xref: /netbsd/lib/libc/citrus/modules/citrus_utf7.c (revision 6550d01e)
1 /*	$NetBSD: citrus_utf7.c,v 1.5 2006/08/23 12:57:24 tnozaki Exp $	*/
2 
3 /*-
4  * Copyright (c)2004, 2005 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 
30 #include <sys/cdefs.h>
31 #if defined(LIB_SCCS) && !defined(lint)
32 __RCSID("$NetBSD: citrus_utf7.c,v 1.5 2006/08/23 12:57:24 tnozaki Exp $");
33 #endif /* LIB_SCCS and not lint */
34 
35 #include <assert.h>
36 #include <errno.h>
37 #include <string.h>
38 #include <stdio.h>
39 #include <stdint.h>
40 #include <stdlib.h>
41 #include <limits.h>
42 #include <wchar.h>
43 
44 #include "citrus_namespace.h"
45 #include "citrus_types.h"
46 #include "citrus_module.h"
47 #include "citrus_ctype.h"
48 #include "citrus_stdenc.h"
49 #include "citrus_utf7.h"
50 
51 /* ----------------------------------------------------------------------
52  * private stuffs used by templates
53  */
54 
55 typedef struct {
56 	uint16_t	cell[0x80];
57 #define	EI_MASK		UINT16_C(0xff)
58 #define EI_DIRECT	UINT16_C(0x100)
59 #define EI_OPTION	UINT16_C(0x200)
60 #define EI_SPACE	UINT16_C(0x400)
61 } _UTF7EncodingInfo;
62 
63 typedef struct {
64 	unsigned int
65 		mode: 1,	/* whether base64 mode */
66 		bits: 4,	/* need to hold 0 - 15 */
67 		cache: 22,	/* 22 = BASE64_BIT + UTF16_BIT */
68 		surrogate: 1;	/* whether surrogate pair or not */
69 	int chlen;
70 	char ch[4]; /* BASE64_IN, 3 * 6 = 18, most closed to UTF16_BIT */
71 } _UTF7State;
72 
73 typedef struct {
74 	_UTF7EncodingInfo	ei;
75 	struct {
76 		/* for future multi-locale facility */
77 		_UTF7State	s_mblen;
78 		_UTF7State	s_mbrlen;
79 		_UTF7State	s_mbrtowc;
80 		_UTF7State	s_mbtowc;
81 		_UTF7State	s_mbsrtowcs;
82 		_UTF7State	s_wcrtomb;
83 		_UTF7State	s_wcsrtombs;
84 		_UTF7State	s_wctomb;
85 	} states;
86 } _UTF7CTypeInfo;
87 
88 #define	_CEI_TO_EI(_cei_)		(&(_cei_)->ei)
89 #define	_CEI_TO_STATE(_cei_, _func_)	(_cei_)->states.s_##_func_
90 
91 #define	_FUNCNAME(m)			_citrus_UTF7_##m
92 #define	_ENCODING_INFO			_UTF7EncodingInfo
93 #define	_CTYPE_INFO			_UTF7CTypeInfo
94 #define	_ENCODING_STATE			_UTF7State
95 #define	_ENCODING_MB_CUR_MAX(_ei_)		4
96 #define	_ENCODING_IS_STATE_DEPENDENT		1
97 #define	_STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
98 
99 static __inline void
100 /*ARGSUSED*/
101 _citrus_UTF7_init_state(_UTF7EncodingInfo * __restrict ei,
102 	_UTF7State * __restrict s)
103 {
104 	/* ei appears to be unused */
105 	_DIAGASSERT(s != NULL);
106 
107 	memset((void *)s, 0, sizeof(*s));
108 }
109 
110 static __inline void
111 /*ARGSUSED*/
112 _citrus_UTF7_pack_state(_UTF7EncodingInfo * __restrict ei,
113 	void *__restrict pspriv, const _UTF7State * __restrict s)
114 {
115 	/* ei seem to be unused */
116 	_DIAGASSERT(pspriv != NULL);
117 	_DIAGASSERT(s != NULL);
118 
119 	memcpy(pspriv, (const void *)s, sizeof(*s));
120 }
121 
122 static __inline void
123 /*ARGSUSED*/
124 _citrus_UTF7_unpack_state(_UTF7EncodingInfo * __restrict ei,
