Warning: Due to various recent migrations, viewing non-HEAD refs may be broken.

/src/utf8_encode.c (raw)

#include <stdint.h>
#include <stddef.h>
#include "unicode.h"

size_t utf8_encode(char *str, uint32_t ch) {
	size_t len = 0;
	uint8_t first;

	if (ch < 0x80) {
		first = 0;
		len = 1;
	} else if (ch < 0x800) {
		first = 0xc0;
		len = 2;
	} else if (ch < 0x10000) {
		first = 0xe0;
		len = 3;
	} else {
		first = 0xf0;
		len = 4;
	}

	for (size_t i = len - 1; i > 0; --i) {
		str[i] = (ch & 0x3f) | 0x80;
		ch >>= 6;
	}

	str[0] = ch | first;
	return len;
}