Home | History | Annotate | Line # | Download | only in modules
citrus_utf1632.c revision 1.9.6.2
      1  1.9.6.2  tnozaki /*	$NetBSD: citrus_utf1632.c,v 1.9.6.2 2008/06/14 16:01:09 tnozaki Exp $	*/
      2  1.9.6.2  tnozaki 
      3  1.9.6.2  tnozaki /*-
      4  1.9.6.2  tnozaki  * Copyright (c)2003 Citrus Project,
      5  1.9.6.2  tnozaki  * All rights reserved.
      6  1.9.6.2  tnozaki  *
      7  1.9.6.2  tnozaki  * Redistribution and use in source and binary forms, with or without
      8  1.9.6.2  tnozaki  * modification, are permitted provided that the following conditions
      9  1.9.6.2  tnozaki  * are met:
     10  1.9.6.2  tnozaki  * 1. Redistributions of source code must retain the above copyright
     11  1.9.6.2  tnozaki  *    notice, this list of conditions and the following disclaimer.
     12  1.9.6.2  tnozaki  * 2. Redistributions in binary form must reproduce the above copyright
     13  1.9.6.2  tnozaki  *    notice, this list of conditions and the following disclaimer in the
     14  1.9.6.2  tnozaki  *    documentation and/or other materials provided with the distribution.
     15  1.9.6.2  tnozaki  *
     16  1.9.6.2  tnozaki  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     17  1.9.6.2  tnozaki  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     18  1.9.6.2  tnozaki  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     19  1.9.6.2  tnozaki  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     20  1.9.6.2  tnozaki  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     21  1.9.6.2  tnozaki  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     22  1.9.6.2  tnozaki  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     23  1.9.6.2  tnozaki  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     24  1.9.6.2  tnozaki  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     25  1.9.6.2  tnozaki  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     26  1.9.6.2  tnozaki  * SUCH DAMAGE.
     27  1.9.6.2  tnozaki  */
     28  1.9.6.2  tnozaki 
     29  1.9.6.2  tnozaki #include <sys/cdefs.h>
     30  1.9.6.2  tnozaki #if defined(LIBC_SCCS) && !defined(lint)
     31  1.9.6.2  tnozaki __RCSID("$NetBSD: citrus_utf1632.c,v 1.9.6.2 2008/06/14 16:01:09 tnozaki Exp $");
     32  1.9.6.2  tnozaki #endif /* LIBC_SCCS and not lint */
     33  1.9.6.2  tnozaki 
     34  1.9.6.2  tnozaki #include <assert.h>
     35  1.9.6.2  tnozaki #include <errno.h>
     36  1.9.6.2  tnozaki #include <string.h>
     37  1.9.6.2  tnozaki #include <stdio.h>
     38  1.9.6.2  tnozaki #include <stdlib.h>
     39  1.9.6.2  tnozaki #include <stddef.h>
     40  1.9.6.2  tnozaki #include <limits.h>
     41  1.9.6.2  tnozaki #include <wchar.h>
     42  1.9.6.2  tnozaki #include <sys/types.h>
     43  1.9.6.2  tnozaki #include <machine/endian.h>
     44  1.9.6.2  tnozaki 
     45  1.9.6.2  tnozaki #include "citrus_namespace.h"
