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 // the error value returned by wxMBConv methods
33 #define wxCONV_FAILED ((size_t)-1)
35 // the default value for some length parameters meaning that the string is
37 #define wxNO_LEN ((size_t)-1)
39 // ----------------------------------------------------------------------------
40 // wxMBConv (abstract base class for conversions)
41 // ----------------------------------------------------------------------------
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
48 // You also have to implement Clone() to allow copying the conversions
51 // And you might need to override GetMBNulLen() as well.
52 class WXDLLIMPEXP_BASE wxMBConv
55 // The functions doing actual conversion from/to narrow to/from wide
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.
64 // In the special case when dstLen is 0 (outputBuf may be NULL then) the
65 // return value is the length of the needed buffer but nothing happens
66 // otherwise. If srcLen is wxNO_LEN, the entire string, up to and
67 // including the trailing NUL(s), is converted, otherwise exactly srcLen
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);
78 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
79 const char *src
, size_t srcLen
= wxNO_LEN
) const;
81 virtual size_t FromWChar(char *dst
, size_t dstLen
,
82 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
85 // Convenience functions for translating NUL-terminated strings: returns
86 // the buffer containing the converted string or NULL pointer if the
88 const wxWCharBuffer
cMB2WC(const char *in
) const;
89 const wxCharBuffer
cWC2MB(const wchar_t *in
) const;
91 // Convenience functions for converting strings which may contain embedded
92 // NULs and don't have to be NUL-terminated.
94 // inLen is the length of the buffer including trailing NUL if any: if the
95 // last 4 bytes of the buffer are all NULs, these functions are more
96 // efficient as they avoid copying the string, but otherwise a copy is made
97 // internally which could be quite bad for (very) long strings.
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)
104 cMB2WC(const char *in
, size_t inLen
, size_t *outLen
) const;
106 cWC2MB(const wchar_t *in
, size_t inLen
, size_t *outLen
) const;
108 // convenience functions for converting MB or WC to/from wxWin default
110 const wxWCharBuffer
cMB2WX(const char *psz
) const { return cMB2WC(psz
); }
111 const wxCharBuffer
cWX2MB(const wchar_t *psz
) const { return cWC2MB(psz
); }
112 const wchar_t* cWC2WX(const wchar_t *psz
) const { return psz
; }
113 const wchar_t* cWX2WC(const wchar_t *psz
) const { return psz
; }
115 const char* cMB2WX(const char *psz
) const { return psz
; }
116 const char* cWX2MB(const char *psz
) const { return psz
; }
117 const wxCharBuffer
cWC2WX(const wchar_t *psz
) const { return cWC2MB(psz
); }
118 const wxWCharBuffer
cWX2WC(const char *psz
) const { return cMB2WC(psz
); }
119 #endif // Unicode/ANSI
121 // this function is used in the implementation of cMB2WC() to distinguish
122 // between the following cases:
124 // a) var width encoding with strings terminated by a single NUL
125 // (usual multibyte encodings): return 1 in this case
126 // b) fixed width encoding with 2 bytes/char and so terminated by
127 // 2 NULs (UTF-16/UCS-2 and variants): return 2 in this case
128 // c) fixed width encoding with 4 bytes/char and so terminated by
129 // 4 NULs (UTF-32/UCS-4 and variants): return 4 in this case
131 // anything else is not supported currently and -1 should be returned
132 virtual size_t GetMBNulLen() const { return 1; }
134 // return the maximal value currently returned by GetMBNulLen() for any
136 static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; }
138 #if wxUSE_UNICODE_UTF8
139 // return true if the converter's charset is UTF-8, i.e. char* strings
140 // decoded using this object can be directly copied to wxString's internal
141 // storage without converting to WC and than back to UTF-8 MB string
142 virtual bool IsUTF8() const { return false; }
145 // The old conversion functions. The existing classes currently mostly
146 // implement these ones but we're in transition to using To/FromWChar()
147 // instead and any new classes should implement just the new functions.
148 // For now, however, we provide default implementation of To/FromWChar() in
149 // this base class in terms of MB2WC/WC2MB() to avoid having to rewrite all
150 // the conversions at once.
152 // On success, the return value is the length (i.e. the number of
153 // characters, not bytes) not counting the trailing NUL(s) of the converted
154 // string. On failure, (size_t)-1 is returned. In the special case when
155 // outputBuf is NULL the return value is the same one but nothing is
156 // written to the buffer.
158 // Note that outLen is the length of the output buffer, not the length of
159 // the input (which is always supposed to be terminated by one or more
160 // NULs, as appropriate for the encoding)!
