xref: /netbsd-src/lib/libc/citrus/modules/citrus_utf1632.c (revision 8b0f9554ff8762542c4defc4f70e1eb76fb508fa) 
1 /*	$NetBSD: citrus_utf1632.c,v 1.7 2006/10/27 14:13:55 tnozaki 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 1.7 2006/10/27 14:13:55 tnozaki 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 		if (wc >= 0xD800 && wc <= 0xDFFF)
225 			goto ilseq;
226 	}
227 
228 
229 	*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_t wc, _UTF1632State *psenc,
251 			     size_t *nresult)
252 {
253 	int ret;
254 	wchar_t wc2;
255 
256 	_DIAGASSERT(ei != NULL);
257 	_DIAGASSERT(nresult != 0);
258 	_DIAGASSERT(s != NULL);
259 
260 	wc2 = 0;
261 	if ((ei->mode & _MODE_UTF32)==0) {
262 		/* UTF16 */
263 		if (wc>0xFFFF) {
264 			/* surrogate */
265 			if (wc>0x10FFFF) {
266 				ret = EILSEQ;
267 				goto err;
268 			}
269 			if (n < 4) {
270 				ret = E2BIG;
271 				goto err;
272 			}
273 			wc -= 0x10000;
274 			wc2 = (wc & 0x3FF) | 0xDC00;
275 			wc = (wc>>10) | 0xD800;
276 			*nresult = (size_t)4;
277 		} else {
278 			if (n < 2) {
279 				ret = E2BIG;
280 				goto err;
281 			}
282 			*nresult = (size_t)2;
283 		}
284 
285 surrogate:
286 		switch (ei->preffered_endian) {
287 		case _ENDIAN_BIG:
288 			s[1] = wc;
289 			s[0] = (wc >>= 8);
290 			break;
291 		case _ENDIAN_LITTLE:
292 			s[0] = wc;
293 			s[1] = (wc >>= 8);
294 			break;
295 		}
296 		if (wc2!=0) {
297 			wc = wc2;
298 			wc2 = 0;
299 			s += 2;
300 			goto surrogate;
301 		}
302 	} else {
303 		/* UTF32 */
304 		if (wc >= 0xD800 && wc <= 0xDFFF)
305 			goto err;
306 		if (n < 4) {
307 			ret = E2BIG;
308 			goto err;
309 		}
310 		switch (ei->preffered_endian) {
311 		case _ENDIAN_BIG:
312 			s[3] = wc;
313 			s[2] = (wc >>= 8);
314 			s[1] = (wc >>= 8);
315 			s[0] = (wc >>= 8);
316 			break;
317 		case _ENDIAN_LITTLE:
318 			s[0] = wc;
319 			s[1] = (wc >>= 8);
320 			s[2] = (wc >>= 8);
321 			s[3] = (wc >>= 8);
322 			break;
323 		}
324 		*nresult = (size_t)4;
325 	}
326 
327 	return 0;
328 
329 err:
330 	*nresult = (size_t)-1;
331 	return ret;
332 }
333 
334 static void
335 parse_variable(_UTF1632EncodingInfo * __restrict ei,
336 	       const void * __restrict var, size_t lenvar)
337 {
338 #define MATCH(x, act)						\
339 do {								\
340 	if (lenvar >= (sizeof(#x)-1) &&				\
341 	    _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) {	\
342 		act;						\
343 		lenvar -= sizeof(#x)-1;				\
344 		p += sizeof(#x)-1;				\
345 	}							\
346 } while (/*CONSTCOND*/0)
347 	const char *p;
348 	p = var;
349 	while (lenvar>0) {
350 		switch (*p) {
351 		case 'B':
352 		case 'b':
353 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
354 			break;
355 		case 'L':
356 		case 'l':
357 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
358 			break;
359 		case 'F':
360 		case 'f':
361 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
362 			break;
363 		case 'U':
364 		case 'u':
365 			MATCH(utf32, ei->mode |= _MODE_UTF32);
366 			break;
367 		}
368 		p++;
369 		lenvar--;
370 	}
371 }
372 
373 static int
374 /*ARGSUSED*/
375 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
376 				     const void * __restrict var,
377 				     size_t lenvar)
378 {
379 	_DIAGASSERT(ei != NULL);
380 
381 	memset((void *)ei, 0, sizeof(*ei));
382 
383 	parse_variable(ei, var, lenvar);
384 
385 	if ((ei->mode&_MODE_UTF32)==0)
386 		ei->cur_max = 6; /* endian + surrogate */
387 	else
388 		ei->cur_max = 8; /* endian + normal */
389 
390 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
391 #if BYTE_ORDER == BIG_ENDIAN
392 		ei->preffered_endian = _ENDIAN_BIG;
393 #else
394 		ei->preffered_endian = _ENDIAN_LITTLE;
395 #endif
396 	}
397 
398 	return (0);
399 }
400 
401 static void
402 /*ARGSUSED*/
403 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei)
404 {
405 }
406 
407 static __inline int
408 /*ARGSUSED*/
409 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei,
410 			      _csid_t * __restrict csid,
411 			      _index_t * __restrict idx,
412 			      _wc_t wc)
413 {
414 
415 	_DIAGASSERT(csid != NULL && idx != NULL);
416 
417 	*csid = 0;
418 	*idx = (_index_t)wc;
419 
420 	return (0);
421 }
422 
423 static __inline int
424 /*ARGSUSED*/
425 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei,
426 			      _wc_t * __restrict wc,
427 			      _csid_t csid, _index_t idx)
428 {
429 
430 	_DIAGASSERT(wc != NULL);
431 
432 	if (csid != 0)
433 		return (EILSEQ);
434 
435 	*wc = (_wc_t)idx;
436 
437 	return (0);
438 }
439 
440 static __inline int
441 /*ARGSUSED*/
442 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei,
443 					      _UTF1632State * __restrict psenc,
444 					      int * __restrict rstate)
445 {
446 
447 	if (psenc->chlen == 0)
448 		*rstate = _STDENC_SDGEN_INITIAL;
449 	else
450 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
451 
452 	return 0;
453 }
454 
455 /* ----------------------------------------------------------------------
456  * public interface for stdenc
457  */
458 
459 _CITRUS_STDENC_DECLS(UTF1632);
460 _CITRUS_STDENC_DEF_OPS(UTF1632);
461 
462 #include "citrus_stdenc_template.h"
463