1 // Locale support (codecvt) -*- C++ -*-
3 // Copyright (C) 2000, 2001, 2002, 2003, 2004 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 /// @brief Extension to use icov for dealing with character encodings.
43 // This includes 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
);
91 // 21.1.2 traits typedefs
93 // typedef STATE_T state_type
94 // requires: state_type shall meet the requirements of
95 // CopyConstructible types (20.1.3)
96 // NB: This does not preseve the actual state of the conversion
97 // descriptor member, but it does duplicate the encoding
99 __enc_traits(const __enc_traits
& __obj
): _M_in_desc(0), _M_out_desc(0)
101 strncpy(_M_int_enc
, __obj
._M_int_enc
, _S_max_size
);
102 strncpy(_M_ext_enc
, __obj
._M_ext_enc
, _S_max_size
);
103 _M_ext_bom
= __obj
._M_ext_bom
;
104 _M_int_bom
= __obj
._M_int_bom
;
109 // Need assignment operator as well.
111 operator=(const __enc_traits
& __obj
)
113 strncpy(_M_int_enc
, __obj
._M_int_enc
, _S_max_size
);
114 strncpy(_M_ext_enc
, __obj
._M_ext_enc
, _S_max_size
);
115 _M_ext_bom
= __obj
._M_ext_bom
;
116 _M_int_bom
= __obj
._M_int_bom
;
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(__N("__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(__N("__enc_traits::_M_init "
141 "creating iconv output descriptor failed"));
148 const __desc_type __err
= reinterpret_cast<iconv_t
>(-1);
149 if (_M_in_desc
&& _M_in_desc
!= __err
)
151 iconv_close(_M_in_desc
);
154 if (_M_out_desc
&& _M_out_desc
!= __err
)
156 iconv_close(_M_out_desc
);
164 const __desc_type __err
= reinterpret_cast<iconv_t
>(-1);
165 bool __test
= _M_in_desc
&& _M_in_desc
!= __err
;
166 __test
&= _M_out_desc
&& _M_out_desc
!= __err
;
171 _M_get_in_descriptor()
172 { return &_M_in_desc
; }
175 _M_get_out_descriptor()
176 { return &_M_out_desc
; }
179 _M_get_external_bom()
180 { return _M_ext_bom
; }
183 _M_get_internal_bom()
184 { return _M_int_bom
; }
187 _M_get_internal_enc()
188 { return _M_int_enc
; }
191 _M_get_external_enc()
192 { return _M_ext_enc
; }
195 /// Partial specialization.
196 // This specialization takes advantage of iconv to provide code
197 // conversions between a large number of character encodings.
198 template<typename _InternT
, typename _ExternT
>
199 class codecvt
<_InternT
, _ExternT
, __enc_traits
>
200 : public __codecvt_abstract_base
<_InternT
, _ExternT
, __enc_traits
>
204 typedef codecvt_base::result result
;
205 typedef _InternT intern_type
;
206 typedef _ExternT extern_type
;
207 typedef __enc_traits state_type
;
208 typedef __enc_traits::__desc_type __desc_type
;
209 typedef __enc_traits __enc_type
;
212 static locale::id id
;
215 codecvt(size_t __refs
= 0)
216 : __codecvt_abstract_base
<intern_type
, extern_type
, state_type
>(__refs
)
220 codecvt(__enc_type
* __enc
, size_t __refs
= 0)
221 : __codecvt_abstract_base
<intern_type
, extern_type
, state_type
>(__refs
)
229 do_out(state_type
& __state
, const intern_type
* __from
,
230 const intern_type
* __from_end
, const intern_type
*& __from_next
,
231 extern_type
* __to
, extern_type
* __to_end
,
232 extern_type
*& __to_next
) const;
235 do_unshift(state_type
& __state
, extern_type
* __to
,
236 extern_type
* __to_end
, extern_type
*& __to_next
) const;
239 do_in(state_type
& __state
, const extern_type
* __from
,
240 const extern_type
* __from_end
, const extern_type
*& __from_next
,
241 intern_type
* __to
, intern_type
* __to_end
,
242 intern_type
*& __to_next
) const;
245 do_encoding() const throw();
248 do_always_noconv() const throw();
251 do_length(state_type
&, const extern_type
* __from
,
252 const extern_type
* __end
, size_t __max
) const;
255 do_max_length() const throw();
258 template<typename _InternT
, typename _ExternT
>
260 codecvt
<_InternT
, _ExternT
, __enc_traits
>::id
;
262 // This adaptor works around the signature problems of the second
263 // argument to iconv(): SUSv2 and others use 'const char**', but glibc 2.2
264 // uses 'char**', which matches the POSIX 1003.1-2001 standard.
