1 /* Elementary Unicode string functions.
2 Copyright (C) 2001-2002, 2005-2021 Free Software Foundation, Inc.
3
4 This file is free software: you can redistribute it and/or modify
5 it under the terms of the GNU Lesser General Public License as
6 published by the Free Software Foundation; either version 2.1 of the
7 License, or (at your option) any later version.
8
9 This file is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU Lesser General Public License for more details.
13
14 You should have received a copy of the GNU Lesser General Public License
15 along with this program. If not, see <https://www.gnu.org/licenses/>. */
16
17 #ifndef _UNISTR_H
18 #define _UNISTR_H
19
20 #include "unitypes.h"
21
22 /* Get bool. */
23 #include <stdbool.h>
24
25 /* Get size_t, ptrdiff_t. */
26 #include <stddef.h>
27
28 /* Get free(). */
29 #include <stdlib.h>
30
31 #ifdef __cplusplus
32 extern "C" {
33 #endif
34
35
36 /* Conventions:
37
38 All functions prefixed with u8_ operate on UTF-8 encoded strings.
39 Their unit is an uint8_t (1 byte).
40
41 All functions prefixed with u16_ operate on UTF-16 encoded strings.
42 Their unit is an uint16_t (a 2-byte word).
43
44 All functions prefixed with u32_ operate on UCS-4 encoded strings.
45 Their unit is an uint32_t (a 4-byte word).
46
47 All argument pairs (s, n) denote a Unicode string s[0..n-1] with exactly
48 n units.
49
50 All arguments starting with "str" and the arguments of functions starting
51 with u8_str/u16_str/u32_str denote a NUL terminated string, i.e. a string
52 which terminates at the first NUL unit. This termination unit is
53 considered part of the string for all memory allocation purposes, but
54 is not considered part of the string for all other logical purposes.
55
56 Functions returning a string result take a (resultbuf, lengthp) argument
57 pair. If resultbuf is not NULL and the result fits into *lengthp units,
58 it is put in resultbuf, and resultbuf is returned. Otherwise, a freshly
59 allocated string is returned. In both cases, *lengthp is set to the
60 length (number of units) of the returned string. In case of error,
61 NULL is returned and errno is set. */
62
63
64 /* Elementary string checks. */
65
66 /* Check whether an UTF-8 string is well-formed.
67 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
68 extern const uint8_t *
69 u8_check (const uint8_t *s, size_t n)
70 _UC_ATTRIBUTE_PURE;
71
72 /* Check whether an UTF-16 string is well-formed.
73 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
74 extern const uint16_t *
75 u16_check (const uint16_t *s, size_t n)
76 _UC_ATTRIBUTE_PURE;
77
78 /* Check whether an UCS-4 string is well-formed.
79 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
80 extern const uint32_t *
81 u32_check (const uint32_t *s, size_t n)
82 _UC_ATTRIBUTE_PURE;
83
84
85 /* Elementary string conversions. */
86
87 /* Convert an UTF-8 string to an UTF-16 string. */
88 extern uint16_t *
89 u8_to_u16 (const uint8_t *s, size_t n, uint16_t *resultbuf,
90 size_t *lengthp);
91
92 /* Convert an UTF-8 string to an UCS-4 string. */
93 extern uint32_t *
94 u8_to_u32 (const uint8_t *s, size_t n, uint32_t *resultbuf,
95 size_t *lengthp);
96
97 /* Convert an UTF-16 string to an UTF-8 string. */
98 extern uint8_t *
99 u16_to_u8 (const uint16_t *s, size_t n, uint8_t *resultbuf,
100 size_t *lengthp);
101
102 /* Convert an UTF-16 string to an UCS-4 string. */
103 extern uint32_t *
104 u16_to_u32 (const uint16_t *s, size_t n, uint32_t *resultbuf,
105 size_t *lengthp);
106
107 /* Convert an UCS-4 string to an UTF-8 string. */
108 extern uint8_t *
109 u32_to_u8 (const uint32_t *s, size_t n, uint8_t *resultbuf,
110 size_t *lengthp);
111
112 /* Convert an UCS-4 string to an UTF-16 string. */
113 extern uint16_t *
114 u32_to_u16 (const uint32_t *s, size_t n, uint16_t *resultbuf,
115 size_t *lengthp);
116
117
118 /* Elementary string functions. */
119
120 /* Return the length (number of units) of the first character in S, which is
121 no longer than N. Return 0 if it is the NUL character. Return -1 upon
122 failure. */
123 /* Similar to mblen(), except that s must not be NULL. */
124 extern int
125 u8_mblen (const uint8_t *s, size_t n)
126 _UC_ATTRIBUTE_PURE;
127 extern int
128 u16_mblen (const uint16_t *s, size_t n)
129 _UC_ATTRIBUTE_PURE;
130 extern int
131 u32_mblen (const uint32_t *s, size_t n)
132 _UC_ATTRIBUTE_PURE;
133
134 /* Return the length (number of units) of the first character in S, putting
135 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd,
136 and an appropriate number of units is returned.