161 virtual size_t MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const;
162 virtual size_t WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const;
165 // make a heap-allocated copy of this object
166 virtual wxMBConv
*Clone() const = 0;
168 // virtual dtor for any base class
172 // ----------------------------------------------------------------------------
173 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
174 // conversion (hence it depends on the current locale)
175 // ----------------------------------------------------------------------------
177 class WXDLLIMPEXP_BASE wxMBConvLibc
: public wxMBConv
180 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
181 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
183 virtual wxMBConv
*Clone() const { return new wxMBConvLibc
; }
185 #if wxUSE_UNICODE_UTF8
186 virtual bool IsUTF8() const { return wxLocaleIsUtf8
; }
192 // ----------------------------------------------------------------------------
193 // wxConvBrokenFileNames is made for Unix in Unicode mode when
194 // files are accidentally written in an encoding which is not
195 // the system encoding. Typically, the system encoding will be
196 // UTF8 but there might be files stored in ISO8859-1 on disk.
197 // ----------------------------------------------------------------------------
199 class WXDLLIMPEXP_BASE wxConvBrokenFileNames
: public wxMBConv
202 wxConvBrokenFileNames(const wxChar
*charset
);
203 wxConvBrokenFileNames(const wxConvBrokenFileNames
& conv
)
205 m_conv(conv
.m_conv
? conv
.m_conv
->Clone() : NULL
)
208 virtual ~wxConvBrokenFileNames() { delete m_conv
; }
210 virtual size_t MB2WC(wchar_t *out
, const char *in
, size_t outLen
) const
212 return m_conv
->MB2WC(out
, in
, outLen
);
215 virtual size_t WC2MB(char *out
, const wchar_t *in
, size_t outLen
) const
217 return m_conv
->WC2MB(out
, in
, outLen
);
220 virtual size_t GetMBNulLen() const
222 // cast needed to call a private function
223 return m_conv
->GetMBNulLen();
226 #if wxUSE_UNICODE_UTF8
227 virtual bool IsUTF8() const { return m_conv
->IsUTF8(); }
230 virtual wxMBConv
*Clone() const { return new wxConvBrokenFileNames(*this); }
233 // the conversion object we forward to
236 DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames
)
241 // ----------------------------------------------------------------------------
242 // wxMBConvUTF7 (for conversion using UTF7 encoding)
243 // ----------------------------------------------------------------------------
245 class WXDLLIMPEXP_BASE wxMBConvUTF7
: public wxMBConv
248 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
249 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
251 virtual wxMBConv
*Clone() const { return new wxMBConvUTF7
; }
254 // ----------------------------------------------------------------------------
255 // wxMBConvUTF8 (for conversion using UTF8 encoding)
256 // ----------------------------------------------------------------------------
258 class WXDLLIMPEXP_BASE wxMBConvUTF8
: public wxMBConv
261 // FIXME-UTF8: split this class into multiple classes, one strict and
262 // other lossy (PUA, OCTAL mappings)
265 MAP_INVALID_UTF8_NOT
= 0,
266 MAP_INVALID_UTF8_TO_PUA
= 1,
267 MAP_INVALID_UTF8_TO_OCTAL
= 2
270 wxMBConvUTF8(int options
= MAP_INVALID_UTF8_NOT
) : m_options(options
) { }
271 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
272 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
274 virtual wxMBConv
*Clone() const { return new wxMBConvUTF8(m_options
); }
276 #if wxUSE_UNICODE_UTF8
277 // NB: other mapping modes are not, strictly speaking, UTF-8, so we can't
278 // take the shortcut in that case
279 virtual bool IsUTF8() const { return m_options
== MAP_INVALID_UTF8_NOT
; }
286 // ----------------------------------------------------------------------------
287 // wxMBConvUTF16Base: for both LE and BE variants
288 // ----------------------------------------------------------------------------
290 class WXDLLIMPEXP_BASE wxMBConvUTF16Base
: public wxMBConv
293 enum { BYTES_PER_CHAR
= 2 };
295 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
298 // return the length of the buffer using srcLen if it's not wxNO_LEN and
299 // computing the length ourselves if it is; also checks that the length is
300 // even if specified as we need an entire number of UTF-16 characters and
301 // returns wxNO_LEN which indicates error if it is odd
302 static size_t GetLength(const char *src
, size_t srcLen
);
305 // ----------------------------------------------------------------------------
306 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
307 // ----------------------------------------------------------------------------
309 class WXDLLIMPEXP_BASE wxMBConvUTF16LE
: public wxMBConvUTF16Base
312 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
313 const char *src
, size_t srcLen
= wxNO_LEN
) const;
314 virtual size_t FromWChar(char *dst
, size_t dstLen
,
315 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
