Home | History | Annotate | Line # | Download | only in modules
citrus_euctw.c revision 1.4
      1  1.4      yamt /*	$NetBSD: citrus_euctw.c,v 1.4 2002/03/28 10:29:11 yamt 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  * Copyright (c)1999 Citrus Project,
     31  1.1  tshiozak  * All rights reserved.
     32  1.1  tshiozak  *
     33  1.1  tshiozak  * Redistribution and use in source and binary forms, with or without
     34  1.1  tshiozak  * modification, are permitted provided that the following conditions
     35  1.1  tshiozak  * are met:
     36  1.1  tshiozak  * 1. Redistributions of source code must retain the above copyright
     37  1.1  tshiozak  *    notice, this list of conditions and the following disclaimer.
     38  1.1  tshiozak  * 2. Redistributions in binary form must reproduce the above copyright
     39  1.1  tshiozak  *    notice, this list of conditions and the following disclaimer in the
     40  1.1  tshiozak  *    documentation and/or other materials provided with the distribution.
     41  1.1  tshiozak  *
     42  1.1  tshiozak  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     43  1.1  tshiozak  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     44  1.1  tshiozak  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     45  1.1  tshiozak  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     46  1.1  tshiozak  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     47  1.1  tshiozak  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     48  1.1  tshiozak  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     49  1.1  tshiozak  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     50  1.1  tshiozak  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     51  1.1  tshiozak  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     52  1.1  tshiozak  * SUCH DAMAGE.
     53  1.1  tshiozak  *
     54  1.1  tshiozak  *	$Citrus: xpg4dl/FreeBSD/lib/libc/locale/euctw.c,v 1.13 2001/06/21 01:51:44 yamt Exp $
     55  1.1  tshiozak  */
     56  1.1  tshiozak 
     57  1.1  tshiozak #include <sys/cdefs.h>
     58  1.1  tshiozak #if defined(LIBC_SCCS) && !defined(lint)
     59  1.4      yamt __RCSID("$NetBSD: citrus_euctw.c,v 1.4 2002/03/28 10:29:11 yamt Exp $");
     60  1.1  tshiozak #endif /* LIBC_SCCS and not lint */
     61  1.1  tshiozak 
     62  1.1  tshiozak #include <assert.h>
     63  1.1  tshiozak #include <errno.h>
     64  1.1  tshiozak #include <string.h>
     65  1.1  tshiozak #include <stdio.h>
     66  1.1  tshiozak #include <stdlib.h>
     67  1.1  tshiozak #include <stddef.h>
     68  1.1  tshiozak #include <locale.h>
     69  1.1  tshiozak #include <wchar.h>
     70  1.1  tshiozak #include <sys/types.h>
     71  1.1  tshiozak #include <limits.h>
     72  1.1  tshiozak #include "citrus_module.h"
     73  1.1  tshiozak #include "citrus_ctype.h"
     74  1.1  tshiozak #include "citrus_euctw.h"
     75  1.1  tshiozak 
     76  1.1  tshiozak 