137 The number of available units, N, must be > 0. */
138 /* Similar to mbtowc(), except that puc and s must not be NULL, n must be > 0,
139 and the NUL character is not treated specially. */
140 /* The variants with _unsafe suffix are for backward compatibility with
141 libunistring versions < 0.9.7. */
142
143 #if GNULIB_UNISTR_U8_MBTOUC_UNSAFE || HAVE_LIBUNISTRING
144 # if !HAVE_INLINE
145 extern int
146 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n);
147 # else
148 extern int
149 u8_mbtouc_unsafe_aux (ucs4_t *puc, const uint8_t *s, size_t n);
150 static inline int
151 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n)
152 {
153 uint8_t c = *s;
154
155 if (c < 0x80)
156 {
157 *puc = c;
158 return 1;
159 }
160 else
161 return u8_mbtouc_unsafe_aux (puc, s, n);
162 }
163 # endif
164 #endif
165
166 #if GNULIB_UNISTR_U16_MBTOUC_UNSAFE || HAVE_LIBUNISTRING
167 # if !HAVE_INLINE
168 extern int
169 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n);
170 # else
171 extern int
172 u16_mbtouc_unsafe_aux (ucs4_t *puc, const uint16_t *s, size_t n);
173 static inline int
174 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n)
175 {
176 uint16_t c = *s;
177
178 if (c < 0xd800 || c >= 0xe000)
179 {
180 *puc = c;
181 return 1;
182 }
183 else
184 return u16_mbtouc_unsafe_aux (puc, s, n);
185 }
186 # endif
187 #endif
188
189 #if GNULIB_UNISTR_U32_MBTOUC_UNSAFE || HAVE_LIBUNISTRING
190 # if !HAVE_INLINE
191 extern int
192 u32_mbtouc_unsafe (ucs4_t *puc, const uint32_t *s, size_t n);
193 # else
194 static inline int
195 u32_mbtouc_unsafe (ucs4_t *puc,
196 const uint32_t *s, _GL_ATTRIBUTE_MAYBE_UNUSED size_t n)
197 {
198 uint32_t c = *s;
199
200 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000))
201 *puc = c;
202 else
203 /* invalid multibyte character */
204 *puc = 0xfffd;
205 return 1;
206 }
207 # endif
208 #endif
209
210 #if GNULIB_UNISTR_U8_MBTOUC || HAVE_LIBUNISTRING
211 # if !HAVE_INLINE
212 extern int
213 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n);
214 # else
215 extern int
216 u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n);
217 static inline int
218 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n)
219 {
220 uint8_t c = *s;
221
222 if (c < 0x80)
223 {
224 *puc = c;
225 return 1;
226 }
227 else
228 return u8_mbtouc_aux (puc, s, n);
229 }
230 # endif
231 #endif
232
233 #if GNULIB_UNISTR_U16_MBTOUC || HAVE_LIBUNISTRING
234 # if !HAVE_INLINE
235 extern int
236 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n);
237 # else
238 extern int
239 u16_mbtouc_aux (ucs4_t *puc, const uint16_t *s, size_t n);
240 static inline int
241 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n)
242 {
243 uint16_t c = *s;
244
245 if (c < 0xd800 || c >= 0xe000)
246 {
247 *puc = c;
248 return 1;
249 }
250 else
251 return u16_mbtouc_aux (puc, s, n);
252 }
253 # endif
254 #endif
255
256 #if GNULIB_UNISTR_U32_MBTOUC || HAVE_LIBUNISTRING
257 # if !HAVE_INLINE
258 extern int
259 u32_mbtouc (ucs4_t *puc, const uint32_t *s, size_t n);
260 # else
261 static inline int
262 u32_mbtouc (ucs4_t *puc, const uint32_t *s,
263 _GL_ATTRIBUTE_MAYBE_UNUSED size_t n)
264 {
265 uint32_t c = *s;
266
267 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000))
268 *puc = c;
269 else
270 /* invalid multibyte character */
271 *puc = 0xfffd;
272 return 1;
273 }
274 # endif
275 #endif
276
277 /* Return the length (number of units) of the first character in S, putting
278 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd,
279 and -1 is returned for an invalid sequence of units, -2 is returned for an
280 incomplete sequence of units.
