Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (C) 2014 The Android Open Source Project |
| 3 | * |
| 4 | * Licensed under the Apache License, Version 2.0 (the "License"); |
| 5 | * you may not use this file except in compliance with the License. |
| 6 | * You may obtain a copy of the License at |
| 7 | * |
| 8 | * http://www.apache.org/licenses/LICENSE-2.0 |
| 9 | * |
| 10 | * Unless required by applicable law or agreed to in writing, software |
| 11 | * distributed under the License is distributed on an "AS IS" BASIS, |
| 12 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 13 | * See the License for the specific language governing permissions and |
| 14 | * limitations under the License. |
| 15 | */ |
| 16 | |
| 17 | |
| 18 | #include <sys/cdefs.h> |
| 19 | #if defined(__BIONIC__) |
| 20 | #define HAVE_UCHAR 1 |
| 21 | #elif defined(__GLIBC__) |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 22 | #define HAVE_UCHAR __GLIBC_PREREQ(2, 16) |
| 23 | #endif |
| 24 | |
| 25 | #include <gtest/gtest.h> |
| 26 | |
| 27 | #include <errno.h> |
| 28 | #include <limits.h> |
| 29 | #include <locale.h> |
| 30 | #include <stdint.h> |
| 31 | |
| 32 | #if HAVE_UCHAR |
| 33 | #include <uchar.h> |
| 34 | #endif |
| 35 | |
| 36 | TEST(uchar, sizeof_uchar_t) { |
| 37 | #if HAVE_UCHAR |
| 38 | EXPECT_EQ(2U, sizeof(char16_t)); |
| 39 | EXPECT_EQ(4U, sizeof(char32_t)); |
| 40 | #else |
| 41 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 42 | #endif |
| 43 | } |
| 44 | |
| 45 | TEST(uchar, start_state) { |
| 46 | #if HAVE_UCHAR |
| 47 | char out[MB_LEN_MAX]; |
| 48 | mbstate_t ps; |
| 49 | |
| 50 | // Any non-initial state is invalid when calling c32rtomb. |
| 51 | memset(&ps, 0, sizeof(ps)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 52 | EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xc2", 1, &ps)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 53 | EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(out, 0x00a2, &ps)); |
| 54 | EXPECT_EQ(EILSEQ, errno); |
| 55 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 56 | // If the first argument to c32rtomb is nullptr or the second is L'\0' the shift |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 57 | // state should be reset. |
| 58 | memset(&ps, 0, sizeof(ps)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 59 | EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xc2", 1, &ps)); |
| 60 | EXPECT_EQ(1U, c32rtomb(nullptr, 0x00a2, &ps)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 61 | EXPECT_TRUE(mbsinit(&ps)); |
| 62 | |
| 63 | memset(&ps, 0, sizeof(ps)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 64 | EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xf0\xa4", 1, &ps)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 65 | EXPECT_EQ(1U, c32rtomb(out, L'\0', &ps)); |
| 66 | EXPECT_TRUE(mbsinit(&ps)); |
| 67 | #else |
| 68 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 69 | #endif |
| 70 | } |
| 71 | |
| 72 | TEST(uchar, c16rtomb_null_out) { |
| 73 | #if HAVE_UCHAR |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 74 | EXPECT_EQ(1U, c16rtomb(nullptr, L'\0', nullptr)); |
| 75 | EXPECT_EQ(1U, c16rtomb(nullptr, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 76 | #else |
| 77 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 78 | #endif |
| 79 | } |
| 80 | |
| 81 | TEST(uchar, c16rtomb_null_char) { |
| 82 | #if HAVE_UCHAR |
| 83 | char bytes[MB_LEN_MAX]; |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 84 | EXPECT_EQ(1U, c16rtomb(bytes, L'\0', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 85 | #else |
| 86 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 87 | #endif |
| 88 | } |
| 89 | |
| 90 | TEST(uchar, c16rtomb) { |
| 91 | #if HAVE_UCHAR |
| 92 | char bytes[MB_LEN_MAX]; |
| 93 | |
| 94 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 95 | EXPECT_EQ(1U, c16rtomb(bytes, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 96 | EXPECT_EQ('h', bytes[0]); |
| 97 | |
| 98 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 99 | uselocale(LC_GLOBAL_LOCALE); |
| 100 | |
