include/wx/strconv.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        strconv.h
   3 // Purpose:     conversion routines for char sets any Unicode
   4 // Author:      Robert Roebling, Ove Kaaven
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Ove Kaaven, Robert Roebling, Vadim Zeitlin
   9 // Licence:     wxWindows licence
  10 ///////////////////////////////////////////////////////////////////////////////
  11
  12 #ifndef _WX_WXSTRCONVH__
  13 #define _WX_WXSTRCONVH__
  14
  15 #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
  16   #pragma interface "strconv.h"
  17 #endif
  18
  19 #include "wx/defs.h"
  20 #include "wx/wxchar.h"
  21 #include "wx/buffer.h"
  22
  23 #ifdef __DIGITALMARS__
  24 #include "typeinfo.h"
  25 #endif
  26
  27 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  28 #  undef __BSEXCPT__
  29 #endif
  30
  31 #include <stdlib.h>
  32
  33 #if wxUSE_WCHAR_T
  34
  35 // ----------------------------------------------------------------------------
  36 // wxMBConv (abstract base class for conversions)
  37 // ----------------------------------------------------------------------------
  38
  39 class WXDLLIMPEXP_BASE wxMBConv
  40 {
  41 public:
  42     // the actual conversion takes place here
  43     //
  44     // note that outputSize is the size of the output buffer, not the length of input
  45     // (the latter is always supposed to be NUL-terminated)
  46     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const = 0;
  47     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const = 0;
  48
  49     // MB <-> WC
  50     const wxWCharBuffer cMB2WC(const char *psz) const;
  51     const wxCharBuffer cWC2MB(const wchar_t *psz) const;
  52
  53     // MB <-> WC for strings with embedded null characters
  54     //
  55     // pszLen length of the input string
  56     // pOutSize gets the final size of the converted string
  57     const wxWCharBuffer cMB2WC(const char *psz, size_t pszLen, size_t* pOutSize) const;
  58     const wxCharBuffer cWC2MB(const wchar_t *psz, size_t pszLen, size_t* pOutSize) const;
  59
  60     // convenience functions for converting MB or WC to/from wxWin default
  61 #if wxUSE_UNICODE
  62     const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
  63     const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }
  64     const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; }
  65     const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; }
  66 #else // ANSI
  67     const char* cMB2WX(const char *psz) const { return psz; }
  68     const char* cWX2MB(const char *psz) const { return psz; }
  69     const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); }
  70     const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); }
  71 #endif // Unicode/ANSI
  72
  73     // virtual dtor for any base class
  74     virtual ~wxMBConv();
  75 };
  76
  77 // ----------------------------------------------------------------------------
  78 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
  79 //              conversion (hence it depends on the current locale)
  80 // ----------------------------------------------------------------------------
  81
  82 class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv
  83 {
  84 public:
  85     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
  86     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
  87 };
  88
  89 // ----------------------------------------------------------------------------
  90 // wxMBConvUTF7 (for conversion using UTF7 encoding)
  91 // ----------------------------------------------------------------------------
  92
  93 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
  94 {
  95 public:
  96     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
  97     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
  98 };
  99
 100 // ----------------------------------------------------------------------------
 101 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 102 // ----------------------------------------------------------------------------
 103
 104 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 105 {
 106 public:
 107     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 108     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 109 };
 110
 111 // ----------------------------------------------------------------------------
 112 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 113 // ----------------------------------------------------------------------------
 114
 115 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv
 116 {
 117 public:
 118     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 119     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 120 };
 121
 122 // ----------------------------------------------------------------------------
 123 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 124 // ----------------------------------------------------------------------------
 125
 126 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv
 127 {
 128 public:
 129     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 130     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 131 };
 132
 133 // ----------------------------------------------------------------------------
 134 // wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding)
 135 // ----------------------------------------------------------------------------
 136
 137 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv
 138 {
 139 public:
 140     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 141     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 142 };
 143
 144 // ----------------------------------------------------------------------------
 145 // wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding)
 146 // ----------------------------------------------------------------------------
 147
 148 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv
 149 {
 150 public:
 151     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 152     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 153 };
 154
 155 // ----------------------------------------------------------------------------
 156 // wxCSConv (for conversion based on loadable char sets)
 157 // ----------------------------------------------------------------------------
 158
 159 #include "wx/fontenc.h"
 160
 161 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 162 {
 163 public:
 164     // we can be created either from charset name or from an encoding constant
 165     // but we can't have both at once
 166     wxCSConv(const wxChar *charset);
 167     wxCSConv(wxFontEncoding encoding);
 168
 169     wxCSConv(const wxCSConv& conv);
 170     virtual ~wxCSConv();
 171
 172     wxCSConv& operator=(const wxCSConv& conv);
 173
 174     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 175     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 176
 177     void Clear() ;
 178
 179 private:
 180     // common part of all ctors
 181     void Init();
 182
 183     // creates m_convReal if necessary
 184     void CreateConvIfNeeded() const;
 185
 186     // do create m_convReal (unconditionally)
 187     wxMBConv *DoCreate() const;
 188
 189     // set the name (may be only called when m_name == NULL), makes copy of
 190     // charset string
 191     void SetName(const wxChar *charset);
 192
 193
 194     // note that we can't use wxString here because of compilation
 195     // dependencies: we're included from wx/string.h
 196     wxChar *m_name;
 197     wxFontEncoding m_encoding;
 198
 199     // use CreateConvIfNeeded() before accessing m_convReal!