     46  1.9.6.2  tnozaki #include "citrus_types.h"
     47  1.9.6.2  tnozaki #include "citrus_module.h"
     48  1.9.6.2  tnozaki #include "citrus_stdenc.h"
     49  1.9.6.2  tnozaki #include "citrus_bcs.h"
     50  1.9.6.2  tnozaki 
     51  1.9.6.2  tnozaki #include "citrus_utf1632.h"
     52  1.9.6.2  tnozaki 
     53  1.9.6.2  tnozaki 
     54  1.9.6.2  tnozaki /* ----------------------------------------------------------------------
     55  1.9.6.2  tnozaki  * private stuffs used by templates
     56  1.9.6.2  tnozaki  */
     57  1.9.6.2  tnozaki 
     58  1.9.6.2  tnozaki typedef struct {
     59  1.9.6.2  tnozaki 	u_int8_t		ch[4];
     60  1.9.6.2  tnozaki 	int			chlen;
     61  1.9.6.2  tnozaki 	int			current_endian;
     62  1.9.6.2  tnozaki } _UTF1632State;
     63  1.9.6.2  tnozaki 
     64  1.9.6.2  tnozaki typedef struct {
     65  1.9.6.2  tnozaki 	int		preffered_endian;
     66  1.9.6.2  tnozaki 	unsigned int	cur_max;
     67  1.9.6.2  tnozaki #define _ENDIAN_UNKNOWN	0
     68  1.9.6.2  tnozaki #define _ENDIAN_BIG	1
     69  1.9.6.2  tnozaki #define _ENDIAN_LITTLE	2
     70  1.9.6.2  tnozaki 	u_int32_t	mode;
     71  1.9.6.2  tnozaki #define _MODE_UTF32		0x00000001U
     72  1.9.6.2  tnozaki #define _MODE_FORCE_ENDIAN	0x00000002U
     73  1.9.6.2  tnozaki } _UTF1632EncodingInfo;
     74  1.9.6.2  tnozaki 
     75  1.9.6.2  tnozaki #define _FUNCNAME(m)			_citrus_UTF1632_##m
     76  1.9.6.2  tnozaki #define _ENCODING_INFO			_UTF1632EncodingInfo
     77  1.9.6.2  tnozaki #define _ENCODING_STATE			_UTF1632State
     78  1.9.6.2  tnozaki #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
     79  1.9.6.2  tnozaki #define _ENCODING_IS_STATE_DEPENDENT	0
     80  1.9.6.2  tnozaki #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
     81  1.9.6.2  tnozaki 
     82  1.9.6.2  tnozaki 
     83  1.9.6.2  tnozaki static __inline void
     84  1.9.6.2  tnozaki /*ARGSUSED*/
     85  1.9.6.2  tnozaki _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei, _UTF1632State *s)
     86  1.9.6.2  tnozaki {
     87  1.9.6.2  tnozaki 	memset(s, 0, sizeof(*s));
     88  1.9.6.2  tnozaki }
     89  1.9.6.2  tnozaki 
     90  1.9.6.2  tnozaki static int
     91  1.9.6.2  tnozaki _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
     92  1.9.6.2  tnozaki 			     const char **s, size_t n, _UTF1632State *psenc,
     93  1.9.6.2  tnozaki 			     size_t *nresult)
     94  1.9.6.2  tnozaki {
     95  1.9.6.2  tnozaki 	int chlenbak, endian, needlen;
     96  1.9.6.2  tnozaki 	wchar_t wc;
     97  1.9.6.2  tnozaki 	size_t result;
     98  1.9.6.2  tnozaki 	const char *s0;
     99  1.9.6.2  tnozaki 
    100  1.9.6.2  tnozaki 	_DIAGASSERT(nresult != 0);
    101  1.9.6.2  tnozaki 	_DIAGASSERT(ei != NULL);
    102  1.9.6.2  tnozaki 	_DIAGASSERT(s != NULL);
    103  1.9.6.2  tnozaki 	_DIAGASSERT(psenc != NULL);
    104  1.9.6.2  tnozaki 
