1 ///////////////////////////////////////////////////////////////////////////////
3 // Purpose: conversion routines for char sets any Unicode
4 // Author: Ove Kaaven, Robert Roebling, Vadim Zeitlin
8 // Copyright: (c) 1998 Ove Kaaven, Robert Roebling
9 // (c) 1998-2006 Vadim Zeitlin
10 // Licence: wxWindows licence
11 ///////////////////////////////////////////////////////////////////////////////
13 #ifndef _WX_STRCONV_H_
14 #define _WX_STRCONV_H_
17 #include "wx/chartype.h"
18 #include "wx/buffer.h"
20 #ifdef __DIGITALMARS__
24 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
30 #endif // ! __WXPALMOS5__
34 class WXDLLIMPEXP_FWD_BASE wxString
;
36 // the error value returned by wxMBConv methods
37 #define wxCONV_FAILED ((size_t)-1)
39 // the default value for some length parameters meaning that the string is
41 #define wxNO_LEN ((size_t)-1)
43 // ----------------------------------------------------------------------------
44 // wxMBConv (abstract base class for conversions)
45 // ----------------------------------------------------------------------------
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
52 // You also have to implement Clone() to allow copying the conversions
55 // And you might need to override GetMBNulLen() as well.
56 class WXDLLIMPEXP_BASE wxMBConv
59 // The functions doing actual conversion from/to narrow to/from wide
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.
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
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);
82 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
83 const char *src
, size_t srcLen
= wxNO_LEN
) const;
85 virtual size_t FromWChar(char *dst
, size_t dstLen
,
86 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
89 // Convenience functions for translating NUL-terminated strings: returns
90 // the buffer containing the converted string or NULL pointer if the
92 const wxWCharBuffer
cMB2WC(const char *in
) const;
93 const wxCharBuffer
cWC2MB(const wchar_t *in
) const;
95 // Convenience functions for converting strings which may contain embedded
96 // NULs and don't have to be NUL-terminated.
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.
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)
108 cMB2WC(const char *in
, size_t inLen
, size_t *outLen
) const;
110 cWC2MB(const wchar_t *in
, size_t inLen
, size_t *outLen
) const;
112 // convenience functions for converting MB or WC to/from wxWin default
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
; }
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
125 // this function is used in the implementation of cMB2WC() to distinguish
126 // between the following cases:
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
135 // anything else is not supported currently and -1 should be returned
136 virtual size_t GetMBNulLen() const { return 1; }
138 // return the maximal value currently returned by GetMBNulLen() for any
140 static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; }
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; }
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.
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.
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;
169 // make a heap-allocated copy of this object
170 virtual wxMBConv
*Clone() const = 0;
172 // virtual dtor for any base class
176 // ----------------------------------------------------------------------------
177 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
178 // conversion (hence it depends on the current locale)
179 // ----------------------------------------------------------------------------
181 class WXDLLIMPEXP_BASE wxMBConvLibc
: public wxMBConv
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;
187 virtual wxMBConv
*Clone() const { return new wxMBConvLibc
; }
189 #if wxUSE_UNICODE_UTF8
190 virtual bool IsUTF8() const { return wxLocaleIsUtf8
; }
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 // ----------------------------------------------------------------------------
203 class WXDLLIMPEXP_BASE wxConvBrokenFileNames
: public wxMBConv
206 wxConvBrokenFileNames(const wxString
& charset
);
207 wxConvBrokenFileNames(const wxConvBrokenFileNames
& conv
)
209 m_conv(conv
.m_conv
? conv
.m_conv
->Clone() : NULL
)
212 virtual ~wxConvBrokenFileNames() { delete m_conv
; }
214 virtual size_t MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const
216 return m_conv
->MB2WC(out
, in
, outLen
);
