1// wstring_convert implementation -*- C++ -*-
2
3// Copyright (C) 2015-2018 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 3, 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// Under Section 7 of GPL version 3, you are granted additional
17// permissions described in the GCC Runtime Library Exception, version
18// 3.1, as published by the Free Software Foundation.
19
20// You should have received a copy of the GNU General Public License and
21// a copy of the GCC Runtime Library Exception along with this program;
22// see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
23// <http://www.gnu.org/licenses/>.
24
25/** @file bits/locale_conv.h
26 * This is an internal header file, included by other library headers.
27 * Do not attempt to use it directly. @headername{locale}
28 */
29
30#ifndef _LOCALE_CONV_H
31#define _LOCALE_CONV_H 1
32
33#if __cplusplus < 201103L
34# include <bits/c++0x_warning.h>
35#else
36
37#include <streambuf>
38#include "stringfwd.h"
39#include "allocator.h"
40#include "codecvt.h"
41#include "unique_ptr.h"
42
43namespace std _GLIBCXX_VISIBILITY(default)
44{
45_GLIBCXX_BEGIN_NAMESPACE_VERSION
46
47 /**
48 * @addtogroup locales
49 * @{
50 */
51
52 template<typename _OutStr, typename _InChar, typename _Codecvt,
53 typename _State, typename _Fn>
54 bool
55 __do_str_codecvt(const _InChar* __first, const _InChar* __last,
56 _OutStr& __outstr, const _Codecvt& __cvt, _State& __state,
57 size_t& __count, _Fn __fn)
58 {
59 if (__first == __last)
60 {
61 __outstr.clear();
62 __count = 0;
63 return true;
64 }
65
66 size_t __outchars = 0;
67 auto __next = __first;
68 const auto __maxlen = __cvt.max_length() + 1;
69
70 codecvt_base::result __result;
71 do
72 {
73 __outstr.resize(__outstr.size() + (__last - __next) * __maxlen);
74 auto __outnext = &__outstr.front() + __outchars;
75 auto const __outlast = &__outstr.back() + 1;
76 __result = (__cvt.*__fn)(__state, __next, __last, __next,
77 __outnext, __outlast, __outnext);
78 __outchars = __outnext - &__outstr.front();
79 }
80 while (__result == codecvt_base::partial && __next != __last
81 && (__outstr.size() - __outchars) < __maxlen);
82
83 if (__result == codecvt_base::error)
84 {
85 __count = __next - __first;
86 return false;
87 }
88
89 if (__result == codecvt_base::noconv)
90 {
91 __outstr.assign(__first, __last);
92 __count = __last - __first;
93 }
94 else
95 {
96 __outstr.resize(__outchars);
97 __count = __next - __first;
98 }
99
100 return true;
101 }
102
103 // Convert narrow character string to wide.
104 template<typename _CharT, typename _Traits, typename _Alloc, typename _State>
105 inline bool
106 __str_codecvt_in(const char* __first, const char* __last,
107 basic_string<_CharT, _Traits, _Alloc>& __outstr,
108 const codecvt<_CharT, char, _State>& __cvt,
109 _State& __state, size_t& __count)
110 {
111 using _Codecvt = codecvt<_CharT, char, _State>;
112 using _ConvFn
113 = codecvt_base::result
114 (_Codecvt::*)(_State&, const char*, const char*, const char*&,
115 _CharT*, _CharT*, _CharT*&) const;
116 _ConvFn __fn = &codecvt<_CharT, char, _State>::in;
117 return __do_str_codecvt(__first, __last, __outstr, __cvt, __state,
118 __count, __fn);
119 }
120
121 template<typename _CharT, typename _Traits, typename _Alloc, typename _State>
122 inline bool
123 __str_codecvt_in(const char* __first, const char* __last,
124 basic_string<_CharT, _Traits, _Alloc>& __outstr,
125 const codecvt<_CharT, char, _State>& __cvt)
126 {
127 _State __state = {};
128 size_t __n;
129 return __str_codecvt_in(__first, __last, __outstr, __cvt, __state, __n);
130 }
131
132 // Convert wide character string to narrow.
