module attributes {tf.versions = {producer = 179 : i32}} {
  func.func @main(%arg0: tensor<128x8xf32> {mhlo.sharding = "bad_sharding"}) {
    func.return
  }
}

func.func @concatConstantTensorsMiddleDim() -> tensor<1x4x3xi32> {
  %cst_0 = arith.constant dense<0> : tensor<1x2x3xi32>
  %cst_1 = arith.constant dense<1> : tensor<1x2x3xi32>
  %0 = "tfl.concatenation"(%cst_0, %cst_1) {axis = 1 : i32, fused_activation_function = "NONE"} : (tensor<1x2x3xi32>, tensor<1x2x3xi32>) -> tensor<1x4x3xi32>
  func.return %0 : tensor<1x4x3xi32>

func.func @testReshapeShapeInference(%arg0: tensor<3x4xi32>) -> tensor<*xi32> {
  %cst = arith.constant dense<[1, 6, 2]> : tensor<3xi32>
  // CHECK: "tfl.reshape"(%arg0, %cst) : (tensor<3x4xi32>, tensor<3xi32>) -> tensor<1x6x2xi32>
  %0 = "tfl.reshape"(%arg0, %cst) : (tensor<3x4xi32>, tensor<3xi32>) -> tensor<*xi32>
  func.return %0 : tensor<*xi32>
}
}

// -----

// CHECK-LABEL: testReshapeShapeInferenceUnknownDim

// CHECK:}

func.func @reshape(%arg0 : tensor<1x128xi32>) -> tensor<4x32x1xi32>{
  %0 = "vhlo.reshape_v1"(%arg0) : (tensor<1x128xi32>) -> tensor<4x32x1xi32>
  func.return %0 : tensor<4x32x1xi32>
}

//CHECK:func.func private @reshape(%arg0: tensor<1x128xi32>) -> tensor<4x32x1xi32> {
//CHECK-NEXT: %0 = "vhlo.reshape_v1"(%arg0) : (tensor<1x128xi32>) -> tensor<4x32x1xi32>
//CHECK-NEXT: return %0 : tensor<4x32x1xi32>

}

// -----

// CHECK-LABEL: func @select_batch_static_r1
func.func @select_batch_static_r1(%arg0: tensor<i1>, %arg1: tensor<2x6x8xi32>, %arg2: tensor<2x6x8xi32>) -> tensor<2x6x8xi32> {
  // CHECK: mhlo.select %arg0, %arg1, %arg2
  %0 = "tf.Select"(%arg0, %arg1, %arg2) : (tensor<i1>, tensor<2x6x8xi32>, tensor<2x6x8xi32>) -> tensor<2x6x8xi32>
  func.return %0: tensor<2x6x8xi32>
}

// -----

  // CHECK-LABEL @lowers_string_ops
  // CHECK-DAG: ml_program.global public @vars.Variable_1([]) : tensor<!tf_type.string>
  func.func @lowers_string_ops(%arg0: tensor<128xi32>, %arg1: tensor<128xi32>, %arg2: tensor<128x1xi32>, %arg3: tensor<128x90xi32>, %arg4: tensor<128x90xi32>, %arg5: tensor<128x90xi32>, %arg6: tensor<128x90x64xf32>, %arg7: tensor<128x90x64xf32>) -> tensor<!tf_type.string> {

// Examples:
//   * If `xla_gather_op_output_type` == tensor<*xf32>, then it returns:
//     tensor<*xf32>.
//   * If `xla_gather_op_output_type` == tensor<3x5xi32> and `collapsed_dims` ==
//     {0}, then it returns: tensor<1x3x5xi32>.
//   * If `xla_gather_op_output_type` == tensor<3x5xf32> and `collapsed_dims` ==
//     {1, 3}, then it returns: tensor<3x1x5x1xf32>.
Type GetSliceOpOutputType(Type xla_gather_op_output_type,

    // CHECK:      metadata
    // CHECK-SAME: args
    // CHECK-SAME: shape
    // CHECK-SAME: dim
    // CHECK-SAME: size: 1
    // CHECK-SAME: dim
    // CHECK-SAME: size: 2
    // CHECK-SAME: dim
    // CHECK-SAME: size: 3

    func.return %0: tensor<1x2x3xi32>
  }

// -----

func.func @testConcatInvalidOperandRankGreater(%arg0: tensor<1x1x2xi32>, %arg1: tensor<1x1x2xi32>) -> tensor<2x2xi32> {
  // expected-error @+1 {{'tfl.concatenation' op rank of operand #0 must be equal to rank of output, expected 2, got 3}}
  %0 = "tfl.concatenation"(%arg0, %arg1) {axis = 0 : i32, fused_activation_function = "NONE"} : (tensor<1x1x2xi32>, tensor<1x1x2xi32>) -> tensor<2x2xi32>
  func.return %0 : tensor<2x2xi32>
}

// CHECK-NEXT:    %[[RESHAPED_1:.*]] = mhlo.reshape %arg1
// CHECK-NEXT:    %[[BMM_0:.*]] = "tfl.batch_matmul"(%[[RESHAPED_0]], %[[RESHAPED_1]]) <{adj_x = false, adj_y = false, asymmetric_quantize_inputs = false}> : {{.*}} -> tensor<1x8xi32>
// CHECK-NEXT:    %[[RESHAPED_BMM:.*]] = mhlo.reshape %[[BMM_0]]
// CHECK-NEXT:    return %[[RESHAPED_BMM]] : tensor<8xi32>
}