    105  1.9.6.2  tnozaki 	s0 = *s;
    106  1.9.6.2  tnozaki 
    107  1.9.6.2  tnozaki 	if (s0 == NULL) {
    108  1.9.6.2  tnozaki 		_citrus_UTF1632_init_state(ei, psenc);
    109  1.9.6.2  tnozaki 		*nresult = 0; /* state independent */
    110  1.9.6.2  tnozaki 		return (0);
    111  1.9.6.2  tnozaki 	}
    112  1.9.6.2  tnozaki 
    113  1.9.6.2  tnozaki 	result = 0;
    114  1.9.6.2  tnozaki 	chlenbak = psenc->chlen;
    115  1.9.6.2  tnozaki 
    116  1.9.6.2  tnozaki refetch:
    117  1.9.6.2  tnozaki 	if ((ei->mode & _MODE_UTF32) != 0 || chlenbak>=2)
    118  1.9.6.2  tnozaki 		needlen = 4;
    119  1.9.6.2  tnozaki 	else
    120  1.9.6.2  tnozaki 		needlen = 2;
    121  1.9.6.2  tnozaki 
    122  1.9.6.2  tnozaki 	while (chlenbak < needlen) {
    123  1.9.6.2  tnozaki 		if (n==0)
    124  1.9.6.2  tnozaki 			goto restart;
    125  1.9.6.2  tnozaki 		psenc->ch[chlenbak++] = *s0++;
    126  1.9.6.2  tnozaki 		n--;
    127  1.9.6.2  tnozaki 		result++;
    128  1.9.6.2  tnozaki 	}
    129  1.9.6.2  tnozaki 
    130  1.9.6.2  tnozaki 	/* judge endian marker */
    131  1.9.6.2  tnozaki 	if ((ei->mode & _MODE_UTF32) == 0) {
    132  1.9.6.2  tnozaki 		/* UTF16 */
    133  1.9.6.2  tnozaki 		if (psenc->ch[0]==0xFE && psenc->ch[1]==0xFF) {
    134  1.9.6.2  tnozaki 			psenc->current_endian = _ENDIAN_BIG;
    135  1.9.6.2  tnozaki 			chlenbak = 0;
    136  1.9.6.2  tnozaki 			goto refetch;
    137  1.9.6.2  tnozaki 		} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE) {
    138  1.9.6.2  tnozaki 			psenc->current_endian = _ENDIAN_LITTLE;
    139  1.9.6.2  tnozaki 			chlenbak = 0;
    140  1.9.6.2  tnozaki 			goto refetch;
    141  1.9.6.2  tnozaki 		}
    142  1.9.6.2  tnozaki 	} else {
    143  1.9.6.2  tnozaki 		/* UTF32 */
    144  1.9.6.2  tnozaki 		if (psenc->ch[0]==0x00 && psenc->ch[1]==0x00 &&
    145  1.9.6.2  tnozaki 		    psenc->ch[2]==0xFE && psenc->ch[3]==0xFF) {
    146  1.9.6.2  tnozaki 			psenc->current_endian = _ENDIAN_BIG;
    147  1.9.6.2  tnozaki 			chlenbak = 0;
    148  1.9.6.2  tnozaki 			goto refetch;
    149  1.9.6.2  tnozaki 		} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE &&
    150  1.9.6.2  tnozaki 			   psenc->ch[2]==0x00 && psenc->ch[3]==0x00) {
    151  1.9.6.2  tnozaki 			psenc->current_endian = _ENDIAN_LITTLE;
    152  1.9.6.2  tnozaki 			chlenbak = 0;
    153  1.9.6.2  tnozaki 			goto refetch;
    154  1.9.6.2  tnozaki 		}
    155  1.9.6.2  tnozaki 	}
    156  1.9.6.2  tnozaki 	if ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
    157  1.9.6.2  tnozaki 	    psenc->current_endian == _ENDIAN_UNKNOWN)
    158  1.9.6.2  tnozaki 		endian = ei->preffered_endian;
    159  1.9.6.2  tnozaki 	else
    160  1.9.6.2  tnozaki 		endian = psenc->current_endian;
    161  1.9.6.2  tnozaki 
    162  1.9.6.2  tnozaki 	/* get wc */
    163  1.9.6.2  tnozaki 	if ((ei->mode & _MODE_UTF32) == 0) {
