libxconv/xconv.c

#include <stdio.h>
#include "charset.h"

static int
add_unicode(char **d, size_t *len, uint16_t u)
{
	if (u < 0x80)
	{
		// 1 byte
		if (*len < 1) return 0;
		(*d)[0] = u;
		(*d)++, (*len)--;
		return 1;
	}
	if (u < 0x800)
	{
		// 2 bytes
		if (*len < 2) return 0;

		(*d)[0] = 0xc0 | ((u >> 6) & 0x1f);
		(*d)[1] = 0x80 | (u & 0x3f);
		*d += 2, *len -= 2;
		return 1;
	}
#if 0
	if (u < 0x10000)
#endif
	{
		// 3 bytes
		if (*len < 3) return 0;

		(*d)[0] = 0xe0 | ((u >> 12) & 0xf);
		(*d)[1] = 0x80 | ((u >> 6) & 0x3f);
		(*d)[2] = 0x80 | (u & 0x3f);
		*d += 3, *len -= 3;
		return 1;
	}
	// 4 byte code point not supported.
	return 0;
}

static int
is_utf_sequence(char *s)
{
	int utf8_len;
	int i;

	if ((*s & 0xe0) == 0xc0)      // U+0080+      110xxxxx
		utf8_len = 1;
	else if ((*s & 0xf0) == 0xe0) // U+0800+      1110xxxx
		utf8_len = 2;
	else if ((*s & 0xf8) == 0xf0) // U+10000+     11110xxx
		utf8_len = 3;
	else return 0;

	for (i = 1; i <= utf8_len; i++)
	{
		// UTF-8 continuation character 10xxxxxx
		if ((s[i] & 0xc0) != 0x80)
			// Invalid UTF-8 sequence
			return 0;
	}
	return 1;
}

size_t
xconv(char *src, char *dst, size_t dstlen)
{
	size_t len = dstlen - 1;
	char *s, *d;
	int changed = 0;
	int i;

	for (s = src, d = dst; *s && len > 0; s++)
	{
		if (!(*s & 0x80))
		{
			*d++ = *s;
			len--;
			continue;
		}

		if (!changed && is_utf_sequence(s))
			return 0;

		changed = 1;
		
		// Check for combined character.
		if ((*s & 0xf0) == 0xc0 && s[1])
		{
			int k = *s & 0xf;
			icc_t *p = iso6937_combined[k];

			for (i = 0; p[i].c; i++)
			{
				if (p[i].c == (s[1] & 0xff))
				{
					add_unicode(&d, &len, p[i].u);
					break;
				}
			}
			s++;
			continue;
		}

		add_unicode(&d, &len, iso6937_map[*s & 0xff]);
	}
	if (!changed)
		return 0;
	
	*d = '\0';
	return dstlen - len - 1;
}