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, Vadim Zeitlin, Vaclav Slavik
   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 #ifndef WX_PRECOMP
  32     #include "wx/intl.h"
  33     #include "wx/log.h"
  34 #endif // WX_PRECOMP
  35
  36 #ifdef __WXMSW__
  37     #include "wx/msw/private.h"
  38 #endif
  39
  40 #include <errno.h>
  41 #include <ctype.h>
  42 #include <string.h>
  43 #include <stdlib.h>
  44
  45 #include "wx/module.h"
  46 #include "wx/strconv.h"
  47
  48 // ----------------------------------------------------------------------------
  49 // globals
  50 // ----------------------------------------------------------------------------
  51
  52 #if wxUSE_WCHAR_T
  53     WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
  54     WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
  55 #else
  56     // stand-ins in absence of wchar_t
  57     WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
  58 #endif // wxUSE_WCHAR_T
  59
  60 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
  61
  62 class wxStrConvModule: public wxModule
  63 {
  64 public:
  65     wxStrConvModule() : wxModule() { }
  66     virtual bool OnInit() { return TRUE; }
  67     virtual void OnExit()
  68     {
  69 #if wxUSE_WCHAR_T
  70         wxConvLocal.Clear();
  71 #endif
  72     }
  73
  74     DECLARE_DYNAMIC_CLASS(wxStrConvModule)
  75 };
  76
  77 IMPLEMENT_DYNAMIC_CLASS(wxStrConvModule, wxModule)
  78
  79
  80 // ----------------------------------------------------------------------------
  81 // headers
  82 // ----------------------------------------------------------------------------
  83
  84 #if wxUSE_WCHAR_T
  85
  86 #ifdef __SALFORDC__
  87     #include <clib.h>
  88 #endif
  89
  90 #ifdef HAVE_ICONV
  91     #include <iconv.h>
  92 #endif
  93
  94 #include "wx/encconv.h"
  95 #include "wx/fontmap.h"
  96
  97 // ----------------------------------------------------------------------------
  98 // macros
  99 // ----------------------------------------------------------------------------
 100
 101 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
 102 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
 103
 104 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
 105 // it might be not defined - assume the most common value
 106 #ifndef SIZEOF_WCHAR_T
 107     #define SIZEOF_WCHAR_T 2
 108 #endif // !defined(SIZEOF_WCHAR_T)
 109
 110 #if SIZEOF_WCHAR_T == 4
 111     #define WC_NAME         "UCS4"
 112     #define WC_BSWAP         BSWAP_UCS4
 113     #ifdef WORDS_BIGENDIAN
 114       #define WC_NAME_BEST  "UCS-4BE"
 115     #else
 116       #define WC_NAME_BEST  "UCS-4LE"
 117     #endif
 118 #elif SIZEOF_WCHAR_T == 2
 119     #define WC_NAME         "UTF16"
 120     #define WC_BSWAP         BSWAP_UTF16
 121     #define WC_UTF16
 122     #ifdef WORDS_BIGENDIAN
 123       #define WC_NAME_BEST  "UTF-16BE"
 124     #else
 125       #define WC_NAME_BEST  "UTF-16LE"
 126     #endif
 127 #else // sizeof(wchar_t) != 2 nor 4
 128     // I don't know what to do about this
 129     #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
 130 #endif
 131
 132 // ============================================================================
 133 // implementation
 134 // ============================================================================
 135
 136 // ----------------------------------------------------------------------------
 137 // UTF-16 en/decoding
 138 // ----------------------------------------------------------------------------
 139
 140 #ifdef WC_UTF16
 141
 142 static size_t encode_utf16(wxUint32 input, wchar_t *output)
 143 {
 144     if (input<=0xffff)
 145     {
 146         if (output) *output++ = (wchar_t) input;
 147         return 1;
 148     }
 149     else if (input>=0x110000)
 150     {
 151         return (size_t)-1;
 152     }
 153     else
 154     {
 155         if (output)
 156         {
 157             *output++ = (wchar_t) ((input >> 10)+0xd7c0);
 158             *output++ = (wchar_t) ((input&0x3ff)+0xdc00);
 159         }
 160         return 2;
 161     }
 162 }
 163
 164 static size_t decode_utf16(const wchar_t* input, wxUint32& output)
 165 {
 166     if ((*input<0xd800) || (*input>0xdfff))
 167     {
 168         output = *input;
 169         return 1;
 170     }
 171     else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
 172     {
 173         output = *input;
 174         return (size_t)-1;
 175     }
 176     else
 177     {
 178         output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
 179         return 2;
 180     }
 181 }
 182
 183 #endif // WC_UTF16
 184
 185 // ----------------------------------------------------------------------------
 186 // wxMBConv
 187 // ----------------------------------------------------------------------------
 188
 189 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 190 {
