Home | History | Annotate | Line # | Download | only in modules
citrus_mskanji.c revision 1.12 
      1 /*	$NetBSD: citrus_mskanji.c,v 1.12 2007/03/05 16:57:06 tnozaki Exp $	*/
      2 
      3 /*-
      4  * Copyright (c)2002 Citrus Project,
      5  * All rights reserved.
      6  *
      7  * Redistribution and use in source and binary forms, with or without
      8  * modification, are permitted provided that the following conditions
      9  * are met:
     10  * 1. Redistributions of source code must retain the above copyright
     11  *    notice, this list of conditions and the following disclaimer.
     12  * 2. Redistributions in binary form must reproduce the above copyright
     13  *    notice, this list of conditions and the following disclaimer in the
     14  *    documentation and/or other materials provided with the distribution.
     15  *
     16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     26  * SUCH DAMAGE.
     27  */
     28 
     29 /*
     30  *    ja_JP.SJIS locale table for BSD4.4/rune
     31  *    version 1.0
     32  *    (C) Sin'ichiro MIYATANI / Phase One, Inc
     33  *    May 12, 1995
     34  *
     35  * Redistribution and use in source and binary forms, with or without
     36  * modification, are permitted provided that the following conditions
     37  * are met:
     38  * 1. Redistributions of source code must retain the above copyright
     39  *    notice, this list of conditions and the following disclaimer.
     40  * 2. Redistributions in binary form must reproduce the above copyright
     41  *    notice, this list of conditions and the following disclaimer in the
     42  *    documentation and/or other materials provided with the distribution.
     43  * 3. All advertising materials mentioning features or use of this software
     44  *    must display the following acknowledgement:
     45  *      This product includes software developed by Phase One, Inc.
     46  * 4. The name of Phase One, Inc. may be used to endorse or promote products
     47  *    derived from this software without specific prior written permission.
     48  *
     49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     59  * SUCH DAMAGE.
     60  */
     61 
     62 
     63 #include <sys/cdefs.h>
     64 #if defined(LIBC_SCCS) && !defined(lint)
     65 __RCSID("$NetBSD: citrus_mskanji.c,v 1.12 2007/03/05 16:57:06 tnozaki Exp $");
     66 #endif /* LIBC_SCCS and not lint */
     67 
     68 #include <assert.h>
     69 #include <errno.h>
     70 #include <string.h>
     71 #include <stdio.h>
     72 #include <stdlib.h>
     73 #include <stddef.h>
     74 #include <locale.h>
     75 #include <wchar.h>
     76 #include <sys/types.h>
     77 #include <limits.h>
     78 
     79 #include "citrus_namespace.h"
     80 #include "citrus_types.h"
     81 #include "citrus_bcs.h"
     82 #include "citrus_module.h"
     83 #include "citrus_ctype.h"
     84 #include "citrus_stdenc.h"
     85 #include "citrus_mskanji.h"
     86 
     87 
     88 /* ----------------------------------------------------------------------
     89  * private stuffs used by templates
     90  */
     91 
     92 typedef struct _MSKanjiState {
     93 	char ch[2];
     94 	int chlen;
     95 } _MSKanjiState;
     96 
     97 typedef struct {
     98 	int mode;
     99 #define MODE_JIS2004	1
    100 } _MSKanjiEncodingInfo;
    101 
    102 typedef struct {
    103 	_MSKanjiEncodingInfo	ei;
    104 	struct {
    105 		/* for future multi-locale facility */
    106 		_MSKanjiState	s_mblen;
    107 		_MSKanjiState	s_mbrlen;
    108 		_MSKanjiState	s_mbrtowc;