316 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16LE
; }
319 // ----------------------------------------------------------------------------
320 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
321 // ----------------------------------------------------------------------------
323 class WXDLLIMPEXP_BASE wxMBConvUTF16BE
: public wxMBConvUTF16Base
326 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
327 const char *src
, size_t srcLen
= wxNO_LEN
) const;
328 virtual size_t FromWChar(char *dst
, size_t dstLen
,
329 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
330 virtual wxMBConv
*Clone() const { return new wxMBConvUTF16BE
; }
333 // ----------------------------------------------------------------------------
334 // wxMBConvUTF32Base: base class for both LE and BE variants
335 // ----------------------------------------------------------------------------
337 class WXDLLIMPEXP_BASE wxMBConvUTF32Base
: public wxMBConv
340 enum { BYTES_PER_CHAR
= 4 };
342 virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR
; }
345 // this is similar to wxMBConvUTF16Base method with the same name except
346 // that, of course, it verifies that length is divisible by 4 if given and
348 static size_t GetLength(const char *src
, size_t srcLen
);
351 // ----------------------------------------------------------------------------
352 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
353 // ----------------------------------------------------------------------------
355 class WXDLLIMPEXP_BASE wxMBConvUTF32LE
: public wxMBConvUTF32Base
358 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
359 const char *src
, size_t srcLen
= wxNO_LEN
) const;
360 virtual size_t FromWChar(char *dst
, size_t dstLen
,
361 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
362 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32LE
; }
365 // ----------------------------------------------------------------------------
366 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
367 // ----------------------------------------------------------------------------
369 class WXDLLIMPEXP_BASE wxMBConvUTF32BE
: public wxMBConvUTF32Base
372 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
373 const char *src
, size_t srcLen
= wxNO_LEN
) const;
374 virtual size_t FromWChar(char *dst
, size_t dstLen
,
375 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
376 virtual wxMBConv
*Clone() const { return new wxMBConvUTF32BE
; }
379 // ----------------------------------------------------------------------------
380 // wxCSConv (for conversion based on loadable char sets)
381 // ----------------------------------------------------------------------------
383 #include "wx/fontenc.h"
385 class WXDLLIMPEXP_BASE wxCSConv
: public wxMBConv
388 // we can be created either from charset name or from an encoding constant
389 // but we can't have both at once
390 wxCSConv(const wxChar
*charset
);
391 wxCSConv(wxFontEncoding encoding
);
393 wxCSConv(const wxCSConv
& conv
);
396 wxCSConv
& operator=(const wxCSConv
& conv
);
398 virtual size_t ToWChar(wchar_t *dst
, size_t dstLen
,
399 const char *src
, size_t srcLen
= wxNO_LEN
) const;
400 virtual size_t FromWChar(char *dst
, size_t dstLen
,
401 const wchar_t *src
, size_t srcLen
= wxNO_LEN
) const;
402 virtual size_t MB2WC(wchar_t *outputBuf
, const char *psz
, size_t outputSize
) const;
403 virtual size_t WC2MB(char *outputBuf
, const wchar_t *psz
, size_t outputSize
) const;
404 virtual size_t GetMBNulLen() const;
406 #if wxUSE_UNICODE_UTF8
407 virtual bool IsUTF8() const;
410 virtual wxMBConv
*Clone() const { return new wxCSConv(*this); }
414 // return true if the conversion could be initilized successfully
418 // common part of all ctors
421 // creates m_convReal if necessary
422 void CreateConvIfNeeded() const;
424 // do create m_convReal (unconditionally)
425 wxMBConv
*DoCreate() const;
427 // set the name (may be only called when m_name == NULL), makes copy of
429 void SetName(const wxChar
*charset
);
432 // note that we can't use wxString here because of compilation
433 // dependencies: we're included from wx/string.h
435 wxFontEncoding m_encoding
;
437 // use CreateConvIfNeeded() before accessing m_convReal!
438 wxMBConv
*m_convReal
;
443 // ----------------------------------------------------------------------------
444 // declare predefined conversion objects
445 // ----------------------------------------------------------------------------
447 // Note: this macro is an implementation detail (see the comment in
448 // strconv.cpp). The wxGet_XXX() and wxGet_XXXPtr() functions shouldn't be
449 // used by user code and neither should XXXPtr, use the wxConvXXX macro
451 #define WX_DECLARE_GLOBAL_CONV(klass, name) \
452 extern WXDLLIMPEXP_DATA_BASE(klass*) name##Ptr; \
453 extern WXDLLIMPEXP_BASE klass* wxGet_##name##Ptr(); \
454 inline klass& wxGet_##name() \
457 name##Ptr = wxGet_##name##Ptr(); \
462 // conversion to be used with all standard functions affected by locale, e.g.
