1 /* Elementary Unicode string functions.
2 Copyright (C) 2001-2002, 2005-2007 Free Software Foundation, Inc.
4 This program is free software; you can redistribute it and/or modify it
5 under the terms of the GNU Library General Public License as published
6 by the Free Software Foundation; either version 2, or (at your option)
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 Library General Public License for more details.
14 You should have received a copy of the GNU Library General Public
15 License along with this program; if not, write to the Free Software
16 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301,
37 All functions prefixed with u8_ operate on UTF-8 encoded strings.
38 Their unit is an uint8_t (1 byte).
40 All functions prefixed with u16_ operate on UTF-16 encoded strings.
41 Their unit is an uint16_t (a 2-byte word).
43 All functions prefixed with u32_ operate on UCS-4 encoded strings.
44 Their unit is an uint32_t (a 4-byte word).
46 All argument pairs (s, n) denote a Unicode string s[0..n-1] with exactly
49 All arguments starting with "str" and the arguments of functions starting
50 with u8_str/u16_str/u32_str denote a NUL terminated string, i.e. a string
51 which terminates at the first NUL unit. This termination unit is
52 considered part of the string for all memory allocation purposes, but
53 is not considered part of the string for all other logical purposes.
55 Functions returning a string result take a (resultbuf, lengthp) argument
56 pair. If resultbuf is not NULL and the result fits into *lengthp units,
57 it is put in resultbuf, and resultbuf is returned. Otherwise, a freshly
58 allocated string is returned. In both cases, *lengthp is set to the
59 length (number of units) of the returned string. In case of error,
60 NULL is returned and errno is set. */
63 /* Elementary string checks. */
65 /* Check whether an UTF-8 string is well-formed.
66 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
67 extern const uint8_t *
68 u8_check (const uint8_t *s, size_t n);
70 /* Check whether an UTF-16 string is well-formed.
71 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
72 extern const uint16_t *
73 u16_check (const uint16_t *s, size_t n);
75 /* Check whether an UCS-4 string is well-formed.
76 Return NULL if valid, or a pointer to the first invalid unit otherwise. */
77 extern const uint32_t *
78 u32_check (const uint32_t *s, size_t n);
81 /* Elementary string conversions. */
83 /* Convert an UTF-8 string to an UTF-16 string. */
85 u8_to_u16 (const uint8_t *s, size_t n, uint16_t *resultbuf,
88 /* Convert an UTF-8 string to an UCS-4 string. */
90 u8_to_u32 (const uint8_t *s, size_t n, uint32_t *resultbuf,
93 /* Convert an UTF-16 string to an UTF-8 string. */
95 u16_to_u8 (const uint16_t *s, size_t n, uint8_t *resultbuf,
98 /* Convert an UTF-16 string to an UCS-4 string. */
100 u16_to_u32 (const uint16_t *s, size_t n, uint32_t *resultbuf,
103 /* Convert an UCS-4 string to an UTF-8 string. */
105 u32_to_u8 (const uint32_t *s, size_t n, uint8_t *resultbuf,
108 /* Convert an UCS-4 string to an UTF-16 string. */
110 u32_to_u16 (const uint32_t *s, size_t n, uint16_t *resultbuf,
114 /* Elementary string functions. */
116 /* Return the length (number of units) of the first character in S, which is
117 no longer than N. Return 0 if it is the NUL character. Return -1 upon
119 /* Similar to mblen(), except that s must not be NULL. */
121 u8_mblen (const uint8_t *s, size_t n);
123 u16_mblen (const uint16_t *s, size_t n);
125 u32_mblen (const uint32_t *s, size_t n);
127 /* Return the length (number of units) of the first character in S, putting
128 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd,
129 and an appropriate number of units is returned.