    109 		_MSKanjiState	s_mbtowc;
    110 		_MSKanjiState	s_mbsrtowcs;
    111 		_MSKanjiState	s_wcrtomb;
    112 		_MSKanjiState	s_wcsrtombs;
    113 		_MSKanjiState	s_wctomb;
    114 	} states;
    115 } _MSKanjiCTypeInfo;
    116 
    117 #define _CEI_TO_EI(_cei_)		(&(_cei_)->ei)
    118 #define _CEI_TO_STATE(_cei_, _func_)	(_cei_)->states.s_##_func_
    119 
    120 #define _FUNCNAME(m)			_citrus_MSKanji_##m
    121 #define _ENCODING_INFO			_MSKanjiEncodingInfo
    122 #define _CTYPE_INFO			_MSKanjiCTypeInfo
    123 #define _ENCODING_STATE			_MSKanjiState
    124 #define _ENCODING_MB_CUR_MAX(_ei_)	2
    125 #define _ENCODING_IS_STATE_DEPENDENT	0
    126 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
    127 
    128 
    129 static int
    130 _mskanji1(int c)
    131 {
    132 
    133 	if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
    134 		return 1;
    135 	else
    136 		return 0;
    137 }
    138 
    139 static int
    140 _mskanji2(int c)
    141 {
    142 
    143 	if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
    144 		return 1;
    145 	else
    146 		return 0;
    147 }
    148 
    149 static __inline void
    150 /*ARGSUSED*/
    151 _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
    152 			   _MSKanjiState * __restrict s)
    153 {
    154 	s->chlen = 0;
    155 }
    156 
    157 static __inline void
    158 /*ARGSUSED*/
    159 _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
    160 			   void * __restrict pspriv,
    161 			   const _MSKanjiState * __restrict s)
    162 {
    163 	memcpy(pspriv, (const void *)s, sizeof(*s));
    164 }
    165 
    166 static __inline void
    167 /*ARGSUSED*/
    168 _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
    169 			     _MSKanjiState * __restrict s,
    170 			     const void * __restrict pspriv)
    171 {
    172 	memcpy((void *)s, pspriv, sizeof(*s));
    173 }
    174 
    175 static int
    176 /*ARGSUSED*/
    177 _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
    178 			     wchar_t * __restrict pwc,
    179 			     const char ** __restrict s, size_t n,
    180 			     _MSKanjiState * __restrict psenc,
    181 			     size_t * __restrict nresult)
    182 {
    183 	wchar_t wchar;
    184 	int len;
    185 	int chlenbak;
    186 	const char *s0;
    187 
    188 	_DIAGASSERT(nresult != 0);
    189 	_DIAGASSERT(ei != NULL);
    190 	_DIAGASSERT(s != NULL);
    191 	_DIAGASSERT(psenc != NULL);
    192 
    193 	s0 = *s;
    194 
    195 	if (s0 == NULL) {
    196 		_citrus_MSKanji_init_state(ei, psenc);
    197 		*nresult = 0; /* state independent */
    198 		return (0);
    199 	}
    200 
    201 	chlenbak = psenc->chlen;
    202 
    203 	/* make sure we have the first byte in the buffer */
    204 	switch (psenc->chlen) {
    205 	case 0:
    206 		if (n < 1)
    207 			goto restart;
    208 		psenc->ch[0] = *s0++;
    209 		psenc->chlen = 1;
    210 		n--;
    211 		break;
    212 	case 1:
    213 		break;
    214 	default:
    215 		/* illegal state */
    216 		goto encoding_error;
    217 	}
    218 
    219 	len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
    220 	while (psenc->chlen < len) {
    221 		if (n < 1)
    222 			goto restart;
    223 		psenc->ch[psenc->chlen] = *s0++;
    224 		psenc->chlen++;
    225 		n--;
    226 	}
    227 
    228 	*s = s0;
    229 
    230 	switch (len) {
    231 	case 1:
    232 		wchar = psenc->ch[0] & 0xff;
    233 		break;
    234 	case 2:
    235 		if (!_mskanji2(psenc->ch[1] & 0xff))
    236 			goto encoding_error;
    237 		wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
