xref: /freebsd/lib/libiconv_modules/UTF1632/citrus_utf1632.c (revision 8d20be1e22095c27faf8fe8b2f0d089739cc742e) 
1 /* $FreeBSD$ */
2 /*	$NetBSD: citrus_utf1632.c,v 1.9 2008/06/14 16:01:08 tnozaki Exp $	*/
3 
4 /*-
5  * Copyright (c)2003 Citrus Project,
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 #include <sys/cdefs.h>
31 #include <sys/endian.h>
32 #include <sys/types.h>
33 
34 #include <assert.h>
35 #include <errno.h>
36 #include <limits.h>
37 #include <stddef.h>
38 #include <stdio.h>
39 #include <stdlib.h>
40 #include <string.h>
41 #include <wchar.h>
42 
43 #include "citrus_namespace.h"
44 #include "citrus_types.h"
45 #include "citrus_module.h"
46 #include "citrus_stdenc.h"
47 #include "citrus_bcs.h"
48 
49 #include "citrus_utf1632.h"
50 
51 
52 /* ----------------------------------------------------------------------
53  * private stuffs used by templates
54  */
55 
56 typedef struct {
57 	int		 chlen;
58 	int		 current_endian;
59 	uint8_t		 ch[4];
60 } _UTF1632State;
61 
62 #define _ENDIAN_UNKNOWN		0
63 #define _ENDIAN_BIG		1
64 #define _ENDIAN_LITTLE		2
65 #if BYTE_ORDER == BIG_ENDIAN
66 #define _ENDIAN_INTERNAL	_ENDIAN_BIG
67 #define _ENDIAN_SWAPPED		_ENDIAN_LITTLE
68 #else
69 #define _ENDIAN_INTERNAL	_ENDIAN_LITTLE
70 #define _ENDIAN_SWAPPED	_ENDIAN_BIG
71 #endif
72 #define _MODE_UTF32		0x00000001U
73 #define _MODE_FORCE_ENDIAN	0x00000002U
74 
75 typedef struct {
76 	int		 preffered_endian;
77 	unsigned int	 cur_max;
78 	uint32_t	 mode;
79 } _UTF1632EncodingInfo;
80 
81 #define _FUNCNAME(m)			_citrus_UTF1632_##m
82 #define _ENCODING_INFO			_UTF1632EncodingInfo
83 #define _ENCODING_STATE			_UTF1632State
84 #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
85 #define _ENCODING_IS_STATE_DEPENDENT	0
86 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
87 
88 
89 static __inline void
90 /*ARGSUSED*/
91 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
92     _UTF1632State *s)
93 {
94 
95 	memset(s, 0, sizeof(*s));
96 }
97 
98 static int
99 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
100     const char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
101 {
102 	const char *s0;
103 	size_t result;
104 	wchar_t wc = L'\0';
105 	int chlenbak, endian, needlen;
106 
107 	s0 = *s;
108 
109 	if (s0 == NULL) {
110 		_citrus_UTF1632_init_state(ei, psenc);
111 		*nresult = 0; /* state independent */
112 		return (0);
113 	}
114 
115 	result = 0;
116 	chlenbak = psenc->chlen;
117 
118 refetch:
119 	needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 2;
120 
121 	while (chlenbak < needlen) {
122 		if (n == 0)
123 			goto restart;
124 		psenc->ch[chlenbak++] = *s0++;
125 		n--;
126 		result++;
127 	}
128 
129 	/* judge endian marker */
130 	if ((ei->mode & _MODE_UTF32) == 0) {
131 		/* UTF16 */
132 		if (psenc->ch[0] == 0xFE && psenc->ch[1] == 0xFF) {
133 			psenc->current_endian = _ENDIAN_BIG;
134 			chlenbak = 0;
135 			goto refetch;
136 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE) {
137 			psenc->current_endian = _ENDIAN_LITTLE;
138 			chlenbak = 0;
139 			goto refetch;
140 		}
141 	} else {
142 		/* UTF32 */
143 		if (psenc->ch[0] == 0x00 && psenc->ch[1] == 0x00 &&
144 		    psenc->ch[2] == 0xFE && psenc->ch[3] == 0xFF) {
