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 #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
  41 #include "wx/debug.h"
  42 #include "wx/strconv.h"
  43 #include "wx/intl.h"
  44 #include "wx/log.h"
  45
  46 // ----------------------------------------------------------------------------
  47 // globals
  48 // ----------------------------------------------------------------------------
  49
  50 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
  51
  52
  53 // ============================================================================
  54 // implementation
  55 // ============================================================================
  56
  57 #if wxUSE_WCHAR_T
  58
  59 #ifdef __SALFORDC__
  60   #include <clib.h>
  61 #endif
  62
  63 #ifdef HAVE_ICONV_H
  64   #include <iconv.h>
  65 #endif
  66
  67 #ifdef __WXMSW__
  68   #include <windows.h>
  69 #endif
  70
  71 #define BSWAP_UCS4(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT32_SWAP_ALWAYS(str[_c]); }
  72 #define BSWAP_UTF16(str, len) { unsigned _c; for (_c=0; _c<len; _c++) str[_c]=wxUINT16_SWAP_ALWAYS(str[_c]); }
  73
  74 // under Unix SIZEOF_WCHAR_T is defined by configure, but under other platforms
  75 // it might be not defined - assume the most common value
  76 #ifndef SIZEOF_WCHAR_T
  77     #define SIZEOF_WCHAR_T 2
  78 #endif // !defined(SIZEOF_WCHAR_T)
  79
  80 #if SIZEOF_WCHAR_T == 4
  81     #define WC_NAME         "UCS4"
  82     #define WC_BSWAP         BSWAP_UCS4
  83     #ifdef WORDS_BIGENDIAN
  84       #define WC_NAME_BEST  "UCS-4BE"
  85     #else
  86       #define WC_NAME_BEST  "UCS-4LE"
  87     #endif
  88 #elif SIZEOF_WCHAR_T == 2
  89     #define WC_NAME         "UTF16"
  90     #define WC_BSWAP         BSWAP_UTF16
  91     #define WC_UTF16
  92     #ifdef WORDS_BIGENDIAN
  93       #define WC_NAME_BEST  "UTF-16BE"
  94     #else
  95       #define WC_NAME_BEST  "UTF-16LE"
  96     #endif
  97 #else // sizeof(wchar_t) != 2 nor 4
  98     // I don't know what to do about this
  99     #error "Weird sizeof(wchar_t): please report your platform details to wx-users mailing list"
 100 #endif
 101
 102
 103 #ifdef WC_UTF16
 104
 105 static size_t encode_utf16(wxUint32 input, wchar_t *output)
 106 {
 107     if (input<=0xffff)
 108     {
 109         if (output) *output++ = input;
 110         return 1;
 111     }
 112     else if (input>=0x110000)
 113     {
 114         return (size_t)-1;
 115     }
 116     else
 117     {
 118         if (output)
 119         {
 120             *output++ = (input >> 10)+0xd7c0;
 121             *output++ = (input&0x3ff)+0xdc00;
 122         }
 123         return 2;
 124     }
 125 }
 126
 127 static size_t decode_utf16(const wchar_t* input, wxUint32& output)
 128 {
 129     if ((*input<0xd800) || (*input>0xdfff))
 130     {
 131         output = *input;
 132         return 1;
 133     }
 134     else if ((input[1]<0xdc00) || (input[1]>=0xdfff))
 135     {
 136         output = *input;
 137         return (size_t)-1;
 138     }
 139     else
 140     {
 141         output = ((input[0] - 0xd7c0) << 10) + (input[1] - 0xdc00);
 142         return 2;
 143     }
 144 }
 145
 146 #endif // WC_UTF16
 147
 148 // ----------------------------------------------------------------------------
 149 // wxMBConv
 150 // ----------------------------------------------------------------------------
 151
 152 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
 153
 154 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 155 {
 156     return wxMB2WC(buf, psz, n);
 157 }
 158
 159 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 160 {
 161     return wxWC2MB(buf, psz, n);
 162 }
 163
 164 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 165 {
 166     if (psz)
 167     {
 168         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
 169         if (nLen == (size_t)-1)
 170             return wxWCharBuffer((wchar_t *) NULL);
 171         wxWCharBuffer buf(nLen);
 172         MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen);
 173         return buf;
 174     }
 175     else
 176         return wxWCharBuffer((wchar_t *) NULL);
 177 }
 178
 179 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
 180 {
 181     if (psz)
 182     {
 183         size_t nLen = WC2MB((char *) NULL, psz, 0);
 184         if (nLen == (size_t)-1)
 185             return wxCharBuffer((char *) NULL);
 186         wxCharBuffer buf(nLen);
