Home | History | Annotate | Line # | Download | only in modules
citrus_utf1632.c revision 1.12.34.2 
      1 /*	$NetBSD: citrus_utf1632.c,v 1.12.34.2 2017/07/21 20:22:29 perseant 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 1.12.34.2 2017/07/21 20:22:29 perseant 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 <limits.h>
     41 #include <wchar.h>
     42 #include <sys/types.h>
     43 #include <machine/endian.h>
     44 
     45 #include "citrus_namespace.h"
     46 #include "citrus_types.h"
     47 #include "citrus_module.h"
     48 #include "citrus_stdenc.h"
     49 #include "citrus_bcs.h"
     50 
     51 #include "citrus_utf1632.h"
     52 
     53 
     54 /* ----------------------------------------------------------------------
     55  * private stuffs used by templates
     56  */
     57 
     58 typedef struct {
     59 	u_int8_t		ch[4];
     60 	int			chlen;
     61 	int			current_endian;
     62 } _UTF1632State;
     63 
     64 typedef struct {
     65 	int		preffered_endian;
     66 	unsigned int	cur_max;
     67 #define _ENDIAN_UNKNOWN	0
     68 #define _ENDIAN_BIG	1
     69 #define _ENDIAN_LITTLE	2
     70 	u_int32_t	mode;
     71 #define _MODE_UTF32		0x00000001U
     72 #define _MODE_FORCE_ENDIAN	0x00000002U
     73 } _UTF1632EncodingInfo;
     74 
     75 #define _FUNCNAME(m)			_citrus_UTF1632_##m
     76 #define _ENCODING_INFO			_UTF1632EncodingInfo
     77 #define _ENCODING_STATE			_UTF1632State
     78 #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
     79 #define _ENCODING_IS_STATE_DEPENDENT	0
     80 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
     81 
     82 #include "citrus_u2k_template.h"
     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_ucs4_t *pwc,
     93 			     const char **s, size_t n, _UTF1632State *psenc,
     94 			     size_t *nresult)
     95 {
     96 	int chlenbak, endian, needlen;
     97 	wchar_kuten_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 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
    132 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
    133 			/* judge endian marker */
    134 			if ((ei->mode & _MODE_UTF32) == 0) {
    135 				/* UTF16 */
    136 				if (psenc->ch[0]==0xFE && psenc->ch[1]==0xFF) {
    137 					psenc->current_endian = _ENDIAN_BIG;
    138 					chlenbak = 0;
    139 					goto refetch;
    140 				} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE) {
    141 					psenc->current_endian = _ENDIAN_LITTLE;
    142 					chlenbak = 0;
    143 					goto refetch;
    144 				}
    145 			} else {
    146 				/* UTF32 */
    147 				if (psenc->ch[0]==0x00 && psenc->ch[1]==0x00 &&
    148 				    psenc->ch[2]==0xFE && psenc->ch[3]==0xFF) {
    149 					psenc->current_endian = _ENDIAN_BIG;
    150 					chlenbak = 0;
    151 					goto refetch;
    152 				} else if (psenc->ch[0]==0xFF && psenc->ch[1]==0xFE &&
    153 					   psenc->ch[2]==0x00 && psenc->ch[3]==0x00) {
    154 					psenc->current_endian = _ENDIAN_LITTLE;
    155 					chlenbak = 0;
    156 					goto refetch;
    157 				}
    158 			}
    159 		}
    160 		psenc->current_endian = ei->preffered_endian;
    161 	}
    162 	endian = psenc->current_endian;
    163 
    164 	/* get wc */
    165 	if ((ei->mode & _MODE_UTF32) == 0) {
    166 		/* UTF16 */
    167 		if (needlen==2) {
    168 			switch (endian) {
    169 			case _ENDIAN_LITTLE:
    170 				wc = (psenc->ch[0] |
    171 				      ((wchar_kuten_t)psenc->ch[1] << 8));
    172 				break;
    173 			case _ENDIAN_BIG:
    174 				wc = (psenc->ch[1] |
    175 				      ((wchar_kuten_t)psenc->ch[0] << 8));
    176 				break;
    177 			default:
    178 				goto ilseq;
    179 			}
    180 			if (wc >= 0xD800 && wc <= 0xDBFF) {
    181 				/* surrogate high */
    182 				needlen=4;
    183 				goto refetch;
    184 			}
    185 		} else {
    186 			/* surrogate low */
    187 			wc -= 0xD800; /* wc : surrogate high (see above) */
    188 			wc <<= 10;
    189 			switch (endian) {