145 			psenc->current_endian = _ENDIAN_BIG;
146 			chlenbak = 0;
147 			goto refetch;
148 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE &&
149 			   psenc->ch[2] == 0x00 && psenc->ch[3] == 0x00) {
150 			psenc->current_endian = _ENDIAN_LITTLE;
151 			chlenbak = 0;
152 			goto refetch;
153 		}
154 	}
155 	endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
156 	    psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
157 	    psenc->current_endian;
158 
159 	/* get wc */
160 	if ((ei->mode & _MODE_UTF32) == 0) {
161 		/* UTF16 */
162 		if (needlen == 2) {
163 			switch (endian) {
164 			case _ENDIAN_LITTLE:
165 				wc = (psenc->ch[0] |
166 				    ((wchar_t)psenc->ch[1] << 8));
167 				break;
168 			case _ENDIAN_BIG:
169 				wc = (psenc->ch[1] |
170 				    ((wchar_t)psenc->ch[0] << 8));
171 				break;
172 			default:
173 				goto ilseq;
174 			}
175 			if (wc >= 0xD800 && wc <= 0xDBFF) {
176 				/* surrogate high */
177 				needlen = 4;
178 				goto refetch;
179 			}
180 		} else {
181 			/* surrogate low */
182 			wc -= 0xD800; /* wc : surrogate high (see above) */
183 			wc <<= 10;
184 			switch (endian) {
185 			case _ENDIAN_LITTLE:
186 				if (psenc->ch[3] < 0xDC || psenc->ch[3] > 0xDF)
187 					goto ilseq;
188 				wc |= psenc->ch[2];
189 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
190 				break;
191 			case _ENDIAN_BIG:
192 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
193 					goto ilseq;
194 				wc |= psenc->ch[3];
195 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
196 				break;
197 			default:
198 				goto ilseq;
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 		default:
218 			goto ilseq;
219 		}
220 		if (wc >= 0xD800 && wc <= 0xDFFF)
221 			goto ilseq;
222 	}
223 
224 
225 	*pwc = wc;
226 	psenc->chlen = 0;
227 	*nresult = result;
228 	*s = s0;
229 
230 	return (0);
231 
232 ilseq:
233 	*nresult = (size_t)-1;
234 	psenc->chlen = 0;
235 	return (EILSEQ);
236 
237 restart:
238 	*nresult = (size_t)-2;
239 	psenc->chlen = chlenbak;
240 	*s = s0;
241 	return (0);
242 }
243 
244 static int
245 _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
246     wchar_t wc, _UTF1632State *psenc, size_t *nresult)
247 {
248 	wchar_t wc2;
249 	static const char _bom[4] = {
250 	    0x00, 0x00, 0xFE, 0xFF,
251 	};
252 	const char *bom = &_bom[0];
253 	size_t cnt;
254 
255 	cnt = (size_t)0;
256 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
257 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
258 			if (ei->mode & _MODE_UTF32)
259 				cnt = 4;
260 			else {
261 				cnt = 2;
262 				bom += 2;
263 			}
264 			if (n < cnt)
265 				goto e2big;
266 			memcpy(s, bom, cnt);
267 			s += cnt, n -= cnt;
268 		}
269 		psenc->current_endian = ei->preffered_endian;
270 	}
271 
272 	wc2 = 0;
273 	if ((ei->mode & _MODE_UTF32)==0) {
274 		/* UTF16 */
275 		if (wc > 0xFFFF) {
276 			/* surrogate */
277 			if (wc > 0x10FFFF)
278 				goto ilseq;
279 			if (n < 4)
280 				goto e2big;
281 			cnt += 4;
282 			wc -= 0x10000;
283 			wc2 = (wc & 0x3FF) | 0xDC00;
284 			wc = (wc>>10) | 0xD800;
285 		} else {
286 			if (n < 2)
287 				goto e2big;
288 			cnt += 2;
289 		}
290 
291 surrogate:
292 		switch (psenc->current_endian) {
293 		case _ENDIAN_BIG:
294 			s[1] = wc;