    238 		break;
    239 	default:
    240 		/* illegal state */
    241 		goto encoding_error;
    242 	}
    243 
    244 	psenc->chlen = 0;
    245 
    246 	if (pwc)
    247 		*pwc = wchar;
    248 
    249 	if (!wchar)
    250 		*nresult = 0;
    251 	else
    252 		*nresult = len - chlenbak;
    253 
    254 	return (0);
    255 
    256 encoding_error:
    257 	psenc->chlen = 0;
    258 	*nresult = (size_t)-1;
    259 	return (EILSEQ);
    260 
    261 restart:
    262 	*nresult = (size_t)-2;
    263 	*s = s0;
    264 	return (0);
    265 }
    266 
    267 
    268 static int
    269 _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
    270 			     char * __restrict s, size_t n, wchar_t wc,
    271 			     _MSKanjiState * __restrict psenc,
    272 			     size_t * __restrict nresult)
    273 {
    274 	int ret;
    275 
    276 	_DIAGASSERT(ei != NULL);
    277 	_DIAGASSERT(psenc != NULL);
    278 	_DIAGASSERT(s != NULL);
    279 
    280 	/* check invalid sequence */
    281 	if (wc & ~0xffff) {
    282 		ret = EILSEQ;
    283 		goto err;
    284 	}
    285 
    286 	if (wc & 0xff00) {
    287 		if (n < 2) {
    288 			ret = E2BIG;
    289 			goto err;
    290 		}
    291 
    292 		s[0] = (wc >> 8) & 0xff;
    293 		s[1] = wc & 0xff;
    294 		if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
    295 			ret = EILSEQ;
    296 			goto err;
    297 		}
    298 
    299 		*nresult = 2;
    300 		return 0;
    301 	} else {
    302 		if (n < 1) {
    303 			ret = E2BIG;
    304 			goto err;
    305 		}
    306 
    307 		s[0] = wc & 0xff;
    308 		if (_mskanji1(s[0] & 0xff)) {
    309 			ret = EILSEQ;
    310 			goto err;
    311 		}
    312 
    313 		*nresult = 1;
    314 		return 0;
    315 	}
    316 
    317 err:
    318 	*nresult = (size_t)-1;
    319 	return ret;
    320 }
    321 
    322 
    323 static __inline int
    324 /*ARGSUSED*/
    325 _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,
    326 			      _csid_t * __restrict csid,
    327 			      _index_t * __restrict idx, wchar_t wc)
    328 {
    329 	_index_t row, col;
    330 	int offset;
    331 
    332 	_DIAGASSERT(csid != NULL && idx != NULL);
    333 
    334 	if ((_wc_t)wc < 0x80) {
    335 		/* ISO-646 */
    336 		*csid = 0;
    337 		*idx = (_index_t)wc;
    338 	} else if ((_wc_t)wc < 0x100) {
    339 		/* KANA */
    340 		*csid = 1;
    341 		*idx = (_index_t)wc & 0x7F;
    342 	} else {
    343 		/* Kanji (containing Gaiji zone) */
    344 		/*
    345 		 * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
    346 		 * 0x8140 - 0x817E -> 0x2121 - 0x215F
    347 		 * 0x8180 - 0x819E -> 0x2160 - 0x217E
    348 		 * 0x819F - 0x81FC -> 0x2221 - 0x227E
    349 		 *
    350 		 * 0x8240 - 0x827E -> 0x2321 - 0x235F
    351 		 *  ...
    352 		 * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
    353 		 *
    354 		 * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
    355 		 *  ...
    356 		 * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
    357 		 *
    358 		 * extended Gaiji zone:
    359 		 * 0xF040 - 0xFCFC
    360 		 *
    361 		 * JIS X0213-plane2:
    362 		 * 0xF040 - 0xF09E -> 0x2121 - 0x217E
    363 		 * 0xF140 - 0xF19E -> 0x2321 - 0x237E
    364 		 * ...
    365 		 * 0xF240 - 0xF29E -> 0x2521 - 0x257E
    366 		 *
    367 		 * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
    368 		 * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
    369 		 * ...
    370 		 * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
    371 		 *
    372 		 * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
    373 		 * ...
