libstdc++-v3/include/ext/codecvt_specializations.h

   1 // Locale support (codecvt) -*- C++ -*-
   2
   3 // Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005
   4 //  Free Software Foundation, Inc.
   5 //
   6 // This file is part of the GNU ISO C++ Library.  This library is free
   7 // software; you can redistribute it and/or modify it under the
   8 // terms of the GNU General Public License as published by the
   9 // Free Software Foundation; either version 2, or (at your option)
  10 // any later version.
  11
  12 // This library is distributed in the hope that it will be useful,
  13 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  14 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15 // GNU General Public License for more details.
  16
  17 // You should have received a copy of the GNU General Public License along
  18 // with this library; see the file COPYING.  If not, write to the Free
  19 // Software Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301,
  20 // USA.
  21
  22 // As a special exception, you may use this file as part of a free software
  23 // library without restriction.  Specifically, if other files instantiate
  24 // templates or use macros or inline functions from this file, or you compile
  25 // this file and link it with other files to produce an executable, this
  26 // file does not by itself cause the resulting executable to be covered by
  27 // the GNU General Public License.  This exception does not however
  28 // invalidate any other reasons why the executable file might be covered by
  29 // the GNU General Public License.
  30
  31 //
  32 // ISO C++ 14882: 22.2.1.5 Template class codecvt
  33 //
  34
  35 // Written by Benjamin Kosnik <bkoz@redhat.com>
  36
  37 /** @file bits/codecvt_specializations.h
  38  *  This file is a GNU extension to the Standard C++ Library.
  39  */
  40
  41   // XXX
  42   // Define this here so codecvt.cc can have _S_max_size definition.
  43 #define _GLIBCXX_USE_ENCODING_STATE 1
  44
  45 _GLIBCXX_BEGIN_NAMESPACE(__gnu_cxx)
  46
  47   /// @brief  Extension to use icov for dealing with character encodings.
  48   // This includes conversions and comparisons between various character
  49   // sets.  This object encapsulates data that may need to be shared between
  50   // char_traits, codecvt and ctype.
  51   class encoding_state
  52   {
  53   public:
  54     // Types:
  55     // NB: A conversion descriptor subsumes and enhances the
  56     // functionality of a simple state type such as mbstate_t.
  57     typedef iconv_t     descriptor_type;
  58
  59   protected:
  60     // Name of internal character set encoding.
  61     std::string         _M_int_enc;
  62
  63     // Name of external character set encoding.
  64     std::string         _M_ext_enc;
  65
  66     // Conversion descriptor between external encoding to internal encoding.
  67     descriptor_type     _M_in_desc;
  68
  69     // Conversion descriptor between internal encoding to external encoding.
  70     descriptor_type     _M_out_desc;
  71
  72     // The byte-order marker for the external encoding, if necessary.
  73     int                 _M_ext_bom;
  74
  75     // The byte-order marker for the internal encoding, if necessary.
  76     int                 _M_int_bom;
  77
  78     // Number of external bytes needed to construct one complete
  79     // character in the internal encoding.
  80     // NB: -1 indicates variable, or stateful, encodings.
  81     int                 _M_bytes;
  82
  83   public:
  84     explicit
  85     encoding_state()
  86     : _M_in_desc(0), _M_out_desc(0), _M_ext_bom(0), _M_int_bom(0), _M_bytes(0)
  87     { }
  88
  89     explicit
  90     encoding_state(const char* __int, const char* __ext,
  91                    int __ibom = 0, int __ebom = 0, int __bytes = 1)
  92     : _M_int_enc(__int), _M_ext_enc(__ext), _M_in_desc(0), _M_out_desc(0),
  93       _M_ext_bom(__ebom), _M_int_bom(__ibom), _M_bytes(__bytes)
  94     { init(); }
  95
  96     // 21.1.2 traits typedefs
  97     // p4
  98     // typedef STATE_T state_type
  99     // requires: state_type shall meet the requirements of
 100     // CopyConstructible types (20.1.3)
 101     // NB: This does not preseve the actual state of the conversion
 102     // descriptor member, but it does duplicate the encoding
 103     // information.
