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         : wxMBConv(),
 195           m_conv(conv.m_conv ? conv.m_conv->Clone() : NULL)
 196     {
 197     }
 198     virtual ~wxConvBrokenFileNames() { delete m_conv; }
 199
 200     virtual size_t MB2WC(wchar_t *out, const char *in, size_t outLen) const
 201     {
 202         return m_conv->MB2WC(out, in, outLen);
 203     }
 204
 205     virtual size_t WC2MB(char *out, const wchar_t *in, size_t outLen) const
 206     {
 207         return m_conv->WC2MB(out, in, outLen);
 208     }
 209
 210     virtual size_t GetMBNulLen() const
 211     {
 212         // cast needed to call a private function
 213         return m_conv->GetMBNulLen();
 214     }
 215
 216     virtual wxMBConv *Clone() const { return new wxConvBrokenFileNames(*this); }
 217
 218 private:
 219     // the conversion object we forward to
 220     wxMBConv *m_conv;
 221
 222     DECLARE_NO_ASSIGN_CLASS(wxConvBrokenFileNames)
 223 };
 224
 225 #endif // __UNIX__
 226
 227 // ----------------------------------------------------------------------------
 228 // wxMBConvUTF7 (for conversion using UTF7 encoding)
 229 // ----------------------------------------------------------------------------
 230
 231 class WXDLLIMPEXP_BASE wxMBConvUTF7 : public wxMBConv
 232 {
 233 public:
 234     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 235     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 236
 237     virtual wxMBConv *Clone() const { return new wxMBConvUTF7; }
 238 };
 239
 240 // ----------------------------------------------------------------------------
 241 // wxMBConvUTF8 (for conversion using UTF8 encoding)
 242 // ----------------------------------------------------------------------------
 243
 244 class WXDLLIMPEXP_BASE wxMBConvUTF8 : public wxMBConv
 245 {
 246 public:
 247     enum
 248     {
 249         MAP_INVALID_UTF8_NOT = 0,
 250         MAP_INVALID_UTF8_TO_PUA = 1,
 251         MAP_INVALID_UTF8_TO_OCTAL = 2
 252     };
 253
 254     wxMBConvUTF8(int options = MAP_INVALID_UTF8_NOT) : m_options(options) { }
 255     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 256     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 257
 258     virtual wxMBConv *Clone() const { return new wxMBConvUTF8(m_options); }
 259
 260 private:
 261     int m_options;
 262 };
 263
 264 // ----------------------------------------------------------------------------
 265 // wxMBConvUTF16Base: for both LE and BE variants
 266 // ----------------------------------------------------------------------------
 267
 268 class WXDLLIMPEXP_BASE wxMBConvUTF16Base : public wxMBConv
 269 {
 270 public:
 271     enum { BYTES_PER_CHAR = 2 };
 272
 273     virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
 274
 275 protected:
 276     // return the length of the buffer using srcLen if it's not wxNO_LEN and
 277     // computing the length ourselves if it is; also checks that the length is
 278     // even if specified as we need an entire number of UTF-16 characters and
 279     // returns wxNO_LEN which indicates error if it is odd
 280     static size_t GetLength(const char *src, size_t srcLen);
 281 };
 282
 283 // ----------------------------------------------------------------------------
 284 // wxMBConvUTF16LE (for conversion using UTF16 Little Endian encoding)
 285 // ----------------------------------------------------------------------------
 286
 287 class WXDLLIMPEXP_BASE wxMBConvUTF16LE : public wxMBConvUTF16Base
 288 {
 289 public:
 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     virtual wxMBConv *Clone() const { return new wxMBConvUTF16LE; }
 295 };
 296
 297 // ----------------------------------------------------------------------------
 298 // wxMBConvUTF16BE (for conversion using UTF16 Big Endian encoding)
 299 // ----------------------------------------------------------------------------
 300
 301 class WXDLLIMPEXP_BASE wxMBConvUTF16BE : public wxMBConvUTF16Base
 302 {
 303 public:
 304     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 305                            const char *src, size_t srcLen = wxNO_LEN) const;
 306     virtual size_t FromWChar(char *dst, size_t dstLen,
 307                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 308     virtual wxMBConv *Clone() const { return new wxMBConvUTF16BE; }
 309 };
 310
 311 // ----------------------------------------------------------------------------
