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
32 class WXDLLIMPEXP_FWD_BASE wxString
;
34 // the error value returned by wxMBConv methods
35 #define wxCONV_FAILED ((size_t)-1)
37 // the default value for some length parameters meaning that the string is
39 #define wxNO_LEN ((size_t)-1)
41 // ----------------------------------------------------------------------------
42 // wxMBConv (abstract base class for conversions)
43 // ----------------------------------------------------------------------------
45 // When deriving a new class from wxMBConv you must reimplement ToWChar() and
46 // FromWChar() methods which are not pure virtual only for historical reasons,
47 // don't let the fact that the existing classes implement MB2WC/WC2MB() instead
50 // You also have to implement Clone() to allow copying the conversions
53 // And you might need to override GetMBNulLen() as well.
54 class WXDLLIMPEXP_BASE wxMBConv
57 // The functions doing actual conversion from/to narrow to/from wide
60 // On success, the return value is the length (i.e. the number of
61 // characters, not bytes) of the converted string including any trailing
62 // L'\0' or (possibly multiple) '\0'(s). If the conversion fails or if
63 // there is not enough space for everything, including the trailing NUL
64 // character(s), in the output buffer, wxCONV_FAILED is returned.
66 // In the special case when dstLen is 0 (outputBuf may be NULL then) the
67 // return value is the length of the needed buffer but nothing happens
68 // otherwise. If srcLen is wxNO_LEN, the entire string, up to and
69 // including the trailing NUL(s), is converted, otherwise exactly srcLen
74 // size_t dstLen = conv.ToWChar(NULL, 0, src);
75 // if ( dstLen != wxCONV_FAILED )
76 // ... handle error ...
77 // wchar_t *wbuf = new wchar_t[dstLen];
78 // conv.ToWChar(wbuf, dstLen, src);
80 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
81 const char *src
, size_t srcLen
= wxNO_LEN
) const;
83 virtual size_t FromWChar(char *dst
, size_t dstLen
,
84 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
87 // Convenience functions for translating NUL-terminated strings: returns
88 // the buffer containing the converted string or NULL pointer if the
90 const wxWCharBuffer
cMB2WC(const char *in
) const;
91 const wxCharBuffer
cWC2MB(const wchar_t *in
) const;
93 // Convenience functions for converting strings which may contain embedded
94 // NULs and don't have to be NUL-terminated.
96 // inLen is the length of the buffer including trailing NUL if any: if the
97 // last 4 bytes of the buffer are all NULs, these functions are more
98 // efficient as they avoid copying the string, but otherwise a copy is made
99 // internally which could be quite bad for (very) long strings.
101 // outLen receives, if not NULL, the length of the converted string or 0 if
102 // the conversion failed (returning 0 and not -1 in this case makes it
103 // difficult to distinguish between failed conversion and empty input but
104 // this is done for backwards compatibility)
106 cMB2WC(const char *in
, size_t inLen
, size_t *outLen
) const;
108 cWC2MB(const wchar_t *in
, size_t inLen
, size_t *outLen
) const;
110 // convenience functions for converting MB or WC to/from wxWin default
112 const wxWCharBuffer
cMB2WX(const char *psz
) const { return cMB2WC(psz
); }
113 const wxCharBuffer
cWX2MB(const wchar_t *psz
) const { return cWC2MB(psz
); }
114 const wchar_t* cWC2WX(const wchar_t *psz
) const { return psz
; }
115 const wchar_t* cWX2WC(const wchar_t *psz
) const { return psz
; }
117 const char* cMB2WX(const char *psz
) const { return psz
; }
118 const char* cWX2MB(const char *psz
) const { return psz
; }
119 const wxCharBuffer
cWC2WX(const wchar_t *psz
) const { return cWC2MB(psz
); }
120 const wxWCharBuffer
cWX2WC(const char *psz
) const { return cMB2WC(psz
); }
121 #endif // Unicode/ANSI
123 // this function is used in the implementation of cMB2WC() to distinguish
124 // between the following cases:
126 // a) var width encoding with strings terminated by a single NUL
127 // (usual multibyte encodings): return 1 in this case
128 // b) fixed width encoding with 2 bytes/char and so terminated by
129 // 2 NULs (UTF-16/UCS-2 and variants): return 2 in this case
130 // c) fixed width encoding with 4 bytes/char and so terminated by
131 // 4 NULs (UTF-32/UCS-4 and variants): return 4 in this case
133 // anything else is not supported currently and -1 should be returned
134 virtual size_t GetMBNulLen() const { return 1; }
136 // return the maximal value currently returned by GetMBNulLen() for any
138 static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; }
140 #if wxUSE_UNICODE_UTF8
141 // return true if the converter's charset is UTF-8, i.e. char* strings
142 // decoded using this object can be directly copied to wxString's internal
143 // storage without converting to WC and than back to UTF-8 MB string
144 virtual bool IsUTF8() const { return false; }
147 // The old conversion functions. The existing classes currently mostly
148 // implement these ones but we're in transition to using To/FromWChar()
149 // instead and any new classes should implement just the new functions.
