xref: /minix3/lib/libc/citrus/modules/citrus_mskanji.c (revision 84d9c625bfea59e274550651111ae9edfdc40fbd)
1*84d9c625SLionel Sambuc /*	$NetBSD: citrus_mskanji.c,v 1.14 2013/05/28 16:57:56 joerg Exp $	*/
22fe8fb19SBen Gras 
32fe8fb19SBen Gras /*-
42fe8fb19SBen Gras  * Copyright (c)2002 Citrus Project,
52fe8fb19SBen Gras  * All rights reserved.
62fe8fb19SBen Gras  *
72fe8fb19SBen Gras  * Redistribution and use in source and binary forms, with or without
82fe8fb19SBen Gras  * modification, are permitted provided that the following conditions
92fe8fb19SBen Gras  * are met:
102fe8fb19SBen Gras  * 1. Redistributions of source code must retain the above copyright
112fe8fb19SBen Gras  *    notice, this list of conditions and the following disclaimer.
122fe8fb19SBen Gras  * 2. Redistributions in binary form must reproduce the above copyright
132fe8fb19SBen Gras  *    notice, this list of conditions and the following disclaimer in the
142fe8fb19SBen Gras  *    documentation and/or other materials provided with the distribution.
152fe8fb19SBen Gras  *
162fe8fb19SBen Gras  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
172fe8fb19SBen Gras  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
182fe8fb19SBen Gras  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
192fe8fb19SBen Gras  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
202fe8fb19SBen Gras  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
212fe8fb19SBen Gras  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
222fe8fb19SBen Gras  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
232fe8fb19SBen Gras  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
242fe8fb19SBen Gras  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
252fe8fb19SBen Gras  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
262fe8fb19SBen Gras  * SUCH DAMAGE.
272fe8fb19SBen Gras  */
282fe8fb19SBen Gras 
292fe8fb19SBen Gras /*
302fe8fb19SBen Gras  *    ja_JP.SJIS locale table for BSD4.4/rune
312fe8fb19SBen Gras  *    version 1.0
322fe8fb19SBen Gras  *    (C) Sin'ichiro MIYATANI / Phase One, Inc
332fe8fb19SBen Gras  *    May 12, 1995
342fe8fb19SBen Gras  *
352fe8fb19SBen Gras  * Redistribution and use in source and binary forms, with or without
362fe8fb19SBen Gras  * modification, are permitted provided that the following conditions
372fe8fb19SBen Gras  * are met:
382fe8fb19SBen Gras  * 1. Redistributions of source code must retain the above copyright
392fe8fb19SBen Gras  *    notice, this list of conditions and the following disclaimer.
402fe8fb19SBen Gras  * 2. Redistributions in binary form must reproduce the above copyright
412fe8fb19SBen Gras  *    notice, this list of conditions and the following disclaimer in the
422fe8fb19SBen Gras  *    documentation and/or other materials provided with the distribution.
432fe8fb19SBen Gras  * 3. All advertising materials mentioning features or use of this software
442fe8fb19SBen Gras  *    must display the following acknowledgement:
452fe8fb19SBen Gras  *      This product includes software developed by Phase One, Inc.
462fe8fb19SBen Gras  * 4. The name of Phase One, Inc. may be used to endorse or promote products
472fe8fb19SBen Gras  *    derived from this software without specific prior written permission.
482fe8fb19SBen Gras  *
492fe8fb19SBen Gras  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
502fe8fb19SBen Gras  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
512fe8fb19SBen Gras  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
522fe8fb19SBen Gras  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
532fe8fb19SBen Gras  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
542fe8fb19SBen Gras  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
552fe8fb19SBen Gras  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
562fe8fb19SBen Gras  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
572fe8fb19SBen Gras  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
582fe8fb19SBen Gras  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
592fe8fb19SBen Gras  * SUCH DAMAGE.
