src/common/strconv.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        strconv.cpp
   3 // Purpose:     Unicode conversion classes
   4 // Author:      Ove Kaaven, Robert Roebling, Vadim Zeitlin, Vaclav Slavik
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1999 Ove Kaaven, Robert Roebling, Vaclav Slavik
   9 //              (c) 2000-2003 Vadim Zeitlin
  10 // Licence:     wxWindows licence
  11 /////////////////////////////////////////////////////////////////////////////
  12
  13 // ============================================================================
  14 // declarations
  15 // ============================================================================
  16
  17 // ----------------------------------------------------------------------------
  18 // headers
  19 // ----------------------------------------------------------------------------
  20
  21 #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
  22   #pragma implementation "strconv.h"
  23 #endif
  24
  25 // For compilers that support precompilation, includes "wx.h".
  26 #include "wx/wxprec.h"
  27
  28 #ifdef __BORLANDC__
  29   #pragma hdrstop
  30 #endif
  31
  32 #ifndef WX_PRECOMP
  33     #include "wx/intl.h"
  34     #include "wx/log.h"
  35 #endif // WX_PRECOMP
  36
  37 #ifdef __WXMSW__
  38     #include "wx/msw/private.h"
  39 #endif
  40
  41 #ifndef __WXWINCE__
  42 #include <errno.h>
  43 #endif
  44
  45 #include <ctype.h>
  46 #include <string.h>
  47 #include <stdlib.h>
  48
  49 #include "wx/module.h"
  50 #include "wx/strconv.h"
  51
  52 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
  53     #define wxHAVE_WIN32_MB2WC
  54 #endif // __WIN32__ but !__WXMICROWIN__
  55
  56 // ----------------------------------------------------------------------------
  57 // globals
  58 // ----------------------------------------------------------------------------
  59
  60 #if wxUSE_WCHAR_T
  61     WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc;
  62     WXDLLIMPEXP_DATA_BASE(wxCSConv) wxConvLocal((const wxChar *)NULL);
  63     WXDLLIMPEXP_DATA_BASE(wxCSConv) wxConvISO8859_1(_T("iso-8859-1"));
  64 #else
  65     // stand-ins in absence of wchar_t
  66     WXDLLIMPEXP_DATA_BASE(wxMBConv) wxConvLibc,
  67                                     wxConvFile,
  68                                     wxConvISO8859_1,
  69                                     wxConvLocal,
  70                                     wxConvUTF8;
  71 #endif // wxUSE_WCHAR_T
  72
  73 WXDLLIMPEXP_DATA_BASE(wxMBConv *) wxConvCurrent = &wxConvLibc;
  74
  75 class wxStrConvModule: public wxModule
  76 {
  77 public:
  78     wxStrConvModule() : wxModule() { }
  79     virtual bool OnInit() { return true; }
  80     virtual void OnExit()
  81     {
  82 #if wxUSE_WCHAR_T
  83          wxConvLocal.Clear();
  84          wxConvISO8859_1.Clear();
  85 #endif
  86     }
  87
  88     DECLARE_DYNAMIC_CLASS(wxStrConvModule)
  89 };
  90
  91 IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
  92
  93
  94 // ----------------------------------------------------------------------------
  95 // headers
  96 // ----------------------------------------------------------------------------
  97
  98 #if wxUSE_WCHAR_T
  99
 100 #ifdef __SALFORDC__
 101     #include <clib.h>
 102 #endif
 103
 104 #ifdef HAVE_ICONV
 105     #include <iconv.h>
 106 #endif
 107
 108 #include "wx/encconv.h"
 109 #include "wx/fontmap.h"
 110
 111 // ----------------------------------------------------------------------------
 112 // macros
 113 // ----------------------------------------------------------------------------
 114
 115 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
 116 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
 117
 118 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
 119 // it might be not defined - assume the most common value
 120 #ifndef SIZEOF_WCHAR_T
 121     #define SIZEOF_WCHAR_T 2
 122 #endif // !defined(SIZEOF_WCHAR_T)
 123
 124 #if SIZEOF_WCHAR_T == 4
 125     #define WC_NAME         "UCS4"
 126     #define WC_BSWAP         BSWAP_UCS4
 127     #ifdef WORDS_BIGENDIAN
 128       #define WC_NAME_BEST  "UCS-4BE"
 129     #else
 130       #define WC_NAME_BEST  "UCS-4LE"
 131     #endif
 132 #elif SIZEOF_WCHAR_T == 2
 133     #define WC_NAME         "UTF16"
 134     #define WC_BSWAP         BSWAP_UTF16
 135     #define WC_UTF16
 136     #ifdef WORDS_BIGENDIAN
 137       #define WC_NAME_BEST  "UTF-16BE"
 138     #else
 139       #define WC_NAME_BEST  "UTF-16LE"
 140     #endif
 141 #else // sizeof(wchar_t) != 2 nor 4
 142     // I don't know what to do about this
 143     #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
 144 #endif
 145
 146 // ============================================================================
 147 // implementation
 148 // ============================================================================
 149
 150 // ----------------------------------------------------------------------------
 151 // UTF-16 en/decoding to/from UCS-4
 152 // ----------------------------------------------------------------------------
 153
 154
 155 static size_t encode_utf16(wxUint32 input, wxUint16 *output)
 156 {
 157     if (input<=0xffff)
 158     {
 159         if (output) *output++ = (wxUint16) input;
 160         return 1;
 161     }
 162     else if (input>=0x110000)
 163     {
 164         return (size_t)-1;
 165     }
 166     else
 167     {
 168         if (output)
 169         {
 170             *output++ = (wxUint16) ((input >> 10)+0xd7c0);
 171             *output++ = (wxUint16) ((input&0x3ff)+0xdc00);
 172         }
 173         return 2;
 174     }
 175 }
 176
 177 static size_t decode_utf16(const wxUint16* input, wxUint32& output)
 178 {
 179     if ((*input<0xd800) || (*input>0xdfff))
 180     {
