unistr.h (23820B)
1 /* DO NOT EDIT! GENERATED AUTOMATICALLY! */ 2 /* Elementary Unicode string functions. 3 Copyright (C) 2001-2002, 2005-2024 Free Software Foundation, Inc. 4 5 This file is free software: you can redistribute it and/or modify 6 it under the terms of the GNU Lesser General Public License as 7 published by the Free Software Foundation; either version 2.1 of the 8 License, or (at your option) any later version. 9 10 This file is distributed in the hope that it will be useful, 11 but WITHOUT ANY WARRANTY; without even the implied warranty of 12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 13 GNU Lesser General Public License for more details. 14 15 You should have received a copy of the GNU Lesser General Public License 16 along with this program. If not, see <https://www.gnu.org/licenses/>. */ 17 18 #ifndef _UNISTR_H 19 #define _UNISTR_H 20 21 #include "unitypes.h" 22 23 /* Get common macros for C. */ 24 #include <unistring/cdefs.h> 25 26 /* Get inline if available. */ 27 #include <unistring/inline.h> 28 29 /* Get bool. */ 30 #include <unistring/stdbool.h> 31 32 /* Get size_t, ptrdiff_t. */ 33 #include <stddef.h> 34 35 /* Get free(). */ 36 #include <stdlib.h> 37 38 #ifdef __cplusplus 39 extern "C" { 40 #endif 41 42 43 /* Conventions: 44 45 All functions prefixed with u8_ operate on UTF-8 encoded strings. 46 Their unit is an uint8_t (1 byte). 47 48 All functions prefixed with u16_ operate on UTF-16 encoded strings. 49 Their unit is an uint16_t (a 2-byte word). 50 51 All functions prefixed with u32_ operate on UCS-4 encoded strings. 52 Their unit is an uint32_t (a 4-byte word). 53 54 All argument pairs (s, n) denote a Unicode string s[0..n-1] with exactly 55 n units. 56 57 All arguments starting with "str" and the arguments of functions starting 58 with u8_str/u16_str/u32_str denote a NUL terminated string, i.e. a string 59 which terminates at the first NUL unit. This termination unit is 60 considered part of the string for all memory allocation purposes, but 61 is not considered part of the string for all other logical purposes. 62 63 Functions returning a string result take a (resultbuf, lengthp) argument 64 pair. If resultbuf is not NULL and the result fits into *lengthp units, 65 it is put in resultbuf, and resultbuf is returned. Otherwise, a freshly 66 allocated string is returned. In both cases, *lengthp is set to the 67 length (number of units) of the returned string. In case of error, 68 NULL is returned and errno is set. */ 69 70 71 /* Elementary string checks. */ 72 73 /* Check whether an UTF-8 string is well-formed. 74 Return NULL if valid, or a pointer to the first invalid unit otherwise. */ 75 extern const uint8_t * 76 u8_check (const uint8_t *s, size_t n) 77 _UC_ATTRIBUTE_PURE; 78 79 /* Check whether an UTF-16 string is well-formed. 80 Return NULL if valid, or a pointer to the first invalid unit otherwise. */ 81 extern const uint16_t * 82 u16_check (const uint16_t *s, size_t n) 83 _UC_ATTRIBUTE_PURE; 84 85 /* Check whether an UCS-4 string is well-formed. 