xref: /freebsd/lib/libc/locale/utf8.c (revision 2546665afcaf0d53dc2c7058fee96354b3680f5a)
1 /*-
2  * Copyright (c) 2002-2004 Tim J. Robbins
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24  * SUCH DAMAGE.
25  */
26 
27 #include <sys/param.h>
28 __FBSDID("$FreeBSD$");
29 
30 #include <errno.h>
31 #include <limits.h>
32 #include <runetype.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <wchar.h>
36 #include "mblocal.h"
37 
38 size_t	_UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t,
39 	    mbstate_t * __restrict);
40 int	_UTF8_mbsinit(const mbstate_t *);
41 size_t	_UTF8_mbsnrtowcs(wchar_t * __restrict, const char ** __restrict,
42 	    size_t, size_t, mbstate_t * __restrict);
43 size_t	_UTF8_wcrtomb(char * __restrict, wchar_t, mbstate_t * __restrict);
44 size_t	_UTF8_wcsnrtombs(char * __restrict, const wchar_t ** __restrict,
45 	    size_t, size_t, mbstate_t * __restrict);
46 
47 typedef struct {
48 	wchar_t	ch;
49 	int	want;
50 	wchar_t	lbound;
51 } _UTF8State;
52 
53 int
54 _UTF8_init(_RuneLocale *rl)
55 {
56 
57 	__mbrtowc = _UTF8_mbrtowc;
58 	__wcrtomb = _UTF8_wcrtomb;
59 	__mbsinit = _UTF8_mbsinit;
60 	__mbsnrtowcs = _UTF8_mbsnrtowcs;
61 	__wcsnrtombs = _UTF8_wcsnrtombs;
62 	_CurrentRuneLocale = rl;
63 	__mb_cur_max = 6;
64 
65 	return (0);
66 }
67 
68 int
69 _UTF8_mbsinit(const mbstate_t *ps)
70 {
71 
72 	return (ps == NULL || ((const _UTF8State *)ps)->want == 0);
73 }
74 
75 size_t
76 _UTF8_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, size_t n,
77     mbstate_t * __restrict ps)
78 {
79 	_UTF8State *us;
80 	int ch, i, mask, want;
81 	wchar_t lbound, wch;
82 
83 	us = (_UTF8State *)ps;
84 
85 	if (us->want < 0 || us->want > 6) {
86 		errno = EINVAL;
87 		return ((size_t)-1);
88 	}
89 
90 	if (s == NULL) {
91 		s = "";
92 		n = 1;
93 		pwc = NULL;
94 	}
95 
96 	if (n == 0)
97 		/* Incomplete multibyte sequence */
98 		return ((size_t)-2);
99 
100 	if (us->want == 0 && ((ch = (unsigned char)*s) & ~0x7f) == 0) {
101 		/* Fast path for plain ASCII characters. */
102 		if (pwc != NULL)
103 			*pwc = ch;
104 		return (ch != '\0' ? 1 : 0);
105 	}
106 
107 	if (us->want == 0) {
108 		/*
109 		 * Determine the number of octets that make up this character
110 		 * from the first octet, and a mask that extracts the
111 		 * interesting bits of the first octet. We already know
112 		 * the character is at least two bytes long.
113 		 *
114 		 * We also specify a lower bound for the character code to
115 		 * detect redundant, non-"shortest form" encodings. For
116 		 * example, the sequence C0 80 is _not_ a legal representation
117 		 * of the null character. This enforces a 1-to-1 mapping
118 		 * between character codes and their multibyte representations.
119 		 */
120 		ch = (unsigned char)*s;
121 		if ((ch & 0x80) == 0) {
122 			mask = 0x7f;
123 			want = 1;
124 			lbound = 0;
125 		} else if ((ch & 0xe0) == 0xc0) {
126 			mask = 0x1f;
127 			want = 2;
128 			lbound = 0x80;
129 		} else if ((ch & 0xf0) == 0xe0) {
130 			mask = 0x0f;
131 			want = 3;
132 			lbound = 0x800;
133 		} else if ((ch & 0xf8) == 0xf0) {
134 			mask = 0x07;
135 			want = 4;
136 			lbound = 0x10000;
137 		} else if ((ch & 0xfc) == 0xf8) {
138 			mask = 0x03;
139 			want = 5;
140 			lbound = 0x200000;
141 		} else if ((ch & 0xfc) == 0xfc) {
142 			mask = 0x01;
143 			want = 6;
144 			lbound = 0x4000000;
145 		} else {
146 			/*
147 			 * Malformed input; input is not UTF-8.
