xref: /freebsd/lib/libiconv_modules/UTF1632/citrus_utf1632.c (revision f6a3b357e9be4c6423c85eff9a847163a0d307c8) 
1 /* $FreeBSD$ */
2 /*	$NetBSD: citrus_utf1632.c,v 1.9 2008/06/14 16:01:08 tnozaki Exp $	*/
3 
4 /*-
5  * SPDX-License-Identifier: BSD-2-Clause
6  *
7  * Copyright (c)2003 Citrus Project,
8  * All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 #include <sys/cdefs.h>
33 #include <sys/endian.h>
34 #include <sys/types.h>
35 
36 #include <assert.h>
37 #include <errno.h>
38 #include <limits.h>
39 #include <stddef.h>
40 #include <stdio.h>
41 #include <stdlib.h>
42 #include <string.h>
43 #include <wchar.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 	int		 chlen;
60 	int		 current_endian;
61 	uint8_t		 ch[4];
62 } _UTF1632State;
63 
64 #define _ENDIAN_UNKNOWN		0
65 #define _ENDIAN_BIG		1
66 #define _ENDIAN_LITTLE		2
67 #if BYTE_ORDER == BIG_ENDIAN
68 #define _ENDIAN_INTERNAL	_ENDIAN_BIG
69 #define _ENDIAN_SWAPPED		_ENDIAN_LITTLE
70 #else
71 #define _ENDIAN_INTERNAL	_ENDIAN_LITTLE
72 #define _ENDIAN_SWAPPED	_ENDIAN_BIG
73 #endif
74 #define _MODE_UTF32		0x00000001U
75 #define _MODE_FORCE_ENDIAN	0x00000002U
76 
77 typedef struct {
78 	int		 preffered_endian;
79 	unsigned int	 cur_max;
80 	uint32_t	 mode;
81 } _UTF1632EncodingInfo;
82 
83 #define _FUNCNAME(m)			_citrus_UTF1632_##m
84 #define _ENCODING_INFO			_UTF1632EncodingInfo
85 #define _ENCODING_STATE			_UTF1632State
86 #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
87 #define _ENCODING_IS_STATE_DEPENDENT	0
88 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
89 
90 
91 static __inline void
92 /*ARGSUSED*/
93 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
94     _UTF1632State *s)
95 {
96 
97 	memset(s, 0, sizeof(*s));
98 }
99 
100 static int
101 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
102     char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
103 {
104 	char *s0;
105 	size_t result;
106 	wchar_t wc = L'\0';
107 	int chlenbak, endian, needlen;
108 
109 	s0 = *s;
110 
111 	if (s0 == NULL) {
112 		_citrus_UTF1632_init_state(ei, psenc);
113 		*nresult = 0; /* state independent */
114 		return (0);
115 	}
116 
117 	result = 0;
118 	chlenbak = psenc->chlen;
119 
120 refetch:
121 	needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 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 	endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
158 	    psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
159 	    psenc->current_endian;
160 
161 	/* get wc */
162 	if ((ei->mode & _MODE_UTF32) == 0) {
163 		/* UTF16 */
164 		if (needlen == 2) {
165 			switch (endian) {
166 			case _ENDIAN_LITTLE:
167 				wc = (psenc->ch[0] |
168 				    ((wchar_t)psenc->ch[1] << 8));
169 				break;
170 			case _ENDIAN_BIG:
171 				wc = (psenc->ch[1] |
172 				    ((wchar_t)psenc->ch[0] << 8));
173 				break;
174 			default:
175 				goto ilseq;
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[3] < 0xDC || psenc->ch[3] > 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[2]<0xDC || psenc->ch[2]>0xDF)
195 					goto ilseq;
196 				wc |= psenc->ch[3];
197 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
198 				break;
199 			default:
200 				goto ilseq;
201 			}
202 			wc += 0x10000;
203 		}
204 	} else {
205 		/* UTF32 */
206 		switch (endian) {
207 		case _ENDIAN_LITTLE:
208 			wc = (psenc->ch[0] |
209 			    ((wchar_t)psenc->ch[1] << 8) |
210 			    ((wchar_t)psenc->ch[2] << 16) |
211 			    ((wchar_t)psenc->ch[3] << 24));
212 			break;
213 		case _ENDIAN_BIG:
214 			wc = (psenc->ch[3] |
215 			    ((wchar_t)psenc->ch[2] << 8) |
216 			    ((wchar_t)psenc->ch[1] << 16) |
217 			    ((wchar_t)psenc->ch[0] << 24));
218 			break;
219 		default:
220 			goto ilseq;
221 		}
222 		if (wc >= 0xD800 && wc <= 0xDFFF)
223 			goto ilseq;
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, size_t *nresult)
249 {
250 	wchar_t wc2;
251 	static const char _bom[4] = {
252 	    0x00, 0x00, 0xFE, 0xFF,
253 	};
254 	const char *bom = &_bom[0];
255 	size_t cnt;
256 
257 	cnt = (size_t)0;
258 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
259 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
