blob: ff76e88c776b1b39a715ae6081a0243bdd83119d [file] [log] [blame]
Rob Clark78178bb2017-09-09 06:47:40 -04001/*
2 * charset conversion utils
3 *
4 * Copyright (c) 2017 Rob Clark
5 *
6 * SPDX-License-Identifier: GPL-2.0+
7 */
8
9#include <common.h>
10#include <charset.h>
11#include <malloc.h>
12
13/*
14 * utf8/utf16 conversion mostly lifted from grub
15 */
16
17size_t utf16_strlen(const uint16_t *in)
18{
19 size_t i;
20 for (i = 0; in[i]; i++);
21 return i;
22}
23
24size_t utf16_strnlen(const uint16_t *in, size_t count)
25{
26 size_t i;
27 for (i = 0; count-- && in[i]; i++);
28 return i;
29}
30
31uint16_t *utf16_strcpy(uint16_t *dest, const uint16_t *src)
32{
33 uint16_t *tmp = dest;
34
35 while ((*dest++ = *src++) != '\0')
36 /* nothing */;
37 return tmp;
38
39}
40
41uint16_t *utf16_strdup(const uint16_t *s)
42{
43 uint16_t *new;
44 if (!s || !(new = malloc((utf16_strlen(s) + 1) * 2)))
45 return NULL;
46 utf16_strcpy(new, s);
47 return new;
48}
49
50/* Convert UTF-16 to UTF-8. */
51uint8_t *utf16_to_utf8(uint8_t *dest, const uint16_t *src, size_t size)
52{
53 uint32_t code_high = 0;
54
55 while (size--) {
56 uint32_t code = *src++;
57
58 if (code_high) {
59 if (code >= 0xDC00 && code <= 0xDFFF) {
60 /* Surrogate pair. */
61 code = ((code_high - 0xD800) << 10) + (code - 0xDC00) + 0x10000;
62
63 *dest++ = (code >> 18) | 0xF0;
64 *dest++ = ((code >> 12) & 0x3F) | 0x80;
65 *dest++ = ((code >> 6) & 0x3F) | 0x80;
66 *dest++ = (code & 0x3F) | 0x80;
67 } else {
68 /* Error... */
69 *dest++ = '?';
70 /* *src may be valid. Don't eat it. */
71 src--;
72 }
73
74 code_high = 0;
75 } else {
76 if (code <= 0x007F) {
77 *dest++ = code;
78 } else if (code <= 0x07FF) {
79 *dest++ = (code >> 6) | 0xC0;
80 *dest++ = (code & 0x3F) | 0x80;
81 } else if (code >= 0xD800 && code <= 0xDBFF) {
82 code_high = code;
83 continue;
84 } else if (code >= 0xDC00 && code <= 0xDFFF) {
85 /* Error... */
86 *dest++ = '?';
87 } else if (code < 0x10000) {
88 *dest++ = (code >> 12) | 0xE0;
89 *dest++ = ((code >> 6) & 0x3F) | 0x80;
90 *dest++ = (code & 0x3F) | 0x80;
91 } else {
92 *dest++ = (code >> 18) | 0xF0;
93 *dest++ = ((code >> 12) & 0x3F) | 0x80;
94 *dest++ = ((code >> 6) & 0x3F) | 0x80;
95 *dest++ = (code & 0x3F) | 0x80;
96 }
97 }
98 }
99
100 return dest;
101}