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 const char *GetMBNul(size_t *nulLen) const
 138     {
 139         // cast needed to call a private function
 140         return ((wxConvBrokenFileNames *)m_conv)->GetMBNul(nulLen);
 141     }
 142
 143
 144     // the conversion object we forward to
 145     wxMBConv *m_conv;
 146 };
 147
 148 #endif // __UNIX__
 149
 150 // ----------------------------------------------------------------------------
 151 // wxMBConvUTF7 (for conversion using UTF7 encoding)
 152 // ----------------------------------------------------------------------------
 153
 154 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 155 {
 156 public:
 157     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 158     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 159 };
 160
 161 // ----------------------------------------------------------------------------
 162 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 163 // ----------------------------------------------------------------------------
 164
 165 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 166 {
 167 public:
 168     enum {
 169         MAP_INVALID_UTF8_NOT = 0,
 170         MAP_INVALID_UTF8_TO_PUA = 1,
 171         MAP_INVALID_UTF8_TO_OCTAL = 2
 172     };
 173
 174     wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
 175     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 176     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 177
 178 private:
 179     int m_options;
 180 };
 181
 182 // ----------------------------------------------------------------------------
 183 // wxMBConvUTF16Base: for both LE and BE variants
 184 // ----------------------------------------------------------------------------
 185
 186 class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv
 187 {
 188 private:
 189     virtual const char *GetMBNul(size_t *nulLen) const
 190     {
 191         *nulLen = 2;
 192         return "\0";
 193     }
 194 };
 195
 196 // ----------------------------------------------------------------------------
 197 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 198 // ----------------------------------------------------------------------------
 199
 200 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base
 201 {
 202 public:
 203     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 204     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 205 };
 206
 207 // ----------------------------------------------------------------------------
 208 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 209 // ----------------------------------------------------------------------------
 210
 211 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base
 212 {
 213 public:
 214     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 215     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 216 };
 217
 218 // ----------------------------------------------------------------------------
 219 // wxMBConvUTF32Base: base class for both LE and BE variants
 220 // ----------------------------------------------------------------------------
 221
 222 class WXDLLIMPEXP_BASE wxMBConvUTF32Base : public wxMBConv
 223 {
 224 private:
 225     virtual const char *GetMBNul(size_t *nulLen) const
 226     {
 227         *nulLen = 4;
 228         return "\0\0\0";
 229     }
 230 };
 231
 232 // ----------------------------------------------------------------------------
 233 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
 234 // ----------------------------------------------------------------------------
 235
 236 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base
 237 {
 238 public:
 239     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 240     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 241 };
 242
 243 // ----------------------------------------------------------------------------
 244 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
 245 // ----------------------------------------------------------------------------
 246
 247 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base
 248 {
 249 public:
 250     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 251     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 252 };
 253
 254 // ----------------------------------------------------------------------------
 255 // wxCSConv (for conversion based on loadable char sets)
 256 // ----------------------------------------------------------------------------
 257
 258 #include "wx/fontenc.h"
 259
 260 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 261 {
 262 public:
 263     // we can be created either from charset name or from an encoding constant
 264     // but we can't have both at once
 265     wxCSConv(const wxChar *charset);
 266     wxCSConv(wxFontEncoding encoding);
 267
 268     wxCSConv(const wxCSConv& conv);
 269     virtual ~wxCSConv();
 270
 271     wxCSConv& operator=(const wxCSConv& conv);
 272
 273     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 274     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 275
 276     void Clear() ;
 277
 278 private:
 279     // common part of all ctors
 280     void Init();
 281
 282     // creates m_convReal if necessary
 283     void CreateConvIfNeeded() const;
 284
 285     // do create m_convReal (unconditionally)
 286     wxMBConv *DoCreate() const;
 287
 288     // set the name (may be only called when m_name == NULL), makes copy of
 289     // charset string
 290     void SetName(const wxChar *charset);
 291
 292     virtual const char *GetMBNul(size_t *nulLen) const;
 293
 294
 295     // note that we can't use wxString here because of compilation
 296     // dependencies: we're included from wx/string.h
 297     wxChar *m_name;
 298     wxFontEncoding m_encoding;
 299
 300     // use CreateConvIfNeeded() before accessing m_convReal!
