libstdc++-v3/config/locale/gnu/codecvt_members.cc

   1 // std::codecvt implementation details, GNU version -*- C++ -*-
   2
   3 // Copyright (C) 2002, 2003 Free Software Foundation, Inc.
   4 //
   5 // This file is part of the GNU ISO C++ Library.  This library is free
   6 // software; you can redistribute it and/or modify it under the
   7 // terms of the GNU General Public License as published by the
   8 // Free Software Foundation; either version 2, or (at your option)
   9 // any later version.
  10
  11 // This library is distributed in the hope that it will be useful,
  12 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 // GNU General Public License for more details.
  15
  16 // You should have received a copy of the GNU General Public License along
  17 // with this library; see the file COPYING.  If not, write to the Free
  18 // Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307,
  19 // USA.
  20
  21 // As a special exception, you may use this file as part of a free software
  22 // library without restriction.  Specifically, if other files instantiate
  23 // templates or use macros or inline functions from this file, or you compile
  24 // this file and link it with other files to produce an executable, this
  25 // file does not by itself cause the resulting executable to be covered by
  26 // the GNU General Public License.  This exception does not however
  27 // invalidate any other reasons why the executable file might be covered by
  28 // the GNU General Public License.
  29
  30 //
  31 // ISO C++ 14882: 22.2.1.5 - Template class codecvt
  32 //
  33
  34 // Written by Benjamin Kosnik <bkoz@redhat.com>
  35
  36 #include <locale>
  37 #include <bits/c++locale_internal.h>
  38
  39 namespace std
  40 {
  41   // Specializations.
  42 #ifdef _GLIBCXX_USE_WCHAR_T
  43   codecvt_base::result
  44   codecvt<wchar_t, char, mbstate_t>::
  45   do_out(state_type& __state, const intern_type* __from,
  46          const intern_type* __from_end, const intern_type*& __from_next,
  47          extern_type* __to, extern_type* __to_end,
  48          extern_type*& __to_next) const
  49   {
  50     result __ret = ok;
  51     state_type __tmp_state(__state);
  52
  53 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
  54     __c_locale __old = __uselocale(_M_c_locale_codecvt);
  55 #endif
  56
  57     // wcsnrtombs is *very* fast but stops if encounters NUL characters:
  58     // in case we fall back to wcrtomb and then continue, in a loop.
  59     // NB: wcsnrtombs is a GNU extension
  60     for (__from_next = __from, __to_next = __to;
  61          __from_next < __from_end && __to_next < __to_end
  62          && __ret == ok;)
  63       {
  64         const intern_type* __from_chunk_end = wmemchr(__from_next, L'\0',
  65                                                       __from_end - __from_next);
  66         if (!__from_chunk_end)
  67           __from_chunk_end = __from_end;
  68
  69         __from = __from_next;
  70         const size_t __conv = wcsnrtombs(__to_next, &__from_next,
  71                                          __from_chunk_end - __from_next,
  72                                          __to_end - __to_next, &__state);
  73         if (__conv == static_cast<size_t>(-1))
  74           {
  75             // In case of error, in order to stop at the exact place we
  76             // have to start again from the beginning with a series of
  77             // wcrtomb.
  78             for (; __from < __from_next; ++__from)
  79               __to_next += wcrtomb(__to_next, *__from, &__tmp_state);
  80             __state = __tmp_state;
  81             __ret = error;
  82           }
  83         else if (__from_next && __from_next < __from_chunk_end)
  84           {
  85             __to_next += __conv;
  86             __ret = partial;
  87           }
  88         else
  89           {
  90             __from_next = __from_chunk_end;
  91             __to_next += __conv;
  92           }
  93
  94         if (__from_next < __from_end && __ret == ok)
  95           {
  96             extern_type __buf[MB_LEN_MAX];
  97             __tmp_state = __state;
  98             const size_t __conv = wcrtomb(__buf, *__from_next, &__tmp_state);
  99             if (__conv > static_cast<size_t>(__to_end - __to_next))
 100               __ret = partial;
 101             else
 102               {
 103                 memcpy(__to_next, __buf, __conv);
 104                 __state = __tmp_state;
 105                 __to_next += __conv;
 106                 ++__from_next;
 107               }
 108           }
 109       }
 110
 111 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 112     __uselocale(__old);
 113 #endif
 114
 115     return __ret;
 116   }
 117
 118   codecvt_base::result
 119   codecvt<wchar_t, char, mbstate_t>::
 120   do_in(state_type& __state, const extern_type* __from,
 121         const extern_type* __from_end, const extern_type*& __from_next,
 122         intern_type* __to, intern_type* __to_end,
 123         intern_type*& __to_next) const
 124   {
 125     result __ret = ok;
 126     state_type __tmp_state(__state);
 127
 128 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 129     __c_locale __old = __uselocale(_M_c_locale_codecvt);
 130 #endif
 131
 132     // mbsnrtowcs is *very* fast but stops if encounters NUL characters:
 133     // in case we store a L'\0' and then continue, in a loop.
