Home | History | Annotate | Line # | Download | only in modules
citrus_utf1632.c revision 1.6 
      1 /*	$NetBSD: citrus_utf1632.c,v 1.6 2006/03/19 01:24:09 christos Exp $	*/
      2 
      3 /*-
      4  * Copyright (c)2003 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 #include <sys/cdefs.h>
     30 #if defined(LIBC_SCCS) && !defined(lint)
     31 __RCSID("$NetBSD: citrus_utf1632.c,v 1.6 2006/03/19 01:24:09 christos Exp $");
     32 #endif /* LIBC_SCCS and not lint */
     33 
     34 #include <assert.h>
     35 #include <errno.h>
     36 #include <string.h>
     37 #include <stdio.h>
     38 #include <stdlib.h>
     39 #include <stddef.h>
     40 #include <locale.h>
     41 #include <limits.h>
     42 #include <wchar.h>
     43 #include <sys/types.h>
     44 #include <machine/endian.h>
     45 
     46 #include "citrus_namespace.h"
     47 #include "citrus_types.h"
     48 #include "citrus_module.h"
     49 #include "citrus_stdenc.h"
     50 #include "citrus_bcs.h"
     51 
     52 #include "citrus_utf1632.h"
     53 
     54 
     55 /* ----------------------------------------------------------------------
     56  * private stuffs used by templates
     57  */
     58 
     59 typedef struct {
     60 	u_int8_t		ch[4];
     61 	int			chlen;
     62 	int			current_endian;
     63 } _UTF1632State;
     64 
     65 typedef struct {
     66 	int		preffered_endian;
     67 	unsigned int	cur_max;
     68 #define _ENDIAN_UNKNOWN	0
     69 #define _ENDIAN_BIG	1
     70 #define _ENDIAN_LITTLE	2
     71 	u_int32_t	mode;
     72 #define _MODE_UTF32		0x00000001U
     73 #define _MODE_FORCE_ENDIAN	0x00000002U
     74 } _UTF1632EncodingInfo;
     75 
     76 #define _FUNCNAME(m)			_citrus_UTF1632_##m
     77 #define _ENCODING_INFO			_UTF1632EncodingInfo
     78 #define _ENCODING_STATE			_UTF1632State
     79 #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
     80 #define _ENCODING_IS_STATE_DEPENDENT	0
     81 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
     82 
     83 
     84 static __inline void
     85 /*ARGSUSED*/
     86 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei, _UTF1632State *s)
     87 {
     88 	memset(s, 0, sizeof(*s));
     89 }
     90 
     91 static int
     92 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
     93 			     const char **s, size_t n, _UTF1632State *psenc,
     94 			     size_t *nresult)
     95 {
     96 	int chlenbak, endian, needlen;
     97 	wchar_t wc;
     98 	size_t result;
     99 	const char *s0;
    100 
    101 	_DIAGASSERT(nresult != 0);
    102 	_DIAGASSERT(ei != NULL);
    103 	_DIAGASSERT(s != NULL);
    104 	_DIAGASSERT(psenc != NULL);
    105 
    106 	s0 = *s;
    107 
    108 	if (s0 == NULL) {
    109 		_citrus_UTF1632_init_state(ei, psenc);
    110 		*nresult = 0; /* state independent */
    111 		return (0);
    112 	}
    113 
    114 	result = 0;
    115 	chlenbak = psenc->chlen;
    116 
    117 refetch:
    118 	if ((ei->mode & _MODE_UTF32) != 0 || chlenbak>=2)
    119 		needlen = 4;
    120 	else
    121 		needlen = 2;
    122 
    123 	while (chlenbak < needlen) {
    124 		if (n==0)
    125 			goto restart;
    126 		psenc->ch[chlenbak++] = *s0++;
    127 		n--;
    128 		result++;
    129 	}
    130 
    131 	/* judge endian marker */
    132 	if ((ei->mode & _MODE_UTF32) == 0) {
    133 		/* UTF16 */
    134 		if (psenc->ch[0]==0xFE && psenc->ch[1]==0xFF) {
    135 			psenc->current_endian = _ENDIAN_BIG;
    136 			chlenbak = 0;
    137 			goto refetch;
    138 		} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE) {
    139 			psenc->current_endian = _ENDIAN_LITTLE;
    140 			chlenbak = 0;
    141 			goto refetch;
    142 		}
    143 	} else {
    144 		/* UTF32 */
    145 		if (psenc->ch[0]==0x00 && psenc->ch[1]==0x00 &&
    146 		    psenc->ch[2]==0xFE && psenc->ch[3]==0xFF) {
    147 			psenc->current_endian = _ENDIAN_BIG;
    148 			chlenbak = 0;
    149 			goto refetch;
    150 		} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE &&
    151 			   psenc->ch[2]==0x00 && psenc->ch[3]==0x00) {