281 The number of available units, N, must be > 0. */
282 /* Similar to u*_mbtouc(), except that the return value gives more details
283 about the failure, similar to mbrtowc(). */
284
285 #if GNULIB_UNISTR_U8_MBTOUCR || HAVE_LIBUNISTRING
286 extern int
287 u8_mbtoucr (ucs4_t *puc, const uint8_t *s, size_t n);
288 #endif
289
290 #if GNULIB_UNISTR_U16_MBTOUCR || HAVE_LIBUNISTRING
291 extern int
292 u16_mbtoucr (ucs4_t *puc, const uint16_t *s, size_t n);
293 #endif
294
295 #if GNULIB_UNISTR_U32_MBTOUCR || HAVE_LIBUNISTRING
296 extern int
297 u32_mbtoucr (ucs4_t *puc, const uint32_t *s, size_t n);
298 #endif
299
300 /* Put the multibyte character represented by UC in S, returning its
301 length. Return -1 upon failure, -2 if the number of available units, N,
302 is too small. The latter case cannot occur if N >= 6/2/1, respectively. */
303 /* Similar to wctomb(), except that s must not be NULL, and the argument n
304 must be specified. */
305
306 #if GNULIB_UNISTR_U8_UCTOMB || HAVE_LIBUNISTRING
307 /* Auxiliary function, also used by u8_chr, u8_strchr, u8_strrchr. */
308 extern int
309 u8_uctomb_aux (uint8_t *s, ucs4_t uc, ptrdiff_t n);
310 # if !HAVE_INLINE
311 extern int
312 u8_uctomb (uint8_t *s, ucs4_t uc, ptrdiff_t n);
313 # else
314 static inline int
315 u8_uctomb (uint8_t *s, ucs4_t uc, ptrdiff_t n)
316 {
317 if (uc < 0x80 && n > 0)
318 {
319 s[0] = uc;
320 return 1;
321 }
322 else
323 return u8_uctomb_aux (s, uc, n);
324 }
325 # endif
326 #endif
327
328 #if GNULIB_UNISTR_U16_UCTOMB || HAVE_LIBUNISTRING
329 /* Auxiliary function, also used by u16_chr, u16_strchr, u16_strrchr. */
330 extern int
331 u16_uctomb_aux (uint16_t *s, ucs4_t uc, ptrdiff_t n);
332 # if !HAVE_INLINE
333 extern int
334 u16_uctomb (uint16_t *s, ucs4_t uc, ptrdiff_t n);
335 # else
336 static inline int
337 u16_uctomb (uint16_t *s, ucs4_t uc, ptrdiff_t n)
338 {
339 if (uc < 0xd800 && n > 0)
340 {
341 s[0] = uc;
342 return 1;
343 }
344 else
345 return u16_uctomb_aux (s, uc, n);
346 }
347 # endif
348 #endif
349
350 #if GNULIB_UNISTR_U32_UCTOMB || HAVE_LIBUNISTRING
351 # if !HAVE_INLINE
352 extern int
353 u32_uctomb (uint32_t *s, ucs4_t uc, ptrdiff_t n);
354 # else
355 static inline int
356 u32_uctomb (uint32_t *s, ucs4_t uc, ptrdiff_t n)
357 {
358 if (uc < 0xd800 || (uc >= 0xe000 && uc < 0x110000))
359 {
360 if (n > 0)
361 {
362 *s = uc;
363 return 1;
364 }
365 else
366 return -2;
367 }
368 else
369 return -1;
370 }
371 # endif
372 #endif
373
374 /* Copy N units from SRC to DEST. */
375 /* Similar to memcpy(). */
376 extern uint8_t *
377 u8_cpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
378 extern uint16_t *
379 u16_cpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
380 extern uint32_t *
381 u32_cpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