130 The number of available units, N, must be > 0. */
131 /* Similar to mbtowc(), except that puc and s must not be NULL, n must be > 0,
132 and the NUL character is not treated specially. */
133 /* The variants with _safe suffix are safe, even if the library is compiled
134 without --enable-safety. */
136 #ifdef GNULIB_UNISTR_U8_MBTOUC_UNSAFE
139 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n);
142 u8_mbtouc_unsafe_aux (ucs4_t *puc, const uint8_t *s, size_t n);
144 u8_mbtouc_unsafe (ucs4_t *puc, const uint8_t *s, size_t n)
154 return u8_mbtouc_unsafe_aux (puc, s, n);
159 #ifdef GNULIB_UNISTR_U16_MBTOUC_UNSAFE
162 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n);
165 u16_mbtouc_unsafe_aux (ucs4_t *puc, const uint16_t *s, size_t n);
167 u16_mbtouc_unsafe (ucs4_t *puc, const uint16_t *s, size_t n)
171 if (c < 0xd800 || c >= 0xe000)
177 return u16_mbtouc_unsafe_aux (puc, s, n);
182 #ifdef GNULIB_UNISTR_U32_MBTOUC_UNSAFE
185 u32_mbtouc_unsafe (ucs4_t *puc, const uint32_t *s, size_t n);
188 u32_mbtouc_unsafe (ucs4_t *puc, const uint32_t *s, size_t n)
192 # if CONFIG_UNICODE_SAFETY
193 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000))
196 # if CONFIG_UNICODE_SAFETY
198 /* invalid multibyte character */
206 #ifdef GNULIB_UNISTR_U8_MBTOUC
209 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n);
212 u8_mbtouc_aux (ucs4_t *puc, const uint8_t *s, size_t n);
214 u8_mbtouc (ucs4_t *puc, const uint8_t *s, size_t n)
224 return u8_mbtouc_aux (puc, s, n);
229 #ifdef GNULIB_UNISTR_U16_MBTOUC
232 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n);
235 u16_mbtouc_aux (ucs4_t *puc, const uint16_t *s, size_t n);
237 u16_mbtouc (ucs4_t *puc, const uint16_t *s, size_t n)
241 if (c < 0xd800 || c >= 0xe000)
247 return u16_mbtouc_aux (puc, s, n);
252 #ifdef GNULIB_UNISTR_U32_MBTOUC
255 u32_mbtouc (ucs4_t *puc, const uint32_t *s, size_t n);
258 u32_mbtouc (ucs4_t *puc, const uint32_t *s, size_t n)
262 if (c < 0xd800 || (c >= 0xe000 && c < 0x110000))
265 /* invalid multibyte character */
272 /* Return the length (number of units) of the first character in S, putting
273 its 'ucs4_t' representation in *PUC. Upon failure, *PUC is set to 0xfffd,
274 and -1 is returned for an invalid sequence of units, -2 is returned for an
275 incomplete sequence of units.