     77  1.1  tshiozak /* ----------------------------------------------------------------------
     78  1.1  tshiozak  * private stuffs used by templates
     79  1.1  tshiozak  */
     80  1.1  tshiozak 
     81  1.1  tshiozak typedef struct {
     82  1.1  tshiozak 	char ch[4];
     83  1.1  tshiozak 	int chlen;
     84  1.1  tshiozak } _EUCTWState;
     85  1.1  tshiozak 
     86  1.1  tshiozak typedef struct {
     87  1.1  tshiozak 	int dummy;
     88  1.1  tshiozak } _EUCTWEncodingInfo;
     89  1.1  tshiozak typedef struct {
     90  1.1  tshiozak 	_EUCTWEncodingInfo	ei;
     91  1.1  tshiozak 	struct {
     92  1.1  tshiozak 		/* for future multi-locale facility */
     93  1.1  tshiozak 		_EUCTWState	s_mblen;
     94  1.1  tshiozak 		_EUCTWState	s_mbrlen;
     95  1.1  tshiozak 		_EUCTWState	s_mbrtowc;
     96  1.1  tshiozak 		_EUCTWState	s_mbtowc;
     97  1.1  tshiozak 		_EUCTWState	s_mbsrtowcs;
     98  1.1  tshiozak 		_EUCTWState	s_wcrtomb;
     99  1.1  tshiozak 		_EUCTWState	s_wcsrtombs;
    100  1.1  tshiozak 		_EUCTWState	s_wctomb;
    101  1.1  tshiozak 	} states;
    102  1.1  tshiozak } _EUCTWCTypeInfo;
    103  1.1  tshiozak 
    104  1.1  tshiozak #define	_SS2	0x008e
    105  1.1  tshiozak #define	_SS3	0x008f
    106  1.1  tshiozak 
    107  1.1  tshiozak #define	_TO_EI(_cl_)			((_EUCTWEncodingInfo *)(_cl_))
    108  1.1  tshiozak #define _TO_CEI(_cl_)			((_EUCTWCTypeInfo *)(_cl_))
    109  1.1  tshiozak #define _TO_STATE(_ps_)			((_EUCTWState *)(_ps_))
    110  1.1  tshiozak #define _CEI_TO_EI(_cei_)		(&(_cei_)->ei)
    111  1.1  tshiozak #define _CEI_TO_STATE(_cei_, _func_)	(_cei_)->states.s_##_func_
    112  1.1  tshiozak 
    113  1.1  tshiozak #define _FUNCNAME(m)			_citrus_EUCTW_##m
    114  1.1  tshiozak #define _ENCODING_INFO			_EUCTWEncodingInfo
    115  1.1  tshiozak #define _CTYPE_INFO			_EUCTWCTypeInfo
    116  1.1  tshiozak #define _ENCODING_STATE			_EUCTWState
    117  1.2      yamt #define _ENCODING_MB_CUR_MAX(_ei_)	4
    118  1.1  tshiozak #define _ENCODING_IS_STATE_DEPENDENT	0
    119  1.4      yamt #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
    120  1.1  tshiozak 
    121  1.1  tshiozak static __inline int
    122  1.1  tshiozak _citrus_EUCTW_cs(u_int c)
    123  1.1  tshiozak {
    124  1.1  tshiozak 	c &= 0xff;
    125  1.1  tshiozak 
    126  1.1  tshiozak 	return ((c & 0x80) ? (c == _SS2 ? 2 : 1) : 0);
    127  1.1  tshiozak }
    128  1.1  tshiozak 
    129  1.1  tshiozak static __inline int
    130  1.1  tshiozak _citrus_EUCTW_count(int cs)
    131  1.1  tshiozak {
    132  1.1  tshiozak 	switch (cs) {
    133  1.1  tshiozak 	case 0:
    134  1.1  tshiozak 		return 1;
    135  1.1  tshiozak 	case 1:
    136  1.1  tshiozak 		return 2;
    137  1.1  tshiozak 	case 2:
    138  1.1  tshiozak 		return 4;
    139  1.1  tshiozak 	case 3:
    140  1.1  tshiozak 		abort();
    141  1.1  tshiozak 		/*NOTREACHED*/
    142  1.1  tshiozak 	}
    143  1.1  tshiozak 	return 0;
    144  1.1  tshiozak }
