%handle_0 = "tf.VarHandleOp"() {container="c", shared_name="a"} : () -> tensor<!tf_type.resource<tensor<1x10xf32>>>

  // CHECK-LABEL: func @lowers_to_stablehlo
  func.func @lowers_to_stablehlo(%arg0: tensor<i32> {tf_saved_model.index_path = []}) -> (tensor<i32> {tf_saved_model.index_path = []})
    attributes {tf_saved_model.exported_names = ["lowers_to_stablehlo"]}
  {
    // CHECK-DAG: [[one:%.*]] = stablehlo.constant dense<1>
    // CHECK-DAG: [[twenty:%.*]] = stablehlo.constant dense<20>
    // CHECK-DAG: [[r3:%.*]] = stablehlo.subtract [[twenty]], %arg0

  "tf_saved_model.asset"() {filename = "assets/table.txt", sym_name = "v"} : () -> ()

  // CHECK: func @f()
  func.func @f(%arg0: tensor<!tf_type.string> {tf_saved_model.bound_input = @v})
  attributes {tf_saved_model.exported_names = ["f"]} {
    %0 = "tf.HashTableV2"() {container = "", device = "", key_dtype = !tf_type.string, shared_name = "", use_node_name_sharing = false, value_dtype = i64} : () -> tensor<!tf_type.resource>

      func.func @init_func1() attributes {tf_saved_model.exported_names = ["init_func1"], tf_saved_model.initializer_type = "init_op"} {
        func.return
      }

      func.func @init_func2() attributes {tf_saved_model.exported_names = ["init_func2"], tf_saved_model.initializer_type = "restore_op"} {
        func.return
      }
    }
  )mlir";

    return
  }

  func.func @main(%arg: tensor<!tf_type.string> {tf_saved_model.index_path = ["__tf_file_prefix"]}) -> ()
      attributes {tf.entry_function = {inputs = "tf_file_prefix:0", outputs = ""}, tf_saved_model.exported_names = ["main"]} {
    tf_executor.graph {
      tf_executor.fetch
    }
    return
  }
}
// Save function should be erased.
// CHECK-NOT: @tf_quant__save

  # CHECK: "tf_saved_model.global_tensor"() <{is_mutable, sym_name = "[[VAR:[a-zA-Z_0-9]+]]", type = tensor<f32>, value = dense<4.200000e+01> : tensor<f32>}> {tf_saved_model.exported_names = ["v42"]} : () -> ()
  # CHECK:      func {{@[a-zA-Z_0-9]+}}(

  # CHECK-SAME:   tensor<*xf32> {tf_saved_model.index_path = [1]})
  # CHECK-SAME: attributes{{.*}}tf_saved_model.exported_names = ["caller"]
  # CHECK:        "tf.StatefulPartitionedCall"{{.*}}f = @[[CALLEE_INTERNAL]]
  #
  # CHECK:      func private @[[CALLEE_INTERNAL]]
  # CHECK-NOT:    tf_saved_model.exported_names

  @tf.function(input_signature=[tf.TensorSpec([], tf.float32)])
  def callee(self, x):
    return x, self.v

// CHECK-SAME: tf.entry_function = {inputs = "serving_default_input2:0,serving_default_input1:0", outputs = "StatefulPartitionedCall:1,StatefulPartitionedCall:0"}
// CHECK-SAME: tf_saved_model.exported_names = ["serving_default"]
module attributes {tf.versions = {bad_consumers = [], min_consumer = 12 : i32, producer = 554 : i32}, tf_saved_model.semantics} {

    ) -> (tensor<1x1xi32> {tf_saved_model.index_path = ["r"]})
      attributes {
        tf.entry_function = {control_outputs = "", inputs = "input:0", outputs = "result:0"},
        tf_saved_model.exported_names = ["serving_default"]
    }
  {
    %0 = "tf.ReadVariableOp"(%arg1) {device = ""} : (tensor<!tf_type.resource<tensor<3x1xi32>>>) -> tensor<3x1xi32>

# Tests handling dupliate functions after V1 control flow is functionalized.

# CHECK:  func {{.*}} tf_saved_model.exported_names = ["key_1"]
# CHECK: "tf.If"
# CHECK-SAME: else_branch = @[[else:[a-zA-Z_0-9]+]]
# CHECK-SAME: then_branch = @[[then:[a-zA-Z_0-9]+]]

# CHECK:  func {{.*}} tf_saved_model.exported_names = ["key_2"]
# CHECK: "tf.If"
# CHECK-SAME: else_branch = @[[else]]
# CHECK-SAME: then_branch = @[[then]]