150 // For now, however, we provide default implementation of To/FromWChar() in
151 // this base class in terms of MB2WC/WC2MB() to avoid having to rewrite all
152 // the conversions at once.
154 // On success, the return value is the length (i.e. the number of
155 // characters, not bytes) not counting the trailing NUL(s) of the converted
156 // string. On failure, (size_t)-1 is returned. In the special case when
157 // outputBuf is NULL the return value is the same one but nothing is
158 // written to the buffer.
160 // Note that outLen is the length of the output buffer, not the length of
161 // the input (which is always supposed to be terminated by one or more
162 // NULs, as appropriate for the encoding)!
163 virtual size_t MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const;
164 virtual size_t WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const;
167 // make a heap-allocated copy of this object
168 virtual wxMBConv
*Clone() const = 0;
170 // virtual dtor for any base class
174 // ----------------------------------------------------------------------------
175 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
176 // conversion (hence it depends on the current locale)
177 // ----------------------------------------------------------------------------
179 class WXDLLIMPEXP_BASE wxMBConvLibc
: public wxMBConv
182 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
183 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
185 virtual wxMBConv
*Clone() const { return new wxMBConvLibc
; }
187 #if wxUSE_UNICODE_UTF8
188 virtual bool IsUTF8() const { return wxLocaleIsUtf8
; }
194 // ----------------------------------------------------------------------------
195 // wxConvBrokenFileNames is made for Unix in Unicode mode when
196 // files are accidentally written in an encoding which is not
197 // the system encoding. Typically, the system encoding will be
198 // UTF8 but there might be files stored in ISO8859-1 on disk.
199 // ----------------------------------------------------------------------------
201 class WXDLLIMPEXP_BASE wxConvBrokenFileNames
: public wxMBConv
204 wxConvBrokenFileNames(const wxString
& charset
);
205 wxConvBrokenFileNames(const wxConvBrokenFileNames
& conv
)
207 m_conv(conv
.m_conv
? conv
.m_conv
->Clone() : NULL
)
210 virtual ~wxConvBrokenFileNames() { delete m_conv
; }
212 virtual size_t MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const
214 return m_conv
->MB2WC(out
, in
, outLen
);
217 virtual size_t WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const
219 return m_conv
->WC2MB(out
, in
, outLen
);
222 virtual size_t GetMBNulLen() const
224 // cast needed to call a private function
225 return m_conv
->GetMBNulLen();
228 #if wxUSE_UNICODE_UTF8
229 virtual bool IsUTF8() const { return m_conv
->IsUTF8(); }
232 virtual wxMBConv
*Clone() const { return new wxConvBrokenFileNames(*this); }
235 // the conversion object we forward to
238 DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames
)
243 // ----------------------------------------------------------------------------
244 // wxMBConvUTF7 (for conversion using UTF7 encoding)
245 // ----------------------------------------------------------------------------
247 class WXDLLIMPEXP_BASE wxMBConvUTF7
: public wxMBConv
250 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
251 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
253 virtual wxMBConv
*Clone() const { return new wxMBConvUTF7
; }
256 // ----------------------------------------------------------------------------
257 // wxMBConvUTF8 (for conversion using UTF8 encoding)
258 // ----------------------------------------------------------------------------
260 // this is the real UTF-8 conversion class, it has to be called "strict UTF-8"
261 // for compatibility reasons: the wxMBConvUTF8 class below also supports lossy
262 // conversions if it is created with non default options
263 class WXDLLIMPEXP_BASE wxMBConvStrictUTF8
: public wxMBConv
266 // compiler-generated default ctor and other methods are ok
268 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
269 const char *src
, size_t srcLen
= wxNO_LEN
) const;
270 virtual size_t FromWChar(char *dst
, size_t dstLen
,
271 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
273 virtual wxMBConv
*Clone() const { return new wxMBConvStrictUTF8(); }
275 #if wxUSE_UNICODE_UTF8
276 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
277 // take the shortcut in that case
278 virtual bool IsUTF8() const { return true; }
282 class WXDLLIMPEXP_BASE wxMBConvUTF8
: public wxMBConvStrictUTF8
287 MAP_INVALID_UTF8_NOT
= 0,
288 MAP_INVALID_UTF8_TO_PUA
= 1,
289 MAP_INVALID_UTF8_TO_OCTAL
= 2
292 wxMBConvUTF8(int options
= MAP_INVALID_UTF8_NOT
) : m_options(options
) { }
293 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
294 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
296 virtual wxMBConv
*Clone() const { return new wxMBConvUTF8(m_options
); }
298 #if wxUSE_UNICODE_UTF8
299 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
300 // take the shortcut in that case
301 virtual bool IsUTF8() const { return m_options
== MAP_INVALID_UTF8_NOT
; }