 187         WC2MB((char *)(const char *) buf, psz, nLen);
 188         return buf;
 189     }
 190     else
 191         return wxCharBuffer((char *) NULL);
 192 }
 193
 194 // ----------------------------------------------------------------------------
 195 // standard file conversion
 196 // ----------------------------------------------------------------------------
 197
 198 WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
 199
 200 // just use the libc conversion for now
 201 size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 202 {
 203     return wxMB2WC(buf, psz, n);
 204 }
 205
 206 size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 207 {
 208     return wxWC2MB(buf, psz, n);
 209 }
 210
 211 // ----------------------------------------------------------------------------
 212 // standard gdk conversion
 213 // ----------------------------------------------------------------------------
 214
 215 #ifdef __WXGTK12__
 216
 217 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
 218
 219 #include <gdk/gdk.h>
 220
 221 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 222 {
 223     if (buf)
 224     {
 225         return gdk_mbstowcs((GdkWChar *)buf, psz, n);
 226     }
 227     else
 228     {
 229         GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
 230         size_t len = gdk_mbstowcs(nbuf, psz, n);
 231         delete[] nbuf;
 232         return len;
 233     }
 234 }
 235
 236 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 237 {
 238     char *mbstr = gdk_wcstombs((GdkWChar *)psz);
 239     size_t len = mbstr ? strlen(mbstr) : 0;
 240     if (buf)
 241     {
 242         if (len > n)
 243             len = n;
 244         memcpy(buf, psz, len);
 245         if (len < n)
 246             buf[len] = 0;
 247     }
 248     return len;
 249 }
 250
 251 #endif // GTK > 1.0
 252
 253 // ----------------------------------------------------------------------------
 254 // UTF-7
 255 // ----------------------------------------------------------------------------
 256
 257 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
 258
 259 #if 0
 260 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 261                         "abcdefghijklmnopqrstuvwxyz"
 262                         "0123456789'(),-./:?";
 263 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
 264 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 265                         "abcdefghijklmnopqrstuvwxyz"
 266                         "0123456789+/";
 267 #endif
 268
 269 // TODO: write actual implementations of UTF-7 here
 270 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
 271                            const char * WXUNUSED(psz),
 272                            size_t WXUNUSED(n)) const
 273 {
 274   return 0;
 275 }
 276
 277 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
 278                            const wchar_t * WXUNUSED(psz),
 279                            size_t WXUNUSED(n)) const
 280 {
 281   return 0;
 282 }
 283
 284 // ----------------------------------------------------------------------------
 285 // UTF-8
 286 // ----------------------------------------------------------------------------
 287
 288 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
 289
 290 static wxUint32 utf8_max[]=
 291     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 292
 293 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 294 {
 295     size_t len = 0;
 296
 297     while (*psz && ((!buf) || (len < n)))
 298     {
 299         unsigned char cc = *psz++, fc = cc;
 300         unsigned cnt;
 301         for (cnt = 0; fc & 0x80; cnt++)
 302             fc <<= 1;
 303         if (!cnt)
 304         {
 305             // plain ASCII char
 306             if (buf)
 307                 *buf++ = cc;
 308             len++;
 309         }
 310         else
 311         {
 312             cnt--;
 313             if (!cnt)
 314             {
 315                 // invalid UTF-8 sequence
 316                 return (size_t)-1;
 317             }
 318             else
 319             {
 320                 unsigned ocnt = cnt - 1;
 321                 wxUint32 res = cc & (0x3f >> cnt);
 322                 while (cnt--)
 323                 {
 324                     cc = *psz++;
 325                     if ((cc & 0xC0) != 0x80)
 326                     {
 327                         // invalid UTF-8 sequence
 328                         return (size_t)-1;
 329                     }