382
383 /* Copy N units from SRC to DEST, guaranteeing correct behavior for
384 overlapping memory areas. */
385 /* Similar to memmove(). */
386 extern uint8_t *
387 u8_move (uint8_t *dest, const uint8_t *src, size_t n);
388 extern uint16_t *
389 u16_move (uint16_t *dest, const uint16_t *src, size_t n);
390 extern uint32_t *
391 u32_move (uint32_t *dest, const uint32_t *src, size_t n);
392
393 /* Set the first N characters of S to UC. UC should be a character that
394 occupies only 1 unit. */
395 /* Similar to memset(). */
396 extern uint8_t *
397 u8_set (uint8_t *s, ucs4_t uc, size_t n);
398 extern uint16_t *
399 u16_set (uint16_t *s, ucs4_t uc, size_t n);
400 extern uint32_t *
401 u32_set (uint32_t *s, ucs4_t uc, size_t n);
402
403 /* Compare S1 and S2, each of length N. */
404 /* Similar to memcmp(). */
405 extern int
406 u8_cmp (const uint8_t *s1, const uint8_t *s2, size_t n)
407 _UC_ATTRIBUTE_PURE;
408 extern int
409 u16_cmp (const uint16_t *s1, const uint16_t *s2, size_t n)
410 _UC_ATTRIBUTE_PURE;
411 extern int
412 u32_cmp (const uint32_t *s1, const uint32_t *s2, size_t n)
413 _UC_ATTRIBUTE_PURE;
414
415 /* Compare S1 and S2. */
416 /* Similar to the gnulib function memcmp2(). */
417 extern int
418 u8_cmp2 (const uint8_t *s1, size_t n1, const uint8_t *s2, size_t n2)
419 _UC_ATTRIBUTE_PURE;
420 extern int
421 u16_cmp2 (const uint16_t *s1, size_t n1, const uint16_t *s2, size_t n2)
422 _UC_ATTRIBUTE_PURE;
423 extern int
424 u32_cmp2 (const uint32_t *s1, size_t n1, const uint32_t *s2, size_t n2)
425 _UC_ATTRIBUTE_PURE;
426
427 /* Search the string at S for UC. */
428 /* Similar to memchr(). */
429 extern uint8_t *
430 u8_chr (const uint8_t *s, size_t n, ucs4_t uc)
431 _UC_ATTRIBUTE_PURE;
432 extern uint16_t *
433 u16_chr (const uint16_t *s, size_t n, ucs4_t uc)
434 _UC_ATTRIBUTE_PURE;
435 extern uint32_t *
436 u32_chr (const uint32_t *s, size_t n, ucs4_t uc)
437 _UC_ATTRIBUTE_PURE;
438
439 /* Count the number of Unicode characters in the N units from S. */
440 /* Similar to mbsnlen(). */
441 extern size_t
442 u8_mbsnlen (const uint8_t *s, size_t n)
443 _UC_ATTRIBUTE_PURE;
444 extern size_t
445 u16_mbsnlen (const uint16_t *s, size_t n)
446 _UC_ATTRIBUTE_PURE;
447 extern size_t
448 u32_mbsnlen (const uint32_t *s, size_t n)
449 _UC_ATTRIBUTE_PURE;
450
451 /* Elementary string functions with memory allocation. */
452
453 /* Make a freshly allocated copy of S, of length N. */
454 extern uint8_t *
455 u8_cpy_alloc (const uint8_t *s, size_t n);
456 extern uint16_t *
457 u16_cpy_alloc (const uint16_t *s, size_t n);
458 extern uint32_t *
459 u32_cpy_alloc (const uint32_t *s, size_t n);
460
461 /* Elementary string functions on NUL terminated strings. */
462
463 /* Return the length (number of units) of the first character in S.
