Home | History | Annotate | Line # | Download | only in modules
citrus_mskanji.c revision 1.14.22.5
      1  1.14.22.5  perseant /*	$NetBSD: citrus_mskanji.c,v 1.14.22.5 2018/01/29 22:08:59 perseant Exp $	*/
      2        1.1  tshiozak 
      3        1.1  tshiozak /*-
      4        1.1  tshiozak  * Copyright (c)2002 Citrus Project,
      5        1.1  tshiozak  * All rights reserved.
      6        1.1  tshiozak  *
      7        1.1  tshiozak  * Redistribution and use in source and binary forms, with or without
      8        1.1  tshiozak  * modification, are permitted provided that the following conditions
      9        1.1  tshiozak  * are met:
     10        1.1  tshiozak  * 1. Redistributions of source code must retain the above copyright
     11        1.1  tshiozak  *    notice, this list of conditions and the following disclaimer.
     12        1.1  tshiozak  * 2. Redistributions in binary form must reproduce the above copyright
     13        1.1  tshiozak  *    notice, this list of conditions and the following disclaimer in the
     14        1.1  tshiozak  *    documentation and/or other materials provided with the distribution.
     15        1.1  tshiozak  *
     16        1.1  tshiozak  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     17        1.1  tshiozak  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     18        1.1  tshiozak  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     19        1.1  tshiozak  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     20        1.1  tshiozak  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     21        1.1  tshiozak  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     22        1.1  tshiozak  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     23        1.1  tshiozak  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     24        1.1  tshiozak  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     25        1.1  tshiozak  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     26        1.1  tshiozak  * SUCH DAMAGE.
     27        1.1  tshiozak  */
     28        1.1  tshiozak 
     29        1.1  tshiozak /*
     30        1.1  tshiozak  *    ja_JP.SJIS locale table for BSD4.4/rune
     31        1.1  tshiozak  *    version 1.0
     32        1.1  tshiozak  *    (C) Sin'ichiro MIYATANI / Phase One, Inc
     33        1.1  tshiozak  *    May 12, 1995
     34        1.1  tshiozak  *
     35        1.1  tshiozak  * Redistribution and use in source and binary forms, with or without
     36        1.1  tshiozak  * modification, are permitted provided that the following conditions
     37        1.1  tshiozak  * are met:
     38        1.1  tshiozak  * 1. Redistributions of source code must retain the above copyright
     39        1.1  tshiozak  *    notice, this list of conditions and the following disclaimer.
     40        1.1  tshiozak  * 2. Redistributions in binary form must reproduce the above copyright
     41        1.1  tshiozak  *    notice, this list of conditions and the following disclaimer in the
     42        1.1  tshiozak  *    documentation and/or other materials provided with the distribution.
     43        1.1  tshiozak  * 3. All advertising materials mentioning features or use of this software
     44        1.1  tshiozak  *    must display the following acknowledgement:
     45        1.1  tshiozak  *      This product includes software developed by Phase One, Inc.
     46        1.1  tshiozak  * 4. The name of Phase One, Inc. may be used to endorse or promote products
     47        1.1  tshiozak  *    derived from this software without specific prior written permission.
     48        1.1  tshiozak  *
     49        1.1  tshiozak  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     50        1.1  tshiozak  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     51        1.1  tshiozak  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     52        1.1  tshiozak  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     53        1.1  tshiozak  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     54        1.1  tshiozak  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     55        1.1  tshiozak  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     56        1.1  tshiozak  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     57        1.1  tshiozak  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     58        1.1  tshiozak  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     59        1.1  tshiozak  * SUCH DAMAGE.
