Line data    Source code

       1             : /* SPDX-License-Identifier: LGPL-2.1+ */
       2             : #pragma once
       3             : 
       4             : #include <stdbool.h>
       5             : #include <stddef.h>
       6             : #include <stdint.h>
       7             : #include <uchar.h>
       8             : 
       9             : #include "macro.h"
      10             : #include "missing_type.h"
      11             : 
      12             : #define UTF8_REPLACEMENT_CHARACTER "\xef\xbf\xbd"
      13             : #define UTF8_BYTE_ORDER_MARK "\xef\xbb\xbf"
      14             : 
      15             : bool unichar_is_valid(char32_t c);
      16             : 
      17             : char *utf8_is_valid(const char *s) _pure_;
      18             : char *ascii_is_valid(const char *s) _pure_;
      19             : char *ascii_is_valid_n(const char *str, size_t len);
      20             : 
      21             : bool utf8_is_printable_newline(const char* str, size_t length, bool newline) _pure_;
      22             : #define utf8_is_printable(str, length) utf8_is_printable_newline(str, length, true)
      23             : 
      24             : char *utf8_escape_invalid(const char *s);
      25             : char *utf8_escape_non_printable_full(const char *str, size_t console_width);
      26           6 : static inline char *utf8_escape_non_printable(const char *str) {
      27           6 :         return utf8_escape_non_printable_full(str, (size_t) -1);
      28             : }
      29             : 
      30             : size_t utf8_encode_unichar(char *out_utf8, char32_t g);
      31             : size_t utf16_encode_unichar(char16_t *out, char32_t c);
      32             : 
      33             : char *utf16_to_utf8(const char16_t *s, size_t length /* bytes! */);
      34             : char16_t *utf8_to_utf16(const char *s, size_t length);
      35             : 
      36             : size_t char16_strlen(const char16_t *s); /* returns the number of 16bit words in the string (not bytes!) */
      37             : 
      38             : int utf8_encoded_valid_unichar(const char *str, size_t length);
      39             : int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar);
      40             : 
      41          81 : static inline bool utf16_is_surrogate(char16_t c) {
      42          81 :         return c >= 0xd800U && c <= 0xdfffU;
      43             : }
      44             : 
      45          95 : static inline bool utf16_is_trailing_surrogate(char16_t c) {
      46          95 :         return c >= 0xdc00U && c <= 0xdfffU;
      47             : }
      48             : 
      49          45 : static inline char32_t utf16_surrogate_pair_to_unichar(char16_t lead, char16_t trail) {
      50          45 :         return ((((char32_t) lead - 0xd800U) << 10) + ((char32_t) trail - 0xdc00U) + 0x10000U);
      51             : }
      52             : 
      53             : size_t utf8_n_codepoints(const char *str);
      54             : size_t utf8_console_width(const char *str);