 181         output = *input;
 182         return 1;
 183     }
 184     else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
 185     {
 186         output = *input;
 187         return (size_t)-1;
 188     }
 189     else
 190     {
 191         output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
 192         return 2;
 193     }
 194 }
 195
 196
 197 // ----------------------------------------------------------------------------
 198 // wxMBConv
 199 // ----------------------------------------------------------------------------
 200
 201 #define IGNORE_LIBC 0
 202
 203 wxMBConv::~wxMBConv()
 204 {
 205     // nothing to do here
 206 }
 207
 208 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 209 {
 210 #if IGNORE_LIBC
 211     if (buf)
 212     {
 213         for (size_t i = 0; i < strlen( psz )+1; i++)
 214             buf[i] = (wchar_t) psz[i];
 215         return strlen( psz );
 216     }
 217     else
 218     {
 219         return strlen( psz );
 220     }
 221 #else
 222     return wxMB2WC(buf, psz, n);
 223 #endif
 224 }
 225
 226 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 227 {
 228 #if IGNORE_LIBC
 229     if (buf)
 230     {
 231         for (size_t i = 0; i < wxStrlen( psz )+1; i++)
 232             buf[i] = (char) psz[i];
 233         return wxStrlen( psz );
 234     }
 235     else
 236     {
 237         return wxStrlen( psz );
 238     }
 239 #else
 240     return wxWC2MB(buf, psz, n);
 241 #endif
 242 }
 243
 244 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 245 {
 246     if ( psz )
 247     {
 248         // calculate the length of the buffer needed first
 249         size_t nLen = MB2WC(NULL, psz, 0);
 250         if ( nLen != (size_t)-1 )
 251         {
 252             // now do the actual conversion
 253             wxWCharBuffer buf(nLen);
 254             MB2WC(buf.data(), psz, nLen + 1); // with the trailing NUL
 255
 256             return buf;
 257         }
 258     }
 259
 260     wxWCharBuffer buf((wchar_t *)NULL);
 261
 262     return buf;
 263 }
 264
 265 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *pwz) const
 266 {
 267     if ( pwz )
 268     {
 269         size_t nLen = WC2MB(NULL, pwz, 0);
 270         if ( nLen != (size_t)-1 )
 271         {
 272             wxCharBuffer buf(nLen+3);       // space for a wxUint32 trailing zero
 273             WC2MB(buf.data(), pwz, nLen + 4);
 274
 275             return buf;
 276         }
 277     }
 278
 279     wxCharBuffer buf((char *)NULL);
 280
 281     return buf;
 282 }
 283
 284 // ----------------------------------------------------------------------------
 285 // UTF-7
 286 // ----------------------------------------------------------------------------
 287
 288 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF7) wxConvUTF7;
 289
 290 #if 0
 291 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 292                         "abcdefghijklmnopqrstuvwxyz"
 293                         "0123456789'(),-./:?";
 294 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
 295 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 296                         "abcdefghijklmnopqrstuvwxyz"
 297                         "0123456789+/";
 298 #endif
 299
 300 // TODO: write actual implementations of UTF-7 here
 301 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
 302                            const char * WXUNUSED(psz),
 303                            size_t WXUNUSED(n)) const
 304 {
 305   return 0;
 306 }
 307
 308 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
 309                            const wchar_t * WXUNUSED(psz),
 310                            size_t WXUNUSED(n)) const
 311 {
 312   return 0;
 313 }
 314
 315 // ----------------------------------------------------------------------------
 316 // UTF-8
 317 // ----------------------------------------------------------------------------
 318
 319 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF8) wxConvUTF8;
 320
 321 static wxUint32 utf8_max[]=
 322     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 323
 324 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 325 {
 326     size_t len = 0;
 327
 328     while (*psz && ((!buf) || (len < n)))
 329     {
 330         unsigned char cc = *psz++, fc = cc;
 331         unsigned cnt;
 332         for (cnt = 0; fc & 0x80; cnt++)
 333             fc <<= 1;
 334         if (!cnt)
 335         {
 336             // plain ASCII char
 337             if (buf)
 338                 *buf++ = cc;
 339             len++;
 340         }
 341         else
 342         {
 343             cnt--;
 344             if (!cnt)
 345             {
 346                 // invalid UTF-8 sequence
 347                 return (size_t)-1;
 348             }
 349             else
 350             {
 351                 unsigned ocnt = cnt - 1;
 352                 wxUint32 res = cc & (0x3f >> cnt);
 353                 while (cnt--)
 354                 {
 355                     cc = *psz++;
 356                     if ((cc & 0xC0) != 0x80)
 357                     {
 358                         // invalid UTF-8 sequence
 359                         return (size_t)-1;
 360                     }
 361                     res = (res << 6) | (cc & 0x3f);
 362                 }
 363                 if (res <= utf8_max[ocnt])
 364                 {
 365                     // illegal UTF-8 encoding
 366                     return (size_t)-1;
 367                 }
 368 #ifdef WC_UTF16
 369                 size_t pa = encode_utf16(res, buf);
 370                 if (pa == (size_t)-1)
 371                   return (size_t)-1;
 372                 if (buf)
 373                     buf += pa;
 374                 len += pa;
 375 #else // !WC_UTF16
 376                 if (buf)
 377                     *buf++ = res;