     60        1.1  tshiozak  */
     61        1.1  tshiozak 
     62        1.1  tshiozak 
     63        1.1  tshiozak #include <sys/cdefs.h>
     64        1.1  tshiozak #if defined(LIBC_SCCS) && !defined(lint)
     65  1.14.22.5  perseant __RCSID("$NetBSD: citrus_mskanji.c,v 1.14.22.5 2018/01/29 22:08:59 perseant Exp $");
     66        1.1  tshiozak #endif /* LIBC_SCCS and not lint */
     67        1.1  tshiozak 
     68        1.1  tshiozak #include <assert.h>
     69        1.1  tshiozak #include <errno.h>
     70        1.1  tshiozak #include <string.h>
     71        1.1  tshiozak #include <stdio.h>
     72        1.1  tshiozak #include <stdlib.h>
     73        1.1  tshiozak #include <stddef.h>
     74        1.1  tshiozak #include <wchar.h>
     75        1.1  tshiozak #include <sys/types.h>
     76        1.1  tshiozak #include <limits.h>
     77        1.6  tshiozak 
     78        1.6  tshiozak #include "citrus_namespace.h"
     79        1.6  tshiozak #include "citrus_types.h"
     80       1.12   tnozaki #include "citrus_bcs.h"
     81        1.1  tshiozak #include "citrus_module.h"
     82        1.1  tshiozak #include "citrus_ctype.h"
     83        1.6  tshiozak #include "citrus_stdenc.h"
     84        1.1  tshiozak #include "citrus_mskanji.h"
     85        1.1  tshiozak 
     86        1.1  tshiozak 
     87        1.1  tshiozak /* ----------------------------------------------------------------------
     88        1.1  tshiozak  * private stuffs used by templates
     89        1.1  tshiozak  */
     90        1.1  tshiozak 
     91        1.1  tshiozak typedef struct _MSKanjiState {
     92        1.1  tshiozak 	char ch[2];
     93        1.1  tshiozak 	int chlen;
     94        1.1  tshiozak } _MSKanjiState;
     95        1.1  tshiozak 
     96        1.1  tshiozak typedef struct {
     97       1.12   tnozaki 	int mode;
     98       1.12   tnozaki #define MODE_JIS2004	1
     99        1.1  tshiozak } _MSKanjiEncodingInfo;
    100        1.1  tshiozak 
    101        1.1  tshiozak typedef struct {
    102        1.1  tshiozak 	_MSKanjiEncodingInfo	ei;
    103        1.1  tshiozak 	struct {
    104        1.1  tshiozak 		/* for future multi-locale facility */
    105        1.1  tshiozak 		_MSKanjiState	s_mblen;
    106        1.1  tshiozak 		_MSKanjiState	s_mbrlen;
    107        1.1  tshiozak 		_MSKanjiState	s_mbrtowc;
    108        1.1  tshiozak 		_MSKanjiState	s_mbtowc;
    109        1.1  tshiozak 		_MSKanjiState	s_mbsrtowcs;
    110       1.14     joerg 		_MSKanjiState	s_mbsnrtowcs;
    111        1.1  tshiozak 		_MSKanjiState	s_wcrtomb;
    112        1.1  tshiozak 		_MSKanjiState	s_wcsrtombs;
    113       1.14     joerg 		_MSKanjiState	s_wcsnrtombs;
    114        1.1  tshiozak 		_MSKanjiState	s_wctomb;
    115        1.1  tshiozak 	} states;
    116        1.1  tshiozak } _MSKanjiCTypeInfo;
    117        1.1  tshiozak 
    118        1.1  tshiozak #define _CEI_TO_EI(_cei_)		(&(_cei_)->ei)
    119        1.1  tshiozak #define _CEI_TO_STATE(_cei_, _func_)	(_cei_)->states.s_##_func_
    120        1.1  tshiozak 
    121        1.1  tshiozak #define _FUNCNAME(m)			_citrus_MSKanji_##m
    122        1.1  tshiozak #define _ENCODING_INFO			_MSKanjiEncodingInfo
    123        1.1  tshiozak #define _CTYPE_INFO			_MSKanjiCTypeInfo
    124        1.1  tshiozak #define _ENCODING_STATE			_MSKanjiState
    125        1.2      yamt #define _ENCODING_MB_CUR_MAX(_ei_)	2
    126        1.1  tshiozak #define _ENCODING_IS_STATE_DEPENDENT	0
    127        1.4      yamt #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
    128        1.1  tshiozak 
    129  1.14.22.3  perseant #ifdef __STDC_ISO_10646__
    130  1.14.22.4  perseant #include "citrus_mskanji_k2u.h"
    131  1.14.22.4  perseant #include "citrus_mskanji_u2k.h"
    132  1.14.22.2  perseant 
    133  1.14.22.2  perseant static __inline int