    190 			case _ENDIAN_LITTLE:
    191 				if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
    192 					goto ilseq;
    193 				wc |= psenc->ch[2];
    194 				wc |= (wchar_kuten_t)(psenc->ch[3] & 3) << 8;
    195 				break;
    196 			case _ENDIAN_BIG:
    197 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
    198 					goto ilseq;
    199 				wc |= psenc->ch[3];
    200 				wc |= (wchar_kuten_t)(psenc->ch[2] & 3) << 8;
    201 				break;
    202 			default:
    203 				goto ilseq;
    204 			}
    205 			wc += 0x10000;
    206 		}
    207 	} else {
    208 		/* UTF32 */
    209 		switch (endian) {
    210 		case _ENDIAN_LITTLE:
    211 			wc = (psenc->ch[0] |
    212 			      ((wchar_kuten_t)psenc->ch[1] << 8) |
    213 			      ((wchar_kuten_t)psenc->ch[2] << 16) |
    214 			      ((wchar_kuten_t)psenc->ch[3] << 24));
    215 			break;
    216 		case _ENDIAN_BIG:
    217 			wc = (psenc->ch[3] |
    218 			      ((wchar_kuten_t)psenc->ch[2] << 8) |
    219 			      ((wchar_kuten_t)psenc->ch[1] << 16) |
    220 			      ((wchar_kuten_t)psenc->ch[0] << 24));
    221 			break;
    222 		default:
    223 			goto ilseq;
    224 		}
    225 		if (wc >= 0xD800 && wc <= 0xDFFF)
    226 			goto ilseq;
    227 	}
    228 
    229 	_citrus_UTF1632_kt2ucs(ei, pwc, wc);
    230 	psenc->chlen = 0;
    231 	*nresult = result;
    232 	*s = s0;
    233 
    234 	return (0);
    235 
    236 ilseq:
    237 	*nresult = (size_t)-1;
    238 	psenc->chlen = 0;
    239 	return (EILSEQ);
    240 
    241 restart:
    242 	*nresult = (size_t)-2;
    243 	psenc->chlen = chlenbak;
    244 	*s = s0;
    245 	return (0);
    246 }
    247 
    248 static int
    249 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
    250 			     wchar_ucs4_t wc, _UTF1632State *psenc,
    251 			     size_t *nresult)
    252 {
    253 	wchar_kuten_t wc2;
    254 	static const char _bom[4] = {
    255 #if BYTE_ORDER == BIG_ENDIAN
    256 	    0x00, 0x00, 0xFE, 0xFF,
    257 #else
    258 	    0xFF, 0xFE, 0x00, 0x00,
    259 #endif
    260 	};
    261 	const char *bom = &_bom[0];
    262 	size_t cnt;
    263 
    264 	_DIAGASSERT(ei != NULL);
    265 	_DIAGASSERT(nresult != 0);
    266 	_DIAGASSERT(s != NULL);
    267 
    268 	_citrus_UTF1632_ucs2kt(ei, &wc, wc);
    269 
    270 	cnt = (size_t)0;
    271 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
    272 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
    273 			if (ei->mode & _MODE_UTF32) {
    274 				cnt = 4;
    275 			} else {
    276 				cnt = 2;
    277 #if BYTE_ORDER == BIG_ENDIAN
    278 				bom += 2;
    279 #endif
    280 			}
    281 			if (n < cnt)
    282 				goto e2big;
    283 			memcpy(s, bom, cnt);
    284 			s += cnt, n -= cnt;
    285 		}
    286 		psenc->current_endian = ei->preffered_endian;
    287 	}
    288 
    289 	wc2 = 0;
    290 	if ((ei->mode & _MODE_UTF32)==0) {
    291 		/* UTF16 */
    292 		if (wc>0xFFFF) {
    293 			/* surrogate */
    294 			if (wc>0x10FFFF)
    295 				goto ilseq;
    296 			if (n < 4)
    297 				goto e2big;
    298 			cnt += 4;
    299 			wc -= 0x10000;
    300 			wc2 = (wc & 0x3FF) | 0xDC00;
    301 			wc = (wc>>10) | 0xD800;
    302 		} else {
    303 			if (n < 2)
    304 				goto e2big;
    305 			cnt += 2;
    306 		}
    307 
    308 surrogate:
    309 		switch (psenc->current_endian) {
    310 		case _ENDIAN_BIG:
    311 			s[1] = wc;
    312 			s[0] = (wc >>= 8);
    313 			break;
    314 		case _ENDIAN_LITTLE:
    315 			s[0] = wc;
    316 			s[1] = (wc >>= 8);
    317 			break;
    318 		}
    319 		if (wc2!=0) {
    320 			wc = wc2;
    321 			wc2 = 0;
    322 			s += 2;
    323 			goto surrogate;
    324 		}
    325 	} else {
    326 		/* UTF32 */
    327 		if (wc >= 0xD800 && wc <= 0xDFFF)
    328 			goto ilseq;
    329 		if (n < 4)
    330 			goto e2big;
    331 		cnt += 4;
    332 		switch (psenc->current_endian) {
    333 		case _ENDIAN_BIG:
    334 			s[3] = wc;
    335 			s[2] = (wc >>= 8);
    336 			s[1] = (wc >>= 8);