 104     encoding_state(const encoding_state& __obj) : _M_in_desc(0), _M_out_desc(0)
 105     { construct(__obj); }
 106
 107     // Need assignment operator as well.
 108     encoding_state&
 109     operator=(const encoding_state& __obj)
 110     {
 111       construct(__obj);
 112       return *this;
 113     }
 114
 115     ~encoding_state()
 116     { destroy(); }
 117
 118     bool
 119     good() const throw()
 120     {
 121       const descriptor_type __err = reinterpret_cast<iconv_t>(-1);
 122       bool __test = _M_in_desc && _M_in_desc != __err;
 123       __test &=  _M_out_desc && _M_out_desc != __err;
 124       return __test;
 125     }
 126
 127     int
 128     character_ratio() const
 129     { return _M_bytes; }
 130
 131     const std::string
 132     internal_encoding() const
 133     { return _M_int_enc; }
 134
 135     int
 136     internal_bom() const
 137     { return _M_int_bom; }
 138
 139     const std::string
 140     external_encoding() const
 141     { return _M_ext_enc; }
 142
 143     int
 144     external_bom() const
 145     { return _M_ext_bom; }
 146
 147     const descriptor_type&
 148     in_descriptor() const
 149     { return _M_in_desc; }
 150
 151     const descriptor_type&
 152     out_descriptor() const
 153     { return _M_out_desc; }
 154
 155   protected:
 156     void
 157     init()
 158     {
 159       const descriptor_type __err = reinterpret_cast<iconv_t>(-1);
 160       const bool __have_encodings = _M_int_enc.size() && _M_ext_enc.size();
 161       if (!_M_in_desc && __have_encodings)
 162         {
 163           _M_in_desc = iconv_open(_M_int_enc.c_str(), _M_ext_enc.c_str());
 164           if (_M_in_desc == __err)
 165             std::__throw_runtime_error(__N("encoding_state::_M_init "
 166                                     "creating iconv input descriptor failed"));
 167         }
 168       if (!_M_out_desc && __have_encodings)
 169         {
 170           _M_out_desc = iconv_open(_M_ext_enc.c_str(), _M_int_enc.c_str());
 171           if (_M_out_desc == __err)
 172             std::__throw_runtime_error(__N("encoding_state::_M_init "
 173                                   "creating iconv output descriptor failed"));
 174         }
 175     }
 176
 177     void
 178     construct(const encoding_state& __obj)
 179     {
 180       destroy();
 181       _M_int_enc = __obj._M_int_enc;
 182       _M_ext_enc = __obj._M_ext_enc;
 183       _M_ext_bom = __obj._M_ext_bom;
 184       _M_int_bom = __obj._M_int_bom;
 185       _M_bytes = __obj._M_bytes;
 186       init();
 187     }
 188
 189     void
 190     destroy() throw()
 191     {
 192       const descriptor_type __err = reinterpret_cast<iconv_t>(-1);
 193       if (_M_in_desc && _M_in_desc != __err)
 194         {
 195           iconv_close(_M_in_desc);
 196           _M_in_desc = 0;
 197         }
 198       if (_M_out_desc && _M_out_desc != __err)
 199         {
 200           iconv_close(_M_out_desc);
 201           _M_out_desc = 0;
 202         }
 203     }
 204   };
 205
 206   /// @brief  encoding_char_traits.
 207   // Custom traits type with encoding_state for the state type, and the
 208   // associated fpos<encoding_state> for the position type, all other
 209   // bits equivalent to the required char_traits instantiations.
 210   template<typename _CharT>
 211     struct encoding_char_traits : public std::char_traits<_CharT>
 212     {
 213       typedef encoding_state                            state_type;
 214       typedef typename std::fpos<state_type>            pos_type;
 215     };
 216
 217 _GLIBCXX_END_NAMESPACE
 218
 219
 220 _GLIBCXX_BEGIN_NAMESPACE(std)
 221
 222   using __gnu_cxx::encoding_state;
 223
 224   /// @brief  codecvt<InternT, _ExternT, encoding_state> specialization.