464 Return 0 if it is the NUL character. Return -1 upon failure. */
465 extern int
466 u8_strmblen (const uint8_t *s)
467 _UC_ATTRIBUTE_PURE;
468 extern int
469 u16_strmblen (const uint16_t *s)
470 _UC_ATTRIBUTE_PURE;
471 extern int
472 u32_strmblen (const uint32_t *s)
473 _UC_ATTRIBUTE_PURE;
474
475 /* Return the length (number of units) of the first character in S, putting
476 its 'ucs4_t' representation in *PUC. Return 0 if it is the NUL
477 character. Return -1 upon failure. */
478 extern int
479 u8_strmbtouc (ucs4_t *puc, const uint8_t *s);
480 extern int
481 u16_strmbtouc (ucs4_t *puc, const uint16_t *s);
482 extern int
483 u32_strmbtouc (ucs4_t *puc, const uint32_t *s);
484
485 /* Forward iteration step. Advances the pointer past the next character,
486 or returns NULL if the end of the string has been reached. Puts the
487 character's 'ucs4_t' representation in *PUC. */
488 extern const uint8_t *
489 u8_next (ucs4_t *puc, const uint8_t *s);
490 extern const uint16_t *
491 u16_next (ucs4_t *puc, const uint16_t *s);
492 extern const uint32_t *
493 u32_next (ucs4_t *puc, const uint32_t *s);
494
495 /* Backward iteration step. Advances the pointer to point to the previous
496 character, or returns NULL if the beginning of the string had been reached.
497 Puts the character's 'ucs4_t' representation in *PUC. */
498 extern const uint8_t *
499 u8_prev (ucs4_t *puc, const uint8_t *s, const uint8_t *start);
500 extern const uint16_t *
501 u16_prev (ucs4_t *puc, const uint16_t *s, const uint16_t *start);
502 extern const uint32_t *
503 u32_prev (ucs4_t *puc, const uint32_t *s, const uint32_t *start);
504
505 /* Return the number of units in S. */
506 /* Similar to strlen(), wcslen(). */
507 extern size_t
508 u8_strlen (const uint8_t *s)
509 _UC_ATTRIBUTE_PURE;
510 extern size_t
511 u16_strlen (const uint16_t *s)
512 _UC_ATTRIBUTE_PURE;
513 extern size_t
514 u32_strlen (const uint32_t *s)
515 _UC_ATTRIBUTE_PURE;
516
517 /* Return the number of units in S, but at most MAXLEN. */
518 /* Similar to strnlen(), wcsnlen(). */
519 extern size_t
520 u8_strnlen (const uint8_t *s, size_t maxlen)
521 _UC_ATTRIBUTE_PURE;
522 extern size_t
523 u16_strnlen (const uint16_t *s, size_t maxlen)
524 _UC_ATTRIBUTE_PURE;
525 extern size_t
526 u32_strnlen (const uint32_t *s, size_t maxlen)
527 _UC_ATTRIBUTE_PURE;
528
529 /* Copy SRC to DEST. */
530 /* Similar to strcpy(), wcscpy(). */
531 extern uint8_t *
532 u8_strcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
533 extern uint16_t *
534 u16_strcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
535 extern uint32_t *
536 u32_strcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
537
538 /* Copy SRC to DEST, returning the address of the terminating NUL in DEST. */
539 /* Similar to stpcpy(). */
540 extern uint8_t *
541 u8_stpcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
542 extern uint16_t *
543 u16_stpcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
544 extern uint32_t *
545 u32_stpcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
546
547 /* Copy no more than N units of SRC to DEST. */
548 /* Similar to strncpy(), wcsncpy(). */
549 extern uint8_t *
550 u8_strncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
551 extern uint16_t *
552 u16_strncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
553 extern uint32_t *
554 u32_strncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
555