 330                     res = (res << 6) | (cc & 0x3f);
 331                 }
 332                 if (res <= utf8_max[ocnt])
 333                 {
 334                     // illegal UTF-8 encoding
 335                     return (size_t)-1;
 336                 }
 337 #ifdef WC_UTF16
 338                 size_t pa = encode_utf16(res, buf);
 339                 if (pa == (size_t)-1)
 340                   return (size_t)-1;
 341                 if (buf)
 342                     buf += pa;
 343                 len += pa;
 344 #else
 345                 if (buf)
 346                     *buf++ = res;
 347                 len++;
 348 #endif
 349             }
 350         }
 351     }
 352     if (buf && (len < n))
 353         *buf = 0;
 354     return len;
 355 }
 356
 357 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 358 {
 359     size_t len = 0;
 360
 361     while (*psz && ((!buf) || (len < n)))
 362     {
 363         wxUint32 cc;
 364 #ifdef WC_UTF16
 365         size_t pa = decode_utf16(psz, cc);
 366         psz += (pa == (size_t)-1) ? 1 : pa;
 367 #else
 368         cc=(*psz++) & 0x7fffffff;
 369 #endif
 370         unsigned cnt;
 371         for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
 372         if (!cnt)
 373         {
 374             // plain ASCII char
 375             if (buf)
 376                 *buf++ = cc;
 377             len++;
 378         }
 379
 380         else
 381         {
 382             len += cnt + 1;
 383             if (buf)
 384             {
 385                 *buf++ = (-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt));
 386                 while (cnt--)
 387                     *buf++ = 0x80 | ((cc >> (cnt * 6)) & 0x3f);
 388             }
 389         }
 390     }
 391
 392     if (buf && (len<n)) *buf = 0;
 393     return len;
 394 }
 395
 396 // ----------------------------------------------------------------------------
 397 // specified character set
 398 // ----------------------------------------------------------------------------
 399
 400 WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
 401
 402 #include "wx/encconv.h"
 403 #include "wx/fontmap.h"
 404
 405 // TODO: add some tables here
 406 // - perhaps common encodings to common codepages (for Win32)
 407 // - perhaps common encodings to objects ("UTF8" -> wxConvUTF8)
 408 // - move wxEncodingConverter meat in here
 409
 410 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 411
 412 // VZ: the new version of wxCharsetToCodepage() is more politically correct
 413 //     and should work on other Windows versions as well but the old version is
 414 //     still needed for !wxUSE_FONTMAP case
 415
 416 extern long wxEncodingToCodepage(wxFontEncoding encoding)
 417 {
 418     // translate encoding into the Windows CHARSET
 419     wxNativeEncodingInfo natveEncInfo;
 420     if ( !wxGetNativeFontEncoding(encoding, &natveEncInfo) )
 421         return -1;
 422
 423     // translate CHARSET to code page
 424     CHARSETINFO csetInfo;
 425     if ( !::TranslateCharsetInfo((DWORD *)(DWORD)natveEncInfo.charset,
 426                                  &csetInfo,
 427                                  TCI_SRCCHARSET) )
 428     {
 429         wxLogLastError(_T("TranslateCharsetInfo(TCI_SRCCHARSET)"));
 430
 431         return -1;
 432     }
 433
 434     return csetInfo.ciACP;
 435 }
 436
 437 #if wxUSE_FONTMAP && wxUSE_GUI
 438
 439 extern long wxCharsetToCodepage(const wxChar *name)
 440 {
 441     // first get the font encoding for this charset
 442     if ( !name )
 443         return -1;
 444
 445     wxFontEncoding enc = wxTheFontMapper->CharsetToEncoding(name, FALSE);
 446     if ( enc == wxFONTENCODING_SYSTEM )
 447         return -1;
 448
 449     // the use the helper function
 450     return wxEncodingToCodepage(enc);
 451 }
 452
 453 #else // old wxCharsetToCodepage() by OK
 454
 455 #include "wx/msw/registry.h"
 456
 457 // this should work if Internet Exploiter is installed
 458 extern long wxCharsetToCodepage(const wxChar *name)
 459 {
 460     if (!name)
 461         return GetACP();
 462
 463     long CP=-1;
 464
 465     wxString cn(name);
 466     do {
 467         wxString path(wxT("MIME\\Database\\Charset\\"));
 468         path += cn;
 469         wxRegKey key(wxRegKey::HKCR, path);
 470
 471         if (!key.Exists()) break;
 472
 473         // two cases: either there's an AliasForCharset string,
 474         // or there are Codepage and InternetEncoding dwords.