    145  1.1  tshiozak 
    146  1.1  tshiozak static __inline void
    147  1.1  tshiozak /*ARGSUSED*/
    148  1.1  tshiozak _citrus_EUCTW_init_state(_EUCTWEncodingInfo * __restrict ei,
    149  1.1  tshiozak 			 _EUCTWState * __restrict s)
    150  1.1  tshiozak {
    151  1.1  tshiozak 	memset(s, 0, sizeof(*s));
    152  1.1  tshiozak }
    153  1.1  tshiozak 
    154  1.1  tshiozak static __inline void
    155  1.1  tshiozak /*ARGSUSED*/
    156  1.1  tshiozak _citrus_EUCTW_pack_state(_EUCTWEncodingInfo * __restrict ei,
    157  1.1  tshiozak 			 void * __restrict pspriv,
    158  1.1  tshiozak 			 const _EUCTWState * __restrict s)
    159  1.1  tshiozak {
    160  1.1  tshiozak 	memcpy(pspriv, (const void *)s, sizeof(*s));
    161  1.1  tshiozak }
    162  1.1  tshiozak 
    163  1.1  tshiozak static __inline void
    164  1.1  tshiozak /*ARGSUSED*/
    165  1.1  tshiozak _citrus_EUCTW_unpack_state(_EUCTWEncodingInfo * __restrict ei,
    166  1.1  tshiozak 			   _EUCTWState * __restrict s,
    167  1.1  tshiozak 			   const void * __restrict pspriv)
    168  1.1  tshiozak {
    169  1.1  tshiozak 	memcpy((void *)s, pspriv, sizeof(*s));
    170  1.1  tshiozak }
    171  1.1  tshiozak 
    172  1.1  tshiozak static int
    173  1.1  tshiozak /*ARGSUSED*/
    174  1.1  tshiozak _citrus_EUCTW_stdencoding_init(_EUCTWEncodingInfo * __restrict ei,
    175  1.1  tshiozak 			       const void * __restrict var, size_t lenvar)
    176  1.1  tshiozak {
    177  1.1  tshiozak 
    178  1.1  tshiozak 	_DIAGASSERT(ei != NULL);
    179  1.1  tshiozak 
    180  1.1  tshiozak 	memset((void *)ei, 0, sizeof(*ei));
    181  1.1  tshiozak 
    182  1.1  tshiozak 	return 0;
    183  1.1  tshiozak }
    184  1.1  tshiozak 
    185  1.1  tshiozak static void
    186  1.1  tshiozak /*ARGSUSED*/
    187  1.1  tshiozak _citrus_EUCTW_stdencoding_uninit(_EUCTWEncodingInfo *ei)
    188  1.1  tshiozak {
    189  1.1  tshiozak }
    190  1.1  tshiozak 
    191  1.1  tshiozak static int
    192  1.1  tshiozak _citrus_EUCTW_mbrtowc_priv(_EUCTWEncodingInfo * __restrict ei,
    193  1.1  tshiozak 			   wchar_t * __restrict pwc,
    194  1.1  tshiozak 			   const char ** __restrict s,
    195  1.1  tshiozak 			   size_t n, _EUCTWState * __restrict psenc,
    196  1.1  tshiozak 			   size_t * __restrict nresult)
    197  1.1  tshiozak {
    198  1.1  tshiozak 	wchar_t wchar;
    199  1.1  tshiozak 	int c, cs;
    200  1.1  tshiozak 	int chlenbak;
    201  1.1  tshiozak 	const char *s0;
    202  1.1  tshiozak 
    203  1.1  tshiozak 	_DIAGASSERT(nresult != 0);
    204  1.1  tshiozak 	_DIAGASSERT(ei != NULL);
    205  1.1  tshiozak 	_DIAGASSERT(psenc != NULL);
    206  1.1  tshiozak 	_DIAGASSERT(s != NULL);
    207  1.1  tshiozak 
    208  1.1  tshiozak 	s0 = *s;
    209  1.1  tshiozak 
    210  1.1  tshiozak 	if (s0 == NULL) {
    211  1.1  tshiozak 		_citrus_EUCTW_init_state(ei, psenc);
    212  1.1  tshiozak 		*nresult = 0; /* state independent */
    213  1.1  tshiozak 		return (0);
