| 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 |  | 
| Elliott Hughes | 18181e6 | 2019-01-30 13:52:36 -0800 | [diff] [blame] | 18 | #include <uchar.h> | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 19 |  | 
|  | 20 | #include <gtest/gtest.h> | 
|  | 21 |  | 
|  | 22 | #include <errno.h> | 
|  | 23 | #include <limits.h> | 
|  | 24 | #include <locale.h> | 
|  | 25 | #include <stdint.h> | 
|  | 26 |  | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 27 | TEST(uchar, sizeof_uchar_t) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 28 | EXPECT_EQ(2U, sizeof(char16_t)); | 
|  | 29 | EXPECT_EQ(4U, sizeof(char32_t)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 30 | } | 
|  | 31 |  | 
|  | 32 | TEST(uchar, start_state) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 33 | char out[MB_LEN_MAX]; | 
|  | 34 | mbstate_t ps; | 
|  | 35 |  | 
|  | 36 | // Any non-initial state is invalid when calling c32rtomb. | 
|  | 37 | memset(&ps, 0, sizeof(ps)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 38 | EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xc2", 1, &ps)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 39 | EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(out, 0x00a2, &ps)); | 
|  | 40 | EXPECT_EQ(EILSEQ, errno); | 
|  | 41 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 42 | // 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] | 43 | // state should be reset. | 
|  | 44 | memset(&ps, 0, sizeof(ps)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 45 | EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xc2", 1, &ps)); | 
|  | 46 | EXPECT_EQ(1U, c32rtomb(nullptr, 0x00a2, &ps)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 47 | EXPECT_TRUE(mbsinit(&ps)); | 
|  | 48 |  | 
|  | 49 | memset(&ps, 0, sizeof(ps)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 50 | 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] | 51 | EXPECT_EQ(1U, c32rtomb(out, L'\0', &ps)); | 
|  | 52 | EXPECT_TRUE(mbsinit(&ps)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 53 | } | 
|  | 54 |  | 
|  | 55 | TEST(uchar, c16rtomb_null_out) { | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 56 | EXPECT_EQ(1U, c16rtomb(nullptr, L'\0', nullptr)); | 
|  | 57 | EXPECT_EQ(1U, c16rtomb(nullptr, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 58 | } | 
|  | 59 |  | 
|  | 60 | TEST(uchar, c16rtomb_null_char) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 61 | char bytes[MB_LEN_MAX]; | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 62 | EXPECT_EQ(1U, c16rtomb(bytes, L'\0', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 63 | } | 
|  | 64 |  | 
|  | 65 | TEST(uchar, c16rtomb) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 66 | char bytes[MB_LEN_MAX]; | 
|  | 67 |  | 
|  | 68 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 69 | EXPECT_EQ(1U, c16rtomb(bytes, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 70 | EXPECT_EQ('h', bytes[0]); | 
|  | 71 |  | 
|  | 72 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 73 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 74 |  | 
|  | 75 | // 1-byte UTF-8. | 
|  | 76 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 77 | EXPECT_EQ(1U, c16rtomb(bytes, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 78 | EXPECT_EQ('h', bytes[0]); | 
|  | 79 | // 2-byte UTF-8. | 
|  | 80 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 81 | EXPECT_EQ(2U, c16rtomb(bytes, 0x00a2, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 82 | EXPECT_EQ('\xc2', bytes[0]); | 
|  | 83 | EXPECT_EQ('\xa2', bytes[1]); | 
|  | 84 | // 3-byte UTF-8. | 
|  | 85 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 86 | EXPECT_EQ(3U, c16rtomb(bytes, 0x20ac, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 87 | EXPECT_EQ('\xe2', bytes[0]); | 
|  | 88 | EXPECT_EQ('\x82', bytes[1]); | 
|  | 89 | EXPECT_EQ('\xac', bytes[2]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 90 | } | 