148 			 */
149 			errno = EILSEQ;
150 			return ((size_t)-1);
151 		}
152 	} else {
153 		want = us->want;
154 		lbound = us->lbound;
155 	}
156 
157 	/*
158 	 * Decode the octet sequence representing the character in chunks
159 	 * of 6 bits, most significant first.
160 	 */
161 	if (us->want == 0)
162 		wch = (unsigned char)*s++ & mask;
163 	else
164 		wch = us->ch;
165 	for (i = (us->want == 0) ? 1 : 0; i < MIN(want, n); i++) {
166 		if ((*s & 0xc0) != 0x80) {
167 			/*
168 			 * Malformed input; bad characters in the middle
169 			 * of a character.
170 			 */
171 			errno = EILSEQ;
172 			return ((size_t)-1);
173 		}
174 		wch <<= 6;
175 		wch |= *s++ & 0x3f;
176 	}
177 	if (i < want) {
178 		/* Incomplete multibyte sequence. */
179 		us->want = want - i;
180 		us->lbound = lbound;
181 		us->ch = wch;
182 		return ((size_t)-2);
183 	}
184 	if (wch < lbound) {
185 		/*
186 		 * Malformed input; redundant encoding.
187 		 */
188 		errno = EILSEQ;
189 		return ((size_t)-1);
190 	}
191 	if (pwc != NULL)
192 		*pwc = wch;
193 	us->want = 0;
194 	return (wch == L'\0' ? 0 : want);
195 }
196 
197 size_t
198 _UTF8_mbsnrtowcs(wchar_t * __restrict dst, const char ** __restrict src,
199     size_t nms, size_t len, mbstate_t * __restrict ps)
200 {
201 	_UTF8State *us;
202 	const char *s;
203 	size_t nchr;
204 	wchar_t wc;
205 	size_t nb;
206 
207 	us = (_UTF8State *)ps;
208 
209 	s = *src;
210 	nchr = 0;
211 
212 	if (dst == NULL) {
213 		/*
214 		 * The fast path in the loop below is not safe if an ASCII
215 		 * character appears as anything but the first byte of a
216 		 * multibyte sequence. Check now to avoid doing it in the loop.
217 		 */
218 		if (nms > 0 && us->want > 0 && (signed char)*s > 0) {
219 			errno = EILSEQ;
220 			return ((size_t)-1);
221 		}
222 		for (;;) {
223 			if (nms > 0 && (signed char)*s > 0)
224 				/*
225 				 * Fast path for plain ASCII characters
226 				 * excluding NUL.
227 				 */
228 				nb = 1;
229 			else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps)) ==
230 			    (size_t)-1)
231 				/* Invalid sequence - mbrtowc() sets errno. */
232 				return ((size_t)-1);
233 			else if (nb == 0 || nb == (size_t)-2)
234 				return (nchr);
235 			s += nb;
236 			nms -= nb;
237 			nchr++;
238 		}
239 		/*NOTREACHED*/
240 	}
241 
242 	/*
243 	 * The fast path in the loop below is not safe if an ASCII
244 	 * character appears as anything but the first byte of a
245 	 * multibyte sequence. Check now to avoid doing it in the loop.
246 	 */
247 	if (nms > 0 && len > 0 && us->want > 0 && (signed char)*s > 0) {
248 		errno = EILSEQ;
249 		return ((size_t)-1);
250 	}
251 	while (len-- > 0) {
252 		if (nms > 0 && (signed char)*s > 0) {
253 			/*
254 			 * Fast path for plain ASCII characters
255 			 * excluding NUL.
256 			 */
257 			*dst = (wchar_t)*s;
258 			nb = 1;
259 		} else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps)) ==
260 		    (size_t)-1) {
261 			*src = s;
262 			return ((size_t)-1);
263 		} else if (nb == (size_t)-2) {
264 			*src = s + nms;
265 			return (nchr);
266 		} else if (nb == 0) {
267 			*src = NULL;
268 			return (nchr);
269 		}
270 		s += nb;
271 		nms -= nb;
272 		nchr++;
273 		dst++;
274 	}
275 	*src = s;
276 	return (nchr);
277 }
278 
279 size_t
280 _UTF8_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps)
281 {
282 	_UTF8State *us;
283 	unsigned char lead;
284 	int i, len;
285 
286 	us = (_UTF8State *)ps;
287 
288 	if (us->want != 0) {
289 		errno = EINVAL;
290 		return ((size_t)-1);
291 	}
292 
293 	if (s == NULL)
294 		/* Reset to initial shift state (no-op) */
295 		return (1);
296 
297 	if ((wc & ~0x7f) == 0) {
298 		/* Fast path for plain ASCII characters. */
299 		*s = (char)wc;
300 		return (1);
301 	}
302 
303 	/*
304 	 * Determine the number of octets needed to represent this character.
