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