 312 // wxMBConvUTF32Base: base class for both LE and BE variants
 313 // ----------------------------------------------------------------------------
 314
 315 class WXDLLIMPEXP_BASE wxMBConvUTF32Base : public wxMBConv
 316 {
 317 public:
 318     enum { BYTES_PER_CHAR = 4 };
 319
 320     virtual size_t GetMBNulLen() const { return BYTES_PER_CHAR; }
 321
 322 protected:
 323     // this is similar to wxMBConvUTF16Base method with the same name except
 324     // that, of course, it verifies that length is divisible by 4 if given and
 325     // not by 2
 326     static size_t GetLength(const char *src, size_t srcLen);
 327 };
 328
 329 // ----------------------------------------------------------------------------
 330 // wxMBConvUTF32LE (for conversion using UTF32 Little Endian encoding)
 331 // ----------------------------------------------------------------------------
 332
 333 class WXDLLIMPEXP_BASE wxMBConvUTF32LE : public wxMBConvUTF32Base
 334 {
 335 public:
 336     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 337                            const char *src, size_t srcLen = wxNO_LEN) const;
 338     virtual size_t FromWChar(char *dst, size_t dstLen,
 339                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 340     virtual wxMBConv *Clone() const { return new wxMBConvUTF32LE; }
 341 };
 342
 343 // ----------------------------------------------------------------------------
 344 // wxMBConvUTF32BE (for conversion using UTF32 Big Endian encoding)
 345 // ----------------------------------------------------------------------------
 346
 347 class WXDLLIMPEXP_BASE wxMBConvUTF32BE : public wxMBConvUTF32Base
 348 {
 349 public:
 350     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 351                            const char *src, size_t srcLen = wxNO_LEN) const;
 352     virtual size_t FromWChar(char *dst, size_t dstLen,
 353                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 354     virtual wxMBConv *Clone() const { return new wxMBConvUTF32BE; }
 355 };
 356
 357 // ----------------------------------------------------------------------------
 358 // wxCSConv (for conversion based on loadable char sets)
 359 // ----------------------------------------------------------------------------
 360
 361 #include "wx/fontenc.h"
 362
 363 class WXDLLIMPEXP_BASE wxCSConv : public wxMBConv
 364 {
 365 public:
 366     // we can be created either from charset name or from an encoding constant
 367     // but we can't have both at once
 368     wxCSConv(const wxChar *charset);
 369     wxCSConv(wxFontEncoding encoding);
 370
 371     wxCSConv(const wxCSConv& conv);
 372     virtual ~wxCSConv();
 373
 374     wxCSConv& operator=(const wxCSConv& conv);
 375
 376     virtual size_t ToWChar(wchar_t *dst, size_t dstLen,
 377                            const char *src, size_t srcLen = wxNO_LEN) const;
 378     virtual size_t FromWChar(char *dst, size_t dstLen,
 379                              const wchar_t *src, size_t srcLen = wxNO_LEN) const;
 380     virtual size_t MB2WC(wchar_t *outputBuf, const char *psz, size_t outputSize) const;
 381     virtual size_t WC2MB(char *outputBuf, const wchar_t *psz, size_t outputSize) const;
 382     virtual size_t GetMBNulLen() const;
 383
 384     virtual wxMBConv *Clone() const { return new wxCSConv(*this); }
 385
 386     void Clear();
 387
 388 private:
 389     // common part of all ctors
 390     void Init();
 391
 392     // creates m_convReal if necessary
 393     void CreateConvIfNeeded() const;
 394
 395     // do create m_convReal (unconditionally)
 396     wxMBConv *DoCreate() const;
 397
 398     // set the name (may be only called when m_name == NULL), makes copy of
 399     // charset string
 400     void SetName(const wxChar *charset);
 401
 402
 403     // note that we can't use wxString here because of compilation
 404     // dependencies: we're included from wx/string.h
 405     wxChar *m_name;
 406     wxFontEncoding m_encoding;
 407
 408     // use CreateConvIfNeeded() before accessing m_convReal!
 409     wxMBConv *m_convReal;
 410     bool m_deferred;
 411 };
 412
 413
 414 // ----------------------------------------------------------------------------
 415 // declare predefined conversion objects
 416 // ----------------------------------------------------------------------------
 417
 418 // conversion to be used with all standard functions affected by locale, e.g.
 419 // strtol(), strftime(), ...
