Minor corrections to ToWChar() comment.
[wxWidgets.git] / include / wx / strconv.h
1 ///////////////////////////////////////////////////////////////////////////////
2 // Name: strconv.h
3 // Purpose: conversion routines for char sets any Unicode
4 // Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin
5 // Modified by:
6 // Created: 29/01/98
7 // RCS-ID: $Id$
8 // Copyright: (c) 1998 Ove Kaaven, Robert Roebling
9 // (c) 1998-2006 Vadim Zeitlin
10 // Licence: wxWindows licence
11 ///////////////////////////////////////////////////////////////////////////////
12
13 #ifndef _WX_STRCONV_H_
14 #define _WX_STRCONV_H_
15
16 #include "wx/defs.h"
17 #include "wx/chartype.h"
18 #include "wx/buffer.h"
19
20 #ifdef __DIGITALMARS__
21 #include "typeinfo.h"
22 #endif
23
24 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
25 # undef __BSEXCPT__
26 #endif
27
28 #ifndef __WXPALMOS5__
29 #include <stdlib.h>
30 #endif // ! __WXPALMOS5__
31
32 #if wxUSE_WCHAR_T
33
34 class WXDLLIMPEXP_FWD_BASE wxString;
35
36 // the error value returned by wxMBConv methods
37 #define wxCONV_FAILED ((size_t)-1)
38
39 // ----------------------------------------------------------------------------
40 // wxMBConv (abstract base class for conversions)
41 // ----------------------------------------------------------------------------
42
43 // When deriving a new class from wxMBConv you must reimplement ToWChar() and
44 // FromWChar() methods which are not pure virtual only for historical reasons,
45 // don't let the fact that the existing classes implement MB2WC/WC2MB() instead
46 // confuse you.
47 //
48 // You also have to implement Clone() to allow copying the conversions
49 // polymorphically.
50 //
51 // And you might need to override GetMBNulLen() as well.
52 class WXDLLIMPEXP_BASE wxMBConv
53 {
54 public:
55 // The functions doing actual conversion from/to narrow to/from wide
56 // character strings.
57 //
58 // On success, the return value is the length (i.e. the number of
59 // characters, not bytes) of the converted string including any trailing
60 // L'\0' or (possibly multiple) '\0'(s). If the conversion fails or if
61 // there is not enough space for everything, including the trailing NUL
62 // character(s), in the output buffer, wxCONV_FAILED is returned.
63 //
64 // In the special case when dst is NULL (the value of dstLen is ignored
65 // then) the return value is the length of the needed buffer but nothing
66 // happens otherwise. If srcLen is wxNO_LEN, the entire string, up to and
67 // including the trailing NUL(s), is converted, otherwise exactly srcLen
68 // bytes are.
69 //
70 // Typical usage:
71 //
72 // size_t dstLen = conv.ToWChar(NULL, 0, src);
73 // if ( dstLen == wxCONV_FAILED )
74 // ... handle error ...
75 // wchar_t *wbuf = new wchar_t[dstLen];
76 // conv.ToWChar(wbuf, dstLen, src);
77 // ... work with wbuf ...
78 // delete [] wbuf;
79 //
80 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
81 const char *src, size_t srcLen = wxNO_LEN) const;
82
83 virtual size_t FromWChar(char *dst, size_t dstLen,
84 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
85
86
87 // Convenience functions for translating NUL-terminated strings: returns
88 // the buffer containing the converted string or NULL pointer if the
89 // conversion failed.
90 const wxWCharBuffer cMB2WC(const char *in) const;
91 const wxCharBuffer cWC2MB(const wchar_t *in) const;
92
93 // Convenience functions for converting strings which may contain embedded
94 // NULs and don't have to be NUL-terminated.
95 //
96 // inLen is the length of the buffer including trailing NUL if any or
97 // wxNO_LEN if the input is NUL-terminated.