305 	 * We always output the shortest sequence possible. Also specify the
306 	 * first few bits of the first octet, which contains the information
307 	 * about the sequence length.
308 	 */
309 	if ((wc & ~0x7f) == 0) {
310 		lead = 0;
311 		len = 1;
312 	} else if ((wc & ~0x7ff) == 0) {
313 		lead = 0xc0;
314 		len = 2;
315 	} else if ((wc & ~0xffff) == 0) {
316 		lead = 0xe0;
317 		len = 3;
318 	} else if ((wc & ~0x1fffff) == 0) {
319 		lead = 0xf0;
320 		len = 4;
321 	} else if ((wc & ~0x3ffffff) == 0) {
322 		lead = 0xf8;
323 		len = 5;
324 	} else if ((wc & ~0x7fffffff) == 0) {
325 		lead = 0xfc;
326 		len = 6;
327 	} else {
328 		errno = EILSEQ;
329 		return ((size_t)-1);
330 	}
331 
332 	/*
333 	 * Output the octets representing the character in chunks
334 	 * of 6 bits, least significant last. The first octet is
335 	 * a special case because it contains the sequence length
336 	 * information.
337 	 */
338 	for (i = len - 1; i > 0; i--) {
339 		s[i] = (wc & 0x3f) | 0x80;
340 		wc >>= 6;
341 	}
342 	*s = (wc & 0xff) | lead;
343 
344 	return (len);
345 }
346 
347 size_t
348 _UTF8_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src,
349     size_t nwc, size_t len, mbstate_t * __restrict ps)
350 {
351 	_UTF8State *us;
352 	char buf[MB_LEN_MAX];
353 	const wchar_t *s;
354 	size_t nbytes;
355 	size_t nb;
356 
357 	us = (_UTF8State *)ps;
358 
359 	if (us->want != 0) {
360 		errno = EINVAL;
361 		return ((size_t)-1);
362 	}
363 
364 	s = *src;
365 	nbytes = 0;
366 
367 	if (dst == NULL) {
368 		while (nwc-- > 0) {
369 			if (0 <= *s && *s < 0x80)
370 				/* Fast path for plain ASCII characters. */
371 				nb = 1;
372 			else if ((nb = _UTF8_wcrtomb(buf, *s, ps)) ==
373 			    (size_t)-1)
374 				/* Invalid character - wcrtomb() sets errno. */
375 				return ((size_t)-1);
376 			if (*s == L'\0')
377 				return (nbytes + nb - 1);
378 			s++;
379 			nbytes += nb;
380 		}
381 		return (nbytes);
382 	}
383 
384 	while (len > 0 && nwc-- > 0) {
385 		if (0 <= *s && *s < 0x80) {
386 			/* Fast path for plain ASCII characters. */
387 			nb = 1;
388 			*dst = *s;
389 		} else if (len > (size_t)MB_CUR_MAX) {
390 			/* Enough space to translate in-place. */
391 			if ((nb = (int)_UTF8_wcrtomb(dst, *s, ps)) < 0) {
392 				*src = s;
393 				return ((size_t)-1);
394 			}
395 		} else {
396 			/*
397 			 * May not be enough space; use temp. buffer.
398 			 */
399 			if ((nb = (int)_UTF8_wcrtomb(buf, *s, ps)) < 0) {
400 				*src = s;
401 				return ((size_t)-1);
402 			}
403 			if (nb > (int)len)
404 				/* MB sequence for character won't fit. */
405 				break;
406 			memcpy(dst, buf, nb);
407 		}
408 		if (*s == L'\0') {
409 			*src = NULL;
410 			return (nbytes + nb - 1);
411 		}
412 		s++;
413 		dst += nb;
414 		len -= nb;
415 		nbytes += nb;
416 	}
417 	*src = s;
418 	return (nbytes);
419 }
420