 378                 len++;
 379 #endif // WC_UTF16/!WC_UTF16
 380             }
 381         }
 382     }
 383     if (buf && (len < n))
 384         *buf = 0;
 385     return len;
 386 }
 387
 388 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 389 {
 390     size_t len = 0;
 391
 392     while (*psz && ((!buf) || (len < n)))
 393     {
 394         wxUint32 cc;
 395 #ifdef WC_UTF16
 396         size_t pa = decode_utf16(psz, cc);
 397         psz += (pa == (size_t)-1) ? 1 : pa;
 398 #else
 399         cc=(*psz++) & 0x7fffffff;
 400 #endif
 401         unsigned cnt;
 402         for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
 403         if (!cnt)
 404         {
 405             // plain ASCII char
 406             if (buf)
 407                 *buf++ = (char) cc;
 408             len++;
 409         }
 410
 411         else
 412         {
 413             len += cnt + 1;
 414             if (buf)
 415             {
 416                 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
 417                 while (cnt--)
 418                     *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
 419             }
 420         }
 421     }
 422
 423     if (buf && (len<n)) *buf = 0;
 424
 425     return len;
 426 }
 427
 428
 429
 430
 431 // ----------------------------------------------------------------------------
 432 // UTF-16
 433 // ----------------------------------------------------------------------------
 434
 435 #ifdef WORDS_BIGENDIAN
 436 #define wxMBConvUTF16straight wxMBConvUTF16BE
 437 #define wxMBConvUTF16swap     wxMBConvUTF16LE
 438 #else
 439 #define wxMBConvUTF16swap     wxMBConvUTF16BE
 440 #define wxMBConvUTF16straight wxMBConvUTF16LE
 441 #endif
 442
 443
 444 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF16LE) wxConvUTF16LE;
 445 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF16BE) wxConvUTF16BE;
 446
 447
 448
 449
 450
 451 #ifdef WC_UTF16
 452
 453
 454 // copy 16bit MB to 16bit String
 455 size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 456 {
 457     size_t len=0;
 458
 459     while (*(wxUint16*)psz && (!buf || len < n))
 460     {
 461         if (buf)
 462             *buf++ = *(wxUint16*)psz;
 463         len++;
 464
 465         psz += sizeof(wxUint16);
 466     }
 467     if (buf && len<n)   *buf=0;
 468
 469     return len;
 470 }
 471
 472
 473 // copy 16bit String to 16bit MB
 474 size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 475 {
 476     size_t len=0;
 477
 478     while (*psz && (!buf || len < n))
 479     {
 480         if (buf)
 481         {
 482             *(wxUint16*)buf = *psz;
 483             buf += sizeof(wxUint16);
 484         }
 485         len += sizeof(wxUint16);
 486         psz++;
 487     }
 488     if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
 489
 490     return len;
 491 }
 492
 493
 494 // swap 16bit MB to 16bit String
 495 size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 496 {
 497     size_t len=0;
 498
 499     while (*(wxUint16*)psz && (!buf || len < n))
 500     {
 501         if (buf)
 502         {
 503             ((char *)buf)[0] = psz[1];
 504             ((char *)buf)[1] = psz[0];
 505             buf++;
 506         }
 507         len++;
 508         psz += sizeof(wxUint16);
 509     }
 510     if (buf && len<n)   *buf=0;
 511
 512     return len;
 513 }
 514
 515
 516 // swap 16bit MB to 16bit String
 517 size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 518 {
 519     size_t len=0;
 520
 521     while (*psz && (!buf || len < n))
 522     {
 523         if (buf)
 524         {
 525             *buf++ = ((char*)psz)[1];
 526             *buf++ = ((char*)psz)[0];
 527         }
 528         len += sizeof(wxUint16);
 529         psz++;
 530     }
 531     if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
 532
 533     return len;
 534 }
 535
 536
 537 #else // WC_UTF16
 538
 539
 540 // copy 16bit MB to 32bit String
 541 size_t wxMBConvUTF16straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 542 {
 543     size_t len=0;
 544
 545     while (*(wxUint16*)psz && (!buf || len < n))
 546     {
 547         wxUint32 cc;
 548         size_t pa=decode_utf16((wxUint16*)psz, cc);
 549         if (pa == (size_t)-1)
 550             return pa;
 551
 552         if (buf)
 553             *buf++ = cc;
 554         len++;
 555         psz += pa * sizeof(wxUint16);
 556     }
 557     if (buf && len<n)   *buf=0;
 558
 559     return len;
 560 }
 561
 562
 563 // copy 32bit String to 16bit MB
 564 size_t wxMBConvUTF16straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 565 {
 566     size_t len=0;
 567
 568     while (*psz && (!buf || len < n))
 569     {
 570         wxUint16 cc[2];
 571         size_t pa=encode_utf16(*psz, cc);
 572
 573         if (pa == (size_t)-1)
 574             return pa;
 575
 576         if (buf)
 577         {
 578             *(wxUint16*)buf = cc[0];
 579             buf += sizeof(wxUint16);
 580             if (pa > 1)
 581             {
 582                 *(wxUint16*)buf = cc[1];
 583                 buf += sizeof(wxUint16);
 584             }
 585         }
 586
 587         len += pa*sizeof(wxUint16);
 588         psz++;
 589     }
 590     if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
 591
 592     return len;
 593 }
 594
 595
 596 // swap 16bit MB to 32bit String
 597 size_t wxMBConvUTF16swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 598 {
 599     size_t len=0;
 600
 601     while (*(wxUint16*)psz && (!buf || len < n))
 602     {
 603         wxUint32 cc;
 604         char tmp[4];
 605         tmp[0]=psz[1];  tmp[1]=psz[0];
 606         tmp[2]=psz[3];  tmp[3]=psz[2];
 607
 608         size_t pa=decode_utf16((wxUint16*)tmp, cc);
 609         if (pa == (size_t)-1)
 610             return pa;
