include/wx/strconv.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        strconv.h
   3 // Purpose:     conversion routines for char sets any Unicode
   4 // Author:      Ove Kaaven, Robert Roebling, Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Ove Kaaven, Robert Roebling
   9 //              (c) 1998-2006 Vadim Zeitlin
  10 // Licence:     wxWindows licence
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 #ifndef _WX_STRCONV_H_
  14 #define _WX_STRCONV_H_
  15
  16 #include "wx/defs.h"
  17 #include "wx/wxchar.h"
  18 #include "wx/buffer.h"
  19
  20 #ifdef __DIGITALMARS__
  21 #include "typeinfo.h"
  22 #endif
  23
  24 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  25 #  undef __BSEXCPT__
  26 #endif
  27
  28 #include <stdlib.h>
  29
  30 #if wxUSE_WCHAR_T
  31
  32 // the error value returned by wxMBConv methods
  33 #define wxCONV_FAILED ((size_t)-1)
  34
  35 // the default value for some length parameters meaning that the string is
  36 // NUL-terminated
  37 #define wxNO_LEN ((size_t)-1)
  38
  39 // ----------------------------------------------------------------------------
  40 // wxMBConv (abstract base class for conversions)
  41 // ----------------------------------------------------------------------------
  42
  43 // When deriving a new class from wxMBConv you must reimplement ToWChar() and
  44 // FromWChar() methods which are not pure virtual only for historical reasons,
  45 // don't let the fact that the existing classes implement MB2WC/WC2MB() instead
  46 // confuse you.
  47 //
  48 // You also have to implement Clone() to allow copying the conversions
  49 // polymorphically.
  50 //
  51 // And you might need to override GetMBNulLen() as well.
  52 class WXDLLIMPEXP_BASE wxMBConv
  53 {
  54 public:
  55     // The functions doing actual conversion from/to narrow to/from wide
  56     // character strings.
  57     //
  58     // On success, the return value is the length (i.e. the number of
  59     // characters, not bytes) of the converted string including any trailing
  60     // L'\0' or (possibly multiple) '\0'(s). If the conversion fails or if
  61     // there is not enough space for everything, including the trailing NUL
  62     // character(s), in the output buffer, wxCONV_FAILED is returned.
  63     //
  64     // In the special case when dstLen is 0 (outputBuf may be NULL then) the
  65     // return value is the length of the needed buffer but nothing happens
  66     // otherwise. If srcLen is wxNO_LEN, the entire string, up to and
  67     // including the trailing NUL(s), is converted, otherwise exactly srcLen
  68     // bytes are.
  69     //
  70     // Typical usage:
  71     //
  72     //          size_t dstLen = conv.ToWChar(NULL, 0, src);
  73     //          if ( dstLen != wxCONV_FAILED )
  74     //              ... handle error ...
  75     //          wchar_t *wbuf = new wchar_t[dstLen];
  76     //          conv.ToWChar(wbuf, dstLen, src);
  77     //
  78     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
  79                            const char *src, size_t srcLen = wxNO_LEN) const;
  80
  81     virtual size_t FromWChar(char *dst, size_t dstLen,
  82                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
  83
  84
  85     // Convenience functions for translating NUL-terminated strings: returns
  86     // the buffer containing the converted string or NULL pointer if the
  87     // conversion failed.
  88     const wxWCharBuffer cMB2WC(const char *in) const;
  89     const wxCharBuffer cWC2MB(const wchar_t *in) const;
  90
  91     // Convenience functions for converting strings which may contain embedded
  92     // NULs and don't have to be NUL-terminated.
  93     //
  94     // inLen is the length of the buffer including trailing NUL if any: if the
  95     // last 4 bytes of the buffer are all NULs, these functions are more
  96     // efficient as they avoid copying the string, but otherwise a copy is made
  97     // internally which could be quite bad for (very) long strings.
