src/common/strconv.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        strconv.cpp
   3 // Purpose:     Unicode conversion classes
   4 // Author:      Ove Kaaven, Robert Roebling, Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1999 Ove Kaaven, Robert Roebling, Vadim Zeitlin
   9 // Licence:     wxWindows license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 // ============================================================================
  13 // declarations
  14 // ============================================================================
  15
  16 // ----------------------------------------------------------------------------
  17 // headers
  18 // ----------------------------------------------------------------------------
  19
  20 #ifdef __GNUG__
  21   #pragma implementation "strconv.h"
  22 #endif
  23
  24 // For compilers that support precompilation, includes "wx.h".
  25 #include "wx/wxprec.h"
  26
  27 #ifdef __BORLANDC__
  28   #pragma hdrstop
  29 #endif
  30
  31 #ifdef __WXMSW__
  32   #include "wx/msw/private.h"
  33 #endif
  34
  35 #include <errno.h>
  36 #include <ctype.h>
  37 #include <string.h>
  38 #include <stdlib.h>
  39
  40 #ifdef __SALFORDC__
  41   #include <clib.h>
  42 #endif
  43
  44 #ifdef HAVE_ICONV_H
  45   #include <iconv.h>
  46 #endif
  47
  48 #ifdef __WXMSW__
  49   #include <windows.h>
  50 #endif
  51
  52 #include "wx/debug.h"
  53 #include "wx/strconv.h"
  54 #include "wx/intl.h"
  55 #include "wx/log.h"
  56
  57 #if defined(WORDS_BIGENDIAN) || defined(__STDC_ISO_10646__)
  58 #define BSWAP_UCS4(str, len)
  59 #define BSWAP_UCS2(str, len)
  60 #else
  61 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
  62 #define BSWAP_UCS2(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
  63 #define WC_NEED_BSWAP
  64 #endif
  65 #define BSWAP_UTF32(str, len) BSWAP_UCS4(str, len)
  66 #define BSWAP_UTF16(str, len) BSWAP_UCS2(str, len)
  67
  68 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
  69 // it might be not defined - assume the most common value
  70 #ifndef SIZEOF_WCHAR_T
  71     #define SIZEOF_WCHAR_T 2
  72 #endif // !defined(SIZEOF_WCHAR_T)
  73
  74 #if SIZEOF_WCHAR_T == 4
  75     #define WC_NAME "UCS4"
  76     #define WC_BSWAP BSWAP_UCS4
  77 #elif SIZEOF_WCHAR_T == 2
  78     #define WC_NAME "UTF16"
  79     #define WC_BSWAP BSWAP_UTF16
  80     #define WC_UTF16
  81 #else // sizeof(wchar_t) != 2 nor 4
  82     // I don't know what to do about this
  83     #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
  84 #endif
  85
  86 // ----------------------------------------------------------------------------
  87 // globals
  88 // ----------------------------------------------------------------------------
  89
  90 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
  91
  92 // ============================================================================
  93 // implementation
  94 // ============================================================================
  95
  96 #if wxUSE_WCHAR_T
  97
  98 #ifdef WC_UTF16
  99
 100 static size_t encode_utf16(wxUint32 input, wchar_t *output)
 101 {
 102     if (input<=0xffff)
 103     {
 104         if (output) *output++ = input;
 105         return 1;
 106     }
 107     else if (input>=0x110000)
 108     {
 109         return (size_t)-1;
 110     }
 111     else
 112     {
 113         if (output)
 114         {
 115             *output++ = (input >> 10)+0xd7c0;
 116             *output++ = (input&0x3ff)+0xdc00;
 117         }
 118         return 2;
 119     }
 120 }
 121
 122 static size_t decode_utf16(const wchar_t* input, wxUint32& output)
 123 {
 124     if ((*input<0xd800) || (*input>0xdfff))
 125     {
 126         output = *input;
 127         return 1;
 128     }
 129     else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
 130     {
 131         output = *input;
 132         return (size_t)-1;
 133     }
 134     else
 135     {
 136         output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
 137         return 2;
 138     }
 139 }
 140
 141 #endif // WC_UTF16
 142
 143 // ----------------------------------------------------------------------------
 144 // wxMBConv
 145 // ----------------------------------------------------------------------------
 146
 147 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