 301     wxMBConv *m_convReal;
 302     bool m_deferred;
 303 };
 304
 305
 306 // ----------------------------------------------------------------------------
 307 // declare predefined conversion objects
 308 // ----------------------------------------------------------------------------
 309
 310 // conversion to be used with all standard functions affected by locale, e.g.
 311 // strtol(), strftime(), ...
 312 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 313
 314 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 315 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 316 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 317 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 318
 319 // conversion used for the file names on the systems where they're not Unicode
 320 // (basically anything except Windows)
 321 //
 322 // this is used by all file functions, can be changed by the application
 323 //
 324 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 325 // under Windows normally)
 326 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 327
 328 // backwards compatible define
 329 #define wxConvFile (*wxConvFileName)
 330
 331 // the current conversion object, may be set to any conversion, is used by
 332 // default in a couple of places inside wx (initially same as wxConvLibc)
 333 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 334
 335 // ???
 336 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 337
 338
 339 // ----------------------------------------------------------------------------
 340 // endianness-dependent conversions
 341 // ----------------------------------------------------------------------------
 342
 343 #ifdef WORDS_BIGENDIAN
 344     typedef wxMBConvUTF16BE wxMBConvUTF16;
 345     typedef wxMBConvUTF32BE wxMBConvUTF32;
 346 #else
 347     typedef wxMBConvUTF16LE wxMBConvUTF16;
 348     typedef wxMBConvUTF32LE wxMBConvUTF32;
 349 #endif
 350
 351 // ----------------------------------------------------------------------------
 352 // filename conversion macros
 353 // ----------------------------------------------------------------------------
 354
 355 // filenames are multibyte on Unix and probably widechar on Windows?
 356 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 357     #define wxMBFILES 1
 358 #else
 359     #define wxMBFILES 0
 360 #endif
 361
 362 #if wxMBFILES && wxUSE_UNICODE
 363     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 364     #define wxFNSTRINGCAST wxMBSTRINGCAST
 365 #else
 366 #if defined( __WXOSX__ ) && wxMBFILES
 367     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 368 #else
 369     #define wxFNCONV(name) name
 370 #endif
 371     #define wxFNSTRINGCAST WXSTRINGCAST
 372 #endif
 373
 374 #else // !wxUSE_WCHAR_T
 375
 376 // ----------------------------------------------------------------------------
 377 // stand-ins in absence of wchar_t
 378 // ----------------------------------------------------------------------------
 379
 380 class WXDLLIMPEXP_BASE wxMBConv
 381 {
 382 public:
 383     const char* cMB2WX(const char *psz) const { return psz; }
 384     const char* cWX2MB(const char *psz) const { return psz; }
 385 };
 386
 387 #define wxConvFile wxConvLocal
 388
 389 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 390                                        wxConvLocal,
 391                                        wxConvISO8859_1,
 392                                        wxConvUTF8;
 393 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 394
 395 #define wxFNCONV(name) name
 396 #define wxFNSTRINGCAST WXSTRINGCAST
 397
 398 #endif
 399   // wxUSE_WCHAR_T
 400
 401 // ----------------------------------------------------------------------------
 402 // macros for the most common conversions
 403 // ----------------------------------------------------------------------------
 404
 405 #if wxUSE_UNICODE
 406     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 407     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 408 #else // ANSI
 409     // no conversions to do
 410     #define wxConvertWX2MB(s)   (s)
 411     #define wxConvertMB2WX(s)   (s)
 412 #endif // Unicode/ANSI
 413
 414 #endif
 415   // _WX_WXSTRCONVH__
 416