lib/mbuiter.h

   1 /* Iterating through multibyte strings: macros for multi-byte encodings.
   2    Copyright (C) 2001, 2005, 2007, 2009-2024 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 3 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 /* Written by Bruno Haible <bruno@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 /* This file uses _GL_INLINE_HEADER_BEGIN, _GL_INLINE,
  93    _GL_ATTRIBUTE_ALWAYS_INLINE.  */
  94 #if !_GL_CONFIG_H_INCLUDED
  95  #error "Please include config.h first."
  96 #endif
  97
  98 #include <assert.h>
  99 #include <stddef.h>
 100 #include <stdlib.h>
 101 #include <string.h>
 102 #include <uchar.h>
 103 #include <wchar.h>
 104
 105 #include "mbchar.h"
 106 #include "strnlen1.h"
 107
 108 _GL_INLINE_HEADER_BEGIN
 109 #ifndef MBUITER_INLINE
 110 # define MBUITER_INLINE _GL_INLINE _GL_ATTRIBUTE_ALWAYS_INLINE
 111 #endif
 112
 113 struct mbuiter_multi
 114 {
 115   #if !GNULIB_MBRTOC32_REGULAR
 116   bool in_shift;        /* true if next byte may not be interpreted as ASCII */
 117                         /* If GNULIB_MBRTOC32_REGULAR, it is always false,
 118                            so optimize it away.  */
 119   #endif
 120   mbstate_t state;      /* if in_shift: current shift state */
 121                         /* If GNULIB_MBRTOC32_REGULAR, it is in an initial state
 122                            before and after every mbuiter_multi_next invocation.
 123                          */
 124   bool next_done;       /* true if mbui_avail has already filled the following */
 125   unsigned int cur_max; /* A cache of MB_CUR_MAX.  */
 126   struct mbchar cur;    /* the current character:
 127         const char *cur.ptr          pointer to current character
 128         The following are only valid after mbui_avail.
 129         size_t cur.bytes             number of bytes of current character
 130         bool cur.wc_valid            true if wc is a valid 32-bit wide character
 131         char32_t cur.wc              if wc_valid: the current character
 132         */
 133 };
 134
 135 MBUITER_INLINE void
 136 mbuiter_multi_next (struct mbuiter_multi *iter)
 137 {
 138   if (iter->next_done)
 139     return;
 140   #if !GNULIB_MBRTOC32_REGULAR
 141   if (iter->in_shift)
 142     goto with_shift;
 143   #endif
 144   /* Handle most ASCII characters quickly, without calling mbrtowc().  */
 145   if (is_basic (*iter->cur.ptr))
 146     {
 147       /* These characters are part of the POSIX portable character set.
 148          For most of them, namely those in the ISO C basic character set,
 149          ISO C 99 guarantees that their wide character code is identical to
 150          their char code.  For the few other ones, this is the case as well,
 151          in all locale encodings that are in use.  The 32-bit wide character
 152          code is the same as well.  */
 153       iter->cur.bytes = 1;
 154       iter->cur.wc = *iter->cur.ptr;
 155       iter->cur.wc_valid = true;
 156     }
 157   else
 158     {
 159       assert (mbsinit (&iter->state));
 160       #if !GNULIB_MBRTOC32_REGULAR
 161       iter->in_shift = true;
 162     with_shift:
 163       #endif
 164       iter->cur.bytes = mbrtoc32 (&iter->cur.wc, iter->cur.ptr,
 165                                   strnlen1 (iter->cur.ptr, iter->cur_max),
 166                                   &iter->state);
 167       if (iter->cur.bytes == (size_t) -1)
 168         {
 169           /* An invalid multibyte sequence was encountered.  */