    214  1.1  tshiozak 	}
    215  1.1  tshiozak 
    216  1.1  tshiozak 	/* make sure we have the first byte in the buffer */
    217  1.1  tshiozak 	switch (psenc->chlen) {
    218  1.1  tshiozak 	case 0:
    219  1.1  tshiozak 		if (n < 1)
    220  1.1  tshiozak 			goto restart;
    221  1.1  tshiozak 		psenc->ch[0] = *s0++;
    222  1.1  tshiozak 		psenc->chlen = 1;
    223  1.1  tshiozak 		n--;
    224  1.1  tshiozak 		break;
    225  1.1  tshiozak 	case 1:
    226  1.1  tshiozak 	case 2:
    227  1.1  tshiozak 		break;
    228  1.1  tshiozak 	default:
    229  1.1  tshiozak 		/* illgeal state */
    230  1.1  tshiozak 		goto ilseq;
    231  1.1  tshiozak 	}
    232  1.1  tshiozak 
    233  1.1  tshiozak 	c = _citrus_EUCTW_count(cs = _citrus_EUCTW_cs(psenc->ch[0] & 0xff));
    234  1.1  tshiozak 	if (c == 0)
    235  1.1  tshiozak 		goto ilseq;
    236  1.1  tshiozak 	while (psenc->chlen < c) {
    237  1.1  tshiozak 		if (n < 1)
    238  1.1  tshiozak 			goto ilseq;
    239  1.1  tshiozak 		psenc->ch[psenc->chlen] = *s0++;
    240  1.1  tshiozak 		psenc->chlen++;
    241  1.1  tshiozak 		n--;
    242  1.1  tshiozak 	}
    243  1.1  tshiozak 
    244  1.1  tshiozak 	wchar = 0;
    245  1.1  tshiozak 	switch (cs) {
    246  1.1  tshiozak 	case 0:
    247  1.1  tshiozak 		if (psenc->ch[0] & 0x80)
    248  1.1  tshiozak 			goto ilseq;
    249  1.1  tshiozak 		wchar = psenc->ch[0] & 0xff;
    250  1.1  tshiozak 		break;
    251  1.1  tshiozak 	case 1:
    252  1.1  tshiozak 		if (!(psenc->ch[0] & 0x80) || !(psenc->ch[1] & 0x80))
    253  1.1  tshiozak 			goto ilseq;
    254  1.1  tshiozak 		wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
    255  1.1  tshiozak 		wchar |= 'G' << 24;
    256  1.1  tshiozak 		break;
    257  1.1  tshiozak 	case 2:
    258  1.1  tshiozak 		if ((u_char)psenc->ch[1] < 0xa1 || 0xa7 < (u_char)psenc->ch[1])
    259  1.1  tshiozak 			goto ilseq;
    260  1.1  tshiozak 		if (!(psenc->ch[2] & 0x80) || !(psenc->ch[3] & 0x80))
    261  1.1  tshiozak 			goto ilseq;
    262  1.1  tshiozak 		wchar = ((psenc->ch[2] & 0xff) << 8) | (psenc->ch[3] & 0xff);
    263  1.1  tshiozak 		wchar |= ('G' + psenc->ch[1] - 0xa1) << 24;
    264  1.1  tshiozak 		break;
    265  1.1  tshiozak 	default:
    266  1.1  tshiozak 		goto ilseq;
    267  1.1  tshiozak 	}
    268  1.1  tshiozak 
    269  1.1  tshiozak 	*s = s0;
    270  1.1  tshiozak 	psenc->chlen = 0;
    271  1.1  tshiozak 
    272  1.1  tshiozak 	if (pwc)
    273  1.1  tshiozak 		*pwc = wchar;
    274  1.1  tshiozak 
    275  1.1  tshiozak 	if (!wchar)
    276  1.1  tshiozak 		*nresult = 0;
    277  1.1  tshiozak 	else
    278  1.1  tshiozak 		*nresult = c - chlenbak;
    279  1.1  tshiozak 
    280  1.1  tshiozak 	return (0);
    281  1.1  tshiozak 
    282  1.1  tshiozak ilseq:
    283  1.1  tshiozak 	psenc->chlen = 0;
    284  1.1  tshiozak 	*nresult = (size_t)-1;
    285  1.1  tshiozak 	return (EILSEQ);
    286  1.1  tshiozak 
    287  1.1  tshiozak restart:
    288  1.1  tshiozak 	*s = s0;
    289  1.1  tshiozak 	*nresult = (size_t)-1;
    290  1.1  tshiozak 	return (0);
    291  1.1  tshiozak }
    292  1.1  tshiozak 
    293  1.1  tshiozak static int
    294  1.1  tshiozak _citrus_EUCTW_wcrtomb_priv(_EUCTWEncodingInfo * __restrict ei,
    295  1.1  tshiozak 			   char * __restrict s, size_t n, wchar_t wc,
    296  1.1  tshiozak 			   _EUCTWState * __restrict psenc,
    297  1.1  tshiozak 			   size_t * __restrict nresult)
    298  1.1  tshiozak {
    299  1.1  tshiozak 	wchar_t cs = wc & 0x7f000080;
    300  1.1  tshiozak 	wchar_t v;
    301  1.1  tshiozak 	int i, len, clen;
    302  1.1  tshiozak 
    303  1.1  tshiozak 	_DIAGASSERT(ei != NULL);
    304  1.1  tshiozak 	_DIAGASSERT(nresult != 0);
    305  1.1  tshiozak 	_DIAGASSERT(s != NULL);
    306  1.1  tshiozak 
    307  1.1  tshiozak 	clen = 1;
    308  1.1  tshiozak 	if (wc & 0x00007f00)
    309  1.1  tshiozak 		clen = 2;
    310  1.1  tshiozak 	if ((wc & 0x007f0000) && !(wc & 0x00800000))
    311  1.1  tshiozak 		clen = 3;
    312  1.1  tshiozak 
    313  1.1  tshiozak 	if (clen == 1 && cs == 0x00000000) {
    314  1.1  tshiozak 		/* ASCII */
    315  1.1  tshiozak 		len = 1;
    316  1.1  tshiozak 		if (n < len)
    317  1.1  tshiozak 			goto ilseq;
    318  1.1  tshiozak 		v = wc & 0x0000007f;
    319  1.1  tshiozak 	} else if (clen == 2 && cs == ('G' << 24)) {
    320  1.1  tshiozak 		/* CNS-11643-1 */
    321  1.1  tshiozak 		len = 2;
    322  1.1  tshiozak 		if (n < len)
    323  1.1  tshiozak 			goto ilseq;
    324  1.1  tshiozak 		v = wc & 0x00007f7f;
    325  1.1  tshiozak 		v |= 0x00008080;
    326  1.1  tshiozak 	} else if (clen == 2 && 'H' <= (cs >> 24) && (cs >> 24) <= 'M') {
    327  1.1  tshiozak 		/* CNS-11643-[2-7] */
    328  1.1  tshiozak 		len = 4;
    329  1.1  tshiozak 		if (n < len)
    330  1.1  tshiozak 			goto ilseq;
    331  1.1  tshiozak 		*s++ = _SS2;
    332  1.1  tshiozak 		*s++ = (cs >> 24) - 'H' + 0xa2;
    333  1.1  tshiozak 		v = wc & 0x00007f7f;
    334  1.1  tshiozak 		v |= 0x00008080;
    335  1.1  tshiozak 	} else
    336  1.1  tshiozak 		goto ilseq;
    337  1.1  tshiozak 
    338  1.1  tshiozak 	i = clen;
    339  1.1  tshiozak 	while (i-- > 0)
    340  1.1  tshiozak 		*s++ = (v >> (i << 3)) & 0xff;
    341  1.1  tshiozak 
    342  1.1  tshiozak 	*nresult = len;
    343  1.1  tshiozak 	return (0);
    344  1.1  tshiozak 
    345  1.1  tshiozak ilseq:
    346  1.1  tshiozak 	*nresult = (size_t)-1;
    347  1.1  tshiozak 	return (EILSEQ);
    348  1.1  tshiozak }
    349  1.1  tshiozak 
    350  1.1  tshiozak /* ----------------------------------------------------------------------
    351  1.1  tshiozak  * public interface for ctype
    352  1.1  tshiozak  */
    353  1.1  tshiozak 
    354  1.1  tshiozak _CITRUS_CTYPE_DECLS(EUCTW);
    355  1.1  tshiozak _CITRUS_CTYPE_DEF_OPS(EUCTW);
    356  1.1  tshiozak 
    357  1.1  tshiozak #include "citrus_ctype_template.h"
    358