 148
 149 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 150 {
 151     return wxMB2WC(buf, psz, n);
 152 }
 153
 154 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 155 {
 156     return wxWC2MB(buf, psz, n);
 157 }
 158
 159 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 160 {
 161     if (psz)
 162     {
 163         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
 164         if (nLen == (size_t)-1)
 165             return wxWCharBuffer((wchar_t *) NULL);
 166         wxWCharBuffer buf(nLen);
 167         MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen);
 168         return buf;
 169     }
 170     else
 171         return wxWCharBuffer((wchar_t *) NULL);
 172 }
 173
 174 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
 175 {
 176     if (psz)
 177     {
 178         size_t nLen = WC2MB((char *) NULL, psz, 0);
 179         if (nLen == (size_t)-1)
 180             return wxCharBuffer((char *) NULL);
 181         wxCharBuffer buf(nLen);
 182         WC2MB((char *)(const char *) buf, psz, nLen);
 183         return buf;
 184     }
 185     else
 186         return wxCharBuffer((char *) NULL);
 187 }
 188
 189 // ----------------------------------------------------------------------------
 190 // standard file conversion
 191 // ----------------------------------------------------------------------------
 192
 193 WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
 194
 195 // just use the libc conversion for now
 196 size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 197 {
 198     return wxMB2WC(buf, psz, n);
 199 }
 200
 201 size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 202 {
 203     return wxWC2MB(buf, psz, n);
 204 }
 205
 206 // ----------------------------------------------------------------------------
 207 // standard gdk conversion
 208 // ----------------------------------------------------------------------------
 209
 210 #ifdef __WXGTK12__
 211
 212 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
 213
 214 #include <gdk/gdk.h>
 215
 216 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 217 {
 218     if (buf)
 219     {
 220         return gdk_mbstowcs((GdkWChar *)buf, psz, n);
 221     }
 222     else
 223     {
 224         GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
 225         size_t len = gdk_mbstowcs(nbuf, psz, n);
 226         delete[] nbuf;
 227         return len;
 228     }
 229 }
 230
 231 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 232 {
 233     char *mbstr = gdk_wcstombs((GdkWChar *)psz);
 234     size_t len = mbstr ? strlen(mbstr) : 0;
 235     if (buf)
 236     {
 237         if (len > n)
 238             len = n;
 239         memcpy(buf, psz, len);
 240         if (len < n)
 241             buf[len] = 0;
 242     }
 243     return len;
 244 }
 245
 246 #endif // GTK > 1.0
 247
 248 // ----------------------------------------------------------------------------
 249 // UTF-7
 250 // ----------------------------------------------------------------------------
 251
 252 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
 253
 254 #if 0
 255 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 256                         "abcdefghijklmnopqrstuvwxyz"
 257                         "0123456789'(),-./:?";
 258 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
 259 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 260                         "abcdefghijklmnopqrstuvwxyz"
 261                         "0123456789+/";
 262 #endif
 263
 264 // TODO: write actual implementations of UTF-7 here
 265 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
 266                            const char * WXUNUSED(psz),
 267                            size_t WXUNUSED(n)) const
 268 {
 269   return 0;
 270 }
 271
 272 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
 273                            const wchar_t * WXUNUSED(psz),
 274                            size_t WXUNUSED(n)) const
 275 {
 276   return 0;
 277 }
 278
 279 // ----------------------------------------------------------------------------
 280 // UTF-8
 281 // ----------------------------------------------------------------------------
 282
 283 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
 284
 285 static wxUint32 utf8_max[]=
 286     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 287
