1 /* Elementary Unicode string functions.
2 Copyright (C) 2001-2002, 2005-2020 Free Software Foundation, Inc.
3
4 This program is free software: you can redistribute it and/or modify it
5 under the terms of the GNU General Public License as published
6 by the Free Software Foundation; either version 3 of the License, or
7 (at your option) any later version.
8
9 This program 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 GNU
12 General Public License for more details.
13
14 You should have received a copy of the GNU 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 common macros for C. */
23 #include "unused-parameter.h"
24
25 /* Get bool. */
26 #include <stdbool.h>
27
28 /* Get size_t. */
29 #include <stddef.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
u8_mbtouc_unsafe(ucs4_t * puc,const uint8_t * s,size_t n)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
u16_mbtouc_unsafe(ucs4_t * puc,const uint16_t * s,size_t n)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
u32_mbtouc_unsafe(ucs4_t * puc,const uint32_t * s,size_t n _GL_UNUSED_PARAMETER)195 u32_mbtouc_unsafe (ucs4_t *puc,
196 const uint32_t *s, size_t n _GL_UNUSED_PARAMETER)
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
u8_mbtouc(ucs4_t * puc,const uint8_t * s,size_t n)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
u16_mbtouc(ucs4_t * puc,const uint16_t * s,size_t n)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
u32_mbtouc(ucs4_t * puc,const uint32_t * s,size_t n _GL_UNUSED_PARAMETER)262 u32_mbtouc (ucs4_t *puc, const uint32_t *s, size_t n _GL_UNUSED_PARAMETER)
263 {
264 uint32_t c = *s;
265
266 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000))
267 *puc = c;
268 else
269 /* invalid multibyte character */
270 *puc = 0xfffd;
271 return 1;
272 }
273 # endif
274 #endif
275
276 /* Return the length (number of units) of the first character in S, putting
277 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd,
278 and -1 is returned for an invalid sequence of units, -2 is returned for an
279 incomplete sequence of units.
280 The number of available units, N, must be > 0. */
281 /* Similar to u*_mbtouc(), except that the return value gives more details
282 about the failure, similar to mbrtowc(). */
283
284 #if GNULIB_UNISTR_U8_MBTOUCR || HAVE_LIBUNISTRING
285 extern int
286 u8_mbtoucr (ucs4_t *puc, const uint8_t *s, size_t n);
287 #endif
288
289 #if GNULIB_UNISTR_U16_MBTOUCR || HAVE_LIBUNISTRING
290 extern int
291 u16_mbtoucr (ucs4_t *puc, const uint16_t *s, size_t n);
292 #endif
293
294 #if GNULIB_UNISTR_U32_MBTOUCR || HAVE_LIBUNISTRING
295 extern int
296 u32_mbtoucr (ucs4_t *puc, const uint32_t *s, size_t n);
297 #endif
298
299 /* Put the multibyte character represented by UC in S, returning its
300 length. Return -1 upon failure, -2 if the number of available units, N,
301 is too small. The latter case cannot occur if N >= 6/2/1, respectively. */
302 /* Similar to wctomb(), except that s must not be NULL, and the argument n
303 must be specified. */
304
305 #if GNULIB_UNISTR_U8_UCTOMB || HAVE_LIBUNISTRING
306 /* Auxiliary function, also used by u8_chr, u8_strchr, u8_strrchr. */
307 extern int
308 u8_uctomb_aux (uint8_t *s, ucs4_t uc, int n);
309 # if !HAVE_INLINE
310 extern int
311 u8_uctomb (uint8_t *s, ucs4_t uc, int n);
312 # else
313 static inline int
u8_uctomb(uint8_t * s,ucs4_t uc,int n)314 u8_uctomb (uint8_t *s, ucs4_t uc, int n)
315 {
316 if (uc < 0x80 && n > 0)
317 {
318 s[0] = uc;
319 return 1;
320 }
321 else
322 return u8_uctomb_aux (s, uc, n);
323 }
324 # endif
325 #endif
326
327 #if GNULIB_UNISTR_U16_UCTOMB || HAVE_LIBUNISTRING
328 /* Auxiliary function, also used by u16_chr, u16_strchr, u16_strrchr. */
329 extern int
330 u16_uctomb_aux (uint16_t *s, ucs4_t uc, int n);
331 # if !HAVE_INLINE
332 extern int
333 u16_uctomb (uint16_t *s, ucs4_t uc, int n);
