Home | History | Annotate | Download | only in lib
      1 /* Iterating through multibyte strings: macros for multi-byte encodings.
      2    Copyright (C) 2001, 2005, 2007, 2009-2012 Free Software Foundation, Inc.
      3 
      4    This program is free software: you can redistribute it and/or modify
      5    it under the terms of the GNU General Public License as published by
      6    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
     12    GNU 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 <http://www.gnu.org/licenses/>.  */
     16 
     17 /* Written by Bruno Haible <bruno (at) clisp.org>.  */
     18 
     19 /* The macros in this file implement forward iteration through a
     20    multi-byte string, without knowing its length a-priori.
     21 
     22    With these macros, an iteration loop that looks like
     23 
     24       char *iter;
     25       for (iter = buf; *iter != '\0'; iter++)
     26         {
     27           do_something (*iter);
     28         }
     29 
     30    becomes
     31 
     32       mbui_iterator_t iter;
     33       for (mbui_init (iter, buf); mbui_avail (iter); mbui_advance (iter))
     34         {
     35           do_something (mbui_cur_ptr (iter), mb_len (mbui_cur (iter)));
     36         }
     37 
     38    The benefit of these macros over plain use of mbrtowc is:
     39    - Handling of invalid multibyte sequences is possible without
     40      making the code more complicated, while still preserving the
     41      invalid multibyte sequences.
     42 
     43    Compared to mbiter.h, the macros here don't need to know the string's
     44    length a-priori.  The downside is that at each step, the look-ahead
     45    that guards against overrunning the terminating '\0' is more expensive.
     46    The mbui_* macros are therefore suitable when there is a high probability
     47    that only the first few multibyte characters need to be inspected.
     48    Whereas the mbi_* macros are better if usually the iteration runs
     49    through the entire string.
     50 
     51    mbui_iterator_t
     52      is a type usable for variable declarations.
     53 
     54    mbui_init (iter, startptr)
     55      initializes the iterator, starting at startptr.
     56 
     57    mbui_avail (iter)
     58      returns true if there are more multibyte characters available before
     59      the end of string is reached. In this case, mbui_cur (iter) is
     60      initialized to the next multibyte character.
     61 
     62    mbui_advance (iter)
     63      advances the iterator by one multibyte character.
     64 
     65    mbui_cur (iter)
     66      returns the current multibyte character, of type mbchar_t.  All the
     67      macros defined in mbchar.h can be used on it.
     68 
     69    mbui_cur_ptr (iter)
     70      return a pointer to the beginning of the current multibyte character.
     71 
     72    mbui_reloc (iter, ptrdiff)
     73      relocates iterator when the string is moved by ptrdiff bytes.
     74 
     75    mbui_copy (&destiter, &srciter)
     76      copies srciter to destiter.
     77 
     78    Here are the function prototypes of the macros.
     79 
     80    extern void          mbui_init (mbui_iterator_t iter, const char *startptr);
     81    extern bool          mbui_avail (mbui_iterator_t iter);
     82    extern void          mbui_advance (mbui_iterator_t iter);
     83    extern mbchar_t      mbui_cur (mbui_iterator_t iter);
     84    extern const char *  mbui_cur_ptr (mbui_iterator_t iter);
     85    extern void          mbui_reloc (mbui_iterator_t iter, ptrdiff_t ptrdiff);
     86    extern void          mbui_copy (mbui_iterator_t *new, const mbui_iterator_t *old);
     87  */
     88 
     89 #ifndef _MBUITER_H
     90 #define _MBUITER_H 1
     91 
     92 #include <assert.h>
     93 #include <stdbool.h>
     94 #include <stddef.h>
     95 #include <stdlib.h>
     96 #include <string.h>
     97 
     98 /* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before
     99    <wchar.h>.
    100    BSD/OS 4.1 has a bug: <stdio.h> and <time.h> must be included before
    101    <wchar.h>.  */
    102 #include <stdio.h>
    103 #include <time.h>
    104 #include <wchar.h>
    105 
    106 #include "mbchar.h"
    107 #include "strnlen1.h"
    108 
    109 _GL_INLINE_HEADER_BEGIN
    110 #ifndef MBUITER_INLINE
    111 # define MBUITER_INLINE _GL_INLINE
    112 #endif
    113 
    114 struct mbuiter_multi
    115 {
    116   bool in_shift;        /* true if next byte may not be interpreted as ASCII */
    117   mbstate_t state;      /* if in_shift: current shift state */
    118   bool next_done;       /* true if mbui_avail has already filled the following */
    119   struct mbchar cur;    /* the current character:
    120         const char *cur.ptr             pointer to current character
    121         The following are only valid after mbui_avail.
    122         size_t cur.bytes                number of bytes of current character
    123         bool cur.wc_valid               true if wc is a valid wide character
    124         wchar_t cur.wc                  if wc_valid: the current character
    125         */
    126 };
    127 
    128 MBUITER_INLINE void
    129 mbuiter_multi_next (struct mbuiter_multi *iter)
    130 {
    131   if (iter->next_done)
    132     return;
    133   if (iter->in_shift)
    134     goto with_shift;
    135   /* Handle most ASCII characters quickly, without calling mbrtowc().  */
    136   if (is_basic (*iter->cur.ptr))
    137     {
    138       /* These characters are part of the basic character set.  ISO C 99
    139          guarantees that their wide character code is identical to their
    140          char code.  */
    141       iter->cur.bytes = 1;
    142       iter->cur.wc = *iter->cur.ptr;
    143       iter->cur.wc_valid = true;
    144     }
    145   else
    146     {
    147       assert (mbsinit (&iter->state));
    148       iter->in_shift = true;
    149     with_shift:
    150       iter->cur.bytes = mbrtowc (&iter->cur.wc, iter->cur.ptr,
    151                                  strnlen1 (iter->cur.ptr, MB_CUR_MAX),
    152                                  &iter->state);
    153       if (iter->cur.bytes == (size_t) -1)
    154         {
    155           /* An invalid multibyte sequence was encountered.  */
    156           iter->cur.bytes = 1;
    157           iter->cur.wc_valid = false;
    158           /* Whether to set iter->in_shift = false and reset iter->state
    159              or not is not very important; the string is bogus anyway.  */
    160         }
    161       else if (iter->cur.bytes == (size_t) -2)
    162         {
    163           /* An incomplete multibyte character at the end.  */
    164           iter->cur.bytes = strlen (iter->cur.ptr);
    165           iter->cur.wc_valid = false;
    166           /* Whether to set iter->in_shift = false and reset iter->state
    167              or not is not important; the string end is reached anyway.  */
    168         }
    169       else
    170         {
    171           if (iter->cur.bytes == 0)
    172             {
    173               /* A null wide character was encountered.  */
    174               iter->cur.bytes = 1;
    175               assert (*iter->cur.ptr == '\0');
    176               assert (iter->cur.wc == 0);
    177             }
    178           iter->cur.wc_valid = true;
    179 
    180           /* When in the initial state, we can go back treating ASCII
    181              characters more quickly.  */
    182           if (mbsinit (&iter->state))
    183             iter->in_shift = false;
    184         }
    185     }
    186   iter->next_done = true;
    187 }
    188 
    189 MBUITER_INLINE void
    190 mbuiter_multi_reloc (struct mbuiter_multi *iter, ptrdiff_t ptrdiff)
    191 {
    192   iter->cur.ptr += ptrdiff;
    193 }
    194 
    195 MBUITER_INLINE void
    196 mbuiter_multi_copy (struct mbuiter_multi *new_iter, const struct mbuiter_multi *old_iter)
    197 {
    198   if ((new_iter->in_shift = old_iter->in_shift))
    199     memcpy (&new_iter->state, &old_iter->state, sizeof (mbstate_t));
    200   else
    201     memset (&new_iter->state, 0, sizeof (mbstate_t));
    202   new_iter->next_done = old_iter->next_done;
    203   mb_copy (&new_iter->cur, &old_iter->cur);
    204 }
    205 
    206 /* Iteration macros.  */
    207 typedef struct mbuiter_multi mbui_iterator_t;
    208 #define mbui_init(iter, startptr) \
    209   ((iter).cur.ptr = (startptr), \
    210    (iter).in_shift = false, memset (&(iter).state, '\0', sizeof (mbstate_t)), \
    211    (iter).next_done = false)
    212 #define mbui_avail(iter) \
    213   (mbuiter_multi_next (&(iter)), !mb_isnul ((iter).cur))
    214 #define mbui_advance(iter) \
    215   ((iter).cur.ptr += (iter).cur.bytes, (iter).next_done = false)
    216 
    217 /* Access to the current character.  */
    218 #define mbui_cur(iter) (iter).cur
    219 #define mbui_cur_ptr(iter) (iter).cur.ptr
    220 
    221 /* Relocation.  */
    222 #define mbui_reloc(iter, ptrdiff) mbuiter_multi_reloc (&iter, ptrdiff)
    223 
    224 /* Copying an iterator.  */
    225 #define mbui_copy mbuiter_multi_copy
    226 
    227 _GL_INLINE_HEADER_END
    228 
    229 #endif /* _MBUITER_H */
    230