 191     return wxMB2WC(buf, psz, n);
 192 }
 193
 194 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 195 {
 196     return wxWC2MB(buf, psz, n);
 197 }
 198
 199 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 200 {
 201     if (psz)
 202     {
 203         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);  // return value excludes /0
 204         if (nLen == (size_t)-1)
 205             return wxWCharBuffer((wchar_t *) NULL);
 206         wxWCharBuffer buf(nLen);                        // this allocates nLen1+
 207         MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen+1);
 208         return buf;
 209     }
 210     else
 211         return wxWCharBuffer((wchar_t *) NULL);
 212 }
 213
 214 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
 215 {
 216     if (psz)
 217     {
 218         size_t nLen = WC2MB((char *) NULL, psz, 0);  // return value excludes /0
 219         if (nLen == (size_t)-1)
 220             return wxCharBuffer((char *) NULL);
 221         wxCharBuffer buf(nLen);                      // this allocates nLen+1
 222         WC2MB((char *)(const char *) buf, psz, nLen+1);
 223         return buf;
 224     }
 225     else
 226         return wxCharBuffer((char *) NULL);
 227 }
 228
 229 // ----------------------------------------------------------------------------
 230 // standard file conversion
 231 // ----------------------------------------------------------------------------
 232
 233 WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
 234
 235 // just use the libc conversion for now
 236 size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 237 {
 238     return wxMB2WC(buf, psz, n);
 239 }
 240
 241 size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 242 {
 243     return wxWC2MB(buf, psz, n);
 244 }
 245
 246 // ----------------------------------------------------------------------------
 247 // standard gdk conversion
 248 // ----------------------------------------------------------------------------
 249
 250 #ifdef __WXGTK12__
 251
 252 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
 253
 254 #include <gdk/gdk.h>
 255
 256 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 257 {
 258     if (buf)
 259     {
 260         return gdk_mbstowcs((GdkWChar *)buf, psz, n);
 261     }
 262     else
 263     {
 264         GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
 265         size_t len = gdk_mbstowcs(nbuf, psz, n);
 266         delete[] nbuf;
 267         return len;
 268     }
 269 }
 270
 271 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 272 {
 273     char *mbstr = gdk_wcstombs((GdkWChar *)psz);
 274     size_t len = mbstr ? strlen(mbstr) : 0;
 275     if (buf)
 276     {
 277         if (len > n)
 278             len = n;
 279         memcpy(buf, psz, len);
 280         if (len < n)
 281             buf[len] = 0;
 282     }
 283     return len;
 284 }
 285
 286 #endif // GTK > 1.0
 287
 288 // ----------------------------------------------------------------------------
 289 // UTF-7
 290 // ----------------------------------------------------------------------------
 291
 292 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
 293
 294 #if 0
 295 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 296                         "abcdefghijklmnopqrstuvwxyz"
 297                         "0123456789'(),-./:?";
 298 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
 299 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 300                         "abcdefghijklmnopqrstuvwxyz"
 301                         "0123456789+/";
 302 #endif
 303
 304 // TODO: write actual implementations of UTF-7 here
 305 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
 306                            const char * WXUNUSED(psz),
 307                            size_t WXUNUSED(n)) const
 308 {
 309   return 0;
 310 }
 311
 312 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
 313                            const wchar_t * WXUNUSED(psz),
 314                            size_t WXUNUSED(n)) const
 315 {
 316   return 0;
 317 }
 318
 319 // ----------------------------------------------------------------------------
 320 // UTF-8
 321 // ----------------------------------------------------------------------------
 322
 323 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
 324
 325 static wxUint32 utf8_max[]=
 326     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 327
 328 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 329 {
 330     size_t len = 0;
 331
 332     while (*psz && ((!buf) || (len < n)))
 333     {
 334         unsigned char cc = *psz++, fc = cc;
 335         unsigned cnt;
 336         for (cnt = 0; fc & 0x80; cnt++)
 337             fc <<= 1;
 338         if (!cnt)
 339         {
 340             // plain ASCII char
 341             if (buf)
 342                 *buf++ = cc;