98 //
99 // outLen receives, if not NULL, the length of the converted string or 0 if
100 // the conversion failed (returning 0 and not -1 in this case makes it
101 // difficult to distinguish between failed conversion and empty input but
102 // this is done for backwards compatibility). Notice that the rules for
103 // whether outLen accounts or not for the last NUL are the same as for
104 // To/FromWChar() above: if inLen is specified, outLen is exactly the
105 // number of characters converted, whether the last one of them was NUL or
106 // not. But if inLen == wxNO_LEN then outLen doesn't account for the last
107 // NUL even though it is present.
108 const wxWCharBuffer
109 cMB2WC(const char *in, size_t inLen, size_t *outLen) const;
110 const wxCharBuffer
111 cWC2MB(const wchar_t *in, size_t inLen, size_t *outLen) const;
112
113 // And yet more convenience functions for converting the entire buffers:
114 // these are the simplest and least error-prone as you never need to bother
115 // with lengths/sizes directly.
116 const wxWCharBuffer cMB2WC(const wxScopedCharBuffer& in) const;
117 const wxCharBuffer cWC2MB(const wxScopedWCharBuffer& in) const;
118
119 // convenience functions for converting MB or WC to/from wxWin default
120 #if wxUSE_UNICODE
121 const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
122 const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }
123 const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; }
124 const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; }
125 #else // ANSI
126 const char* cMB2WX(const char *psz) const { return psz; }
127 const char* cWX2MB(const char *psz) const { return psz; }
128 const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); }
129 const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); }
130 #endif // Unicode/ANSI
131
132 // this function is used in the implementation of cMB2WC() to distinguish
133 // between the following cases:
134 //
135 // a) var width encoding with strings terminated by a single NUL
136 // (usual multibyte encodings): return 1 in this case
137 // b) fixed width encoding with 2 bytes/char and so terminated by
138 // 2 NULs (UTF-16/UCS-2 and variants): return 2 in this case
139 // c) fixed width encoding with 4 bytes/char and so terminated by
140 // 4 NULs (UTF-32/UCS-4 and variants): return 4 in this case
141 //
142 // anything else is not supported currently and -1 should be returned
143 virtual size_t GetMBNulLen() const { return 1; }
144
145 // return the maximal value currently returned by GetMBNulLen() for any
146 // encoding
147 static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; }
148
149 #if wxUSE_UNICODE_UTF8
150 // return true if the converter's charset is UTF-8, i.e. char* strings
151 // decoded using this object can be directly copied to wxString's internal
152 // storage without converting to WC and than back to UTF-8 MB string
153 virtual bool IsUTF8() const { return false; }
154 #endif
155
156 // The old conversion functions. The existing classes currently mostly
157 // implement these ones but we're in transition to using To/FromWChar()
158 // instead and any new classes should implement just the new functions.
159 // For now, however, we provide default implementation of To/FromWChar() in
160 // this base class in terms of MB2WC/WC2MB() to avoid having to rewrite all
161 // the conversions at once.
162 //
163 // On success, the return value is the length (i.e. the number of
164 // characters, not bytes) not counting the trailing NUL(s) of the converted
165 // string. On failure, (size_t)-1 is returned. In the special case when
166 // outputBuf is NULL the return value is the same one but nothing is
167 // written to the buffer.
168 //
169 // Note that outLen is the length of the output buffer, not the length of
170 // the input (which is always supposed to be terminated by one or more
171 // NULs, as appropriate for the encoding)!
172 virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const;
173 virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const;
174
175
176 // make a heap-allocated copy of this object
177 virtual wxMBConv *Clone() const = 0;
178
179 // virtual dtor for any base class
180 virtual ~wxMBConv();
181 };
182
183 // ----------------------------------------------------------------------------
184 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
185 // conversion (hence it depends on the current locale)
186 // ----------------------------------------------------------------------------
187
188 class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv
189 {
190 public:
191 virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
192 virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
193
194 virtual wxMBConv *Clone() const { return new wxMBConvLibc; }
195
196 #if wxUSE_UNICODE_UTF8
197 virtual bool IsUTF8() const { return wxLocaleIsUtf8; }
198 #endif
199 };
200
201 #ifdef __UNIX__
202
203 // ----------------------------------------------------------------------------
204 // wxConvBrokenFileNames is made for Unix in Unicode mode when
205 // files are accidentally written in an encoding which is not
206 // the system encoding. Typically, the system encoding will be
207 // UTF8 but there might be files stored in ISO8859-1 on disk.