602fe8fb19SBen Gras  */
612fe8fb19SBen Gras 
622fe8fb19SBen Gras 
632fe8fb19SBen Gras #include <sys/cdefs.h>
642fe8fb19SBen Gras #if defined(LIBC_SCCS) && !defined(lint)
65*84d9c625SLionel Sambuc __RCSID("$NetBSD: citrus_mskanji.c,v 1.14 2013/05/28 16:57:56 joerg Exp $");
662fe8fb19SBen Gras #endif /* LIBC_SCCS and not lint */
672fe8fb19SBen Gras 
682fe8fb19SBen Gras #include <assert.h>
692fe8fb19SBen Gras #include <errno.h>
702fe8fb19SBen Gras #include <string.h>
712fe8fb19SBen Gras #include <stdio.h>
722fe8fb19SBen Gras #include <stdlib.h>
732fe8fb19SBen Gras #include <stddef.h>
742fe8fb19SBen Gras #include <wchar.h>
752fe8fb19SBen Gras #include <sys/types.h>
762fe8fb19SBen Gras #include <limits.h>
772fe8fb19SBen Gras 
782fe8fb19SBen Gras #include "citrus_namespace.h"
792fe8fb19SBen Gras #include "citrus_types.h"
802fe8fb19SBen Gras #include "citrus_bcs.h"
812fe8fb19SBen Gras #include "citrus_module.h"
822fe8fb19SBen Gras #include "citrus_ctype.h"
832fe8fb19SBen Gras #include "citrus_stdenc.h"
842fe8fb19SBen Gras #include "citrus_mskanji.h"
852fe8fb19SBen Gras 
862fe8fb19SBen Gras 
872fe8fb19SBen Gras /* ----------------------------------------------------------------------
882fe8fb19SBen Gras  * private stuffs used by templates
892fe8fb19SBen Gras  */
902fe8fb19SBen Gras 
912fe8fb19SBen Gras typedef struct _MSKanjiState {
922fe8fb19SBen Gras 	char ch[2];
932fe8fb19SBen Gras 	int chlen;
942fe8fb19SBen Gras } _MSKanjiState;
952fe8fb19SBen Gras 
962fe8fb19SBen Gras typedef struct {
972fe8fb19SBen Gras 	int mode;
982fe8fb19SBen Gras #define MODE_JIS2004	1
992fe8fb19SBen Gras } _MSKanjiEncodingInfo;
1002fe8fb19SBen Gras 
1012fe8fb19SBen Gras typedef struct {
1022fe8fb19SBen Gras 	_MSKanjiEncodingInfo	ei;
1032fe8fb19SBen Gras 	struct {
1042fe8fb19SBen Gras 		/* for future multi-locale facility */
1052fe8fb19SBen Gras 		_MSKanjiState	s_mblen;
1062fe8fb19SBen Gras 		_MSKanjiState	s_mbrlen;
1072fe8fb19SBen Gras 		_MSKanjiState	s_mbrtowc;
1082fe8fb19SBen Gras 		_MSKanjiState	s_mbtowc;
1092fe8fb19SBen Gras 		_MSKanjiState	s_mbsrtowcs;
110*84d9c625SLionel Sambuc 		_MSKanjiState	s_mbsnrtowcs;
1112fe8fb19SBen Gras 		_MSKanjiState	s_wcrtomb;
1122fe8fb19SBen Gras 		_MSKanjiState	s_wcsrtombs;
113*84d9c625SLionel Sambuc 		_MSKanjiState	s_wcsnrtombs;
1142fe8fb19SBen Gras 		_MSKanjiState	s_wctomb;
1152fe8fb19SBen Gras 	} states;
1162fe8fb19SBen Gras } _MSKanjiCTypeInfo;
1172fe8fb19SBen Gras 
1182fe8fb19SBen Gras #define _CEI_TO_EI(_cei_)		(&(_cei_)->ei)
1192fe8fb19SBen Gras #define _CEI_TO_STATE(_cei_, _func_)	(_cei_)->states.s_##_func_
1202fe8fb19SBen Gras 
1212fe8fb19SBen Gras #define _FUNCNAME(m)			_citrus_MSKanji_##m
1222fe8fb19SBen Gras #define _ENCODING_INFO			_MSKanjiEncodingInfo
1232fe8fb19SBen Gras #define _CTYPE_INFO			_MSKanjiCTypeInfo
1242fe8fb19SBen Gras #define _ENCODING_STATE			_MSKanjiState
1252fe8fb19SBen Gras #define _ENCODING_MB_CUR_MAX(_ei_)	2
1262fe8fb19SBen Gras #define _ENCODING_IS_STATE_DEPENDENT	0
1272fe8fb19SBen Gras #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
1282fe8fb19SBen Gras 
1292fe8fb19SBen Gras 