308 // ----------------------------------------------------------------------------
309 // wxMBConvUTF16Base: for both LE and BE variants
310 // ----------------------------------------------------------------------------
312 class WXDLLIMPEXP_BASE wxMBConvUTF16Base
: public wxMBConv
315 enum { BYTES_PER_CHAR
= 2 };
317 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
320 // return the length of the buffer using srcLen if it's not wxNO_LEN and
321 // computing the length ourselves if it is; also checks that the length is
322 // even if specified as we need an entire number of UTF-16 characters and
323 // returns wxNO_LEN which indicates error if it is odd
324 static size_t GetLength(const char *src
, size_t srcLen
);
327 // ----------------------------------------------------------------------------
328 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
329 // ----------------------------------------------------------------------------
331 class WXDLLIMPEXP_BASE wxMBConvUTF16LE
: public wxMBConvUTF16Base
334 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
335 const char *src
, size_t srcLen
= wxNO_LEN
) const;
336 virtual size_t FromWChar(char *dst
, size_t dstLen
,
337 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
338 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16LE
; }
341 // ----------------------------------------------------------------------------
342 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
343 // ----------------------------------------------------------------------------
345 class WXDLLIMPEXP_BASE wxMBConvUTF16BE
: public wxMBConvUTF16Base
348 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
349 const char *src
, size_t srcLen
= wxNO_LEN
) const;
350 virtual size_t FromWChar(char *dst
, size_t dstLen
,
351 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
352 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16BE
; }
355 // ----------------------------------------------------------------------------
356 // wxMBConvUTF32Base: base class for both LE and BE variants
357 // ----------------------------------------------------------------------------
359 class WXDLLIMPEXP_BASE wxMBConvUTF32Base
: public wxMBConv
362 enum { BYTES_PER_CHAR
= 4 };
364 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
367 // this is similar to wxMBConvUTF16Base method with the same name except
368 // that, of course, it verifies that length is divisible by 4 if given and
370 static size_t GetLength(const char *src
, size_t srcLen
);
373 // ----------------------------------------------------------------------------
374 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
375 // ----------------------------------------------------------------------------
377 class WXDLLIMPEXP_BASE wxMBConvUTF32LE
: public wxMBConvUTF32Base
380 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
381 const char *src
, size_t srcLen
= wxNO_LEN
) const;
382 virtual size_t FromWChar(char *dst
, size_t dstLen
,
383 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
384 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32LE
; }
387 // ----------------------------------------------------------------------------
388 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
389 // ----------------------------------------------------------------------------
391 class WXDLLIMPEXP_BASE wxMBConvUTF32BE
: public wxMBConvUTF32Base
394 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
395 const char *src
, size_t srcLen
= wxNO_LEN
) const;
396 virtual size_t FromWChar(char *dst
, size_t dstLen
,
397 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
398 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32BE
; }
401 // ----------------------------------------------------------------------------
402 // wxCSConv (for conversion based on loadable char sets)
403 // ----------------------------------------------------------------------------
405 #include "wx/fontenc.h"
407 class WXDLLIMPEXP_BASE wxCSConv
: public wxMBConv
410 // we can be created either from charset name or from an encoding constant
411 // but we can't have both at once
412 wxCSConv(const wxString
& charset
);
413 wxCSConv(wxFontEncoding encoding
);
415 wxCSConv(const wxCSConv
& conv
);
418 wxCSConv
& operator=(const wxCSConv
& conv
);
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 size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
425 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
426 virtual size_t GetMBNulLen() const;
428 #if wxUSE_UNICODE_UTF8
429 virtual bool IsUTF8() const;
432 virtual wxMBConv
*Clone() const { return new wxCSConv(*this); }
436 // return true if the conversion could be initilized successfully
440 // common part of all ctors
443 // creates m_convReal if necessary
444 void CreateConvIfNeeded() const;
446 // do create m_convReal (unconditionally)
447 wxMBConv
*DoCreate() const;
449 // set the name (may be only called when m_name == NULL), makes copy of
451 void SetName(const char *charset
);
454 // note that we can't use wxString here because of compilation
455 // dependencies: we're included from wx/string.h
457 wxFontEncoding m_encoding
;
459 // use CreateConvIfNeeded() before accessing m_convReal!