295 			s[0] = (wc >>= 8);
296 			break;
297 		case _ENDIAN_LITTLE:
298 			s[0] = wc;
299 			s[1] = (wc >>= 8);
300 			break;
301 		}
302 		if (wc2 != 0) {
303 			wc = wc2;
304 			wc2 = 0;
305 			s += 2;
306 			goto surrogate;
307 		}
308 	} else {
309 		/* UTF32 */
310 		if (wc >= 0xD800 && wc <= 0xDFFF)
311 			goto ilseq;
312 		if (n < 4)
313 			goto e2big;
314 		cnt += 4;
315 		switch (psenc->current_endian) {
316 		case _ENDIAN_BIG:
317 			s[3] = wc;
318 			s[2] = (wc >>= 8);
319 			s[1] = (wc >>= 8);
320 			s[0] = (wc >>= 8);
321 			break;
322 		case _ENDIAN_LITTLE:
323 			s[0] = wc;
324 			s[1] = (wc >>= 8);
325 			s[2] = (wc >>= 8);
326 			s[3] = (wc >>= 8);
327 			break;
328 		}
329 	}
330 	*nresult = cnt;
331 
332 	return (0);
333 
334 ilseq:
335 	*nresult = (size_t)-1;
336 	return (EILSEQ);
337 e2big:
338 	*nresult = (size_t)-1;
339 	return (E2BIG);
340 }
341 
342 static void
343 parse_variable(_UTF1632EncodingInfo * __restrict ei,
344     const void * __restrict var, size_t lenvar)
345 {
346 	const char *p;
347 
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 'i':
360 		case 'I':
361 			MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
362 			break;
363 		case 's':
364 		case 'S':
365 			MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
366 			break;
367 		case 'F':
368 		case 'f':
369 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
370 			break;
371 		case 'U':
372 		case 'u':
373 			MATCH(utf32, ei->mode |= _MODE_UTF32);
374 			break;
375 		}
376 		p++;
377 		lenvar--;
378 	}
379 }
380 
381 static int
382 /*ARGSUSED*/
383 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
384     const void * __restrict var, size_t lenvar)
385 {
386 
387 	memset((void *)ei, 0, sizeof(*ei));
388 
389 	parse_variable(ei, var, lenvar);
390 
391 	ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
392 	/* 6: endian + surrogate */
393 	/* 8: endian + normal */
394 
395 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
396 		ei->preffered_endian = _ENDIAN_BIG;
397 	}
398 
399 	return (0);
400 }
401 
402 static void
403 /*ARGSUSED*/
404 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
405 {
406 
407 }
408 
409 static __inline int
410 /*ARGSUSED*/
411 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
412      _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
413 {
414 
415 	*csid = 0;
416 	*idx = (_index_t)wc;
417 
418 	return (0);
419 }
420 
421 static __inline int
422 /*ARGSUSED*/
423 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
424     _wc_t * __restrict wc, _csid_t csid, _index_t idx)
425 {
426 
427 	if (csid != 0)
428 		return (EILSEQ);
429 
430 	*wc = (_wc_t)idx;
431 
432 	return (0);
433 }
434 
435 static __inline int
436 /*ARGSUSED*/
437 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
438     _UTF1632State * __restrict psenc, int * __restrict rstate)
439 {
440 
441 	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
442 	    _STDENC_SDGEN_INCOMPLETE_CHAR;
443 	return (0);
444 }
445 
446 /* ----------------------------------------------------------------------
447  * public interface for stdenc
448  */
449 
450 _CITRUS_STDENC_DECLS(UTF1632);
451 _CITRUS_STDENC_DEF_OPS(UTF1632);
452 
453 #include "citrus_stdenc_template.h"
454