 225   // This partial specialization takes advantage of iconv to provide
 226   // code conversions between a large number of character encodings.
 227   template<typename _InternT, typename _ExternT>
 228     class codecvt<_InternT, _ExternT, encoding_state>
 229     : public __codecvt_abstract_base<_InternT, _ExternT, encoding_state>
 230     {
 231     public:
 232       // Types:
 233       typedef codecvt_base::result                      result;
 234       typedef _InternT                                  intern_type;
 235       typedef _ExternT                                  extern_type;
 236       typedef __gnu_cxx::encoding_state                 state_type;
 237       typedef state_type::descriptor_type               descriptor_type;
 238
 239       // Data Members:
 240       static locale::id                 id;
 241
 242       explicit
 243       codecvt(size_t __refs = 0)
 244       : __codecvt_abstract_base<intern_type, extern_type, state_type>(__refs)
 245       { }
 246
 247       explicit
 248       codecvt(state_type& __enc, size_t __refs = 0)
 249       : __codecvt_abstract_base<intern_type, extern_type, state_type>(__refs)
 250       { }
 251
 252      protected:
 253       virtual
 254       ~codecvt() { }
 255
 256       virtual result
 257       do_out(state_type& __state, const intern_type* __from,
 258              const intern_type* __from_end, const intern_type*& __from_next,
 259              extern_type* __to, extern_type* __to_end,
 260              extern_type*& __to_next) const;
 261
 262       virtual result
 263       do_unshift(state_type& __state, extern_type* __to,
 264                  extern_type* __to_end, extern_type*& __to_next) const;
 265
 266       virtual result
 267       do_in(state_type& __state, const extern_type* __from,
 268             const extern_type* __from_end, const extern_type*& __from_next,
 269             intern_type* __to, intern_type* __to_end,
 270             intern_type*& __to_next) const;
 271
 272       virtual int
 273       do_encoding() const throw();
 274
 275       virtual bool
 276       do_always_noconv() const throw();
 277
 278       virtual int
 279       do_length(state_type&, const extern_type* __from,
 280                 const extern_type* __end, size_t __max) const;
 281
 282       virtual int
 283       do_max_length() const throw();
 284     };
 285
 286   template<typename _InternT, typename _ExternT>
 287     locale::id
 288     codecvt<_InternT, _ExternT, encoding_state>::id;
 289
 290   // This adaptor works around the signature problems of the second
 291   // argument to iconv():  SUSv2 and others use 'const char**', but glibc 2.2
 292   // uses 'char**', which matches the POSIX 1003.1-2001 standard.
 293   // Using this adaptor, g++ will do the work for us.
 294   template<typename _T>
 295     inline size_t
 296     __iconv_adaptor(size_t(*__func)(iconv_t, _T, size_t*, char**, size_t*),
 297                     iconv_t __cd, char** __inbuf, size_t* __inbytes,
 298                     char** __outbuf, size_t* __outbytes)
 299     { return __func(__cd, (_T)__inbuf, __inbytes, __outbuf, __outbytes); }
 300
 301   template<typename _InternT, typename _ExternT>
 302     codecvt_base::result
 303     codecvt<_InternT, _ExternT, encoding_state>::
 304     do_out(state_type& __state, const intern_type* __from,
 305            const intern_type* __from_end, const intern_type*& __from_next,
 306            extern_type* __to, extern_type* __to_end,
 307            extern_type*& __to_next) const
 308     {
 309       result __ret = codecvt_base::error;
 310       if (__state.good())
 311         {
 312           const descriptor_type& __desc = __state.out_descriptor();
 313           const size_t __fmultiple = sizeof(intern_type);
 314           size_t __fbytes = __fmultiple * (__from_end - __from);
 315           const size_t __tmultiple = sizeof(extern_type);
 316           size_t __tbytes = __tmultiple * (__to_end - __to);
 317
 318           // Argument list for iconv specifies a byte sequence. Thus,
 319           // all to/from arrays must be brutally casted to char*.
