|
| 1 | +/* Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved. |
| 2 | +
|
| 3 | +Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | +you may not use this file except in compliance with the License. |
| 5 | +You may obtain a copy of the License at |
| 6 | +
|
| 7 | + http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | +
|
| 9 | +Unless required by applicable law or agreed to in writing, software |
| 10 | +distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See |
| 12 | +the License for the specific language governing permissions and |
| 13 | +limitations under the License. */ |
| 14 | + |
| 15 | +#include <gtest/gtest.h> |
| 16 | +#include <memory> |
| 17 | + |
| 18 | +#include "paddle/phi/api/include/strings_api.h" |
| 19 | +#include "paddle/phi/api/lib/utils/allocator.h" |
| 20 | +#include "paddle/phi/backends/all_context.h" |
| 21 | +#include "paddle/phi/core/kernel_registry.h" |
| 22 | +#include "paddle/phi/core/string_tensor.h" |
| 23 | + |
| 24 | +PD_DECLARE_KERNEL(strings_lower, CPU, ALL_LAYOUT); |
| 25 | +PD_DECLARE_KERNEL(strings_upper, CPU, ALL_LAYOUT); |
| 26 | + |
| 27 | +namespace paddle { |
| 28 | +namespace tests { |
| 29 | + |
| 30 | +using phi::CPUPlace; |
| 31 | +using phi::StringTensor; |
| 32 | +using phi::StringTensorMeta; |
| 33 | + |
| 34 | +TEST(API, case_convert) { |
| 35 | + auto cpu = CPUPlace(); |
| 36 | + const auto alloc = |
| 37 | + std::make_shared<paddle::experimental::DefaultAllocator>(cpu); |
| 38 | + // 1. create tensor |
| 39 | + const phi::DDim dims({1, 2}); |
| 40 | + StringTensorMeta meta(dims); |
| 41 | + auto cpu_strings_x = std::make_shared<phi::StringTensor>( |
| 42 | + alloc.get(), phi::StringTensorMeta(meta)); |
| 43 | + phi::DeviceContextPool& pool = phi::DeviceContextPool::Instance(); |
| 44 | + auto* dev_ctx = pool.Get(phi::CPUPlace()); |
| 45 | + |
| 46 | + pstring* cpu_strings_x_data = |
| 47 | + dev_ctx-> template Alloc<pstring>(cpu_strings_x. get()); |
| 48 | + std::string strs[] = {"A Short Pstring.", |
| 49 | + "A Large Pstring Whose Length Is Longer Than 22."}; |
| 50 | + for (int i = 0; i < 2; ++i) { |
| 51 | + cpu_strings_x_data[i] = strs[i]; |
| 52 | + } |
| 53 | + // 2. get expected results |
| 54 | + std::string expected_results[] = {strs[0], strs[0], strs[1], strs[1]}; |
| 55 | + std::transform( |
| 56 | + strs[0].begin(), strs[0].end(), expected_results[0].begin(), ::tolower); |
| 57 | + std::transform( |
| 58 | + strs[0].begin(), strs[0].end(), expected_results[1].begin(), ::toupper); |
| 59 | + std::transform( |
| 60 | + strs[1].begin(), strs[1].end(), expected_results[2].begin(), ::tolower); |
| 61 | + std::transform( |
| 62 | + strs[1].begin(), strs[1].end(), expected_results[3].begin(), ::toupper); |
| 63 | + // 3. test API, ascii encoding |
| 64 | + paddle::experimental::Tensor x(cpu_strings_x); |
| 65 | + auto lower_out = paddle::experimental::strings::lower(x, false); |
| 66 | + auto upper_out = paddle::experimental::strings::upper(x, false); |
| 67 | + |