    134  1.14.22.2  perseant /*ARGSUSED*/
    135  1.14.22.2  perseant _FUNCNAME(ucs2kt)(_ENCODING_INFO * __restrict ei,
    136  1.14.22.2  perseant 		  wchar_kuten_t * __restrict ktp, wchar_ucs4_t wc)
    137  1.14.22.2  perseant {
    138  1.14.22.2  perseant 	struct unicode2kuten_lookup *uk;
    139  1.14.22.2  perseant 
    140  1.14.22.2  perseant 	_DIAGASSERT(ktp != NULL);
    141  1.14.22.2  perseant 
    142  1.14.22.2  perseant 	/* US-ASCII are not in the list */
    143  1.14.22.2  perseant 	if (wc < 0x80) {
    144  1.14.22.2  perseant 		*ktp = wc;
    145  1.14.22.2  perseant 		return 0;
    146  1.14.22.2  perseant 	}
    147  1.14.22.2  perseant 
    148  1.14.22.4  perseant 	*ktp = citrus_trie_lookup(&__mskanji_u2k_header, wc);
    149  1.14.22.2  perseant 	return 0;
    150  1.14.22.2  perseant }
    151  1.14.22.2  perseant 
    152  1.14.22.2  perseant static __inline int
    153  1.14.22.2  perseant /*ARGSUSED*/
    154  1.14.22.2  perseant _FUNCNAME(kt2ucs)(_ENCODING_INFO * __restrict ei,
    155  1.14.22.2  perseant 		  wchar_ucs4_t * __restrict up, wchar_kuten_t kt)
    156  1.14.22.2  perseant {
    157  1.14.22.2  perseant 	_csid_t csid;
    158  1.14.22.2  perseant 	_index_t idx;
    159  1.14.22.2  perseant 	struct unicode2kuten_lookup *uk, *table;
    160  1.14.22.2  perseant 
    161  1.14.22.2  perseant 	_DIAGASSERT(up != NULL);
    162  1.14.22.2  perseant 
    163  1.14.22.2  perseant 	table = NULL;
    164  1.14.22.2  perseant 
    165  1.14.22.2  perseant 	/* Special cases */
    166  1.14.22.2  perseant 	if (kt == 0x5c) { /* backslash -> Yen sign */
    167  1.14.22.2  perseant 		*up = 0x00A5;
    168  1.14.22.2  perseant 		return 0;
    169  1.14.22.2  perseant 	}
    170  1.14.22.2  perseant 	if (kt == 0x7E) { /* tilde -> overline */
    171  1.14.22.2  perseant 		*up = 0x203E;
    172  1.14.22.2  perseant 		return 0;
    173  1.14.22.2  perseant 	}
    174  1.14.22.2  perseant 	if (kt < 0x80) {
    175  1.14.22.2  perseant 		*up = kt;
    176  1.14.22.2  perseant 		return 0;
    177  1.14.22.2  perseant 	}
    178  1.14.22.2  perseant 
    179  1.14.22.4  perseant 	*up = citrus_trie_lookup(&__mskanji_k2u_header, kt);
    180  1.14.22.2  perseant 	return 0;
    181  1.14.22.2  perseant }
    182  1.14.22.3  perseant #else
    183  1.14.22.3  perseant #include "citrus_u2k_template.h"
    184  1.14.22.3  perseant #endif
    185        1.1  tshiozak 
    186        1.1  tshiozak static int
    187        1.1  tshiozak _mskanji1(int c)
    188        1.1  tshiozak {
    189        1.1  tshiozak 
    190       1.10  tshiozak 	if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
    191        1.1  tshiozak 		return 1;
    192        1.1  tshiozak 	else
    193        1.1  tshiozak 		return 0;
    194        1.1  tshiozak }
    195        1.1  tshiozak 
    196        1.1  tshiozak static int
    197        1.1  tshiozak _mskanji2(int c)
    198        1.1  tshiozak {
    199        1.1  tshiozak 
    200        1.1  tshiozak 	if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
    201        1.1  tshiozak 		return 1;
    202        1.1  tshiozak 	else
    203        1.1  tshiozak 		return 0;
    204        1.1  tshiozak }
    205        1.1  tshiozak 
    206        1.1  tshiozak static __inline void
    207        1.1  tshiozak /*ARGSUSED*/
    208        1.1  tshiozak _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
    209        1.1  tshiozak 			   _MSKanjiState * __restrict s)
    210        1.1  tshiozak {
    211       1.12   tnozaki 	s->chlen = 0;
    212        1.1  tshiozak }
    213        1.1  tshiozak 
    214        1.1  tshiozak static __inline void
    215        1.1  tshiozak /*ARGSUSED*/
    216        1.1  tshiozak _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
    217        1.1  tshiozak 			   void * __restrict pspriv,
    218        1.1  tshiozak 			   const _MSKanjiState * __restrict s)
