diff options
Diffstat (limited to 'lib9p/utf8.h')
| -rw-r--r-- | lib9p/utf8.h | 34 |
1 files changed, 34 insertions, 0 deletions
diff --git a/lib9p/utf8.h b/lib9p/utf8.h new file mode 100644 index 0000000..5ffd674 --- /dev/null +++ b/lib9p/utf8.h @@ -0,0 +1,34 @@ +/* lib9p/utf8.h - Internal UTF-8 validation + * + * Copyright (C) 2024-2025 Luke T. Shumaker <lukeshu@lukeshu.com> + * SPDX-License-Identifier: AGPL-3.0-or-later + */ + +#ifndef _LIB9P_UTF8_H_ +#define _LIB9P_UTF8_H_ + +static inline bool _is_valid_utf8(uint8_t *str, size_t len, bool forbid_nul) { + uint32_t ch; + uint8_t chlen; + assert(str); + for (size_t pos = 0; pos < len;) { + if ((str[pos] & 0b10000000) == 0b00000000) { ch = str[pos] & 0b01111111; chlen = 1; } + else if ((str[pos] & 0b11100000) == 0b11000000) { ch = str[pos] & 0b00011111; chlen = 2; } + else if ((str[pos] & 0b11110000) == 0b11100000) { ch = str[pos] & 0b00001111; chlen = 3; } + else if ((str[pos] & 0b11111000) == 0b11110000) { ch = str[pos] & 0b00000111; chlen = 4; } + else return false; + if ((ch == 0 && (chlen != 1 || forbid_nul)) || pos + chlen > len) return false; + for (uint8_t i = 1; i < chlen; i++) { + if ((str[pos+i] & 0b11000000) != 0b10000000) return false; + ch = (ch << 6) | (str[pos+i] & 0b00111111); + } + if (ch > 0x10FFFF) return false; + pos += chlen; + } + return true; +} + +#define is_valid_utf8(str, len) _is_valid_utf8(str, len, false) +#define is_valid_utf8_without_nul(str, len) _is_valid_utf8(str, len, true) + +#endif /* _LIB9P_UTF8_H_ */ |