  98     //
  99     // outLen receives, if not NULL, the length of the converted string or 0 if
 100     // the conversion failed (returning 0 and not -1 in this case makes it
 101     // difficult to distinguish between failed conversion and empty input but
 102     // this is done for backwards compatibility)
 103     const wxWCharBuffer
 104         cMB2WC(const char *in, size_t inLen, size_t *outLen) const;
 105     const wxCharBuffer
 106         cWC2MB(const wchar_t *in, size_t inLen, size_t *outLen) const;
 107
 108     // convenience functions for converting MB or WC to/from wxWin default
 109 #if wxUSE_UNICODE
 110     const wxWCharBuffer cMB2WX(const char *psz) const { return cMB2WC(psz); }
 111     const wxCharBuffer cWX2MB(const wchar_t *psz) const { return cWC2MB(psz); }
 112     const wchar_t* cWC2WX(const wchar_t *psz) const { return psz; }
 113     const wchar_t* cWX2WC(const wchar_t *psz) const { return psz; }
 114 #else // ANSI
 115     const char* cMB2WX(const char *psz) const { return psz; }
 116     const char* cWX2MB(const char *psz) const { return psz; }
 117     const wxCharBuffer cWC2WX(const wchar_t *psz) const { return cWC2MB(psz); }
 118     const wxWCharBuffer cWX2WC(const char *psz) const { return cMB2WC(psz); }
 119 #endif // Unicode/ANSI
 120
 121     // this function is used in the implementation of cMB2WC() to distinguish
 122     // between the following cases:
 123     //
 124     //      a) var width encoding with strings terminated by a single NUL
 125     //         (usual multibyte encodings): return 1 in this case
 126     //      b) fixed width encoding with 2 bytes/char and so terminated by
 127     //         2 NULs (UTF-16/UCS-2 and variants): return 2 in this case
 128     //      c) fixed width encoding with 4 bytes/char and so terminated by
 129     //         4 NULs (UTF-32/UCS-4 and variants): return 4 in this case
 130     //
 131     // anything else is not supported currently and -1 should be returned
 132     virtual size_t GetMBNulLen() const { return 1; }
 133
 134     // return the maximal value currently returned by GetMBNulLen() for any
 135     // encoding
 136     static size_t GetMaxMBNulLen() { return 4 /* for UTF-32 */; }
 137
 138
 139     // The old conversion functions. The existing classes currently mostly
 140     // implement these ones but we're in transition to using To/FromWChar()
 141     // instead and any new classes should implement just the new functions.
 142     // For now, however, we provide default implementation of To/FromWChar() in
 143     // this base class in terms of MB2WC/WC2MB() to avoid having to rewrite all
 144     // the conversions at once.
 145     //
 146     // On success, the return value is the length (i.e. the number of
 147     // characters, not bytes) not counting the trailing NUL(s) of the converted
 148     // string. On failure, (size_t)-1 is returned. In the special case when
 149     // outputBuf is NULL the return value is the same one but nothing is
 150     // written to the buffer.
 151     //
 152     // Note that outLen is the length of the output buffer, not the length of
 153     // the input (which is always supposed to be terminated by one or more
 154     // NULs, as appropriate for the encoding)!
 155     virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const;
 156     virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const;
 157
 158
 159     // make a heap-allocated copy of this object
 160     virtual wxMBConv *Clone() const = 0;
 161
 162     // virtual dtor for any base class
 163     virtual ~wxMBConv();
 164 };
 165
 166 // ----------------------------------------------------------------------------
 167 // wxMBConvLibc uses standard mbstowcs() and wcstombs() functions for
 168 //              conversion (hence it depends on the current locale)
 169 // ----------------------------------------------------------------------------
 170
 171 class WXDLLIMPEXP_BASE wxMBConvLibc : public wxMBConv
 172 {
 173 public:
 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     virtual wxMBConv *Clone() const { return new wxMBConvLibc; }
 178 };
 179
 180 #ifdef __UNIX__
 181
 182 // ----------------------------------------------------------------------------
 183 // wxConvBrokenFileNames is made for Unix in Unicode mode when
 184 // files are accidentally written in an encoding which is not
 185 // the system encoding. Typically, the system encoding will be
 186 // UTF8 but there might be files stored in ISO8859-1 on disk.