125 	_UTF7State * __restrict s, const void * __restrict pspriv)
126 {
127 	/* ei seem to be unused */
128 	_DIAGASSERT(s != NULL);
129 	_DIAGASSERT(pspriv != NULL);
130 
131 	memcpy((void *)s, pspriv, sizeof(*s));
132 }
133 
134 static const char base64[] =
135 	"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
136 	"abcdefghijklmnopqrstuvwxyz"
137 	"0123456789+/";
138 
139 static const char direct[] =
140 	"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
141 	"abcdefghijklmnopqrstuvwxyz"
142 	"0123456789(),-./:?";
143 
144 static const char option[] = "!\"#$%&';<=>@[]^_`{|}";
145 static const char spaces[] = " \t\r\n";
146 
147 #define	BASE64_BIT	6
148 #define	UTF16_BIT	16
149 
150 #define	BASE64_MAX	0x3f
151 #define	UTF16_MAX	UINT16_C(0xffff)
152 #define	UTF32_MAX	UINT32_C(0x10ffff)
153 
154 #define	BASE64_IN	'+'
155 #define	BASE64_OUT	'-'
156 
157 #define	SHIFT7BIT(c)	((c) >> 7)
158 #define	ISSPECIAL(c)	((c) == '\0' || (c) == BASE64_IN)
159 
160 #define	FINDLEN(ei, c) \
161 	(SHIFT7BIT((c)) ? -1 : (((ei)->cell[(c)] & EI_MASK) - 1))
162 
163 #define	ISDIRECT(ei, c)	(!SHIFT7BIT((c)) && (ISSPECIAL((c)) || \
164 	ei->cell[(c)] & (EI_DIRECT | EI_OPTION | EI_SPACE)))
165 
166 #define	ISSAFE(ei, c)	(!SHIFT7BIT((c)) && (ISSPECIAL((c)) || \
167 	(c < 0x80 && ei->cell[(c)] & (EI_DIRECT | EI_SPACE))))
168 
169 /* surrogate pair */
170 #define	SRG_BASE	UINT32_C(0x10000)
171 #define	HISRG_MIN	UINT16_C(0xd800)
172 #define	HISRG_MAX	UINT16_C(0xdbff)
173 #define	LOSRG_MIN	UINT16_C(0xdc00)
174 #define	LOSRG_MAX	UINT16_C(0xdfff)
175 
176 static int
177 _citrus_UTF7_mbtoutf16(_UTF7EncodingInfo * __restrict ei,
178 	uint16_t * __restrict u16, const char ** __restrict s, size_t n,
179 	_UTF7State * __restrict psenc, size_t * __restrict nresult)
180 {
181 	_UTF7State sv;
182 	const char *s0;
183 	int i, done, len;
184 
185 	_DIAGASSERT(ei != NULL);
186 	_DIAGASSERT(s != NULL && *s != NULL);
187 	_DIAGASSERT(psenc != NULL);
188 
189 	s0 = *s;
190 	sv = *psenc;
191 
192 	for (i = 0, done = 0; done == 0; i++) {
193 		_DIAGASSERT(i <= psenc->chlen);
194 		if (i == psenc->chlen) {
195 			if (n-- < 1) {
196 				*nresult = (size_t)-2;
197 				*s = s0;
198 				sv.chlen = psenc->chlen;
199 				*psenc = sv;
200 				return 0;
201 			}
202 			psenc->ch[psenc->chlen++] = *s0++;
203 		}
204 		if (SHIFT7BIT((int)psenc->ch[i]))
205 			goto ilseq;
206 		if (!psenc->mode) {
207 			if (psenc->bits > 0 || psenc->cache > 0)
208 				return EINVAL;
209 			if (psenc->ch[i] == BASE64_IN) {
210 				psenc->mode = 1;
211 			} else {
212 				if (!ISDIRECT(ei, (int)psenc->ch[i]))
213 					goto ilseq;
214 				*u16 = (uint16_t)psenc->ch[i];
215 				done = 1;
216 				continue;
217 			}
218 		} else {
219 			if (psenc->ch[i] == BASE64_OUT && psenc->cache == 0) {
220 				psenc->mode = 0;
221 				*u16 = (uint16_t)BASE64_IN;
222 				done = 1;
223 				continue;
224 			}
225 			len = FINDLEN(ei, (int)psenc->ch[i]);