1302fe8fb19SBen Gras static int
_mskanji1(int c)1312fe8fb19SBen Gras _mskanji1(int c)
1322fe8fb19SBen Gras {
1332fe8fb19SBen Gras 
1342fe8fb19SBen Gras 	if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
1352fe8fb19SBen Gras 		return 1;
1362fe8fb19SBen Gras 	else
1372fe8fb19SBen Gras 		return 0;
1382fe8fb19SBen Gras }
1392fe8fb19SBen Gras 
1402fe8fb19SBen Gras static int
_mskanji2(int c)1412fe8fb19SBen Gras _mskanji2(int c)
1422fe8fb19SBen Gras {
1432fe8fb19SBen Gras 
1442fe8fb19SBen Gras 	if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
1452fe8fb19SBen Gras 		return 1;
1462fe8fb19SBen Gras 	else
1472fe8fb19SBen Gras 		return 0;
1482fe8fb19SBen Gras }
1492fe8fb19SBen Gras 
1502fe8fb19SBen Gras static __inline void
1512fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,_MSKanjiState * __restrict s)1522fe8fb19SBen Gras _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
1532fe8fb19SBen Gras 			   _MSKanjiState * __restrict s)
1542fe8fb19SBen Gras {
1552fe8fb19SBen Gras 	s->chlen = 0;
1562fe8fb19SBen Gras }
1572fe8fb19SBen Gras 
1582fe8fb19SBen Gras static __inline void
1592fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,void * __restrict pspriv,const _MSKanjiState * __restrict s)1602fe8fb19SBen Gras _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
1612fe8fb19SBen Gras 			   void * __restrict pspriv,
1622fe8fb19SBen Gras 			   const _MSKanjiState * __restrict s)
1632fe8fb19SBen Gras {
1642fe8fb19SBen Gras 	memcpy(pspriv, (const void *)s, sizeof(*s));
1652fe8fb19SBen Gras }
1662fe8fb19SBen Gras 
1672fe8fb19SBen Gras static __inline void
1682fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,_MSKanjiState * __restrict s,const void * __restrict pspriv)1692fe8fb19SBen Gras _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
1702fe8fb19SBen Gras 			     _MSKanjiState * __restrict s,
1712fe8fb19SBen Gras 			     const void * __restrict pspriv)
1722fe8fb19SBen Gras {
1732fe8fb19SBen Gras 	memcpy((void *)s, pspriv, sizeof(*s));
1742fe8fb19SBen Gras }
1752fe8fb19SBen Gras 
1762fe8fb19SBen Gras static int
1772fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,wchar_t * __restrict pwc,const char ** __restrict s,size_t n,_MSKanjiState * __restrict psenc,size_t * __restrict nresult)1782fe8fb19SBen Gras _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
1792fe8fb19SBen Gras 			     wchar_t * __restrict pwc,
1802fe8fb19SBen Gras 			     const char ** __restrict s, size_t n,
1812fe8fb19SBen Gras 			     _MSKanjiState * __restrict psenc,
1822fe8fb19SBen Gras 			     size_t * __restrict nresult)
1832fe8fb19SBen Gras {
1842fe8fb19SBen Gras 	wchar_t wchar;
1852fe8fb19SBen Gras 	int len;
1862fe8fb19SBen Gras 	int chlenbak;
1872fe8fb19SBen Gras 	const char *s0;
1882fe8fb19SBen Gras 
1892fe8fb19SBen Gras 	_DIAGASSERT(nresult != 0);
1902fe8fb19SBen Gras 	_DIAGASSERT(ei != NULL);
1912fe8fb19SBen Gras 	_DIAGASSERT(s != NULL);
1922fe8fb19SBen Gras 	_DIAGASSERT(psenc != NULL);
1932fe8fb19SBen Gras 
1942fe8fb19SBen Gras 	s0 = *s;
1952fe8fb19SBen Gras 
1962fe8fb19SBen Gras 	if (s0 == NULL) {