 475         // The InternetEncoding gives us the actual encoding,
 476         // the Codepage just says which Windows character set to
 477         // use when displaying the data.
 478         if (key.HasValue(wxT("InternetEncoding")) &&
 479             key.QueryValue(wxT("InternetEncoding"), &CP)) break;
 480
 481         // no encoding, see if it's an alias
 482         if (!key.HasValue(wxT("AliasForCharset")) ||
 483             !key.QueryValue(wxT("AliasForCharset"), cn)) break;
 484     } while (1);
 485
 486     return CP;
 487 }
 488
 489 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
 490
 491 #endif // Win32
 492
 493 class wxCharacterSet
 494 {
 495 public:
 496     wxCharacterSet(const wxChar*name)
 497         : cname(name) {}
 498     virtual ~wxCharacterSet()
 499         {}
 500     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 501         { return (size_t)-1; }
 502     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 503         { return (size_t)-1; }
 504     virtual bool usable()
 505         { return FALSE; }
 506 public:
 507     const wxChar*cname;
 508 };
 509
 510 class ID_CharSet : public wxCharacterSet
 511 {
 512 public:
 513     ID_CharSet(const wxChar *name,wxMBConv *cnv)
 514         : wxCharacterSet(name), work(cnv) {}
 515
 516     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 517         { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
 518
 519     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 520         { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
 521
 522     bool usable()
 523         { return work!=NULL; }
 524 public:
 525     wxMBConv*work;
 526 };
 527
 528
 529 #ifdef HAVE_ICONV_H
 530
 531 bool g_wcNeedsSwap = FALSE;
 532 static const char *g_wcCharset = NULL;
 533
 534 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
 535 //     if output buffer is _exactly_ as big as needed. Such case is (unless there's
 536 //     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
 537 //     (which means error) and says there are 0 bytes left in the input buffer --
 538 //     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
 539 //     this alternative test for iconv() failure.
 540 //     [This bug does not appear in glibc 2.2.]
 541 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
 542 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
 543                                      (errno != E2BIG || bufLeft != 0))
 544 #else
 545 #define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
 546 #endif
 547
 548 class IC_CharSet : public wxCharacterSet
 549 {
 550 public:
 551     IC_CharSet(const wxChar *name)
 552         : wxCharacterSet(name)
 553     {
 554         // check for charset that represents wchar_t:
 555         if (g_wcCharset == NULL)
 556         {
 557             g_wcNeedsSwap = FALSE;
 558
 559             // try charset with explicit bytesex info (e.g. "UCS-4LE"):
 560             g_wcCharset = WC_NAME_BEST;
 561             m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name));
 562
 563             if (m2w == (iconv_t)-1)
 564             {
 565                 // try charset w/o bytesex info (e.g. "UCS4")
 566                 // and check for bytesex ourselves:
 567                 g_wcCharset = WC_NAME;
 568                 m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name));
 569
 570                 // last bet, try if it knows WCHAR_T pseudo-charset
 571                 if (m2w == (iconv_t)-1)
 572                 {
 573                     g_wcCharset = "WCHAR_T";
 574                     m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name));
 575                 }
 576
 577                 if (m2w != (iconv_t)-1)
 578                 {
 579                     char    buf[2], *bufPtr;
 580                     wchar_t wbuf[2], *wbufPtr;
 581                     size_t  insz, outsz;
 582                     size_t  res;
 583
 584                     buf[0] = 'A';
 585                     buf[1] = 0;
 586                     wbuf[0] = 0;
 587                     insz = 2;