226 			if (len < 0) {
227 				if (psenc->bits >= BASE64_BIT)
228 					return EINVAL;
229 				psenc->mode = 0;
230 				psenc->bits = psenc->cache = 0;
231 				if (psenc->ch[i] != BASE64_OUT) {
232 					if (!ISDIRECT(ei, (int)psenc->ch[i]))
233 						goto ilseq;
234 					*u16 = (uint16_t)psenc->ch[i];
235 					done = 1;
236 				}
237 			} else {
238 				psenc->cache =
239 				    (psenc->cache << BASE64_BIT) | len;
240 				switch (psenc->bits) {
241 				case 0: case 2: case 4: case 6: case 8:
242 					psenc->bits += BASE64_BIT;
243 					break;
244 				case 10: case 12: case 14:
245 					psenc->bits -= (UTF16_BIT - BASE64_BIT);
246 					*u16 = (psenc->cache >> psenc->bits)
247 					    & UTF16_MAX;
248 					done = 1;
249 					break;
250 				default:
251 					return EINVAL;
252 				}
253 			}
254 		}
255 	}
256 
257 	if (psenc->chlen > i)
258 		return EINVAL;
259 	psenc->chlen = 0;
260 	*nresult = (size_t)((*u16 == 0) ? 0 : s0 - *s);
261 	*s = s0;
262 
263 	return 0;
264 
265 ilseq:
266 	*nresult = (size_t)-1;
267 	return EILSEQ;
268 }
269 
270 static int
271 _citrus_UTF7_mbrtowc_priv(_UTF7EncodingInfo * __restrict ei,
272 	wchar_t * __restrict pwc, const char ** __restrict s, size_t n,
273 	_UTF7State * __restrict psenc, size_t * __restrict nresult)
274 {
275 	const char *s0;
276 	uint32_t u32;
277 	uint16_t hi, lo;
278 	size_t siz, nr;
279 	int err;
280 
281 	_DIAGASSERT(ei != NULL);
282 	/* pwc may be null */
283 	_DIAGASSERT(s != NULL);
284 	_DIAGASSERT(psenc != NULL);
285 
286 	if (*s == NULL) {
287 		_citrus_UTF7_init_state(ei, psenc);
288 		*nresult = (size_t)_ENCODING_IS_STATE_DEPENDENT;
289 		return 0;
290 	}
291 	s0 = *s;
292 	if (psenc->surrogate) {
293 		hi = (psenc->cache >> 2) & UTF16_MAX;
294 		if (hi < HISRG_MIN || hi > HISRG_MAX)
295 			return EINVAL;
296 		siz = 0;
297 	} else {
298 		err = _citrus_UTF7_mbtoutf16(ei, &hi, &s0, n, psenc, &nr);
299 		if (nr == (size_t)-1 || nr == (size_t)-2) {
300 			*nresult = nr;
301 			return err;
302 		}
303 		if (err != 0)
304 			return err;
305 		n -= nr;
306 		siz = nr;
307 		if (hi < HISRG_MIN || hi > HISRG_MAX) {
308 			u32 = (uint32_t)hi;
309 			goto done;
310 		}
311 		psenc->surrogate = 1;
312 	}
313 	err = _citrus_UTF7_mbtoutf16(ei, &lo, &s0, n, psenc, &nr);
314 	if (nr == (size_t)-1 || nr == (size_t)-2) {
315 		*nresult = nr;
316 		return err;
317 	}
318 	if (err != 0)
319 		return err;
320 	hi -= HISRG_MIN;
321 	lo -= LOSRG_MIN;
322 	u32 = (hi << 10 | lo) + SRG_BASE;
323 	siz += nr;
324 done:
325 	*s = s0;
326 	if (pwc != NULL)
327 		*pwc = (wchar_t)u32;
328 	if (u32 == (uint32_t)0) {
329 		*nresult = (size_t)0;
330 		_citrus_UTF7_init_state(ei, psenc);
331 	} else {
332 		*nresult = siz;
333 		psenc->surrogate = 0;
334 	}
335 	return err;
336 }
337 
338 static int
339 _citrus_UTF7_utf16tomb(_UTF7EncodingInfo * __restrict ei,
340 	char * __restrict s, size_t n, uint16_t u16,
341 	_UTF7State * __restrict psenc, size_t * __restrict nresult)