    337 			s[0] = (wc >>= 8);
    338 			break;
    339 		case _ENDIAN_LITTLE:
    340 			s[0] = wc;
    341 			s[1] = (wc >>= 8);
    342 			s[2] = (wc >>= 8);
    343 			s[3] = (wc >>= 8);
    344 			break;
    345 		}
    346 	}
    347 	*nresult = cnt;
    348 
    349 	return 0;
    350 
    351 ilseq:
    352 	*nresult = (size_t)-1;
    353 	return EILSEQ;
    354 e2big:
    355 	*nresult = (size_t)-1;
    356 	return E2BIG;
    357 }
    358 
    359 static void
    360 parse_variable(_UTF1632EncodingInfo * __restrict ei,
    361 	       const void * __restrict var, size_t lenvar)
    362 {
    363 #define MATCH(x, act)						\
    364 do {								\
    365 	if (lenvar >= (sizeof(#x)-1) &&				\
    366 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
    367 		act;						\
    368 		lenvar -= sizeof(#x)-1;				\
    369 		p += sizeof(#x)-1;				\
    370 	}							\
    371 } while (/*CONSTCOND*/0)
    372 	const char *p;
    373 	p = var;
    374 	while (lenvar>0) {
    375 		switch (*p) {
    376 		case 'B':
    377 		case 'b':
    378 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
    379 			break;
    380 		case 'L':
    381 		case 'l':
    382 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
    383 			break;
    384 		case 'F':
    385 		case 'f':
    386 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
    387 			break;
    388 		case 'U':
    389 		case 'u':
    390 			MATCH(utf32, ei->mode |= _MODE_UTF32);
    391 			break;
    392 		}
    393 		p++;
    394 		lenvar--;
    395 	}
    396 }
    397 
    398 static int
    399 /*ARGSUSED*/
    400 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
    401 				     const void * __restrict var,
    402 				     size_t lenvar)
    403 {
    404 	_DIAGASSERT(ei != NULL);
    405 
    406 	memset((void *)ei, 0, sizeof(*ei));
    407 
    408 	parse_variable(ei, var, lenvar);
    409 
    410 	if ((ei->mode&_MODE_UTF32)==0)
    411 		ei->cur_max = 6; /* endian + surrogate */
    412 	else
    413 		ei->cur_max = 8; /* endian + normal */
    414 
    415 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
    416 #if BYTE_ORDER == BIG_ENDIAN
    417 		ei->preffered_endian = _ENDIAN_BIG;
    418 #else
    419 		ei->preffered_endian = _ENDIAN_LITTLE;
    420 #endif
    421 	}
    422 
    423 	return (0);
    424 }
    425 
    426 static void
    427 /*ARGSUSED*/
    428 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
    429 {
    430 }
    431 
    432 static int
    433 /*ARGSUSED*/
    434 _citrus_UTF1632_stdenc_wctocs(struct _citrus_stdenc *ce,
    435 			      _csid_t * __restrict csid,
    436 			      _index_t * __restrict idx,
    437 			      wchar_kuten_t wc)
    438 {
    439 
    440 	_DIAGASSERT(csid != NULL && idx != NULL);
    441 
    442 	*csid = 0;
    443 	*idx = (_index_t)wc;
    444 
    445 	return (0);
    446 }
    447 
    448 static int
    449 /*ARGSUSED*/
    450 _citrus_UTF1632_stdenc_cstowc(struct _citrus_stdenc *ce,
    451 			      wchar_kuten_t * __restrict wc,
    452 			      _csid_t csid, _index_t idx)
    453 {
    454 
    455 	_DIAGASSERT(wc != NULL);
    456 
    457 	if (csid != 0)
    458 		return (EILSEQ);
    459 
    460 	*wc = (wchar_kuten_t)idx;
    461 
    462 	return (0);
    463 }
    464 
    465 static __inline int
    466 /*ARGSUSED*/
    467 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei,
    468 					      _UTF1632State * __restrict psenc,
    469 					      int * __restrict rstate)
    470 {
    471 
    472 	if (psenc->chlen == 0)
    473 		*rstate = _STDENC_SDGEN_INITIAL;
    474 	else
    475 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
    476 
    477 	return 0;
    478 }
    479 
    480 /* ----------------------------------------------------------------------
    481  * public interface for stdenc
    482  */
    483 
    484 _CITRUS_STDENC_DECLS(UTF1632);
    485 _CITRUS_STDENC_DEF_OPS(UTF1632);
    486 
    487 #include "citrus_stdenc_template.h"
    488