 320           char* __cto = reinterpret_cast<char*>(__to);
 321           char* __cfrom;
 322           size_t __conv;
 323
 324           // Some encodings need a byte order marker as the first item
 325           // in the byte stream, to designate endian-ness. The default
 326           // value for the byte order marker is NULL, so if this is
 327           // the case, it's not necessary and we can just go on our
 328           // merry way.
 329           int __int_bom = __state.internal_bom();
 330           if (__int_bom)
 331             {
 332               size_t __size = __from_end - __from;
 333               intern_type* __cfixed = static_cast<intern_type*>
 334                 (__builtin_alloca(sizeof(intern_type) * (__size + 1)));
 335               __cfixed[0] = static_cast<intern_type>(__int_bom);
 336               char_traits<intern_type>::copy(__cfixed + 1, __from, __size);
 337               __cfrom = reinterpret_cast<char*>(__cfixed);
 338               __conv = __iconv_adaptor(iconv, __desc, &__cfrom,
 339                                         &__fbytes, &__cto, &__tbytes);
 340             }
 341           else
 342             {
 343               intern_type* __cfixed = const_cast<intern_type*>(__from);
 344               __cfrom = reinterpret_cast<char*>(__cfixed);
 345               __conv = __iconv_adaptor(iconv, __desc, &__cfrom, &__fbytes,
 346                                        &__cto, &__tbytes);
 347             }
 348
 349           if (__conv != size_t(-1))
 350             {
 351               __from_next = reinterpret_cast<const intern_type*>(__cfrom);
 352               __to_next = reinterpret_cast<extern_type*>(__cto);
 353               __ret = codecvt_base::ok;
 354             }
 355           else
 356             {
 357               if (__fbytes < __fmultiple * (__from_end - __from))
 358                 {
 359                   __from_next = reinterpret_cast<const intern_type*>(__cfrom);
 360                   __to_next = reinterpret_cast<extern_type*>(__cto);
 361                   __ret = codecvt_base::partial;
 362                 }
 363               else
 364                 __ret = codecvt_base::error;
 365             }
 366         }
 367       return __ret;
 368     }
 369
 370   template<typename _InternT, typename _ExternT>
 371     codecvt_base::result
 372     codecvt<_InternT, _ExternT, encoding_state>::
 373     do_unshift(state_type& __state, extern_type* __to,
 374                extern_type* __to_end, extern_type*& __to_next) const
 375     {
 376       result __ret = codecvt_base::error;
 377       if (__state.good())
 378         {
 379           const descriptor_type& __desc = __state.in_descriptor();
 380           const size_t __tmultiple = sizeof(intern_type);
 381           size_t __tlen = __tmultiple * (__to_end - __to);
 382
 383           // Argument list for iconv specifies a byte sequence. Thus,
 384           // all to/from arrays must be brutally casted to char*.
 385           char* __cto = reinterpret_cast<char*>(__to);
 386           size_t __conv = __iconv_adaptor(iconv,__desc, NULL, NULL,
 387                                           &__cto, &__tlen);
 388
 389           if (__conv != size_t(-1))
 390             {
 391               __to_next = reinterpret_cast<extern_type*>(__cto);
 392               if (__tlen == __tmultiple * (__to_end - __to))
 393                 __ret = codecvt_base::noconv;
 394               else if (__tlen == 0)
 395                 __ret = codecvt_base::ok;
 396               else
 397                 __ret = codecvt_base::partial;
 398             }
 399           else
 400             __ret = codecvt_base::error;
 401         }
 402       return __ret;
 403     }
 404
 405   template<typename _InternT, typename _ExternT>
 406     codecvt_base::result
 407     codecvt<_InternT, _ExternT, encoding_state>::
 408     do_in(state_type& __state, const extern_type* __from,
 409           const extern_type* __from_end, const extern_type*& __from_next,
 410           intern_type* __to, intern_type* __to_end,
 411           intern_type*& __to_next) const
 412     {
 413       result __ret = codecvt_base::error;
 414       if (__state.good())
 415         {
 416           const descriptor_type& __desc = __state.in_descriptor();
 417           const size_t __fmultiple = sizeof(extern_type);
 418           size_t __flen = __fmultiple * (__from_end - __from);
 419           const size_t __tmultiple = sizeof(intern_type);
 420           size_t __tlen = __tmultiple * (__to_end - __to);
 421
 422           // Argument list for iconv specifies a byte sequence. Thus,
 423           // all to/from arrays must be brutally casted to char*.
