]> git.ipfire.org Git - thirdparty/util-linux.git/blob - lib/encode.c
Merge branch 'dmesg/kmsg-file' of https://github.com/t-8ch/util-linux
[thirdparty/util-linux.git] / lib / encode.c
1 /*
2 * Based on code from libblkid,
3 *
4 * Copyright (C) 2008 Kay Sievers <kay.sievers@vrfy.org>
5 * Copyright (C) 2009 Karel Zak <kzak@redhat.com>
6 * Copyright (C) 2020 Pali Rohár <pali.rohar@gmail.com>
7 *
8 * This file may be redistributed under the terms of the
9 * GNU Lesser General Public License.
10 */
11 #include "c.h"
12 #include "encode.h"
13
14 size_t ul_encode_to_utf8(int enc, unsigned char *dest, size_t len,
15 const unsigned char *src, size_t count)
16 {
17 size_t i, j;
18 uint32_t c;
19 uint16_t c2;
20
21 for (j = i = 0; i < count; i++) {
22 if (enc == UL_ENCODE_UTF16LE) {
23 if (i+2 > count)
24 break;
25 c = (src[i+1] << 8) | src[i];
26 i++;
27 } else if (enc == UL_ENCODE_UTF16BE) {
28 if (i+2 > count)
29 break;
30 c = (src[i] << 8) | src[i+1];
31 i++;
32 } else if (enc == UL_ENCODE_LATIN1) {
33 c = src[i];
34 } else {
35 return 0;
36 }
37 if ((enc == UL_ENCODE_UTF16LE || enc == UL_ENCODE_UTF16BE) &&
38 c >= 0xD800 && c <= 0xDBFF && i+2 < count) {
39 if (enc == UL_ENCODE_UTF16LE)
40 c2 = (src[i+2] << 8) | src[i+1];
41 else
42 c2 = (src[i+1] << 8) | src[i+2];
43 if (c2 >= 0xDC00 && c2 <= 0xDFFF) {
44 c = 0x10000 + ((c - 0xD800) << 10) + (c2 - 0xDC00);
45 i += 2;
46 }
47 }
48 if (c == 0) {
49 dest[j] = '\0';
50 break;
51 }
52
53 if (c < 0x80) {
54 if (j+1 >= len)
55 break;
56 dest[j++] = (uint8_t) c;
57 } else if (c < 0x800) {
58 if (j+2 >= len)
59 break;
60 dest[j++] = (uint8_t) (0xc0 | (c >> 6));
61 dest[j++] = (uint8_t) (0x80 | (c & 0x3f));
62 } else if (c < 0x10000) {
63 if (j+3 >= len)
64 break;
65 dest[j++] = (uint8_t) (0xe0 | (c >> 12));
66 dest[j++] = (uint8_t) (0x80 | ((c >> 6) & 0x3f));
67 dest[j++] = (uint8_t) (0x80 | (c & 0x3f));
68 } else {
69 if (j+4 >= len)
70 break;
71 dest[j++] = (uint8_t) (0xf0 | (c >> 18));
72 dest[j++] = (uint8_t) (0x80 | ((c >> 12) & 0x3f));
73 dest[j++] = (uint8_t) (0x80 | ((c >> 6) & 0x3f));
74 dest[j++] = (uint8_t) (0x80 | (c & 0x3f));
75 }
76 }
77 dest[j] = '\0';
78 return j;
79 }