219 virtual size_t WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const
221 return m_conv
->WC2MB(out
, in
, outLen
);
224 virtual size_t GetMBNulLen() const
226 // cast needed to call a private function
227 return m_conv
->GetMBNulLen();
230 #if wxUSE_UNICODE_UTF8
231 virtual bool IsUTF8() const { return m_conv
->IsUTF8(); }
234 virtual wxMBConv
*Clone() const { return new wxConvBrokenFileNames(*this); }
237 // the conversion object we forward to
240 DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames
)
245 // ----------------------------------------------------------------------------
246 // wxMBConvUTF7 (for conversion using UTF7 encoding)
247 // ----------------------------------------------------------------------------
249 class WXDLLIMPEXP_BASE wxMBConvUTF7
: public wxMBConv
252 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
253 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
255 virtual wxMBConv
*Clone() const { return new wxMBConvUTF7
; }
258 // ----------------------------------------------------------------------------
259 // wxMBConvUTF8 (for conversion using UTF8 encoding)
260 // ----------------------------------------------------------------------------
262 // this is the real UTF-8 conversion class, it has to be called "strict UTF-8"
263 // for compatibility reasons: the wxMBConvUTF8 class below also supports lossy
264 // conversions if it is created with non default options
265 class WXDLLIMPEXP_BASE wxMBConvStrictUTF8
: public wxMBConv
268 // compiler-generated default ctor and other methods are ok
270 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
271 const char *src
, size_t srcLen
= wxNO_LEN
) const;
272 virtual size_t FromWChar(char *dst
, size_t dstLen
,
273 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
275 virtual wxMBConv
*Clone() const { return new wxMBConvStrictUTF8(); }
277 #if wxUSE_UNICODE_UTF8
278 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
279 // take the shortcut in that case
280 virtual bool IsUTF8() const { return true; }
284 class WXDLLIMPEXP_BASE wxMBConvUTF8
: public wxMBConvStrictUTF8
289 MAP_INVALID_UTF8_NOT
= 0,
290 MAP_INVALID_UTF8_TO_PUA
= 1,
291 MAP_INVALID_UTF8_TO_OCTAL
= 2
294 wxMBConvUTF8(int options
= MAP_INVALID_UTF8_NOT
) : m_options(options
) { }
296 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
297 const char *src
, size_t srcLen
= wxNO_LEN
) const;
298 virtual size_t FromWChar(char *dst
, size_t dstLen
,
299 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
301 virtual wxMBConv
*Clone() const { return new wxMBConvUTF8(m_options
); }
303 #if wxUSE_UNICODE_UTF8
304 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
305 // take the shortcut in that case
306 virtual bool IsUTF8() const { return m_options
== MAP_INVALID_UTF8_NOT
; }
313 // ----------------------------------------------------------------------------
314 // wxMBConvUTF16Base: for both LE and BE variants
315 // ----------------------------------------------------------------------------
317 class WXDLLIMPEXP_BASE wxMBConvUTF16Base
: public wxMBConv
320 enum { BYTES_PER_CHAR
= 2 };
322 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
325 // return the length of the buffer using srcLen if it's not wxNO_LEN and
326 // computing the length ourselves if it is; also checks that the length is
327 // even if specified as we need an entire number of UTF-16 characters and
328 // returns wxNO_LEN which indicates error if it is odd
329 static size_t GetLength(const char *src
, size_t srcLen
);
332 // ----------------------------------------------------------------------------
333 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
334 // ----------------------------------------------------------------------------
336 class WXDLLIMPEXP_BASE wxMBConvUTF16LE
: public wxMBConvUTF16Base
339 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
340 const char *src
, size_t srcLen
= wxNO_LEN
) const;
341 virtual size_t FromWChar(char *dst
, size_t dstLen
,
342 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
343 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16LE
; }
346 // ----------------------------------------------------------------------------
347 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
348 // ----------------------------------------------------------------------------
350 class WXDLLIMPEXP_BASE wxMBConvUTF16BE
: public wxMBConvUTF16Base
353 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
354 const char *src
, size_t srcLen
= wxNO_LEN
) const;