| 101 | // 1-byte UTF-8. |
| 102 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 103 | EXPECT_EQ(1U, c16rtomb(bytes, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 104 | EXPECT_EQ('h', bytes[0]); |
| 105 | // 2-byte UTF-8. |
| 106 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 107 | EXPECT_EQ(2U, c16rtomb(bytes, 0x00a2, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 108 | EXPECT_EQ('\xc2', bytes[0]); |
| 109 | EXPECT_EQ('\xa2', bytes[1]); |
| 110 | // 3-byte UTF-8. |
| 111 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 112 | EXPECT_EQ(3U, c16rtomb(bytes, 0x20ac, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 113 | EXPECT_EQ('\xe2', bytes[0]); |
| 114 | EXPECT_EQ('\x82', bytes[1]); |
| 115 | EXPECT_EQ('\xac', bytes[2]); |
| 116 | #else |
| 117 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 118 | #endif |
| 119 | } |
| 120 | |
| 121 | TEST(uchar, c16rtomb_surrogate) { |
| 122 | #if HAVE_UCHAR |
| 123 | char bytes[MB_LEN_MAX]; |
| 124 | |
| 125 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 126 | EXPECT_EQ(0U, c16rtomb(bytes, 0xdbea, nullptr)); |
| 127 | EXPECT_EQ(4U, c16rtomb(bytes, 0xdfcd, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 128 | EXPECT_EQ('\xf4', bytes[0]); |
| 129 | EXPECT_EQ('\x8a', bytes[1]); |
| 130 | EXPECT_EQ('\xaf', bytes[2]); |
| 131 | EXPECT_EQ('\x8d', bytes[3]); |
| 132 | #else |
| 133 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 134 | #endif |
| 135 | } |
| 136 | |
| 137 | TEST(uchar, c16rtomb_invalid) { |
| 138 | #if HAVE_UCHAR |
| 139 | char bytes[MB_LEN_MAX]; |
| 140 | |
| 141 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 142 | EXPECT_EQ(static_cast<size_t>(-1), c16rtomb(bytes, 0xdfcd, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 143 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 144 | EXPECT_EQ(0U, c16rtomb(bytes, 0xdbea, nullptr)); |
| 145 | EXPECT_EQ(static_cast<size_t>(-1), c16rtomb(bytes, 0xdbea, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 146 | #else |
| 147 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 148 | #endif |
| 149 | } |
| 150 | |
| 151 | TEST(uchar, mbrtoc16_null) { |
| 152 | #if HAVE_UCHAR |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 153 | ASSERT_EQ(0U, mbrtoc16(nullptr, nullptr, 0, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 154 | #else |
| 155 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 156 | #endif |
| 157 | } |
| 158 | |
| 159 | TEST(uchar, mbrtoc16_zero_len) { |
| 160 | #if HAVE_UCHAR |
| 161 | char16_t out; |
| 162 | |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 163 | out = L'x'; |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 164 | ASSERT_EQ(0U, mbrtoc16(&out, "hello", 0, nullptr)); |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 165 | ASSERT_EQ(L'x', out); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 166 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 167 | ASSERT_EQ(0U, mbrtoc16(&out, "hello", 0, nullptr)); |
| 168 | ASSERT_EQ(0U, mbrtoc16(&out, "", 0, nullptr)); |
| 169 | ASSERT_EQ(1U, mbrtoc16(&out, "hello", 1, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 170 | ASSERT_EQ(L'h', out); |
| 171 | #else |
| 172 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 173 | #endif |
| 174 | } |
| 175 | |
| 176 | TEST(uchar, mbrtoc16) { |
| 177 | #if HAVE_UCHAR |
| 178 | char16_t out; |
| 179 | |
| 180 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 181 | uselocale(LC_GLOBAL_LOCALE); |
| 182 | |
| 183 | // 1-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 184 | ASSERT_EQ(1U, mbrtoc16(&out, "abcdef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 185 | ASSERT_EQ(L'a', out); |
| 186 | // 2-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 187 | ASSERT_EQ(2U, mbrtoc16(&out, "\xc2\xa2" "cdef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 188 | ASSERT_EQ(static_cast<char16_t>(0x00a2), out); |
| 189 | // 3-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 190 | ASSERT_EQ(3U, mbrtoc16(&out, "\xe2\x82\xac" "def", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 191 | ASSERT_EQ(static_cast<char16_t>(0x20ac), out); |