 611
 612         if (buf)
 613             *buf++ = cc;
 614
 615         len++;
 616         psz += pa * sizeof(wxUint16);
 617     }
 618     if (buf && len<n)   *buf=0;
 619
 620     return len;
 621 }
 622
 623
 624 // swap 32bit String to 16bit MB
 625 size_t wxMBConvUTF16swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 626 {
 627     size_t len=0;
 628
 629     while (*psz && (!buf || len < n))
 630     {
 631         wxUint16 cc[2];
 632         size_t pa=encode_utf16(*psz, cc);
 633
 634         if (pa == (size_t)-1)
 635             return pa;
 636
 637         if (buf)
 638         {
 639             *buf++ = ((char*)cc)[1];
 640             *buf++ = ((char*)cc)[0];
 641             if (pa > 1)
 642             {
 643                 *buf++ = ((char*)cc)[3];
 644                 *buf++ = ((char*)cc)[2];
 645             }
 646         }
 647
 648         len += pa*sizeof(wxUint16);
 649         psz++;
 650     }
 651     if (buf && len<=n-sizeof(wxUint16))   *(wxUint16*)buf=0;
 652
 653     return len;
 654 }
 655
 656 #endif // WC_UTF16
 657
 658
 659 // ----------------------------------------------------------------------------
 660 // UTF-32
 661 // ----------------------------------------------------------------------------
 662
 663 #ifdef WORDS_BIGENDIAN
 664 #define wxMBConvUTF32straight  wxMBConvUTF32BE
 665 #define wxMBConvUTF32swap      wxMBConvUTF32LE
 666 #else
 667 #define wxMBConvUTF32swap      wxMBConvUTF32BE
 668 #define wxMBConvUTF32straight  wxMBConvUTF32LE
 669 #endif
 670
 671
 672 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32LE) wxConvUTF32LE;
 673 WXDLLIMPEXP_DATA_BASE(wxMBConvUTF32BE) wxConvUTF32BE;
 674
 675
 676 #ifdef WC_UTF16
 677
 678 // copy 32bit MB to 16bit String
 679 size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 680 {
 681     size_t len=0;
 682
 683     while (*(wxUint32*)psz && (!buf || len < n))
 684     {
 685         wxUint16 cc[2];
 686
 687         size_t pa=encode_utf16(*(wxUint32*)psz, cc);
 688         if (pa == (size_t)-1)
 689             return pa;
 690
 691         if (buf)
 692         {
 693             *buf++ = cc[0];
 694             if (pa > 1)
 695                 *buf++ = cc[1];
 696         }
 697         len += pa;
 698         psz += sizeof(wxUint32);
 699     }
 700     if (buf && len<n)   *buf=0;
 701
 702     return len;
 703 }
 704
 705
 706 // copy 16bit String to 32bit MB
 707 size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 708 {
 709     size_t len=0;
 710
 711     while (*psz && (!buf || len < n))
 712     {
 713         wxUint32 cc;
 714
 715         size_t pa=decode_utf16(psz, cc);
 716         if (pa == (size_t)-1)
 717             return pa;
 718
 719         if (buf)
 720         {
 721             *(wxUint32*)buf = cc;
 722             buf += sizeof(wxUint32);
 723         }
 724         len += sizeof(wxUint32);
 725         psz += pa;
 726     }
 727     if (buf && len<=n-sizeof(wxUint32))   *(wxUint32*)buf=0;
 728
 729     return len;
 730 }
 731
 732
 733
 734 // swap 32bit MB to 16bit String
 735 size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 736 {
 737     size_t len=0;
 738
 739     while (*(wxUint32*)psz && (!buf || len < n))
 740     {
 741         char tmp[4];
 742         tmp[0] = psz[3];   tmp[1] = psz[2];
 743         tmp[2] = psz[1];   tmp[3] = psz[0];
 744
 745
 746         wxUint16 cc[2];
 747
 748         size_t pa=encode_utf16(*(wxUint32*)tmp, cc);
 749         if (pa == (size_t)-1)
 750             return pa;
 751
 752         if (buf)
 753         {
 754             *buf++ = cc[0];
 755             if (pa > 1)
 756                 *buf++ = cc[1];
 757         }
 758         len += pa;
 759         psz += sizeof(wxUint32);
 760     }
 761     if (buf && len<n)   *buf=0;
 762
 763     return len;
 764 }
 765
 766
 767 // swap 16bit String to 32bit MB
 768 size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 769 {
 770     size_t len=0;
 771
 772     while (*psz && (!buf || len < n))
 773     {
 774         char cc[4];
 775
 776         size_t pa=decode_utf16(psz, *(wxUint32*)cc);
 777         if (pa == (size_t)-1)
 778             return pa;
 779
 780         if (buf)
 781         {
 782             *buf++ = cc[3];
 783             *buf++ = cc[2];
 784             *buf++ = cc[1];
 785             *buf++ = cc[0];
 786         }
 787         len += sizeof(wxUint32);
 788         psz += pa;
 789     }
 790     if (buf && len<=n-sizeof(wxUint32))   *(wxUint32*)buf=0;
 791
 792     return len;
 793 }
 794
 795 #else // WC_UTF16
 796
 797
 798 // copy 32bit MB to 32bit String
 799 size_t wxMBConvUTF32straight::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 800 {
 801     size_t len=0;
 802
 803     while (*(wxUint32*)psz && (!buf || len < n))
 804     {
 805         if (buf)
 806             *buf++ = *(wxUint32*)psz;
 807         len++;
 808         psz += sizeof(wxUint32);
 809     }
 810     if (buf && len<n)   *buf=0;
 811
 812     return len;
 813 }
 814
 815
 816 // copy 32bit String to 32bit MB
 817 size_t wxMBConvUTF32straight::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 818 {
 819     size_t len=0;
 820
 821     while (*psz && (!buf || len < n))
 822     {
 823         if (buf)
 824         {
 825             *(wxUint32*)buf = *psz;
 826             buf += sizeof(wxUint32);
 827         }
 828
 829         len += sizeof(wxUint32);
 830         psz++;
 831     }
 832
 833     if (buf && len<=n-sizeof(wxUint32))   *(wxUint32*)buf=0;
 834
 835     return len;
 836 }
 837
 838
 839 // swap 32bit MB to 32bit String
 840 size_t wxMBConvUTF32swap::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 841 {
 842     size_t len=0;
 843
 844     while (*(wxUint32*)psz && (!buf || len < n))
 845     {
 846         if (buf)
 847         {
