// CHECK:      {
// CHECK-NEXT:     version: 3,
// CHECK-NEXT:     operator_codes: [ {
// CHECK-NEXT:       deprecated_builtin_code: 35,
// CHECK-NEXT:       version: 1,
// CHECK-NEXT:       builtin_code: UNIDIRECTIONAL_SEQUENCE_RNN
// CHECK-NEXT:     } ],
// CHECK-NEXT:     subgraphs: [ {
// CHECK-NEXT:       tensors: [ {
// CHECK-NEXT:         shape: [ 4 ],
// CHECK-NEXT:         buffer: 1,
// CHECK-NEXT:         name: "arg0",

  CUSTOM = 32,
  EMBEDDING_LOOKUP_SPARSE = 33,
  PAD = 34,
  UNIDIRECTIONAL_SEQUENCE_RNN = 35,
  GATHER = 36,
  BATCH_TO_SPACE_ND = 37,
  SPACE_TO_BATCH_ND = 38,
  TRANSPOSE = 39,
  MEAN = 40,
  SUB = 41,
  DIV = 42,
  SQUEEZE = 43,
  UNIDIRECTIONAL_SEQUENCE_LSTM = 44,
  STRIDED_SLICE = 45,
  BIDIRECTIONAL_SEQUENCE_RNN = 46,
  EXP = 47,
  TOPK_V2 = 48,
  SPLIT = 49,
  LOG_SOFTMAX = 50,

  CUSTOM = 32,
  EMBEDDING_LOOKUP_SPARSE = 33,
  PAD = 34,
  UNIDIRECTIONAL_SEQUENCE_RNN = 35,
  GATHER = 36,
  BATCH_TO_SPACE_ND = 37,
  SPACE_TO_BATCH_ND = 38,
  TRANSPOSE = 39,
  MEAN = 40,
  SUB = 41,
  DIV = 42,
  SQUEEZE = 43,
  UNIDIRECTIONAL_SEQUENCE_LSTM = 44,
  STRIDED_SLICE = 45,
  BIDIRECTIONAL_SEQUENCE_RNN = 46,
  EXP = 47,
  TOPK_V2 = 48,
  SPLIT = 49,
  LOG_SOFTMAX = 50,

  CUSTOM = 32,
  EMBEDDING_LOOKUP_SPARSE = 33,
  PAD = 34,
  UNIDIRECTIONAL_SEQUENCE_RNN = 35,
  GATHER = 36,
  BATCH_TO_SPACE_ND = 37,
  SPACE_TO_BATCH_ND = 38,
  TRANSPOSE = 39,
  MEAN = 40,
  SUB = 41,
  DIV = 42,
  SQUEEZE = 43,
  UNIDIRECTIONAL_SEQUENCE_LSTM = 44,
  STRIDED_SLICE = 45,
  BIDIRECTIONAL_SEQUENCE_RNN = 46,
  EXP = 47,
  TOPK_V2 = 48,
  SPLIT = 49,
  LOG_SOFTMAX = 50,

  // CHECK: "tfl.unidirectional_sequence_rnn"(%arg0, %arg1, %arg2, %arg3, %arg4) <{fused_activation_function = "NONE", time_major = false}> : (tensor<?xf32>, tensor<?xf32>, tensor<?xf32>, tensor<?xf32>, tensor<?x?xf32>) -> tensor<?xf32>

// CHECK-DAG:       [[VAL_2:%.*]] = arith.constant dense<0.000000e+00> : tensor<28xf32>
// CHECK-DAG:       [[VAL_3:%.*]] = arith.constant dense<0.000000e+00> : tensor<1x28xf32>
// CHECK:           [[VAL_4:%.*]] = "tfl.unidirectional_sequence_rnn"([[VAL_0]], [[VAL_1]], [[VAL_1]], [[VAL_2]], [[VAL_3]]) <{fused_activation_function = "TANH", time_major = true}> : (tensor<28x1x28xf32>, tensor<28x28xf32>, tensor<28x28xf32>, tensor<28xf32>, tensor<1x28xf32>) -> tensor<28x1x28xf32>

              22, 23, 24, 25, 26, 27, 28, 33, 40, 41, 42, 43, 44, 45, 46, 47};
    }
  }];
}

// UnidirectionalSequenceRNN op.
def TFL_UnidirectionalSequenceRNNOp : TFL_Op<"unidirectional_sequence_rnn", [
    TFL_OperandHasRank<4, 2>,
    PredOpTrait<"input and output must have same element type",
      TFL_TCresVTEtIsSameAsOp<0, 0>>,

    if (op->getNumOperands() != 5) {
      op->emitError()
          << "We're expecting 5 inputs for UnidirectionalSequenceRNN, only "
          << op->getNumOperands() << " provided";
      return failure();
    }

    if (op->getNumResults() != 2) {
      op->emitError()
          << "We're expecting 2 inputs for UnidirectionalSequenceRNN, only "
          << op->getNumResults() << " found";
      return failure();
    }

    "'LastState' type: DT_FLOAT } output_arg: { name: 'Output' type: DT_FLOAT} "
    "attr : { name: '_tflite_input_indices' type: 'list(int)'}";

constexpr mlir::StringRef kUnidirectionalSequenceRnnOp =
    "name: 'UnidirectionalSequenceRnn' input_arg: {name: 'Input' type: "
    "DT_FLOAT} input_arg: { name: 'Weights' type: DT_FLOAT } "
    "input_arg: { name: 'RecurrentWeights' type: DT_FLOAT } input_arg: { "
    "name: 'Bias' type: DT_FLOAT} "

    "CUSTOM",
    "EMBEDDING_LOOKUP_SPARSE",
    "PAD",
    "UNIDIRECTIONAL_SEQUENCE_RNN",
    "GATHER",
    "BATCH_TO_SPACE_ND",
    "SPACE_TO_BATCH_ND",
    "TRANSPOSE",
    "MEAN",
    "SUB",
    "DIV",
    "SQUEEZE",
    "UNIDIRECTIONAL_SEQUENCE_LSTM",
    "STRIDED_SLICE",
    "BIDIRECTIONAL_SEQUENCE_RNN",
    "EXP",
    "TOPK_V2",
    "SPLIT",
    "LOG_SOFTMAX",