86 Return NULL if valid, or a pointer to the first invalid unit otherwise. */ 87 extern const uint32_t * 88 u32_check (const uint32_t *s, size_t n) 89 _UC_ATTRIBUTE_PURE; 90 91 92 /* Elementary string conversions. */ 93 94 /* Convert an UTF-8 string to an UTF-16 string. */ 95 extern uint16_t * 96 u8_to_u16 (const uint8_t *s, size_t n, uint16_t *resultbuf, 97 size_t *lengthp); 98 99 /* Convert an UTF-8 string to an UCS-4 string. */ 100 extern uint32_t * 101 u8_to_u32 (const uint8_t *s, size_t n, uint32_t *resultbuf, 102 size_t *lengthp); 103 104 /* Convert an UTF-16 string to an UTF-8 string. */ 105 extern uint8_t * 106 u16_to_u8 (const uint16_t *s, size_t n, uint8_t *resultbuf, 107 size_t *lengthp); 108 109 /* Convert an UTF-16 string to an UCS-4 string. */ 110 extern uint32_t * 111 u16_to_u32 (const uint16_t *s, size_t n, uint32_t *resultbuf, 112 size_t *lengthp); 113 114 /* Convert an UCS-4 string to an UTF-8 string. */ 115 extern uint8_t * 116 u32_to_u8 (const uint32_t *s, size_t n, uint8_t *resultbuf, 117 size_t *lengthp); 118 119 /* Convert an UCS-4 string to an UTF-16 string. */ 120 extern uint16_t * 121 u32_to_u16 (const uint32_t *s, size_t n, uint16_t *resultbuf, 122 size_t *lengthp); 123 124 125 /* Elementary string functions. */ 126 127 /* Return the length (number of units) of the first character in S, which is 128 no longer than N. Return 0 if it is the NUL character. Return -1 upon 129 failure. */ 130 /* Similar to mblen(), except that s must not be NULL. */ 131 extern int 132 u8_mblen (const uint8_t *s, size_t n) 133 _UC_ATTRIBUTE_PURE; 134 extern int 135 u16_mblen (const uint16_t *s, size_t n) 136 _UC_ATTRIBUTE_PURE; 137 extern int 138 u32_mblen (const uint32_t *s, size_t n) 139 _UC_ATTRIBUTE_PURE; 140 141 /* Return the length (number of units) of the first character in S, putting 142 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd, 143 and an appropriate number of units is returned. 144 The number of available units, N, must be > 0. */ 145 /* Similar to mbtowc(), except that puc and s must not be NULL, n must be > 0, 146 and the NUL character is not treated specially. */ 147 /* The variants with _unsafe suffix are for backward compatibility with 148 libunistring versions < 0.9.7. */ 149 150 #if 1 151 # if !UNISTRING_HAVE_INLINE 152 extern int 153 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n); 154 # else 155 extern int 156 u8_mbtouc_unsafe_aux (ucs4_t *puc, const uint8_t *s, size_t n); 157 static inline int 158 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n) 159 { 160 uint8_t c = *s; 161 162 if (c < 0x80) 163 { 164 *puc = c; 165 return 1; 166 } 167 else 168 return u8_mbtouc_unsafe_aux (puc, s, n); 169 } 170 # endif 171 #endif 172 173 #if 1 174 # if !UNISTRING_HAVE_INLINE 175 extern int 176 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n); 177 # else 178 extern int 179 u16_mbtouc_unsafe_aux (ucs4_t *puc, const uint16_t *s, size_t n); 180 static inline int 181 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n) 182 { 183 uint16_t c = *s; 184 185 if (c < 0xd800 || c >= 0xe000) 186 { 187 *puc = c; 188 return 1; 189 } 190 else 191 return u16_mbtouc_unsafe_aux (puc, s, n); 192 } 193 # endif 194 #endif 195 196 #if 1 197 # if !UNISTRING_HAVE_INLINE 198 extern int 199 u32_mbtouc_unsafe (ucs4_t *puc, const uint32_t *s, size_t n); 200 # else 201 static inline int 202 u32_mbtouc_unsafe (ucs4_t *puc, 203 const uint32_t *s, _GL_ATTRIBUTE_MAYBE_UNUSED size_t n) 204 { 205 uint32_t c = *s; 206 207 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000)) 208 *puc = c; 209 else 210 /* invalid multibyte character */ 211 *puc = 0xfffd; 212 return 1; 213 } 214 # endif 215 #endif 216 217 #if 1 218 # if !UNISTRING_HAVE_INLINE 219 extern int 220 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n); 221 # else 222 extern int 223 u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n); 224 static inline int 225 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n) 