    219        1.1  tshiozak {
    220        1.1  tshiozak 	memcpy(pspriv, (const void *)s, sizeof(*s));
    221        1.1  tshiozak }
    222        1.1  tshiozak 
    223        1.1  tshiozak static __inline void
    224        1.1  tshiozak /*ARGSUSED*/
    225        1.1  tshiozak _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
    226        1.1  tshiozak 			     _MSKanjiState * __restrict s,
    227        1.1  tshiozak 			     const void * __restrict pspriv)
    228        1.1  tshiozak {
    229        1.1  tshiozak 	memcpy((void *)s, pspriv, sizeof(*s));
    230        1.1  tshiozak }
    231        1.1  tshiozak 
    232        1.1  tshiozak static int
    233        1.1  tshiozak /*ARGSUSED*/
    234        1.1  tshiozak _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
    235  1.14.22.2  perseant 			     wchar_ucs4_t * __restrict pwc,
    236        1.1  tshiozak 			     const char ** __restrict s, size_t n,
    237        1.1  tshiozak 			     _MSKanjiState * __restrict psenc,
    238        1.1  tshiozak 			     size_t * __restrict nresult)
    239        1.1  tshiozak {
    240  1.14.22.2  perseant 	wchar_kuten_t wchar;
    241        1.1  tshiozak 	int len;
    242        1.1  tshiozak 	int chlenbak;
    243        1.1  tshiozak 	const char *s0;
    244        1.1  tshiozak 
    245        1.1  tshiozak 	_DIAGASSERT(nresult != 0);
    246        1.1  tshiozak 	_DIAGASSERT(ei != NULL);
    247        1.1  tshiozak 	_DIAGASSERT(s != NULL);
    248        1.1  tshiozak 	_DIAGASSERT(psenc != NULL);
    249        1.1  tshiozak 
    250        1.1  tshiozak 	s0 = *s;
    251        1.1  tshiozak 
    252        1.1  tshiozak 	if (s0 == NULL) {
    253        1.1  tshiozak 		_citrus_MSKanji_init_state(ei, psenc);
    254        1.1  tshiozak 		*nresult = 0; /* state independent */
    255        1.1  tshiozak 		return (0);
    256        1.1  tshiozak 	}
    257        1.1  tshiozak 
    258        1.1  tshiozak 	chlenbak = psenc->chlen;
    259        1.1  tshiozak 
    260        1.1  tshiozak 	/* make sure we have the first byte in the buffer */
    261        1.1  tshiozak 	switch (psenc->chlen) {
    262        1.1  tshiozak 	case 0:
    263        1.1  tshiozak 		if (n < 1)
    264        1.1  tshiozak 			goto restart;
    265        1.1  tshiozak 		psenc->ch[0] = *s0++;
    266        1.1  tshiozak 		psenc->chlen = 1;
    267        1.1  tshiozak 		n--;
    268        1.1  tshiozak 		break;
    269        1.1  tshiozak 	case 1:
    270        1.1  tshiozak 		break;
    271        1.1  tshiozak 	default:
    272        1.1  tshiozak 		/* illegal state */
    273        1.1  tshiozak 		goto encoding_error;
    274        1.1  tshiozak 	}
    275        1.1  tshiozak 
    276        1.1  tshiozak 	len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
    277        1.1  tshiozak 	while (psenc->chlen < len) {
    278        1.1  tshiozak 		if (n < 1)
    279        1.1  tshiozak 			goto restart;
    280        1.1  tshiozak 		psenc->ch[psenc->chlen] = *s0++;
    281        1.1  tshiozak 		psenc->chlen++;
    282        1.1  tshiozak 		n--;
    283        1.1  tshiozak 	}
    284        1.1  tshiozak 
    285        1.1  tshiozak 	*s = s0;
    286        1.1  tshiozak 
    287        1.1  tshiozak 	switch (len) {
    288        1.1  tshiozak 	case 1:
    289        1.1  tshiozak 		wchar = psenc->ch[0] & 0xff;
    290        1.1  tshiozak 		break;
    291        1.1  tshiozak 	case 2:
    292        1.1  tshiozak 		if (!_mskanji2(psenc->ch[1] & 0xff))
    293        1.1  tshiozak 			goto encoding_error;
    294        1.1  tshiozak 		wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
    295        1.1  tshiozak 		break;
    296        1.1  tshiozak 	default:
    297        1.1  tshiozak 		/* illegal state */
    298        1.1  tshiozak 		goto encoding_error;