 343             len++;
 344         }
 345         else
 346         {
 347             cnt--;
 348             if (!cnt)
 349             {
 350                 // invalid UTF-8 sequence
 351                 return (size_t)-1;
 352             }
 353             else
 354             {
 355                 unsigned ocnt = cnt - 1;
 356                 wxUint32 res = cc & (0x3f >> cnt);
 357                 while (cnt--)
 358                 {
 359                     cc = *psz++;
 360                     if ((cc & 0xC0) != 0x80)
 361                     {
 362                         // invalid UTF-8 sequence
 363                         return (size_t)-1;
 364                     }
 365                     res = (res << 6) | (cc & 0x3f);
 366                 }
 367                 if (res <= utf8_max[ocnt])
 368                 {
 369                     // illegal UTF-8 encoding
 370                     return (size_t)-1;
 371                 }
 372 #ifdef WC_UTF16
 373                 size_t pa = encode_utf16(res, buf);
 374                 if (pa == (size_t)-1)
 375                   return (size_t)-1;
 376                 if (buf)
 377                     buf += pa;
 378                 len += pa;
 379 #else // !WC_UTF16
 380                 if (buf)
 381                     *buf++ = res;
 382                 len++;
 383 #endif // WC_UTF16/!WC_UTF16
 384             }
 385         }
 386     }
 387     if (buf && (len < n))
 388         *buf = 0;
 389     return len;
 390 }
 391
 392 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 393 {
 394     size_t len = 0;
 395
 396     while (*psz && ((!buf) || (len < n)))
 397     {
 398         wxUint32 cc;
 399 #ifdef WC_UTF16
 400         size_t pa = decode_utf16(psz, cc);
 401         psz += (pa == (size_t)-1) ? 1 : pa;
 402 #else
 403         cc=(*psz++) & 0x7fffffff;
 404 #endif
 405         unsigned cnt;
 406         for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
 407         if (!cnt)
 408         {
 409             // plain ASCII char
 410             if (buf)
 411                 *buf++ = (char) cc;
 412             len++;
 413         }
 414
 415         else
 416         {
 417             len += cnt + 1;
 418             if (buf)
 419             {
 420                 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
 421                 while (cnt--)
 422                     *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
 423             }
 424         }
 425     }
 426
 427     if (buf && (len<n)) *buf = 0;
 428
 429     return len;
 430 }
 431
 432 // ============================================================================
 433 // wxCharacterSet and derived classes
 434 // ============================================================================
 435
 436 // ----------------------------------------------------------------------------
 437 // wxCharacterSet is the ABC for the classes below
 438 // ----------------------------------------------------------------------------
 439
 440 class wxCharacterSet
 441 {
 442 public:
 443     wxCharacterSet(const wxChar*name) : cname(name) {}
 444     virtual ~wxCharacterSet() {}
 445     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0;
 446     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0;
 447     virtual bool usable() const = 0;
 448 public:
 449     const wxChar*cname;
 450 };
 451
 452 // ----------------------------------------------------------------------------
 453 // ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
 454 // ----------------------------------------------------------------------------
 455
 456 class ID_CharSet : public wxCharacterSet
 457 {
 458 public:
 459     ID_CharSet(const wxChar *name, wxMBConv *cnv)
 460         : wxCharacterSet(name), work(cnv) {}
 461
 462     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 463         { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
 464
 465     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 466         { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
 467
 468     bool usable() const
 469         { return work!=NULL; }
 470 public:
 471     wxMBConv*work;
 472 };
 473
 474
 475 // ============================================================================
 476 // The classes doing conversion using the iconv_xxx() functions
 477 // ============================================================================
 478
 479 #ifdef HAVE_ICONV
 480
 481 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
 482 //     if output buffer is _exactly_ as big as needed. Such case is (unless there's
 483 //     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
 484 //     (which means error) and says there are 0 bytes left in the input buffer --
 485 //     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
 486 //     this alternative test for iconv() failure.