1972fe8fb19SBen Gras 		_citrus_MSKanji_init_state(ei, psenc);
1982fe8fb19SBen Gras 		*nresult = 0; /* state independent */
1992fe8fb19SBen Gras 		return (0);
2002fe8fb19SBen Gras 	}
2012fe8fb19SBen Gras 
2022fe8fb19SBen Gras 	chlenbak = psenc->chlen;
2032fe8fb19SBen Gras 
2042fe8fb19SBen Gras 	/* make sure we have the first byte in the buffer */
2052fe8fb19SBen Gras 	switch (psenc->chlen) {
2062fe8fb19SBen Gras 	case 0:
2072fe8fb19SBen Gras 		if (n < 1)
2082fe8fb19SBen Gras 			goto restart;
2092fe8fb19SBen Gras 		psenc->ch[0] = *s0++;
2102fe8fb19SBen Gras 		psenc->chlen = 1;
2112fe8fb19SBen Gras 		n--;
2122fe8fb19SBen Gras 		break;
2132fe8fb19SBen Gras 	case 1:
2142fe8fb19SBen Gras 		break;
2152fe8fb19SBen Gras 	default:
2162fe8fb19SBen Gras 		/* illegal state */
2172fe8fb19SBen Gras 		goto encoding_error;
2182fe8fb19SBen Gras 	}
2192fe8fb19SBen Gras 
2202fe8fb19SBen Gras 	len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
2212fe8fb19SBen Gras 	while (psenc->chlen < len) {
2222fe8fb19SBen Gras 		if (n < 1)
2232fe8fb19SBen Gras 			goto restart;
2242fe8fb19SBen Gras 		psenc->ch[psenc->chlen] = *s0++;
2252fe8fb19SBen Gras 		psenc->chlen++;
2262fe8fb19SBen Gras 		n--;
2272fe8fb19SBen Gras 	}
2282fe8fb19SBen Gras 
2292fe8fb19SBen Gras 	*s = s0;
2302fe8fb19SBen Gras 
2312fe8fb19SBen Gras 	switch (len) {
2322fe8fb19SBen Gras 	case 1:
2332fe8fb19SBen Gras 		wchar = psenc->ch[0] & 0xff;
2342fe8fb19SBen Gras 		break;
2352fe8fb19SBen Gras 	case 2:
2362fe8fb19SBen Gras 		if (!_mskanji2(psenc->ch[1] & 0xff))
2372fe8fb19SBen Gras 			goto encoding_error;
2382fe8fb19SBen Gras 		wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
2392fe8fb19SBen Gras 		break;
2402fe8fb19SBen Gras 	default:
2412fe8fb19SBen Gras 		/* illegal state */
2422fe8fb19SBen Gras 		goto encoding_error;
2432fe8fb19SBen Gras 	}
2442fe8fb19SBen Gras 
2452fe8fb19SBen Gras 	psenc->chlen = 0;
2462fe8fb19SBen Gras 
2472fe8fb19SBen Gras 	if (pwc)
2482fe8fb19SBen Gras 		*pwc = wchar;
2492fe8fb19SBen Gras 
2502fe8fb19SBen Gras 	if (!wchar)
2512fe8fb19SBen Gras 		*nresult = 0;
2522fe8fb19SBen Gras 	else
2532fe8fb19SBen Gras 		*nresult = len - chlenbak;
2542fe8fb19SBen Gras 
2552fe8fb19SBen Gras 	return (0);
2562fe8fb19SBen Gras 
2572fe8fb19SBen Gras encoding_error:
2582fe8fb19SBen Gras 	psenc->chlen = 0;
2592fe8fb19SBen Gras 	*nresult = (size_t)-1;
2602fe8fb19SBen Gras 	return (EILSEQ);
2612fe8fb19SBen Gras 
2622fe8fb19SBen Gras restart:
2632fe8fb19SBen Gras 	*nresult = (size_t)-2;
2642fe8fb19SBen Gras 	*s = s0;
2652fe8fb19SBen Gras 	return (0);
2662fe8fb19SBen Gras }
2672fe8fb19SBen Gras 
2682fe8fb19SBen Gras 
2692fe8fb19SBen Gras static int
_citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,char * __restrict s,size_t n,wchar_t wc,_MSKanjiState * __restrict psenc,size_t * __restrict nresult)2702fe8fb19SBen Gras _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
2712fe8fb19SBen Gras 			     char * __restrict s, size_t n, wchar_t wc,
2722fe8fb19SBen Gras 			     _MSKanjiState * __restrict psenc,
2732fe8fb19SBen Gras 			     size_t * __restrict nresult)
2742fe8fb19SBen Gras {