    164  1.9.6.2  tnozaki 		/* UTF16 */
    165  1.9.6.2  tnozaki 		if (needlen==2) {
    166  1.9.6.2  tnozaki 			switch (endian) {
    167  1.9.6.2  tnozaki 			case _ENDIAN_LITTLE:
    168  1.9.6.2  tnozaki 				wc = (psenc->ch[0] |
    169  1.9.6.2  tnozaki 				      ((wchar_t)psenc->ch[1] << 8));
    170  1.9.6.2  tnozaki 				break;
    171  1.9.6.2  tnozaki 			case _ENDIAN_BIG:
    172  1.9.6.2  tnozaki 				wc = (psenc->ch[1] |
    173  1.9.6.2  tnozaki 				      ((wchar_t)psenc->ch[0] << 8));
    174  1.9.6.2  tnozaki 				break;
    175  1.9.6.2  tnozaki 			default:
    176  1.9.6.2  tnozaki 				goto ilseq;
    177  1.9.6.2  tnozaki 			}
    178  1.9.6.2  tnozaki 			if (wc >= 0xD800 && wc <= 0xDBFF) {
    179  1.9.6.2  tnozaki 				/* surrogate high */
    180  1.9.6.2  tnozaki 				needlen=4;
    181  1.9.6.2  tnozaki 				goto refetch;
    182  1.9.6.2  tnozaki 			}
    183  1.9.6.2  tnozaki 		} else {
    184  1.9.6.2  tnozaki 			/* surrogate low */
    185  1.9.6.2  tnozaki 			wc -= 0xD800; /* wc : surrogate high (see above) */
    186  1.9.6.2  tnozaki 			wc <<= 10;
    187  1.9.6.2  tnozaki 			switch (endian) {
    188  1.9.6.2  tnozaki 			case _ENDIAN_LITTLE:
    189  1.9.6.2  tnozaki 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
    190  1.9.6.2  tnozaki 					goto ilseq;
    191  1.9.6.2  tnozaki 				wc |= psenc->ch[2];
    192  1.9.6.2  tnozaki 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
    193  1.9.6.2  tnozaki 				break;
    194  1.9.6.2  tnozaki 			case _ENDIAN_BIG:
    195  1.9.6.2  tnozaki 				if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
    196  1.9.6.2  tnozaki 					goto ilseq;
    197  1.9.6.2  tnozaki 				wc |= psenc->ch[3];
    198  1.9.6.2  tnozaki 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
    199  1.9.6.2  tnozaki 				break;
    200  1.9.6.2  tnozaki 			default:
    201  1.9.6.2  tnozaki 				goto ilseq;
    202  1.9.6.2  tnozaki 			}
    203  1.9.6.2  tnozaki 			wc += 0x10000;
    204  1.9.6.2  tnozaki 		}
    205  1.9.6.2  tnozaki 	} else {
    206  1.9.6.2  tnozaki 		/* UTF32 */
    207  1.9.6.2  tnozaki 		switch (endian) {
    208  1.9.6.2  tnozaki 		case _ENDIAN_LITTLE:
    209  1.9.6.2  tnozaki 			wc = (psenc->ch[0] |
    210  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[1] << 8) |
    211  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[2] << 16) |
    212  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[3] << 24));
    213  1.9.6.2  tnozaki 			break;
    214  1.9.6.2  tnozaki 		case _ENDIAN_BIG:
    215  1.9.6.2  tnozaki 			wc = (psenc->ch[3] |
    216  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[2] << 8) |
    217  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[1] << 16) |
    218  1.9.6.2  tnozaki 			      ((wchar_t)psenc->ch[0] << 24));
    219  1.9.6.2  tnozaki 			break;
    220  1.9.6.2  tnozaki 		default:
    221  1.9.6.2  tnozaki 			goto ilseq;
    222  1.9.6.2  tnozaki 		}
    223  1.9.6.2  tnozaki 		if (wc >= 0xD800 && wc <= 0xDFFF)
    224  1.9.6.2  tnozaki 			goto ilseq;
    225  1.9.6.2  tnozaki 	}
    226  1.9.6.2  tnozaki 
    227  1.9.6.2  tnozaki 
    228  1.9.6.2  tnozaki 	*pwc = wc;
    229  1.9.6.2  tnozaki 	psenc->chlen = 0;
    230  1.9.6.2  tnozaki 	*nresult = result;
    231  1.9.6.2  tnozaki 	*s = s0;
    232  1.9.6.2  tnozaki 
    233  1.9.6.2  tnozaki 	return (0);
    234  1.9.6.2  tnozaki 
    235  1.9.6.2  tnozaki ilseq:
    236  1.9.6.2  tnozaki 	*nresult = (size_t)-1;
    237  1.9.6.2  tnozaki 	psenc->chlen = 0;
    238  1.9.6.2  tnozaki 	return (EILSEQ);
    239  1.9.6.2  tnozaki 
    240  1.9.6.2  tnozaki restart:
    241  1.9.6.2  tnozaki 	*nresult = (size_t)-2;
    242  1.9.6.2  tnozaki 	psenc->chlen = chlenbak;
    243  1.9.6.2  tnozaki 	*s = s0;
    244  1.9.6.2  tnozaki 	return (0);
    245  1.9.6.2  tnozaki }
    246  1.9.6.2  tnozaki 
    247  1.9.6.2  tnozaki static int
    248  1.9.6.2  tnozaki _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
    249  1.9.6.2  tnozaki 			     wchar_t wc, _UTF1632State *psenc,
    250  1.9.6.2  tnozaki 			     size_t *nresult)
    251  1.9.6.2  tnozaki {
    252  1.9.6.2  tnozaki 	int ret;
    253  1.9.6.2  tnozaki 	wchar_t wc2;
    254  1.9.6.2  tnozaki 	static const char _bom[4] = {
    255  1.9.6.2  tnozaki #if BYTE_ORDER == BIG_ENDIAN
    256  1.9.6.2  tnozaki 	    0x00, 0x00, 0xFE, 0xFF,
    257  1.9.6.2  tnozaki #else
    258  1.9.6.2  tnozaki 	    0xFF, 0xFE, 0x00, 0x00,
    259  1.9.6.2  tnozaki #endif
    260  1.9.6.2  tnozaki 	};
    261  1.9.6.2  tnozaki 	const char *bom = &_bom[0];
    262  1.9.6.2  tnozaki 	size_t cnt;
    263  1.9.6.2  tnozaki 
    264  1.9.6.2  tnozaki 	_DIAGASSERT(ei != NULL);
    265  1.9.6.2  tnozaki 	_DIAGASSERT(nresult != 0);
    266  1.9.6.2  tnozaki 	_DIAGASSERT(s != NULL);
    267  1.9.6.2  tnozaki 
    268  1.9.6.2  tnozaki 	cnt = (size_t)0;
    269  1.9.6.2  tnozaki 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
    270  1.9.6.2  tnozaki 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
    271  1.9.6.2  tnozaki 			if (ei->mode & _MODE_UTF32) {
    272  1.9.6.2  tnozaki 				cnt = 4;
    273  1.9.6.2  tnozaki 			} else {
    274  1.9.6.2  tnozaki 				cnt = 2;
    275  1.9.6.2  tnozaki #if BYTE_ORDER == BIG_ENDIAN
    276  1.9.6.2  tnozaki 				bom += 2;
    277  1.9.6.2  tnozaki #endif
    278  1.9.6.2  tnozaki 			}
    279  1.9.6.2  tnozaki 			if (n < cnt)
    280  1.9.6.2  tnozaki 				goto e2big;
    281  1.9.6.2  tnozaki 			memcpy(s, bom, cnt);
    282  1.9.6.2  tnozaki 			s += cnt, n -= cnt;
    283  1.9.6.2  tnozaki 		}
    284  1.9.6.2  tnozaki 		psenc->current_endian = ei->preffered_endian;
    285  1.9.6.2  tnozaki 	}
    286  1.9.6.2  tnozaki 
    287  1.9.6.2  tnozaki 	wc2 = 0;
    288  1.9.6.2  tnozaki 	if ((ei->mode & _MODE_UTF32)==0) {