|  | 91 |  | 
|  | 92 | TEST(uchar, c16rtomb_surrogate) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 93 | char bytes[MB_LEN_MAX]; | 
|  | 94 |  | 
|  | 95 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 96 | EXPECT_EQ(0U, c16rtomb(bytes, 0xdbea, nullptr)); | 
|  | 97 | EXPECT_EQ(4U, c16rtomb(bytes, 0xdfcd, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 98 | EXPECT_EQ('\xf4', bytes[0]); | 
|  | 99 | EXPECT_EQ('\x8a', bytes[1]); | 
|  | 100 | EXPECT_EQ('\xaf', bytes[2]); | 
|  | 101 | EXPECT_EQ('\x8d', bytes[3]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 102 | } | 
|  | 103 |  | 
|  | 104 | TEST(uchar, c16rtomb_invalid) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 105 | char bytes[MB_LEN_MAX]; | 
|  | 106 |  | 
|  | 107 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 108 | EXPECT_EQ(static_cast<size_t>(-1), c16rtomb(bytes, 0xdfcd, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 109 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 110 | EXPECT_EQ(0U, c16rtomb(bytes, 0xdbea, nullptr)); | 
|  | 111 | EXPECT_EQ(static_cast<size_t>(-1), c16rtomb(bytes, 0xdbea, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 112 | } | 
|  | 113 |  | 
|  | 114 | TEST(uchar, mbrtoc16_null) { | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 115 | ASSERT_EQ(0U, mbrtoc16(nullptr, nullptr, 0, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 116 | } | 
|  | 117 |  | 
|  | 118 | TEST(uchar, mbrtoc16_zero_len) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 119 | char16_t out; | 
|  | 120 |  | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 121 | out = L'x'; | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 122 | ASSERT_EQ(0U, mbrtoc16(&out, "hello", 0, nullptr)); | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 123 | ASSERT_EQ(L'x', out); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 124 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 125 | ASSERT_EQ(0U, mbrtoc16(&out, "hello", 0, nullptr)); | 
|  | 126 | ASSERT_EQ(0U, mbrtoc16(&out, "", 0, nullptr)); | 
|  | 127 | ASSERT_EQ(1U, mbrtoc16(&out, "hello", 1, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 128 | ASSERT_EQ(L'h', out); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 129 | } | 
|  | 130 |  | 
|  | 131 | TEST(uchar, mbrtoc16) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 132 | char16_t out; | 
|  | 133 |  | 
|  | 134 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 135 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 136 |  | 
|  | 137 | // 1-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 138 | ASSERT_EQ(1U, mbrtoc16(&out, "abcdef", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 139 | ASSERT_EQ(L'a', out); | 
|  | 140 | // 2-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 141 | ASSERT_EQ(2U, mbrtoc16(&out, "\xc2\xa2" "cdef", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 142 | ASSERT_EQ(static_cast<char16_t>(0x00a2), out); | 
|  | 143 | // 3-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 144 | ASSERT_EQ(3U, mbrtoc16(&out, "\xe2\x82\xac" "def", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 145 | ASSERT_EQ(static_cast<char16_t>(0x20ac), out); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 146 | } | 
|  | 147 |  | 
|  | 148 | TEST(uchar, mbrtoc16_surrogate) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 149 | char16_t out; | 
|  | 150 |  | 
|  | 151 | ASSERT_EQ(static_cast<size_t>(-3), | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 152 | mbrtoc16(&out, "\xf4\x8a\xaf\x8d", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 153 | ASSERT_EQ(static_cast<char16_t>(0xdbea), out); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 154 | ASSERT_EQ(4U, mbrtoc16(&out, "\xf4\x8a\xaf\x8d" "ef", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 155 | ASSERT_EQ(static_cast<char16_t>(0xdfcd), out); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 156 | } | 