260 			if (ei->mode & _MODE_UTF32)
261 				cnt = 4;
262 			else {
263 				cnt = 2;
264 				bom += 2;
265 			}
266 			if (n < cnt)
267 				goto e2big;
268 			memcpy(s, bom, cnt);
269 			s += cnt, n -= cnt;
270 		}
271 		psenc->current_endian = ei->preffered_endian;
272 	}
273 
274 	wc2 = 0;
275 	if ((ei->mode & _MODE_UTF32)==0) {
276 		/* UTF16 */
277 		if (wc > 0xFFFF) {
278 			/* surrogate */
279 			if (wc > 0x10FFFF)
280 				goto ilseq;
281 			if (n < 4)
282 				goto e2big;
283 			cnt += 4;
284 			wc -= 0x10000;
285 			wc2 = (wc & 0x3FF) | 0xDC00;
286 			wc = (wc>>10) | 0xD800;
287 		} else {
288 			if (n < 2)
289 				goto e2big;
290 			cnt += 2;
291 		}
292 
293 surrogate:
294 		switch (psenc->current_endian) {
295 		case _ENDIAN_BIG:
296 			s[1] = wc;
297 			s[0] = (wc >>= 8);
298 			break;
299 		case _ENDIAN_LITTLE:
300 			s[0] = wc;
301 			s[1] = (wc >>= 8);
302 			break;
303 		}
304 		if (wc2 != 0) {
305 			wc = wc2;
306 			wc2 = 0;
307 			s += 2;
308 			goto surrogate;
309 		}
310 	} else {
311 		/* UTF32 */
312 		if (wc >= 0xD800 && wc <= 0xDFFF)
313 			goto ilseq;
314 		if (n < 4)
315 			goto e2big;
316 		cnt += 4;
317 		switch (psenc->current_endian) {
318 		case _ENDIAN_BIG:
319 			s[3] = wc;
320 			s[2] = (wc >>= 8);
321 			s[1] = (wc >>= 8);
322 			s[0] = (wc >>= 8);
323 			break;
324 		case _ENDIAN_LITTLE:
325 			s[0] = wc;
326 			s[1] = (wc >>= 8);
327 			s[2] = (wc >>= 8);
328 			s[3] = (wc >>= 8);
329 			break;
330 		}
331 	}
332 	*nresult = cnt;
333 
334 	return (0);
335 
336 ilseq:
337 	*nresult = (size_t)-1;
338 	return (EILSEQ);
339 e2big:
340 	*nresult = (size_t)-1;
341 	return (E2BIG);
342 }
343 
344 static void
345 parse_variable(_UTF1632EncodingInfo * __restrict ei,
346     const void * __restrict var, size_t lenvar)
347 {
348 	const char *p;
349 
350 	p = var;
351 	while (lenvar > 0) {
352 		switch (*p) {
353 		case 'B':
354 		case 'b':
355 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
356 			break;
357 		case 'L':
358 		case 'l':
359 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
360 			break;
361 		case 'i':
362 		case 'I':
363 			MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
364 			break;
365 		case 's':
366 		case 'S':
367 			MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
368 			break;
369 		case 'F':
370 		case 'f':
371 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
372 			break;
373 		case 'U':
374 		case 'u':
375 			MATCH(utf32, ei->mode |= _MODE_UTF32);
376 			break;
377 		}
378 		p++;
379 		lenvar--;
380 	}
381 }
382 
383 static int
384 /*ARGSUSED*/
385 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
386     const void * __restrict var, size_t lenvar)
387 {
388 
389 	memset((void *)ei, 0, sizeof(*ei));
390 
391 	parse_variable(ei, var, lenvar);
392 
393 	ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
394 	/* 6: endian + surrogate */
395 	/* 8: endian + normal */
396 
397 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
398 		ei->preffered_endian = _ENDIAN_BIG;
399 	}
400 
401 	return (0);
402 }
403 
404 static void
405 /*ARGSUSED*/
406 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
407 {
408 
409 }
410 
411 static __inline int
412 /*ARGSUSED*/
413 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
414      _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
415 {
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 __unused,
426     _wc_t * __restrict wc, _csid_t csid, _index_t idx)
427 {
428 
429 	if (csid != 0)
430 		return (EILSEQ);
431 
432 	*wc = (_wc_t)idx;
433 
434 	return (0);
435 }
436 
437 static __inline int
438 /*ARGSUSED*/
439 _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
440     _UTF1632State * __restrict psenc, int * __restrict rstate)
441 {
442 
443 	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
444 	    _STDENC_SDGEN_INCOMPLETE_CHAR;
445 	return (0);
446 }
447 
448 /* ----------------------------------------------------------------------
449  * public interface for stdenc
450  */
451 
452 _CITRUS_STDENC_DECLS(UTF1632);
453 _CITRUS_STDENC_DEF_OPS(UTF1632);
454 
455 #include "citrus_stdenc_template.h"
456