Andreas Gampe | 027444b | 2017-03-31 12:49:07 -0700 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (C) 2017 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 | #ifndef ART_TEST_TI_AGENT_TI_UTF_H_ |
| 18 | #define ART_TEST_TI_AGENT_TI_UTF_H_ |
| 19 | |
| 20 | #include <inttypes.h> |
| 21 | #include <string.h> |
| 22 | |
| 23 | #include "android-base/logging.h" |
| 24 | |
| 25 | namespace art { |
| 26 | namespace ti { |
| 27 | |
| 28 | inline size_t CountModifiedUtf8Chars(const char* utf8, size_t byte_count) { |
| 29 | DCHECK_LE(byte_count, strlen(utf8)); |
| 30 | size_t len = 0; |
| 31 | const char* end = utf8 + byte_count; |
| 32 | for (; utf8 < end; ++utf8) { |
| 33 | int ic = *utf8; |
| 34 | len++; |
| 35 | if (LIKELY((ic & 0x80) == 0)) { |
| 36 | // One-byte encoding. |
| 37 | continue; |
| 38 | } |
| 39 | // Two- or three-byte encoding. |
| 40 | utf8++; |
| 41 | if ((ic & 0x20) == 0) { |
| 42 | // Two-byte encoding. |
| 43 | continue; |
| 44 | } |
| 45 | utf8++; |
| 46 | if ((ic & 0x10) == 0) { |
| 47 | // Three-byte encoding. |
| 48 | continue; |
| 49 | } |
| 50 | |
| 51 | // Four-byte encoding: needs to be converted into a surrogate |
| 52 | // pair. |
| 53 | utf8++; |
| 54 | len++; |
| 55 | } |
| 56 | return len; |
| 57 | } |
| 58 | |
| 59 | inline uint16_t GetTrailingUtf16Char(uint32_t maybe_pair) { |
| 60 | return static_cast<uint16_t>(maybe_pair >> 16); |
| 61 | } |
| 62 | |
| 63 | inline uint16_t GetLeadingUtf16Char(uint32_t maybe_pair) { |
| 64 | return static_cast<uint16_t>(maybe_pair & 0x0000FFFF); |
| 65 | } |
| 66 | |
| 67 | inline uint32_t GetUtf16FromUtf8(const char** utf8_data_in) { |
| 68 | const uint8_t one = *(*utf8_data_in)++; |
| 69 | if ((one & 0x80) == 0) { |
| 70 | // one-byte encoding |
| 71 | return one; |
| 72 | } |
| 73 | |
| 74 | const uint8_t two = *(*utf8_data_in)++; |
| 75 | if ((one & 0x20) == 0) { |
| 76 | // two-byte encoding |
| 77 | return ((one & 0x1f) << 6) | (two & 0x3f); |
| 78 | } |
| 79 | |
| 80 | const uint8_t three = *(*utf8_data_in)++; |
| 81 | if ((one & 0x10) == 0) { |
| 82 | return ((one & 0x0f) << 12) | ((two & 0x3f) << 6) | (three & 0x3f); |
| 83 | } |
| 84 | |
| 85 | // Four byte encodings need special handling. We'll have |
| 86 | // to convert them into a surrogate pair. |
| 87 | const uint8_t four = *(*utf8_data_in)++; |
| 88 | |
| 89 | // Since this is a 4 byte UTF-8 sequence, it will lie between |
| 90 | // U+10000 and U+1FFFFF. |
| 91 | // |
| 92 | // TODO: What do we do about values in (U+10FFFF, U+1FFFFF) ? The |
| 93 | // spec says they're invalid but nobody appears to check for them. |
| 94 | const uint32_t code_point = ((one & 0x0f) << 18) | ((two & 0x3f) << 12) |
| 95 | | ((three & 0x3f) << 6) | (four & 0x3f); |
| 96 | |
| 97 | uint32_t surrogate_pair = 0; |
| 98 | // Step two: Write out the high (leading) surrogate to the bottom 16 bits |
| 99 | // of the of the 32 bit type. |
| 100 | surrogate_pair |= ((code_point >> 10) + 0xd7c0) & 0xffff; |
| 101 | // Step three : Write out the low (trailing) surrogate to the top 16 bits. |
| 102 | surrogate_pair |= ((code_point & 0x03ff) + 0xdc00) << 16; |
| 103 | |
| 104 | return surrogate_pair; |
| 105 | } |
| 106 | |