226 { 227 uint8_t c = *s; 228 229 if (c < 0x80) 230 { 231 *puc = c; 232 return 1; 233 } 234 else 235 return u8_mbtouc_aux (puc, s, n); 236 } 237 # endif 238 #endif 239 240 #if 1 241 # if !UNISTRING_HAVE_INLINE 242 extern int 243 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n); 244 # else 245 extern int 246 u16_mbtouc_aux (ucs4_t *puc, const uint16_t *s, size_t n); 247 static inline int 248 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n) 249 { 250 uint16_t c = *s; 251 252 if (c < 0xd800 || c >= 0xe000) 253 { 254 *puc = c; 255 return 1; 256 } 257 else 258 return u16_mbtouc_aux (puc, s, n); 259 } 260 # endif 261 #endif 262 263 #if 1 264 # if !UNISTRING_HAVE_INLINE 265 extern int 266 u32_mbtouc (ucs4_t *puc, const uint32_t *s, size_t n); 267 # else 268 static inline int 269 u32_mbtouc (ucs4_t *puc, const uint32_t *s, 270 _GL_ATTRIBUTE_MAYBE_UNUSED size_t n) 271 { 272 uint32_t c = *s; 273 274 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000)) 275 *puc = c; 276 else 277 /* invalid multibyte character */ 278 *puc = 0xfffd; 279 return 1; 280 } 281 # endif 282 #endif 283 284 /* Return the length (number of units) of the first character in S, putting 285 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd, 286 and -1 is returned for an invalid sequence of units, -2 is returned for an 287 incomplete sequence of units. 288 The number of available units, N, must be > 0. */ 289 /* Similar to u*_mbtouc(), except that the return value gives more details 290 about the failure, similar to mbrtowc(). */ 291 292 #if 1 293 extern int 294 u8_mbtoucr (ucs4_t *puc, const uint8_t *s, size_t n); 295 #endif 296 297 #if 1 298 extern int 299 u16_mbtoucr (ucs4_t *puc, const uint16_t *s, size_t n); 300 #endif 301 302 #if 1 303 extern int 304 u32_mbtoucr (ucs4_t *puc, const uint32_t *s, size_t n); 305 #endif 306 307 /* Put the multibyte character represented by UC in S, returning its 308 length. Return -1 upon failure, -2 if the number of available units, N, 309 is too small. The latter case cannot occur if N >= 6/2/1, respectively. */ 310 /* Similar to wctomb(), except that s must not be NULL, and the argument n 311 must be specified. */ 312 313 #if 1 314 /* Auxiliary function, also used by u8_chr, u8_strchr, u8_strrchr. */ 315 extern int 316 u8_uctomb_aux (uint8_t *s, ucs4_t uc, ptrdiff_t n); 317 # if !UNISTRING_HAVE_INLINE 318 extern int 319 u8_uctomb (uint8_t *s, ucs4_t uc, ptrdiff_t n); 320 # else 321 static inline int 322 u8_uctomb (uint8_t *s, ucs4_t uc, ptrdiff_t n) 323 { 324 if (uc < 0x80 && n > 0) 325 { 326 s[0] = uc; 327 return 1; 328 } 329 else 330 return u8_uctomb_aux (s, uc, n); 331 } 332 # endif 333 #endif 334 335 #if 1 336 /* Auxiliary function, also used by u16_chr, u16_strchr, u16_strrchr. */ 337 extern int 338 u16_uctomb_aux (uint16_t *s, ucs4_t uc, ptrdiff_t n); 339 # if !UNISTRING_HAVE_INLINE 340 extern int 341 u16_uctomb (uint16_t *s, ucs4_t uc, ptrdiff_t n); 342 # else 343 static inline int 344 u16_uctomb (uint16_t *s, ucs4_t uc, ptrdiff_t n) 345 { 346 if (uc < 0xd800 && n > 0) 347 { 348 s[0] = uc; 349 return 1; 350 } 351 else 352 return u16_uctomb_aux (s, uc, n); 353 } 354 # endif 355 #endif 356 357 #if 1 358 # if !UNISTRING_HAVE_INLINE 359 extern int 360 u32_uctomb (uint32_t *s, ucs4_t uc, ptrdiff_t n); 361 # else 362 static inline int 363 u32_uctomb (uint32_t *s, ucs4_t uc, ptrdiff_t n) 364 { 365 if (uc < 0xd800 || (uc >= 0xe000 && uc < 0x110000)) 366 { 367 if (n > 0) 368 { 369 *s = uc; 370 return 