|  | 157 |  | 
|  | 158 | TEST(uchar, mbrtoc16_reserved_range) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 159 | char16_t out; | 
|  | 160 | ASSERT_EQ(static_cast<size_t>(-1), | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 161 | mbrtoc16(&out, "\xf0\x80\xbf\xbf", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 162 | } | 
|  | 163 |  | 
|  | 164 | TEST(uchar, mbrtoc16_beyond_range) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 165 | char16_t out; | 
|  | 166 | ASSERT_EQ(static_cast<size_t>(-1), | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 167 | mbrtoc16(&out, "\xf5\x80\x80\x80", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 168 | } | 
|  | 169 |  | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 170 | void test_mbrtoc16_incomplete(mbstate_t* ps) { | 
|  | 171 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 172 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 173 |  | 
|  | 174 | char16_t out; | 
|  | 175 | // 2-byte UTF-8. | 
|  | 176 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xc2", 1, ps)); | 
|  | 177 | ASSERT_EQ(1U, mbrtoc16(&out, "\xa2" "cdef", 5, ps)); | 
|  | 178 | ASSERT_EQ(static_cast<char16_t>(0x00a2), out); | 
|  | 179 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 180 | // 3-byte UTF-8. | 
|  | 181 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xe2", 1, ps)); | 
|  | 182 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\x82", 1, ps)); | 
|  | 183 | ASSERT_EQ(1U, mbrtoc16(&out, "\xac" "def", 4, ps)); | 
|  | 184 | ASSERT_EQ(static_cast<char16_t>(0x20ac), out); | 
|  | 185 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 186 | // 4-byte UTF-8. | 
|  | 187 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xf4", 1, ps)); | 
|  | 188 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\x8a\xaf", 2, ps)); | 
|  | 189 | ASSERT_EQ(static_cast<size_t>(-3), mbrtoc16(&out, "\x8d" "ef", 3, ps)); | 
|  | 190 | ASSERT_EQ(static_cast<char16_t>(0xdbea), out); | 
|  | 191 | ASSERT_EQ(1U, mbrtoc16(&out, "\x80" "ef", 3, ps)); | 
|  | 192 | ASSERT_EQ(static_cast<char16_t>(0xdfcd), out); | 
|  | 193 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 194 |  | 
|  | 195 | // Invalid 2-byte | 
|  | 196 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xc2", 1, ps)); | 
|  | 197 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc16(&out, "\x20" "cdef", 5, ps)); | 
|  | 198 | ASSERT_EQ(EILSEQ, errno); | 
|  | 199 | } | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 200 |  | 
|  | 201 | TEST(uchar, mbrtoc16_incomplete) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 202 | mbstate_t ps; | 
|  | 203 | memset(&ps, 0, sizeof(ps)); | 
|  | 204 |  | 
|  | 205 | test_mbrtoc16_incomplete(&ps); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 206 | test_mbrtoc16_incomplete(nullptr); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 207 | } | 
|  | 208 |  | 
|  | 209 | TEST(uchar, c32rtomb) { | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 210 | EXPECT_EQ(1U, c32rtomb(nullptr, L'\0', nullptr)); | 
|  | 211 | EXPECT_EQ(1U, c32rtomb(nullptr, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 212 |  | 
|  | 213 | char bytes[MB_LEN_MAX]; | 
|  | 214 |  | 
| Elliott Hughes | 697f42a | 2017-07-14 17:00:05 -0700 | [diff] [blame] | 215 | memset(bytes, 1, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 216 | EXPECT_EQ(1U, c32rtomb(bytes, L'\0', nullptr)); | 
| Elliott Hughes | 697f42a | 2017-07-14 17:00:05 -0700 | [diff] [blame] | 217 | EXPECT_EQ('\0', bytes[0]); | 
|  | 218 | EXPECT_EQ('\x01', bytes[1]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 219 |  | 
|  | 220 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 221 | EXPECT_EQ(1U, c32rtomb(bytes, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 222 | EXPECT_EQ('h', bytes[0]); | 
|  | 223 |  | 
|  | 224 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 225 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 226 |  | 
