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 #include "wx/defs.h"
  16 #include "wx/wxchar.h"
  17 #include "wx/buffer.h"
  18
  19 #ifdef __DIGITALMARS__
  20 #include "typeinfo.h"
  21 #endif
  22
  23 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  24 #  undef __BSEXCPT__
  25 #endif
  26
  27 #include <stdlib.h>
  28
  29 #if wxUSE_WCHAR_T
  30
  31 // ----------------------------------------------------------------------------
  32 // wxMBConv (abstract base class for conversions)
  33 // ----------------------------------------------------------------------------
  34
  35 class WXDLLIMPEXP_BASE wxMBConv
  36 {
  37 public:
  38     // The functions doing actual conversion. On success, the return value is
  39     // the length (i.e. the number of characters, not bytes, and not counting
  40     // the trailing L'\0') of the converted string. On failure, (size_t)-1 is
  41     // returned. In the special case when outputBuf is NULL the return value is
  42     // the same one but nothing is written to the buffer.
  43     //
  44     // Note that outLen is the length of the output buffer, not the length of
  45     // the input (which is always supposed to be terminated by one or more
  46     // NULs, as appropriate for the encoding)!
  47     virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const = 0;
  48     virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const = 0;
  49
  50     // MB <-> WC
  51     const wxWCharBuffer cMB2WC(const char *in) const;
  52     const wxCharBuffer cWC2MB(const wchar_t *in) const;
  53
  54     // Functions converting strings which may contain embedded NULs and don't
  55     // have to be NUL-terminated.
  56     //
  57     // inLen is the length of the buffer including trailing NUL if any: if the
  58     // last 4 bytes of the buffer are all NULs, these functions are more
  59     // efficient as they avoid copying the string, but otherwise a copy is made
  60     // internally which could be quite bad for (very) long strings.
  61     //
  62     // outLen receives, if not NULL, the length of the converted string or 0 if
  63     // the conversion failed (returning 0 and not -1 in this case makes it
  64     // difficult to distinguish between failed conversion and empty input but
  65     // this is done for backwards compatibility)
  66     const wxWCharBuffer
  67         cMB2WC(const char *in, size_t inLen, size_t *outLen) const;
  68     const wxCharBuffer
  69         cWC2MB(const wchar_t *in, size_t inLen, size_t *outLen) const;
  70
  71     // convenience functions for converting MB or WC to/from wxWin default
  72 #if wxUSE_UNICODE
  73     const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
  74     const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }
  75     const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; }
  76     const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; }
  77 #else // ANSI
  78     const char* cMB2WX(const char *psz) const { return psz; }
  79     const char* cWX2MB(const char *psz) const { return psz; }
  80     const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); }
  81     const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); }
  82 #endif // Unicode/ANSI
  83
  84     // virtual dtor for any base class
  85     virtual ~wxMBConv();
  86
  87 private:
  88     // this function must return the multibyte representation of L'\0'
  89     //
  90     // on error, nulLen should be set to -1
  91     virtual const char *GetMBNul(size_t *nulLen) const
  92     {
  93         *nulLen = 1;
  94
  95         return "";
  96     }
  97 };
  98
  99 // ----------------------------------------------------------------------------
 100 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
 101 //              conversion (hence it depends on the current locale)
 102 // ----------------------------------------------------------------------------
 103
 104 class WXDLLIMPEXP_BASE wxMBConvLibc : 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 #ifdef __UNIX__
 112
 113 // ----------------------------------------------------------------------------
 114 // wxConvBrokenFileNames is made for Unix in Unicode mode when
 115 // files are accidentally written in an encoding which is not
 116 // the system encoding. Typically, the system encoding will be
 117 // UTF8 but there might be files stored in ISO8859-1 on disk.