2752fe8fb19SBen Gras 	int ret;
2762fe8fb19SBen Gras 
2772fe8fb19SBen Gras 	_DIAGASSERT(ei != NULL);
2782fe8fb19SBen Gras 	_DIAGASSERT(psenc != NULL);
2792fe8fb19SBen Gras 	_DIAGASSERT(s != NULL);
2802fe8fb19SBen Gras 
2812fe8fb19SBen Gras 	/* check invalid sequence */
2822fe8fb19SBen Gras 	if (wc & ~0xffff) {
2832fe8fb19SBen Gras 		ret = EILSEQ;
2842fe8fb19SBen Gras 		goto err;
2852fe8fb19SBen Gras 	}
2862fe8fb19SBen Gras 
2872fe8fb19SBen Gras 	if (wc & 0xff00) {
2882fe8fb19SBen Gras 		if (n < 2) {
2892fe8fb19SBen Gras 			ret = E2BIG;
2902fe8fb19SBen Gras 			goto err;
2912fe8fb19SBen Gras 		}
2922fe8fb19SBen Gras 
2932fe8fb19SBen Gras 		s[0] = (wc >> 8) & 0xff;
2942fe8fb19SBen Gras 		s[1] = wc & 0xff;
2952fe8fb19SBen Gras 		if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
2962fe8fb19SBen Gras 			ret = EILSEQ;
2972fe8fb19SBen Gras 			goto err;
2982fe8fb19SBen Gras 		}
2992fe8fb19SBen Gras 
3002fe8fb19SBen Gras 		*nresult = 2;
3012fe8fb19SBen Gras 		return 0;
3022fe8fb19SBen Gras 	} else {
3032fe8fb19SBen Gras 		if (n < 1) {
3042fe8fb19SBen Gras 			ret = E2BIG;
3052fe8fb19SBen Gras 			goto err;
3062fe8fb19SBen Gras 		}
3072fe8fb19SBen Gras 
3082fe8fb19SBen Gras 		s[0] = wc & 0xff;
3092fe8fb19SBen Gras 		if (_mskanji1(s[0] & 0xff)) {
3102fe8fb19SBen Gras 			ret = EILSEQ;
3112fe8fb19SBen Gras 			goto err;
3122fe8fb19SBen Gras 		}
3132fe8fb19SBen Gras 
3142fe8fb19SBen Gras 		*nresult = 1;
3152fe8fb19SBen Gras 		return 0;
3162fe8fb19SBen Gras 	}
3172fe8fb19SBen Gras 
3182fe8fb19SBen Gras err:
3192fe8fb19SBen Gras 	*nresult = (size_t)-1;
3202fe8fb19SBen Gras 	return ret;
3212fe8fb19SBen Gras }
3222fe8fb19SBen Gras 
3232fe8fb19SBen Gras 
3242fe8fb19SBen Gras static __inline int
3252fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,_csid_t * __restrict csid,_index_t * __restrict idx,wchar_t wc)3262fe8fb19SBen Gras _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,
3272fe8fb19SBen Gras 			      _csid_t * __restrict csid,
3282fe8fb19SBen Gras 			      _index_t * __restrict idx, wchar_t wc)
3292fe8fb19SBen Gras {
3302fe8fb19SBen Gras 	_index_t row, col;
3312fe8fb19SBen Gras 	int offset;
3322fe8fb19SBen Gras 
3332fe8fb19SBen Gras 	_DIAGASSERT(csid != NULL && idx != NULL);
3342fe8fb19SBen Gras 
3352fe8fb19SBen Gras 	if ((_wc_t)wc < 0x80) {
3362fe8fb19SBen Gras 		/* ISO-646 */
3372fe8fb19SBen Gras 		*csid = 0;
3382fe8fb19SBen Gras 		*idx = (_index_t)wc;
3392fe8fb19SBen Gras 	} else if ((_wc_t)wc < 0x100) {
3402fe8fb19SBen Gras 		/* KANA */
3412fe8fb19SBen Gras 		*csid = 1;
3422fe8fb19SBen Gras 		*idx = (_index_t)wc & 0x7F;
3432fe8fb19SBen Gras 	} else {
3442fe8fb19SBen Gras 		/* Kanji (containing Gaiji zone) */
3452fe8fb19SBen Gras 		/*
3462fe8fb19SBen Gras 		 * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
3472fe8fb19SBen Gras 		 * 0x8140 - 0x817E -> 0x2121 - 0x215F
3482fe8fb19SBen Gras 		 * 0x8180 - 0x819E -> 0x2160 - 0x217E
3492fe8fb19SBen Gras 		 * 0x819F - 0x81FC -> 0x2221 - 0x227E
3502fe8fb19SBen Gras 		 *
3512fe8fb19SBen Gras 		 * 0x8240 - 0x827E -> 0x2321 - 0x235F
3522fe8fb19SBen Gras 		 *  ...