276 The number of available units, N, must be > 0. */
277 /* Similar to u*_mbtouc(), except that the return value gives more details
278 about the failure, similar to mbrtowc(). */
280 #ifdef GNULIB_UNISTR_U8_MBTOUCR
282 u8_mbtoucr (ucs4_t *puc, const uint8_t *s, size_t n);
285 #ifdef GNULIB_UNISTR_U16_MBTOUCR
287 u16_mbtoucr (ucs4_t *puc, const uint16_t *s, size_t n);
290 #ifdef GNULIB_UNISTR_U32_MBTOUCR
292 u32_mbtoucr (ucs4_t *puc, const uint32_t *s, size_t n);
295 /* Put the multibyte character represented by UC in S, returning its
296 length. Return -1 upon failure, -2 if the number of available units, N,
297 is too small. The latter case cannot occur if N >= 6/2/1, respectively. */
298 /* Similar to wctomb(), except that s must not be NULL, and the argument n
299 must be specified. */
301 #ifdef GNULIB_UNISTR_U8_UCTOMB
304 u8_uctomb (uint8_t *s, ucs4_t uc, int n);
307 u8_uctomb_aux (uint8_t *s, ucs4_t uc, int n);
309 u8_uctomb (uint8_t *s, ucs4_t uc, int n)
311 if (uc < 0x80 && n > 0)
317 return u8_uctomb_aux (s, uc, n);
322 #ifdef GNULIB_UNISTR_U16_UCTOMB
325 u16_uctomb (uint16_t *s, ucs4_t uc, int n);
328 u16_uctomb_aux (uint16_t *s, ucs4_t uc, int n);
330 u16_uctomb (uint16_t *s, ucs4_t uc, int n)
332 if (uc < 0xd800 && n > 0)
338 return u16_uctomb_aux (s, uc, n);
343 #ifdef GNULIB_UNISTR_U32_UCTOMB
346 u32_uctomb (uint32_t *s, ucs4_t uc, int n);
349 u32_uctomb (uint32_t *s, ucs4_t uc, int n)
351 if (uc < 0xd800 || (uc >= 0xe000 && uc < 0x110000))
367 /* Copy N units from SRC to DEST. */
368 /* Similar to memcpy(). */
370 u8_cpy (uint8_t *dest, const uint8_t *src, size_t n);
372 u16_cpy (uint16_t *dest, const uint16_t *src, size_t n);
374 u32_cpy (uint32_t *dest, const uint32_t *src, size_t n);
376 /* Copy N units from SRC to DEST, guaranteeing correct behavior for
377 overlapping memory areas. */
378 /* Similar to memmove(). */
380 u8_move (uint8_t *dest, const uint8_t *src, size_t n);
382 u16_move (uint16_t *dest, const uint16_t *src, size_t n);
384 u32_move (uint32_t *dest, const uint32_t *src, size_t n);
386 /* Set the first N characters of S to UC. UC should be a character that
387 occupies only 1 unit. */
388 /* Similar to memset(). */
390 u8_set (uint8_t *s, ucs4_t uc, size_t n);
392 u16_set (uint16_t *s, ucs4_t uc, size_t n);
394 u32_set (uint32_t *s, ucs4_t uc, size_t n);
396 /* Compare S1 and S2, each of length N. */
397 /* Similar to memcmp(). */
399 u8_cmp (const uint8_t *s1, const uint8_t *s2, size_t n);
401 u16_cmp (const uint16_t *s1, const uint16_t *s2, size_t n);
403 u32_cmp (const uint32_t *s1, const uint32_t *s2, size_t n);
405 /* Search the string at S for UC. */
406 /* Similar to memchr(). */
408 u8_chr (const uint8_t *s, size_t n, ucs4_t uc);
410 u16_chr (const uint16_t *s, size_t n, ucs4_t uc);
412 u32_chr (const uint32_t *s, size_t n, ucs4_t uc);
414 /* Elementary string functions with memory allocation. */
416 /* Make a freshly allocated copy of S, of length N. */
418 u8_cpy_alloc (const uint8_t *s, size_t n);
420 u16_cpy_alloc (const uint16_t *s, size_t n);
422 u32_cpy_alloc (const uint32_t *s, size_t n);
424 /* Elementary string functions on NUL terminated strings. */
426 /* Return the length (number of units) of the first character in S.
427 Return 0 if it is the NUL character. Return -1 upon failure. */
429 u8_strmblen (const uint8_t *s);
431 u16_strmblen (const uint16_t *s);
433 u32_strmblen (const uint32_t *s);
435 /* Return the length (number of units) of the first character in S, putting
436 its 'ucs4_t' representation in *PUC. Return 0 if it is the NUL
437 character. Return -1 upon failure. */
439 u8_strmbtouc (ucs4_t *puc, const uint8_t *s);
441 u16_strmbtouc (ucs4_t *puc, const uint16_t *s);
443 u32_strmbtouc (ucs4_t *puc, const uint32_t *s);
445 /* Forward iteration step. Advances the pointer past the next character,
446 or returns NULL if the end of the string has been reached. Puts the
447 character's 'ucs4_t' representation in *PUC. */
448 extern const uint8_t *
449 u8_next (ucs4_t *puc, const uint8_t *s);
450 extern const uint16_t *
451 u16_next (ucs4_t *puc, const uint16_t *s);
452 extern const uint32_t *
453 u32_next (ucs4_t *puc, const uint32_t *s);
455 /* Backward iteration step. Advances the pointer to point to the previous
456 character, or returns NULL if the beginning of the string had been reached.
