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 	unsigned int	 cur_min;
81 	uint32_t	 mode;
82 } _UTF1632EncodingInfo;
83 
84 #define _FUNCNAME(m)			_citrus_UTF1632_##m
85 #define _ENCODING_INFO			_UTF1632EncodingInfo
86 #define _ENCODING_STATE			_UTF1632State
87 #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
88 #define _ENCODING_MB_CUR_MIN(_ei_)	((_ei_)->cur_min)
89 #define _ENCODING_IS_STATE_DEPENDENT	0
90 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
91 
92 
93 static __inline void
94 /*ARGSUSED*/
95 _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
96     _UTF1632State *s)
97 {
98 
99 	memset(s, 0, sizeof(*s));
100 }
101 
102 static int
103 _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
104     char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
105 {
106 	char *s0;
107 	size_t result;
108 	wchar_t wc = L'\0';
109 	int chlenbak, endian, needlen;
110 
111 	s0 = *s;
112 
113 	if (s0 == NULL) {
114 		_citrus_UTF1632_init_state(ei, psenc);
115 		*nresult = 0; /* state independent */
116 		return (0);
117 	}
118 
119 	result = 0;
120 	chlenbak = psenc->chlen;
121 
122 refetch:
123 	needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 2;
124 
125 	while (chlenbak < needlen) {
126 		if (n == 0)
127 			goto restart;
128 		psenc->ch[chlenbak++] = *s0++;
129 		n--;
130 		result++;
131 	}
132 
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 	endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
160 	    psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
161 	    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[3] < 0xDC || psenc->ch[3] > 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[2]<0xDC || psenc->ch[2]>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, size_t *nresult)
251 {
252 	wchar_t wc2;
253 	static const char _bom[4] = {
254 	    0x00, 0x00, 0xFE, 0xFF,
255 	};
256 	const char *bom = &_bom[0];
257 	size_t cnt;
258 
259 	cnt = (size_t)0;
260 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
261 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
262 			if (ei->mode & _MODE_UTF32)
263 				cnt = 4;
264 			else {
265 				cnt = 2;
266 				bom += 2;
267 			}
268 			if (n < cnt)
269 				goto e2big;
270 			memcpy(s, bom, cnt);
271 			s += cnt, n -= cnt;
272 		}
273 		psenc->current_endian = ei->preffered_endian;
274 	}
275 
276 	wc2 = 0;
277 	if ((ei->mode & _MODE_UTF32)==0) {
278 		/* UTF16 */
279 		if (wc > 0xFFFF) {
280 			/* surrogate */
281 			if (wc > 0x10FFFF)
282 				goto ilseq;
283 			if (n < 4)
284 				goto e2big;
285 			cnt += 4;
286 			wc -= 0x10000;
287 			wc2 = (wc & 0x3FF) | 0xDC00;
288 			wc = (wc>>10) | 0xD800;
289 		} else {
290 			if (n < 2)
291 				goto e2big;
292 			cnt += 2;
293 		}
294 
295 surrogate:
296 		switch (psenc->current_endian) {
297 		case _ENDIAN_BIG:
298 			s[1] = wc;
299 			s[0] = (wc >>= 8);
300 			break;
301 		case _ENDIAN_LITTLE:
302 			s[0] = wc;
303 			s[1] = (wc >>= 8);
304 			break;
305 		}
306 		if (wc2 != 0) {
307 			wc = wc2;
308 			wc2 = 0;
309 			s += 2;
310 			goto surrogate;
311 		}
312 	} else {
313 		/* UTF32 */
314 		if (wc >= 0xD800 && wc <= 0xDFFF)
315 			goto ilseq;
316 		if (n < 4)
317 			goto e2big;
318 		cnt += 4;
319 		switch (psenc->current_endian) {
320 		case _ENDIAN_BIG:
321 			s[3] = wc;
322 			s[2] = (wc >>= 8);
323 			s[1] = (wc >>= 8);
324 			s[0] = (wc >>= 8);
325 			break;
326 		case _ENDIAN_LITTLE:
327 			s[0] = wc;
328 			s[1] = (wc >>= 8);
329 			s[2] = (wc >>= 8);
330 			s[3] = (wc >>= 8);
331 			break;
332 		}
333 	}
334 	*nresult = cnt;
335 
336 	return (0);
337 
338 ilseq:
339 	*nresult = (size_t)-1;
340 	return (EILSEQ);
341 e2big:
342 	*nresult = (size_t)-1;
343 	return (E2BIG);
344 }
345 
346 static void
347 parse_variable(_UTF1632EncodingInfo * __restrict ei,
348     const void * __restrict var, size_t lenvar)
349 {
350 	const char *p;
351 
352 	p = var;
353 	while (lenvar > 0) {
354 		switch (*p) {
355 		case 'B':
356 		case 'b':
357 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
358 			break;
359 		case 'L':
360 		case 'l':
361 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
362 			break;
363 		case 'i':
364 		case 'I':
365 			MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
366 			break;
367 		case 's':
368 		case 'S':
369 			MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
370 			break;
371 		case 'F':
372 		case 'f':
373 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
374 			break;
375 		case 'U':
376 		case 'u':
377 			MATCH(utf32, ei->mode |= _MODE_UTF32);
378 			break;
379 		}
380 		p++;
381 		lenvar--;
382 	}
383 }
384 
385 static int
386 /*ARGSUSED*/
387 _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
388     const void * __restrict var, size_t lenvar)
389 {
390 
391 	memset((void *)ei, 0, sizeof(*ei));
392 
393 	parse_variable(ei, var, lenvar);
394 
395 	ei->cur_min = ((ei->mode&_MODE_UTF32) == 0) ? 2 : 4;
396 	ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
397 	/* 6: endian + surrogate */
398 	/* 8: endian + normal */
399 
400 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
401 		ei->preffered_endian = _ENDIAN_BIG;
402 	}
403 
404 	return (0);
405 }
406 
407 static void
408 /*ARGSUSED*/
409 _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
410 {
411 
412 }
413 
414 static __inline int
415 /*ARGSUSED*/
416 _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
417      _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
418 {
419 
420 	*csid = 0;
421 	*idx = (_index_t)wc;
422 
423 	return (0);
424 }
425 
426 static __inline int
427 /*ARGSUSED*/
428 _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
429     _wc_t * __restrict wc, _csid_t csid, _index_t idx)
430 {
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 __unused,
443     _UTF1632State * __restrict psenc, int * __restrict rstate)
444 {
445 
446 	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
447 	    _STDENC_SDGEN_INCOMPLETE_CHAR;
448 	return (0);
449 }
450 
451 /* ----------------------------------------------------------------------
452  * public interface for stdenc
453  */
454 
455 _CITRUS_STDENC_DECLS(UTF1632);
456 _CITRUS_STDENC_DEF_OPS(UTF1632);
457 
458 #include "citrus_stdenc_template.h"
459