265 // Using this adaptor, g++ will do the work for us.
266 template<typename _T
>
268 __iconv_adaptor(size_t(*__func
)(iconv_t
, _T
, size_t*, char**, size_t*),
269 iconv_t __cd
, char** __inbuf
, size_t* __inbytes
,
270 char** __outbuf
, size_t* __outbytes
)
271 { return __func(__cd
, (_T
)__inbuf
, __inbytes
, __outbuf
, __outbytes
); }
273 template<typename _InternT
, typename _ExternT
>
275 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
276 do_out(state_type
& __state
, const intern_type
* __from
,
277 const intern_type
* __from_end
, const intern_type
*& __from_next
,
278 extern_type
* __to
, extern_type
* __to_end
,
279 extern_type
*& __to_next
) const
281 result __ret
= codecvt_base::error
;
282 if (__state
._M_good())
284 typedef state_type::__desc_type __desc_type
;
285 const __desc_type
* __desc
= __state
._M_get_out_descriptor();
286 const size_t __fmultiple
= sizeof(intern_type
);
287 size_t __fbytes
= __fmultiple
* (__from_end
- __from
);
288 const size_t __tmultiple
= sizeof(extern_type
);
289 size_t __tbytes
= __tmultiple
* (__to_end
- __to
);
291 // Argument list for iconv specifies a byte sequence. Thus,
292 // all to/from arrays must be brutally casted to char*.
293 char* __cto
= reinterpret_cast<char*>(__to
);
297 // Some encodings need a byte order marker as the first item
298 // in the byte stream, to designate endian-ness. The default
299 // value for the byte order marker is NULL, so if this is
300 // the case, it's not necessary and we can just go on our
302 int __int_bom
= __state
._M_get_internal_bom();
305 size_t __size
= __from_end
- __from
;
306 intern_type
* __cfixed
= static_cast<intern_type
*>(__builtin_alloca(sizeof(intern_type
) * (__size
+ 1)));
307 __cfixed
[0] = static_cast<intern_type
>(__int_bom
);
308 char_traits
<intern_type
>::copy(__cfixed
+ 1, __from
, __size
);
309 __cfrom
= reinterpret_cast<char*>(__cfixed
);
310 __conv
= __iconv_adaptor(iconv
, *__desc
, &__cfrom
,
311 &__fbytes
, &__cto
, &__tbytes
);
315 intern_type
* __cfixed
= const_cast<intern_type
*>(__from
);
316 __cfrom
= reinterpret_cast<char*>(__cfixed
);
317 __conv
= __iconv_adaptor(iconv
, *__desc
, &__cfrom
, &__fbytes
,
321 if (__conv
!= size_t(-1))
323 __from_next
= reinterpret_cast<const intern_type
*>(__cfrom
);
324 __to_next
= reinterpret_cast<extern_type
*>(__cto
);
325 __ret
= codecvt_base::ok
;
329 if (__fbytes
< __fmultiple
* (__from_end
- __from
))
331 __from_next
= reinterpret_cast<const intern_type
*>(__cfrom
);
332 __to_next
= reinterpret_cast<extern_type
*>(__cto
);
333 __ret
= codecvt_base::partial
;
336 __ret
= codecvt_base::error
;
342 template<typename _InternT
, typename _ExternT
>
344 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
345 do_unshift(state_type
& __state
, extern_type
* __to
,
346 extern_type
* __to_end
, extern_type
*& __to_next
) const
348 result __ret
= codecvt_base::error
;
349 if (__state
._M_good())
351 typedef state_type::__desc_type __desc_type
;
352 const __desc_type
* __desc
= __state
._M_get_in_descriptor();
353 const size_t __tmultiple
= sizeof(intern_type
);
354 size_t __tlen
= __tmultiple
* (__to_end
- __to
);
356 // Argument list for iconv specifies a byte sequence. Thus,
357 // all to/from arrays must be brutally casted to char*.