1; 371 } 372 else 373 return -2; 374 } 375 else 376 return -1; 377 } 378 # endif 379 #endif 380 381 /* Copy N units from SRC to DEST. */ 382 /* Similar to memcpy(). */ 383 extern uint8_t * 384 u8_cpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n); 385 extern uint16_t * 386 u16_cpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n); 387 extern uint32_t * 388 u32_cpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n); 389 390 /* Copy N units from SRC to DEST, returning pointer after last written unit. */ 391 /* Similar to mempcpy(). */ 392 extern uint8_t * 393 u8_pcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n); 394 extern uint16_t * 395 u16_pcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n); 396 extern uint32_t * 397 u32_pcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n); 398 399 /* Copy N units from SRC to DEST, guaranteeing correct behavior for 400 overlapping memory areas. */ 401 /* Similar to memmove(). */ 402 extern uint8_t * 403 u8_move (uint8_t *dest, const uint8_t *src, size_t n); 404 extern uint16_t * 405 u16_move (uint16_t *dest, const uint16_t *src, size_t n); 406 extern uint32_t * 407 u32_move (uint32_t *dest, const uint32_t *src, size_t n); 408 409 /* Set the first N characters of S to UC. UC should be a character that 410 occupies only 1 unit. */ 411 /* Similar to memset(). */ 412 extern uint8_t * 413 u8_set (uint8_t *s, ucs4_t uc, size_t n); 414 extern uint16_t * 415 u16_set (uint16_t *s, ucs4_t uc, size_t n); 416 extern uint32_t * 417 u32_set (uint32_t *s, ucs4_t uc, size_t n); 418 419 /* Compare S1 and S2, each of length N. */ 420 /* Similar to memcmp(). */ 421 extern int 422 u8_cmp (const uint8_t *s1, const uint8_t *s2, size_t n) 423 _UC_ATTRIBUTE_PURE; 424 extern int 425 u16_cmp (const uint16_t *s1, const uint16_t *s2, size_t n) 426 _UC_ATTRIBUTE_PURE; 427 extern int 428 u32_cmp (const uint32_t *s1, const uint32_t *s2, size_t n) 429 _UC_ATTRIBUTE_PURE; 430 431 /* Compare S1 and S2. */ 432 /* Similar to the gnulib function memcmp2(). */ 433 extern int 434 u8_cmp2 (const uint8_t *s1, size_t n1, const uint8_t *s2, size_t n2) 435 _UC_ATTRIBUTE_PURE; 436 extern int 437 u16_cmp2 (const uint16_t *s1, size_t n1, const uint16_t *s2, size_t n2) 438 _UC_ATTRIBUTE_PURE; 439 extern int 440 u32_cmp2 (const uint32_t *s1, size_t n1, const uint32_t *s2, size_t n2) 441 _UC_ATTRIBUTE_PURE; 442 443 /* Search the string at S for UC. */ 444 /* Similar to memchr(). */ 445 extern uint8_t * 446 u8_chr (const uint8_t *s, size_t n, ucs4_t uc) 447 _UC_ATTRIBUTE_PURE; 448 extern uint16_t * 449 u16_chr (const uint16_t *s, size_t n, ucs4_t uc) 450 _UC_ATTRIBUTE_PURE; 451 extern uint32_t * 452 u32_chr (const uint32_t *s, size_t n, ucs4_t uc) 453 _UC_ATTRIBUTE_PURE; 454 455 /* Count the number of Unicode characters in the N units from S. */ 456 /* Similar to mbsnlen(). */ 457 extern size_t 458 u8_mbsnlen (const uint8_t *s, size_t n) 459 _UC_ATTRIBUTE_PURE; 460 extern size_t 461 u16_mbsnlen (const uint16_t *s, size_t n) 462 _UC_ATTRIBUTE_PURE; 463 extern size_t 464 u32_mbsnlen (const uint32_t *s, size_t n) 465 _UC_ATTRIBUTE_PURE; 466 467 /* Elementary string functions with memory allocation. */ 468 469 /* Make a freshly allocated copy of S, of length N. */ 470 extern uint8_t * 471 u8_cpy_alloc (const uint8_t *s, size_t n); 472 extern uint16_t * 473 u16_cpy_alloc (const uint16_t *s, size_t n); 474 extern uint32_t * 475 u32_cpy_alloc (const uint32_t *s, size_t n); 476 477 /* Elementary string functions on NUL terminated