    289  1.9.6.2  tnozaki 		/* UTF16 */
    290  1.9.6.2  tnozaki 		if (wc>0xFFFF) {
    291  1.9.6.2  tnozaki 			/* surrogate */
    292  1.9.6.2  tnozaki 			if (wc>0x10FFFF)
    293  1.9.6.2  tnozaki 				goto ilseq;
    294  1.9.6.2  tnozaki 			if (n < 4)
    295  1.9.6.2  tnozaki 				goto e2big;
    296  1.9.6.2  tnozaki 			cnt += 4;
    297  1.9.6.2  tnozaki 			wc -= 0x10000;
    298  1.9.6.2  tnozaki 			wc2 = (wc & 0x3FF) | 0xDC00;
    299  1.9.6.2  tnozaki 			wc = (wc>>10) | 0xD800;
    300  1.9.6.2  tnozaki 		} else {
    301  1.9.6.2  tnozaki 			if (n < 2)
    302  1.9.6.2  tnozaki 				goto e2big;
    303  1.9.6.2  tnozaki 			cnt += 2;
    304  1.9.6.2  tnozaki 		}
    305  1.9.6.2  tnozaki 
    306  1.9.6.2  tnozaki surrogate:
    307  1.9.6.2  tnozaki 		switch (psenc->current_endian) {
    308  1.9.6.2  tnozaki 		case _ENDIAN_BIG:
    309  1.9.6.2  tnozaki 			s[1] = wc;
    310  1.9.6.2  tnozaki 			s[0] = (wc >>= 8);
    311  1.9.6.2  tnozaki 			break;
    312  1.9.6.2  tnozaki 		case _ENDIAN_LITTLE:
    313  1.9.6.2  tnozaki 			s[0] = wc;
    314  1.9.6.2  tnozaki 			s[1] = (wc >>= 8);
    315  1.9.6.2  tnozaki 			break;
    316  1.9.6.2  tnozaki 		}
    317  1.9.6.2  tnozaki 		if (wc2!=0) {
    318  1.9.6.2  tnozaki 			wc = wc2;
    319  1.9.6.2  tnozaki 			wc2 = 0;
    320  1.9.6.2  tnozaki 			s += 2;
    321  1.9.6.2  tnozaki 			goto surrogate;
    322  1.9.6.2  tnozaki 		}
    323  1.9.6.2  tnozaki 	} else {
    324  1.9.6.2  tnozaki 		/* UTF32 */
    325  1.9.6.2  tnozaki 		if (wc >= 0xD800 && wc <= 0xDFFF)
    326  1.9.6.2  tnozaki 			goto ilseq;
    327  1.9.6.2  tnozaki 		if (n < 4)
    328  1.9.6.2  tnozaki 			goto e2big;
    329  1.9.6.2  tnozaki 		cnt += 4;
    330  1.9.6.2  tnozaki 		switch (psenc->current_endian) {
    331  1.9.6.2  tnozaki 		case _ENDIAN_BIG:
    332  1.9.6.2  tnozaki 			s[3] = wc;
    333  1.9.6.2  tnozaki 			s[2] = (wc >>= 8);
    334  1.9.6.2  tnozaki 			s[1] = (wc >>= 8);
    335  1.9.6.2  tnozaki 			s[0] = (wc >>= 8);
    336  1.9.6.2  tnozaki 			break;
    337  1.9.6.2  tnozaki 		case _ENDIAN_LITTLE:
    338  1.9.6.2  tnozaki 			s[0] = wc;
    339  1.9.6.2  tnozaki 			s[1] = (wc >>= 8);
    340  1.9.6.2  tnozaki 			s[2] = (wc >>= 8);
    341  1.9.6.2  tnozaki 			s[3] = (wc >>= 8);
    342  1.9.6.2  tnozaki 			break;
    343  1.9.6.2  tnozaki 		}
    344  1.9.6.2  tnozaki 	}
    345  1.9.6.2  tnozaki 	*nresult = cnt;
    346  1.9.6.2  tnozaki 
    347  1.9.6.2  tnozaki 	return 0;
    348  1.9.6.2  tnozaki 
    349  1.9.6.2  tnozaki ilseq:
    350  1.9.6.2  tnozaki 	*nresult = (size_t)-1;
    351  1.9.6.2  tnozaki 	return EILSEQ;
    352  1.9.6.2  tnozaki e2big:
    353  1.9.6.2  tnozaki 	*nresult = (size_t)-1;
    354  1.9.6.2  tnozaki 	return E2BIG;
    355  1.9.6.2  tnozaki }
    356  1.9.6.2  tnozaki 
    357  1.9.6.2  tnozaki static void
    358  1.9.6.2  tnozaki parse_variable(_UTF1632EncodingInfo * __restrict ei,