208 // ----------------------------------------------------------------------------
209
210 class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv
211 {
212 public:
213 wxConvBrokenFileNames(const wxString& charset);
214 wxConvBrokenFileNames(const wxConvBrokenFileNames& conv)
215 : wxMBConv(),
216 m_conv(conv.m_conv ? conv.m_conv->Clone() : NULL)
217 {
218 }
219 virtual ~wxConvBrokenFileNames() { delete m_conv; }
220
221 virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const
222 {
223 return m_conv->MB2WC(out, in, outLen);
224 }
225
226 virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const
227 {
228 return m_conv->WC2MB(out, in, outLen);
229 }
230
231 virtual size_t GetMBNulLen() const
232 {
233 // cast needed to call a private function
234 return m_conv->GetMBNulLen();
235 }
236
237 #if wxUSE_UNICODE_UTF8
238 virtual bool IsUTF8() const { return m_conv->IsUTF8(); }
239 #endif
240
241 virtual wxMBConv *Clone() const { return new wxConvBrokenFileNames(*this); }
242
243 private:
244 // the conversion object we forward to
245 wxMBConv *m_conv;
246
247 wxDECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames);
248 };
249
250 #endif // __UNIX__
251
252 // ----------------------------------------------------------------------------
253 // wxMBConvUTF7 (for conversion using UTF7 encoding)
254 // ----------------------------------------------------------------------------
255
256 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
257 {
258 public:
259 wxMBConvUTF7() { }
260
261 // compiler-generated copy ctor, assignment operator and dtor are ok
262 // (assuming it's ok to copy the shift state -- not really sure about it)
263
264 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
265 const char *src, size_t srcLen = wxNO_LEN) const;
266 virtual size_t FromWChar(char *dst, size_t dstLen,
267 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
268
269 virtual wxMBConv *Clone() const { return new wxMBConvUTF7; }
270
271 private:
272 // UTF-7 decoder/encoder may be in direct mode or in shifted mode after a
273 // '+' (and until the '-' or any other non-base64 character)
274 struct StateMode
275 {
276 enum Mode
277 {
278 Direct, // pass through state
279 Shifted // after a '+' (and before '-')
280 };
281 };
282
283 // the current decoder state: this is only used by ToWChar() if srcLen
284 // parameter is not wxNO_LEN, when working on the entire NUL-terminated
285 // strings we neither update nor use the state
286 class DecoderState : private StateMode
287 {
288 private:
289 // current state: this one is private as we want to enforce the use of
290 // ToDirect/ToShifted() methods below
291 Mode mode;
292
293 public:
294 // the initial state is direct
295 DecoderState() { mode = Direct; }
296
297 // switch to/from shifted mode
298 void ToDirect() { mode = Direct; }
299 void ToShifted() { mode = Shifted; accum = bit = 0; isLSB = false; }
300
301 bool IsDirect() const { return mode == Direct; }
302 bool IsShifted() const { return mode == Shifted; }
303
304
305 // these variables are only used in shifted mode
306
307 unsigned int accum; // accumulator of the bit we've already got
308 unsigned int bit; // the number of bits consumed mod 8
309 unsigned char msb; // the high byte of UTF-16 word
310 bool isLSB; // whether we're decoding LSB or MSB of UTF-16 word
311 };
312
313 DecoderState m_stateDecoder;
314
315
316 // encoder state is simpler as we always receive entire Unicode characters
317 // on input
318 class EncoderState : private StateMode
319 {
320 private:
321 Mode mode;
322
323 public:
324 EncoderState() { mode = Direct; }
325
326 void ToDirect() { mode = Direct; }
327 void ToShifted() { mode = Shifted; accum = bit = 0; }
328
329 bool IsDirect() const { return mode == Direct; }
330 bool IsShifted() const { return mode == Shifted; }
331
332 unsigned int accum;
333 unsigned int bit;
334 };
335
336 EncoderState m_stateEncoder;
337 };
338
339 // ----------------------------------------------------------------------------
340 // wxMBConvUTF8 (for conversion using UTF8 encoding)
341 // ----------------------------------------------------------------------------
342
343 // this is the real UTF-8 conversion class, it has to be called "strict UTF-8"
