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     enum {
 108         MAP_INVALID_UTF8_NOT = 0,
 109         MAP_INVALID_UTF8_TO_PUA = 1,
 110         MAP_INVALID_UTF8_TO_OCTAL = 2
 111     };
 112
 113     wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
 114     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 115     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 116
 117 private:
 118     int m_options;
 119 };
 120
 121 // ----------------------------------------------------------------------------
 122 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 123 // ----------------------------------------------------------------------------
 124
 125 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv
 126 {
 127 public:
 128     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 129     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 130 };
 131
 132 // ----------------------------------------------------------------------------
 133 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 134 // ----------------------------------------------------------------------------
 135
 136 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv
 137 {
 138 public:
 139     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 140     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 141 };
 142
 143 // ----------------------------------------------------------------------------
 144 // wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding)
 145 // ----------------------------------------------------------------------------
 146
 147 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv
 148 {
 149 public:
 150     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 151     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 152 };
 153
 154 // ----------------------------------------------------------------------------
 155 // wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding)
 156 // ----------------------------------------------------------------------------
 157
 158 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv
 159 {
 160 public:
 161     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 162     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 163 };
 164
 165 // ----------------------------------------------------------------------------
 166 // wxCSConv (for conversion based on loadable char sets)
 167 // ----------------------------------------------------------------------------
 168
 169 #include "wx/fontenc.h"
 170
 171 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 172 {
 173 public:
 174     // we can be created either from charset name or from an encoding constant
 175     // but we can't have both at once
 176     wxCSConv(const wxChar *charset);
 177     wxCSConv(wxFontEncoding encoding);
 178
 179     wxCSConv(const wxCSConv& conv);
 180     virtual ~wxCSConv();
 181
 182     wxCSConv& operator=(const wxCSConv& conv);
 183
 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     void Clear() ;
 188
 189 private:
 190     // common part of all ctors
 191     void Init();
 192
 193     // creates m_convReal if necessary
 194     void CreateConvIfNeeded() const;
 195
 196     // do create m_convReal (unconditionally)
 197     wxMBConv *DoCreate() const;
 198
 199     // set the name (may be only called when m_name == NULL), makes copy of
 200     // charset string
 201     void SetName(const wxChar *charset);
 202
 203
 204     // note that we can't use wxString here because of compilation
 205     // dependencies: we're included from wx/string.h
 206     wxChar *m_name;
 207     wxFontEncoding m_encoding;
 208
 209     // use CreateConvIfNeeded() before accessing m_convReal!
 210     wxMBConv *m_convReal;
 211     bool m_deferred;
 212 };
 213
 214 #ifdef __WXGTK20__
 215
 216 // ----------------------------------------------------------------------------
 217 // wxConvBrokenFileNames is made for GTK2 in Unicode mode when
 218 // files are accidentally written in an encoding which is not
 219 // the system encoding. Typically, the system encoding will be
 220 // UTF8 but there might be files stored in ISO8859-1 on disk.
 221 // ----------------------------------------------------------------------------
 222
 223 class wxConvBrokenFileNames : public wxMBConv
 224 {
 225 public:
 226     wxConvBrokenFileNames();
 227     virtual ~wxConvBrokenFileNames() { delete m_conv; }
 228
 229     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 230     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 231
 232 private:
 233     // the conversion object we forward to
 234     wxMBConv *m_conv;
 235 };
 236 #endif
 237 // __WXGTK20__
 238
 239 // ----------------------------------------------------------------------------
 240 // declare predefined conversion objects
 241 // ----------------------------------------------------------------------------
 242
 243 // conversion to be used with all standard functions affected by locale, e.g.
 244 // strtol(), strftime(), ...
 245 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 246
 247 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 248 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 249 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 250 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 251
 252 // conversion used for the file names on the systems where they're not Unicode
 253 // (basically anything except Windows)
 254 //
 255 // this is used by all file functions, can be changed by the application
 256 //
 257 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 258 // under Windows normally)
 259 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 260
 261 // backwards compatible define
 262 #define wxConvFile (*wxConvFileName)
 263
 264 // the current conversion object, may be set to any conversion, is used by
 265 // default in a couple of places inside wx (initially same as wxConvLibc)
 266 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 267
 268 // ???
 269 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 270
 271
 272 // ----------------------------------------------------------------------------
 273 // endianness-dependent conversions
 274 // ----------------------------------------------------------------------------
 275
 276 #ifdef WORDS_BIGENDIAN
 277     typedef wxMBConvUTF16BE wxMBConvUTF16;
 278     typedef wxMBConvUTF32BE wxMBConvUTF32;
 279 #else
 280     typedef wxMBConvUTF16LE wxMBConvUTF16;
 281     typedef wxMBConvUTF32LE wxMBConvUTF32;
 282 #endif
 283
 284 // ----------------------------------------------------------------------------
 285 // filename conversion macros
 286 // ----------------------------------------------------------------------------
 287
 288 // filenames are multibyte on Unix and probably widechar on Windows?
 289 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 290     #define wxMBFILES 1
 291 #else
 292     #define wxMBFILES 0
 293 #endif
 294
 295 #if wxMBFILES && wxUSE_UNICODE
 296     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 297     #define wxFNSTRINGCAST wxMBSTRINGCAST
 298 #else
 299 #if defined( __WXOSX__ ) && wxMBFILES
 300     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 301 #else
 302     #define wxFNCONV(name) name
 303 #endif
 304     #define wxFNSTRINGCAST WXSTRINGCAST
 305 #endif
 306
 307 #else // !wxUSE_WCHAR_T
 308
 309 // ----------------------------------------------------------------------------
 310 // stand-ins in absence of wchar_t
 311 // ----------------------------------------------------------------------------
 312
 313 class WXDLLIMPEXP_BASE wxMBConv
 314 {
 315 public:
 316     const char* cMB2WX(const char *psz) const { return psz; }
 317     const char* cWX2MB(const char *psz) const { return psz; }
 318 };
 319
 320 #define wxConvFile wxConvLocal
 321
 322 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 323                                        wxConvLocal,
 324                                        wxConvISO8859_1,
 325                                        wxConvUTF8;
 326 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 327
 328 #define wxFNCONV(name) name
 329 #define wxFNSTRINGCAST WXSTRINGCAST
 330
 331 #endif
 332   // wxUSE_WCHAR_T
 333
 334 // ----------------------------------------------------------------------------
 335 // macros for the most common conversions
 336 // ----------------------------------------------------------------------------
 337
 338 #if wxUSE_UNICODE
 339     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 340     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 341 #else // ANSI
 342     // no conversions to do
 343     #define wxConvertWX2MB(s)   (s)
 344     #define wxConvertMB2WX(s)   (s)
 345 #endif // Unicode/ANSI
 346
 347 #endif
 348   // _WX_WXSTRCONVH__
 349