    299        1.1  tshiozak 	}
    300        1.1  tshiozak 
    301        1.1  tshiozak 	psenc->chlen = 0;
    302        1.1  tshiozak 
    303        1.1  tshiozak 	if (pwc)
    304  1.14.22.2  perseant 	        _citrus_MSKanji_kt2ucs(ei, pwc, wchar);
    305        1.1  tshiozak 
    306        1.1  tshiozak 	if (!wchar)
    307        1.1  tshiozak 		*nresult = 0;
    308        1.1  tshiozak 	else
    309        1.1  tshiozak 		*nresult = len - chlenbak;
    310        1.1  tshiozak 
    311        1.1  tshiozak 	return (0);
    312        1.1  tshiozak 
    313        1.1  tshiozak encoding_error:
    314        1.1  tshiozak 	psenc->chlen = 0;
    315        1.1  tshiozak 	*nresult = (size_t)-1;
    316        1.1  tshiozak 	return (EILSEQ);
    317        1.1  tshiozak 
    318        1.1  tshiozak restart:
    319        1.1  tshiozak 	*nresult = (size_t)-2;
    320        1.1  tshiozak 	*s = s0;
    321        1.1  tshiozak 	return (0);
    322        1.1  tshiozak }
    323        1.1  tshiozak 
    324        1.1  tshiozak 
    325        1.1  tshiozak static int
    326        1.1  tshiozak _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
    327  1.14.22.2  perseant 			     char * __restrict s, size_t n, wchar_ucs4_t wc,
    328        1.1  tshiozak 			     _MSKanjiState * __restrict psenc,
    329        1.1  tshiozak 			     size_t * __restrict nresult)
    330        1.1  tshiozak {
    331        1.6  tshiozak 	int ret;
    332        1.1  tshiozak 
    333        1.1  tshiozak 	_DIAGASSERT(ei != NULL);
    334        1.1  tshiozak 	_DIAGASSERT(psenc != NULL);
    335        1.1  tshiozak 	_DIAGASSERT(s != NULL);
    336        1.6  tshiozak 
    337  1.14.22.2  perseant 	_citrus_MSKanji_ucs2kt(ei, &wc, wc);
    338  1.14.22.2  perseant 
    339        1.1  tshiozak 	/* check invalid sequence */
    340        1.6  tshiozak 	if (wc & ~0xffff) {
    341        1.6  tshiozak 		ret = EILSEQ;
    342        1.6  tshiozak 		goto err;
    343        1.6  tshiozak 	}
    344        1.1  tshiozak 
    345        1.1  tshiozak 	if (wc & 0xff00) {
    346        1.6  tshiozak 		if (n < 2) {
    347        1.6  tshiozak 			ret = E2BIG;
    348        1.6  tshiozak 			goto err;
    349        1.6  tshiozak 		}
    350        1.1  tshiozak 
    351        1.1  tshiozak 		s[0] = (wc >> 8) & 0xff;
    352        1.1  tshiozak 		s[1] = wc & 0xff;
    353        1.6  tshiozak 		if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
    354        1.6  tshiozak 			ret = EILSEQ;
    355        1.6  tshiozak 			goto err;
    356        1.6  tshiozak 		}
    357        1.1  tshiozak 
    358        1.1  tshiozak 		*nresult = 2;
    359        1.6  tshiozak 		return 0;
    360        1.1  tshiozak 	} else {
    361        1.8      yamt 		if (n < 1) {
    362        1.8      yamt 			ret = E2BIG;
    363        1.8      yamt 			goto err;
    364        1.8      yamt 		}
    365        1.8      yamt 
    366        1.1  tshiozak 		s[0] = wc & 0xff;
    367        1.6  tshiozak 		if (_mskanji1(s[0] & 0xff)) {
    368        1.6  tshiozak 			ret = EILSEQ;
    369        1.6  tshiozak 			goto err;
    370        1.6  tshiozak 		}
    371        1.1  tshiozak 
    372        1.1  tshiozak 		*nresult = 1;
    373        1.6  tshiozak 		return 0;
    374        1.1  tshiozak 	}
    375        1.1  tshiozak 
    376        1.6  tshiozak err:
    377        1.1  tshiozak 	*nresult = (size_t)-1;
    378        1.6  tshiozak 	return ret;
    379        1.1  tshiozak }
    380        1.1  tshiozak 
    381        1.1  tshiozak 
    382  1.14.22.1  perseant static int
    383        1.6  tshiozak /*ARGSUSED*/
    384  1.14.22.1  perseant _citrus_MSKanji_stdenc_wctocs(struct _citrus_stdenc *ce,
    385        1.6  tshiozak 			      _csid_t * __restrict csid,
    386  1.14.22.2  perseant 			      _index_t * __restrict idx, wchar_kuten_t wc)