    359  1.9.6.2  tnozaki 	       const void * __restrict var, size_t lenvar)
    360  1.9.6.2  tnozaki {
    361  1.9.6.2  tnozaki #define MATCH(x, act)						\
    362  1.9.6.2  tnozaki do {								\
    363  1.9.6.2  tnozaki 	if (lenvar >= (sizeof(#x)-1) &&				\
    364  1.9.6.2  tnozaki 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
    365  1.9.6.2  tnozaki 		act;						\
    366  1.9.6.2  tnozaki 		lenvar -= sizeof(#x)-1;				\
    367  1.9.6.2  tnozaki 		p += sizeof(#x)-1;				\
    368  1.9.6.2  tnozaki 	}							\
    369  1.9.6.2  tnozaki } while (/*CONSTCOND*/0)
    370  1.9.6.2  tnozaki 	const char *p;
    371  1.9.6.2  tnozaki 	p = var;
    372  1.9.6.2  tnozaki 	while (lenvar>0) {
    373  1.9.6.2  tnozaki 		switch (*p) {
    374  1.9.6.2  tnozaki 		case 'B':
    375  1.9.6.2  tnozaki 		case 'b':
    376  1.9.6.2  tnozaki 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
    377  1.9.6.2  tnozaki 			break;
    378  1.9.6.2  tnozaki 		case 'L':
    379  1.9.6.2  tnozaki 		case 'l':
    380  1.9.6.2  tnozaki 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
    381  1.9.6.2  tnozaki 			break;
    382  1.9.6.2  tnozaki 		case 'F':
    383  1.9.6.2  tnozaki 		case 'f':
    384  1.9.6.2  tnozaki 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
    385  1.9.6.2  tnozaki 			break;
    386  1.9.6.2  tnozaki 		case 'U':
    387  1.9.6.2  tnozaki 		case 'u':
    388  1.9.6.2  tnozaki 			MATCH(utf32, ei->mode |= _MODE_UTF32);
    389  1.9.6.2  tnozaki 			break;
    390  1.9.6.2  tnozaki 		}
    391  1.9.6.2  tnozaki 		p++;
    392  1.9.6.2  tnozaki 		lenvar--;
    393  1.9.6.2  tnozaki 	}
    394  1.9.6.2  tnozaki }
    395  1.9.6.2  tnozaki 
    396  1.9.6.2  tnozaki static int
    397  1.9.6.2  tnozaki /*ARGSUSED*/
    398  1.9.6.2  tnozaki _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
    399  1.9.6.2  tnozaki 				     const void * __restrict var,
    400  1.9.6.2  tnozaki 				     size_t lenvar)
    401  1.9.6.2  tnozaki {
    402  1.9.6.2  tnozaki 	_DIAGASSERT(ei != NULL);
    403  1.9.6.2  tnozaki 
    404  1.9.6.2  tnozaki 	memset((void *)ei, 0, sizeof(*ei));
    405  1.9.6.2  tnozaki 
    406  1.9.6.2  tnozaki 	parse_variable(ei, var, lenvar);
    407  1.9.6.2  tnozaki 
    408  1.9.6.2  tnozaki 	if ((ei->mode&_MODE_UTF32)==0)
    409  1.9.6.2  tnozaki 		ei->cur_max = 6; /* endian + surrogate */
    410  1.9.6.2  tnozaki 	else
    411  1.9.6.2  tnozaki 		ei->cur_max = 8; /* endian + normal */
    412  1.9.6.2  tnozaki 
    413  1.9.6.2  tnozaki 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
    414  1.9.6.2  tnozaki #if BYTE_ORDER == BIG_ENDIAN
    415  1.9.6.2  tnozaki 		ei->preffered_endian = _ENDIAN_BIG;
    416  1.9.6.2  tnozaki #else
    417  1.9.6.2  tnozaki 		ei->preffered_endian = _ENDIAN_LITTLE;
    418  1.9.6.2  tnozaki #endif
    419  1.9.6.2  tnozaki 	}