 848             ((char *)buf)[0] = psz[3];
 849             ((char *)buf)[1] = psz[2];
 850             ((char *)buf)[2] = psz[1];
 851             ((char *)buf)[3] = psz[0];
 852             buf++;
 853         }
 854         len++;
 855         psz += sizeof(wxUint32);
 856     }
 857     if (buf && len<n)   *buf=0;
 858
 859     return len;
 860 }
 861
 862
 863 // swap 32bit String to 32bit MB
 864 size_t wxMBConvUTF32swap::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 865 {
 866     size_t len=0;
 867
 868     while (*psz && (!buf || len < n))
 869     {
 870         if (buf)
 871         {
 872             *buf++ = ((char *)psz)[3];
 873             *buf++ = ((char *)psz)[2];
 874             *buf++ = ((char *)psz)[1];
 875             *buf++ = ((char *)psz)[0];
 876         }
 877         len += sizeof(wxUint32);
 878         psz++;
 879     }
 880     if (buf && len<=n-sizeof(wxUint32))   *(wxUint32*)buf=0;
 881
 882     return len;
 883 }
 884
 885
 886 #endif // WC_UTF16
 887
 888
 889 // ============================================================================
 890 // The classes doing conversion using the iconv_xxx() functions
 891 // ============================================================================
 892
 893 #ifdef HAVE_ICONV
 894
 895 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
 896 //     if output buffer is _exactly_ as big as needed. Such case is (unless there's
 897 //     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
 898 //     (which means error) and says there are 0 bytes left in the input buffer --
 899 //     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
 900 //     this alternative test for iconv() failure.
 901 //     [This bug does not appear in glibc 2.2.]
 902 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
 903 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
 904                                      (errno != E2BIG || bufLeft != 0))
 905 #else
 906 #define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
 907 #endif
 908
 909 #define ICONV_CHAR_CAST(x)  ((ICONV_CONST char **)(x))
 910
 911 // ----------------------------------------------------------------------------
 912 // wxMBConv_iconv: encapsulates an iconv character set
 913 // ----------------------------------------------------------------------------
 914
 915 class wxMBConv_iconv : public wxMBConv
 916 {
 917 public:
 918     wxMBConv_iconv(const wxChar *name);
 919     virtual ~wxMBConv_iconv();
 920
 921     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
 922     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
 923
 924     bool IsOk() const
 925         { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
 926
 927 protected:
 928     // the iconv handlers used to translate from multibyte to wide char and in
 929     // the other direction
 930     iconv_t m2w,
 931             w2m;
 932
 933 private:
 934     // the name (for iconv_open()) of a wide char charset -- if none is
 935     // available on this machine, it will remain NULL
 936     static const char *ms_wcCharsetName;
 937
 938     // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
 939     // different endian-ness than the native one
 940     static bool ms_wcNeedsSwap;
 941 };
 942
 943 const char *wxMBConv_iconv::ms_wcCharsetName = NULL;
 944 bool wxMBConv_iconv::ms_wcNeedsSwap = false;
 945
 946 wxMBConv_iconv::wxMBConv_iconv(const wxChar *name)
 947 {
 948     // Do it the hard way
 949     char cname[100];
 950     for (size_t i = 0; i < wxStrlen(name)+1; i++)
 951         cname[i] = (char) name[i];
 952
 953     // check for charset that represents wchar_t:
 954     if (ms_wcCharsetName == NULL)
 955     {
 956         ms_wcNeedsSwap = false;
 957
 958         // try charset with explicit bytesex info (e.g. "UCS-4LE"):
 959         ms_wcCharsetName = WC_NAME_BEST;
 960         m2w = iconv_open(ms_wcCharsetName, cname);
 961
 962         if (m2w == (iconv_t)-1)
 963         {
 964             // try charset w/o bytesex info (e.g. "UCS4")
 965             // and check for bytesex ourselves:
 966             ms_wcCharsetName = WC_NAME;
 967             m2w = iconv_open(ms_wcCharsetName, cname);
 968
 969             // last bet, try if it knows WCHAR_T pseudo-charset
 970             if (m2w == (iconv_t)-1)
 971             {
 972                 ms_wcCharsetName = "WCHAR_T";
 973                 m2w = iconv_open(ms_wcCharsetName, cname);
 974             }
 975
 976             if (m2w != (iconv_t)-1)
 977             {
 978                 char    buf[2], *bufPtr;
 979                 wchar_t wbuf[2], *wbufPtr;
 980                 size_t  insz, outsz;
 981                 size_t  res;
 982
 983                 buf[0] = 'A';
 984                 buf[1] = 0;
 985                 wbuf[0] = 0;
 986                 insz = 2;
 987                 outsz = SIZEOF_WCHAR_T * 2;
 988                 wbufPtr = wbuf;
 989                 bufPtr = buf;
 990
 991                 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
 992                             (char**)&wbufPtr, &outsz);
 993
 994                 if (ICONV_FAILED(res, insz))
 995                 {
 996                     ms_wcCharsetName = NULL;
 997                     wxLogLastError(wxT("iconv"));
 998                     wxLogError(_("Conversion to charset '%s' doesn't work."), name);
 999                 }
1000                 else
1001                 {
1002                     ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
1003                 }
1004             }
1005             else
1006             {
1007                 ms_wcCharsetName = NULL;
1008
1009                 // VS: we must not output an error here, since wxWindows will safely
1010                 //     fall back to using wxEncodingConverter.