| 68 | + auto lower_tensor = |
| 69 | + std::dynamic_pointer_cast<phi::StringTensor>(lower_out.impl()); |
| 70 | + auto upper_tensor = |
| 71 | + std::dynamic_pointer_cast<phi::StringTensor>(upper_out.impl()); |
| 72 | + ASSERT_EQ(lower_tensor->dims(), dims); |
| 73 | + ASSERT_EQ(upper_tensor->dims(), dims); |
| 74 | + |
| 75 | + auto lower_tensor_ptr = lower_tensor->data(); |
| 76 | + auto upper_tensor_ptr = upper_tensor->data(); |
| 77 | + |
| 78 | + const std::string cpu_results[] = {lower_tensor_ptr[0].data(), |
| 79 | + upper_tensor_ptr[0].data(), |
| 80 | + lower_tensor_ptr[1].data(), |
| 81 | + upper_tensor_ptr[1].data()}; |
| 82 | + |
| 83 | + for (int i = 0; i < 4; ++i) { |
| 84 | + ASSERT_EQ(cpu_results[i], expected_results[i]); |
| 85 | + } |
| 86 | +} |
| 87 | + |
| 88 | +TEST(API, case_convert_utf8) { |
| 89 | + auto cpu = CPUPlace(); |
| 90 | + const auto alloc = |
| 91 | + std::make_shared<paddle::experimental::DefaultAllocator>(cpu); |
| 92 | + // 1. create tensor |
| 93 | + const phi::DDim dims({1, 2}); |
| 94 | + StringTensorMeta meta(dims); |
| 95 | + auto cpu_strings_x = std::make_shared<phi::StringTensor>( |
| 96 | + alloc.get(), phi::StringTensorMeta(meta)); |
| 97 | + phi::DeviceContextPool& pool = phi::DeviceContextPool::Instance(); |
| 98 | + auto* dev_ctx = pool.Get(phi::CPUPlace()); |
| 99 | + |
| 100 | + pstring* cpu_strings_x_data = |
| 101 | + dev_ctx-> template Alloc<pstring>(cpu_strings_x. get()); |
| 102 | + std::string strs[] = {"óÓsscHloëË", "óÓsscHloëËóÓsscHloëËóÓsscHloëË"}; |
| 103 | + for (int i = 0; i < 2; ++i) { |
| 104 | + cpu_strings_x_data[i] = strs[i]; |
| 105 | + } |
| 106 | + // 2. get expected results |
| 107 | + std::string expected_results[] = {"óósschloëë", |
| 108 | + "ÓÓSSCHLOËË", |
| 109 | + "óósschloëëóósschloëëóósschloëë", |
| 110 | + "ÓÓSSCHLOËËÓÓSSCHLOËËÓÓSSCHLOËË"}; |
| 111 | + // 3. test API, ascii encoding |
| 112 | + paddle::experimental::Tensor x(cpu_strings_x); |
| 113 | + auto lower_out = paddle::experimental::strings::lower(x, true); |
| 114 | + auto upper_out = paddle::experimental::strings::upper(x, true); |
| 115 | + |
| 116 | + auto lower_tensor = |
| 117 | + std::dynamic_pointer_cast<phi::StringTensor>(lower_out.impl()); |
| 118 | + auto upper_tensor = |
| 119 | + std::dynamic_pointer_cast<phi::StringTensor>(upper_out.impl()); |
| 120 | + ASSERT_EQ(lower_tensor->dims(), dims); |
| 121 | + ASSERT_EQ(upper_tensor->dims(), dims); |
| 122 | + |
| 123 | + auto lower_tensor_ptr = lower_tensor->data(); |
| 124 | + auto upper_tensor_ptr = upper_tensor->data(); |
| 125 | + |
| 126 | + const char* cpu_results[] = {lower_tensor_ptr[0].data(), |
| 127 | + upper_tensor_ptr[0].data(), |
| 128 | + lower_tensor_ptr[1].data(), |
| 129 | + upper_tensor_ptr[1].data()}; |
| 130 | + |
| 131 | + for (int i = 0; i < 4; ++i) { |
| 132 | + ASSERT_EQ(std::string(cpu_results[i]), expected_results[i]); |
| 133 | + } |
| 134 | +} |
| 135 | + |
| 136 | +} // namespace tests |
| 137 | +} // namespace paddle |
0 commit comments