 118 // ----------------------------------------------------------------------------
 119
 120 class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv
 121 {
 122 public:
 123     wxConvBrokenFileNames(const wxChar *charset);
 124     virtual ~wxConvBrokenFileNames() { delete m_conv; }
 125
 126     virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const
 127     {
 128         return m_conv->MB2WC(out, in, outLen);
 129     }
 130
 131     virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const
 132     {
 133         return m_conv->WC2MB(out, in, outLen);
 134     }
 135
 136 private:
 137     virtual wxCharBuffer GetMBNul(size_t *nulLen) const
 138     {
 139         return m_conv->GetMBNul(nulLen);
 140     }
 141
 142
 143     // the conversion object we forward to
 144     wxMBConv *m_conv;
 145 };
 146
 147 #endif // __UNIX__
 148
 149 // ----------------------------------------------------------------------------
 150 // wxMBConvUTF7 (for conversion using UTF7 encoding)
 151 // ----------------------------------------------------------------------------
 152
 153 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 154 {
 155 public:
 156     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 157     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 158 };
 159
 160 // ----------------------------------------------------------------------------
 161 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 162 // ----------------------------------------------------------------------------
 163
 164 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 165 {
 166 public:
 167     enum {
 168         MAP_INVALID_UTF8_NOT = 0,
 169         MAP_INVALID_UTF8_TO_PUA = 1,
 170         MAP_INVALID_UTF8_TO_OCTAL = 2
 171     };
 172
 173     wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
 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 private:
 178     int m_options;
 179 };
 180
 181 // ----------------------------------------------------------------------------
 182 // wxMBConvUTF16Base: for both LE and BE variants
 183 // ----------------------------------------------------------------------------
 184
 185 class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv
 186 {
 187 private:
 188     virtual const char *GetMBNul(size_t *nulLen) const
 189     {
 190         *nulLen = 2;
 191         return "\0";
 192     }
 193 };
 194
 195 // ----------------------------------------------------------------------------
 196 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 197 // ----------------------------------------------------------------------------
 198
 199 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base
 200 {
 201 public:
 202     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 203     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 204 };
 205
 206 // ----------------------------------------------------------------------------
 207 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 208 // ----------------------------------------------------------------------------
 209
 210 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base
 211 {
 212 public:
 213     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 214     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 215 };
 216
 217 // ----------------------------------------------------------------------------
 218 // wxMBConvUTF32Base: base class for both LE and BE variants
 219 // ----------------------------------------------------------------------------
 220
 221 class WXDLLIMPEXP_BASE wxMBConvUTF32Base : public wxMBConv
 222 {
 223 private:
 224     virtual const char *GetMBNul(size_t *nulLen) const
 225     {
 226         *nulLen = 4;
 227         return "\0\0\0";
 228     }
 229 };
 230
 231 // ----------------------------------------------------------------------------
 232 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
 233 // ----------------------------------------------------------------------------
 234
 235 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base
 236 {
 237 public:
 238     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 239     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 240 };
 241
 242 // ----------------------------------------------------------------------------
 243 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
 244 // ----------------------------------------------------------------------------
 245
 246 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base
 247 {
 248 public:
 249     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 250     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 251 };
 252
 253 // ----------------------------------------------------------------------------
 254 // wxCSConv (for conversion based on loadable char sets)
 255 // ----------------------------------------------------------------------------
 256
 257 #include "wx/fontenc.h"
 258
 259 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 260 {
 261 public:
 262     // we can be created either from charset name or from an encoding constant
 263     // but we can't have both at once
 264     wxCSConv(const wxChar *charset);
 265     wxCSConv(wxFontEncoding encoding);
 266
 267     wxCSConv(const wxCSConv& conv);
 268     virtual ~wxCSConv();
 269
 270     wxCSConv& operator=(const wxCSConv& conv);
 271
 272     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 273     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 274
 275     void Clear() ;
 276
 277 private:
 278     // common part of all ctors
 279     void Init();
 280
 281     // creates m_convReal if necessary
 282     void CreateConvIfNeeded() const;
 283
 284     // do create m_convReal (unconditionally)
 285     wxMBConv *DoCreate() const;
 286
 287     // set the name (may be only called when m_name == NULL), makes copy of
 288     // charset string
 289     void SetName(const wxChar *charset);
 290
 291     virtual const char *GetMBNul(size_t *nulLen) const;
 292
 293
 294     // note that we can't use wxString here because of compilation
 295     // dependencies: we're included from wx/string.h
 296     wxChar *m_name;
 297     wxFontEncoding m_encoding;
 298
 299     // use CreateConvIfNeeded() before accessing m_convReal!