| 107 | inline void ConvertUtf16ToModifiedUtf8(char* utf8_out, |
| 108 | size_t byte_count, |
| 109 | const uint16_t* utf16_in, |
| 110 | size_t char_count) { |
| 111 | if (LIKELY(byte_count == char_count)) { |
| 112 | // Common case where all characters are ASCII. |
| 113 | const uint16_t *utf16_end = utf16_in + char_count; |
| 114 | for (const uint16_t *p = utf16_in; p < utf16_end;) { |
| 115 | *utf8_out++ = static_cast<char>(*p++); |
| 116 | } |
| 117 | return; |
| 118 | } |
| 119 | |
| 120 | // String contains non-ASCII characters. |
| 121 | while (char_count--) { |
| 122 | const uint16_t ch = *utf16_in++; |
| 123 | if (ch > 0 && ch <= 0x7f) { |
| 124 | *utf8_out++ = ch; |
| 125 | } else { |
| 126 | // Char_count == 0 here implies we've encountered an unpaired |
| 127 | // surrogate and we have no choice but to encode it as 3-byte UTF |
| 128 | // sequence. Note that unpaired surrogates can occur as a part of |
| 129 | // "normal" operation. |
| 130 | if ((ch >= 0xd800 && ch <= 0xdbff) && (char_count > 0)) { |
| 131 | const uint16_t ch2 = *utf16_in; |
| 132 | |
| 133 | // Check if the other half of the pair is within the expected |
| 134 | // range. If it isn't, we will have to emit both "halves" as |
| 135 | // separate 3 byte sequences. |
| 136 | if (ch2 >= 0xdc00 && ch2 <= 0xdfff) { |
| 137 | utf16_in++; |
| 138 | char_count--; |
| 139 | const uint32_t code_point = (ch << 10) + ch2 - 0x035fdc00; |
| 140 | *utf8_out++ = (code_point >> 18) | 0xf0; |
| 141 | *utf8_out++ = ((code_point >> 12) & 0x3f) | 0x80; |
| 142 | *utf8_out++ = ((code_point >> 6) & 0x3f) | 0x80; |
| 143 | *utf8_out++ = (code_point & 0x3f) | 0x80; |
| 144 | continue; |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | if (ch > 0x07ff) { |
| 149 | // Three byte encoding. |
| 150 | *utf8_out++ = (ch >> 12) | 0xe0; |
| 151 | *utf8_out++ = ((ch >> 6) & 0x3f) | 0x80; |
| 152 | *utf8_out++ = (ch & 0x3f) | 0x80; |
| 153 | } else /*(ch > 0x7f || ch == 0)*/ { |
| 154 | // Two byte encoding. |
| 155 | *utf8_out++ = (ch >> 6) | 0xc0; |
| 156 | *utf8_out++ = (ch & 0x3f) | 0x80; |
| 157 | } |
| 158 | } |
| 159 | } |
| 160 | } |
| 161 | |
| 162 | inline size_t CountUtf8Bytes(const uint16_t* chars, size_t char_count) { |
| 163 | size_t result = 0; |
| 164 | const uint16_t *end = chars + char_count; |
| 165 | while (chars < end) { |
| 166 | const uint16_t ch = *chars++; |
| 167 | if (LIKELY(ch != 0 && ch < 0x80)) { |
| 168 | result++; |
| 169 | continue; |
| 170 | } |
| 171 | if (ch < 0x800) { |
| 172 | result += 2; |
| 173 | continue; |
| 174 | } |
| 175 | if (ch >= 0xd800 && ch < 0xdc00) { |
| 176 | if (chars < end) { |
| 177 | const uint16_t ch2 = *chars; |
| 178 | // If we find a properly paired surrogate, we emit it as a 4 byte |
| 179 | // UTF sequence. If we find an unpaired leading or trailing surrogate, |
| 180 | // we emit it as a 3 byte sequence like would have done earlier. |
| 181 | if (ch2 >= 0xdc00 && ch2 < 0xe000) { |
| 182 | chars++; |
| 183 | result += 4; |
| 184 | continue; |
| 185 | } |
| 186 | } |
| 187 | } |
| 188 | result += 3; |
| 189 | } |
| 190 | return result; |
| 191 | } |
| 192 | |
| 193 | } // namespace ti |
| 194 | } // namespace art |
| 195 | |
| 196 | #endif // ART_TEST_TI_AGENT_TI_UTF_H_ |