 187 // ----------------------------------------------------------------------------
 188
 189 class WXDLLIMPEXP_BASE wxConvBrokenFileNames : public wxMBConv
 190 {
 191 public:
 192     wxConvBrokenFileNames(const wxChar *charset);
 193     wxConvBrokenFileNames(const wxConvBrokenFileNames& conv)
 194         : m_conv(conv.m_conv ? conv.m_conv->Clone() : NULL)
 195     {
 196     }
 197     virtual ~wxConvBrokenFileNames() { delete m_conv; }
 198
 199     virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const
 200     {
 201         return m_conv->MB2WC(out, in, outLen);
 202     }
 203
 204     virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const
 205     {
 206         return m_conv->WC2MB(out, in, outLen);
 207     }
 208
 209     virtual size_t GetMBNulLen() const
 210     {
 211         // cast needed to call a private function
 212         return m_conv->GetMBNulLen();
 213     }
 214
 215     virtual wxMBConv *Clone() const { return new wxConvBrokenFileNames(*this); }
 216
 217 private:
 218     // the conversion object we forward to
 219     wxMBConv *m_conv;
 220
 221     DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames)
 222 };
 223
 224 #endif // __UNIX__
 225
 226 // ----------------------------------------------------------------------------
 227 // wxMBConvUTF7 (for conversion using UTF7 encoding)
 228 // ----------------------------------------------------------------------------
 229
 230 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 231 {
 232 public:
 233     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 234     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 235
 236     virtual wxMBConv *Clone() const { return new wxMBConvUTF7; }
 237 };
 238
 239 // ----------------------------------------------------------------------------
 240 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 241 // ----------------------------------------------------------------------------
 242
 243 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 244 {
 245 public:
 246     enum
 247     {
 248         MAP_INVALID_UTF8_NOT = 0,
 249         MAP_INVALID_UTF8_TO_PUA = 1,
 250         MAP_INVALID_UTF8_TO_OCTAL = 2
 251     };
 252
 253     wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
 254     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 255     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 256
 257     virtual wxMBConv *Clone() const { return new wxMBConvUTF8(m_options); }
 258
 259 private:
 260     int m_options;
 261 };
 262
 263 // ----------------------------------------------------------------------------
 264 // wxMBConvUTF16Base: for both LE and BE variants
 265 // ----------------------------------------------------------------------------
 266
 267 class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv
 268 {
 269 public:
 270     enum { BYTES_PER_CHAR = 2 };
 271
 272     virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
 273
 274 protected:
 275     // return the length of the buffer using srcLen if it's not wxNO_LEN and
 276     // computing the length ourselves if it is; also checks that the length is
 277     // even if specified as we need an entire number of UTF-16 characters and
 278     // returns wxNO_LEN which indicates error if it is odd
 279     static size_t GetLength(const char *src, size_t srcLen);
 280 };
 281
 282 // ----------------------------------------------------------------------------
 283 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 284 // ----------------------------------------------------------------------------
 285
 286 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base
 287 {
 288 public:
 289 #if SIZEOF_WCHAR_T == 2
 290     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 291                            const char *src, size_t srcLen = wxNO_LEN) const;
 292     virtual size_t FromWChar(char *dst, size_t dstLen,
 293                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 294 #else
 295     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 296     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 297 #endif
 298     virtual wxMBConv *Clone() const { return new wxMBConvUTF16LE; }
 299 };
 300
 301 // ----------------------------------------------------------------------------
 302 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 303 // ----------------------------------------------------------------------------
 304
 305 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base
 306 {
 307 public:
 308 #if SIZEOF_WCHAR_T == 2
 309     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 310                            const char *src, size_t srcLen = wxNO_LEN) const;
 311     virtual size_t FromWChar(char *dst, size_t dstLen,
 312                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 313 #else
 314     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 315     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 316 #endif
 317     virtual wxMBConv *Clone() const { return new wxMBConvUTF16BE; }
 318 };
 319
 320 // ----------------------------------------------------------------------------
 321 // wxMBConvUTF32Base: base class for both LE and BE variants
 322 // ----------------------------------------------------------------------------
 323
 324 class WXDLLIMPEXP_BASE wxMBConvUTF32Base : public wxMBConv
 325 {
 326 public:
 327     enum { BYTES_PER_CHAR = 4 };
 328
 329     virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
 330
 331 protected:
 332     // this is similar to wxMBConvUTF16Base method with the same name except
 333     // that, of course, it verifies that length is divisible by 4 if given and
 334     // not by 2
 335     static size_t GetLength(const char *src, size_t srcLen);
 336 };
 337
 338 // ----------------------------------------------------------------------------
 339 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
 340 // ----------------------------------------------------------------------------
 341
 342 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base
 343 {
 344 public:
 345 #if SIZEOF_WCHAR_T == 2
 346     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 347                            const char *src, size_t srcLen = wxNO_LEN) const;
 348     virtual size_t FromWChar(char *dst, size_t dstLen,
 349                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 350 #else
 351     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 352     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 353 #endif
 354     virtual wxMBConv *Clone() const { return new wxMBConvUTF32LE; }
 355 };
 356
 357 // ----------------------------------------------------------------------------
 358 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
 359 // ----------------------------------------------------------------------------
 360
 361 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base
 362 {
 363 public:
 364 #if SIZEOF_WCHAR_T == 2
 365     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 366                            const char *src, size_t srcLen = wxNO_LEN) const;
 367     virtual size_t FromWChar(char *dst, size_t dstLen,
 368                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 369 #else
 370     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 371     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 372 #endif
 373     virtual wxMBConv *Clone() const { return new wxMBConvUTF32BE; }
 374 };
 375
 376 // ----------------------------------------------------------------------------
 377 // wxCSConv (for conversion based on loadable char sets)
 378 // ----------------------------------------------------------------------------
 379
 380 #include "wx/fontenc.h"
 381
 382 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 383 {
 384 public:
 385     // we can be created either from charset name or from an encoding constant
 386     // but we can't have both at once
 387     wxCSConv(const wxChar *charset);
 388     wxCSConv(wxFontEncoding encoding);
 389
 390     wxCSConv(const wxCSConv& conv);
 391     virtual ~wxCSConv();
 392
 393     wxCSConv& operator=(const wxCSConv& conv);
 394
 395     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 396     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 397     virtual size_t GetMBNulLen() const;
 398     virtual wxMBConv *Clone() const { return new wxCSConv(*this); }
 399
 400     void Clear();
 401
 402 private:
 403     // common part of all ctors
 404     void Init();
 405
 406     // creates m_convReal if necessary
 407     void CreateConvIfNeeded() const;
 408
 409     // do create m_convReal (unconditionally)
 410     wxMBConv *DoCreate() const;
 411
 412     // set the name (may be only called when m_name == NULL), makes copy of
 413     // charset string
 414     void SetName(const wxChar *charset);
 415
 416
 417     // note that we can't use wxString here because of compilation
 418     // dependencies: we're included from wx/string.h
 419     wxChar *m_name;
 420     wxFontEncoding m_encoding;
 421
 422     // use CreateConvIfNeeded() before accessing m_convReal!