| 192 | #else |
| 193 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 194 | #endif |
| 195 | } |
| 196 | |
| 197 | TEST(uchar, mbrtoc16_surrogate) { |
| 198 | #if HAVE_UCHAR |
| 199 | char16_t out; |
| 200 | |
| 201 | ASSERT_EQ(static_cast<size_t>(-3), |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 202 | mbrtoc16(&out, "\xf4\x8a\xaf\x8d", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 203 | ASSERT_EQ(static_cast<char16_t>(0xdbea), out); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 204 | ASSERT_EQ(4U, mbrtoc16(&out, "\xf4\x8a\xaf\x8d" "ef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 205 | ASSERT_EQ(static_cast<char16_t>(0xdfcd), out); |
| 206 | #else |
| 207 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 208 | #endif |
| 209 | } |
| 210 | |
| 211 | TEST(uchar, mbrtoc16_reserved_range) { |
| 212 | #if HAVE_UCHAR |
| 213 | char16_t out; |
| 214 | ASSERT_EQ(static_cast<size_t>(-1), |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 215 | mbrtoc16(&out, "\xf0\x80\xbf\xbf", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 216 | #else |
| 217 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 218 | #endif |
| 219 | } |
| 220 | |
| 221 | TEST(uchar, mbrtoc16_beyond_range) { |
| 222 | #if HAVE_UCHAR |
| 223 | char16_t out; |
| 224 | ASSERT_EQ(static_cast<size_t>(-1), |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 225 | mbrtoc16(&out, "\xf5\x80\x80\x80", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 226 | #else |
| 227 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 228 | #endif |
| 229 | } |
| 230 | |
| 231 | #if HAVE_UCHAR |
| 232 | void test_mbrtoc16_incomplete(mbstate_t* ps) { |
| 233 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 234 | uselocale(LC_GLOBAL_LOCALE); |
| 235 | |
| 236 | char16_t out; |
| 237 | // 2-byte UTF-8. |
| 238 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xc2", 1, ps)); |
| 239 | ASSERT_EQ(1U, mbrtoc16(&out, "\xa2" "cdef", 5, ps)); |
| 240 | ASSERT_EQ(static_cast<char16_t>(0x00a2), out); |
| 241 | ASSERT_TRUE(mbsinit(ps)); |
| 242 | // 3-byte UTF-8. |
| 243 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xe2", 1, ps)); |
| 244 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\x82", 1, ps)); |
| 245 | ASSERT_EQ(1U, mbrtoc16(&out, "\xac" "def", 4, ps)); |
| 246 | ASSERT_EQ(static_cast<char16_t>(0x20ac), out); |
| 247 | ASSERT_TRUE(mbsinit(ps)); |
| 248 | // 4-byte UTF-8. |
| 249 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xf4", 1, ps)); |
| 250 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\x8a\xaf", 2, ps)); |
| 251 | ASSERT_EQ(static_cast<size_t>(-3), mbrtoc16(&out, "\x8d" "ef", 3, ps)); |
| 252 | ASSERT_EQ(static_cast<char16_t>(0xdbea), out); |
| 253 | ASSERT_EQ(1U, mbrtoc16(&out, "\x80" "ef", 3, ps)); |
| 254 | ASSERT_EQ(static_cast<char16_t>(0xdfcd), out); |
| 255 | ASSERT_TRUE(mbsinit(ps)); |
| 256 | |
| 257 | // Invalid 2-byte |
| 258 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xc2", 1, ps)); |
| 259 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc16(&out, "\x20" "cdef", 5, ps)); |
| 260 | ASSERT_EQ(EILSEQ, errno); |
| 261 | } |
| 262 | #endif |
| 263 | |
| 264 | TEST(uchar, mbrtoc16_incomplete) { |
| 265 | #if HAVE_UCHAR |
| 266 | mbstate_t ps; |
| 267 | memset(&ps, 0, sizeof(ps)); |
| 268 | |
| 269 | test_mbrtoc16_incomplete(&ps); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 270 | test_mbrtoc16_incomplete(nullptr); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 271 | #else |
| 272 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 273 | #endif |
| 274 | } |
| 275 | |
| 276 | TEST(uchar, c32rtomb) { |
| 277 | #if HAVE_UCHAR |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 278 | EXPECT_EQ(1U, c32rtomb(nullptr, L'\0', nullptr)); |
| 279 | EXPECT_EQ(1U, c32rtomb(nullptr, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 280 | |
| 281 | char bytes[MB_LEN_MAX]; |
| 282 | |
Elliott Hughes | 697f42a | 2017-07-14 17:00:05 -0700 | [diff] [blame] | 283 | memset(bytes, 1, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 284 | EXPECT_EQ(1U, c32rtomb(bytes, L'\0', nullptr)); |