334 # else
335 static inline int
u16_uctomb(uint16_t * s,ucs4_t uc,int n)336 u16_uctomb (uint16_t *s, ucs4_t uc, int n)
337 {
338 if (uc < 0xd800 && n > 0)
339 {
340 s[0] = uc;
341 return 1;
342 }
343 else
344 return u16_uctomb_aux (s, uc, n);
345 }
346 # endif
347 #endif
348
349 #if GNULIB_UNISTR_U32_UCTOMB || HAVE_LIBUNISTRING
350 # if !HAVE_INLINE
351 extern int
352 u32_uctomb (uint32_t *s, ucs4_t uc, int n);
353 # else
354 static inline int
u32_uctomb(uint32_t * s,ucs4_t uc,int n)355 u32_uctomb (uint32_t *s, ucs4_t uc, int n)
356 {
357 if (uc < 0xd800 || (uc >= 0xe000 && uc < 0x110000))
358 {
359 if (n > 0)
360 {
361 *s = uc;
362 return 1;
363 }
364 else
365 return -2;
366 }
367 else
368 return -1;
369 }
370 # endif
371 #endif
372
373 /* Copy N units from SRC to DEST. */
374 /* Similar to memcpy(). */
375 extern uint8_t *
376 u8_cpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
377 extern uint16_t *
378 u16_cpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
379 extern uint32_t *
380 u32_cpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
381
382 /* Copy N units from SRC to DEST, guaranteeing correct behavior for
383 overlapping memory areas. */
384 /* Similar to memmove(). */
385 extern uint8_t *
386 u8_move (uint8_t *dest, const uint8_t *src, size_t n);
387 extern uint16_t *
388 u16_move (uint16_t *dest, const uint16_t *src, size_t n);
389 extern uint32_t *
390 u32_move (uint32_t *dest, const uint32_t *src, size_t n);
391
392 /* Set the first N characters of S to UC. UC should be a character that
393 occupies only 1 unit. */
394 /* Similar to memset(). */
395 extern uint8_t *
396 u8_set (uint8_t *s, ucs4_t uc, size_t n);
397 extern uint16_t *
398 u16_set (uint16_t *s, ucs4_t uc, size_t n);
399 extern uint32_t *
400 u32_set (uint32_t *s, ucs4_t uc, size_t n);
401
402 /* Compare S1 and S2, each of length N. */
403 /* Similar to memcmp(). */
404 extern int
405 u8_cmp (const uint8_t *s1, const uint8_t *s2, size_t n)
406 _UC_ATTRIBUTE_PURE;
407 extern int
408 u16_cmp (const uint16_t *s1, const uint16_t *s2, size_t n)
409 _UC_ATTRIBUTE_PURE;
410 extern int
411 u32_cmp (const uint32_t *s1, const uint32_t *s2, size_t n)
412 _UC_ATTRIBUTE_PURE;
413
414 /* Compare S1 and S2. */
415 /* Similar to the gnulib function memcmp2(). */
416 extern int
417 u8_cmp2 (const uint8_t *s1, size_t n1, const uint8_t *s2, size_t n2)
418 _UC_ATTRIBUTE_PURE;
419 extern int
420 u16_cmp2 (const uint16_t *s1, size_t n1, const uint16_t *s2, size_t n2)
421 _UC_ATTRIBUTE_PURE;
422 extern int
423 u32_cmp2 (const uint32_t *s1, size_t n1, const uint32_t *s2, size_t n2)
424 _UC_ATTRIBUTE_PURE;
425
426 /* Search the string at S for UC. */
427 /* Similar to memchr(). */
428 extern uint8_t *
429 u8_chr (const uint8_t *s, size_t n, ucs4_t uc)
430 _UC_ATTRIBUTE_PURE;
431 extern uint16_t *
432 u16_chr (const uint16_t *s, size_t n, ucs4_t uc)
433 _UC_ATTRIBUTE_PURE;
434 extern uint32_t *
435 u32_chr (const uint32_t *s, size_t n, ucs4_t uc)
436 _UC_ATTRIBUTE_PURE;
437
438 /* Count the number of Unicode characters in the N units from S. */
439 /* Similar to mbsnlen(). */
440 extern size_t
441 u8_mbsnlen (const uint8_t *s, size_t n)
442 _UC_ATTRIBUTE_PURE;
443 extern size_t
444 u16_mbsnlen (const uint16_t *s, size_t n)
445 _UC_ATTRIBUTE_PURE;
446 extern size_t
447 u32_mbsnlen (const uint32_t *s, size_t n)
448 _UC_ATTRIBUTE_PURE;
449
450 /* Elementary string functions with memory allocation. */
451
452 /* Make a freshly allocated copy of S, of length N. */
453 extern uint8_t *
454 u8_cpy_alloc (const uint8_t *s, size_t n);
455 extern uint16_t *
456 u16_cpy_alloc (const uint16_t *s, size_t n);
457 extern uint32_t *
458 u32_cpy_alloc (const uint32_t *s, size_t n);
459
460 /* Elementary string functions on NUL terminated strings. */
461
462 /* Return the length (number of units) of the first character in S.