 420 extern WXDLLIMPEXP_DATA_BASE(wxMBConv&) wxConvLibc;
 421
 422 // conversion ISO-8859-1/UTF-7/UTF-8 <-> wchar_t
 423 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvISO8859_1;
 424 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7&) wxConvUTF7;
 425 extern WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8&) wxConvUTF8;
 426
 427 // conversion used for the file names on the systems where they're not Unicode
 428 // (basically anything except Windows)
 429 //
 430 // this is used by all file functions, can be changed by the application
 431 //
 432 // by default UTF-8 under Mac OS X and wxConvLibc elsewhere (but it's not used
 433 // under Windows normally)
 434 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvFileName;
 435
 436 // backwards compatible define
 437 #define wxConvFile (*wxConvFileName)
 438
 439 // the current conversion object, may be set to any conversion, is used by
 440 // default in a couple of places inside wx (initially same as wxConvLibc)
 441 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 442
 443 // the conversion corresponding to the current locale
 444 extern WXDLLIMPEXP_DATA_BASE(wxCSConv&) wxConvLocal;
 445
 446 // the conversion corresponding to the encoding of the standard UI elements
 447 //
 448 // by default this is the same as wxConvLocal but may be changed if the program
 449 // needs to use a fixed encoding
 450 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvUI;
 451
 452 // ----------------------------------------------------------------------------
 453 // endianness-dependent conversions
 454 // ----------------------------------------------------------------------------
 455
 456 #ifdef WORDS_BIGENDIAN
 457     typedef wxMBConvUTF16BE wxMBConvUTF16;
 458     typedef wxMBConvUTF32BE wxMBConvUTF32;
 459 #else
 460     typedef wxMBConvUTF16LE wxMBConvUTF16;
 461     typedef wxMBConvUTF32LE wxMBConvUTF32;
 462 #endif
 463
 464 // ----------------------------------------------------------------------------
 465 // filename conversion macros
 466 // ----------------------------------------------------------------------------
 467
 468 // filenames are multibyte on Unix and widechar on Windows
 469 #if defined(__UNIX__) || defined(__WXMAC__)
 470     #define wxMBFILES 1
 471 #else
 472     #define wxMBFILES 0
 473 #endif
 474
 475 #if wxMBFILES && wxUSE_UNICODE
 476     #define wxFNCONV(name) wxConvFileName->cWX2MB(name)
 477     #define wxFNSTRINGCAST wxMBSTRINGCAST
 478 #else
 479 #if defined( __WXOSX__ ) && wxMBFILES
 480     #define wxFNCONV(name) wxConvFileName->cWC2MB( wxConvLocal.cWX2WC(name) )
 481 #else
 482     #define wxFNCONV(name) name
 483 #endif
 484     #define wxFNSTRINGCAST WXSTRINGCAST
 485 #endif
 486
 487 #else // !wxUSE_WCHAR_T
 488
 489 // ----------------------------------------------------------------------------
 490 // stand-ins in absence of wchar_t
 491 // ----------------------------------------------------------------------------
 492
 493 class WXDLLIMPEXP_BASE wxMBConv
 494 {
 495 public:
 496     const char* cMB2WX(const char *psz) const { return psz; }
 497     const char* cWX2MB(const char *psz) const { return psz; }
 498 };
 499
 500 #define wxConvFile wxConvLocal
 501
 502 extern WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
 503                                        wxConvLocal,
 504                                        wxConvISO8859_1,
 505                                        wxConvUTF8;
 506 extern WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent;
 507
 508 #define wxFNCONV(name) name
 509 #define wxFNSTRINGCAST WXSTRINGCAST
 510
 511 #endif
 512   // wxUSE_WCHAR_T
 513
 514 // ----------------------------------------------------------------------------
 515 // macros for the most common conversions
 516 // ----------------------------------------------------------------------------
 517
 518 #if wxUSE_UNICODE
 519     #define wxConvertWX2MB(s)   wxConvCurrent->cWX2MB(s)
 520     #define wxConvertMB2WX(s)   wxConvCurrent->cMB2WX(s)
 521 #else // ANSI
 522     // no conversions to do
 523     #define wxConvertWX2MB(s)   (s)
 524     #define wxConvertMB2WX(s)   (s)
 525 #endif // Unicode/ANSI
 526
 527 #endif // _WX_STRCONV_H_
 528