 588                     outsz = SIZEOF_WCHAR_T * 2;
 589                     wbufPtr = wbuf;
 590                     bufPtr = buf;
 591
 592                     #ifdef WX_ICONV_TAKES_CHAR
 593                     res = iconv(m2w, (char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz);
 594                     #else
 595                     res = iconv(m2w, (const char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz);
 596                     #endif
 597                     if (ICONV_FAILED(res, insz))
 598                     {
 599                         g_wcCharset = NULL;
 600                         wxLogLastError(wxT("iconv"));
 601                         wxLogError(_("Convertion to charset '%s' doesn't work."), name);
 602                     }
 603                     else
 604                     {
 605                         g_wcNeedsSwap = (wbuf[0] != (wchar_t)buf[0]);
 606                     }
 607                 }
 608                 else
 609                 {
 610                     g_wcCharset = NULL;
 611                     wxLogError(_("Don't know how to convert to/from charset '%s'."), name);
 612                 }
 613             }
 614             wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), g_wcCharset, g_wcNeedsSwap);
 615         }
 616         else
 617             m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name));
 618
 619         w2m = iconv_open(wxConvLibc.cWX2MB(name), g_wcCharset);
 620     }
 621
 622     ~IC_CharSet()
 623     {
 624         if ( m2w != (iconv_t)-1 )
 625             iconv_close(m2w);
 626         if ( w2m != (iconv_t)-1 )
 627             iconv_close(w2m);
 628     }
 629
 630     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 631     {
 632         size_t inbuf = strlen(psz);
 633         size_t outbuf = n * SIZEOF_WCHAR_T;
 634         size_t res, cres;
 635         // VS: Use these instead of psz, buf because iconv() modifies its arguments:
 636         wchar_t *bufPtr = buf;
 637         const char *pszPtr = psz;
 638
 639         if (buf)
 640         {
 641             // have destination buffer, convert there
 642 #ifdef WX_ICONV_TAKES_CHAR
 643             cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
 644 #else
 645             cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
 646 #endif
 647             res = n - (outbuf / SIZEOF_WCHAR_T);
 648
 649             if (g_wcNeedsSwap)
 650             {
 651                 // convert to native endianness
 652                 WC_BSWAP(buf /* _not_ bufPtr */, res)
 653             }
 654         }
 655         else
 656         {
 657             // no destination buffer... convert using temp buffer
 658             // to calculate destination buffer requirement
 659             wchar_t tbuf[8];
 660             res = 0;
 661             do {
 662                 bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T;
 663 #ifdef WX_ICONV_TAKES_CHAR
 664                 cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
 665 #else
 666                 cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
 667 #endif
 668                 res += 8-(outbuf/SIZEOF_WCHAR_T);
 669             } while ((cres==(size_t)-1) && (errno==E2BIG));
 670         }
 671
 672         if (ICONV_FAILED(cres, inbuf))
 673         {
 674             //VS: it is ok if iconv fails, hence trace only
 675             wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 676             return (size_t)-1;
 677         }
 678
 679         return res;
 680     }
 681
 682     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 683     {
 684 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 685         size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
 686 #else
 687         size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T;
 688 #endif
 689         size_t outbuf = n;
 690         size_t res, cres;
 691
 692         wchar_t *tmpbuf = 0;
 693
 694         if (g_wcNeedsSwap)
 695         {
 696             // need to copy to temp buffer to switch endianness
 697             // this absolutely doesn't rock!
