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     // actual conversion for strings with embedded null characters
  50     //
  51     // outputSize is the size of the output buffer
  52     // pszLen is the length of the input string (including all but last null character)
  53     size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize, size_t pszLen) const;
  54     size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize, size_t pszLen) const;
  55
  56     // MB <-> WC
  57     const wxWCharBuffer cMB2WC(const char *psz) const;
  58     const wxCharBuffer cWC2MB(const wchar_t *psz) 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 // not very accurately named because it is not necessarily of type wxMBConvLibc
  90 // (but the name can't eb changed because of backwards compatibility) default
  91 // conversion
  92 WXDLLIMPEXP_DATA_BASE(extern wxMBConv&) wxConvLibc;
  93
  94 // ----------------------------------------------------------------------------
  95 // wxMBConvUTF7 (for conversion using UTF7 encoding)
  96 // ----------------------------------------------------------------------------
  97
  98 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
  99 {
 100 public:
 101     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 102     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 103 };
 104
 105 WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF7&) wxConvUTF7;
 106
 107 // ----------------------------------------------------------------------------
 108 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 109 // ----------------------------------------------------------------------------
 110
 111 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 112 {
 113 public:
 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
 118 WXDLLIMPEXP_DATA_BASE(extern wxMBConvUTF8&) wxConvUTF8;
 119
 120 // ----------------------------------------------------------------------------
 121 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 122 // ----------------------------------------------------------------------------
 123
 124 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConv
 125 {
 126 public:
 127     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 128     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 129 };
 130
 131 // ----------------------------------------------------------------------------
 132 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 133 // ----------------------------------------------------------------------------
 134
 135 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConv
 136 {
 137 public:
 138     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 139     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 140 };
 141
 142 // ----------------------------------------------------------------------------
 143 // wxMBConvUCS4LE (for conversion using UTF32 Little Endian encoding)
 144 // ----------------------------------------------------------------------------
 145
 146 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConv
 147 {
 148 public:
 149     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 150     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 151 };
 152
 153 // ----------------------------------------------------------------------------
 154 // wxMBConvUCS4BE (for conversion using UTF32 Big Endian encoding)
 155 // ----------------------------------------------------------------------------
 156
 157 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConv
 158 {
 159 public:
 160     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 161     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 162 };
 163
 164 // ----------------------------------------------------------------------------
 165 // wxCSConv (for conversion based on loadable char sets)
 166 // ----------------------------------------------------------------------------
 167
 168 #include "wx/fontenc.h"
 169
 170 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 171 {
 172 public:
 173     // we can be created either from charset name or from an encoding constant
 174     // but we can't have both at once
 175     wxCSConv(const wxChar *charset);
 176     wxCSConv(wxFontEncoding encoding);
 177
 178     wxCSConv(const wxCSConv& conv);
 179     virtual ~wxCSConv();
 180
 181     wxCSConv& operator=(const wxCSConv& conv);
 182
 183     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 184     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 185
 186     void Clear() ;
 187
 188 private:
 189     // common part of all ctors
 190     void Init();
 191
 192     // creates m_convReal if necessary
 193     void CreateConvIfNeeded() const;
 194
 195     // do create m_convReal (unconditionally)
 196     wxMBConv *DoCreate() const;
 197
 198     // set the name (may be only called when m_name == NULL), makes copy of
 199     // charset string
 200     void SetName(const wxChar *charset);
 201
 202
 203     // note that we can't use wxString here because of compilation
 204     // dependencies: we're included from wx/string.h
 205     wxChar *m_name;
 206     wxFontEncoding m_encoding;
 207
 208     // use CreateConvIfNeeded() before accessing m_convReal!
 209     wxMBConv *m_convReal;
 210     bool m_deferred;
 211 };
 212
 213 #define wxConvFile wxConvLocal
 214 WXDLLIMPEXP_DATA_BASE(extern wxCSConv&) wxConvLocal;
 215 WXDLLIMPEXP_DATA_BASE(extern wxCSConv&) wxConvISO8859_1;
 216 WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent;
 217
 218 // ----------------------------------------------------------------------------
 219 // endianness-dependent conversions
 220 // ----------------------------------------------------------------------------
 221
 222 #ifdef WORDS_BIGENDIAN
 223     typedef wxMBConvUTF16BE wxMBConvUTF16;
 224     typedef wxMBConvUTF32BE wxMBConvUTF32;
 225 #else
 226     typedef wxMBConvUTF16LE wxMBConvUTF16;
 227     typedef wxMBConvUTF32LE wxMBConvUTF32;
 228 #endif
 229
 230 // ----------------------------------------------------------------------------
 231 // filename conversion macros
 232 // ----------------------------------------------------------------------------
 233
 234 // filenames are multibyte on Unix and probably widechar on Windows?
 235 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 236     #define wxMBFILES 1
 237 #else
 238     #define wxMBFILES 0
 239 #endif
 240
 241 #if wxMBFILES && wxUSE_UNICODE
 242     #define wxFNCONV(name) wxConvFile.cWX2MB(name)
 243     #define wxFNSTRINGCAST wxMBSTRINGCAST
 244 #else
 245     #define wxFNCONV(name) name
 246     #define wxFNSTRINGCAST WXSTRINGCAST
 247 #endif
 248
 249 #else
 250   // !wxUSE_WCHAR_T
 251
 252 // ----------------------------------------------------------------------------
 253 // stand-ins in absence of wchar_t
 254 // ----------------------------------------------------------------------------
 255
 256 class WXDLLIMPEXP_BASE wxMBConv
 257 {
 258 public:
 259     const char* cMB2WX(const char *psz) const { return psz; }
 260     const char* cWX2MB(const char *psz) const { return psz; }
 261 };
 262
 263 #define wxConvFile wxConvLocal
 264
 265 WXDLLIMPEXP_DATA_BASE(extern wxMBConv) wxConvLibc,
 266                                        wxConvLocal,
 267                                        wxConvISO8859_1,
 268                                        wxConvUTF8;
 269 WXDLLIMPEXP_DATA_BASE(extern wxMBConv *) wxConvCurrent;
 270
 271 #define wxFNCONV(name) name
 272 #define wxFNSTRINGCAST WXSTRINGCAST
 273
 274 #endif
 275   // wxUSE_WCHAR_T
 276
 277 // ----------------------------------------------------------------------------
 278 // macros for the most common conversions
 279 // ----------------------------------------------------------------------------
 280
 281 #if wxUSE_UNICODE
 282     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 283     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 284 #else // ANSI
 285     // no conversions to do
 286     #define wxConvertWX2MB(s)   (s)
 287     #define wxConvertMB2WX(s)   (s)
 288 #endif // Unicode/ANSI
 289
 290 #endif
 291   // _WX_WXSTRCONVH__
 292