 423     wxMBConv *m_convReal;
 424     bool m_deferred;
 425 };
 426
 427
 428 // ----------------------------------------------------------------------------
 429 // declare predefined conversion objects
 430 // ----------------------------------------------------------------------------
 431
 432 // conversion to be used with all standard functions affected by locale, e.g.
 433 // strtol(), strftime(), ...
 434 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 435
 436 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 437 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 438 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 439 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 440
 441 // conversion used for the file names on the systems where they're not Unicode
 442 // (basically anything except Windows)
 443 //
 444 // this is used by all file functions, can be changed by the application
 445 //
 446 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 447 // under Windows normally)
 448 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 449
 450 // backwards compatible define
 451 #define wxConvFile (*wxConvFileName)
 452
 453 // the current conversion object, may be set to any conversion, is used by
 454 // default in a couple of places inside wx (initially same as wxConvLibc)
 455 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 456
 457 // ???
 458 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 459
 460
 461 // ----------------------------------------------------------------------------
 462 // endianness-dependent conversions
 463 // ----------------------------------------------------------------------------
 464
 465 #ifdef WORDS_BIGENDIAN
 466     typedef wxMBConvUTF16BE wxMBConvUTF16;
 467     typedef wxMBConvUTF32BE wxMBConvUTF32;
 468 #else
 469     typedef wxMBConvUTF16LE wxMBConvUTF16;
 470     typedef wxMBConvUTF32LE wxMBConvUTF32;
 471 #endif
 472
 473 // ----------------------------------------------------------------------------
 474 // filename conversion macros
 475 // ----------------------------------------------------------------------------
 476
 477 // filenames are multibyte on Unix and probably widechar on Windows?
 478 #if defined(__UNIX__) || defined(__BORLANDC__) || defined(__WXMAC__ )
 479     #define wxMBFILES 1
 480 #else
 481     #define wxMBFILES 0
 482 #endif
 483
 484 #if wxMBFILES && wxUSE_UNICODE
 485     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 486     #define wxFNSTRINGCAST wxMBSTRINGCAST
 487 #else
 488 #if defined( __WXOSX__ ) && wxMBFILES
 489     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 490 #else
 491     #define wxFNCONV(name) name
 492 #endif
 493     #define wxFNSTRINGCAST WXSTRINGCAST
 494 #endif
 495
 496 #else // !wxUSE_WCHAR_T
 497
 498 // ----------------------------------------------------------------------------
 499 // stand-ins in absence of wchar_t
 500 // ----------------------------------------------------------------------------
 501
 502 class WXDLLIMPEXP_BASE wxMBConv
 503 {
 504 public:
 505     const char* cMB2WX(const char *psz) const { return psz; }
 506     const char* cWX2MB(const char *psz) const { return psz; }
 507 };
 508
 509 #define wxConvFile wxConvLocal
 510
 511 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 512                                        wxConvLocal,
 513                                        wxConvISO8859_1,
 514                                        wxConvUTF8;
 515 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 516
 517 #define wxFNCONV(name) name
 518 #define wxFNSTRINGCAST WXSTRINGCAST
 519
 520 #endif
 521   // wxUSE_WCHAR_T
 522
 523 // ----------------------------------------------------------------------------
 524 // macros for the most common conversions
 525 // ----------------------------------------------------------------------------
 526
 527 #if wxUSE_UNICODE
 528     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 529     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 530 #else // ANSI
 531     // no conversions to do
 532     #define wxConvertWX2MB(s)   (s)
 533     #define wxConvertMB2WX(s)   (s)
 534 #endif // Unicode/ANSI
 535
 536 #endif // _WX_STRCONV_H_
 537