xref: /netbsd-src/lib/libc/citrus/modules/citrus_utf1632.c (revision aaf4ece63a859a04e37cf3a7229b5fab0157cc06) 
1 /*	$NetBSD: citrus_utf1632.c,v 1.4 2005/10/29 18:02:04 tshiozak 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 1.4 2005/10/29 18:02:04 tshiozak 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 <sys/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 			}
177 			if (wc >= 0xD800 && wc <= 0xDBFF) {
178 				/* surrogate high */
179 				needlen=4;
180 				goto refetch;
181 			}
182 		} else {
183 			/* surrogate low */
184 			wc -= 0xD800; /* wc : surrogate high (see above) */
185 			wc <<= 10;
186 			switch (endian) {
187 			case _ENDIAN_LITTLE:
188 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
189 					goto ilseq;
190 				wc |= psenc->ch[2];
191 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
192 				break;
193 			case _ENDIAN_BIG:
194 				if (psenc->ch[3]<0xDC || psenc->ch[3]>0xDF)
195 					goto ilseq;
196 				wc |= psenc->ch[3];
197 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
198 				break;
199 			}
200 			wc += 0x10000;
201 		}
202 	} else {
203 		/* UTF32 */
204 		switch (endian) {
205 		case _ENDIAN_LITTLE:
206 			wc = (psenc->ch[0] |
207 			      ((wchar_t)psenc->ch[1] << 8) |
208 			      ((wchar_t)psenc->ch[2] << 16) |
209 			      ((wchar_t)psenc->ch[3] << 24));
210 			break;
211 		case _ENDIAN_BIG:
212 			wc = (psenc->ch[3] |
213 			      ((wchar_t)psenc->ch[2] << 8) |
214 			      ((wchar_t)psenc->ch[1] << 16) |
215 			      ((wchar_t)psenc->ch[0] << 24));
216 			break;
217 		}
218 	}
219 
220 
221 	*pwc = wc;
222 	psenc->chlen = 0;
223 	*nresult = result;
224 	*s = s0;
225 
226 	return (0);
227 
228 ilseq:
229 	*nresult = (size_t)-1;
230 	psenc->chlen = 0;
231 	return (EILSEQ);
232 
233 restart:
234 	*nresult = (size_t)-2;
235 	psenc->chlen = chlenbak;
236 	*s = s0;
237 	return (0);
238 }
239 
240 static int
241 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
242 			     wchar_t wc, _UTF1632State *psenc,
243 			     size_t *nresult)
244 {
245 	int ret;
246 	wchar_t wc2;
247 
248 	_DIAGASSERT(ei != NULL);
249 	_DIAGASSERT(nresult != 0);
250 	_DIAGASSERT(s != NULL);
251 
252 	wc2 = 0;
253 	if ((ei->mode & _MODE_UTF32)==0) {
254 		/* UTF16 */
255 		if (wc>0xFFFF) {
256 			/* surrogate */
257 			if (wc>0x10FFFF) {
258 				ret = EILSEQ;
259 				goto err;
260 			}
261 			if (n < 4) {
262 				ret = E2BIG;
263 				goto err;
264 			}
265 			wc -= 0x10000;
266 			wc2 = (wc & 0x3FF) | 0xDC00;
267 			wc = (wc>>10) | 0xD800;
268 			*nresult = (size_t)4;
269 		} else {
270 			if (n < 2) {
271 				ret = E2BIG;
272 				goto err;
273 			}
274 			*nresult = (size_t)2;
275 		}
276 
277 surrogate:
278 		switch (ei->preffered_endian) {
279 		case _ENDIAN_BIG:
280 			s[1] = wc;
281 			s[0] = (wc >>= 8);
282 			break;
283 		case _ENDIAN_LITTLE:
284 			s[0] = wc;
285 			s[1] = (wc >>= 8);
286 			break;
287 		}
288 		if (wc2!=0) {
289 			wc = wc2;
290 			wc2 = 0;
291 			s += 2;
292 			goto surrogate;
293 		}
294 	} else {
295 		/* UTF32 */
296 		if (n < 4) {
297 			ret = E2BIG;
298 			goto err;
299 		}
300 		switch (ei->preffered_endian) {
301 		case _ENDIAN_BIG:
302 			s[3] = wc;
303 			s[2] = (wc >>= 8);
304 			s[1] = (wc >>= 8);
305 			s[0] = (wc >>= 8);
306 			break;