 200     wxMBConv *m_convReal;
 201     bool m_deferred;
 202 };
 203
 204 // ----------------------------------------------------------------------------
 205 // declare predefined conversion objects
 206 // ----------------------------------------------------------------------------
 207
 208 // conversion to be used with all standard functions affected by locale, e.g.
 209 // strtol(), strftime(), ...
 210 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 211
 212 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 213 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 214 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 215 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 216
 217 // conversion used for the file names on the systems where they're not Unicode
 218 // (basically anything except Windows)
 219 //
 220 // this is used by all file functions, can be changed by the application
 221 //
 222 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 223 // under Windows normally)
 224 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 225
 226 // backwards compatible define
 227 #define wxConvFile (*wxConvFileName)
 228
 229 // the current conversion object, may be set to any conversion, is used by
 230 // default in a couple of places inside wx (initially same as wxConvLibc)
 231 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 232
 233 // ???
 234 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 235
 236
 237 // ----------------------------------------------------------------------------
 238 // endianness-dependent conversions
 239 // ----------------------------------------------------------------------------
 240
 241 #ifdef WORDS_BIGENDIAN
 242     typedef wxMBConvUTF16BE wxMBConvUTF16;
 243     typedef wxMBConvUTF32BE wxMBConvUTF32;
 244 #else
 245     typedef wxMBConvUTF16LE wxMBConvUTF16;
 246     typedef wxMBConvUTF32LE wxMBConvUTF32;
 247 #endif
 248
 249 // ----------------------------------------------------------------------------
 250 // filename conversion macros
 251 // ----------------------------------------------------------------------------
 252
 253 // filenames are multibyte on Unix and probably widechar on Windows?
 254 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 255     #define wxMBFILES 1
 256 #else
 257     #define wxMBFILES 0
 258 #endif
 259
 260 #if wxMBFILES && wxUSE_UNICODE
 261     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 262     #define wxFNSTRINGCAST wxMBSTRINGCAST
 263 #else
 264 #if defined( __WXOSX__ ) && wxMBFILES
 265     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 266 #else
 267     #define wxFNCONV(name) name
 268 #endif
 269     #define wxFNSTRINGCAST WXSTRINGCAST
 270 #endif
 271
 272 #else // !wxUSE_WCHAR_T
 273
 274 // ----------------------------------------------------------------------------
 275 // stand-ins in absence of wchar_t
 276 // ----------------------------------------------------------------------------
 277
 278 class WXDLLIMPEXP_BASE wxMBConv
 279 {
 280 public:
 281     const char* cMB2WX(const char *psz) const { return psz; }
 282     const char* cWX2MB(const char *psz) const { return psz; }
 283 };
 284
 285 #define wxConvFile wxConvLocal
 286
 287 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 288                                        wxConvLocal,
 289                                        wxConvISO8859_1,
 290                                        wxConvUTF8;
 291 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 292
 293 #define wxFNCONV(name) name
 294 #define wxFNSTRINGCAST WXSTRINGCAST
 295
 296 #endif
 297   // wxUSE_WCHAR_T
 298
 299 // ----------------------------------------------------------------------------
 300 // macros for the most common conversions
 301 // ----------------------------------------------------------------------------
 302
 303 #if wxUSE_UNICODE
 304     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 305     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 306 #else // ANSI
 307     // no conversions to do
 308     #define wxConvertWX2MB(s)   (s)
 309     #define wxConvertMB2WX(s)   (s)
 310 #endif // Unicode/ANSI
 311
 312 #endif
 313   // _WX_WXSTRCONVH__
 314