 288 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 289 {
 290     size_t len = 0;
 291
 292     while (*psz && ((!buf) || (len < n)))
 293     {
 294         unsigned char cc = *psz++, fc = cc;
 295         unsigned cnt;
 296         for (cnt = 0; fc & 0x80; cnt++)
 297             fc <<= 1;
 298         if (!cnt)
 299         {
 300             // plain ASCII char
 301             if (buf)
 302                 *buf++ = cc;
 303             len++;
 304         }
 305         else
 306         {
 307             cnt--;
 308             if (!cnt)
 309             {
 310                 // invalid UTF-8 sequence
 311                 return (size_t)-1;
 312             }
 313             else
 314             {
 315                 unsigned ocnt = cnt - 1;
 316                 wxUint32 res = cc & (0x3f >> cnt);
 317                 while (cnt--)
 318                 {
 319                     cc = *psz++;
 320                     if ((cc & 0xC0) != 0x80)
 321                     {
 322                         // invalid UTF-8 sequence
 323                         return (size_t)-1;
 324                     }
 325                     res = (res << 6) | (cc & 0x3f);
 326                 }
 327                 if (res <= utf8_max[ocnt])
 328                 {
 329                     // illegal UTF-8 encoding
 330                     return (size_t)-1;
 331                 }
 332 #ifdef WC_UTF16
 333                 size_t pa = encode_utf16(res, buf);
 334                 if (pa == (size_t)-1)
 335                   return (size_t)-1;
 336                 if (buf)
 337                     buf += pa;
 338                 len += pa;
 339 #else
 340                 if (buf)
 341                     *buf++ = res;
 342                 len++;
 343 #endif
 344             }
 345         }
 346     }
 347     if (buf && (len < n))
 348         *buf = 0;
 349     return len;
 350 }
 351
 352 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 353 {
 354     size_t len = 0;
 355
 356     while (*psz && ((!buf) || (len < n)))
 357     {
 358         wxUint32 cc;
 359 #ifdef WC_UTF16
 360         size_t pa = decode_utf16(psz, cc);
 361         psz += (pa == (size_t)-1) ? 1 : pa;
 362 #else
 363         cc=(*psz++) & 0x7fffffff;
 364 #endif
 365         unsigned cnt;
 366         for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
 367         if (!cnt)
 368         {
 369             // plain ASCII char
 370             if (buf)
 371                 *buf++ = cc;
 372             len++;
 373         }
 374
 375         else
 376         {
 377             len += cnt + 1;
 378             if (buf)
 379             {
 380                 *buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt));
 381                 while (cnt--)
 382                     *buf++ = 0x80 | ((cc >> (cnt * 6)) & 0x3f);
 383             }
 384         }
 385     }
 386
 387     if (buf && (len<n)) *buf = 0;
 388     return len;
 389 }
 390
 391 // ----------------------------------------------------------------------------
 392 // specified character set
 393 // ----------------------------------------------------------------------------
 394
 395 WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
 396
 397 #include "wx/encconv.h"
 398 #include "wx/fontmap.h"
 399
 400 // TODO: add some tables here
 401 // - perhaps common encodings to common codepages (for Win32)
 402 // - perhaps common encodings to objects ("UTF8" -> wxConvUTF8)
 403 // - move wxEncodingConverter meat in here
 404
 405 #ifdef __WIN32__
 406 #include "wx/msw/registry.h"
 407 // this should work if M$ Internet Exploiter is installed
 408 static long CharsetToCodepage(const wxChar *name)
 409 {
 410     if (!name)
 411         return GetACP();
 412
 413     long CP=-1;
 414
 415     wxString cn(name);
 416     do {
 417         wxString path(wxT("MIME\\Database\\Charset\\"));
 418         path += cn;
 419         wxRegKey key(wxRegKey::HKCR, path);
 420
 421         if (!key.Exists()) break;
 422
 423         // two cases: either there's an AliasForCharset string,
 424         // or there are Codepage and InternetEncoding dwords.
 425         // The InternetEncoding gives us the actual encoding,
 426         // the Codepage just says which Windows character set to
 427         // use when displaying the data.