 424           char* __cto = reinterpret_cast<char*>(__to);
 425           char* __cfrom;
 426           size_t __conv;
 427
 428           // Some encodings need a byte order marker as the first item
 429           // in the byte stream, to designate endian-ness. The default
 430           // value for the byte order marker is NULL, so if this is
 431           // the case, it's not necessary and we can just go on our
 432           // merry way.
 433           int __ext_bom = __state.external_bom();
 434           if (__ext_bom)
 435             {
 436               size_t __size = __from_end - __from;
 437               extern_type* __cfixed =  static_cast<extern_type*>
 438                 (__builtin_alloca(sizeof(extern_type) * (__size + 1)));
 439               __cfixed[0] = static_cast<extern_type>(__ext_bom);
 440               char_traits<extern_type>::copy(__cfixed + 1, __from, __size);
 441               __cfrom = reinterpret_cast<char*>(__cfixed);
 442               __conv = __iconv_adaptor(iconv, __desc, &__cfrom,
 443                                        &__flen, &__cto, &__tlen);
 444             }
 445           else
 446             {
 447               extern_type* __cfixed = const_cast<extern_type*>(__from);
 448               __cfrom = reinterpret_cast<char*>(__cfixed);
 449               __conv = __iconv_adaptor(iconv, __desc, &__cfrom,
 450                                        &__flen, &__cto, &__tlen);
 451             }
 452
 453
 454           if (__conv != size_t(-1))
 455             {
 456               __from_next = reinterpret_cast<const extern_type*>(__cfrom);
 457               __to_next = reinterpret_cast<intern_type*>(__cto);
 458               __ret = codecvt_base::ok;
 459             }
 460           else
 461             {
 462               if (__flen < static_cast<size_t>(__from_end - __from))
 463                 {
 464                   __from_next = reinterpret_cast<const extern_type*>(__cfrom);
 465                   __to_next = reinterpret_cast<intern_type*>(__cto);
 466                   __ret = codecvt_base::partial;
 467                 }
 468               else
 469                 __ret = codecvt_base::error;
 470             }
 471         }
 472       return __ret;
 473     }
 474
 475   template<typename _InternT, typename _ExternT>
 476     int
 477     codecvt<_InternT, _ExternT, encoding_state>::
 478     do_encoding() const throw()
 479     {
 480       int __ret = 0;
 481       if (sizeof(_ExternT) <= sizeof(_InternT))
 482         __ret = sizeof(_InternT) / sizeof(_ExternT);
 483       return __ret;
 484     }
 485
 486   template<typename _InternT, typename _ExternT>
 487     bool
 488     codecvt<_InternT, _ExternT, encoding_state>::
 489     do_always_noconv() const throw()
 490     { return false; }
 491
 492   template<typename _InternT, typename _ExternT>
 493     int
 494     codecvt<_InternT, _ExternT, encoding_state>::
 495     do_length(state_type&, const extern_type* __from,
 496               const extern_type* __end, size_t __max) const
 497     { return std::min(__max, static_cast<size_t>(__end - __from)); }
 498
 499   // _GLIBCXX_RESOLVE_LIB_DEFECTS
 500   // 74.  Garbled text for codecvt::do_max_length
 501   template<typename _InternT, typename _ExternT>
 502     int
 503     codecvt<_InternT, _ExternT, encoding_state>::
 504     do_max_length() const throw()
 505     { return 1; }
 506
 507 _GLIBCXX_END_NAMESPACE
 508