1 // Locale support (codecvt) -*- C++ -*-
3 // Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
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)
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.
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,
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.
31 // ISO C++ 14882: 22.2.1.5 Template class codecvt
34 // Warning: this file is not meant for user inclusion. Use <locale>.
36 // Written by Benjamin Kosnik <bkoz@cygnus.com>
39 // Define this here so codecvt.cc can have _S_max_size definition.
40 #define _GLIBCXX_USE___ENC_TRAITS 1
42 // Extension to use icov for dealing with character encodings,
43 // including conversions and comparisons between various character
44 // sets. This object encapsulates data that may need to be shared between
45 // char_traits, codecvt and ctype.
50 // NB: A conversion descriptor subsumes and enhances the
51 // functionality of a simple state type such as mbstate_t.
52 typedef iconv_t __desc_type;
56 // Max size of charset encoding name
57 static const int _S_max_size = 32;
58 // Name of internal character set encoding.
59 char _M_int_enc[_S_max_size];
60 // Name of external character set encoding.
61 char _M_ext_enc[_S_max_size];
63 // Conversion descriptor between external encoding to internal encoding.
64 __desc_type _M_in_desc;
65 // Conversion descriptor between internal encoding to external encoding.
66 __desc_type _M_out_desc;
68 // Details the byte-order marker for the external encoding, if necessary.
71 // Details the byte-order marker for the internal encoding, if necessary.
75 explicit __enc_traits()
76 : _M_in_desc(0), _M_out_desc(0), _M_ext_bom(0), _M_int_bom(0)
78 memset(_M_int_enc, 0, _S_max_size);
79 memset(_M_ext_enc, 0, _S_max_size);
82 explicit __enc_traits(const char* __int, const char* __ext,
83 int __ibom = 0, int __ebom = 0)
84 : _M_in_desc(0), _M_out_desc(0), _M_ext_bom(__ebom), _M_int_bom(__ibom)
86 strncpy(_M_int_enc, __int, _S_max_size);
87 strncpy(_M_ext_enc, __ext, _S_max_size);
90 // 21.1.2 traits typedefs
92 // typedef STATE_T state_type
93 // requires: state_type shall meet the requirements of
94 // CopyConstructible types (20.1.3)
95 __enc_traits(const __enc_traits& __obj): _M_in_desc(0), _M_out_desc(0)
97 strncpy(_M_int_enc, __obj._M_int_enc, _S_max_size);
98 strncpy(_M_ext_enc, __obj._M_ext_enc, _S_max_size);
99 _M_ext_bom = __obj._M_ext_bom;
100 _M_int_bom = __obj._M_int_bom;
103 // Need assignment operator as well.
105 operator=(const __enc_traits& __obj)
107 strncpy(_M_int_enc, __obj._M_int_enc, _S_max_size);
108 strncpy(_M_ext_enc, __obj._M_ext_enc, _S_max_size);
111 _M_ext_bom = __obj._M_ext_bom;
112 _M_int_bom = __obj._M_int_bom;
118 __desc_type __err = reinterpret_cast<iconv_t>(-1);
119 if (_M_in_desc && _M_in_desc != __err)
120 iconv_close(_M_in_desc);
121 if (_M_out_desc && _M_out_desc != __err)
122 iconv_close(_M_out_desc);
128 const __desc_type __err = reinterpret_cast<iconv_t>(-1);
131 _M_in_desc = iconv_open(_M_int_enc, _M_ext_enc);
132 if (_M_in_desc == __err)
133 __throw_runtime_error("__enc_traits::_M_init "
134 "creating iconv input descriptor failed");
138 _M_out_desc = iconv_open(_M_ext_enc, _M_int_enc);
139 if (_M_out_desc == __err)
140 __throw_runtime_error("__enc_traits::_M_init "
141 "creating iconv output descriptor failed");
148 const __desc_type __err = reinterpret_cast<iconv_t>(-1);
149 bool __test = _M_in_desc && _M_in_desc != __err;
150 __test &= _M_out_desc && _M_out_desc != __err;
155 _M_get_in_descriptor()
156 { return &_M_in_desc; }
159 _M_get_out_descriptor()
160 { return &_M_out_desc; }
163 _M_get_external_bom()
164 { return _M_ext_bom; }
167 _M_get_internal_bom()
168 { return _M_int_bom; }
171 _M_get_internal_enc()
172 { return _M_int_enc; }
175 _M_get_external_enc()
176 { return _M_ext_enc; }
179 // Partial specialization
180 // This specialization takes advantage of iconv to provide code
181 // conversions between a large number of character encodings.