344 // for compatibility reasons: the wxMBConvUTF8 class below also supports lossy
345 // conversions if it is created with non default options
346 class WXDLLIMPEXP_BASE wxMBConvStrictUTF8 : public wxMBConv
347 {
348 public:
349 // compiler-generated default ctor and other methods are ok
350
351 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
352 const char *src, size_t srcLen = wxNO_LEN) const;
353 virtual size_t FromWChar(char *dst, size_t dstLen,
354 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
355
356 virtual wxMBConv *Clone() const { return new wxMBConvStrictUTF8(); }
357
358 #if wxUSE_UNICODE_UTF8
359 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
360 // take the shortcut in that case
361 virtual bool IsUTF8() const { return true; }
362 #endif
363 };
364
365 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConvStrictUTF8
366 {
367 public:
368 enum
369 {
370 MAP_INVALID_UTF8_NOT = 0,
371 MAP_INVALID_UTF8_TO_PUA = 1,
372 MAP_INVALID_UTF8_TO_OCTAL = 2
373 };
374
375 wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
376
377 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
378 const char *src, size_t srcLen = wxNO_LEN) const;
379 virtual size_t FromWChar(char *dst, size_t dstLen,
380 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
381
382 virtual wxMBConv *Clone() const { return new wxMBConvUTF8(m_options); }
383
384 #if wxUSE_UNICODE_UTF8
385 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
386 // take the shortcut in that case
387 virtual bool IsUTF8() const { return m_options == MAP_INVALID_UTF8_NOT; }
388 #endif
389
390 private:
391 int m_options;
392 };
393
394 // ----------------------------------------------------------------------------
395 // wxMBConvUTF16Base: for both LE and BE variants
396 // ----------------------------------------------------------------------------
397
398 class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv
399 {
400 public:
401 enum { BYTES_PER_CHAR = 2 };
402
403 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
404
405 protected:
406 // return the length of the buffer using srcLen if it's not wxNO_LEN and
407 // computing the length ourselves if it is; also checks that the length is
408 // even if specified as we need an entire number of UTF-16 characters and
409 // returns wxNO_LEN which indicates error if it is odd
410 static size_t GetLength(const char *src, size_t srcLen);
411 };
412
413 // ----------------------------------------------------------------------------
414 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
415 // ----------------------------------------------------------------------------
416
417 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base
418 {
419 public:
420 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
421 const char *src, size_t srcLen = wxNO_LEN) const;
422 virtual size_t FromWChar(char *dst, size_t dstLen,
423 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
424 virtual wxMBConv *Clone() const { return new wxMBConvUTF16LE; }
425 };
426
427 // ----------------------------------------------------------------------------
428 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
429 // ----------------------------------------------------------------------------
430
431 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base
432 {
433 public:
434 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
435 const char *src, size_t srcLen = wxNO_LEN) const;
436 virtual size_t FromWChar(char *dst, size_t dstLen,
437 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
438 virtual wxMBConv *Clone() const { return new wxMBConvUTF16BE; }
439 };
440
441 // ----------------------------------------------------------------------------
442 // wxMBConvUTF32Base: base class for both LE and BE variants
443 // ----------------------------------------------------------------------------
444
445 class WXDLLIMPEXP_BASE wxMBConvUTF32Base : public wxMBConv
446 {
447 public:
448 enum { BYTES_PER_CHAR = 4 };
449
450 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
451
452 protected:
453 // this is similar to wxMBConvUTF16Base method with the same name except
454 // that, of course, it verifies that length is divisible by 4 if given and
455 // not by 2
456 static size_t GetLength(const char *src, size_t srcLen);