1011                 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
1012                 //wxLogError(
1013             }
1014         }
1015         wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
1016     }
1017     else // we already have ms_wcCharsetName
1018     {
1019         m2w = iconv_open(ms_wcCharsetName, cname);
1020     }
1021
1022     // NB: don't ever pass NULL to iconv_open(), it may crash!
1023     if ( ms_wcCharsetName )
1024     {
1025         w2m = iconv_open( cname, ms_wcCharsetName);
1026     }
1027     else
1028     {
1029         w2m = (iconv_t)-1;
1030     }
1031 }
1032
1033 wxMBConv_iconv::~wxMBConv_iconv()
1034 {
1035     if ( m2w != (iconv_t)-1 )
1036         iconv_close(m2w);
1037     if ( w2m != (iconv_t)-1 )
1038         iconv_close(w2m);
1039 }
1040
1041 size_t wxMBConv_iconv::MB2WC(wchar_t *buf, const char *psz, size_t n)
1042 {
1043     size_t inbuf = strlen(psz);
1044     size_t outbuf = n * SIZEOF_WCHAR_T;
1045     size_t res, cres;
1046     // VS: Use these instead of psz, buf because iconv() modifies its arguments:
1047     wchar_t *bufPtr = buf;
1048     const char *pszPtr = psz;
1049
1050     if (buf)
1051     {
1052         // have destination buffer, convert there
1053         cres = iconv(m2w,
1054                      ICONV_CHAR_CAST(&pszPtr), &inbuf,
1055                      (char**)&bufPtr, &outbuf);
1056         res = n - (outbuf / SIZEOF_WCHAR_T);
1057
1058         if (ms_wcNeedsSwap)
1059         {
1060             // convert to native endianness
1061             WC_BSWAP(buf /* _not_ bufPtr */, res)
1062         }
1063
1064         // NB: iconv was given only strlen(psz) characters on input, and so
1065         //     it couldn't convert the trailing zero. Let's do it ourselves
1066         //     if there's some room left for it in the output buffer.
1067         if (res < n)
1068             buf[res] = 0;
1069     }
1070     else
1071     {
1072         // no destination buffer... convert using temp buffer
1073         // to calculate destination buffer requirement
1074         wchar_t tbuf[8];
1075         res = 0;
1076         do {
1077             bufPtr = tbuf;
1078             outbuf = 8*SIZEOF_WCHAR_T;
1079
1080             cres = iconv(m2w,
1081                          ICONV_CHAR_CAST(&pszPtr), &inbuf,
1082                          (char**)&bufPtr, &outbuf );
1083
1084             res += 8-(outbuf/SIZEOF_WCHAR_T);
1085         } while ((cres==(size_t)-1) && (errno==E2BIG));
1086     }
1087
1088     if (ICONV_FAILED(cres, inbuf))
1089     {
1090         //VS: it is ok if iconv fails, hence trace only
1091         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1092         return (size_t)-1;
1093     }
1094
1095     return res;
1096 }
1097
1098 size_t wxMBConv_iconv::WC2MB(char *buf, const wchar_t *psz, size_t n)
1099 {
1100     size_t inbuf = wxWcslen(psz) * SIZEOF_WCHAR_T;
1101     size_t outbuf = n;
1102     size_t res, cres;
1103
1104     wchar_t *tmpbuf = 0;
1105
1106     if (ms_wcNeedsSwap)
1107     {
1108         // need to copy to temp buffer to switch endianness
1109         // this absolutely doesn't rock!
1110         // (no, doing WC_BSWAP twice on the original buffer won't help, as it
1111         //  could be in read-only memory, or be accessed in some other thread)
1112         tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
1113         memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
1114         WC_BSWAP(tmpbuf, inbuf)
1115         psz=tmpbuf;
1116     }
1117
1118     if (buf)
1119     {
1120         // have destination buffer, convert there
1121         cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1122
1123         res = n-outbuf;
1124
1125         // NB: iconv was given only wcslen(psz) characters on input, and so
1126         //     it couldn't convert the trailing zero. Let's do it ourselves
1127         //     if there's some room left for it in the output buffer.
1128         if (res < n)
1129             buf[0] = 0;
1130     }
1131     else
1132     {
1133         // no destination buffer... convert using temp buffer
1134         // to calculate destination buffer requirement
1135         char tbuf[16];
1136         res = 0;
1137         do {
1138             buf = tbuf; outbuf = 16;
1139
1140             cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
1141
1142             res += 16 - outbuf;
1143         } while ((cres==(size_t)-1) && (errno==E2BIG));
1144     }
1145
1146     if (ms_wcNeedsSwap)
1147     {
1148         free(tmpbuf);
1149     }
1150
1151     if (ICONV_FAILED(cres, inbuf))
1152     {
1153         //VS: it is ok if iconv fails, hence trace only
1154         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
1155         return (size_t)-1;
1156     }
1157
1158     return res;
1159 }
1160
1161 #endif // HAVE_ICONV
1162
1163
1164 // ============================================================================
1165 // Win32 conversion classes
1166 // ============================================================================
1167
1168 #ifdef wxHAVE_WIN32_MB2WC
1169
1170 // from utils.cpp
1171 extern WXDLLIMPEXP_BASE long wxCharsetToCodepage(const wxChar *charset);
1172 extern WXDLLIMPEXP_BASE long wxEncodingToCodepage(wxFontEncoding encoding);
1173
1174 class wxMBConv_win32 : public wxMBConv
1175 {
1176 public:
1177     wxMBConv_win32(const wxChar* name)
1178         {
1179             m_CodePage = wxCharsetToCodepage(name);
1180         }
1181
1182     wxMBConv_win32(wxFontEncoding encoding)
1183         {
1184             m_CodePage = wxEncodingToCodepage(encoding);
1185         }
1186
1187     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
1188     {
1189         const size_t len = ::MultiByteToWideChar
1190                              (
1191                                 m_CodePage,     // code page
1192                                 0,              // flags (none)
1193                                 psz,            // input string
1194                                 -1,             // its length (NUL-terminated)
1195                                 buf,            // output string
1196                                 buf ? n : 0     // size of output buffer
1197                              );
1198
1199         // note that it returns # of written chars for buf != NULL and *size*
1200         // of the needed buffer for buf == NULL
1201         return len ? (buf ? len : len - 1) : (size_t)-1;
1202     }
1203
1204     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
1205     {
1206         const size_t len = ::WideCharToMultiByte
1207                              (
1208                                 m_CodePage,     // code page
1209                                 0,              // flags (none)
1210                                 psz,            // input string
1211                                 -1,             // it is (wide) NUL-terminated