strings. */ 478 479 /* Return the length (number of units) of the first character in S. 480 Return 0 if it is the NUL character. Return -1 upon failure. */ 481 extern int 482 u8_strmblen (const uint8_t *s) 483 _UC_ATTRIBUTE_PURE; 484 extern int 485 u16_strmblen (const uint16_t *s) 486 _UC_ATTRIBUTE_PURE; 487 extern int 488 u32_strmblen (const uint32_t *s) 489 _UC_ATTRIBUTE_PURE; 490 491 /* Return the length (number of units) of the first character in S, putting 492 its 'ucs4_t' representation in *PUC. Return 0 if it is the NUL 493 character. Return -1 upon failure. */ 494 extern int 495 u8_strmbtouc (ucs4_t *puc, const uint8_t *s); 496 extern int 497 u16_strmbtouc (ucs4_t *puc, const uint16_t *s); 498 extern int 499 u32_strmbtouc (ucs4_t *puc, const uint32_t *s); 500 501 /* Forward iteration step. Advances the pointer past the next character, 502 or returns NULL if the end of the string has been reached. Puts the 503 character's 'ucs4_t' representation in *PUC. */ 504 extern const uint8_t * 505 u8_next (ucs4_t *puc, const uint8_t *s); 506 extern const uint16_t * 507 u16_next (ucs4_t *puc, const uint16_t *s); 508 extern const uint32_t * 509 u32_next (ucs4_t *puc, const uint32_t *s); 510 511 /* Backward iteration step. Advances the pointer to point to the previous 512 character, or returns NULL if the beginning of the string had been reached. 513 Puts the character's 'ucs4_t' representation in *PUC. */ 514 extern const uint8_t * 515 u8_prev (ucs4_t *puc, const uint8_t *s, const uint8_t *start); 516 extern const uint16_t * 517 u16_prev (ucs4_t *puc, const uint16_t *s, const uint16_t *start); 518 extern const uint32_t * 519 u32_prev (ucs4_t *puc, const uint32_t *s, const uint32_t *start); 520 521 /* Return the number of units in S. */ 522 /* Similar to strlen(), wcslen(). */ 523 extern size_t 524 u8_strlen (const uint8_t *s) 525 _UC_ATTRIBUTE_PURE; 526 extern size_t 527 u16_strlen (const uint16_t *s) 528 _UC_ATTRIBUTE_PURE; 529 extern size_t 530 u32_strlen (const uint32_t *s) 531 _UC_ATTRIBUTE_PURE; 532 533 /* Return the number of units in S, but at most MAXLEN. */ 534 /* Similar to strnlen(), wcsnlen(). */ 535 extern size_t 536 u8_strnlen (const uint8_t *s, size_t maxlen) 537 _UC_ATTRIBUTE_PURE; 538 extern size_t 539 u16_strnlen (const uint16_t *s, size_t maxlen) 540 _UC_ATTRIBUTE_PURE; 541 extern size_t 542 u32_strnlen (const uint32_t *s, size_t maxlen) 543 _UC_ATTRIBUTE_PURE; 544 545 /* Copy SRC to DEST. */ 546 /* Similar to strcpy(), wcscpy(). */ 547 extern uint8_t * 548 u8_strcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src); 549 extern uint16_t * 550 u16_strcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src); 551 extern uint32_t * 552 u32_strcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src); 553 554 /* Copy SRC to DEST, returning the address of the terminating NUL in DEST. */ 555 /* Similar to stpcpy(). */ 556 extern uint8_t * 557 u8_stpcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src); 558 extern uint16_t * 559 u16_stpcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src); 560 extern uint32_t * 561 u32_stpcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src); 562 563 /* Copy no more than N units of SRC to DEST. */ 564 /* Similar to strncpy(), wcsncpy(). */ 565 extern uint8_t * 566 u8_strncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n); 567 extern uint16_t * 568 u16_strncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n); 569 extern uint32_t * 570 