 170           iter->cur.bytes = 1;
 171           iter->cur.wc_valid = false;
 172           /* Allow the next invocation to continue from a sane state.  */
 173           #if !GNULIB_MBRTOC32_REGULAR
 174           iter->in_shift = false;
 175           #endif
 176           mbszero (&iter->state);
 177         }
 178       else if (iter->cur.bytes == (size_t) -2)
 179         {
 180           /* An incomplete multibyte character at the end.  */
 181           iter->cur.bytes = strlen (iter->cur.ptr);
 182           iter->cur.wc_valid = false;
 183           /* Whether to set iter->in_shift = false and reset iter->state
 184              or not is not important; the string end is reached anyway.  */
 185         }
 186       else
 187         {
 188           if (iter->cur.bytes == 0)
 189             {
 190               /* A null wide character was encountered.  */
 191               iter->cur.bytes = 1;
 192               assert (*iter->cur.ptr == '\0');
 193               assert (iter->cur.wc == 0);
 194             }
 195           #if !GNULIB_MBRTOC32_REGULAR
 196           else if (iter->cur.bytes == (size_t) -3)
 197             /* The previous multibyte sequence produced an additional 32-bit
 198                wide character.  */
 199             iter->cur.bytes = 0;
 200           #endif
 201           iter->cur.wc_valid = true;
 202
 203           /* When in an initial state, we can go back treating ASCII
 204              characters more quickly.  */
 205           #if !GNULIB_MBRTOC32_REGULAR
 206           if (mbsinit (&iter->state))
 207             iter->in_shift = false;
 208           #endif
 209         }
 210     }
 211   iter->next_done = true;
 212 }
 213
 214 MBUITER_INLINE void
 215 mbuiter_multi_reloc (struct mbuiter_multi *iter, ptrdiff_t ptrdiff)
 216 {
 217   iter->cur.ptr += ptrdiff;
 218 }
 219
 220 MBUITER_INLINE void
 221 mbuiter_multi_copy (struct mbuiter_multi *new_iter, const struct mbuiter_multi *old_iter)
 222 {
 223   #if !GNULIB_MBRTOC32_REGULAR
 224   if ((new_iter->in_shift = old_iter->in_shift))
 225     memcpy (&new_iter->state, &old_iter->state, sizeof (mbstate_t));
 226   else
 227   #endif
 228     mbszero (&new_iter->state);
 229   new_iter->next_done = old_iter->next_done;
 230   new_iter->cur_max = old_iter->cur_max;
 231   mb_copy (&new_iter->cur, &old_iter->cur);
 232 }
 233
 234 /* Iteration macros.  */
 235 typedef struct mbuiter_multi mbui_iterator_t;
 236 #if !GNULIB_MBRTOC32_REGULAR
 237 #define mbui_init(iter, startptr) \
 238   ((iter).cur.ptr = (startptr), \
 239    (iter).in_shift = false, mbszero (&(iter).state), \
 240    (iter).next_done = false, \
 241    (iter).cur_max = MB_CUR_MAX)
 242 #else
 243 /* Optimized: no in_shift.  */
 244 #define mbui_init(iter, startptr) \
 245   ((iter).cur.ptr = (startptr), \
 246    mbszero (&(iter).state), \
 247    (iter).next_done = false, \
 248    (iter).cur_max = MB_CUR_MAX)
 249 #endif
 250 #define mbui_avail(iter) \
 251   (mbuiter_multi_next (&(iter)), !mb_isnul ((iter).cur))
 252 #define mbui_advance(iter) \
 253   ((iter).cur.ptr += (iter).cur.bytes, (iter).next_done = false)
 254
 255 /* Access to the current character.  */
 256 #define mbui_cur(iter) (iter).cur
 257 #define mbui_cur_ptr(iter) (iter).cur.ptr
 258
 259 /* Relocation.  */
 260 #define mbui_reloc(iter, ptrdiff) mbuiter_multi_reloc (&iter, ptrdiff)
 261
 262 /* Copying an iterator.  */
 263 #define mbui_copy mbuiter_multi_copy
 264
 265 _GL_INLINE_HEADER_END
 266
 267 #endif /* _MBUITER_H */