133 template<typename _CharT, typename _Traits, typename _Alloc, typename _State>
134 inline bool
135 __str_codecvt_out(const _CharT* __first, const _CharT* __last,
136 basic_string<char, _Traits, _Alloc>& __outstr,
137 const codecvt<_CharT, char, _State>& __cvt,
138 _State& __state, size_t& __count)
139 {
140 using _Codecvt = codecvt<_CharT, char, _State>;
141 using _ConvFn
142 = codecvt_base::result
143 (_Codecvt::*)(_State&, const _CharT*, const _CharT*, const _CharT*&,
144 char*, char*, char*&) const;
145 _ConvFn __fn = &codecvt<_CharT, char, _State>::out;
146 return __do_str_codecvt(__first, __last, __outstr, __cvt, __state,
147 __count, __fn);
148 }
149
150 template<typename _CharT, typename _Traits, typename _Alloc, typename _State>
151 inline bool
152 __str_codecvt_out(const _CharT* __first, const _CharT* __last,
153 basic_string<char, _Traits, _Alloc>& __outstr,
154 const codecvt<_CharT, char, _State>& __cvt)
155 {
156 _State __state = {};
157 size_t __n;
158 return __str_codecvt_out(__first, __last, __outstr, __cvt, __state, __n);
159 }
160
161#ifdef _GLIBCXX_USE_WCHAR_T
162
163_GLIBCXX_BEGIN_NAMESPACE_CXX11
164
165 /// String conversions
166 template<typename _Codecvt, typename _Elem = wchar_t,
167 typename _Wide_alloc = allocator<_Elem>,
168 typename _Byte_alloc = allocator<char>>
169 class wstring_convert
170 {
171 public:
172 typedef basic_string<char, char_traits<char>, _Byte_alloc> byte_string;
173 typedef basic_string<_Elem, char_traits<_Elem>, _Wide_alloc> wide_string;
174 typedef typename _Codecvt::state_type state_type;
175 typedef typename wide_string::traits_type::int_type int_type;
176
177 /** Default constructor.
178 *
179 * @param __pcvt The facet to use for conversions.
180 *
181 * Takes ownership of @p __pcvt and will delete it in the destructor.
182 */
183 explicit
184 wstring_convert(_Codecvt* __pcvt = new _Codecvt()) : _M_cvt(__pcvt)
185 {
186 if (!_M_cvt)
187 __throw_logic_error("wstring_convert");
188 }
189
190 /** Construct with an initial converstion state.
191 *
192 * @param __pcvt The facet to use for conversions.
193 * @param __state Initial conversion state.
194 *
195 * Takes ownership of @p __pcvt and will delete it in the destructor.
196 * The object's conversion state will persist between conversions.
197 */
198 wstring_convert(_Codecvt* __pcvt, state_type __state)
199 : _M_cvt(__pcvt), _M_state(__state), _M_with_cvtstate(true)
200 {
201 if (!_M_cvt)
202 __throw_logic_error("wstring_convert");
203 }
204
205 /** Construct with error strings.
206 *
207 * @param __byte_err A string to return on failed conversions.
208 * @param __wide_err A wide string to return on failed conversions.
209 */
210 explicit
211 wstring_convert(const byte_string& __byte_err,
212 const wide_string& __wide_err = wide_string())
213 : _M_cvt(new _Codecvt),
214 _M_byte_err_string(__byte_err), _M_wide_err_string(__wide_err),
215 _M_with_strings(true)
216 {
217 if (!_M_cvt)
218 __throw_logic_error("wstring_convert");
219 }
220
221 ~wstring_convert() = default;
222
223 // _GLIBCXX_RESOLVE_LIB_DEFECTS
224 // 2176. Special members for wstring_convert and wbuffer_convert
225 wstring_convert(const wstring_convert&) = delete;
226 wstring_convert& operator=(const wstring_convert&) = delete;
227
228 /// @{ Convert from bytes.
229 wide_string
230 from_bytes(char __byte)
231 {
232 char __bytes[2] = { __byte };
233 return from_bytes(__bytes, __bytes+1);
234 }
235
236 wide_string
237 from_bytes(const char* __ptr)
238 { return from_bytes(__ptr, __ptr+char_traits<char>::length(__ptr)); }
239
240 wide_string
241 from_bytes(const byte_string& __str)
242 {
243 auto __ptr = __str.data();
244 return from_bytes(__ptr, __ptr + __str.size());
245 }
246
247 wide_string
248 from_bytes(const char* __first, const char* __last)
249 {
250 if (!_M_with_cvtstate)
251 _M_state = state_type();
252 wide_string __out{ _M_wide_err_string.get_allocator() };
253 if (__str_codecvt_in(__first, __last, __out, *_M_cvt, _M_state,
254 _M_count))
255 return __out;
256 if (_M_with_strings)
257 return _M_wide_err_string;
258 __throw_range_error("wstring_convert::from_bytes");
259 }
260 /// @}
261
262 /// @{ Convert to bytes.