355 virtual size_t FromWChar(char *dst
, size_t dstLen
,
356 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
357 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16BE
; }
360 // ----------------------------------------------------------------------------
361 // wxMBConvUTF32Base: base class for both LE and BE variants
362 // ----------------------------------------------------------------------------
364 class WXDLLIMPEXP_BASE wxMBConvUTF32Base
: public wxMBConv
367 enum { BYTES_PER_CHAR
= 4 };
369 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
372 // this is similar to wxMBConvUTF16Base method with the same name except
373 // that, of course, it verifies that length is divisible by 4 if given and
375 static size_t GetLength(const char *src
, size_t srcLen
);
378 // ----------------------------------------------------------------------------
379 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
380 // ----------------------------------------------------------------------------
382 class WXDLLIMPEXP_BASE wxMBConvUTF32LE
: public wxMBConvUTF32Base
385 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
386 const char *src
, size_t srcLen
= wxNO_LEN
) const;
387 virtual size_t FromWChar(char *dst
, size_t dstLen
,
388 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
389 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32LE
; }
392 // ----------------------------------------------------------------------------
393 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
394 // ----------------------------------------------------------------------------
396 class WXDLLIMPEXP_BASE wxMBConvUTF32BE
: public wxMBConvUTF32Base
399 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
400 const char *src
, size_t srcLen
= wxNO_LEN
) const;
401 virtual size_t FromWChar(char *dst
, size_t dstLen
,
402 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
403 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32BE
; }
406 // ----------------------------------------------------------------------------
407 // wxCSConv (for conversion based on loadable char sets)
408 // ----------------------------------------------------------------------------
410 #include "wx/fontenc.h"
412 class WXDLLIMPEXP_BASE wxCSConv
: public wxMBConv
415 // we can be created either from charset name or from an encoding constant
416 // but we can't have both at once
417 wxCSConv(const wxString
& charset
);
418 wxCSConv(wxFontEncoding encoding
);
420 wxCSConv(const wxCSConv
& conv
);
423 wxCSConv
& operator=(const wxCSConv
& conv
);
425 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
426 const char *src
, size_t srcLen
= wxNO_LEN
) const;
427 virtual size_t FromWChar(char *dst
, size_t dstLen
,
428 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
429 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
430 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
431 virtual size_t GetMBNulLen() const;
433 #if wxUSE_UNICODE_UTF8
434 virtual bool IsUTF8() const;
437 virtual wxMBConv
*Clone() const { return new wxCSConv(*this); }
441 // return true if the conversion could be initilized successfully
445 // common part of all ctors
448 // creates m_convReal if necessary
449 void CreateConvIfNeeded() const;
451 // do create m_convReal (unconditionally)
452 wxMBConv
*DoCreate() const;
454 // set the name (may be only called when m_name == NULL), makes copy of
456 void SetName(const char *charset
);
459 // note that we can't use wxString here because of compilation
460 // dependencies: we're included from wx/string.h
462 wxFontEncoding m_encoding
;
464 // use CreateConvIfNeeded() before accessing m_convReal!
465 wxMBConv
*m_convReal
;
470 // ----------------------------------------------------------------------------
471 // declare predefined conversion objects
472 // ----------------------------------------------------------------------------
474 // Note: this macro is an implementation detail (see the comment in
475 // strconv.cpp). The wxGet_XXX() and wxGet_XXXPtr() functions shouldn't be
476 // used by user code and neither should XXXPtr, use the wxConvXXX macro
478 #define WX_DECLARE_GLOBAL_CONV(klass, name) \
479 extern WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr; \
480 extern WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr(); \
481 inline klass& wxGet_##name() \
484 name##Ptr = wxGet_##name##Ptr(); \
489 // conversion to be used with all standard functions affected by locale, e.g.
490 // strtol(), strftime(), ...