457 Puts the character's 'ucs4_t' representation in *PUC. */
458 extern const uint8_t *
459 u8_prev (ucs4_t *puc, const uint8_t *s, const uint8_t *start);
460 extern const uint16_t *
461 u16_prev (ucs4_t *puc, const uint16_t *s, const uint16_t *start);
462 extern const uint32_t *
463 u32_prev (ucs4_t *puc, const uint32_t *s, const uint32_t *start);
465 /* Return the number of units in S. */
466 /* Similar to strlen(), wcslen(). */
468 u8_strlen (const uint8_t *s);
470 u16_strlen (const uint16_t *s);
472 u32_strlen (const uint32_t *s);
474 /* Return the number of units in S, but at most MAXLEN. */
475 /* Similar to strnlen(), wcsnlen(). */
477 u8_strnlen (const uint8_t *s, size_t maxlen);
479 u16_strnlen (const uint16_t *s, size_t maxlen);
481 u32_strnlen (const uint32_t *s, size_t maxlen);
483 /* Copy SRC to DEST. */
484 /* Similar to strcpy(), wcscpy(). */
486 u8_strcpy (uint8_t *dest, const uint8_t *src);
488 u16_strcpy (uint16_t *dest, const uint16_t *src);
490 u32_strcpy (uint32_t *dest, const uint32_t *src);
492 /* Copy SRC to DEST, returning the address of the terminating NUL in DEST. */
493 /* Similar to stpcpy(). */
495 u8_stpcpy (uint8_t *dest, const uint8_t *src);
497 u16_stpcpy (uint16_t *dest, const uint16_t *src);
499 u32_stpcpy (uint32_t *dest, const uint32_t *src);
501 /* Copy no more than N units of SRC to DEST. */
502 /* Similar to strncpy(), wcsncpy(). */
504 u8_strncpy (uint8_t *dest, const uint8_t *src, size_t n);
506 u16_strncpy (uint16_t *dest, const uint16_t *src, size_t n);
508 u32_strncpy (uint32_t *dest, const uint32_t *src, size_t n);
510 /* Copy no more than N characters of SRC to DEST, returning the address of
511 the last character written into DEST. */
512 /* Similar to stpncpy(). */
514 u8_stpncpy (uint8_t *dest, const uint8_t *src, size_t n);
516 u16_stpncpy (uint16_t *dest, const uint16_t *src, size_t n);
518 u32_stpncpy (uint32_t *dest, const uint32_t *src, size_t n);
520 /* Append SRC onto DEST. */
521 /* Similar to strcat(), wcscat(). */
523 u8_strcat (uint8_t *dest, const uint8_t *src);
525 u16_strcat (uint16_t *dest, const uint16_t *src);
527 u32_strcat (uint32_t *dest, const uint32_t *src);
529 /* Append no more than N units of SRC onto DEST. */
530 /* Similar to strncat(), wcsncat(). */
532 u8_strncat (uint8_t *dest, const uint8_t *src, size_t n);
534 u16_strncat (uint16_t *dest, const uint16_t *src, size_t n);
536 u32_strncat (uint32_t *dest, const uint32_t *src, size_t n);
538 /* Compare S1 and S2. */
539 /* Similar to strcmp(), wcscmp(). */
541 u8_strcmp (const uint8_t *s1, const uint8_t *s2);
543 u16_strcmp (const uint16_t *s1, const uint16_t *s2);
545 u32_strcmp (const uint32_t *s1, const uint32_t *s2);
547 /* Compare no more than N units of S1 and S2. */
548 /* Similar to strncmp(), wcsncmp(). */
550 u8_strncmp (const uint8_t *s1, const uint8_t *s2, size_t n);
552 u16_strncmp (const uint16_t *s1, const uint16_t *s2, size_t n);
