1 /*- 2 * Copyright (c) 2002-2004 Tim J. Robbins 3 * All rights reserved. 4 * 5 * Copyright (c) 2011 The FreeBSD Foundation 6 * All rights reserved. 7 * Portions of this software were developed by David Chisnall 8 * under sponsorship from the FreeBSD Foundation. 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 * $FreeBSD: head/lib/libc/locale/utf8.c 227753 2011-11-20 14:45:42Z theraven $ 32 */ 33 34 #include <sys/param.h> 35 36 #include <errno.h> 37 #include <limits.h> 38 #include <runetype.h> 39 #include <stdlib.h> 40 #include <string.h> 41 #include <wchar.h> 42 #include "mblocal.h" 43 44 extern int __mb_sb_limit; 45 46 static size_t _UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict, 47 size_t, mbstate_t * __restrict); 48 static int _UTF8_mbsinit(const mbstate_t *); 49 static size_t _UTF8_mbsnrtowcs(wchar_t * __restrict, 50 const char ** __restrict, size_t, size_t, 51 mbstate_t * __restrict); 52 static size_t _UTF8_wcrtomb(char * __restrict, wchar_t, 53 mbstate_t * __restrict); 54 static size_t _UTF8_wcsnrtombs(char * __restrict, const wchar_t ** __restrict, 55 size_t, size_t, mbstate_t * __restrict); 56 57 typedef struct { 58 wchar_t ch; 59 int want; 60 wchar_t lbound; 61 } _UTF8State; 62 63 int 64 _UTF8_init(struct xlocale_ctype *l, _RuneLocale *rl) 65 { 66 67 l->__mbrtowc = _UTF8_mbrtowc; 68 l->__wcrtomb = _UTF8_wcrtomb; 69 l->__mbsinit = _UTF8_mbsinit; 70 l->__mbsnrtowcs = _UTF8_mbsnrtowcs; 71 l->__wcsnrtombs = _UTF8_wcsnrtombs; 72 l->runes = rl; 73 l->__mb_cur_max = 6; 74 /* 75 * UCS-4 encoding used as the internal representation, so 76 * slots 0x0080-0x00FF are occuped and must be excluded 77 * from the single byte ctype by setting the limit. 78 */ 79 l->__mb_sb_limit = 128; 80 81 return (0); 82 } 83 84 static int 85 _UTF8_mbsinit(const mbstate_t *ps) 86 { 87 88 return (ps == NULL || ((const _UTF8State *)ps)->want == 0); 89 } 90 91 static size_t 92 _UTF8_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, size_t n, 93 mbstate_t * __restrict ps) 94 { 95 _UTF8State *us; 96 int ch, i, mask, want; 97 wchar_t lbound, wch; 98 99 us = (_UTF8State *)ps; 100 101 if (us->want < 0 || us->want > 6) { 102 errno = EINVAL; 103 return ((size_t)-1); 104 } 105 106 if (s == NULL) { 107 s = ""; 108 n = 1; 109 pwc = NULL; 110 } 111 112 if (n == 0) 113 /* Incomplete multibyte sequence */ 114 return ((size_t)-2); 115 116 if (us->want == 0 && ((ch = (unsigned char)*s) & ~0x7f) == 0) { 117 /* Fast path for plain ASCII characters. */ 118 if (pwc != NULL) 119 *pwc = ch; 120 return (ch != '\0' ? 1 : 0); 121 } 122 123 if (us->want == 0) { 124 /* 125 * Determine the number of octets that make up this character 126 * from the first octet, and a mask that extracts the 127 * interesting bits of the first octet. We already know 128 * the character is at least two bytes long. 129 * 130 * We also specify a lower bound for the character code to 131 * detect redundant, non-"shortest form" encodings. For 132 * example, the sequence C0 80 is _not_ a legal representation 133 * of the null character. This enforces a 1-to-1 mapping 134 * between character codes and their multibyte representations. 