463 // strtol(), strftime(), ...
464 WX_DECLARE_GLOBAL_CONV(wxMBConv
, wxConvLibc
)
465 #define wxConvLibc wxGet_wxConvLibc()
467 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
468 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvISO8859_1
)
469 #define wxConvISO8859_1 wxGet_wxConvISO8859_1()
471 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF8
, wxConvUTF8
)
472 #define wxConvUTF8 wxGet_wxConvUTF8()
474 WX_DECLARE_GLOBAL_CONV(wxMBConvUTF7
, wxConvUTF7
)
475 #define wxConvUTF7 wxGet_wxConvUTF7()
477 // conversion used for the file names on the systems where they're not Unicode
478 // (basically anything except Windows)
480 // this is used by all file functions, can be changed by the application
482 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
483 // under Windows normally)
484 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvFileName
;
486 // backwards compatible define
487 #define wxConvFile (*wxConvFileName)
489 // the current conversion object, may be set to any conversion, is used by
490 // default in a couple of places inside wx (initially same as wxConvLibc)
491 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
493 // the conversion corresponding to the current locale
494 WX_DECLARE_GLOBAL_CONV(wxCSConv
, wxConvLocal
)
495 #define wxConvLocal wxGet_wxConvLocal()
497 // the conversion corresponding to the encoding of the standard UI elements
499 // by default this is the same as wxConvLocal but may be changed if the program
500 // needs to use a fixed encoding
501 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvUI
;
503 #undef WX_DECLARE_GLOBAL_CONV
505 // ----------------------------------------------------------------------------
506 // endianness-dependent conversions
507 // ----------------------------------------------------------------------------
509 #ifdef WORDS_BIGENDIAN
510 typedef wxMBConvUTF16BE wxMBConvUTF16
;
511 typedef wxMBConvUTF32BE wxMBConvUTF32
;
513 typedef wxMBConvUTF16LE wxMBConvUTF16
;
514 typedef wxMBConvUTF32LE wxMBConvUTF32
;
517 // ----------------------------------------------------------------------------
518 // filename conversion macros
519 // ----------------------------------------------------------------------------
521 // filenames are multibyte on Unix and widechar on Windows
522 #if defined(__UNIX__) || defined(__WXMAC__)
528 #if wxMBFILES && wxUSE_UNICODE
529 #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
530 #define wxFNSTRINGCAST wxMBSTRINGCAST
532 #if defined( __WXOSX__ ) && wxMBFILES
533 #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
535 #define wxFNCONV(name) name
537 #define wxFNSTRINGCAST WXSTRINGCAST
540 #else // !wxUSE_WCHAR_T
542 // ----------------------------------------------------------------------------
543 // stand-ins in absence of wchar_t
544 // ----------------------------------------------------------------------------
546 class WXDLLIMPEXP_BASE wxMBConv
549 const char* cMB2WX(const char *psz
) const { return psz
; }
550 const char* cWX2MB(const char *psz
) const { return psz
; }
553 #define wxConvFile wxConvLocal
555 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
) wxConvLibc
,
559 extern WXDLLIMPEXP_DATA_BASE(wxMBConv
*) wxConvCurrent
;
561 #define wxFNCONV(name) name
562 #define wxFNSTRINGCAST WXSTRINGCAST
567 // ----------------------------------------------------------------------------
568 // macros for the most common conversions
569 // ----------------------------------------------------------------------------
572 #define wxConvertWX2MB(s) wxConvCurrent->cWX2MB(s)
573 #define wxConvertMB2WX(s) wxConvCurrent->cMB2WX(s)
575 // these functions should be used when the conversions really, really have
576 // to succeed (usually because we pass their results to a standard C
577 // function which would crash if we passed NULL to it), so these functions
578 // always return a valid pointer if their argument is non-NULL
580 // this function safety is achieved by trying wxConvLibc first, wxConvUTF8
581 // next if it fails and, finally, wxConvISO8859_1 which always succeeds
582 extern WXDLLIMPEXP_BASE wxWCharBuffer
wxSafeConvertMB2WX(const char *s
);
584 // this function uses wxConvLibc and wxConvUTF8(MAP_INVALID_UTF8_TO_OCTAL)
586 extern WXDLLIMPEXP_BASE wxCharBuffer
wxSafeConvertWX2MB(const wchar_t *ws
);
588 // no conversions to do
589 #define wxConvertWX2MB(s) (s)
590 #define wxConvertMB2WX(s) (s)
591 #define wxSafeConvertMB2WX(s) (s)
592 #define wxSafeConvertWX2MB(s) (s)
593 #endif // Unicode/ANSI
595 #endif // _WX_STRCONV_H_