457 };
458
459 // ----------------------------------------------------------------------------
460 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
461 // ----------------------------------------------------------------------------
462
463 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base
464 {
465 public:
466 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
467 const char *src, size_t srcLen = wxNO_LEN) const;
468 virtual size_t FromWChar(char *dst, size_t dstLen,
469 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
470 virtual wxMBConv *Clone() const { return new wxMBConvUTF32LE; }
471 };
472
473 // ----------------------------------------------------------------------------
474 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
475 // ----------------------------------------------------------------------------
476
477 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base
478 {
479 public:
480 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
481 const char *src, size_t srcLen = wxNO_LEN) const;
482 virtual size_t FromWChar(char *dst, size_t dstLen,
483 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
484 virtual wxMBConv *Clone() const { return new wxMBConvUTF32BE; }
485 };
486
487 // ----------------------------------------------------------------------------
488 // wxCSConv (for conversion based on loadable char sets)
489 // ----------------------------------------------------------------------------
490
491 #include "wx/fontenc.h"
492
493 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
494 {
495 public:
496 // we can be created either from charset name or from an encoding constant
497 // but we can't have both at once
498 wxCSConv(const wxString& charset);
499 wxCSConv(wxFontEncoding encoding);
500
501 wxCSConv(const wxCSConv& conv);
502 virtual ~wxCSConv();
503
504 wxCSConv& operator=(const wxCSConv& conv);
505
506 virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
507 const char *src, size_t srcLen = wxNO_LEN) const;
508 virtual size_t FromWChar(char *dst, size_t dstLen,
509 const wchar_t *src, size_t srcLen = wxNO_LEN) const;
510 virtual size_t GetMBNulLen() const;
511
512 #if wxUSE_UNICODE_UTF8
513 virtual bool IsUTF8() const;
514 #endif
515
516 virtual wxMBConv *Clone() const { return new wxCSConv(*this); }
517
518 void Clear();
519
520 // return true if the conversion could be initialized successfully
521 bool IsOk() const;
522
523 private:
524 // common part of all ctors
525 void Init();
526
527 // creates m_convReal if necessary
528 void CreateConvIfNeeded() const;
529
530 // do create m_convReal (unconditionally)
531 wxMBConv *DoCreate() const;
532
533 // set the name (may be only called when m_name == NULL), makes copy of
534 // the charset string
535 void SetName(const char *charset);
536
537
538 // m_name may be NULL in which case m_encoding should be used
539 //
540 // note that we can't use wxString here because of compilation
541 // dependencies: we're included from wx/string.h
542 char *m_name;
543
544 // may be wxFONTENCODING_SYSTEM in which case m_name is used
545 //
546 // if m_name is NULL, then we should use the default system encoding
547 wxFontEncoding m_encoding;
548
549 // use CreateConvIfNeeded() before accessing m_convReal!
550 wxMBConv *m_convReal;
551 bool m_deferred;
552 };
553
554
555 // ----------------------------------------------------------------------------
556 // declare predefined conversion objects
557 // ----------------------------------------------------------------------------
558
559 // Note: this macro is an implementation detail (see the comment in
560 // strconv.cpp). The wxGet_XXX() and wxGet_XXXPtr() functions shouldn't be
561 // used by user code and neither should XXXPtr, use the wxConvXXX macro
562 // instead.
563 #define WX_DECLARE_GLOBAL_CONV(klass, name) \
564 extern WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr; \
565 extern WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr(); \
566 inline klass& wxGet_##name() \
567 { \
568 if ( !name##Ptr ) \
569 name##Ptr = wxGet_##name##Ptr(); \
570 return *name##Ptr; \
571 }
572
573
574 // conversion to be used with all standard functions affected by locale, e.g.
575 // strtol(), strftime(), ...