358 char* __cto
= reinterpret_cast<char*>(__to
);
359 size_t __conv
= __iconv_adaptor(iconv
,*__desc
, NULL
, NULL
,
362 if (__conv
!= size_t(-1))
364 __to_next
= reinterpret_cast<extern_type
*>(__cto
);
365 if (__tlen
== __tmultiple
* (__to_end
- __to
))
366 __ret
= codecvt_base::noconv
;
367 else if (__tlen
== 0)
368 __ret
= codecvt_base::ok
;
370 __ret
= codecvt_base::partial
;
373 __ret
= codecvt_base::error
;
378 template<typename _InternT
, typename _ExternT
>
380 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
381 do_in(state_type
& __state
, const extern_type
* __from
,
382 const extern_type
* __from_end
, const extern_type
*& __from_next
,
383 intern_type
* __to
, intern_type
* __to_end
,
384 intern_type
*& __to_next
) const
386 result __ret
= codecvt_base::error
;
387 if (__state
._M_good())
389 typedef state_type::__desc_type __desc_type
;
390 const __desc_type
* __desc
= __state
._M_get_in_descriptor();
391 const size_t __fmultiple
= sizeof(extern_type
);
392 size_t __flen
= __fmultiple
* (__from_end
- __from
);
393 const size_t __tmultiple
= sizeof(intern_type
);
394 size_t __tlen
= __tmultiple
* (__to_end
- __to
);
396 // Argument list for iconv specifies a byte sequence. Thus,
397 // all to/from arrays must be brutally casted to char*.
398 char* __cto
= reinterpret_cast<char*>(__to
);
402 // Some encodings need a byte order marker as the first item
403 // in the byte stream, to designate endian-ness. The default
404 // value for the byte order marker is NULL, so if this is
405 // the case, it's not necessary and we can just go on our
407 int __ext_bom
= __state
._M_get_external_bom();
410 size_t __size
= __from_end
- __from
;
411 extern_type
* __cfixed
= static_cast<extern_type
*>(__builtin_alloca(sizeof(extern_type
) * (__size
+ 1)));
412 __cfixed
[0] = static_cast<extern_type
>(__ext_bom
);
413 char_traits
<extern_type
>::copy(__cfixed
+ 1, __from
, __size
);
414 __cfrom
= reinterpret_cast<char*>(__cfixed
);
415 __conv
= __iconv_adaptor(iconv
, *__desc
, &__cfrom
,
416 &__flen
, &__cto
, &__tlen
);
420 extern_type
* __cfixed
= const_cast<extern_type
*>(__from
);
421 __cfrom
= reinterpret_cast<char*>(__cfixed
);
422 __conv
= __iconv_adaptor(iconv
, *__desc
, &__cfrom
,
423 &__flen
, &__cto
, &__tlen
);
427 if (__conv
!= size_t(-1))
429 __from_next
= reinterpret_cast<const extern_type
*>(__cfrom
);
430 __to_next
= reinterpret_cast<intern_type
*>(__cto
);
431 __ret
= codecvt_base::ok
;
435 if (__flen
< static_cast<size_t>(__from_end
- __from
))
437 __from_next
= reinterpret_cast<const extern_type
*>(__cfrom
);
438 __to_next
= reinterpret_cast<intern_type
*>(__cto
);
439 __ret
= codecvt_base::partial
;
442 __ret
= codecvt_base::error
;
448 template<typename _InternT
, typename _ExternT
>
450 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
451 do_encoding() const throw()
454 if (sizeof(_ExternT
) <= sizeof(_InternT
))
455 __ret
= sizeof(_InternT
)/sizeof(_ExternT
);
459 template<typename _InternT
, typename _ExternT
>
461 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
462 do_always_noconv() const throw()
465 template<typename _InternT
, typename _ExternT
>
467 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
468 do_length(state_type
&, const extern_type
* __from
,
469 const extern_type
* __end
, size_t __max
) const
470 { return std::min(__max
, static_cast<size_t>(__end
- __from
)); }
472 // _GLIBCXX_RESOLVE_LIB_DEFECTS
473 // 74. Garbled text for codecvt::do_max_length
474 template<typename _InternT
, typename _ExternT
>
476 codecvt
<_InternT
, _ExternT
, __enc_traits
>::
477 do_max_length() const throw()