    152 			psenc->current_endian = _ENDIAN_LITTLE;
    153 			chlenbak = 0;
    154 			goto refetch;
    155 		}
    156 	}
    157 	if ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
    158 	    psenc->current_endian == _ENDIAN_UNKNOWN)
    159 		endian = ei->preffered_endian;
    160 	else
    161 		endian = psenc->current_endian;
    162 
    163 	/* get wc */
    164 	if ((ei->mode & _MODE_UTF32) == 0) {
    165 		/* UTF16 */
    166 		if (needlen==2) {
    167 			switch (endian) {
    168 			case _ENDIAN_LITTLE:
    169 				wc = (psenc->ch[0] |
    170 				      ((wchar_t)psenc->ch[1] << 8));
    171 				break;
    172 			case _ENDIAN_BIG:
    173 				wc = (psenc->ch[1] |
    174 				      ((wchar_t)psenc->ch[0] << 8));
    175 				break;
    176 			default:
    177 				goto ilseq;
    178 			}
    179 			if (wc >= 0xD800 && wc <= 0xDBFF) {
    180 				/* surrogate high */
    181 				needlen=4;
    182 				goto refetch;
    183 			}
    184 		} else {
    185 			/* surrogate low */
    186 			wc -= 0xD800; /* wc : surrogate high (see above) */
    187 			wc <<= 10;
    188 			switch (endian) {
    189 			case _ENDIAN_LITTLE:
    190 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
    191 					goto ilseq;
    192 				wc |= psenc->ch[2];
    193 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
    194 				break;
    195 			case _ENDIAN_BIG:
    196 				if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
    197 					goto ilseq;
    198 				wc |= psenc->ch[3];
    199 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
    200 				break;
    201 			default:
    202 				goto ilseq;
    203 			}
    204 			wc += 0x10000;
    205 		}
    206 	} else {
    207 		/* UTF32 */
    208 		switch (endian) {
    209 		case _ENDIAN_LITTLE:
    210 			wc = (psenc->ch[0] |
    211 			      ((wchar_t)psenc->ch[1] << 8) |
    212 			      ((wchar_t)psenc->ch[2] << 16) |
    213 			      ((wchar_t)psenc->ch[3] << 24));
    214 			break;
    215 		case _ENDIAN_BIG:
    216 			wc = (psenc->ch[3] |
    217 			      ((wchar_t)psenc->ch[2] << 8) |
    218 			      ((wchar_t)psenc->ch[1] << 16) |
    219 			      ((wchar_t)psenc->ch[0] << 24));
    220 			break;
    221 		default:
    222 			goto ilseq;
    223 		}
    224 	}
    225 
    226 
    227 	*pwc = wc;
    228 	psenc->chlen = 0;
    229 	*nresult = result;
    230 	*s = s0;
    231 
    232 	return (0);
    233 
    234 ilseq:
    235 	*nresult = (size_t)-1;
    236 	psenc->chlen = 0;
    237 	return (EILSEQ);
    238 
    239 restart:
    240 	*nresult = (size_t)-2;
    241 	psenc->chlen = chlenbak;
    242 	*s = s0;
    243 	return (0);
    244 }
    245 
    246 static int
    247 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
    248 			     wchar_t wc, _UTF1632State *psenc,
    249 			     size_t *nresult)
    250 {
    251 	int ret;
    252 	wchar_t wc2;
    253 
    254 	_DIAGASSERT(ei != NULL);
    255 	_DIAGASSERT(nresult != 0);
    256 	_DIAGASSERT(s != NULL);
    257 
    258 	wc2 = 0;
    259 	if ((ei->mode & _MODE_UTF32)==0) {
    260 		/* UTF16 */
    261 		if (wc>0xFFFF) {
    262 			/* surrogate */
    263 			if (wc>0x10FFFF) {
    264 				ret = EILSEQ;
    265 				goto err;
    266 			}
    267 			if (n < 4) {
    268 				ret = E2BIG;
    269 				goto err;
    270 			}
    271 			wc -= 0x10000;
    272 			wc2 = (wc & 0x3FF) | 0xDC00;
    273 			wc = (wc>>10) | 0xD800;
    274 			*nresult = (size_t)4;
    275 		} else {
    276 			if (n < 2) {
    277 				ret = E2BIG;
    278 				goto err;
    279 			}
    280 			*nresult = (size_t)2;
    281 		}
    282 
    283 surrogate:
    284 		switch (ei->preffered_endian) {
    285 		case _ENDIAN_BIG:
    286 			s[1] = wc;
    287 			s[0] = (wc >>= 8);
    288 			break;
    289 		case _ENDIAN_LITTLE:
    290 			s[0] = wc;
    291 			s[1] = (wc >>= 8);
    292 			break;
    293 		}
    294 		if (wc2!=0) {
    295 			wc = wc2;
    296 			wc2 = 0;
    297 			s += 2;
    298 			goto surrogate;
    299 		}
    300 	} else {
    301 		/* UTF32 */
    302 		if (n < 4) {
    303 			ret = E2BIG;
    304 			goto err;