263 byte_string
264 to_bytes(_Elem __wchar)
265 {
266 _Elem __wchars[2] = { __wchar };
267 return to_bytes(__wchars, __wchars+1);
268 }
269
270 byte_string
271 to_bytes(const _Elem* __ptr)
272 {
273 return to_bytes(__ptr, __ptr+wide_string::traits_type::length(__ptr));
274 }
275
276 byte_string
277 to_bytes(const wide_string& __wstr)
278 {
279 auto __ptr = __wstr.data();
280 return to_bytes(__ptr, __ptr + __wstr.size());
281 }
282
283 byte_string
284 to_bytes(const _Elem* __first, const _Elem* __last)
285 {
286 if (!_M_with_cvtstate)
287 _M_state = state_type();
288 byte_string __out{ _M_byte_err_string.get_allocator() };
289 if (__str_codecvt_out(__first, __last, __out, *_M_cvt, _M_state,
290 _M_count))
291 return __out;
292 if (_M_with_strings)
293 return _M_byte_err_string;
294 __throw_range_error("wstring_convert::to_bytes");
295 }
296 /// @}
297
298 // _GLIBCXX_RESOLVE_LIB_DEFECTS
299 // 2174. wstring_convert::converted() should be noexcept
300 /// The number of elements successfully converted in the last conversion.
301 size_t converted() const noexcept { return _M_count; }
302
303 /// The final conversion state of the last conversion.
304 state_type state() const { return _M_state; }
305
306 private:
307 unique_ptr<_Codecvt> _M_cvt;
308 byte_string _M_byte_err_string;
309 wide_string _M_wide_err_string;
310 state_type _M_state = state_type();
311 size_t _M_count = 0;
312 bool _M_with_cvtstate = false;
313 bool _M_with_strings = false;
314 };
315
316_GLIBCXX_END_NAMESPACE_CXX11
317
318 /// Buffer conversions
319 template<typename _Codecvt, typename _Elem = wchar_t,
320 typename _Tr = char_traits<_Elem>>
321 class wbuffer_convert : public basic_streambuf<_Elem, _Tr>
322 {
323 typedef basic_streambuf<_Elem, _Tr> _Wide_streambuf;
324
325 public:
326 typedef typename _Codecvt::state_type state_type;
327
328 /** Default constructor.
329 *
330 * @param __bytebuf The underlying byte stream buffer.
331 * @param __pcvt The facet to use for conversions.
332 * @param __state Initial conversion state.
333 *
334 * Takes ownership of @p __pcvt and will delete it in the destructor.
335 */
336 explicit
337 wbuffer_convert(streambuf* __bytebuf = 0, _Codecvt* __pcvt = new _Codecvt,
338 state_type __state = state_type())
339 : _M_buf(__bytebuf), _M_cvt(__pcvt), _M_state(__state)
340 {
341 if (!_M_cvt)
342 __throw_logic_error("wbuffer_convert");
343
344 _M_always_noconv = _M_cvt->always_noconv();
345
346 if (_M_buf)
347 {
348 this->setp(_M_put_area, _M_put_area + _S_buffer_length);
349 this->setg(_M_get_area + _S_putback_length,
350 _M_get_area + _S_putback_length,
351 _M_get_area + _S_putback_length);
352 }
353 }
354
355 ~wbuffer_convert() = default;
356
357 // _GLIBCXX_RESOLVE_LIB_DEFECTS
358 // 2176. Special members for wstring_convert and wbuffer_convert
359 wbuffer_convert(const wbuffer_convert&) = delete;
360 wbuffer_convert& operator=(const wbuffer_convert&) = delete;
361
362 streambuf* rdbuf() const noexcept { return _M_buf; }
363
364 streambuf*
365 rdbuf(streambuf *__bytebuf) noexcept
366 {
367 auto __prev = _M_buf;
368 _M_buf = __bytebuf;
369 return __prev;
370 }
371
372 /// The conversion state following the last conversion.