463 Return 0 if it is the NUL character. Return -1 upon failure. */
464 extern int
465 u8_strmblen (const uint8_t *s)
466 _UC_ATTRIBUTE_PURE;
467 extern int
468 u16_strmblen (const uint16_t *s)
469 _UC_ATTRIBUTE_PURE;
470 extern int
471 u32_strmblen (const uint32_t *s)
472 _UC_ATTRIBUTE_PURE;
473
474 /* Return the length (number of units) of the first character in S, putting
475 its 'ucs4_t' representation in *PUC. Return 0 if it is the NUL
476 character. Return -1 upon failure. */
477 extern int
478 u8_strmbtouc (ucs4_t *puc, const uint8_t *s);
479 extern int
480 u16_strmbtouc (ucs4_t *puc, const uint16_t *s);
481 extern int
482 u32_strmbtouc (ucs4_t *puc, const uint32_t *s);
483
484 /* Forward iteration step. Advances the pointer past the next character,
485 or returns NULL if the end of the string has been reached. Puts the
486 character's 'ucs4_t' representation in *PUC. */
487 extern const uint8_t *
488 u8_next (ucs4_t *puc, const uint8_t *s);
489 extern const uint16_t *
490 u16_next (ucs4_t *puc, const uint16_t *s);
491 extern const uint32_t *
492 u32_next (ucs4_t *puc, const uint32_t *s);
493
494 /* Backward iteration step. Advances the pointer to point to the previous
495 character, or returns NULL if the beginning of the string had been reached.
496 Puts the character's 'ucs4_t' representation in *PUC. */
497 extern const uint8_t *
498 u8_prev (ucs4_t *puc, const uint8_t *s, const uint8_t *start);
499 extern const uint16_t *
500 u16_prev (ucs4_t *puc, const uint16_t *s, const uint16_t *start);
501 extern const uint32_t *
502 u32_prev (ucs4_t *puc, const uint32_t *s, const uint32_t *start);
503
504 /* Return the number of units in S. */
505 /* Similar to strlen(), wcslen(). */
506 extern size_t
507 u8_strlen (const uint8_t *s)
508 _UC_ATTRIBUTE_PURE;
509 extern size_t
510 u16_strlen (const uint16_t *s)
511 _UC_ATTRIBUTE_PURE;
512 extern size_t
513 u32_strlen (const uint32_t *s)
514 _UC_ATTRIBUTE_PURE;
515
516 /* Return the number of units in S, but at most MAXLEN. */
517 /* Similar to strnlen(), wcsnlen(). */
518 extern size_t
519 u8_strnlen (const uint8_t *s, size_t maxlen)
520 _UC_ATTRIBUTE_PURE;
521 extern size_t
522 u16_strnlen (const uint16_t *s, size_t maxlen)
523 _UC_ATTRIBUTE_PURE;
524 extern size_t
525 u32_strnlen (const uint32_t *s, size_t maxlen)
526 _UC_ATTRIBUTE_PURE;
527
528 /* Copy SRC to DEST. */
529 /* Similar to strcpy(), wcscpy(). */
530 extern uint8_t *
531 u8_strcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
532 extern uint16_t *
533 u16_strcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
534 extern uint32_t *
535 u32_strcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
536
537 /* Copy SRC to DEST, returning the address of the terminating NUL in DEST. */
538 /* Similar to stpcpy(). */
539 extern uint8_t *
540 u8_stpcpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
541 extern uint16_t *
542 u16_stpcpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
543 extern uint32_t *
544 u32_stpcpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
545
546 /* Copy no more than N units of SRC to DEST. */
547 /* Similar to strncpy(), wcsncpy(). */
548 extern uint8_t *
549 u8_strncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
550 extern uint16_t *
551 u16_strncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
552 extern uint32_t *
553 u32_strncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
