xref: /freebsd/lib/libc/locale/utf8.c (revision 6fd05b64b5b65dd4ba9b86482a0634a5f0b96c29)
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 <runetype.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <wchar.h>
35 #include "mblocal.h"
36 
37 size_t	_UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict, size_t,
38 	    mbstate_t * __restrict);
39 int	_UTF8_mbsinit(const mbstate_t *);
40 size_t	_UTF8_wcrtomb(char * __restrict, wchar_t, mbstate_t * __restrict);
41 
42 typedef struct {
43 	wchar_t	ch;
44 	int	want;
45 	wchar_t	lbound;
46 } _UTF8State;
47 
48 int
49 _UTF8_init(_RuneLocale *rl)
50 {
51 
52 	__mbrtowc = _UTF8_mbrtowc;
53 	__wcrtomb = _UTF8_wcrtomb;
54 	__mbsinit = _UTF8_mbsinit;
55 	_CurrentRuneLocale = rl;
56 	__mb_cur_max = 6;
57 
58 	return (0);
59 }
60 
61 int
62 _UTF8_mbsinit(const mbstate_t *ps)
63 {
64 
65 	return (ps == NULL || ((const _UTF8State *)ps)->want == 0);
66 }
67 
68 size_t
69 _UTF8_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, size_t n,
70     mbstate_t * __restrict ps)
71 {
72 	_UTF8State *us;
73 	int ch, i, mask, want;
74 	wchar_t lbound, wch;
75 
76 	us = (_UTF8State *)ps;
77 
78 	if (us->want < 0 || us->want > 6) {
79 		errno = EINVAL;
80 		return ((size_t)-1);
81 	}
82 
83 	if (s == NULL) {
84 		s = "";
85 		n = 1;
86 		pwc = NULL;
87 	}
88 
89 	if (n == 0)
90 		/* Incomplete multibyte sequence */
91 		return ((size_t)-2);
92 
93 	if (us->want == 0 && ((ch = (unsigned char)*s) & ~0x7f) == 0) {
94 		/* Fast path for plain ASCII characters. */
95 		if (pwc != NULL)
96 			*pwc = ch;
97 		return (ch != '\0' ? 1 : 0);
98 	}
99 
100 	if (us->want == 0) {
101 		/*
102 		 * Determine the number of octets that make up this character
103 		 * from the first octet, and a mask that extracts the
104 		 * interesting bits of the first octet. We already know
105 		 * the character is at least two bytes long.
106 		 *
107 		 * We also specify a lower bound for the character code to
108 		 * detect redundant, non-"shortest form" encodings. For
109 		 * example, the sequence C0 80 is _not_ a legal representation
110 		 * of the null character. This enforces a 1-to-1 mapping
111 		 * between character codes and their multibyte representations.
112 		 */
113 		ch = (unsigned char)*s;
114 		if ((ch & 0x80) == 0) {
115 			mask = 0x7f;
116 			want = 1;
117 			lbound = 0;
118 		} else if ((ch & 0xe0) == 0xc0) {
119 			mask = 0x1f;
120 			want = 2;
121 			lbound = 0x80;
122 		} else if ((ch & 0xf0) == 0xe0) {
123 			mask = 0x0f;
124 			want = 3;
125 			lbound = 0x800;
126 		} else if ((ch & 0xf8) == 0xf0) {
127 			mask = 0x07;
128 			want = 4;
129 			lbound = 0x10000;
130 		} else if ((ch & 0xfc) == 0xf8) {
131 			mask = 0x03;
132 			want = 5;
133 			lbound = 0x200000;
134 		} else if ((ch & 0xfc) == 0xfc) {
135 			mask = 0x01;
136 			want = 6;
137 			lbound = 0x4000000;
138 		} else {
139 			/*
140 			 * Malformed input; input is not UTF-8.
141 			 */
142 			errno = EILSEQ;
143 			return ((size_t)-1);
144 		}
145 	} else {
146 		want = us->want;
147 		lbound = us->lbound;
148 	}
149 
150 	/*
151 	 * Decode the octet sequence representing the character in chunks
152 	 * of 6 bits, most significant first.
153 	 */
154 	if (us->want == 0)
155 		wch = (unsigned char)*s++ & mask;
156 	else
157 		wch = us->ch;
158 	for (i = (us->want == 0) ? 1 : 0; i < MIN(want, n); i++) {
159 		if ((*s & 0xc0) != 0x80) {
160 			/*
161 			 * Malformed input; bad characters in the middle
162 			 * of a character.
163 			 */
164 			errno = EILSEQ;
165 			return ((size_t)-1);
166 		}
167 		wch <<= 6;
168 		wch |= *s++ & 0x3f;
169 	}
170 	if (i < want) {
171 		/* Incomplete multibyte sequence. */
172 		us->want = want - i;
173 		us->lbound = lbound;
174 		us->ch = wch;
175 		return ((size_t)-2);
176 	}
177 	if (wch < lbound) {
178 		/*
179 		 * Malformed input; redundant encoding.
180 		 */
181 		errno = EILSEQ;
182 		return ((size_t)-1);
183 	}
184 	if (pwc != NULL)
185 		*pwc = wch;
186 	us->want = 0;
187 	return (wch == L'\0' ? 0 : want);
188 }
189 
190 size_t
191 _UTF8_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps)
192 {
193 	_UTF8State *us;
194 	unsigned char lead;
195 	int i, len;
196 
197 	us = (_UTF8State *)ps;
198 
199 	if (us->want != 0) {
200 		errno = EINVAL;
201 		return ((size_t)-1);
202 	}
203 
204 	if (s == NULL)
205 		/* Reset to initial shift state (no-op) */
206 		return (1);
207 
208 	if ((wc & ~0x7f) == 0) {
209 		/* Fast path for plain ASCII characters. */
210 		*s = (char)wc;
211 		return (1);
212 	}
213 
214 	/*
215 	 * Determine the number of octets needed to represent this character.
216 	 * We always output the shortest sequence possible. Also specify the
217 	 * first few bits of the first octet, which contains the information
218 	 * about the sequence length.
219 	 */
220 	if ((wc & ~0x7f) == 0) {
221 		lead = 0;
222 		len = 1;
223 	} else if ((wc & ~0x7ff) == 0) {
224 		lead = 0xc0;
225 		len = 2;
226 	} else if ((wc & ~0xffff) == 0) {
227 		lead = 0xe0;
228 		len = 3;
229 	} else if ((wc & ~0x1fffff) == 0) {
230 		lead = 0xf0;
231 		len = 4;
232 	} else if ((wc & ~0x3ffffff) == 0) {
233 		lead = 0xf8;
234 		len = 5;
235 	} else if ((wc & ~0x7fffffff) == 0) {
236 		lead = 0xfc;
237 		len = 6;
238 	} else {
239 		errno = EILSEQ;
240 		return ((size_t)-1);
241 	}
242 
243 	/*
244 	 * Output the octets representing the character in chunks
245 	 * of 6 bits, least significant last. The first octet is
246 	 * a special case because it contains the sequence length
247 	 * information.
248 	 */
249 	for (i = len - 1; i > 0; i--) {
250 		s[i] = (wc & 0x3f) | 0x80;
251 		wc >>= 6;
252 	}
253 	*s = (wc & 0xff) | lead;
254 
255 	return (len);
256 }
257