 487 //     [This bug does not appear in glibc 2.2.]
 488 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
 489 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
 490                                      (errno != E2BIG || bufLeft != 0))
 491 #else
 492 #define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
 493 #endif
 494
 495 #define ICONV_CHAR_CAST(x)  ((ICONV_CONST char **)(x))
 496
 497 // ----------------------------------------------------------------------------
 498 // IC_CharSet: encapsulates an iconv character set
 499 // ----------------------------------------------------------------------------
 500
 501 class IC_CharSet : public wxCharacterSet
 502 {
 503 public:
 504     IC_CharSet(const wxChar *name);
 505     virtual ~IC_CharSet();
 506
 507     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
 508     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
 509
 510     bool usable() const
 511         { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
 512
 513 protected:
 514     // the iconv handlers used to translate from multibyte to wide char and in
 515     // the other direction
 516     iconv_t m2w,
 517             w2m;
 518
 519 private:
 520     // the name (for iconv_open()) of a wide char charset - if none is
 521     // available on this machine, it will remain NULL
 522     static const char *ms_wcCharsetName;
 523
 524     // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
 525     // different endian-ness than the native one
 526     static bool ms_wcNeedsSwap;
 527 };
 528
 529 const char *IC_CharSet::ms_wcCharsetName = NULL;
 530 bool IC_CharSet::ms_wcNeedsSwap = FALSE;
 531
 532 IC_CharSet::IC_CharSet(const wxChar *name)
 533           : wxCharacterSet(name)
 534 {
 535     // check for charset that represents wchar_t:
 536     if (ms_wcCharsetName == NULL)
 537     {
 538         ms_wcNeedsSwap = FALSE;
 539
 540         // try charset with explicit bytesex info (e.g. "UCS-4LE"):
 541         ms_wcCharsetName = WC_NAME_BEST;
 542         m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
 543
 544         if (m2w == (iconv_t)-1)
 545         {
 546             // try charset w/o bytesex info (e.g. "UCS4")
 547             // and check for bytesex ourselves:
 548             ms_wcCharsetName = WC_NAME;
 549             m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
 550
 551             // last bet, try if it knows WCHAR_T pseudo-charset
 552             if (m2w == (iconv_t)-1)
 553             {
 554                 ms_wcCharsetName = "WCHAR_T";
 555                 m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
 556             }
 557
 558             if (m2w != (iconv_t)-1)
 559             {
 560                 char    buf[2], *bufPtr;
 561                 wchar_t wbuf[2], *wbufPtr;
 562                 size_t  insz, outsz;
 563                 size_t  res;
 564
 565                 buf[0] = 'A';
 566                 buf[1] = 0;
 567                 wbuf[0] = 0;
 568                 insz = 2;
 569                 outsz = SIZEOF_WCHAR_T * 2;
 570                 wbufPtr = wbuf;
 571                 bufPtr = buf;
 572
 573                 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
 574                             (char**)&wbufPtr, &outsz);
 575
 576                 if (ICONV_FAILED(res, insz))
 577                 {
 578                     ms_wcCharsetName = NULL;
 579                     wxLogLastError(wxT("iconv"));
 580                     wxLogError(_("Convertion to charset '%s' doesn't work."), name);
 581                 }
 582                 else
 583                 {
 584                     ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
 585                 }
 586             }
 587             else
 588             {
 589                 ms_wcCharsetName = NULL;
 590
 591                 // VS: we must not output an error here, since wxWindows will safely
 592                 //     fall back to using wxEncodingConverter.
