1 // SPDX-License-Identifier: GPL-2.0+
3 * charset conversion utils
5 * Copyright (c) 2017 Rob Clark
12 * utf8/utf16 conversion mostly lifted from grub
15 size_t utf16_strlen(const uint16_t *in)
18 for (i = 0; in[i]; i++);
22 size_t utf16_strnlen(const uint16_t *in, size_t count)
25 for (i = 0; count-- && in[i]; i++);
29 uint16_t *utf16_strcpy(uint16_t *dest, const uint16_t *src)
33 while ((*dest++ = *src++) != '\0')
39 uint16_t *utf16_strdup(const uint16_t *s)
42 if (!s || !(new = malloc((utf16_strlen(s) + 1) * 2)))
48 /* Convert UTF-16 to UTF-8. */
49 uint8_t *utf16_to_utf8(uint8_t *dest, const uint16_t *src, size_t size)
51 uint32_t code_high = 0;
54 uint32_t code = *src++;
57 if (code >= 0xDC00 && code <= 0xDFFF) {
59 code = ((code_high - 0xD800) << 10) + (code - 0xDC00) + 0x10000;
61 *dest++ = (code >> 18) | 0xF0;
62 *dest++ = ((code >> 12) & 0x3F) | 0x80;
63 *dest++ = ((code >> 6) & 0x3F) | 0x80;
64 *dest++ = (code & 0x3F) | 0x80;
68 /* *src may be valid. Don't eat it. */
76 } else if (code <= 0x07FF) {
77 *dest++ = (code >> 6) | 0xC0;
78 *dest++ = (code & 0x3F) | 0x80;
79 } else if (code >= 0xD800 && code <= 0xDBFF) {
82 } else if (code >= 0xDC00 && code <= 0xDFFF) {
85 } else if (code < 0x10000) {
86 *dest++ = (code >> 12) | 0xE0;
87 *dest++ = ((code >> 6) & 0x3F) | 0x80;
88 *dest++ = (code & 0x3F) | 0x80;
90 *dest++ = (code >> 18) | 0xF0;
91 *dest++ = ((code >> 12) & 0x3F) | 0x80;
92 *dest++ = ((code >> 6) & 0x3F) | 0x80;
93 *dest++ = (code & 0x3F) | 0x80;
101 uint16_t *utf8_to_utf16(uint16_t *dest, const uint8_t *src, size_t size)
110 /* Exit on zero byte */
113 } else if (*src <= 0xbf) {
116 } else if (*src <= 0xdf) {
117 code = *src++ & 0x1f;
119 } else if (*src <= 0xef) {
120 code = *src++ & 0x0f;
122 } else if (*src <= 0xf7) {
123 code = *src++ & 0x07;
130 for (; extension_bytes && size; --size, --extension_bytes) {
131 if ((*src & 0xc0) == 0x80) {
133 code |= *src++ & 0x3f;
143 if (code < 0x10000) {
147 * Simplified expression for
148 * (((code - 0x10000) >> 10) & 0x3ff) | 0xd800
150 *dest++ = (code >> 10) + 0xd7c0;
151 *dest++ = (code & 0x3ff) | 0xdc00;