342 {
343 	int bits, i;
344 
345 	_DIAGASSERT(ei != NULL);
346 	_DIAGASSERT(psenc != NULL);
347 
348 	if (psenc->chlen != 0 || psenc->bits > BASE64_BIT)
349 		return EINVAL;
350 
351 	if (ISSAFE(ei, u16)) {
352 		if (psenc->mode) {
353 			if (psenc->bits > 0) {
354 				bits = BASE64_BIT - psenc->bits;
355 				i = (psenc->cache << bits) & BASE64_MAX;
356 				psenc->ch[psenc->chlen++] = base64[i];
357 				psenc->bits = psenc->cache = 0;
358 			}
359 			if (u16 == BASE64_OUT || FINDLEN(ei, u16) >= 0)
360 				psenc->ch[psenc->chlen++] = BASE64_OUT;
361 			psenc->mode = 0;
362 		}
363 		if (psenc->bits != 0)
364 			return EINVAL;
365 		psenc->ch[psenc->chlen++] = (char)u16;
366 		if (u16 == BASE64_IN)
367 			psenc->ch[psenc->chlen++] = BASE64_OUT;
368 	} else {
369 		if (!psenc->mode) {
370 			if (psenc->bits > 0)
371 				return EINVAL;
372 			psenc->ch[psenc->chlen++] = BASE64_IN;
373 			psenc->mode = 1;
374 		}
375 		psenc->cache = (psenc->cache << UTF16_BIT) | u16;
376 		bits = UTF16_BIT + psenc->bits;
377 		psenc->bits = bits % BASE64_BIT;
378 		while ((bits -= BASE64_BIT) >= 0) {
379 			i = (psenc->cache >> bits) & BASE64_MAX;
380 			psenc->ch[psenc->chlen++] = base64[i];
381 		}
382 	}
383 	memcpy(s, psenc->ch, psenc->chlen);
384 	*nresult = psenc->chlen;
385 	psenc->chlen = 0;
386 
387 	return 0;
388 }
389 
390 static int
391 _citrus_UTF7_wcrtomb_priv(_UTF7EncodingInfo * __restrict ei,
392 	char * __restrict s, size_t n, wchar_t wchar,
393 	_UTF7State * __restrict psenc, size_t * __restrict nresult)
394 {
395 	uint32_t u32;
396 	uint16_t u16[2];
397 	int err, len, i;
398 	size_t siz, nr;
399 
400 	_DIAGASSERT(ei != NULL);
401 	_DIAGASSERT(s != NULL);
402 	_DIAGASSERT(psenc != NULL);
403 	_DIAGASSERT(nresult != NULL);
404 
405 	u32 = (uint32_t)wchar;
406 	if (u32 <= UTF16_MAX) {
407 		u16[0] = (uint16_t)u32;
408 		len = 1;
409 	} else if (u32 <= UTF32_MAX) {
410 		u32 -= SRG_BASE;
411 		u16[0] = (u32 >> 10) + HISRG_MIN;
412 		u16[1] = ((uint16_t)(u32 & UINT32_C(0x3ff))) + LOSRG_MIN;
413 		len = 2;
414 	} else {
415 		*nresult = (size_t)-1;
416 		return EILSEQ;
417 	}
418 	siz = 0;
419 	for (i = 0; i < len; ++i) {
420 		err = _citrus_UTF7_utf16tomb(ei, s, n, u16[i], psenc, &nr);
421 		if (err != 0)
422 			return err; /* XXX: state has been modified */
423 		s += nr;
424 		n -= nr;
425 		siz += nr;
426 	}
427 	*nresult = siz;
428 
429 	return 0;
430 }
431 
432 static int
433 /* ARGSUSED */
434 _citrus_UTF7_put_state_reset(_UTF7EncodingInfo * __restrict ei,
435 	char * __restrict s, size_t n, _UTF7State * __restrict psenc,
436 	size_t * __restrict nresult)
437 {
438 	int bits, pos;
439 
440 	_DIAGASSERT(ei != NULL);
441 	_DIAGASSERT(s != NULL);
442 	_DIAGASSERT(psenc != NULL);
443 	_DIAGASSERT(nresult != NULL);
444 
445 	if (psenc->chlen != 0 || psenc->bits > BASE64_BIT || psenc->surrogate)
446 		return EINVAL;
447 
448 	if (psenc->mode) {
449 		if (psenc->bits > 0) {