554
555 /* Copy no more than N units of SRC to DEST. Return a pointer past the last
556 non-NUL unit written into DEST. */
557 /* Similar to stpncpy(). */
558 extern uint8_t *
559 u8_stpncpy (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
560 extern uint16_t *
561 u16_stpncpy (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
562 extern uint32_t *
563 u32_stpncpy (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
564
565 /* Append SRC onto DEST. */
566 /* Similar to strcat(), wcscat(). */
567 extern uint8_t *
568 u8_strcat (uint8_t *_UC_RESTRICT dest, const uint8_t *src);
569 extern uint16_t *
570 u16_strcat (uint16_t *_UC_RESTRICT dest, const uint16_t *src);
571 extern uint32_t *
572 u32_strcat (uint32_t *_UC_RESTRICT dest, const uint32_t *src);
573
574 /* Append no more than N units of SRC onto DEST. */
575 /* Similar to strncat(), wcsncat(). */
576 extern uint8_t *
577 u8_strncat (uint8_t *_UC_RESTRICT dest, const uint8_t *src, size_t n);
578 extern uint16_t *
579 u16_strncat (uint16_t *_UC_RESTRICT dest, const uint16_t *src, size_t n);
580 extern uint32_t *
581 u32_strncat (uint32_t *_UC_RESTRICT dest, const uint32_t *src, size_t n);
582
583 /* Compare S1 and S2. */
584 /* Similar to strcmp(), wcscmp(). */
585 #ifdef __sun
586 /* Avoid a collision with the u8_strcmp() function in Solaris 11 libc. */
587 extern int
588 u8_strcmp_gnu (const uint8_t *s1, const uint8_t *s2)
589 _UC_ATTRIBUTE_PURE;
590 # define u8_strcmp u8_strcmp_gnu
591 #else
592 extern int
593 u8_strcmp (const uint8_t *s1, const uint8_t *s2)
594 _UC_ATTRIBUTE_PURE;
595 #endif
596 extern int
597 u16_strcmp (const uint16_t *s1, const uint16_t *s2)
598 _UC_ATTRIBUTE_PURE;
599 extern int
600 u32_strcmp (const uint32_t *s1, const uint32_t *s2)
601 _UC_ATTRIBUTE_PURE;
602
603 /* Compare S1 and S2 using the collation rules of the current locale.
604 Return -1 if S1 < S2, 0 if S1 = S2, 1 if S1 > S2.
605 Upon failure, set errno and return any value. */
606 /* Similar to strcoll(), wcscoll(). */
607 extern int
608 u8_strcoll (const uint8_t *s1, const uint8_t *s2);
609 extern int
610 u16_strcoll (const uint16_t *s1, const uint16_t *s2);
611 extern int
612 u32_strcoll (const uint32_t *s1, const uint32_t *s2);
613
614 /* Compare no more than N units of S1 and S2. */
615 /* Similar to strncmp(), wcsncmp(). */
616 extern int
617 u8_strncmp (const uint8_t *s1, const uint8_t *s2, size_t n)
618 _UC_ATTRIBUTE_PURE;
619 extern int
620 u16_strncmp (const uint16_t *s1, const uint16_t *s2, size_t n)
621 _UC_ATTRIBUTE_PURE;
622 extern int
623 u32_strncmp (const uint32_t *s1, const uint32_t *s2, size_t n)
624 _UC_ATTRIBUTE_PURE;
625
626 /* Duplicate S, returning an identical malloc'd string. */
627 /* Similar to strdup(), wcsdup(). */
628 extern uint8_t *
629 u8_strdup (const uint8_t *s);
630 extern uint16_t *
631 u16_strdup (const uint16_t *s);
632 extern uint32_t *
633 u32_strdup (const uint32_t *s);
634
635 /* Find the first occurrence of UC in STR. */
636 /* Similar to strchr(), wcschr(). */
637 extern uint8_t *
638 u8_strchr (const uint8_t *str, ucs4_t uc)
639 _UC_ATTRIBUTE_PURE;
640 extern uint16_t *
641 u16_strchr (const uint16_t *str, ucs4_t uc)
642 _UC_ATTRIBUTE_PURE;
643 extern uint32_t *
644 u32_strchr (const uint32_t *str, ucs4_t uc)
645 _UC_ATTRIBUTE_PURE;
646
647 /* Find the last occurrence of UC in STR. */
648 /* Similar to strrchr(), wcsrchr(). */
649 extern uint8_t *