|  | 227 | // 1-byte UTF-8. | 
|  | 228 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 229 | EXPECT_EQ(1U, c32rtomb(bytes, L'h', nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 230 | EXPECT_EQ('h', bytes[0]); | 
|  | 231 | // 2-byte UTF-8. | 
|  | 232 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 233 | EXPECT_EQ(2U, c32rtomb(bytes, 0x00a2, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 234 | EXPECT_EQ('\xc2', bytes[0]); | 
|  | 235 | EXPECT_EQ('\xa2', bytes[1]); | 
|  | 236 | // 3-byte UTF-8. | 
|  | 237 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 238 | EXPECT_EQ(3U, c32rtomb(bytes, 0x20ac, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 239 | EXPECT_EQ('\xe2', bytes[0]); | 
|  | 240 | EXPECT_EQ('\x82', bytes[1]); | 
|  | 241 | EXPECT_EQ('\xac', bytes[2]); | 
|  | 242 | // 4-byte UTF-8. | 
|  | 243 | memset(bytes, 0, sizeof(bytes)); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 244 | EXPECT_EQ(4U, c32rtomb(bytes, 0x24b62, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 245 | EXPECT_EQ('\xf0', bytes[0]); | 
|  | 246 | EXPECT_EQ('\xa4', bytes[1]); | 
|  | 247 | EXPECT_EQ('\xad', bytes[2]); | 
|  | 248 | EXPECT_EQ('\xa2', bytes[3]); | 
|  | 249 | // Invalid code point. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 250 | EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(bytes, 0xffffffff, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 251 | EXPECT_EQ(EILSEQ, errno); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 252 | } | 
|  | 253 |  | 
| Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 254 | TEST(uchar, mbrtoc32_valid_non_characters) { | 
| Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 255 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 256 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 257 |  | 
|  | 258 | char32_t out[8] = {}; | 
|  | 259 | ASSERT_EQ(3U, mbrtoc32(out, "\xef\xbf\xbe", 3, nullptr)); | 
|  | 260 | ASSERT_EQ(0xfffeU, out[0]); | 
|  | 261 | ASSERT_EQ(3U, mbrtoc32(out, "\xef\xbf\xbf", 3, nullptr)); | 
|  | 262 | ASSERT_EQ(0xffffU, out[0]); | 
| Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 263 | } | 
|  | 264 |  | 
|  | 265 | TEST(uchar, mbrtoc32_out_of_range) { | 
| Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 266 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 267 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 268 |  | 
|  | 269 | char32_t out[8] = {}; | 
|  | 270 | errno = 0; | 
|  | 271 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf5\x80\x80\x80", 4, nullptr)); | 
|  | 272 | ASSERT_EQ(EILSEQ, errno); | 
| Elliott Hughes | 402c762 | 2018-07-06 17:18:05 -0700 | [diff] [blame] | 273 | } | 
|  | 274 |  | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 275 | TEST(uchar, mbrtoc32) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 276 | char32_t out[8]; | 
|  | 277 |  | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 278 | out[0] = L'x'; | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 279 | ASSERT_EQ(0U, mbrtoc32(out, "hello", 0, nullptr)); | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 280 | ASSERT_EQ(static_cast<char32_t>(L'x'), out[0]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 281 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 282 | ASSERT_EQ(0U, mbrtoc32(out, "hello", 0, nullptr)); | 
|  | 283 | ASSERT_EQ(0U, mbrtoc32(out, "", 0, nullptr)); | 
|  | 284 | ASSERT_EQ(1U, mbrtoc32(out, "hello", 1, nullptr)); | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 285 | ASSERT_EQ(static_cast<char32_t>(L'h'), out[0]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 286 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 287 | ASSERT_EQ(0U, mbrtoc32(nullptr, "hello", 0, nullptr)); | 
|  | 288 | ASSERT_EQ(0U, mbrtoc32(nullptr, "", 0, nullptr)); | 