3532fe8fb19SBen Gras 		 * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
3542fe8fb19SBen Gras 		 *
3552fe8fb19SBen Gras 		 * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
3562fe8fb19SBen Gras 		 *  ...
3572fe8fb19SBen Gras 		 * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
3582fe8fb19SBen Gras 		 *
3592fe8fb19SBen Gras 		 * extended Gaiji zone:
3602fe8fb19SBen Gras 		 * 0xF040 - 0xFCFC
3612fe8fb19SBen Gras 		 *
3622fe8fb19SBen Gras 		 * JIS X0213-plane2:
3632fe8fb19SBen Gras 		 * 0xF040 - 0xF09E -> 0x2121 - 0x217E
3642fe8fb19SBen Gras 		 * 0xF140 - 0xF19E -> 0x2321 - 0x237E
3652fe8fb19SBen Gras 		 * ...
3662fe8fb19SBen Gras 		 * 0xF240 - 0xF29E -> 0x2521 - 0x257E
3672fe8fb19SBen Gras 		 *
3682fe8fb19SBen Gras 		 * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
3692fe8fb19SBen Gras 		 * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
3702fe8fb19SBen Gras 		 * ...
3712fe8fb19SBen Gras 		 * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
3722fe8fb19SBen Gras 		 *
3732fe8fb19SBen Gras 		 * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
3742fe8fb19SBen Gras 		 * ...
3752fe8fb19SBen Gras 		 * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
3762fe8fb19SBen Gras 		 */
3772fe8fb19SBen Gras 		row = ((_wc_t)wc >> 8) & 0xFF;
3782fe8fb19SBen Gras 		col = (_wc_t)wc & 0xFF;
3792fe8fb19SBen Gras 		if (!_mskanji1(row) || !_mskanji2(col))
3802fe8fb19SBen Gras 			return EILSEQ;
3812fe8fb19SBen Gras 		if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
3822fe8fb19SBen Gras 			*csid = 2;
3832fe8fb19SBen Gras 			offset = 0x81;
3842fe8fb19SBen Gras 		} else {
3852fe8fb19SBen Gras 			*csid = 3;
3862fe8fb19SBen Gras 			if ((_wc_t)wc <= 0xF49E) {
3872fe8fb19SBen Gras 				offset = (_wc_t)wc >= 0xF29F ||
3882fe8fb19SBen Gras 				  ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
3892fe8fb19SBen Gras 				    ? 0xED : 0xF0;
3902fe8fb19SBen Gras 			} else
3912fe8fb19SBen Gras 				offset = 0xCE;
3922fe8fb19SBen Gras 		}
3932fe8fb19SBen Gras 		row -= offset;
3942fe8fb19SBen Gras 		if (row >= 0x5F)
3952fe8fb19SBen Gras 			row -= 0x40;
3962fe8fb19SBen Gras 		row = row * 2 + 0x21;
3972fe8fb19SBen Gras 		col -= 0x1F;
3982fe8fb19SBen Gras 		if (col >= 0x61)
3992fe8fb19SBen Gras 			col -= 1;
4002fe8fb19SBen Gras 		if (col > 0x7E) {
4012fe8fb19SBen Gras 			row += 1;
4022fe8fb19SBen Gras 			col -= 0x5E;
4032fe8fb19SBen Gras 		}
4042fe8fb19SBen Gras 		*idx = ((_index_t)row << 8) | col;
4052fe8fb19SBen Gras 	}
4062fe8fb19SBen Gras 
4072fe8fb19SBen Gras 	return 0;
4082fe8fb19SBen Gras }
4092fe8fb19SBen Gras 
4102fe8fb19SBen Gras static __inline int
4112fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,wchar_t * __restrict wc,_csid_t csid,_index_t idx)4122fe8fb19SBen Gras _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,