135 */ 136 ch = (unsigned char)*s; 137 if ((ch & 0x80) == 0) { 138 mask = 0x7f; 139 want = 1; 140 lbound = 0; 141 } else if ((ch & 0xe0) == 0xc0) { 142 mask = 0x1f; 143 want = 2; 144 lbound = 0x80; 145 } else if ((ch & 0xf0) == 0xe0) { 146 mask = 0x0f; 147 want = 3; 148 lbound = 0x800; 149 } else if ((ch & 0xf8) == 0xf0) { 150 mask = 0x07; 151 want = 4; 152 lbound = 0x10000; 153 } else if ((ch & 0xfc) == 0xf8) { 154 mask = 0x03; 155 want = 5; 156 lbound = 0x200000; 157 } else if ((ch & 0xfe) == 0xfc) { 158 mask = 0x01; 159 want = 6; 160 lbound = 0x4000000; 161 } else { 162 /* 163 * Malformed input; input is not UTF-8. 164 */ 165 errno = EILSEQ; 166 return ((size_t)-1); 167 } 168 } else { 169 want = us->want; 170 lbound = us->lbound; 171 } 172 173 /* 174 * Decode the octet sequence representing the character in chunks 175 * of 6 bits, most significant first. 176 */ 177 if (us->want == 0) 178 wch = (unsigned char)*s++ & mask; 179 else 180 wch = us->ch; 181 for (i = (us->want == 0) ? 1 : 0; i < MIN(want, n); i++) { 182 if ((*s & 0xc0) != 0x80) { 183 /* 184 * Malformed input; bad characters in the middle 185 * of a character. 186 */ 187 errno = EILSEQ; 188 return ((size_t)-1); 189 } 190 wch <<= 6; 191 wch |= *s++ & 0x3f; 192 } 193 if (i < want) { 194 /* Incomplete multibyte sequence. */ 195 us->want = want - i; 196 us->lbound = lbound; 197 us->ch = wch; 198 return ((size_t)-2); 199 } 200 if (wch < lbound) { 201 /* 202 * Malformed input; redundant encoding. 203 */ 204 errno = EILSEQ; 205 return ((size_t)-1); 206 } 207 if (pwc != NULL) 208 *pwc = wch; 209 us->want = 0; 210 return (wch == L'\0' ? 0 : want); 211 } 212 213 static size_t 214 _UTF8_mbsnrtowcs(wchar_t * __restrict dst, const char ** __restrict src, 215 size_t nms, size_t len, mbstate_t * __restrict ps) 216 { 217 _UTF8State *us; 218 const char *s; 219 size_t nchr; 220 wchar_t wc; 221 size_t nb; 222 223 us = (_UTF8State *)ps; 224 225 s = *src; 226 nchr = 0; 227 228 if (dst == NULL) { 229 /* 230 * The fast path in the loop below is not safe if an ASCII 231 * character appears as anything but the first byte of a 232 * multibyte sequence. Check now to avoid doing it in the loop. 233 */ 234 if (nms > 0 && us->want > 0 && (signed char)*s > 0) { 235 errno = EILSEQ; 236 return ((size_t)-1); 237 } 238 for (;;) { 239 if (nms > 0 && (signed char)*s > 0) 240 /* 241 * Fast path for plain ASCII characters 242 * excluding NUL. 243 */ 244 nb = 1; 245 else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps)) == 246 (size_t)-1) 247 /* Invalid sequence - mbrtowc() sets errno. */ 248 return ((size_t)-1); 249 else if (nb == 0 || nb == (size_t)-2) 250 return (nchr); 251 s += nb; 252 nms -= nb; 253 nchr++; 254 } 255 /*NOTREACHED*/ 256 } 257 258 /* 259 * The fast path in the loop below is not safe if an ASCII 260 * character appears as anything but the first byte of a 261 * multibyte sequence. Check now to avoid doing it in the loop. 262 */ 263 if (nms > 0 && len > 0 && us->want > 0 && (signed char)*s > 0) { 264 errno = EILSEQ; 265 return ((size_t)-1); 266 } 267 while (len-- > 0) { 268 if (nms > 0 && (signed char)*s > 0) { 269 /* 270 * Fast path for plain ASCII characters 271 * excluding NUL. 272 */ 273 *dst = (wchar_t)*s; 274 nb = 1; 275 } else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps)) == 276 (size_t)-1) { 277 *src = s; 278 return ((size_t)-1); 279 } else if (nb == (size_t)-2) { 280 *src = s + nms; 281 return (nchr); 282 } else if (nb == 0) { 283 *src = NULL; 