1212                                 buf,            // output buffer
1213                                 buf ? n : 0,    // and its size
1214                                 NULL,           // default "replacement" char
1215                                 NULL            // [out] was it used?
1216                              );
1217
1218         // see the comment above!
1219         return len ? (buf ? len : len - 1) : (size_t)-1;
1220     }
1221
1222     bool IsOk() const
1223         { return m_CodePage != -1; }
1224
1225 public:
1226     long m_CodePage;
1227 };
1228
1229 #endif // wxHAVE_WIN32_MB2WC
1230
1231
1232 // ============================================================================
1233 // wxEncodingConverter based conversion classes
1234 // ============================================================================
1235
1236 #if wxUSE_FONTMAP
1237
1238 class wxMBConv_wxwin : public wxMBConv
1239 {
1240 private:
1241     void Init()
1242     {
1243         m_ok = m2w.Init(m_enc, wxFONTENCODING_UNICODE) &&
1244                w2m.Init(wxFONTENCODING_UNICODE, m_enc);
1245     }
1246
1247 public:
1248     // temporarily just use wxEncodingConverter stuff,
1249     // so that it works while a better implementation is built
1250     wxMBConv_wxwin(const wxChar* name)
1251     {
1252         if (name)
1253             m_enc = wxFontMapper::Get()->CharsetToEncoding(name, false);
1254         else
1255             m_enc = wxFONTENCODING_SYSTEM;
1256
1257         Init();
1258     }
1259
1260     wxMBConv_wxwin(wxFontEncoding enc)
1261     {
1262         m_enc = enc;
1263
1264         Init();
1265     }
1266
1267     size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
1268     {
1269         size_t inbuf = strlen(psz);
1270         if (buf)
1271             m2w.Convert(psz,buf);
1272         return inbuf;
1273     }
1274
1275     size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
1276     {
1277         const size_t inbuf = wxWcslen(psz);
1278         if (buf)
1279             w2m.Convert(psz,buf);
1280
1281         return inbuf;
1282     }
1283
1284     bool IsOk() const { return m_ok; }
1285
1286 public:
1287     wxFontEncoding m_enc;
1288     wxEncodingConverter m2w, w2m;
1289
1290     // were we initialized successfully?
1291     bool m_ok;
1292
1293     DECLARE_NO_COPY_CLASS(wxMBConv_wxwin)
1294 };
1295
1296 #endif // wxUSE_FONTMAP
1297
1298 // ============================================================================
1299 // wxCSConv implementation
1300 // ============================================================================
1301
1302 void wxCSConv::Init()
1303 {
1304     m_name = NULL;
1305     m_convReal =  NULL;
1306     m_deferred = true;
1307 }
1308
1309 // find a valid value for the encoding
1310 void wxCSConv::SetEncoding()
1311 {
1312 #if wxUSE_INTL
1313     m_encoding = wxLocale::GetSystemEncoding();
1314 #else
1315     m_encoding = wxFONTENCODING_SYSTEM;
1316 #endif
1317 }
1318
1319 wxCSConv::wxCSConv(const wxChar *charset)
1320 {
1321     Init();
1322
1323     if ( charset )
1324     {
1325         // not used
1326         m_encoding = wxFONTENCODING_SYSTEM;
1327
1328         SetName(charset);
1329     }
1330     else // no charset specified
1331     {
1332         SetEncoding();
1333     }
1334 }
1335
1336 wxCSConv::wxCSConv(wxFontEncoding encoding)
1337 {
1338     if ( encoding == wxFONTENCODING_MAX ||
1339             encoding == wxFONTENCODING_DEFAULT )
1340     {
1341         wxFAIL_MSG( _T("invalid encoding value in wxCSConv ctor") );
1342
1343         encoding = wxFONTENCODING_SYSTEM;
1344     }
1345
1346     Init();
1347
1348     if ( encoding == wxFONTENCODING_SYSTEM )
1349     {
1350         SetEncoding();
1351     }
1352     else // have valid encoding, use it
1353     {
1354         m_encoding = encoding;
1355     }
1356 }
1357
1358 wxCSConv::~wxCSConv()
1359 {
1360     Clear();
1361 }
1362
1363 wxCSConv::wxCSConv(const wxCSConv& conv)
1364         : wxMBConv()
1365 {
1366     Init();
1367
1368     SetName(conv.m_name);
1369     m_encoding = conv.m_encoding;
1370 }
1371
1372 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
1373 {
1374     Clear();
1375
1376     SetName(conv.m_name);
1377     m_encoding = conv.m_encoding;
1378
1379     return *this;
1380 }
1381
1382 void wxCSConv::Clear()
1383 {
1384     free(m_name);
1385     delete m_convReal;
1386
1387     m_name = NULL;
1388     m_convReal = NULL;
1389 }
1390
1391 void wxCSConv::SetName(const wxChar *charset)
1392 {
1393     if (charset)
1394     {
1395         m_name = wxStrdup(charset);
1396         m_deferred = true;
1397     }
1398 }
1399
1400 static inline bool DoesntNeedConv(wxFontEncoding enc)
1401 {
1402     return enc == wxFONTENCODING_DEFAULT ||
