8ff8d59957d6502cf15fd646d4b6a5e32b4b2b69
[oweals/u-boot.git] / lib / charset.c
1 // SPDX-License-Identifier: GPL-2.0+
2 /*
3  *  charset conversion utils
4  *
5  *  Copyright (c) 2017 Rob Clark
6  */
7
8 #include <charset.h>
9 #include <malloc.h>
10
11 /*
12  * utf8/utf16 conversion mostly lifted from grub
13  */
14
15 size_t u16_strlen(const u16 *in)
16 {
17         size_t i;
18         for (i = 0; in[i]; i++);
19         return i;
20 }
21
22 size_t u16_strnlen(const u16 *in, size_t count)
23 {
24         size_t i;
25         for (i = 0; count-- && in[i]; i++);
26         return i;
27 }
28
29 uint16_t *utf16_strcpy(uint16_t *dest, const uint16_t *src)
30 {
31         uint16_t *tmp = dest;
32
33         while ((*dest++ = *src++) != '\0')
34                 /* nothing */;
35         return tmp;
36
37 }
38
39 uint16_t *utf16_strdup(const uint16_t *s)
40 {
41         uint16_t *new;
42
43         if (!s)
44                 return NULL;
45         new = malloc((u16_strlen(s) + 1) * 2);
46         if (!new)
47                 return NULL;
48         utf16_strcpy(new, s);
49         return new;
50 }
51
52 /* Convert UTF-16 to UTF-8.  */
53 uint8_t *utf16_to_utf8(uint8_t *dest, const uint16_t *src, size_t size)
54 {
55         uint32_t code_high = 0;
56
57         while (size--) {
58                 uint32_t code = *src++;
59
60                 if (code_high) {
61                         if (code >= 0xDC00 && code <= 0xDFFF) {
62                                 /* Surrogate pair.  */
63                                 code = ((code_high - 0xD800) << 10) + (code - 0xDC00) + 0x10000;
64
65                                 *dest++ = (code >> 18) | 0xF0;
66                                 *dest++ = ((code >> 12) & 0x3F) | 0x80;
67                                 *dest++ = ((code >> 6) & 0x3F) | 0x80;
68                                 *dest++ = (code & 0x3F) | 0x80;
69                         } else {
70                                 /* Error...  */
71                                 *dest++ = '?';
72                                 /* *src may be valid. Don't eat it.  */
73                                 src--;
74                         }
75
76                         code_high = 0;
77                 } else {
78                         if (code <= 0x007F) {
79                                 *dest++ = code;
80                         } else if (code <= 0x07FF) {
81                                 *dest++ = (code >> 6) | 0xC0;
82                                 *dest++ = (code & 0x3F) | 0x80;
83                         } else if (code >= 0xD800 && code <= 0xDBFF) {
84                                 code_high = code;
85                                 continue;
86                         } else if (code >= 0xDC00 && code <= 0xDFFF) {
87                                 /* Error... */
88                                 *dest++ = '?';
89                         } else if (code < 0x10000) {
90                                 *dest++ = (code >> 12) | 0xE0;
91                                 *dest++ = ((code >> 6) & 0x3F) | 0x80;
92                                 *dest++ = (code & 0x3F) | 0x80;
93                         } else {
94                                 *dest++ = (code >> 18) | 0xF0;
95                                 *dest++ = ((code >> 12) & 0x3F) | 0x80;
96                                 *dest++ = ((code >> 6) & 0x3F) | 0x80;
97                                 *dest++ = (code & 0x3F) | 0x80;
98                         }
99                 }
100         }
101
102         return dest;
103 }
104
105 uint16_t *utf8_to_utf16(uint16_t *dest, const uint8_t *src, size_t size)
106 {
107         while (size--) {
108                 int extension_bytes;
109                 uint32_t code;
110
111                 extension_bytes = 0;
112                 if (*src <= 0x7f) {
113                         code = *src++;
114                         /* Exit on zero byte */
115                         if (!code)
116                                 size = 0;
117                 } else if (*src <= 0xbf) {
118                         /* Illegal code */
119                         code = '?';
120                 } else if (*src <= 0xdf) {
121                         code = *src++ & 0x1f;
122                         extension_bytes = 1;
123                 } else if (*src <= 0xef) {
124                         code = *src++ & 0x0f;
125                         extension_bytes = 2;
126                 } else if (*src <= 0xf7) {
127                         code = *src++ & 0x07;
128                         extension_bytes = 3;
129                 } else {
130                         /* Illegal code */
131                         code = '?';
132                 }
133
134                 for (; extension_bytes && size; --size, --extension_bytes) {
135                         if ((*src & 0xc0) == 0x80) {
136                                 code <<= 6;
137                                 code |= *src++ & 0x3f;
138                         } else {
139                                 /* Illegal code */
140                                 code = '?';
141                                 ++src;
142                                 --size;
143                                 break;
144                         }
145                 }
146
147                 if (code < 0x10000) {
148                         *dest++ = code;
149                 } else {
150                         /*
151                          * Simplified expression for
152                          * (((code - 0x10000) >> 10) & 0x3ff) | 0xd800
153                          */
154                         *dest++ = (code >> 10) + 0xd7c0;
155                         *dest++ = (code & 0x3ff) | 0xdc00;
156                 }
157         }
158         return dest;
159 }