650 u8_strrchr (const uint8_t *str, ucs4_t uc)
651 _UC_ATTRIBUTE_PURE;
652 extern uint16_t *
653 u16_strrchr (const uint16_t *str, ucs4_t uc)
654 _UC_ATTRIBUTE_PURE;
655 extern uint32_t *
656 u32_strrchr (const uint32_t *str, ucs4_t uc)
657 _UC_ATTRIBUTE_PURE;
658
659 /* Return the length of the initial segment of STR which consists entirely
660 of Unicode characters not in REJECT. */
661 /* Similar to strcspn(), wcscspn(). */
662 extern size_t
663 u8_strcspn (const uint8_t *str, const uint8_t *reject)
664 _UC_ATTRIBUTE_PURE;
665 extern size_t
666 u16_strcspn (const uint16_t *str, const uint16_t *reject)
667 _UC_ATTRIBUTE_PURE;
668 extern size_t
669 u32_strcspn (const uint32_t *str, const uint32_t *reject)
670 _UC_ATTRIBUTE_PURE;
671
672 /* Return the length of the initial segment of STR which consists entirely
673 of Unicode characters in ACCEPT. */
674 /* Similar to strspn(), wcsspn(). */
675 extern size_t
676 u8_strspn (const uint8_t *str, const uint8_t *accept)
677 _UC_ATTRIBUTE_PURE;
678 extern size_t
679 u16_strspn (const uint16_t *str, const uint16_t *accept)
680 _UC_ATTRIBUTE_PURE;
681 extern size_t
682 u32_strspn (const uint32_t *str, const uint32_t *accept)
683 _UC_ATTRIBUTE_PURE;
684
685 /* Find the first occurrence in STR of any character in ACCEPT. */
686 /* Similar to strpbrk(), wcspbrk(). */
687 extern uint8_t *
688 u8_strpbrk (const uint8_t *str, const uint8_t *accept)
689 _UC_ATTRIBUTE_PURE;
690 extern uint16_t *
691 u16_strpbrk (const uint16_t *str, const uint16_t *accept)
692 _UC_ATTRIBUTE_PURE;
693 extern uint32_t *
694 u32_strpbrk (const uint32_t *str, const uint32_t *accept)
695 _UC_ATTRIBUTE_PURE;
696
697 /* Find the first occurrence of NEEDLE in HAYSTACK. */
698 /* Similar to strstr(), wcsstr(). */
699 extern uint8_t *
700 u8_strstr (const uint8_t *haystack, const uint8_t *needle)
701 _UC_ATTRIBUTE_PURE;
702 extern uint16_t *
703 u16_strstr (const uint16_t *haystack, const uint16_t *needle)
704 _UC_ATTRIBUTE_PURE;
705 extern uint32_t *
706 u32_strstr (const uint32_t *haystack, const uint32_t *needle)
707 _UC_ATTRIBUTE_PURE;
708
709 /* Test whether STR starts with PREFIX. */
710 extern bool
711 u8_startswith (const uint8_t *str, const uint8_t *prefix)
712 _UC_ATTRIBUTE_PURE;
713 extern bool
714 u16_startswith (const uint16_t *str, const uint16_t *prefix)
715 _UC_ATTRIBUTE_PURE;
716 extern bool
717 u32_startswith (const uint32_t *str, const uint32_t *prefix)
718 _UC_ATTRIBUTE_PURE;
719
720 /* Test whether STR ends with SUFFIX. */
721 extern bool
722 u8_endswith (const uint8_t *str, const uint8_t *suffix)
723 _UC_ATTRIBUTE_PURE;
724 extern bool
725 u16_endswith (const uint16_t *str, const uint16_t *suffix)
726 _UC_ATTRIBUTE_PURE;
727 extern bool
728 u32_endswith (const uint32_t *str, const uint32_t *suffix)
729 _UC_ATTRIBUTE_PURE;
730
731 /* Divide STR into tokens separated by characters in DELIM.
732 This interface is actually more similar to wcstok than to strtok. */
733 /* Similar to strtok_r(), wcstok(). */
734 extern uint8_t *
735 u8_strtok (uint8_t *_UC_RESTRICT str, const uint8_t *delim,
736 uint8_t **ptr);
737 extern uint16_t *
738 u16_strtok (uint16_t *_UC_RESTRICT str, const uint16_t *delim,
739 uint16_t **ptr);
740 extern uint32_t *
741 u32_strtok (uint32_t *_UC_RESTRICT str, const uint32_t *delim,
742 uint32_t **ptr);
743
744
745 #ifdef __cplusplus
746 }
747 #endif
748
749 #endif /* _UNISTR_H */
750