182 template<typename _InternT, typename _ExternT>
183 class codecvt<_InternT, _ExternT, __enc_traits>
184 : public __codecvt_abstract_base<_InternT, _ExternT, __enc_traits>
188 typedef codecvt_base::result result;
189 typedef _InternT intern_type;
190 typedef _ExternT extern_type;
191 typedef __enc_traits state_type;
192 typedef __enc_traits::__desc_type __desc_type;
193 typedef __enc_traits __enc_type;
196 static locale::id id;
199 codecvt(size_t __refs = 0)
200 : __codecvt_abstract_base<intern_type, extern_type, state_type>(__refs)
204 codecvt(__enc_type* __enc, size_t __refs = 0)
205 : __codecvt_abstract_base<intern_type, extern_type, state_type>(__refs)
213 do_out(state_type& __state, const intern_type* __from,
214 const intern_type* __from_end, const intern_type*& __from_next,
215 extern_type* __to, extern_type* __to_end,
216 extern_type*& __to_next) const;
219 do_unshift(state_type& __state, extern_type* __to,
220 extern_type* __to_end, extern_type*& __to_next) const;
223 do_in(state_type& __state, const extern_type* __from,
224 const extern_type* __from_end, const extern_type*& __from_next,
225 intern_type* __to, intern_type* __to_end,
226 intern_type*& __to_next) const;
229 do_encoding() const throw();
232 do_always_noconv() const throw();
235 do_length(state_type&, const extern_type* __from,
236 const extern_type* __end, size_t __max) const;
239 do_max_length() const throw();
242 template<typename _InternT, typename _ExternT>
244 codecvt<_InternT, _ExternT, __enc_traits>::id;
246 // This adaptor works around the signature problems of the second
247 // argument to iconv(): SUSv2 and others use 'const char**', but glibc 2.2
248 // uses 'char**', which matches the POSIX 1003.1-2001 standard.
249 // Using this adaptor, g++ will do the work for us.
250 template<typename _T>
252 __iconv_adaptor(size_t(*__func)(iconv_t, _T, size_t*, char**, size_t*),
253 iconv_t __cd, char** __inbuf, size_t* __inbytes,
254 char** __outbuf, size_t* __outbytes)
255 { return __func(__cd, (_T)__inbuf, __inbytes, __outbuf, __outbytes); }
257 template<typename _InternT, typename _ExternT>
259 codecvt<_InternT, _ExternT, __enc_traits>::
260 do_out(state_type& __state, const intern_type* __from,
261 const intern_type* __from_end, const intern_type*& __from_next,
262 extern_type* __to, extern_type* __to_end,
263 extern_type*& __to_next) const
265 result __ret = codecvt_base::error;
266 if (__state._M_good())
268 typedef state_type::__desc_type __desc_type;
269 const __desc_type* __desc = __state._M_get_out_descriptor();
270 const size_t __fmultiple = sizeof(intern_type);
271 size_t __fbytes = __fmultiple * (__from_end - __from);
272 const size_t __tmultiple = sizeof(extern_type);
273 size_t __tbytes = __tmultiple * (__to_end - __to);
275 // Argument list for iconv specifies a byte sequence. Thus,
276 // all to/from arrays must be brutally casted to char*.
277 char* __cto = reinterpret_cast<char*>(__to);
281 // Some encodings need a byte order marker as the first item
282 // in the byte stream, to designate endian-ness. The default
283 // value for the byte order marker is NULL, so if this is
284 // the case, it's not necessary and we can just go on our
286 int __int_bom = __state._M_get_internal_bom();
289 size_t __size = __from_end - __from;
290 intern_type* __cfixed = static_cast<intern_type*>(__builtin_alloca(sizeof(intern_type) * (__size + 1)));
291 __cfixed[0] = static_cast<intern_type>(__int_bom);
292 char_traits<intern_type>::copy(__cfixed + 1, __from, __size);
293 __cfrom = reinterpret_cast<char*>(__cfixed);
294 __conv = __iconv_adaptor(iconv, *__desc, &__cfrom,
295 &__fbytes, &__cto, &__tbytes);
299 intern_type* __cfixed = const_cast<intern_type*>(__from);
300 __cfrom = reinterpret_cast<char*>(__cfixed);
301 __conv = __iconv_adaptor(iconv, *__desc, &__cfrom, &__fbytes,
305 if (__conv != size_t(-1))
307 __from_next = reinterpret_cast<const intern_type*>(__cfrom);
308 __to_next = reinterpret_cast<extern_type*>(__cto);
309 __ret = codecvt_base::ok;
313 if (__fbytes < __fmultiple * (__from_end - __from))
315 __from_next = reinterpret_cast<const intern_type*>(__cfrom);
316 __to_next = reinterpret_cast<extern_type*>(__cto);
317 __ret = codecvt_base::partial;
320 __ret = codecvt_base::error;
326 template<typename _InternT, typename _ExternT>
328 codecvt<_InternT, _ExternT, __enc_traits>::
329 do_unshift(state_type& __state, extern_type* __to,
330 extern_type* __to_end, extern_type*& __to_next) const
332 result __ret = codecvt_base::error;
333 if (__state._M_good())
335 typedef state_type::__desc_type __desc_type;
336 const __desc_type* __desc = __state._M_get_in_descriptor();
337 const size_t __tmultiple = sizeof(intern_type);
338 size_t __tlen = __tmultiple * (__to_end - __to);
340 // Argument list for iconv specifies a byte sequence. Thus,
341 // all to/from arrays must be brutally casted to char*.