284 return (nchr); 285 } 286 s += nb; 287 nms -= nb; 288 nchr++; 289 dst++; 290 } 291 *src = s; 292 return (nchr); 293 } 294 295 static size_t 296 _UTF8_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps) 297 { 298 _UTF8State *us; 299 unsigned char lead; 300 int i, len; 301 302 us = (_UTF8State *)ps; 303 304 if (us->want != 0) { 305 errno = EINVAL; 306 return ((size_t)-1); 307 } 308 309 if (s == NULL) 310 /* Reset to initial shift state (no-op) */ 311 return (1); 312 313 if ((wc & ~0x7f) == 0) { 314 /* Fast path for plain ASCII characters. */ 315 *s = (char)wc; 316 return (1); 317 } 318 319 /* 320 * Determine the number of octets needed to represent this character. 321 * We always output the shortest sequence possible. Also specify the 322 * first few bits of the first octet, which contains the information 323 * about the sequence length. 324 */ 325 if ((wc & ~0x7f) == 0) { 326 lead = 0; 327 len = 1; 328 } else if ((wc & ~0x7ff) == 0) { 329 lead = 0xc0; 330 len = 2; 331 } else if ((wc & ~0xffff) == 0) { 332 lead = 0xe0; 333 len = 3; 334 } else if ((wc & ~0x1fffff) == 0) { 335 lead = 0xf0; 336 len = 4; 337 } else if ((wc & ~0x3ffffff) == 0) { 338 lead = 0xf8; 339 len = 5; 340 } else if ((wc & ~0x7fffffff) == 0) { 341 lead = 0xfc; 342 len = 6; 343 } else { 344 errno = EILSEQ; 345 return ((size_t)-1); 346 } 347 348 /* 349 * Output the octets representing the character in chunks 350 * of 6 bits, least significant last. The first octet is 351 * a special case because it contains the sequence length 352 * information. 353 */ 354 for (i = len - 1; i > 0; i--) { 355 s[i] = (wc & 0x3f) | 0x80; 356 wc >>= 6; 357 } 358 *s = (wc & 0xff) | lead; 359 360 return (len); 361 } 362 363 static size_t 364 _UTF8_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src, 365 size_t nwc, size_t len, mbstate_t * __restrict ps) 366 { 367 _UTF8State *us; 368 char buf[MB_LEN_MAX]; 369 const wchar_t *s; 370 size_t nbytes; 371 size_t nb; 372 373 us = (_UTF8State *)ps; 374 375 if (us->want != 0) { 376 errno = EINVAL; 377 return ((size_t)-1); 378 } 379 380 s = *src; 381 nbytes = 0; 382 383 if (dst == NULL) { 384 while (nwc-- > 0) { 385 if (0 <= *s && *s < 0x80) 386 /* Fast path for plain ASCII characters. */ 387 nb = 1; 388 else if ((nb = _UTF8_wcrtomb(buf, *s, ps)) == 389 (size_t)-1) 390 /* Invalid character - wcrtomb() sets errno. */ 391 return ((size_t)-1); 392 if (*s == L'\0') 393 return (nbytes + nb - 1); 394 s++; 395 nbytes += nb; 396 } 397 return (nbytes); 398 } 399 400 while (len > 0 && nwc-- > 0) { 401 if (0 <= *s && *s < 0x80) { 402 /* Fast path for plain ASCII characters. */ 403 nb = 1; 404 *dst = *s; 405 } else if (len > (size_t)MB_CUR_MAX) { 406 /* Enough space to translate in-place. */ 407 if ((nb = _UTF8_wcrtomb(dst, *s, ps)) == (size_t)-1) { 408 *src = s; 409 return ((size_t)-1); 410 } 411 } else { 412 /* 413 * May not be enough space; use temp. buffer. 414 */ 415 if ((nb = _UTF8_wcrtomb(buf, *s, ps)) == (size_t)-1) { 416 *src = s; 417 return ((size_t)-1); 418 } 419 if (nb > (int)len) 420 /* MB sequence for character won't fit. */ 421 break; 422 memcpy(dst, buf, nb); 423 } 424 if (*s == L'\0') { 425 *src = NULL; 426 return (nbytes + nb - 1); 427 } 428 s++; 429 dst += nb; 430 len -= nb; 431 nbytes += nb; 432 } 433 *src = s; 434 return (nbytes); 435 } 436