 300     wxMBConv *m_convReal;
 301     bool m_deferred;
 302 };
 303
 304
 305 // ----------------------------------------------------------------------------
 306 // declare predefined conversion objects
 307 // ----------------------------------------------------------------------------
 308
 309 // conversion to be used with all standard functions affected by locale, e.g.
 310 // strtol(), strftime(), ...
 311 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 312
 313 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 314 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 315 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 316 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 317
 318 // conversion used for the file names on the systems where they're not Unicode
 319 // (basically anything except Windows)
 320 //
 321 // this is used by all file functions, can be changed by the application
 322 //
 323 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 324 // under Windows normally)
 325 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 326
 327 // backwards compatible define
 328 #define wxConvFile (*wxConvFileName)
 329
 330 // the current conversion object, may be set to any conversion, is used by
 331 // default in a couple of places inside wx (initially same as wxConvLibc)
 332 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 333
 334 // ???
 335 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 336
 337
 338 // ----------------------------------------------------------------------------
 339 // endianness-dependent conversions
 340 // ----------------------------------------------------------------------------
 341
 342 #ifdef WORDS_BIGENDIAN
 343     typedef wxMBConvUTF16BE wxMBConvUTF16;
 344     typedef wxMBConvUTF32BE wxMBConvUTF32;
 345 #else
 346     typedef wxMBConvUTF16LE wxMBConvUTF16;
 347     typedef wxMBConvUTF32LE wxMBConvUTF32;
 348 #endif
 349
 350 // ----------------------------------------------------------------------------
 351 // filename conversion macros
 352 // ----------------------------------------------------------------------------
 353
 354 // filenames are multibyte on Unix and probably widechar on Windows?
 355 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 356     #define wxMBFILES 1
 357 #else
 358     #define wxMBFILES 0
 359 #endif
 360
 361 #if wxMBFILES && wxUSE_UNICODE
 362     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 363     #define wxFNSTRINGCAST wxMBSTRINGCAST
 364 #else
 365 #if defined( __WXOSX__ ) && wxMBFILES
 366     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 367 #else
 368     #define wxFNCONV(name) name
 369 #endif
 370     #define wxFNSTRINGCAST WXSTRINGCAST
 371 #endif
 372
 373 #else // !wxUSE_WCHAR_T
 374
 375 // ----------------------------------------------------------------------------
 376 // stand-ins in absence of wchar_t
 377 // ----------------------------------------------------------------------------
 378
 379 class WXDLLIMPEXP_BASE wxMBConv
 380 {
 381 public:
 382     const char* cMB2WX(const char *psz) const { return psz; }
 383     const char* cWX2MB(const char *psz) const { return psz; }
 384 };
 385
 386 #define wxConvFile wxConvLocal
 387
 388 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 389                                        wxConvLocal,
 390                                        wxConvISO8859_1,
 391                                        wxConvUTF8;
 392 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 393
 394 #define wxFNCONV(name) name
 395 #define wxFNSTRINGCAST WXSTRINGCAST
 396
 397 #endif
 398   // wxUSE_WCHAR_T
 399
 400 // ----------------------------------------------------------------------------
 401 // macros for the most common conversions
 402 // ----------------------------------------------------------------------------
 403
 404 #if wxUSE_UNICODE
 405     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 406     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 407 #else // ANSI
 408     // no conversions to do
 409     #define wxConvertWX2MB(s)   (s)
 410     #define wxConvertMB2WX(s)   (s)
 411 #endif // Unicode/ANSI
 412
 413 #endif
 414   // _WX_WXSTRCONVH__
 415