* @param utf8 input utf-8 code unit sequence * @param length length of utf-8 code unit sequence * @return codepoint of input utf8 code unit sequence */
| 123 | * @return codepoint of input utf8 code unit sequence |
| 124 | */ |
| 125 | [[nodiscard]] constexpr char32_t utf8_to_utf32(ochar8_t const* const utf8, const u64 length) noexcept |
| 126 | { |
| 127 | if (!utf8) |
| 128 | return 0; |
| 129 | const ochar8_t c0 = utf8[0]; |
| 130 | const ochar8_t lead_mask = get_utf8_mask(length); |
| 131 | char32_t utf32 = c0 & lead_mask; |
| 132 | constexpr ochar8_t following_mask = get_utf8_mask(0); |
| 133 | for (u64 i = 1; i < length; ++i) |
| 134 | { |
| 135 | constexpr u64 bits_following = 6; |
| 136 | const ochar8_t ci = utf8[i]; |
| 137 | utf32 <<= bits_following; |
| 138 | utf32 |= ci & following_mask; |
| 139 | } |
| 140 | return utf32; |
| 141 | } |
| 142 | |
| 143 | /** |
| 144 | * @param utf32 input utf-32 code unit |
no test coverage detected