    387        1.6  tshiozak {
    388  1.14.22.2  perseant 	_MSKanjiEncodingInfo *ei;
    389        1.6  tshiozak 	_index_t row, col;
    390       1.12   tnozaki 	int offset;
    391        1.6  tshiozak 
    392  1.14.22.1  perseant 	_DIAGASSERT(ce != NULL && csid != NULL && idx != NULL);
    393  1.14.22.1  perseant 
    394  1.14.22.1  perseant 	ei = (_ENCODING_INFO *)(ce->ce_closure);
    395  1.14.22.1  perseant 	_DIAGASSERT(ei != NULL);
    396        1.6  tshiozak 
    397        1.6  tshiozak 	if ((_wc_t)wc < 0x80) {
    398        1.6  tshiozak 		/* ISO-646 */
    399        1.6  tshiozak 		*csid = 0;
    400        1.6  tshiozak 		*idx = (_index_t)wc;
    401        1.6  tshiozak 	} else if ((_wc_t)wc < 0x100) {
    402        1.6  tshiozak 		/* KANA */
    403        1.6  tshiozak 		*csid = 1;
    404        1.6  tshiozak 		*idx = (_index_t)wc & 0x7F;
    405       1.12   tnozaki 	} else {
    406        1.6  tshiozak 		/* Kanji (containing Gaiji zone) */
    407        1.6  tshiozak 		/*
    408        1.6  tshiozak 		 * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
    409        1.6  tshiozak 		 * 0x8140 - 0x817E -> 0x2121 - 0x215F
    410        1.6  tshiozak 		 * 0x8180 - 0x819E -> 0x2160 - 0x217E
    411        1.6  tshiozak 		 * 0x819F - 0x81FC -> 0x2221 - 0x227E
    412        1.6  tshiozak 		 *
    413        1.6  tshiozak 		 * 0x8240 - 0x827E -> 0x2321 - 0x235F
    414        1.6  tshiozak 		 *  ...
    415        1.6  tshiozak 		 * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
    416        1.6  tshiozak 		 *
    417        1.6  tshiozak 		 * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
    418        1.6  tshiozak 		 *  ...
    419        1.6  tshiozak 		 * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
    420        1.6  tshiozak 		 *
    421        1.6  tshiozak 		 * extended Gaiji zone:
    422        1.6  tshiozak 		 * 0xF040 - 0xFCFC
    423       1.12   tnozaki 		 *
    424       1.12   tnozaki 		 * JIS X0213-plane2:
    425       1.12   tnozaki 		 * 0xF040 - 0xF09E -> 0x2121 - 0x217E
    426       1.12   tnozaki 		 * 0xF140 - 0xF19E -> 0x2321 - 0x237E
    427       1.12   tnozaki 		 * ...
    428       1.12   tnozaki 		 * 0xF240 - 0xF29E -> 0x2521 - 0x257E
    429       1.12   tnozaki 		 *
    430       1.12   tnozaki 		 * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
    431       1.12   tnozaki 		 * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
    432       1.12   tnozaki 		 * ...
    433       1.12   tnozaki 		 * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
    434       1.12   tnozaki 		 *
    435       1.12   tnozaki 		 * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
    436       1.12   tnozaki 		 * ...
    437       1.12   tnozaki 		 * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
    438        1.6  tshiozak 		 */
    439       1.12   tnozaki 		row = ((_wc_t)wc >> 8) & 0xFF;
    440       1.12   tnozaki 		col = (_wc_t)wc & 0xFF;
    441       1.12   tnozaki 		if (!_mskanji1(row) || !_mskanji2(col))
    442       1.12   tnozaki 			return EILSEQ;
    443       1.12   tnozaki 		if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
    444       1.12   tnozaki 			*csid = 2;
    445       1.12   tnozaki 			offset = 0x81;
    446       1.12   tnozaki 		} else {
    447       1.12   tnozaki 			*csid = 3;
    448       1.12   tnozaki 			if ((_wc_t)wc <= 0xF49E) {
    449       1.12   tnozaki 				offset = (_wc_t)wc >= 0xF29F ||
    450       1.12   tnozaki 				  ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
    451       1.12   tnozaki 				    ? 0xED : 0xF0;
    452       1.12   tnozaki 			} else
    453       1.12   tnozaki 				offset = 0xCE;
    454       1.12   tnozaki 		}
    455       1.12   tnozaki 		row -= offset;
    456        1.6  tshiozak 		if (row >= 0x5F)
    457        1.6  tshiozak 			row -= 0x40;
