/*
   Convertors for UTF-8

   Name: UTF-8                                                    [RFC2044]
   MIBenum: 106
   Source: RFC 2044
   Alias: 
*/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>
#include "xlocale.h"
#include "xwchar.h"

int _UTF8_init __P((_LocaleInfo *loc, const char *encoding));
static void   _UTF8_stateinit __P((mbstate_t *));
static int    _UTF8_mbsinit   __P((const mbstate_t *));
static size_t _UTF8_mbrtowc __P((wchar_t*, const char*, size_t, mbstate_t *));
static size_t _UTF8_wcrtomb __P((char *, wchar_t, mbstate_t *));

int
_UTF8_init(loc, encoding)
	 _LocaleInfo *loc;
	 const char *encoding;
{
	if (!strcasecmp(encoding, "UTF-8")) {
		loc->mbcurmax  = 6;
		loc->stateinit = _UTF8_stateinit;
		loc->mbsinit   = _UTF8_mbsinit;
		loc->mbrtowc   = _UTF8_mbrtowc;
		loc->wcrtomb   = _UTF8_wcrtomb;
		return 0;
	} else
		return -1;
}

static void
_UTF8_stateinit(state)
	 mbstate_t *state;
{
}

static int
_UTF8_mbsinit(state)
	 const mbstate_t *state;
{
	return 1;
}

static size_t
_UTF8_mbrtowc(pwc, s, len, mbs)
	 wchar_t *pwc;
	 const char *s;
	 size_t len;
	 mbstate_t *mbs;
{
	u_char *in = (u_char *)s;
	int c, c2, c3, c4, c5, c6;
	wchar_t wc;

	if (s == NULL) {			/* initial state */
/*		_UTF8_stateinit(mbs);	*/
		return 0; 				/* Shift_Jis has no states */
	}

	if (len < 1)
		return -1;
	else if (!((c = *in++) & 0x80)) 	/* US ASCII */
		wc = c;
	else if (!(c & 0x40)) {
		goto err;
	} else if (!(c & 0x20)) {
		if (len < 2 || ((c2 = *in++) & 0xc0) != 0x80)
			goto err;
		wc = (c & 0x1f) << 6 | (c2 & 0x3f);
	} else if (!(c & 0x10)) {
		if (len < 3 || ((c2 = *in++) & 0xc0) != 0x80 || 
			((c3 = *in++) & 0xc0) != 0x80)
			goto err;
		wc = (c & 0x1f) << 12 | (c2 & 0x3f) << 6 | (c3 & 0x3f) << 6;
	} else if (!(c & 0x08)) {
		if (len < 4 || ((c2 = *in++) & 0xc0) != 0x80 ||
			((c3 = *in++) & 0xc0) != 0x80 || ((c4 = *in++) & 0xc0) != 0x80)
			goto err;
		wc = (c & 0x1f) << 18 | (c2 & 0x3f) << 12 | (c3 & 0x3f) << 6 |
			(c4 & 0x3f);
	} else if (!(c & 0x04)) {
		if (len < 5 || ((c2 = *in++) & 0xc0) != 0x80 ||
			((c3 = *in++) & 0xc0) != 0x80 || ((c4 = *in++) & 0xc0) != 0x80 ||
			((c5 = *in++) & 0xc0) != 0x80)
			goto err;
		wc = (c & 0x1f) << 24 | (c2 & 0x3f) << 18 | (c3 & 0x3f) << 12 |
			(c4 & 0x3f) << 6 | (c5 & 0x3f);
	} else if (!(c & 0x02)) {
		if (len < 6 || ((c2 = *in++) & 0xc0) != 0x80 ||
			((c3 = *in++) & 0xc0) != 0x80 || ((c4 = *in++) & 0xc0) != 0x80 ||
			((c5 = *in++) & 0xc0) != 0x80 || ((c6 = *in++) & 0xc0) != 0x80)
			goto err;
		wc = (c & 0x1f) << 30 | (c2 & 0x3f) << 24 | (c3 & 0x3f) << 18 |
			(c4 & 0x3f) << 12 | (c5 & 0x3f) << 6 | (c6 & 0x3f);
	} else 
		goto err;

	if (pwc)
		*pwc = wc;
	return ((const char*)in - s);	/* return length */

 err:
	return -1;
}

static size_t
_UTF8_wcrtomb(s, wc, wcs)
	 char *s;
	 wchar_t wc;
	 mbstate_t *wcs;
{
	return 1;
}