custom_embedding_matmul"
}
library {
  function {
    signature {
      name: "custom_relu"
    }
    attr {
      key: "_implements"
      value {
        func {
          name: "tensorflow.relu"
        }
      }
    }
  }
  function {
    signature {
      name: "custom_embedding_matmul"
    }
    attr {
      key: "_implements"

  auto cleanup_tensors = absl::MakeCleanup([outs]() {
    for (auto t : outs) {
      t->Unref();
    }
  });
  return Unflatten(outs, concrete_fn.output_signature);
}

StatusOr<Function::ConcreteFunction> Function::GetConcreteFunction(
    TaggedValue value) const {
  if (concrete_fns_.empty()) {
    return tensorflow::errors::FailedPrecondition(
        "No registered ConcreteFunctions.");
  }

See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/

#include "tensorflow/compiler/mlir/tfrt/function/function.h"

#include "absl/log/log.h"
#include "absl/strings/match.h"
#include "mlir/IR/OperationSupport.h"  // from @llvm-project
#include "mlir/Pass/PassManager.h"  // from @llvm-project

  // to proactively transfer the result to host since the consumer op (or
  // function) of the result may still be on TPU.
  // TODO(b/194081364): remove this option once we unify servo TPU serving
  // result transfer behavior.
  bool tpu_transfer_result_to_host = false;
};

// Compile MLIR generated by tf.function in TF dialect into BEF.
Status CompileTFMLIRToBEF(const TfrtFunctionCompileOptions& options,

// RUN: tf-mlir-translate -mlir-to-graphdef %s -tf-graph-as-function -o - | FileCheck %s

func.func @main(%arg0: tensor<*x!tf_type.resource>, %arg1: tensor<*x!tf_type.resource<tensor<3x3x1x32xf32>>>, %arg2: tensor<*xf32>, %arg3: tensor<2x4x6x8xi32>) -> (tensor<f32>, tensor<f32>)
attributes {tf.entry_function = {inputs = "args_0,args_1,args_2,args_3", outputs = "rets_0_RetVal,rets_1_RetVal"}} {
  %graph:2 = tf_executor.graph {