lib/mbiterf.h

   1 /* Iterating through multibyte strings, faster: 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 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 /* Written by Bruno Haible <bruno@clisp.org>,
  18    with insights from Paul Eggert.  */
  19
  20 /* The macros in this file implement forward iteration through a
  21    multi-byte string.
  22
  23    With these macros, an iteration loop that looks like
  24
  25       char *iter;
  26       for (iter = buf; iter < buf + buflen; iter++)
  27         {
  28           do_something (*iter);
  29         }
  30
  31    becomes
  32
  33       const char *buf_end = buf + buflen;
  34       mbif_state_t state;
  35       [const] char *iter;
  36       for (mbif_init (state), iter = buf; mbif_avail (state, iter, buf_end); )
  37         {
  38           mbchar_t cur = mbif_next (state, iter, buf_end);
  39           // Note: Here always mb_ptr (cur) == iter.
  40           do_something (iter, mb_len (cur));
  41           iter += mb_len (cur);
  42         }
  43
  44    The benefit of these macros over plain use of mbrtowc or mbrtoc32 is:
  45    - Handling of invalid multibyte sequences is possible without
  46      making the code more complicated, while still preserving the
  47      invalid multibyte sequences.
  48
  49    The benefit of these macros over those from mbiter.h is that it
  50    produces faster code with today's optimizing compilers (because mbif_next
  51    returns its result by value).
  52
  53    mbif_state_t
  54      is a type usable for variable declarations.
  55
  56    mbif_init (state)
  57      initializes the state.
  58
  59    mbif_avail (state, iter, endptr)
  60      returns true if another loop round is needed.
  61
  62    mbif_next (state, iter, endptr)
  63      returns the next multibyte character.
  64      It asssumes that the state is initialized and that iter < endptr.
  65
  66    Here are the function prototypes of the macros.
  67
  68    extern void      mbif_init (mbif_state_t state);
  69    extern bool      mbif_avail (mbif_state_t state, const char *iter, const char *endptr);
  70    extern mbchar_t  mbif_next (mbif_state_t state, const char *iter, const char *endptr);
  71  */
  72
  73 #ifndef _MBITERF_H
  74 #define _MBITERF_H 1
  75
  76 /* This file uses _GL_INLINE_HEADER_BEGIN, _GL_INLINE,
  77    _GL_ATTRIBUTE_ALWAYS_INLINE.  */
  78 #if !_GL_CONFIG_H_INCLUDED
  79  #error "Please include config.h first."
  80 #endif
  81
  82 #include <assert.h>
  83 #include <stddef.h>
  84 #include <string.h>
  85 #include <uchar.h>
  86 #include <wchar.h>
  87
  88 #include "mbchar.h"
  89
  90 _GL_INLINE_HEADER_BEGIN
  91 #ifndef MBITERF_INLINE
  92 # define MBITERF_INLINE _GL_INLINE _GL_ATTRIBUTE_ALWAYS_INLINE
  93 #endif
  94
  95 struct mbif_state
  96 {
  97   #if !GNULIB_MBRTOC32_REGULAR
  98   bool in_shift;        /* true if next byte may not be interpreted as ASCII */
  99                         /* If GNULIB_MBRTOC32_REGULAR, it is always false,
 100                            so optimize it away.  */
 101   #endif
 102   mbstate_t state;      /* if in_shift: current shift state */
 103                         /* If GNULIB_MBRTOC32_REGULAR, it is in an initial state
 104                            before and after every mbiterf_next invocation.
 105                          */
 106 };
 107
 108 MBITERF_INLINE mbchar_t
 109 mbiterf_next (struct mbif_state *ps, const char *iter, const char *endptr)
 110 {
 111   #if !GNULIB_MBRTOC32_REGULAR
 112   if (ps->in_shift)
 113     goto with_shift;
 114   #endif
 115   /* Handle most ASCII characters quickly, without calling mbrtowc().  */
 116   if (is_basic (*iter))
 117     {
 118       /* These characters are part of the POSIX portable character set.
 119          For most of them, namely those in the ISO C basic character set,
 120          ISO C 99 guarantees that their wide character code is identical to
 121          their char code.  For the few other ones, this is the case as well,
 122          in all locale encodings that are in use.  The 32-bit wide character
 123          code is the same as well.  */
 124       return (mbchar_t) { .ptr = iter, .bytes = 1, .wc_valid = true, .wc = *iter };
 125     }
 126   else
 127     {
 128       assert (mbsinit (&ps->state));
 129       #if !GNULIB_MBRTOC32_REGULAR
 130       ps->in_shift = true;
 131     with_shift:;
 132       #endif
 133       size_t bytes;
 134       char32_t wc;
 135       bytes = mbrtoc32 (&wc, iter, endptr - iter, &ps->state);
 136       if (bytes == (size_t) -1)
 137         {
 138           /* An invalid multibyte sequence was encountered.  */
 139           /* Allow the next invocation to continue from a sane state.  */
 140           #if !GNULIB_MBRTOC32_REGULAR
 141           ps->in_shift = false;
 142           #endif
 143           mbszero (&ps->state);
 144           return (mbchar_t) { .ptr = iter, .bytes = 1, .wc_valid = false };
 145         }
 146       else if (bytes == (size_t) -2)
 147         {
 148           /* An incomplete multibyte character at the end.  */
 149           #if !GNULIB_MBRTOC32_REGULAR
 150           ps->in_shift = false;
 151           #endif
 152           /* Whether to reset ps->state or not is not important; the string end
 153              is reached anyway.  */
 154           return (mbchar_t) { .ptr = iter, .bytes = endptr - iter, .wc_valid = false };
 155         }
 156       else
 157         {
 158           if (bytes == 0)
 159             {
 160               /* A null wide character was encountered.  */
 161               bytes = 1;
 162               assert (*iter == '\0');
 163               assert (wc == 0);
 164             }
 165           #if !GNULIB_MBRTOC32_REGULAR
 166           else if (bytes == (size_t) -3)
 167             /* The previous multibyte sequence produced an additional 32-bit
 168                wide character.  */
 169             bytes = 0;
 170           #endif
 171
 172           /* When in an initial state, we can go back treating ASCII
 173              characters more quickly.  */
 174           #if !GNULIB_MBRTOC32_REGULAR
 175           if (mbsinit (&ps->state))
 176             ps->in_shift = false;
 177           #endif
 178           return (mbchar_t) { .ptr = iter, .bytes = bytes, .wc_valid = true, .wc = wc };
 179         }
 180     }
 181 }
 182
 183 /* Iteration macros.  */
 184 typedef struct mbif_state mbif_state_t;
 185 #if !GNULIB_MBRTOC32_REGULAR
 186 #define mbif_init(st) \
 187   ((st).in_shift = false, mbszero (&(st).state))
 188 #else
 189 /* Optimized: no in_shift.  */
 190 #define mbif_init(st) \
 191   (mbszero (&(st).state))
 192 #endif
 193 #if !GNULIB_MBRTOC32_REGULAR
 194 #define mbif_avail(st, iter, endptr) ((st).in_shift || ((iter) < (endptr)))
 195 #else
 196 /* Optimized: no in_shift.  */
 197 #define mbif_avail(st, iter, endptr) ((iter) < (endptr))
 198 #endif
 199 #define mbif_next(st, iter, endptr) \
 200   mbiterf_next (&(st), (iter), (endptr))
 201
 202 _GL_INLINE_HEADER_END
 203
 204 #endif /* _MBITERF_H */