342 char* __cto = reinterpret_cast<char*>(__to);
343 size_t __conv = __iconv_adaptor(iconv,*__desc, NULL, NULL,
346 if (__conv != size_t(-1))
348 __to_next = reinterpret_cast<extern_type*>(__cto);
349 if (__tlen == __tmultiple * (__to_end - __to))
350 __ret = codecvt_base::noconv;
351 else if (__tlen == 0)
352 __ret = codecvt_base::ok;
354 __ret = codecvt_base::partial;
357 __ret = codecvt_base::error;
362 template<typename _InternT, typename _ExternT>
364 codecvt<_InternT, _ExternT, __enc_traits>::
365 do_in(state_type& __state, const extern_type* __from,
366 const extern_type* __from_end, const extern_type*& __from_next,
367 intern_type* __to, intern_type* __to_end,
368 intern_type*& __to_next) const
370 result __ret = codecvt_base::error;
371 if (__state._M_good())
373 typedef state_type::__desc_type __desc_type;
374 const __desc_type* __desc = __state._M_get_in_descriptor();
375 const size_t __fmultiple = sizeof(extern_type);
376 size_t __flen = __fmultiple * (__from_end - __from);
377 const size_t __tmultiple = sizeof(intern_type);
378 size_t __tlen = __tmultiple * (__to_end - __to);
380 // Argument list for iconv specifies a byte sequence. Thus,
381 // all to/from arrays must be brutally casted to char*.
382 char* __cto = reinterpret_cast<char*>(__to);
386 // Some encodings need a byte order marker as the first item
387 // in the byte stream, to designate endian-ness. The default
388 // value for the byte order marker is NULL, so if this is
389 // the case, it's not necessary and we can just go on our
391 int __ext_bom = __state._M_get_external_bom();
394 size_t __size = __from_end - __from;
395 extern_type* __cfixed = static_cast<extern_type*>(__builtin_alloca(sizeof(extern_type) * (__size + 1)));
396 __cfixed[0] = static_cast<extern_type>(__ext_bom);
397 char_traits<extern_type>::copy(__cfixed + 1, __from, __size);
398 __cfrom = reinterpret_cast<char*>(__cfixed);
399 __conv = __iconv_adaptor(iconv, *__desc, &__cfrom,
400 &__flen, &__cto, &__tlen);
404 extern_type* __cfixed = const_cast<extern_type*>(__from);
405 __cfrom = reinterpret_cast<char*>(__cfixed);
406 __conv = __iconv_adaptor(iconv, *__desc, &__cfrom,
407 &__flen, &__cto, &__tlen);
411 if (__conv != size_t(-1))
413 __from_next = reinterpret_cast<const extern_type*>(__cfrom);
414 __to_next = reinterpret_cast<intern_type*>(__cto);
415 __ret = codecvt_base::ok;
419 if (__flen < static_cast<size_t>(__from_end - __from))
421 __from_next = reinterpret_cast<const extern_type*>(__cfrom);
422 __to_next = reinterpret_cast<intern_type*>(__cto);
423 __ret = codecvt_base::partial;
426 __ret = codecvt_base::error;
432 template<typename _InternT, typename _ExternT>
434 codecvt<_InternT, _ExternT, __enc_traits>::
435 do_encoding() const throw()
438 if (sizeof(_ExternT) <= sizeof(_InternT))
439 __ret = sizeof(_InternT)/sizeof(_ExternT);
443 template<typename _InternT, typename _ExternT>
445 codecvt<_InternT, _ExternT, __enc_traits>::
446 do_always_noconv() const throw()
449 template<typename _InternT, typename _ExternT>
451 codecvt<_InternT, _ExternT, __enc_traits>::
452 do_length(state_type&, const extern_type* __from,
453 const extern_type* __end, size_t __max) const
454 { return std::min(__max, static_cast<size_t>(__end - __from)); }
456 // _GLIBCXX_RESOLVE_LIB_DEFECTS
457 // 74. Garbled text for codecvt::do_max_length
458 template<typename _InternT, typename _ExternT>
460 codecvt<_InternT, _ExternT, __enc_traits>::
461 do_max_length() const throw()