 428         if (key.HasValue(wxT("InternetEncoding")) &&
 429             key.QueryValue(wxT("InternetEncoding"), &CP)) break;
 430
 431         // no encoding, see if it's an alias
 432         if (!key.HasValue(wxT("AliasForCharset")) ||
 433             !key.QueryValue(wxT("AliasForCharset"), cn)) break;
 434     } while (1);
 435
 436     return CP;
 437 }
 438 #endif
 439
 440 class wxCharacterSet
 441 {
 442 public:
 443     wxCharacterSet(const wxChar*name)
 444         : cname(name) {}
 445     virtual ~wxCharacterSet()
 446         {}
 447     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 448         { return (size_t)-1; }
 449     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 450         { return (size_t)-1; }
 451     virtual bool usable()
 452         { return FALSE; }
 453 public:
 454     const wxChar*cname;
 455 };
 456
 457 class ID_CharSet : public wxCharacterSet
 458 {
 459 public:
 460     ID_CharSet(const wxChar *name,wxMBConv *cnv)
 461         : wxCharacterSet(name), work(cnv) {}
 462
 463     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 464         { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
 465
 466     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 467         { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
 468
 469     bool usable()
 470         { return work!=NULL; }
 471 public:
 472     wxMBConv*work;
 473 };
 474
 475
 476 #ifdef HAVE_ICONV_H
 477
 478 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
 479 //     if output buffer is _exactly_ as big as needed. Such case is (unless there's
 480 //     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
 481 //     (which means error) and says there are 0 bytes left in the input buffer --
 482 //     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
 483 //     this alternative test for iconv() failure.
 484 //     [This bug does not appear in glibc 2.2.]
 485 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
 486 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
 487                                      (errno != E2BIG || bufLeft != 0))
 488 #else
 489 #define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
 490 #endif
 491
 492 class IC_CharSet : public wxCharacterSet
 493 {
 494 public:
 495     IC_CharSet(const wxChar *name)
 496         : wxCharacterSet(name)
 497     {
 498         m2w = iconv_open(WC_NAME, wxConvLibc.cWX2MB(cname));
 499         w2m = iconv_open(wxConvLibc.cWX2MB(cname), WC_NAME);
 500     }
 501
 502     ~IC_CharSet()
 503     {
 504         if ( m2w != (iconv_t)-1 )
 505             iconv_close(m2w);
 506         if ( w2m != (iconv_t)-1 )
 507             iconv_close(w2m);
 508     }
 509
 510     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 511     {
 512         size_t inbuf = strlen(psz);
 513         size_t outbuf = n * SIZEOF_WCHAR_T;
 514         size_t res, cres;
 515         // VS: Use these instead of psz, buf because iconv() modifies its arguments:
 516         wchar_t *bufPtr = buf;
 517         const char *pszPtr = psz;
 518
 519         if (buf)
 520         {
 521             // have destination buffer, convert there
 522 #ifdef WX_ICONV_TAKES_CHAR
 523             cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
 524 #else
 525             cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
 526 #endif
 527             res = n - (outbuf / SIZEOF_WCHAR_T);
 528             // convert to native endianness
 529 #ifdef WC_NEED_BSWAP
 530             WC_BSWAP(buf /* _not_ bufPtr */, res)
 531 #endif
 532         }
 533         else
 534         {
 535             // no destination buffer... convert using temp buffer
 536             // to calculate destination buffer requirement
 537             wchar_t tbuf[8];
 538             res = 0;
 539             do {
 540                 bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T;
 541 #ifdef WX_ICONV_TAKES_CHAR
 542                 cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
 543 #else
 544                 cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
 545 #endif
 546                 res += 8-(outbuf/SIZEOF_WCHAR_T);
 547             } while ((cres==(size_t)-1) && (errno==E2BIG));
 548         }
 549
 550         if (ICONV_FAILED(cres, inbuf))
 551             return (size_t)-1;
 552
 553         return res;
 554     }
 555
 556     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 557     {
 558 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 559         size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
 560 #else
 561         size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T;
 562 #endif
 563         size_t outbuf = n;
 564         size_t res, cres;
 565
 566 #ifdef WC_NEED_BSWAP
 567         // need to copy to temp buffer to switch endianness
 568         // this absolutely doesn't rock!