    458        1.6  tshiozak 		row = row * 2 + 0x21;
    459       1.12   tnozaki 		col -= 0x1F;
    460        1.6  tshiozak 		if (col >= 0x61)
    461        1.6  tshiozak 			col -= 1;
    462        1.6  tshiozak 		if (col > 0x7E) {
    463        1.6  tshiozak 			row += 1;
    464        1.6  tshiozak 			col -= 0x5E;
    465        1.6  tshiozak 		}
    466        1.6  tshiozak 		*idx = ((_index_t)row << 8) | col;
    467       1.12   tnozaki 	}
    468        1.6  tshiozak 
    469        1.6  tshiozak 	return 0;
    470        1.6  tshiozak }
    471        1.6  tshiozak 
    472  1.14.22.1  perseant static int
    473        1.6  tshiozak /*ARGSUSED*/
    474  1.14.22.1  perseant _citrus_MSKanji_stdenc_cstowc(struct _citrus_stdenc *ce,
    475  1.14.22.2  perseant 			      wchar_kuten_t * __restrict wc,
    476        1.6  tshiozak 			      _csid_t csid, _index_t idx)
    477        1.6  tshiozak {
    478        1.6  tshiozak 	u_int32_t row, col;
    479       1.12   tnozaki 	int offset;
    480  1.14.22.1  perseant 	_MSKanjiEncodingInfo *ei;
    481        1.6  tshiozak 
    482  1.14.22.1  perseant 	_DIAGASSERT(wc != NULL && ce != NULL);
    483  1.14.22.1  perseant 
    484  1.14.22.1  perseant 	ei = (_ENCODING_INFO *)(ce->ce_closure);
    485  1.14.22.1  perseant 	_DIAGASSERT(ei != NULL);
    486        1.6  tshiozak 
    487        1.6  tshiozak 	switch (csid) {
    488        1.6  tshiozak 	case 0:
    489        1.6  tshiozak 		/* ISO-646 */
    490        1.6  tshiozak 		if (idx >= 0x80)
    491        1.6  tshiozak 			return EILSEQ;
    492  1.14.22.2  perseant 		*wc = (wchar_kuten_t)idx;
    493        1.6  tshiozak 		break;
    494        1.6  tshiozak 	case 1:
    495        1.6  tshiozak 		/* kana */
    496        1.6  tshiozak 		if (idx >= 0x80)
    497        1.6  tshiozak 			return EILSEQ;
    498  1.14.22.2  perseant 		*wc = (wchar_kuten_t)idx + 0x80;
    499        1.6  tshiozak 		break;
    500       1.12   tnozaki 	case 3:
    501       1.12   tnozaki 		if ((ei->mode & MODE_JIS2004) == 0)
    502       1.12   tnozaki 			return EILSEQ;
    503       1.12   tnozaki 	/*FALLTHROUGH*/
    504        1.6  tshiozak 	case 2:
    505        1.6  tshiozak 		/* kanji */
    506        1.6  tshiozak 		row = (idx >> 8);
    507       1.12   tnozaki 		if (row < 0x21)
    508       1.12   tnozaki 			return EILSEQ;
    509       1.12   tnozaki 		if (csid == 3) {
    510       1.12   tnozaki 			if (row <= 0x2F)
    511       1.12   tnozaki 				offset = (row == 0x22 || row >= 0x26)
    512       1.12   tnozaki 				    ? 0xED : 0xF0;
    513       1.12   tnozaki 			else if (row >= 0x4D && row <= 0x7E)
    514       1.12   tnozaki 				offset = 0xCE;
    515       1.12   tnozaki 			else
    516       1.12   tnozaki 				return EILSEQ;
    517       1.12   tnozaki 		} else {
    518       1.12   tnozaki 			if (row > 0x97)
    519       1.12   tnozaki 				return EILSEQ;
    520       1.12   tnozaki 			offset = (row < 0x5F) ? 0x81 : 0xC1;
    521       1.12   tnozaki 		}
    522       1.12   tnozaki 		col = idx & 0xFF;
    523       1.12   tnozaki 		if (col < 0x21 || col > 0x7E)
    524        1.6  tshiozak 			return EILSEQ;
    525        1.6  tshiozak 		row -= 0x21; col -= 0x21;
    526       1.12   tnozaki 		if ((row & 1) == 0) {
    527        1.6  tshiozak 			col += 0x40;
    528       1.12   tnozaki 			if (col >= 0x7F)
    529        1.6  tshiozak 				col += 1;
    530        1.6  tshiozak 		} else
    531        1.6  tshiozak 			col += 0x9F;
    532       1.12   tnozaki 		row = row / 2 + offset;
    533  1.14.22.2  perseant 		*wc = ((wchar_kuten_t)row << 8) | col;
    534        1.6  tshiozak 		break;
    535        1.6  tshiozak 	default:
    536        1.6  tshiozak 		return EILSEQ;
    537        1.6  tshiozak 	}
    538        1.6  tshiozak 
    539        1.6  tshiozak 	return 0;
    540        1.6  tshiozak }