450 			if (n-- < 1)
451 				return E2BIG;
452 			bits = BASE64_BIT - psenc->bits;
453 			pos = (psenc->cache << bits) & BASE64_MAX;
454 			psenc->ch[psenc->chlen++] = base64[pos];
455 			psenc->ch[psenc->chlen++] = BASE64_OUT;
456 			psenc->bits = psenc->cache = 0;
457 		}
458 		psenc->mode = 0;
459 	}
460 	if (psenc->bits != 0)
461 		return EINVAL;
462 	if (n-- < 1)
463 		return E2BIG;
464 
465 	_DIAGASSERT(n >= psenc->chlen);
466 	*nresult = (size_t)psenc->chlen;
467 	if (psenc->chlen > 0) {
468 		memcpy(s, psenc->ch, psenc->chlen);
469 		psenc->chlen = 0;
470 	}
471 
472 	return 0;
473 }
474 
475 static __inline int
476 /*ARGSUSED*/
477 _citrus_UTF7_stdenc_wctocs(_UTF7EncodingInfo * __restrict ei,
478 			   _csid_t * __restrict csid,
479 			   _index_t * __restrict idx, wchar_t wc)
480 {
481 	/* ei seem to be unused */
482 	_DIAGASSERT(csid != NULL);
483 	_DIAGASSERT(idx != NULL);
484 
485 	*csid = 0;
486 	*idx = (_index_t)wc;
487 
488 	return 0;
489 }
490 
491 static __inline int
492 /*ARGSUSED*/
493 _citrus_UTF7_stdenc_cstowc(_UTF7EncodingInfo * __restrict ei,
494 			   wchar_t * __restrict wc,
495 			   _csid_t csid, _index_t idx)
496 {
497 	/* ei seem to be unused */
498 	_DIAGASSERT(wc != NULL);
499 
500 	if (csid != 0)
501 		return EILSEQ;
502 	*wc = (wchar_t)idx;
503 
504 	return 0;
505 }
506 
507 static __inline int
508 /*ARGSUSED*/
509 _citrus_UTF7_stdenc_get_state_desc_generic(_UTF7EncodingInfo * __restrict ei,
510 					   _UTF7State * __restrict psenc,
511 					   int * __restrict rstate)
512 {
513 
514 	if (psenc->chlen == 0)
515 		*rstate = _STDENC_SDGEN_INITIAL;
516 	else
517 		*rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
518 
519 	return 0;
520 }
521 
522 static void
523 /*ARGSUSED*/
524 _citrus_UTF7_encoding_module_uninit(_UTF7EncodingInfo *ei)
525 {
526 	/* ei seems to be unused */
527 }
528 
529 static int
530 /*ARGSUSED*/
531 _citrus_UTF7_encoding_module_init(_UTF7EncodingInfo * __restrict ei,
532 				  const void * __restrict var, size_t lenvar)
533 {
534 	const char *s;
535 
536 	_DIAGASSERT(ei != NULL);
537 	/* var may be null */
538 
539 	memset(ei, 0, sizeof(*ei));
540 
541 #define FILL(str, flag)				\
542 do {						\
543 	for (s = str; *s != '\0'; s++)		\
544 		ei->cell[*s & 0x7f] |= flag;	\
545 } while (/*CONSTCOND*/0)
546 
547 	FILL(base64, (s - base64) + 1);
548 	FILL(direct, EI_DIRECT);
549 	FILL(option, EI_OPTION);
550 	FILL(spaces, EI_SPACE);
551 
552 	return 0;
553 }
554 
555 /* ----------------------------------------------------------------------
556  * public interface for ctype
557  */
558 
559 _CITRUS_CTYPE_DECLS(UTF7);
560 _CITRUS_CTYPE_DEF_OPS(UTF7);
561 
562 #include "citrus_ctype_template.h"
563 
564 /* ----------------------------------------------------------------------
565  * public interface for stdenc
566  */
567 
568 _CITRUS_STDENC_DECLS(UTF7);
569 _CITRUS_STDENC_DEF_OPS(UTF7);
570 
571 #include "citrus_stdenc_template.h"
572