460 wxMBConv
*m_convReal
;
465 // ----------------------------------------------------------------------------
466 // declare predefined conversion objects
467 // ----------------------------------------------------------------------------
469 // Note: this macro is an implementation detail (see the comment in
470 // strconv.cpp). The wxGet_XXX() and wxGet_XXXPtr() functions shouldn't be
471 // used by user code and neither should XXXPtr, use the wxConvXXX macro
473 #define WX_DECLARE_GLOBAL_CONV(klass, name) \
474 extern WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr; \
475 extern WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr(); \
476 inline klass& wxGet_##name() \
479 name##Ptr = wxGet_##name##Ptr(); \
484 // conversion to be used with all standard functions affected by locale, e.g.
485 // strtol(), strftime(), ...
486 WX_DECLARE_GLOBAL_CONV(wxMBConv
, wxConvLibc
)
487 #define wxConvLibc wxGet_wxConvLibc()
489 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
490 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvISO8859_1
)
491 #define wxConvISO8859_1 wxGet_wxConvISO8859_1()
493 WX_DECLARE_GLOBAL_CONV(wxMBConvStrictUTF8
, wxConvUTF8
)
494 #define wxConvUTF8 wxGet_wxConvUTF8()
496 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7
, wxConvUTF7
)
497 #define wxConvUTF7 wxGet_wxConvUTF7()
499 // conversion used for the file names on the systems where they're not Unicode
500 // (basically anything except Windows)
502 // this is used by all file functions, can be changed by the application
504 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
505 // under Windows normally)
506 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvFileName
;
508 // backwards compatible define
509 #define wxConvFile (*wxConvFileName)
511 // the current conversion object, may be set to any conversion, is used by
512 // default in a couple of places inside wx (initially same as wxConvLibc)
513 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
515 // the conversion corresponding to the current locale
516 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvLocal
)
517 #define wxConvLocal wxGet_wxConvLocal()
519 // the conversion corresponding to the encoding of the standard UI elements
521 // by default this is the same as wxConvLocal but may be changed if the program
522 // needs to use a fixed encoding
523 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvUI
;
525 #undef WX_DECLARE_GLOBAL_CONV
527 // ----------------------------------------------------------------------------
528 // endianness-dependent conversions
529 // ----------------------------------------------------------------------------
531 #ifdef WORDS_BIGENDIAN
532 typedef wxMBConvUTF16BE wxMBConvUTF16
;
533 typedef wxMBConvUTF32BE wxMBConvUTF32
;
535 typedef wxMBConvUTF16LE wxMBConvUTF16
;
536 typedef wxMBConvUTF32LE wxMBConvUTF32
;
539 // ----------------------------------------------------------------------------
540 // filename conversion macros
541 // ----------------------------------------------------------------------------
543 // filenames are multibyte on Unix and widechar on Windows
544 #if wxMBFILES && wxUSE_UNICODE
545 #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
546 #define wxFNSTRINGCAST wxMBSTRINGCAST
548 #if defined( __WXOSX__ ) && wxMBFILES
549 #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
551 #define wxFNCONV(name) name
553 #define wxFNSTRINGCAST WXSTRINGCAST
556 #else // !wxUSE_WCHAR_T
558 // ----------------------------------------------------------------------------
559 // stand-ins in absence of wchar_t
560 // ----------------------------------------------------------------------------
562 class WXDLLIMPEXP_BASE wxMBConv
565 const char* cMB2WX(const char *psz
) const { return psz
; }
566 const char* cWX2MB(const char *psz
) const { return psz
; }
569 #define wxConvFile wxConvLocal
571 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
) wxConvLibc
,
575 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
577 #define wxFNCONV(name) name
578 #define wxFNSTRINGCAST WXSTRINGCAST
583 // ----------------------------------------------------------------------------
584 // macros for the most common conversions
585 // ----------------------------------------------------------------------------
588 #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s)
589 #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s)
591 // these functions should be used when the conversions really, really have
592 // to succeed (usually because we pass their results to a standard C
593 // function which would crash if we passed NULL to it), so these functions
594 // always return a valid pointer if their argument is non-NULL
596 // this function safety is achieved by trying wxConvLibc first, wxConvUTF8
597 // next if it fails and, finally, wxConvISO8859_1 which always succeeds
598 extern WXDLLIMPEXP_BASE wxWCharBuffer
wxSafeConvertMB2WX(const char *s
);
600 // this function uses wxConvLibc and wxConvUTF8(MAP_INVALID_UTF8_TO_OCTAL)
602 extern WXDLLIMPEXP_BASE wxCharBuffer
wxSafeConvertWX2MB(const wchar_t *ws
);
604 // no conversions to do
605 #define wxConvertWX2MB(s) (s)
606 #define wxConvertMB2WX(s) (s)
607 #define wxSafeConvertMB2WX(s) (s)
608 #define wxSafeConvertWX2MB(s) (s)
609 #endif // Unicode/ANSI
611 #endif // _WX_STRCONV_H_