 134     // NB: mbsnrtowcs is a GNU extension
 135     for (__from_next = __from, __to_next = __to;
 136          __from_next < __from_end && __to_next < __to_end
 137          && __ret == ok;)
 138       {
 139         const extern_type* __from_chunk_end;
 140         __from_chunk_end = static_cast<const extern_type*>(memchr(__from_next, '\0',
 141                                                                   __from_end
 142                                                                   - __from_next));
 143         if (!__from_chunk_end)
 144           __from_chunk_end = __from_end;
 145
 146         __from = __from_next;
 147         size_t __conv = mbsnrtowcs(__to_next, &__from_next,
 148                                    __from_chunk_end - __from_next,
 149                                    __to_end - __to_next, &__state);
 150         if (__conv == static_cast<size_t>(-1))
 151           {
 152             // In case of error, in order to stop at the exact place we
 153             // have to start again from the beginning with a series of
 154             // mbrtowc.
 155             for (;; ++__to_next, __from += __conv)
 156               {
 157                 __conv = mbrtowc(__to_next, __from, __from_end - __from,
 158                                  &__tmp_state);
 159                 if (__conv == static_cast<size_t>(-1)
 160                     || __conv == static_cast<size_t>(-2))
 161                   break;
 162               }
 163             __from_next = __from;
 164             __state = __tmp_state;
 165             __ret = error;
 166           }
 167         else if (__from_next && __from_next < __from_chunk_end)
 168           {
 169             // It is unclear what to return in this case (see DR 382).
 170             __to_next += __conv;
 171             __ret = partial;
 172           }
 173         else
 174           {
 175             __from_next = __from_chunk_end;
 176             __to_next += __conv;
 177           }
 178
 179         if (__from_next < __from_end && __ret == ok)
 180           {
 181             if (__to_next < __to_end)
 182               {
 183                 // XXX Probably wrong for stateful encodings
 184                 __tmp_state = __state;
 185                 ++__from_next;
 186                 *__to_next++ = L'\0';
 187               }
 188             else
 189               __ret = partial;
 190           }
 191       }
 192
 193 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 194     __uselocale(__old);
 195 #endif
 196
 197     return __ret;
 198   }
 199
 200   int
 201   codecvt<wchar_t, char, mbstate_t>::
 202   do_encoding() const throw()
 203   {
 204     // XXX This implementation assumes that the encoding is
 205     // stateless and is either single-byte or variable-width.
 206     int __ret = 0;
 207 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 208     __c_locale __old = __uselocale(_M_c_locale_codecvt);
 209 #endif
 210     if (MB_CUR_MAX == 1)
 211       __ret = 1;
 212 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 213     __uselocale(__old);
 214 #endif
 215     return __ret;
 216   }
 217
 218   int
 219   codecvt<wchar_t, char, mbstate_t>::
 220   do_max_length() const throw()
 221   {
 222 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 223     __c_locale __old = __uselocale(_M_c_locale_codecvt);
 224 #endif
 225     // XXX Probably wrong for stateful encodings.
 226     int __ret = MB_CUR_MAX;
 227 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 228     __uselocale(__old);
 229 #endif
 230     return __ret;
 231   }
 232
 233   int
 234   codecvt<wchar_t, char, mbstate_t>::
 235   do_length(state_type& __state, const extern_type* __from,
 236             const extern_type* __end, size_t __max) const
 237   {
 238     int __ret = 0;
 239     state_type __tmp_state(__state);
 240
 241 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 242     __c_locale __old = __uselocale(_M_c_locale_codecvt);
 243 #endif
 244
 245     // mbsnrtowcs is *very* fast but stops if encounters NUL characters:
 246     // in case we advance past it and then continue, in a loop.
 247     // NB: mbsnrtowcs is a GNU extension
 248
 249     // A dummy internal buffer is needed in order for mbsnrtocws to consider
 250     // its fourth parameter (it wouldn't with NULL as first parameter).
 251     wchar_t* __to = static_cast<wchar_t*>(__builtin_alloca(sizeof(wchar_t)
 252                                                            * __max));
 253     while (__from < __end && __max)
 254       {
 255         const extern_type* __from_chunk_end;
 256         __from_chunk_end = static_cast<const extern_type*>(memchr(__from, '\0',
 257                                                                   __end
 258                                                                   - __from));
 259         if (!__from_chunk_end)
 260           __from_chunk_end = __end;
 261
 262         const extern_type* __tmp_from = __from;
 263         size_t __conv = mbsnrtowcs(__to, &__from,
 264                                    __from_chunk_end - __from,
 265                                    __max, &__state);
 266         if (__conv == static_cast<size_t>(-1))
 267           {
 268             // In case of error, in order to stop at the exact place we
 269             // have to start again from the beginning with a series of
 270             // mbrtowc.
 271             for (__from = __tmp_from;; __from += __conv)
 272               {
 273                 __conv = mbrtowc(NULL, __from, __end - __from,
 274                                  &__tmp_state);
 275                 if (__conv == static_cast<size_t>(-1)
 276                     || __conv == static_cast<size_t>(-2))
 277                   break;
 278               }
 279             __state = __tmp_state;
 280             __ret += __from - __tmp_from;
 281             break;
 282           }
 283         if (!__from)
 284           __from = __from_chunk_end;
 285
 286         __ret += __from - __tmp_from;
 287         __max -= __conv;
 288
 289         if (__from < __end && __max)
 290           {
 291             // XXX Probably wrong for stateful encodings
 292             __tmp_state = __state;
 293             ++__from;
 294             ++__ret;
 295             --__max;
 296           }
 297       }
 298
 299 #if __GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ > 2)
 300     __uselocale(__old);
 301 #endif
 302
 303     return __ret;
 304   }
 305 #endif
 306 }