|  | 289 | ASSERT_EQ(1U, mbrtoc32(nullptr, "hello", 1, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 290 |  | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 291 | ASSERT_EQ(0U, mbrtoc32(nullptr, nullptr, 0, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 292 |  | 
|  | 293 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 294 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 295 |  | 
|  | 296 | // 1-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 297 | ASSERT_EQ(1U, mbrtoc32(out, "abcdef", 6, nullptr)); | 
| Alexander Ivchenko | 68b0166 | 2014-06-11 16:20:54 +0400 | [diff] [blame] | 298 | ASSERT_EQ(static_cast<char32_t>(L'a'), out[0]); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 299 | // 2-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 300 | ASSERT_EQ(2U, mbrtoc32(out, "\xc2\xa2" "cdef", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 301 | ASSERT_EQ(static_cast<char32_t>(0x00a2), out[0]); | 
|  | 302 | // 3-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 303 | ASSERT_EQ(3U, mbrtoc32(out, "\xe2\x82\xac" "def", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 304 | ASSERT_EQ(static_cast<char32_t>(0x20ac), out[0]); | 
|  | 305 | // 4-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 306 | ASSERT_EQ(4U, mbrtoc32(out, "\xf0\xa4\xad\xa2" "ef", 6, nullptr)); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 307 | ASSERT_EQ(static_cast<char32_t>(0x24b62), out[0]); | 
|  | 308 | #if defined(__BIONIC__) // glibc allows this. | 
|  | 309 | // Illegal 5-byte UTF-8. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 310 | 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] | 311 | ASSERT_EQ(EILSEQ, errno); | 
|  | 312 | #endif | 
|  | 313 | // Illegal over-long sequence. | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 314 | 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] | 315 | ASSERT_EQ(EILSEQ, errno); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 316 | } | 
|  | 317 |  | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 318 | void test_mbrtoc32_incomplete(mbstate_t* ps) { | 
|  | 319 | ASSERT_STREQ("C.UTF-8", setlocale(LC_CTYPE, "C.UTF-8")); | 
|  | 320 | uselocale(LC_GLOBAL_LOCALE); | 
|  | 321 |  | 
|  | 322 | char32_t out; | 
|  | 323 | // 2-byte UTF-8. | 
|  | 324 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xc2", 1, ps)); | 
|  | 325 | ASSERT_EQ(1U, mbrtoc32(&out, "\xa2" "cdef", 5, ps)); | 
|  | 326 | ASSERT_EQ(static_cast<char32_t>(0x00a2), out); | 
|  | 327 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 328 | // 3-byte UTF-8. | 
|  | 329 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xe2", 1, ps)); | 
|  | 330 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\x82", 1, ps)); | 
|  | 331 | ASSERT_EQ(1U, mbrtoc32(&out, "\xac" "def", 4, ps)); | 
|  | 332 | ASSERT_EQ(static_cast<char32_t>(0x20ac), out); | 
|  | 333 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 334 | // 4-byte UTF-8. | 
|  | 335 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xf0", 1, ps)); | 
|  | 336 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xa4\xad", 2, ps)); | 
|  | 337 | ASSERT_EQ(1U, mbrtoc32(&out, "\xa2" "ef", 3, ps)); | 
|  | 338 | ASSERT_EQ(static_cast<char32_t>(0x24b62), out); | 
|  | 339 | ASSERT_TRUE(mbsinit(ps)); | 
|  | 340 |  | 
|  | 341 | // Invalid 2-byte | 
|  | 342 | ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xc2", 1, ps)); | 
|  | 343 | ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(&out, "\x20" "cdef", 5, ps)); | 
|  | 344 | ASSERT_EQ(EILSEQ, errno); | 
|  | 345 | } | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 346 |  | 
|  | 347 | TEST(uchar, mbrtoc32_incomplete) { | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 348 | mbstate_t ps; | 
|  | 349 | memset(&ps, 0, sizeof(ps)); | 
|  | 350 |  | 
|  | 351 | test_mbrtoc32_incomplete(&ps); | 
| Yi Kong | 32bc0fc | 2018-08-02 17:31:13 -0700 | [diff] [blame] | 352 | test_mbrtoc32_incomplete(nullptr); | 
| Dan Albert | 7a7f995 | 2014-06-02 11:33:04 -0700 | [diff] [blame] | 353 | } |