4132fe8fb19SBen Gras 			      wchar_t * __restrict wc,
4142fe8fb19SBen Gras 			      _csid_t csid, _index_t idx)
4152fe8fb19SBen Gras {
4162fe8fb19SBen Gras 	u_int32_t row, col;
4172fe8fb19SBen Gras 	int offset;
4182fe8fb19SBen Gras 
4192fe8fb19SBen Gras 	_DIAGASSERT(wc != NULL);
4202fe8fb19SBen Gras 
4212fe8fb19SBen Gras 	switch (csid) {
4222fe8fb19SBen Gras 	case 0:
4232fe8fb19SBen Gras 		/* ISO-646 */
4242fe8fb19SBen Gras 		if (idx >= 0x80)
4252fe8fb19SBen Gras 			return EILSEQ;
4262fe8fb19SBen Gras 		*wc = (wchar_t)idx;
4272fe8fb19SBen Gras 		break;
4282fe8fb19SBen Gras 	case 1:
4292fe8fb19SBen Gras 		/* kana */
4302fe8fb19SBen Gras 		if (idx >= 0x80)
4312fe8fb19SBen Gras 			return EILSEQ;
4322fe8fb19SBen Gras 		*wc = (wchar_t)idx + 0x80;
4332fe8fb19SBen Gras 		break;
4342fe8fb19SBen Gras 	case 3:
4352fe8fb19SBen Gras 		if ((ei->mode & MODE_JIS2004) == 0)
4362fe8fb19SBen Gras 			return EILSEQ;
4372fe8fb19SBen Gras 	/*FALLTHROUGH*/
4382fe8fb19SBen Gras 	case 2:
4392fe8fb19SBen Gras 		/* kanji */
4402fe8fb19SBen Gras 		row = (idx >> 8);
4412fe8fb19SBen Gras 		if (row < 0x21)
4422fe8fb19SBen Gras 			return EILSEQ;
4432fe8fb19SBen Gras 		if (csid == 3) {
4442fe8fb19SBen Gras 			if (row <= 0x2F)
4452fe8fb19SBen Gras 				offset = (row == 0x22 || row >= 0x26)
4462fe8fb19SBen Gras 				    ? 0xED : 0xF0;
4472fe8fb19SBen Gras 			else if (row >= 0x4D && row <= 0x7E)
4482fe8fb19SBen Gras 				offset = 0xCE;
4492fe8fb19SBen Gras 			else
4502fe8fb19SBen Gras 				return EILSEQ;
4512fe8fb19SBen Gras 		} else {
4522fe8fb19SBen Gras 			if (row > 0x97)
4532fe8fb19SBen Gras 				return EILSEQ;
4542fe8fb19SBen Gras 			offset = (row < 0x5F) ? 0x81 : 0xC1;
4552fe8fb19SBen Gras 		}
4562fe8fb19SBen Gras 		col = idx & 0xFF;
4572fe8fb19SBen Gras 		if (col < 0x21 || col > 0x7E)
4582fe8fb19SBen Gras 			return EILSEQ;
4592fe8fb19SBen Gras 		row -= 0x21; col -= 0x21;
4602fe8fb19SBen Gras 		if ((row & 1) == 0) {
4612fe8fb19SBen Gras 			col += 0x40;
4622fe8fb19SBen Gras 			if (col >= 0x7F)
4632fe8fb19SBen Gras 				col += 1;
4642fe8fb19SBen Gras 		} else
4652fe8fb19SBen Gras 			col += 0x9F;
4662fe8fb19SBen Gras 		row = row / 2 + offset;
4672fe8fb19SBen Gras 		*wc = ((wchar_t)row << 8) | col;
4682fe8fb19SBen Gras 		break;
4692fe8fb19SBen Gras 	default:
4702fe8fb19SBen Gras 		return EILSEQ;
4712fe8fb19SBen Gras 	}
4722fe8fb19SBen Gras 
4732fe8fb19SBen Gras 	return 0;
4742fe8fb19SBen Gras }
4752fe8fb19SBen Gras 
4762fe8fb19SBen Gras static __inline int
4772fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,_MSKanjiState * __restrict psenc,int * __restrict rstate)4782fe8fb19SBen Gras _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
4792fe8fb19SBen Gras 					      _MSKanjiState * __restrict psenc,
4802fe8fb19SBen Gras 					      int * __restrict rstate)
4812fe8fb19SBen Gras {
4822fe8fb19SBen Gras 