556 /* Copy no more than N units of SRC to DEST. Return a pointer past the last
557 non-NUL unit written into DEST. */
558 /* Similar to stpncpy(). */
559 extern uint8_t *
560 u8_stpncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
561 extern uint16_t *
562 u16_stpncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
563 extern uint32_t *
564 u32_stpncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
565
566 /* Append SRC onto DEST. */
567 /* Similar to strcat(), wcscat(). */
568 extern uint8_t *
569 u8_strcat (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
570 extern uint16_t *
571 u16_strcat (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
572 extern uint32_t *
573 u32_strcat (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
574
575 /* Append no more than N units of SRC onto DEST. */
576 /* Similar to strncat(), wcsncat(). */
577 extern uint8_t *
578 u8_strncat (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
579 extern uint16_t *
580 u16_strncat (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
581 extern uint32_t *
582 u32_strncat (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
583
584 /* Compare S1 and S2. */
585 /* Similar to strcmp(), wcscmp(). */
586 #ifdef __sun
587 /* Avoid a collision with the u8_strcmp() function in Solaris 11 libc. */
588 extern int
589 u8_strcmp_gnu (const uint8_t *s1, const uint8_t *s2)
590 _UC_ATTRIBUTE_PURE;
591 # define u8_strcmp u8_strcmp_gnu
592 #else
593 extern int
594 u8_strcmp (const uint8_t *s1, const uint8_t *s2)
595 _UC_ATTRIBUTE_PURE;
596 #endif
597 extern int
598 u16_strcmp (const uint16_t *s1, const uint16_t *s2)
599 _UC_ATTRIBUTE_PURE;
600 extern int
601 u32_strcmp (const uint32_t *s1, const uint32_t *s2)
602 _UC_ATTRIBUTE_PURE;
603
604 /* Compare S1 and S2 using the collation rules of the current locale.
605 Return -1 if S1 < S2, 0 if S1 = S2, 1 if S1 > S2.
606 Upon failure, set errno and return any value. */
607 /* Similar to strcoll(), wcscoll(). */
608 extern int
609 u8_strcoll (const uint8_t *s1, const uint8_t *s2);
610 extern int
611 u16_strcoll (const uint16_t *s1, const uint16_t *s2);
612 extern int
613 u32_strcoll (const uint32_t *s1, const uint32_t *s2);
614
615 /* Compare no more than N units of S1 and S2. */
616 /* Similar to strncmp(), wcsncmp(). */
617 extern int
618 u8_strncmp (const uint8_t *s1, const uint8_t *s2, size_t n)
619 _UC_ATTRIBUTE_PURE;
620 extern int
621 u16_strncmp (const uint16_t *s1, const uint16_t *s2, size_t n)
622 _UC_ATTRIBUTE_PURE;
623 extern int
624 u32_strncmp (const uint32_t *s1, const uint32_t *s2, size_t n)
625 _UC_ATTRIBUTE_PURE;
626
627 /* Duplicate S, returning an identical malloc'd string. */
628 /* Similar to strdup(), wcsdup(). */
629 extern uint8_t *
630 u8_strdup (const uint8_t *s)
631 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE;
632 extern uint16_t *
633 u16_strdup (const uint16_t *s)
634 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE;
635 extern uint32_t *
636 u32_strdup (const uint32_t *s)
637 _GL_ATTRIBUTE_MALLOC _GL_ATTRIBUTE_DEALLOC_FREE;
638
639 /* Find the first occurrence of UC in STR. */
640 /* Similar to strchr(), wcschr(). */
641 extern uint8_t *
642 u8_strchr (const uint8_t *str, ucs4_t uc)
643 _UC_ATTRIBUTE_PURE;
644 extern uint16_t *
645 u16_strchr (const uint16_t *str, ucs4_t uc)
646 _UC_ATTRIBUTE_PURE;
647 extern uint32_t *
648 u32_strchr (const uint32_t *str, ucs4_t uc)
649 _UC_ATTRIBUTE_PURE;
650
651 /* Find the last occurrence of UC in STR. */