Elliott Hughes | 697f42a | 2017-07-14 17:00:05 -0700 | [diff] [blame] | 285 | EXPECT_EQ('\0', bytes[0]); |
| 286 | EXPECT_EQ('\x01', bytes[1]); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 287 | |
| 288 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 289 | EXPECT_EQ(1U, c32rtomb(bytes, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 290 | EXPECT_EQ('h', bytes[0]); |
| 291 | |
| 292 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 293 | uselocale(LC_GLOBAL_LOCALE); |
| 294 | |
| 295 | // 1-byte UTF-8. |
| 296 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 297 | EXPECT_EQ(1U, c32rtomb(bytes, L'h', nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 298 | EXPECT_EQ('h', bytes[0]); |
| 299 | // 2-byte UTF-8. |
| 300 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 301 | EXPECT_EQ(2U, c32rtomb(bytes, 0x00a2, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 302 | EXPECT_EQ('\xc2', bytes[0]); |
| 303 | EXPECT_EQ('\xa2', bytes[1]); |
| 304 | // 3-byte UTF-8. |
| 305 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 306 | EXPECT_EQ(3U, c32rtomb(bytes, 0x20ac, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 307 | EXPECT_EQ('\xe2', bytes[0]); |
| 308 | EXPECT_EQ('\x82', bytes[1]); |
| 309 | EXPECT_EQ('\xac', bytes[2]); |
| 310 | // 4-byte UTF-8. |
| 311 | memset(bytes, 0, sizeof(bytes)); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 312 | EXPECT_EQ(4U, c32rtomb(bytes, 0x24b62, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 313 | EXPECT_EQ('\xf0', bytes[0]); |
| 314 | EXPECT_EQ('\xa4', bytes[1]); |
| 315 | EXPECT_EQ('\xad', bytes[2]); |
| 316 | EXPECT_EQ('\xa2', bytes[3]); |
| 317 | // Invalid code point. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 318 | EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(bytes, 0xffffffff, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 319 | EXPECT_EQ(EILSEQ, errno); |
| 320 | #else |
| 321 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 322 | #endif |
| 323 | } |
| 324 | |
Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 325 | TEST(uchar, mbrtoc32_valid_non_characters) { |
| 326 | #if HAVE_UCHAR |
| 327 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 328 | uselocale(LC_GLOBAL_LOCALE); |
| 329 | |
| 330 | char32_t out[8] = {}; |
| 331 | ASSERT_EQ(3U, mbrtoc32(out, "\xef\xbf\xbe", 3, nullptr)); |
| 332 | ASSERT_EQ(0xfffeU, out[0]); |
| 333 | ASSERT_EQ(3U, mbrtoc32(out, "\xef\xbf\xbf", 3, nullptr)); |
| 334 | ASSERT_EQ(0xffffU, out[0]); |
| 335 | #else |
| 336 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 337 | #endif |
| 338 | } |
| 339 | |
| 340 | TEST(uchar, mbrtoc32_out_of_range) { |
| 341 | #if HAVE_UCHAR |
| 342 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 343 | uselocale(LC_GLOBAL_LOCALE); |
| 344 | |
| 345 | char32_t out[8] = {}; |
| 346 | errno = 0; |
| 347 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf5\x80\x80\x80", 4, nullptr)); |
| 348 | ASSERT_EQ(EILSEQ, errno); |
| 349 | #else |
| 350 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 351 | #endif |
| 352 | } |
| 353 | |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 354 | TEST(uchar, mbrtoc32) { |
| 355 | #if HAVE_UCHAR |
| 356 | char32_t out[8]; |
| 357 | |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 358 | out[0] = L'x'; |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 359 | ASSERT_EQ(0U, mbrtoc32(out, "hello", 0, nullptr)); |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 360 | ASSERT_EQ(static_cast<char32_t>(L'x'), out[0]); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 361 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 362 | ASSERT_EQ(0U, mbrtoc32(out, "hello", 0, nullptr)); |
| 363 | ASSERT_EQ(0U, mbrtoc32(out, "", 0, nullptr)); |
| 364 | ASSERT_EQ(1U, mbrtoc32(out, "hello", 1, nullptr)); |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 365 | ASSERT_EQ(static_cast<char32_t>(L'h'), out[0]); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 366 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 367 | ASSERT_EQ(0U, mbrtoc32(nullptr, "hello", 0, nullptr)); |