    541        1.6  tshiozak 
    542       1.11  tshiozak static __inline int
    543       1.11  tshiozak /*ARGSUSED*/
    544       1.11  tshiozak _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
    545       1.11  tshiozak 					      _MSKanjiState * __restrict psenc,
    546       1.11  tshiozak 					      int * __restrict rstate)
    547       1.11  tshiozak {
    548       1.11  tshiozak 
    549       1.11  tshiozak 	if (psenc->chlen == 0)
    550       1.11  tshiozak 		*rstate = _STDENC_SDGEN_INITIAL;
    551       1.11  tshiozak 	else
    552       1.11  tshiozak 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
    553       1.11  tshiozak 
    554       1.11  tshiozak 	return 0;
    555       1.11  tshiozak }
    556       1.11  tshiozak 
    557        1.1  tshiozak static int
    558        1.1  tshiozak /*ARGSUSED*/
    559        1.6  tshiozak _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo *  __restrict ei,
    560        1.6  tshiozak 				     const void * __restrict var,
    561        1.6  tshiozak 				     size_t lenvar)
    562        1.1  tshiozak {
    563       1.12   tnozaki 	const char *p;
    564        1.1  tshiozak 
    565        1.9   tnozaki 	_DIAGASSERT(ei != NULL);
    566        1.1  tshiozak 
    567  1.14.22.5  perseant #ifdef __STDC_ISO_10646__
    568  1.14.22.5  perseant         citrus_trie_init(&__mskanji_u2k_header, __mskanji_u2k_sizes, __mskanji_u2k_data);
    569  1.14.22.5  perseant         citrus_trie_init(&__mskanji_k2u_header, __mskanji_k2u_sizes, __mskanji_k2u_data);
    570  1.14.22.5  perseant #endif
    571  1.14.22.5  perseant 
    572       1.12   tnozaki 	p = var;
    573       1.12   tnozaki #define MATCH(x, act)						\
    574       1.12   tnozaki do {								\
    575       1.12   tnozaki 	if (lenvar >= (sizeof(#x)-1) &&				\
    576       1.12   tnozaki 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
    577       1.12   tnozaki 		act;						\
    578       1.12   tnozaki 		lenvar -= sizeof(#x)-1;				\
    579       1.12   tnozaki 		p += sizeof(#x)-1;				\
    580       1.12   tnozaki 	}							\
    581       1.12   tnozaki } while (/*CONSTCOND*/0)
    582       1.12   tnozaki 	memset((void *)ei, 0, sizeof(*ei));
    583       1.12   tnozaki 	while (lenvar > 0) {
    584       1.12   tnozaki 		switch (_bcs_toupper(*p)) {
    585       1.12   tnozaki 		case 'J':
    586       1.12   tnozaki 			MATCH(JIS2004, ei->mode |= MODE_JIS2004);
    587       1.12   tnozaki 			break;
    588       1.12   tnozaki 		}
    589       1.12   tnozaki 		++p;
    590       1.12   tnozaki 		--lenvar;
    591       1.12   tnozaki 	}
    592       1.12   tnozaki 
    593       1.12   tnozaki 	return 0;
    594        1.1  tshiozak }
    595        1.1  tshiozak 
    596        1.1  tshiozak static void
    597        1.6  tshiozak _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
    598        1.1  tshiozak {
    599        1.1  tshiozak }
    600        1.1  tshiozak 
    601        1.1  tshiozak /* ----------------------------------------------------------------------
    602        1.1  tshiozak  * public interface for ctype
    603        1.1  tshiozak  */
    604        1.1  tshiozak 
    605        1.1  tshiozak _CITRUS_CTYPE_DECLS(MSKanji);
    606        1.1  tshiozak _CITRUS_CTYPE_DEF_OPS(MSKanji);
    607        1.1  tshiozak 
    608        1.1  tshiozak #include "citrus_ctype_template.h"
    609        1.6  tshiozak 
    610        1.6  tshiozak /* ----------------------------------------------------------------------
    611        1.6  tshiozak  * public interface for stdenc
    612        1.6  tshiozak  */
    613        1.6  tshiozak 
    614        1.6  tshiozak _CITRUS_STDENC_DECLS(MSKanji);
    615        1.6  tshiozak _CITRUS_STDENC_DEF_OPS(MSKanji);
    616        1.6  tshiozak 
    617        1.6  tshiozak #include "citrus_stdenc_template.h"
    618