u32_strncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n); 571 572 /* Copy no more than N units of SRC to DEST. Return a pointer past the last 573 non-NUL unit written into DEST. */ 574 /* Similar to stpncpy(). */ 575 extern uint8_t * 576 u8_stpncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n); 577 extern uint16_t * 578 u16_stpncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n); 579 extern uint32_t * 580 u32_stpncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n); 581 582 /* Append SRC onto DEST. */ 583 /* Similar to strcat(), wcscat(). */ 584 extern uint8_t * 585 u8_strcat (uint8_t *_UC_RESTRICT dest, const uint8_t *src); 586 extern uint16_t * 587 u16_strcat (uint16_t *_UC_RESTRICT dest, const uint16_t *src); 588 extern uint32_t * 589 u32_strcat (uint32_t *_UC_RESTRICT dest, const uint32_t *src); 590 591 /* Append no more than N units of SRC onto DEST. */ 592 /* Similar to strncat(), wcsncat(). */ 593 extern uint8_t * 594 u8_strncat (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n); 595 extern uint16_t * 596 u16_strncat (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n); 597 extern uint32_t * 598 u32_strncat (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n); 599 600 /* Compare S1 and S2. */ 601 /* Similar to strcmp(), wcscmp(). */ 602 #ifdef __sun 603 /* Avoid a collision with the u8_strcmp() function in Solaris 11 libc. */ 604 extern int 605 u8_strcmp_gnu (const uint8_t *s1, const uint8_t *s2) 606 _UC_ATTRIBUTE_PURE; 607 # define u8_strcmp u8_strcmp_gnu 608 #else 609 extern int 610 u8_strcmp (const uint8_t *s1, const uint8_t *s2) 611 _UC_ATTRIBUTE_PURE; 612 #endif 613 extern int 614 u16_strcmp (const uint16_t *s1, const uint16_t *s2) 615 _UC_ATTRIBUTE_PURE; 616 extern int 617 u32_strcmp (const uint32_t *s1, const uint32_t *s2) 618 _UC_ATTRIBUTE_PURE; 619 620 /* Compare S1 and S2 using the collation rules of the current locale. 621 Return -1 if S1 < S2, 0 if S1 = S2, 1 if S1 > S2. 622 Upon failure, set errno and return any value. */ 623 /* Similar to strcoll(), wcscoll(). */ 624 extern int 625 u8_strcoll (const uint8_t *s1, const uint8_t *s2); 626 extern int 627 u16_strcoll (const uint16_t *s1, const uint16_t *s2); 628 extern int 629 u32_strcoll (const uint32_t *s1, const uint32_t *s2); 630 631 /* Compare no more than N units of S1 and S2. */ 632 /* Similar to strncmp(), wcsncmp(). */ 633 extern int 634 u8_strncmp (const uint8_t *s1, const uint8_t *s2, size_t n) 635 _UC_ATTRIBUTE_PURE; 636 extern int 637 u16_strncmp (const uint16_t *s1, const uint16_t *s2, size_t n) 638 _UC_ATTRIBUTE_PURE; 639 extern int 640 u32_strncmp (const uint32_t *s1, const uint32_t *s2, size_t n) 641 _UC_ATTRIBUTE_PURE; 642 643 /* Duplicate S, returning an identical malloc'd string. */ 644 /* Similar to strdup(), wcsdup(). */ 645 extern uint8_t * 646 u8_strdup (const uint8_t *s) 647 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE; 648 extern uint16_t * 649 u16_strdup (const uint16_t *s) 650 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE; 651 extern uint32_t * 652 u32_strdup (const uint32_t *s) 653 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE; 654 655 /* Find the first occurrence of UC in STR. */ 656 /* Similar to strchr(), wcschr(). */ 657 extern uint8_t * 658 u8_strchr (const uint8_t *str, ucs4_t uc) 659 _UC_ATTRIBUTE_PURE; 660 extern uint16_t * 661 u16_strchr (const uint16_t *str, ucs4_t uc) 662 _UC_ATTRIBUTE_PURE; 663 extern uint32_t * 664 u32_strchr (const uint32_t *str, ucs4_t uc) 665 _UC_ATTRIBUTE_PURE; 666 667 /* Find the last occurrence of UC in STR. */ 668 /* Similar to strrchr(), wcsrchr(). */ 669 extern uint8_t * 670 u8_strrchr (const uint8_t *str, ucs4_t uc) 671 _UC_ATTRIBUTE_PURE; 672 extern uint16_t * 673 u16_strrchr (const uint16_t *str, ucs4_t uc) 674 _UC_ATTRIBUTE_PURE; 675 extern uint32_t * 676 u32_strrchr (const uint32_t *str, ucs4_t uc) 677 _UC_ATTRIBUTE_PURE; 678 679 /* Return the length of the initial segment of STR which consists entirely 680 of Unicode characters not in REJECT. */ 681 /* Similar to strcspn(), wcscspn(). */ 682 extern size_t 683 u8_strcspn (const uint8_t *str, const uint8_t *reject) 684 _UC_ATTRIBUTE_PURE; 685 extern size_t 686 u16_strcspn (const uint16_t *str, const uint16_t *reject) 687 _UC_ATTRIBUTE_PURE; 688 extern size_t 689 u32_strcspn (const uint32_t *str, const uint32_t *reject) 690 _UC_ATTRIBUTE_PURE; 691 692 /* Return the length of the initial segment of STR which consists entirely 693 of Unicode characters in ACCEPT. */ 694 /* Similar to strspn(), wcsspn(). */ 695 extern size_t 696 u8_strspn (const uint8_t *str, const uint8_t *accept) 697 _UC_ATTRIBUTE_PURE; 698 extern size_t 699 u16_strspn (const uint16_t *str, const uint16_t *accept) 700 _UC_ATTRIBUTE_PURE; 701 extern size_t 702 u32_strspn (const uint32_t *str, const uint32_t *accept) 703 _UC_ATTRIBUTE_PURE; 704 705 /* Find the first occurrence in STR of any character in ACCEPT. */ 706 /* Similar to strpbrk(), wcspbrk(). */ 707 extern uint8_t * 708 u8_strpbrk (const uint8_t *str, const uint8_t *accept) 709 _UC_ATTRIBUTE_PURE; 710 extern uint16_t * 711 u16_strpbrk (const uint16_t *str, const uint16_t *accept) 712 _UC_ATTRIBUTE_PURE; 713 extern uint32_t * 714 u32_strpbrk (const uint32_t *str, const uint32_t *accept) 715 _UC_ATTRIBUTE_PURE; 716 717 /* Find the first occurrence of NEEDLE in HAYSTACK. */ 718 /* Similar to strstr(), wcsstr(). */ 719 extern uint8_t * 720 u8_strstr (const uint8_t *haystack, const uint8_t *needle) 721 _UC_ATTRIBUTE_PURE; 722 extern uint16_t * 723 u16_strstr (const uint16_t *haystack, const uint16_t *needle) 724 _UC_ATTRIBUTE_PURE; 725 extern uint32_t * 726 u32_strstr (const uint32_t *haystack, const uint32_t *needle) 727 _UC_ATTRIBUTE_PURE; 728 729 /* Test whether STR starts with PREFIX. */ 730 extern bool 731 u8_startswith (const uint8_t *str, const uint8_t *prefix) 732 _UC_ATTRIBUTE_PURE; 733 extern bool 734 u16_startswith (const uint16_t *str, const uint16_t *prefix) 735 _UC_ATTRIBUTE_PURE; 736 extern bool 737 u32_startswith (const uint32_t *str, const uint32_t *prefix) 738 _UC_ATTRIBUTE_PURE; 739 740 /* Test whether STR ends with SUFFIX. */ 741 extern bool 742 u8_endswith (const uint8_t *str, const uint8_t *suffix) 743 _UC_ATTRIBUTE_PURE; 744 extern bool 745 u16_endswith (const uint16_t *str, const uint16_t *suffix) 746 _UC_ATTRIBUTE_PURE; 747 extern bool 748 u32_endswith (const uint32_t *str, const uint32_t *suffix) 749 _UC_ATTRIBUTE_PURE; 750 751 /* Divide STR into tokens separated by characters in DELIM. 752 This interface is actually more similar to wcstok than to strtok. */ 753 /* Similar to strtok_r(), wcstok(). */ 754 extern uint8_t * 755 u8_strtok (uint8_t *_UC_RESTRICT str, const uint8_t *delim, 756 uint8_t **ptr); 757 extern uint16_t * 758 u16_strtok (uint16_t *_UC_RESTRICT str, const uint16_t *delim, 759 uint16_t **ptr); 760 extern uint32_t * 761 u32_strtok (uint32_t *_UC_RESTRICT str, const uint32_t *delim, 762 uint32_t **ptr); 763 764 765 #ifdef __cplusplus 766 } 767 #endif 768 769 #endif /* _UNISTR_H */