    420  1.9.6.2  tnozaki 
    421  1.9.6.2  tnozaki 	return (0);
    422  1.9.6.2  tnozaki }
    423  1.9.6.2  tnozaki 
    424  1.9.6.2  tnozaki static void
    425  1.9.6.2  tnozaki /*ARGSUSED*/
    426  1.9.6.2  tnozaki _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
    427  1.9.6.2  tnozaki {
    428  1.9.6.2  tnozaki }
    429  1.9.6.2  tnozaki 
    430  1.9.6.2  tnozaki static __inline int
    431  1.9.6.2  tnozaki /*ARGSUSED*/
    432  1.9.6.2  tnozaki _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei,
    433  1.9.6.2  tnozaki 			      _csid_t * __restrict csid,
    434  1.9.6.2  tnozaki 			      _index_t * __restrict idx,
    435  1.9.6.2  tnozaki 			      _wc_t wc)
    436  1.9.6.2  tnozaki {
    437  1.9.6.2  tnozaki 
    438  1.9.6.2  tnozaki 	_DIAGASSERT(csid != NULL && idx != NULL);
    439  1.9.6.2  tnozaki 
    440  1.9.6.2  tnozaki 	*csid = 0;
    441  1.9.6.2  tnozaki 	*idx = (_index_t)wc;
    442  1.9.6.2  tnozaki 
    443  1.9.6.2  tnozaki 	return (0);
    444  1.9.6.2  tnozaki }
    445  1.9.6.2  tnozaki 
    446  1.9.6.2  tnozaki static __inline int
    447  1.9.6.2  tnozaki /*ARGSUSED*/
    448  1.9.6.2  tnozaki _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei,
    449  1.9.6.2  tnozaki 			      _wc_t * __restrict wc,
    450  1.9.6.2  tnozaki 			      _csid_t csid, _index_t idx)
    451  1.9.6.2  tnozaki {
    452  1.9.6.2  tnozaki 
    453  1.9.6.2  tnozaki 	_DIAGASSERT(wc != NULL);
    454  1.9.6.2  tnozaki 
    455  1.9.6.2  tnozaki 	if (csid != 0)
    456  1.9.6.2  tnozaki 		return (EILSEQ);
    457  1.9.6.2  tnozaki 
    458  1.9.6.2  tnozaki 	*wc = (_wc_t)idx;
    459  1.9.6.2  tnozaki 
    460  1.9.6.2  tnozaki 	return (0);
    461  1.9.6.2  tnozaki }
    462  1.9.6.2  tnozaki 
    463  1.9.6.2  tnozaki static __inline int
    464  1.9.6.2  tnozaki /*ARGSUSED*/
    465  1.9.6.2  tnozaki _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei,
    466  1.9.6.2  tnozaki 					      _UTF1632State * __restrict psenc,
    467  1.9.6.2  tnozaki 					      int * __restrict rstate)
    468  1.9.6.2  tnozaki {
    469  1.9.6.2  tnozaki 
    470  1.9.6.2  tnozaki 	if (psenc->chlen == 0)
    471  1.9.6.2  tnozaki 		*rstate = _STDENC_SDGEN_INITIAL;
    472  1.9.6.2  tnozaki 	else
    473  1.9.6.2  tnozaki 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
    474  1.9.6.2  tnozaki 
    475  1.9.6.2  tnozaki 	return 0;
    476  1.9.6.2  tnozaki }
    477  1.9.6.2  tnozaki 
    478  1.9.6.2  tnozaki /* ----------------------------------------------------------------------
    479  1.9.6.2  tnozaki  * public interface for stdenc
    480  1.9.6.2  tnozaki  */
    481  1.9.6.2  tnozaki 
    482  1.9.6.2  tnozaki _CITRUS_STDENC_DECLS(UTF1632);
    483  1.9.6.2  tnozaki _CITRUS_STDENC_DEF_OPS(UTF1632);
    484  1.9.6.2  tnozaki 
    485  1.9.6.2  tnozaki #include "citrus_stdenc_template.h"
    486