    374 		 * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
    375 		 */
    376 		row = ((_wc_t)wc >> 8) & 0xFF;
    377 		col = (_wc_t)wc & 0xFF;
    378 		if (!_mskanji1(row) || !_mskanji2(col))
    379 			return EILSEQ;
    380 		if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
    381 			*csid = 2;
    382 			offset = 0x81;
    383 		} else {
    384 			*csid = 3;
    385 			if ((_wc_t)wc <= 0xF49E) {
    386 				offset = (_wc_t)wc >= 0xF29F ||
    387 				  ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
    388 				    ? 0xED : 0xF0;
    389 			} else
    390 				offset = 0xCE;
    391 		}
    392 		row -= offset;
    393 		if (row >= 0x5F)
    394 			row -= 0x40;
    395 		row = row * 2 + 0x21;
    396 		col -= 0x1F;
    397 		if (col >= 0x61)
    398 			col -= 1;
    399 		if (col > 0x7E) {
    400 			row += 1;
    401 			col -= 0x5E;
    402 		}
    403 		*idx = ((_index_t)row << 8) | col;
    404 	}
    405 
    406 	return 0;
    407 }
    408 
    409 static __inline int
    410 /*ARGSUSED*/
    411 _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,
    412 			      wchar_t * __restrict wc,
    413 			      _csid_t csid, _index_t idx)
    414 {
    415 	u_int32_t row, col;
    416 	int offset;
    417 
    418 	_DIAGASSERT(wc != NULL);
    419 
    420 	switch (csid) {
    421 	case 0:
    422 		/* ISO-646 */
    423 		if (idx >= 0x80)
    424 			return EILSEQ;
    425 		*wc = (wchar_t)idx;
    426 		break;
    427 	case 1:
    428 		/* kana */
    429 		if (idx >= 0x80)
    430 			return EILSEQ;
    431 		*wc = (wchar_t)idx + 0x80;
    432 		break;
    433 	case 3:
    434 		if ((ei->mode & MODE_JIS2004) == 0)
    435 			return EILSEQ;
    436 	/*FALLTHROUGH*/
    437 	case 2:
    438 		/* kanji */
    439 		row = (idx >> 8);
    440 		if (row < 0x21)
    441 			return EILSEQ;
    442 		if (csid == 3) {
    443 			if (row <= 0x2F)
    444 				offset = (row == 0x22 || row >= 0x26)
    445 				    ? 0xED : 0xF0;
    446 			else if (row >= 0x4D && row <= 0x7E)
    447 				offset = 0xCE;
    448 			else
    449 				return EILSEQ;
    450 		} else {
    451 			if (row > 0x97)
    452 				return EILSEQ;
    453 			offset = (row < 0x5F) ? 0x81 : 0xC1;
    454 		}
    455 		col = idx & 0xFF;
    456 		if (col < 0x21 || col > 0x7E)
    457 			return EILSEQ;
    458 		row -= 0x21; col -= 0x21;
    459 		if ((row & 1) == 0) {
    460 			col += 0x40;
    461 			if (col >= 0x7F)
    462 				col += 1;
    463 		} else
    464 			col += 0x9F;
    465 		row = row / 2 + offset;
    466 		*wc = ((wchar_t)row << 8) | col;
    467 		break;
    468 	default:
    469 		return EILSEQ;
    470 	}
    471 
    472 	return 0;
    473 }
    474 
    475 static __inline int
    476 /*ARGSUSED*/
    477 _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
    478 					      _MSKanjiState * __restrict psenc,
    479 					      int * __restrict rstate)
    480 {
    481 
    482 	if (psenc->chlen == 0)
    483 		*rstate = _STDENC_SDGEN_INITIAL;
    484 	else
    485 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
    486 
    487 	return 0;
    488 }
    489 
    490 static int
    491 /*ARGSUSED*/
    492 _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo *  __restrict ei,
    493 				     const void * __restrict var,
    494 				     size_t lenvar)
    495 {
    496 	const char *p;
    497 
    498 	_DIAGASSERT(ei != NULL);
    499 
    500 	p = var;
    501 #define MATCH(x, act)						\
    502 do {								\
    503 	if (lenvar >= (sizeof(#x)-1) &&				\
    504 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
    505 		act;						\
    506 		lenvar -= sizeof(#x)-1;				\
    507 		p += sizeof(#x)-1;				\
    508 	}							\
    509 } while (/*CONSTCOND*/0)
    510 	memset((void *)ei, 0, sizeof(*ei));
    511 	while (lenvar > 0) {
    512 		switch (_bcs_toupper(*p)) {
    513 		case 'J':
    514 			MATCH(JIS2004, ei->mode |= MODE_JIS2004);
    515 			break;
    516 		}
    517 		++p;
    518 		--lenvar;
    519 	}
    520 
    521 	return 0;
    522 }
    523 
    524 static void
    525 _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
    526 {
    527 }
    528 
    529 /* ----------------------------------------------------------------------
    530  * public interface for ctype
    531  */
    532 
    533 _CITRUS_CTYPE_DECLS(MSKanji);
    534 _CITRUS_CTYPE_DEF_OPS(MSKanji);
    535 
    536 #include "citrus_ctype_template.h"
    537 
    538 /* ----------------------------------------------------------------------
    539  * public interface for stdenc
    540  */
    541 
    542 _CITRUS_STDENC_DECLS(MSKanji);
    543 _CITRUS_STDENC_DEF_OPS(MSKanji);
    544 
    545 #include "citrus_stdenc_template.h"
    546