373 state_type state() const noexcept { return _M_state; }
374
375 protected:
376 int
377 sync()
378 { return _M_buf && _M_conv_put() && !_M_buf->pubsync() ? 0 : -1; }
379
380 typename _Wide_streambuf::int_type
381 overflow(typename _Wide_streambuf::int_type __out)
382 {
383 if (!_M_buf || !_M_conv_put())
384 return _Tr::eof();
385 else if (!_Tr::eq_int_type(__out, _Tr::eof()))
386 return this->sputc(__out);
387 return _Tr::not_eof(__out);
388 }
389
390 typename _Wide_streambuf::int_type
391 underflow()
392 {
393 if (!_M_buf)
394 return _Tr::eof();
395
396 if (this->gptr() < this->egptr() || (_M_buf && _M_conv_get()))
397 return _Tr::to_int_type(*this->gptr());
398 else
399 return _Tr::eof();
400 }
401
402 streamsize
403 xsputn(const typename _Wide_streambuf::char_type* __s, streamsize __n)
404 {
405 if (!_M_buf || __n == 0)
406 return 0;
407 streamsize __done = 0;
408 do
409 {
410 auto __nn = std::min<streamsize>(this->epptr() - this->pptr(),
411 __n - __done);
412 _Tr::copy(this->pptr(), __s + __done, __nn);
413 this->pbump(__nn);
414 __done += __nn;
415 } while (__done < __n && _M_conv_put());
416 return __done;
417 }
418
419 private:
420 // fill the get area from converted contents of the byte stream buffer
421 bool
422 _M_conv_get()
423 {
424 const streamsize __pb1 = this->gptr() - this->eback();
425 const streamsize __pb2 = _S_putback_length;
426 const streamsize __npb = std::min(__pb1, __pb2);
427
428 _Tr::move(_M_get_area + _S_putback_length - __npb,
429 this->gptr() - __npb, __npb);
430
431 streamsize __nbytes = sizeof(_M_get_buf) - _M_unconv;
432 __nbytes = std::min(__nbytes, _M_buf->in_avail());
433 if (__nbytes < 1)
434 __nbytes = 1;
435 __nbytes = _M_buf->sgetn(_M_get_buf + _M_unconv, __nbytes);
436 if (__nbytes < 1)
437 return false;
438 __nbytes += _M_unconv;
439
440 // convert _M_get_buf into _M_get_area
441
442 _Elem* __outbuf = _M_get_area + _S_putback_length;
443 _Elem* __outnext = __outbuf;
444 const char* __bnext = _M_get_buf;
445
446 codecvt_base::result __result;
447 if (_M_always_noconv)
448 __result = codecvt_base::noconv;
449 else
450 {
451 _Elem* __outend = _M_get_area + _S_buffer_length;
452
453 __result = _M_cvt->in(_M_state,
454 __bnext, __bnext + __nbytes, __bnext,
455 __outbuf, __outend, __outnext);
456 }
457
458 if (__result == codecvt_base::noconv)
459 {
460 // cast is safe because noconv means _Elem is same type as char
461 auto __get_buf = reinterpret_cast<const _Elem*>(_M_get_buf);
462 _Tr::copy(__outbuf, __get_buf, __nbytes);
463 _M_unconv = 0;
464 return true;
465 }
466
467 if ((_M_unconv = _M_get_buf + __nbytes - __bnext))
468 char_traits<char>::move(_M_get_buf, __bnext, _M_unconv);
469
470 this->setg(__outbuf, __outbuf, __outnext);
471
472 return __result != codecvt_base::error;
473 }
474
475 // unused
476 bool
477 _M_put(...)
478 { return false; }
479
480 bool
481 _M_put(const char* __p, streamsize __n)
482 {
483 if (_M_buf->sputn(__p, __n) < __n)
484 return false;
485 return true;
486 }
487
488 // convert the put area and write to the byte stream buffer
489 bool
490 _M_conv_put()
491 {
492 _Elem* const __first = this->pbase();
493 const _Elem* const __last = this->pptr();
494 const streamsize __pending = __last - __first;
495
496 if (_M_always_noconv)
497 return _M_put(__first, __pending);
498
499 char __outbuf[2 * _S_buffer_length];
500
501 const _Elem* __next = __first;
502 const _Elem* __start;
503 do
504 {
505 __start = __next;
506 char* __outnext = __outbuf;
507 char* const __outlast = __outbuf + sizeof(__outbuf);
508 auto __result = _M_cvt->out(_M_state, __next, __last, __next,
509 __outnext, __outlast, __outnext);
510 if (__result == codecvt_base::error)
511 return false;
512 else if (__result == codecvt_base::noconv)
513 return _M_put(__next, __pending);
514
515 if (!_M_put(__outbuf, __outnext - __outbuf))
516 return false;
517 }
518 while (__next != __last && __next != __start);
519
520 if (__next != __last)
521 _Tr::move(__first, __next, __last - __next);
522
523 this->pbump(__first - __next);
524 return __next != __first;
525 }
526
527 streambuf* _M_buf;
528 unique_ptr<_Codecvt> _M_cvt;
529 state_type _M_state;
530
531 static const streamsize _S_buffer_length = 32;
532 static const streamsize _S_putback_length = 3;
533 _Elem _M_put_area[_S_buffer_length];
534 _Elem _M_get_area[_S_buffer_length];
535 streamsize _M_unconv = 0;
536 char _M_get_buf[_S_buffer_length-_S_putback_length];
537 bool _M_always_noconv;
538 };
539
540#endif // _GLIBCXX_USE_WCHAR_T
541
542 /// @} group locales
543
544_GLIBCXX_END_NAMESPACE_VERSION
545} // namespace
546
547#endif // __cplusplus
548
549#endif /* _LOCALE_CONV_H */
550