1403             enc == wxFONTENCODING_SYSTEM ||
1404              enc == wxFONTENCODING_ISO8859_1;
1405 }
1406
1407 wxMBConv *wxCSConv::DoCreate() const
1408 {
1409 #if wxUSE_FONTMAP
1410     wxFontMapper * const fontMapper = wxFontMapper::Get();
1411
1412     wxFontEncoding encFromName = m_name ? fontMapper->CharsetToEncoding(m_name)
1413                                         : wxFONTENCODING_SYSTEM;
1414 #endif // wxUSE_FONTMAP
1415
1416     // check for the special case of ASCII charset
1417     if ( (!m_name && DoesntNeedConv(m_encoding))
1418 #if wxUSE_FONTMAP
1419             || (m_name && DoesntNeedConv(encFromName))
1420 #endif // wxUSE_FONTMAP
1421        )
1422     {
1423         // don't convert at all
1424         return NULL;
1425     }
1426
1427     // we trust OS to do conversion better than we can so try external
1428     // conversion methods first
1429     //
1430     // the full order is:
1431     //      1. OS conversion (iconv() under Unix or Win32 API)
1432     //      2. hard coded conversions for UTF
1433     //      3. wxEncodingConverter as fall back
1434
1435     // step (1)
1436 #ifdef HAVE_ICONV
1437     if ( m_name )
1438     {
1439         wxMBConv_iconv *conv = new wxMBConv_iconv(m_name);
1440         if ( conv->IsOk() )
1441             return conv;
1442
1443         delete conv;
1444     }
1445 #endif // HAVE_ICONV
1446
1447 #ifdef wxHAVE_WIN32_MB2WC
1448     {
1449         wxMBConv_win32 *conv = m_name ? new wxMBConv_win32(m_name)
1450                                       : new wxMBConv_win32(m_encoding);
1451         if ( conv->IsOk() )
1452             return conv;
1453
1454         delete conv;
1455     }
1456 #endif // wxHAVE_WIN32_MB2WC
1457
1458     // step (2)
1459     wxFontEncoding enc = m_encoding;
1460 #if wxUSE_FONTMAP
1461     if ( enc == wxFONTENCODING_SYSTEM )
1462         enc = encFromName;
1463 #endif // wxUSE_FONTMAP
1464
1465     switch ( enc )
1466     {
1467         case wxFONTENCODING_UTF7:
1468              return new wxMBConvUTF7;
1469
1470         case wxFONTENCODING_UTF8:
1471              return new wxMBConvUTF8;
1472
1473         case wxFONTENCODING_UTF16:
1474              return new wxMBConvUTF16;
1475
1476         case wxFONTENCODING_UTF16BE:
1477              return new wxMBConvUTF16BE;
1478
1479         case wxFONTENCODING_UTF16LE:
1480              return new wxMBConvUTF16LE;
1481
1482         case wxFONTENCODING_UTF32:
1483              return new wxMBConvUTF32;
1484
1485         case wxFONTENCODING_UTF32BE:
1486              return new wxMBConvUTF32BE;
1487
1488         case wxFONTENCODING_UTF32LE:
1489              return new wxMBConvUTF32LE;
1490
1491         default:
1492              // nothing to do but put here to suppress gcc warnings
1493              ;
1494     }
1495
1496     // step (3)
1497 #if wxUSE_FONTMAP
1498     {
1499         wxMBConv_wxwin *conv = m_name ? new wxMBConv_wxwin(m_name)
1500                                       : new wxMBConv_wxwin(m_encoding);
1501         if ( conv->IsOk() )
1502             return conv;
1503
1504         delete conv;
1505     }
1506 #endif // wxUSE_FONTMAP
1507
1508     wxLogError(_("Cannot convert from the charset '%s'!"),
1509                m_name ? m_name
1510                       :
1511 #if wxUSE_FONTMAP
1512                          wxFontMapper::GetEncodingDescription(m_encoding).c_str()
1513 #else // !wxUSE_FONTMAP
1514                          wxString::Format(_("encoding %s"), m_encoding).c_str()
1515 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
1516               );
1517
1518     return NULL;
1519 }
1520
1521 void wxCSConv::CreateConvIfNeeded() const
1522 {
1523     if ( m_deferred )
1524     {
1525         wxCSConv *self = (wxCSConv *)this; // const_cast
1526         self->m_convReal = DoCreate();
1527         self->m_deferred = false;
1528     }
1529 }
1530
1531 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1532 {
1533     CreateConvIfNeeded();
1534
1535     if (m_convReal)
1536         return m_convReal->MB2WC(buf, psz, n);
1537
1538     // latin-1 (direct)
1539     size_t len = strlen(psz);
1540
1541     if (buf)
1542     {
1543         for (size_t c = 0; c <= len; c++)
1544             buf[c] = (unsigned char)(psz[c]);
1545     }
1546
1547     return len;
1548 }
1549
1550 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1551 {
1552     CreateConvIfNeeded();
1553
1554     if (m_convReal)
1555         return m_convReal->WC2MB(buf, psz, n);
1556
1557     // latin-1 (direct)
1558     const size_t len = wxWcslen(psz);
1559     if (buf)
1560     {
1561         for (size_t c = 0; c <= len; c++)
1562             buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
1563     }
1564
1565     return len;
1566 }
1567
1568 #endif // wxUSE_WCHAR_T
1569
1570