OXYGENHUB
◆ main x16-emulator / src / utf8_encode.h
1.3 KiB · Objective-C · last changed chore(): initial commit 15 days ago source blame history raw
1 
#pragma once
2
3 
#include <stdint.h>
4
5 
/**
6 
 * Encode a code point using UTF-8
7 
 *
8 
 * @author Ondřej Hruška <[email protected]>
9 
 * @license MIT
10 
 *
11 
 * @param out - output buffer (min 5 characters), will be 0-terminated
12 
 * @param utf - code point 0-0x10FFFF
13 
 * @return number of bytes on success, 0 on failure (also produces U+FFFD, which uses 3 bytes)
14 
 */
15 
static int utf8_encode(char *out, uint32_t utf)
16 
{
17 
	if (utf <= 0x7F) {
18 
		// Plain ASCII
19 
		out[0] = (char) utf;
20 
		out[1] = 0;
21 
		return 1;
22 
	}
23 
	else if (utf <= 0x07FF) {
24 
		// 2-byte unicode
25 
		out[0] = (char) (((utf >> 6) & 0x1F) | 0xC0);
26 
		out[1] = (char) (((utf >> 0) & 0x3F) | 0x80);
27 
		out[2] = 0;
28 
		return 2;
29 
	}
30 
	else if (utf <= 0xFFFF) {
31 
		// 3-byte unicode
32 
		out[0] = (char) (((utf >> 12) & 0x0F) | 0xE0);
33 
		out[1] = (char) (((utf >>  6) & 0x3F) | 0x80);
34 
		out[2] = (char) (((utf >>  0) & 0x3F) | 0x80);
35 
		out[3] = 0;
36 
		return 3;
37 
	}
38 
	else if (utf <= 0x10FFFF) {
39 
		// 4-byte unicode
40 
		out[0] = (char) (((utf >> 18) & 0x07) | 0xF0);
41 
		out[1] = (char) (((utf >> 12) & 0x3F) | 0x80);
42 
		out[2] = (char) (((utf >>  6) & 0x3F) | 0x80);
43 
		out[3] = (char) (((utf >>  0) & 0x3F) | 0x80);
44 
		out[4] = 0;
45 
		return 4;
46 
	}
47 
	else {
48 
		// error - use replacement character
49 
		out[0] = (char) 0xEF;
50 
		out[1] = (char) 0xBF;
51 
		out[2] = (char) 0xBD;
52 
		out[3] = 0;
53 
		return 0;
54 
	}
55 
}