 593                 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
 594                 //wxLogError(
 595             }
 596         }
 597         wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
 598     }
 599     else // we already have ms_wcCharsetName
 600     {
 601         m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
 602     }
 603
 604     // NB: don't ever pass NULL to iconv_open(), it may crash!
 605     if ( ms_wcCharsetName )
 606     {
 607         w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName);
 608     }
 609     else
 610     {
 611         w2m = (iconv_t)-1;
 612     }
 613 }
 614
 615 IC_CharSet::~IC_CharSet()
 616 {
 617     if ( m2w != (iconv_t)-1 )
 618         iconv_close(m2w);
 619     if ( w2m != (iconv_t)-1 )
 620         iconv_close(w2m);
 621 }
 622
 623 size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
 624 {
 625     size_t inbuf = strlen(psz);
 626     size_t outbuf = n * SIZEOF_WCHAR_T;
 627     size_t res, cres;
 628     // VS: Use these instead of psz, buf because iconv() modifies its arguments:
 629     wchar_t *bufPtr = buf;
 630     const char *pszPtr = psz;
 631
 632     if (buf)
 633     {
 634         // have destination buffer, convert there
 635         cres = iconv(m2w,
 636                      ICONV_CHAR_CAST(&pszPtr), &inbuf,
 637                      (char**)&bufPtr, &outbuf);
 638         res = n - (outbuf / SIZEOF_WCHAR_T);
 639
 640         if (ms_wcNeedsSwap)
 641         {
 642             // convert to native endianness
 643             WC_BSWAP(buf /* _not_ bufPtr */, res)
 644         }
 645     }
 646     else
 647     {
 648         // no destination buffer... convert using temp buffer
 649         // to calculate destination buffer requirement
 650         wchar_t tbuf[8];
 651         res = 0;
 652         do {
 653             bufPtr = tbuf;
 654             outbuf = 8*SIZEOF_WCHAR_T;
 655
 656             cres = iconv(m2w,
 657                          ICONV_CHAR_CAST(&pszPtr), &inbuf,
 658                          (char**)&bufPtr, &outbuf );
 659
 660             res += 8-(outbuf/SIZEOF_WCHAR_T);
 661         } while ((cres==(size_t)-1) && (errno==E2BIG));
 662     }
 663
 664     if (ICONV_FAILED(cres, inbuf))
 665     {
 666         //VS: it is ok if iconv fails, hence trace only
 667         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 668         return (size_t)-1;
 669     }
 670
 671     return res;
 672 }
 673
 674 size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
 675 {
 676 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 677     size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
 678 #else
 679     size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T;
 680 #endif
 681     size_t outbuf = n;
 682     size_t res, cres;
 683
 684     wchar_t *tmpbuf = 0;
 685
 686     if (ms_wcNeedsSwap)
 687     {
 688         // need to copy to temp buffer to switch endianness
 689         // this absolutely doesn't rock!