554 u32_strncmp (const uint32_t *s1, const uint32_t *s2, size_t n);
556 /* Duplicate S, returning an identical malloc'd string. */
557 /* Similar to strdup(), wcsdup(). */
559 u8_strdup (const uint8_t *s);
561 u16_strdup (const uint16_t *s);
563 u32_strdup (const uint32_t *s);
565 /* Find the first occurrence of UC in STR. */
566 /* Similar to strchr(), wcschr(). */
568 u8_strchr (const uint8_t *str, ucs4_t uc);
570 u16_strchr (const uint16_t *str, ucs4_t uc);
572 u32_strchr (const uint32_t *str, ucs4_t uc);
574 /* Find the last occurrence of UC in STR. */
575 /* Similar to strrchr(), wcsrchr(). */
577 u8_strrchr (const uint8_t *str, ucs4_t uc);
579 u16_strrchr (const uint16_t *str, ucs4_t uc);
581 u32_strrchr (const uint32_t *str, ucs4_t uc);
583 /* Return the length of the initial segment of STR which consists entirely
584 of Unicode characters not in REJECT. */
585 /* Similar to strcspn(), wcscspn(). */
587 u8_strcspn (const uint8_t *str, const uint8_t *reject);
589 u16_strcspn (const uint16_t *str, const uint16_t *reject);
591 u32_strcspn (const uint32_t *str, const uint32_t *reject);
593 /* Return the length of the initial segment of STR which consists entirely
594 of Unicode characters in ACCEPT. */
595 /* Similar to strspn(), wcsspn(). */
597 u8_strspn (const uint8_t *str, const uint8_t *accept);
599 u16_strspn (const uint16_t *str, const uint16_t *accept);
601 u32_strspn (const uint32_t *str, const uint32_t *accept);
603 /* Find the first occurrence in STR of any character in ACCEPT. */
604 /* Similar to strpbrk(), wcspbrk(). */
606 u8_strpbrk (const uint8_t *str, const uint8_t *accept);
608 u16_strpbrk (const uint16_t *str, const uint16_t *accept);
610 u32_strpbrk (const uint32_t *str, const uint32_t *accept);
612 /* Find the first occurrence of NEEDLE in HAYSTACK. */
613 /* Similar to strstr(), wcsstr(). */
615 u8_strstr (const uint8_t *haystack, const uint8_t *needle);
617 u16_strstr (const uint16_t *haystack, const uint16_t *needle);
619 u32_strstr (const uint32_t *haystack, const uint32_t *needle);
621 /* Test whether STR starts with PREFIX. */
623 u8_startswith (const uint8_t *str, const uint8_t *prefix);
625 u16_startswith (const uint16_t *str, const uint16_t *prefix);
627 u32_startswith (const uint32_t *str, const uint32_t *prefix);
629 /* Test whether STR ends with SUFFIX. */
631 u8_endswith (const uint8_t *str, const uint8_t *suffix);
633 u16_endswith (const uint16_t *str, const uint16_t *suffix);
635 u32_endswith (const uint32_t *str, const uint32_t *suffix);
637 /* Divide STR into tokens separated by characters in DELIM.
638 This interface is actually more similar to wcstok than to strtok. */
639 /* Similar to strtok_r(), wcstok(). */
641 u8_strtok (uint8_t *str, const uint8_t *delim, uint8_t **ptr);
643 u16_strtok (uint16_t *str, const uint16_t *delim, uint16_t **ptr);
645 u32_strtok (uint32_t *str, const uint32_t *delim, uint32_t **ptr);
652 #endif /* _UNISTR_H */