// CHECK: %[[RES_PERM:.*]] = "tf.Const"() <{value = dense<[0, 2, 3, 1]> : tensor<4xi32>}>
  // CHECK: "tf.FusedBatchNormV3"(%arg0, {{.*}} <{data_format = "NCHW"
  // CHECK: %[[RES_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%y, %[[RES_PERM]])
  // CHECK: return %[[RES_TRANSPOSE]]

  // Transpose NCHW -> NHWC
  %0 = "tf.Const"() {value = dense<[0, 2, 3, 1]> : tensor<4xi32>} : () -> tensor<4xi32>

// CHECK-LABEL: func @transposeConv2D
func.func @transposeConv2D(%input: tensor<1x3x32x32xf32>, %filter: tensor<1x1x3x8xf32>) -> tensor<1x8x7x6xf32> {

  // CHECK: %[[ARG_PERM:.*]] = "tf.Const"() <{value = dense<[0, 2, 3, 1]> : tensor<4xi64>}>
  // CHECK: %[[ARG_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%arg0, %[[ARG_PERM]])

  // CHECK: %[[CONV2D:[0-9]*]] = "tf.Conv2D"(%[[ARG_TRANSPOSE]], %arg1)
  // CHECK-SAME: data_format = "NHWC"

  // CHECK-SAME: padding = "EXPLICIT"
  // CHECK-SAME: strides = [5, 8, 6, 7]
  // CHECK-SAME: (tensor<1x3x32x32xf32>, tensor<1x1x3x8xf32>) -> tensor<1x8x7x7xf32>

  // CHECK: %[[RES_PERM:.*]] = "tf.Const"() <{value = dense<[0, 2, 3, 1]> : tensor<4xi64>}>
  // CHECK: %[[RES_TRANSPOSE:[0-9]*]] = "tf.Transpose"(%[[CONV2D]], %[[RES_PERM]])
  // CHECK: return %[[RES_TRANSPOSE]]

  %0 = "tf.Conv2D"(%input, %filter)
       {
         data_format = "NHWC",

00000040  00 00 00 00 00 00 00 00  00 00 00 00 00 32 cc a9  |.............2..|
00000050  cc a8 c0 2b c0 2f c0 2c  c0 30 c0 09 c0 13 c0 0a  |...+./.,.0......|
00000060  c0 14 00 9c 00 9d 00 2f  00 35 c0 12 00 0a c0 23  |......./.5.....#|
00000070  c0 27 00 3c c0 07 c0 11  00 05 13 03 13 01 13 02  |.'.<............|
00000080  01 00 00 79 00 0b 00 02  01 00 ff 01 00 01 00 00  |...y............|

Documentation = "https://fastapi.tiangolo.com/"
Repository = "https://github.com/tiangolo/fastapi"

[project.optional-dependencies]

standard = [
    "fastapi-cli >=0.0.2",
    # For the test client
    "httpx >=0.23.0",
    # For templates
    "jinja2 >=2.11.2",
    # For forms and file uploads
    "python-multipart >=0.0.7",
    # For UJSONResponse
    "ujson >=4.0.1,!=4.0.2,!=4.1.0,!=4.2.0,!=4.3.0,!=5.0.0,!=5.1.0",

  return %2 : tensor<1x8x4x4xf32>
}
// CHECK-DAG: %[[CONST:.+]] = stablehlo.constant {{.*}} : tensor<3x3x8x8xf32>
// CHECK-DAG: %[[TRANSPOSE_0:.+]] = stablehlo.transpose %[[ARG]], dims = [0, 2, 3, 1] : (tensor<1x8x4x4xf32>) -> tensor<1x4x4x8xf32>

  const SmallVector<int64_t> permutation = {0, 2, 3, 1};

  const SmallVector<int> permutation_result = Permute<int>(arr, permutation);
  EXPECT_THAT(permutation_result, ElementsAre(0, 1, 2, 3));
}

TEST(PermutationTest, PermuteFourStringElements) {
  const SmallVector<std::string> arr = {"a", "b", "c", "d"};
  const SmallVector<int64_t> permutation = {0, 2, 3, 1};

  const SmallVector<std::string> permutation_result =

00000040  00 00 00 00 00 00 00 00  00 00 00 00 00 32 cc a9  |.............2..|
00000050  cc a8 c0 2b c0 2f c0 2c  c0 30 c0 09 c0 13 c0 0a  |...+./.,.0......|
00000060  c0 14 00 9c 00 9d 00 2f  00 35 c0 12 00 0a c0 23  |......./.5.....#|
00000070  c0 27 00 3c c0 07 c0 11  00 05 13 03 13 01 13 02  |.'.<............|
00000080  01 00 00 7f 00 0b 00 02  01 00 ff 01 00 01 00 00  |................|

}

// CHECK-LABEL:   func.func @avg_pool2d_1(
// CHECK-SAME:                            %[[VAL_0:.*]]: tensor<1x3x6x6xf32>) -> tensor<*xf32> {
// CHECK:           %[[VAL_1:.*]] = arith.constant dense<[0, 2, 3, 1]> : tensor<4xi32>
// CHECK:           %[[VAL_2:.*]] = "tfl.transpose"(%[[VAL_0]], %[[VAL_1]]) : (tensor<1x3x6x6xf32>, tensor<4xi32>) -> tensor<1x6x6x3xf32>
// CHECK:           %[[VAL_3:.*]] = arith.constant dense<0> : tensor<4x2xi32>

}

SmallVector<int64_t, 4> GetDataFormatPermutation(StringRef from, StringRef to) {
  if (from == "NHWC" && to == "NCHW") {
    return {0, 3, 1, 2};
  } else if (from == "NCHW" && to == "NHWC") {
    return {0, 2, 3, 1};
  } else {
    return {};
  }
}

// Shuffle elements in the `attr` according to the permutation. Optional
// `inner_size` allows to shuffle array attributes created from rank 2 tensors