 690         // (no, doing WC_BSWAP twice on the original buffer won't help, as it
 691         //  could be in read-only memory, or be accessed in some other thread)
 692         tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
 693         memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
 694         WC_BSWAP(tmpbuf, inbuf)
 695         psz=tmpbuf;
 696     }
 697
 698     if (buf)
 699     {
 700         // have destination buffer, convert there
 701         cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 702
 703         res = n-outbuf;
 704     }
 705     else
 706     {
 707         // no destination buffer... convert using temp buffer
 708         // to calculate destination buffer requirement
 709         char tbuf[16];
 710         res = 0;
 711         do {
 712             buf = tbuf; outbuf = 16;
 713
 714             cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 715
 716             res += 16 - outbuf;
 717         } while ((cres==(size_t)-1) && (errno==E2BIG));
 718     }
 719
 720     if (ms_wcNeedsSwap)
 721     {
 722         free(tmpbuf);
 723     }
 724
 725     if (ICONV_FAILED(cres, inbuf))
 726     {
 727         //VS: it is ok if iconv fails, hence trace only
 728         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 729         return (size_t)-1;
 730     }
 731
 732     return res;
 733 }
 734
 735 #endif // HAVE_ICONV
 736
 737 // ============================================================================
 738 // Win32 conversion classes
 739 // ============================================================================
 740
 741 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 742
 743 extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp
 744
 745 class CP_CharSet : public wxCharacterSet
 746 {
 747 public:
 748     CP_CharSet(const wxChar* name)
 749         : wxCharacterSet(name)
 750         {
 751             m_CodePage = wxCharsetToCodepage(name);
 752         }
 753
 754     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 755     {
 756         size_t len =
 757             MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0);
 758         //VS: returns # of written chars for buf!=NULL and *size*
 759         //    needed buffer for buf==NULL
 760         return len ? (buf ? len : len-1) : (size_t)-1;
 761     }
 762
 763     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 764     {
 765         size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf,
 766                                          buf ? n : 0, NULL, NULL);
 767         //VS: returns # of written chars for buf!=NULL and *size*
 768         //    needed buffer for buf==NULL
 769         return len ? (buf ? len : len-1) : (size_t)-1;
 770     }
 771
 772     bool usable() const
 773         { return m_CodePage != -1; }
 774
 775 public:
 776     long m_CodePage;
 777 };
 778 #endif // __WIN32__
 779
 780 // ============================================================================
 781 // wxEncodingConverter based conversion classes
 782 // ============================================================================
 783
 784 #if wxUSE_FONTMAP
 785
 786 class EC_CharSet : public wxCharacterSet
 787 {
 788 public:
 789     // temporarily just use wxEncodingConverter stuff,
 790     // so that it works while a better implementation is built
 791     EC_CharSet(const wxChar* name) : wxCharacterSet(name),
 792                                      enc(wxFONTENCODING_SYSTEM)
 793     {
 794         if (name)
 795             enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE);
 796
 797         m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) &&
 798                w2m.Init(wxFONTENCODING_UNICODE, enc);
 799     }
 800
 801     size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
 802     {
 803         size_t inbuf = strlen(psz);
 804         if (buf)
 805             m2w.Convert(psz,buf);
 806         return inbuf;
 807     }
 808
 809     size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
 810     {
 811 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
 812     || ( defined(__MWERKS__) && defined(__WXMSW__) )
 813         size_t inbuf = std::wcslen(psz);
 814 #else
 815         size_t inbuf = ::wcslen(psz);
 816 #endif
 817         if (buf)
 818             w2m.Convert(psz,buf);
 819
 820         return inbuf;
 821     }
 822
 823     bool usable() const { return m_ok; }
 824
 825 public:
 826     wxFontEncoding enc;
 827     wxEncodingConverter m2w, w2m;
 828
 829     // were we initialized successfully?