4832fe8fb19SBen Gras 	if (psenc->chlen == 0)
4842fe8fb19SBen Gras 		*rstate = _STDENC_SDGEN_INITIAL;
4852fe8fb19SBen Gras 	else
4862fe8fb19SBen Gras 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
4872fe8fb19SBen Gras 
4882fe8fb19SBen Gras 	return 0;
4892fe8fb19SBen Gras }
4902fe8fb19SBen Gras 
4912fe8fb19SBen Gras static int
4922fe8fb19SBen Gras /*ARGSUSED*/
_citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo * __restrict ei,const void * __restrict var,size_t lenvar)4932fe8fb19SBen Gras _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo *  __restrict ei,
4942fe8fb19SBen Gras 				     const void * __restrict var,
4952fe8fb19SBen Gras 				     size_t lenvar)
4962fe8fb19SBen Gras {
4972fe8fb19SBen Gras 	const char *p;
4982fe8fb19SBen Gras 
4992fe8fb19SBen Gras 	_DIAGASSERT(ei != NULL);
5002fe8fb19SBen Gras 
5012fe8fb19SBen Gras 	p = var;
5022fe8fb19SBen Gras #define MATCH(x, act)						\
5032fe8fb19SBen Gras do {								\
5042fe8fb19SBen Gras 	if (lenvar >= (sizeof(#x)-1) &&				\
5052fe8fb19SBen Gras 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
5062fe8fb19SBen Gras 		act;						\
5072fe8fb19SBen Gras 		lenvar -= sizeof(#x)-1;				\
5082fe8fb19SBen Gras 		p += sizeof(#x)-1;				\
5092fe8fb19SBen Gras 	}							\
5102fe8fb19SBen Gras } while (/*CONSTCOND*/0)
5112fe8fb19SBen Gras 	memset((void *)ei, 0, sizeof(*ei));
5122fe8fb19SBen Gras 	while (lenvar > 0) {
5132fe8fb19SBen Gras 		switch (_bcs_toupper(*p)) {
5142fe8fb19SBen Gras 		case 'J':
5152fe8fb19SBen Gras 			MATCH(JIS2004, ei->mode |= MODE_JIS2004);
5162fe8fb19SBen Gras 			break;
5172fe8fb19SBen Gras 		}
5182fe8fb19SBen Gras 		++p;
5192fe8fb19SBen Gras 		--lenvar;
5202fe8fb19SBen Gras 	}
5212fe8fb19SBen Gras 
5222fe8fb19SBen Gras 	return 0;
5232fe8fb19SBen Gras }
5242fe8fb19SBen Gras 
5252fe8fb19SBen Gras static void
_citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo * ei)5262fe8fb19SBen Gras _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
5272fe8fb19SBen Gras {
5282fe8fb19SBen Gras }
5292fe8fb19SBen Gras 
5302fe8fb19SBen Gras /* ----------------------------------------------------------------------
5312fe8fb19SBen Gras  * public interface for ctype
5322fe8fb19SBen Gras  */
5332fe8fb19SBen Gras 
5342fe8fb19SBen Gras _CITRUS_CTYPE_DECLS(MSKanji);
5352fe8fb19SBen Gras _CITRUS_CTYPE_DEF_OPS(MSKanji);
5362fe8fb19SBen Gras 
5372fe8fb19SBen Gras #include "citrus_ctype_template.h"
5382fe8fb19SBen Gras 
5392fe8fb19SBen Gras /* ----------------------------------------------------------------------
5402fe8fb19SBen Gras  * public interface for stdenc
5412fe8fb19SBen Gras  */
5422fe8fb19SBen Gras 
5432fe8fb19SBen Gras _CITRUS_STDENC_DECLS(MSKanji);
5442fe8fb19SBen Gras _CITRUS_STDENC_DEF_OPS(MSKanji);
5452fe8fb19SBen Gras 
5462fe8fb19SBen Gras #include "citrus_stdenc_template.h"
547