576 WX_DECLARE_GLOBAL_CONV(wxMBConv, wxConvLibc)
577 #define wxConvLibc wxGet_wxConvLibc()
578
579 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
580 WX_DECLARE_GLOBAL_CONV(wxCSConv, wxConvISO8859_1)
581 #define wxConvISO8859_1 wxGet_wxConvISO8859_1()
582
583 WX_DECLARE_GLOBAL_CONV(wxMBConvStrictUTF8, wxConvUTF8)
584 #define wxConvUTF8 wxGet_wxConvUTF8()
585
586 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7, wxConvUTF7)
587 #define wxConvUTF7 wxGet_wxConvUTF7()
588
589 // conversion used for the file names on the systems where they're not Unicode
590 // (basically anything except Windows)
591 //
592 // this is used by all file functions, can be changed by the application
593 //
594 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
595 // under Windows normally)
596 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
597
598 // backwards compatible define
599 #define wxConvFile (*wxConvFileName)
600
601 // the current conversion object, may be set to any conversion, is used by
602 // default in a couple of places inside wx (initially same as wxConvLibc)
603 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
604
605 // the conversion corresponding to the current locale
606 WX_DECLARE_GLOBAL_CONV(wxCSConv, wxConvLocal)
607 #define wxConvLocal wxGet_wxConvLocal()
608
609 // the conversion corresponding to the encoding of the standard UI elements
610 //
611 // by default this is the same as wxConvLocal but may be changed if the program
612 // needs to use a fixed encoding
613 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI;
614
615 #undef WX_DECLARE_GLOBAL_CONV
616
617 // ----------------------------------------------------------------------------
618 // endianness-dependent conversions
619 // ----------------------------------------------------------------------------
620
621 #ifdef WORDS_BIGENDIAN
622 typedef wxMBConvUTF16BE wxMBConvUTF16;
623 typedef wxMBConvUTF32BE wxMBConvUTF32;
624 #else
625 typedef wxMBConvUTF16LE wxMBConvUTF16;
626 typedef wxMBConvUTF32LE wxMBConvUTF32;
627 #endif
628
629 // ----------------------------------------------------------------------------
630 // filename conversion macros
631 // ----------------------------------------------------------------------------
632
633 // filenames are multibyte on Unix and widechar on Windows
634 #if wxMBFILES && wxUSE_UNICODE
635 #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
636 #define wxFNSTRINGCAST wxMBSTRINGCAST
637 #else
638 #if defined( __WXOSX_OR_COCOA__ ) && wxMBFILES
639 #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
640 #else
641 #define wxFNCONV(name) name
642 #endif
643 #define wxFNSTRINGCAST WXSTRINGCAST
644 #endif
645
646 #else // !wxUSE_WCHAR_T
647
648 // ----------------------------------------------------------------------------
649 // stand-ins in absence of wchar_t
650 // ----------------------------------------------------------------------------
651
652 class WXDLLIMPEXP_BASE wxMBConv
653 {
654 public:
655 const char* cMB2WX(const char *psz) const { return psz; }
656 const char* cWX2MB(const char *psz) const { return psz; }
657 };
658
659 #define wxConvFile wxConvLocal
660
661 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
662 wxConvLocal,
663 wxConvISO8859_1,
664 wxConvUTF8;
665 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
666
667 #define wxFNCONV(name) name
668 #define wxFNSTRINGCAST WXSTRINGCAST
669
670 #endif
671 // wxUSE_WCHAR_T
672
673 // ----------------------------------------------------------------------------
674 // macros for the most common conversions
675 // ----------------------------------------------------------------------------
676
677 #if wxUSE_UNICODE
678 #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s)
679 #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s)
680
681 // these functions should be used when the conversions really, really have
682 // to succeed (usually because we pass their results to a standard C
683 // function which would crash if we passed NULL to it), so these functions
684 // always return a valid pointer if their argument is non-NULL
685
686 // this function safety is achieved by trying wxConvLibc first, wxConvUTF8
687 // next if it fails and, finally, wxConvISO8859_1 which always succeeds
688 extern WXDLLIMPEXP_BASE wxWCharBuffer wxSafeConvertMB2WX(const char *s);
689
690 // this function uses wxConvLibc and wxConvUTF8(MAP_INVALID_UTF8_TO_OCTAL)
691 // if it fails
692 extern WXDLLIMPEXP_BASE wxCharBuffer wxSafeConvertWX2MB(const wchar_t *ws);
693 #else // ANSI
694 // no conversions to do
695 #define wxConvertWX2MB(s) (s)
696 #define wxConvertMB2WX(s) (s)
697 #define wxSafeConvertMB2WX(s) (s)
698 #define wxSafeConvertWX2MB(s) (s)
699 #endif // Unicode/ANSI
700
701 #endif // _WX_STRCONV_H_
702