 569         // (no, doing WC_BSWAP twice on the original buffer won't help, as it
 570         //  could be in read-only memory, or be accessed in some other thread)
 571         wchar_t *tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
 572         memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
 573         WC_BSWAP(tmpbuf, inbuf)
 574         psz=tmpbuf;
 575 #endif
 576         if (buf)
 577         {
 578             // have destination buffer, convert there
 579 #ifdef WX_ICONV_TAKES_CHAR
 580             cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
 581 #else
 582             cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
 583 #endif
 584             res = n-outbuf;
 585         }
 586         else
 587         {
 588             // no destination buffer... convert using temp buffer
 589             // to calculate destination buffer requirement
 590             char tbuf[16];
 591             res = 0;
 592             do {
 593                 buf = tbuf; outbuf = 16;
 594 #ifdef WX_ICONV_TAKES_CHAR
 595                 cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
 596 #else
 597                 cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
 598 #endif
 599                 res += 16 - outbuf;
 600             } while ((cres==(size_t)-1) && (errno==E2BIG));
 601         }
 602 #ifdef WC_NEED_BSWAP
 603         free(tmpbuf);
 604 #endif
 605         if (ICONV_FAILED(cres, inbuf))
 606             return (size_t)-1;
 607
 608         return res;
 609     }
 610
 611     bool usable()
 612         { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
 613
 614 public:
 615     iconv_t m2w, w2m;
 616 };
 617 #endif
 618
 619 #ifdef __WIN32__
 620 class CP_CharSet : public wxCharacterSet
 621 {
 622 public:
 623     CP_CharSet(const wxChar*name)
 624         : wxCharacterSet(name), CodePage(CharsetToCodepage(name)) {}
 625
 626     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 627     {
 628         size_t len =
 629             MultiByteToWideChar(CodePage, 0, psz, -1, buf, buf ? n : 0);
 630         return len ? len : (size_t)-1;
 631     }
 632
 633     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 634     {
 635         size_t len = WideCharToMultiByte(CodePage, 0, psz, -1, buf,
 636                                          buf ? n : 0, NULL, NULL);
 637         return len ? len : (size_t)-1;
 638     }
 639
 640     bool usable()
 641         { return CodePage != -1; }
 642
 643 public:
 644     long CodePage;
 645 };
 646 #endif
 647
 648 class EC_CharSet : public wxCharacterSet
 649 {
 650 public:
 651     // temporarily just use wxEncodingConverter stuff,
 652     // so that it works while a better implementation is built
 653     EC_CharSet(const wxChar*name) : wxCharacterSet(name),
 654                                     enc(wxFONTENCODING_SYSTEM)
 655     {
 656         if (name)
 657             enc = wxTheFontMapper->CharsetToEncoding(name, FALSE);
 658         m2w.Init(enc, wxFONTENCODING_UNICODE);
 659         w2m.Init(wxFONTENCODING_UNICODE, enc);
 660     }
 661
 662     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 663     {
 664         size_t inbuf = strlen(psz);
 665         if (buf)
 666             m2w.Convert(psz,buf);
 667         return inbuf;
 668     }
 669
 670     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 671     {
 672 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 673         size_t inbuf = std::wcslen(psz);
 674 #else
 675         size_t inbuf = ::wcslen(psz);
 676 #endif
 677         if (buf)
 678             w2m.Convert(psz,buf);
 679
 680         return inbuf;
 681     }
 682
 683     bool usable()
 684         { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); }
 685
 686 public:
 687     wxFontEncoding enc;
 688     wxEncodingConverter m2w, w2m;
 689 };
 690
 691 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
 692 {
 693     wxCharacterSet *cset = NULL;
 694     if (name)
 695     {
 696         if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
 697         {
 698             cset = new ID_CharSet(name, &wxConvUTF8);
 699         }
 700         else
 701         {
 702 #ifdef HAVE_ICONV_H
 703             cset = new IC_CharSet(name); // may not take NULL