    305 		}
    306 		switch (ei->preffered_endian) {
    307 		case _ENDIAN_BIG:
    308 			s[3] = wc;
    309 			s[2] = (wc >>= 8);
    310 			s[1] = (wc >>= 8);
    311 			s[0] = (wc >>= 8);
    312 			break;
    313 		case _ENDIAN_LITTLE:
    314 			s[0] = wc;
    315 			s[1] = (wc >>= 8);
    316 			s[2] = (wc >>= 8);
    317 			s[3] = (wc >>= 8);
    318 			break;
    319 		}
    320 		*nresult = (size_t)4;
    321 	}
    322 
    323 	return 0;
    324 
    325 err:
    326 	*nresult = (size_t)-1;
    327 	return ret;
    328 }
    329 
    330 static void
    331 parse_variable(_UTF1632EncodingInfo * __restrict ei,
    332 	       const void * __restrict var, size_t lenvar)
    333 {
    334 #define MATCH(x, act)						\
    335 do {								\
    336 	if (lenvar >= (sizeof(#x)-1) &&				\
    337 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
    338 		act;						\
    339 		lenvar -= sizeof(#x)-1;				\
    340 		p += sizeof(#x)-1;				\
    341 	}							\
    342 } while (/*CONSTCOND*/0)
    343 	const char *p;
    344 	p = var;
    345 	while (lenvar>0) {
    346 		switch (*p) {
    347 		case 'B':
    348 		case 'b':
    349 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
    350 			break;
    351 		case 'L':
    352 		case 'l':
    353 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
    354 			break;
    355 		case 'F':
    356 		case 'f':
    357 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
    358 			break;
    359 		case 'U':
    360 		case 'u':
    361 			MATCH(utf32, ei->mode |= _MODE_UTF32);
    362 			break;
    363 		}
    364 		p++;
    365 		lenvar--;
    366 	}
    367 }
    368 
    369 static int
    370 /*ARGSUSED*/
    371 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
    372 				     const void * __restrict var,
    373 				     size_t lenvar)
    374 {
    375 	_DIAGASSERT(ei != NULL);
    376 
    377 	memset((void *)ei, 0, sizeof(*ei));
    378 
    379 	parse_variable(ei, var, lenvar);
    380 
    381 	if ((ei->mode&_MODE_UTF32)==0)
    382 		ei->cur_max = 6; /* endian + surrogate */
    383 	else
    384 		ei->cur_max = 8; /* endian + normal */
    385 
    386 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
    387 #if BYTE_ORDER == BIG_ENDIAN
    388 		ei->preffered_endian = _ENDIAN_BIG;
    389 #else
    390 		ei->preffered_endian = _ENDIAN_LITTLE;
    391 #endif
    392 	}
    393 
    394 	return (0);
    395 }
    396 
    397 static void
    398 /*ARGSUSED*/
    399 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
    400 {
    401 }
    402 
    403 static __inline int
    404 /*ARGSUSED*/
    405 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei,
    406 			      _csid_t * __restrict csid,
    407 			      _index_t * __restrict idx,
    408 			      _wc_t wc)
    409 {
    410 
    411 	_DIAGASSERT(csid != NULL && idx != NULL);
    412 
    413 	*csid = 0;
    414 	*idx = (_index_t)wc;
    415 
    416 	return (0);
    417 }
    418 
    419 static __inline int
    420 /*ARGSUSED*/
    421 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei,
    422 			      _wc_t * __restrict wc,
    423 			      _csid_t csid, _index_t idx)
    424 {
    425 
    426 	_DIAGASSERT(wc != NULL);
    427 
    428 	if (csid != 0)
    429 		return (EILSEQ);
    430 
    431 	*wc = (_wc_t)idx;
    432 
    433 	return (0);
    434 }
    435 
    436 static __inline int
    437 /*ARGSUSED*/
    438 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei,
    439 					      _UTF1632State * __restrict psenc,
    440 					      int * __restrict rstate)
    441 {
    442 
    443 	if (psenc->chlen == 0)
    444 		*rstate = _STDENC_SDGEN_INITIAL;
    445 	else
    446 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
    447 
    448 	return 0;
    449 }
    450 
    451 /* ----------------------------------------------------------------------
    452  * public interface for stdenc
    453  */
    454 
    455 _CITRUS_STDENC_DECLS(UTF1632);
    456 _CITRUS_STDENC_DEF_OPS(UTF1632);
    457 
    458 #include "citrus_stdenc_template.h"
    459