491 WX_DECLARE_GLOBAL_CONV(wxMBConv
, wxConvLibc
)
492 #define wxConvLibc wxGet_wxConvLibc()
494 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
495 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvISO8859_1
)
496 #define wxConvISO8859_1 wxGet_wxConvISO8859_1()
498 WX_DECLARE_GLOBAL_CONV(wxMBConvStrictUTF8
, wxConvUTF8
)
499 #define wxConvUTF8 wxGet_wxConvUTF8()
501 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7
, wxConvUTF7
)
502 #define wxConvUTF7 wxGet_wxConvUTF7()
504 // conversion used for the file names on the systems where they're not Unicode
505 // (basically anything except Windows)
507 // this is used by all file functions, can be changed by the application
509 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
510 // under Windows normally)
511 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvFileName
;
513 // backwards compatible define
514 #define wxConvFile (*wxConvFileName)
516 // the current conversion object, may be set to any conversion, is used by
517 // default in a couple of places inside wx (initially same as wxConvLibc)
518 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
520 // the conversion corresponding to the current locale
521 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvLocal
)
522 #define wxConvLocal wxGet_wxConvLocal()
524 // the conversion corresponding to the encoding of the standard UI elements
526 // by default this is the same as wxConvLocal but may be changed if the program
527 // needs to use a fixed encoding
528 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvUI
;
530 #undef WX_DECLARE_GLOBAL_CONV
532 // ----------------------------------------------------------------------------
533 // endianness-dependent conversions
534 // ----------------------------------------------------------------------------
536 #ifdef WORDS_BIGENDIAN
537 typedef wxMBConvUTF16BE wxMBConvUTF16
;
538 typedef wxMBConvUTF32BE wxMBConvUTF32
;
540 typedef wxMBConvUTF16LE wxMBConvUTF16
;
541 typedef wxMBConvUTF32LE wxMBConvUTF32
;
544 // ----------------------------------------------------------------------------
545 // filename conversion macros
546 // ----------------------------------------------------------------------------
548 // filenames are multibyte on Unix and widechar on Windows
549 #if wxMBFILES && wxUSE_UNICODE
550 #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
551 #define wxFNSTRINGCAST wxMBSTRINGCAST
553 #if defined( __WXOSX__ ) && wxMBFILES
554 #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
556 #define wxFNCONV(name) name
558 #define wxFNSTRINGCAST WXSTRINGCAST
561 #else // !wxUSE_WCHAR_T
563 // ----------------------------------------------------------------------------
564 // stand-ins in absence of wchar_t
565 // ----------------------------------------------------------------------------
567 class WXDLLIMPEXP_BASE wxMBConv
570 const char* cMB2WX(const char *psz
) const { return psz
; }
571 const char* cWX2MB(const char *psz
) const { return psz
; }
574 #define wxConvFile wxConvLocal
576 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
) wxConvLibc
,
580 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
582 #define wxFNCONV(name) name
583 #define wxFNSTRINGCAST WXSTRINGCAST
588 // ----------------------------------------------------------------------------
589 // macros for the most common conversions
590 // ----------------------------------------------------------------------------
593 #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s)
594 #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s)
596 // these functions should be used when the conversions really, really have
597 // to succeed (usually because we pass their results to a standard C
598 // function which would crash if we passed NULL to it), so these functions
599 // always return a valid pointer if their argument is non-NULL
601 // this function safety is achieved by trying wxConvLibc first, wxConvUTF8
602 // next if it fails and, finally, wxConvISO8859_1 which always succeeds
603 extern WXDLLIMPEXP_BASE wxWCharBuffer
wxSafeConvertMB2WX(const char *s
);
605 // this function uses wxConvLibc and wxConvUTF8(MAP_INVALID_UTF8_TO_OCTAL)
607 extern WXDLLIMPEXP_BASE wxCharBuffer
wxSafeConvertWX2MB(const wchar_t *ws
);
609 // no conversions to do
610 #define wxConvertWX2MB(s) (s)
611 #define wxConvertMB2WX(s) (s)
612 #define wxSafeConvertMB2WX(s) (s)
613 #define wxSafeConvertWX2MB(s) (s)
614 #endif // Unicode/ANSI
616 #endif // _WX_STRCONV_H_