 830     bool m_ok;
 831 };
 832
 833 #endif // wxUSE_FONTMAP
 834
 835 // ----------------------------------------------------------------------------
 836 // the function creating the wxCharacterSet for the specified charset on the
 837 // current system, trying all possibilities
 838 // ----------------------------------------------------------------------------
 839
 840 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
 841 {
 842     // check for the special case of ASCII charset
 843 #if wxUSE_FONTMAP
 844     if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT )
 845 #else // wxUSE_FONTMAP
 846     if ( !name )
 847 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
 848     {
 849         // don't convert at all
 850         return NULL;
 851     }
 852
 853     // the test above must have taken care of this case
 854     wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") );
 855
 856     wxCharacterSet *cset;
 857
 858     if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
 859     {
 860         cset = new ID_CharSet(name, &wxConvUTF8);
 861     }
 862     else
 863     {
 864 #ifdef HAVE_ICONV
 865         cset = new IC_CharSet(name);
 866 #else // !HAVE_ICONV
 867         cset = NULL;
 868 #endif // HAVE_ICONV/!HAVE_ICONV
 869     }
 870
 871     // it can only be NULL in this case
 872 #ifndef HAVE_ICONV
 873     if ( cset )
 874 #endif // !HAVE_ICONV
 875     {
 876         if ( cset->usable() )
 877             return cset;
 878
 879         delete cset;
 880         cset = NULL;
 881     }
 882
 883 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 884     cset = new CP_CharSet(name);
 885     if ( cset->usable() )
 886         return cset;
 887
 888     delete cset;
 889     cset = NULL;
 890 #endif // __WIN32__
 891
 892 #if wxUSE_FONTMAP
 893     cset = new EC_CharSet(name);
 894     if ( cset->usable() )
 895         return cset;
 896
 897     delete cset;
 898     cset = NULL;
 899 #endif // wxUSE_FONTMAP
 900
 901     wxLogError(_("Cannot convert from encoding '%s'!"), name);
 902
 903     return NULL;
 904 }
 905
 906 // ============================================================================
 907 // wxCSConv implementation
 908 // ============================================================================
 909
 910 wxCSConv::wxCSConv(const wxChar *charset)
 911 {
 912     m_name = (wxChar *)NULL;
 913     m_cset = (wxCharacterSet *) NULL;
 914     m_deferred = TRUE;
 915
 916     SetName(charset);
 917 }
 918
 919 wxCSConv::~wxCSConv()
 920 {
 921     Clear();
 922 }
 923
 924 wxCSConv::wxCSConv(const wxCSConv& conv)
 925     : wxMBConv()
 926 {
 927     Clear();
 928     SetName(conv.m_name);
 929 }
 930
 931 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 932 {
 933     Clear();
 934     SetName(conv.m_name);
 935     return *this;
 936 }
 937
 938 void wxCSConv::Clear()
 939 {
 940     if (m_name)
 941         free(m_name);
 942     if (m_cset)
 943         delete m_cset;
 944     m_name = NULL;
 945     m_cset = NULL;
 946 }
 947
 948 void wxCSConv::SetName(const wxChar *charset)
 949 {
 950     if (charset)
 951     {
 952         m_name = wxStrdup(charset);
 953         m_deferred = TRUE;
 954     }
 955 }
 956
 957 void wxCSConv::LoadNow()
 958 {
 959     if (m_deferred)
 960     {
 961         if ( !m_name )
 962         {
 963             wxString name = wxLocale::GetSystemEncodingName();
 964             if ( !name.empty() )
 965                 SetName(name);
 966         }
 967
 968         // wxGetCharacterSet() complains about NULL name
 969         m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
 970         m_deferred = FALSE;
 971     }
 972 }
 973
 974 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 975 {
 976     ((wxCSConv *)this)->LoadNow(); // discard constness
 977
 978     if (m_cset)
 979         return m_cset->MB2WC(buf, psz, n);
 980
 981     // latin-1 (direct)
 982     size_t len = strlen(psz);
 983
 984     if (buf)
 985     {
 986         for (size_t c = 0; c <= len; c++)
 987             buf[c] = (unsigned char)(psz[c]);
 988     }
 989
 990     return len;
 991 }
 992
 993 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 994 {
 995     ((wxCSConv *)this)->LoadNow(); // discard constness
 996
 997     if (m_cset)
 998         return m_cset->WC2MB(buf, psz, n);
 999
1000     // latin-1 (direct)
1001 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
1002     || ( defined(__MWERKS__) && defined(__WXMSW__) )
1003     size_t len=std::wcslen(psz);
1004 #else
1005     size_t len=::wcslen(psz);
1006 #endif
1007     if (buf)
1008     {
1009         for (size_t c = 0; c <= len; c++)
1010             buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
1011     }
1012
1013     return len;
1014 }
1015
1016 #endif // wxUSE_WCHAR_T
1017
1018