]>
Commit | Line | Data |
---|---|---|
db9ecf05 | 1 | /* SPDX-License-Identifier: LGPL-2.1-or-later */ |
c2f1db8f | 2 | #pragma once |
7f110ff9 | 3 | |
31f7bf19 | 4 | #include <stdbool.h> |
11c3a366 TA |
5 | #include <stddef.h> |
6 | #include <stdint.h> | |
c932fb71 | 7 | #include <uchar.h> |
31f7bf19 | 8 | |
7f110ff9 | 9 | #include "macro.h" |
38e0c63d | 10 | #include "missing_type.h" |
7f110ff9 | 11 | |
550a40ec | 12 | #define UTF8_REPLACEMENT_CHARACTER "\xef\xbf\xbd" |
9dd7ea9a | 13 | #define UTF8_BYTE_ORDER_MARK "\xef\xbb\xbf" |
550a40ec | 14 | |
c932fb71 | 15 | bool unichar_is_valid(char32_t c); |
f3ee6297 | 16 | |
80ab31a4 ZJS |
17 | char *utf8_is_valid_n(const char *str, size_t len_bytes) _pure_; |
18 | static inline char *utf8_is_valid(const char *s) { | |
f5fbe71d | 19 | return utf8_is_valid_n(s, SIZE_MAX); |
80ab31a4 | 20 | } |
7f110ff9 | 21 | char *ascii_is_valid(const char *s) _pure_; |
294a3121 | 22 | char *ascii_is_valid_n(const char *str, size_t len); |
7f110ff9 | 23 | |
9b49a3b4 ZJS |
24 | int utf8_to_ascii(const char *str, char replacement_char, char **ret); |
25 | ||
618727da | 26 | bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newline) _pure_; |
6ed62be0 LP |
27 | #define utf8_is_printable(str, length) utf8_is_printable_newline(str, length, true) |
28 | ||
29 | char *utf8_escape_invalid(const char *s); | |
fc96e5c0 | 30 | char *utf8_escape_non_printable_full(const char *str, size_t console_width, bool force_ellipsis); |
da88f542 | 31 | static inline char *utf8_escape_non_printable(const char *str) { |
fc96e5c0 | 32 | return utf8_escape_non_printable_full(str, SIZE_MAX, false); |
da88f542 | 33 | } |
ba961854 | 34 | |
c932fb71 | 35 | size_t utf8_encode_unichar(char *out_utf8, char32_t g); |
80b0a597 LP |
36 | size_t utf16_encode_unichar(char16_t *out, char32_t c); |
37 | ||
2ac2ff3f | 38 | char *utf16_to_utf8(const char16_t *s, size_t length /* bytes! */); |
80b0a597 LP |
39 | char16_t *utf8_to_utf16(const char *s, size_t length); |
40 | ||
da890466 | 41 | size_t char16_strlen(const char16_t *s); /* returns the number of 16-bit words in the string (not bytes!) */ |
02a36bc9 | 42 | |
92e068b4 | 43 | int utf8_encoded_valid_unichar(const char *str, size_t length); |
c932fb71 | 44 | int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar); |
04166cb7 | 45 | |
c932fb71 | 46 | static inline bool utf16_is_surrogate(char16_t c) { |
07667be7 | 47 | return c >= 0xd800U && c <= 0xdfffU; |
04166cb7 TG |
48 | } |
49 | ||
c932fb71 | 50 | static inline bool utf16_is_trailing_surrogate(char16_t c) { |
07667be7 | 51 | return c >= 0xdc00U && c <= 0xdfffU; |
04166cb7 TG |
52 | } |
53 | ||
c932fb71 | 54 | static inline char32_t utf16_surrogate_pair_to_unichar(char16_t lead, char16_t trail) { |
07667be7 | 55 | return ((((char32_t) lead - 0xd800U) << 10) + ((char32_t) trail - 0xdc00U) + 0x10000U); |
04166cb7 | 56 | } |
65ee8660 LP |
57 | |
58 | size_t utf8_n_codepoints(const char *str); | |
3f536d5b | 59 | size_t utf8_console_width(const char *str); |