307 		case _ENDIAN_LITTLE:
308 			s[0] = wc;
309 			s[1] = (wc >>= 8);
310 			s[2] = (wc >>= 8);
311 			s[3] = (wc >>= 8);
312 			break;
313 		}
314 		*nresult = (size_t)4;
315 	}
316 
317 	return 0;
318 
319 err:
320 	*nresult = (size_t)-1;
321 	return ret;
322 }
323 
324 static void
325 parse_variable(_UTF1632EncodingInfo * __restrict ei,
326 	       const void * __restrict var, size_t lenvar)
327 {
328 #define MATCH(x, act)						\
329 do {								\
330 	if (lenvar >= (sizeof(#x)-1) &&				\
331 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
332 		act;						\
333 		lenvar -= sizeof(#x)-1;				\
334 		p += sizeof(#x)-1;				\
335 	}							\
336 } while (/*CONSTCOND*/0)
337 	const char *p;
338 	p = var;
339 	while (lenvar>0) {
340 		switch (*p) {
341 		case 'B':
342 		case 'b':
343 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
344 			break;
345 		case 'L':
346 		case 'l':
347 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
348 			break;
349 		case 'F':
350 		case 'f':
351 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
352 			break;
353 		case 'U':
354 		case 'u':
355 			MATCH(utf32, ei->mode |= _MODE_UTF32);
356 			break;
357 		}
358 		p++;
359 		lenvar--;
360 	}
361 }
362 
363 static int
364 /*ARGSUSED*/
365 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
366 				     const void * __restrict var,
367 				     size_t lenvar)
368 {
369 	_DIAGASSERT(ei != NULL);
370 
371 	memset((void *)ei, 0, sizeof(*ei));
372 
373 	parse_variable(ei, var, lenvar);
374 
375 	if ((ei->mode&_MODE_UTF32)==0)
376 		ei->cur_max = 6; /* endian + surrogate */
377 	else
378 		ei->cur_max = 8; /* endian + normal */
379 
380 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
381 #if BYTE_ORDER == BIG_ENDIAN
382 		ei->preffered_endian = _ENDIAN_BIG;
383 #else
384 		ei->preffered_endian = _ENDIAN_LITTLE;
385 #endif
386 	}
387 
388 	return (0);
389 }
390 
391 static void
392 /*ARGSUSED*/
393 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
394 {
395 }
396 
397 static __inline int
398 /*ARGSUSED*/
399 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei,
400 			      _csid_t * __restrict csid,
401 			      _index_t * __restrict idx,
402 			      _wc_t wc)
403 {
404 
405 	_DIAGASSERT(csid != NULL && idx != NULL);
406 
407 	*csid = 0;
408 	*idx = (_index_t)wc;
409 
410 	return (0);
411 }
412 
413 static __inline int
414 /*ARGSUSED*/
415 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei,
416 			      _wc_t * __restrict wc,
417 			      _csid_t csid, _index_t idx)
418 {
419 
420 	_DIAGASSERT(wc != NULL);
421 
422 	if (csid != 0)
423 		return (EILSEQ);
424 
425 	*wc = (_wc_t)idx;
426 
427 	return (0);
428 }
429 
430 static __inline int
431 /*ARGSUSED*/
432 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei,
433 					      _UTF1632State * __restrict psenc,
434 					      int * __restrict rstate)
435 {
436 
437 	if (psenc->chlen == 0)
438 		*rstate = _STDENC_SDGEN_INITIAL;
439 	else
440 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
441 
442 	return 0;
443 }
444 
445 /* ----------------------------------------------------------------------
446  * public interface for stdenc
447  */
448 
449 _CITRUS_STDENC_DECLS(UTF1632);
450 _CITRUS_STDENC_DEF_OPS(UTF1632);
451 
452 #include "citrus_stdenc_template.h"
453