| 368 | ASSERT_EQ(0U, mbrtoc32(nullptr, "", 0, nullptr)); |
| 369 | ASSERT_EQ(1U, mbrtoc32(nullptr, "hello", 1, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 370 | |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 371 | ASSERT_EQ(0U, mbrtoc32(nullptr, nullptr, 0, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 372 | |
| 373 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 374 | uselocale(LC_GLOBAL_LOCALE); |
| 375 | |
| 376 | // 1-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 377 | ASSERT_EQ(1U, mbrtoc32(out, "abcdef", 6, nullptr)); |
Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 378 | ASSERT_EQ(static_cast<char32_t>(L'a'), out[0]); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 379 | // 2-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 380 | ASSERT_EQ(2U, mbrtoc32(out, "\xc2\xa2" "cdef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 381 | ASSERT_EQ(static_cast<char32_t>(0x00a2), out[0]); |
| 382 | // 3-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 383 | ASSERT_EQ(3U, mbrtoc32(out, "\xe2\x82\xac" "def", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 384 | ASSERT_EQ(static_cast<char32_t>(0x20ac), out[0]); |
| 385 | // 4-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 386 | ASSERT_EQ(4U, mbrtoc32(out, "\xf0\xa4\xad\xa2" "ef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 387 | ASSERT_EQ(static_cast<char32_t>(0x24b62), out[0]); |
| 388 | #if defined(__BIONIC__) // glibc allows this. |
| 389 | // Illegal 5-byte UTF-8. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 390 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf8\xa1\xa2\xa3\xa4" "f", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 391 | ASSERT_EQ(EILSEQ, errno); |
| 392 | #endif |
| 393 | // Illegal over-long sequence. |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 394 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf0\x82\x82\xac" "ef", 6, nullptr)); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 395 | ASSERT_EQ(EILSEQ, errno); |
| 396 | #else |
| 397 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 398 | #endif |
| 399 | } |
| 400 | |
| 401 | #if HAVE_UCHAR |
| 402 | void test_mbrtoc32_incomplete(mbstate_t* ps) { |
| 403 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); |
| 404 | uselocale(LC_GLOBAL_LOCALE); |
| 405 | |
| 406 | char32_t out; |
| 407 | // 2-byte UTF-8. |
| 408 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xc2", 1, ps)); |
| 409 | ASSERT_EQ(1U, mbrtoc32(&out, "\xa2" "cdef", 5, ps)); |
| 410 | ASSERT_EQ(static_cast<char32_t>(0x00a2), out); |
| 411 | ASSERT_TRUE(mbsinit(ps)); |
| 412 | // 3-byte UTF-8. |
| 413 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xe2", 1, ps)); |
| 414 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\x82", 1, ps)); |
| 415 | ASSERT_EQ(1U, mbrtoc32(&out, "\xac" "def", 4, ps)); |
| 416 | ASSERT_EQ(static_cast<char32_t>(0x20ac), out); |
| 417 | ASSERT_TRUE(mbsinit(ps)); |
| 418 | // 4-byte UTF-8. |
| 419 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xf0", 1, ps)); |
| 420 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xa4\xad", 2, ps)); |
| 421 | ASSERT_EQ(1U, mbrtoc32(&out, "\xa2" "ef", 3, ps)); |
| 422 | ASSERT_EQ(static_cast<char32_t>(0x24b62), out); |
| 423 | ASSERT_TRUE(mbsinit(ps)); |
| 424 | |
| 425 | // Invalid 2-byte |
| 426 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xc2", 1, ps)); |
| 427 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(&out, "\x20" "cdef", 5, ps)); |
| 428 | ASSERT_EQ(EILSEQ, errno); |
| 429 | } |
| 430 | #endif |
| 431 | |
| 432 | TEST(uchar, mbrtoc32_incomplete) { |
| 433 | #if HAVE_UCHAR |
| 434 | mbstate_t ps; |
| 435 | memset(&ps, 0, sizeof(ps)); |
| 436 | |
| 437 | test_mbrtoc32_incomplete(&ps); |
Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame^] | 438 | test_mbrtoc32_incomplete(nullptr); |
Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 439 | #else |
| 440 | GTEST_LOG_(INFO) << "uchar.h is unavailable.\n"; |
| 441 | #endif |
| 442 | } |