652 /* Similar to strrchr(), wcsrchr(). */
653 extern uint8_t *
654 u8_strrchr (const uint8_t *str, ucs4_t uc)
655 _UC_ATTRIBUTE_PURE;
656 extern uint16_t *
657 u16_strrchr (const uint16_t *str, ucs4_t uc)
658 _UC_ATTRIBUTE_PURE;
659 extern uint32_t *
660 u32_strrchr (const uint32_t *str, ucs4_t uc)
661 _UC_ATTRIBUTE_PURE;
662
663 /* Return the length of the initial segment of STR which consists entirely
664 of Unicode characters not in REJECT. */
665 /* Similar to strcspn(), wcscspn(). */
666 extern size_t
667 u8_strcspn (const uint8_t *str, const uint8_t *reject)
668 _UC_ATTRIBUTE_PURE;
669 extern size_t
670 u16_strcspn (const uint16_t *str, const uint16_t *reject)
671 _UC_ATTRIBUTE_PURE;
672 extern size_t
673 u32_strcspn (const uint32_t *str, const uint32_t *reject)
674 _UC_ATTRIBUTE_PURE;
675
676 /* Return the length of the initial segment of STR which consists entirely
677 of Unicode characters in ACCEPT. */
678 /* Similar to strspn(), wcsspn(). */
679 extern size_t
680 u8_strspn (const uint8_t *str, const uint8_t *accept)
681 _UC_ATTRIBUTE_PURE;
682 extern size_t
683 u16_strspn (const uint16_t *str, const uint16_t *accept)
684 _UC_ATTRIBUTE_PURE;
685 extern size_t
686 u32_strspn (const uint32_t *str, const uint32_t *accept)
687 _UC_ATTRIBUTE_PURE;
688
689 /* Find the first occurrence in STR of any character in ACCEPT. */
690 /* Similar to strpbrk(), wcspbrk(). */
691 extern uint8_t *
692 u8_strpbrk (const uint8_t *str, const uint8_t *accept)
693 _UC_ATTRIBUTE_PURE;
694 extern uint16_t *
695 u16_strpbrk (const uint16_t *str, const uint16_t *accept)
696 _UC_ATTRIBUTE_PURE;
697 extern uint32_t *
698 u32_strpbrk (const uint32_t *str, const uint32_t *accept)
699 _UC_ATTRIBUTE_PURE;
700
701 /* Find the first occurrence of NEEDLE in HAYSTACK. */
702 /* Similar to strstr(), wcsstr(). */
703 extern uint8_t *
704 u8_strstr (const uint8_t *haystack, const uint8_t *needle)
705 _UC_ATTRIBUTE_PURE;
706 extern uint16_t *
707 u16_strstr (const uint16_t *haystack, const uint16_t *needle)
708 _UC_ATTRIBUTE_PURE;
709 extern uint32_t *
710 u32_strstr (const uint32_t *haystack, const uint32_t *needle)
711 _UC_ATTRIBUTE_PURE;
712
713 /* Test whether STR starts with PREFIX. */
714 extern bool
715 u8_startswith (const uint8_t *str, const uint8_t *prefix)
716 _UC_ATTRIBUTE_PURE;
717 extern bool
718 u16_startswith (const uint16_t *str, const uint16_t *prefix)
719 _UC_ATTRIBUTE_PURE;
720 extern bool
721 u32_startswith (const uint32_t *str, const uint32_t *prefix)
722 _UC_ATTRIBUTE_PURE;
723
724 /* Test whether STR ends with SUFFIX. */
725 extern bool
726 u8_endswith (const uint8_t *str, const uint8_t *suffix)
727 _UC_ATTRIBUTE_PURE;
728 extern bool
729 u16_endswith (const uint16_t *str, const uint16_t *suffix)
730 _UC_ATTRIBUTE_PURE;
731 extern bool
732 u32_endswith (const uint32_t *str, const uint32_t *suffix)
733 _UC_ATTRIBUTE_PURE;
734
735 /* Divide STR into tokens separated by characters in DELIM.
736 This interface is actually more similar to wcstok than to strtok. */
737 /* Similar to strtok_r(), wcstok(). */
738 extern uint8_t *
739 u8_strtok (uint8_t *_UC_RESTRICT str, const uint8_t *delim,
740 uint8_t **ptr);
741 extern uint16_t *
742 u16_strtok (uint16_t *_UC_RESTRICT str, const uint16_t *delim,
743 uint16_t **ptr);
744 extern uint32_t *
745 u32_strtok (uint32_t *_UC_RESTRICT str, const uint32_t *delim,
746 uint32_t **ptr);
747
748
749 #ifdef __cplusplus
750 }
751 #endif
752
753 #endif /* _UNISTR_H */