 704 #endif
 705         }
 706     }
 707
 708     if (cset && cset->usable()) return cset;
 709     if (cset)
 710     {
 711         delete cset;
 712         cset = NULL;
 713     }
 714
 715 #ifdef __WIN32__
 716     cset = new CP_CharSet(name); // may take NULL
 717     if (cset->usable())
 718         return cset;
 719
 720     delete cset;
 721 #endif // __WIN32__
 722
 723     cset = new EC_CharSet(name);
 724     if (cset->usable())
 725         return cset;
 726
 727     delete cset;
 728     wxLogError(_("Unknown encoding '%s'!"), name);
 729     return NULL;
 730 }
 731
 732 wxCSConv::wxCSConv(const wxChar *charset)
 733 {
 734     m_name = (wxChar *)NULL;
 735     m_cset = (wxCharacterSet *) NULL;
 736     m_deferred = TRUE;
 737
 738     SetName(charset);
 739 }
 740
 741 wxCSConv::~wxCSConv()
 742 {
 743     free(m_name);
 744     delete m_cset;
 745 }
 746
 747 void wxCSConv::SetName(const wxChar *charset)
 748 {
 749     if (charset)
 750     {
 751         m_name = wxStrdup(charset);
 752         m_deferred = TRUE;
 753     }
 754 }
 755
 756 void wxCSConv::LoadNow()
 757 {
 758     if (m_deferred)
 759     {
 760         if ( !m_name )
 761         {
 762             wxString name = wxLocale::GetSystemEncodingName();
 763             if ( !name.empty() )
 764                 SetName(name);
 765         }
 766
 767         m_cset = wxGetCharacterSet(m_name);
 768         m_deferred = FALSE;
 769     }
 770 }
 771
 772 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 773 {
 774     ((wxCSConv *)this)->LoadNow(); // discard constness
 775
 776     if (m_cset)
 777         return m_cset->MB2WC(buf, psz, n);
 778
 779     // latin-1 (direct)
 780     size_t len = strlen(psz);
 781
 782     if (buf)
 783     {
 784         for (size_t c = 0; c <= len; c++)
 785             buf[c] = (unsigned char)(psz[c]);
 786     }
 787
 788     return len;
 789 }
 790
 791 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 792 {
 793     ((wxCSConv *)this)->LoadNow(); // discard constness
 794
 795     if (m_cset)
 796         return m_cset->WC2MB(buf, psz, n);
 797
 798     // latin-1 (direct)
 799 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 800     size_t len=std::wcslen(psz);
 801 #else
 802     size_t len=::wcslen(psz);
 803 #endif
 804     if (buf)
 805     {
 806         for (size_t c = 0; c <= len; c++)
 807             buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
 808     }
 809
 810     return len;
 811 }
 812
 813 #ifdef HAVE_ICONV_H
 814
 815 class IC_CharSetConverter
 816 {
 817 public:
 818     IC_CharSetConverter(IC_CharSet *from, IC_CharSet *to)
 819     {
 820         cnv = iconv_open(wxConvLibc.cWX2MB(to->cname),
 821                          wxConvLibc.cWX2MB(from->cname));
 822     }
 823
 824     ~IC_CharSetConverter()
 825     {
 826         if (cnv != (iconv_t)-1)
 827             iconv_close(cnv);
 828     }
 829
 830     size_t Convert(char *buf, const char *psz, size_t n)
 831     {
 832         size_t inbuf = strlen(psz);
 833         size_t outbuf = n;
 834 #ifdef WX_ICONV_TAKES_CHAR
 835         size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf );
 836 #else
 837         size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf );
 838 #endif
 839         if (res == (size_t)-1)
 840             return (size_t)-1;
 841         return (n - outbuf);
 842     }
 843
 844 public:
 845     iconv_t cnv;
 846 };
 847
 848 #endif // HAVE_ICONV_H
 849
 850 class EC_CharSetConverter
 851 {
 852 public:
 853     EC_CharSetConverter(EC_CharSet*from,EC_CharSet*to)
 854         { cnv.Init(from->enc,to->enc); }
 855
 856     size_t Convert(char*buf, const char*psz, size_t n)
 857     {
 858         size_t inbuf = strlen(psz);
 859         if (buf) cnv.Convert(psz,buf);
 860         return inbuf;
 861     }
 862
 863 public:
 864     wxEncodingConverter cnv;
 865 };
 866
 867 #else // !wxUSE_WCHAR_T
 868
 869 // ----------------------------------------------------------------------------
 870 // stand-ins in absence of wchar_t
 871 // ----------------------------------------------------------------------------
 872
 873 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
 874
 875 #endif // wxUSE_WCHAR_T
 876
 877