 698             // (no, doing WC_BSWAP twice on the original buffer won't help, as it
 699             //  could be in read-only memory, or be accessed in some other thread)
 700             tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
 701             memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
 702             WC_BSWAP(tmpbuf, inbuf)
 703             psz=tmpbuf;
 704         }
 705
 706         if (buf)
 707         {
 708             // have destination buffer, convert there
 709 #ifdef WX_ICONV_TAKES_CHAR
 710             cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
 711 #else
 712             cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
 713 #endif
 714             res = n-outbuf;
 715         }
 716         else
 717         {
 718             // no destination buffer... convert using temp buffer
 719             // to calculate destination buffer requirement
 720             char tbuf[16];
 721             res = 0;
 722             do {
 723                 buf = tbuf; outbuf = 16;
 724 #ifdef WX_ICONV_TAKES_CHAR
 725                 cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
 726 #else
 727                 cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
 728 #endif
 729                 res += 16 - outbuf;
 730             } while ((cres==(size_t)-1) && (errno==E2BIG));
 731         }
 732
 733         if (g_wcNeedsSwap)
 734         {
 735             free(tmpbuf);
 736         }
 737
 738         if (ICONV_FAILED(cres, inbuf))
 739         {
 740             //VS: it is ok if iconv fails, hence trace only
 741             wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 742             return (size_t)-1;
 743         }
 744
 745         return res;
 746     }
 747
 748     bool usable()
 749         { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
 750
 751 protected:
 752     iconv_t m2w, w2m;
 753 };
 754 #endif
 755
 756 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 757 class CP_CharSet : public wxCharacterSet
 758 {
 759 public:
 760     CP_CharSet(const wxChar* name)
 761         : wxCharacterSet(name)
 762         {
 763             m_CodePage = wxCharsetToCodepage(name);
 764         }
 765
 766     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 767     {
 768         size_t len =
 769             MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0);
 770         //VS: returns # of written chars for buf!=NULL and *size*
 771         //    needed buffer for buf==NULL
 772         return len ? (buf ? len : len-1) : (size_t)-1;
 773     }
 774
 775     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 776     {
 777         size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf,
 778                                          buf ? n : 0, NULL, NULL);
 779         //VS: returns # of written chars for buf!=NULL and *size*
 780         //    needed buffer for buf==NULL
 781         return len ? (buf ? len : len-1) : (size_t)-1;
 782     }
 783
 784     bool usable()
 785         { return m_CodePage != -1; }
 786
 787 public:
 788     long m_CodePage;
 789 };
 790 #endif // __WIN32__
 791
 792 #if wxUSE_FONTMAP
 793
 794 class EC_CharSet : public wxCharacterSet
 795 {
 796 public:
 797     // temporarily just use wxEncodingConverter stuff,
 798     // so that it works while a better implementation is built
 799     EC_CharSet(const wxChar* name) : wxCharacterSet(name),
 800                                      enc(wxFONTENCODING_SYSTEM)
 801     {
 802         if (name)
 803             enc = wxTheFontMapper->CharsetToEncoding(name, FALSE);
 804         m2w.Init(enc, wxFONTENCODING_UNICODE);
 805         w2m.Init(wxFONTENCODING_UNICODE, enc);
 806     }
 807
 808     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 809     {
 810         size_t inbuf = strlen(psz);
 811         if (buf)
 812             m2w.Convert(psz,buf);
 813         return inbuf;
 814     }
 815
 816     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 817     {
 818 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
 819     || ( defined(__MWERKS__) && defined(__WXMSW__) )
 820         size_t inbuf = std::wcslen(psz);
 821 #else
 822         size_t inbuf = ::wcslen(psz);
 823 #endif
 824         if (buf)
 825             w2m.Convert(psz,buf);
 826
 827         return inbuf;
 828     }
 829
 830     bool usable()
 831         { return (enc!=wxFONTENCODING_SYSTEM) && (enc!=wxFONTENCODING_DEFAULT); }
 832
 833 public:
 834     wxFontEncoding enc;
 835     wxEncodingConverter m2w, w2m;
 836 };
 837
 838 #endif // wxUSE_FONTMAP
 839
 840 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
 841 {
 842     wxCharacterSet *cset = NULL;
 843     if (name)
 844     {
 845         if (wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
 846         {
 847             cset = new ID_CharSet(name, &wxConvUTF8);
 848         }
 849         else
 850         {
 851 #ifdef HAVE_ICONV_H
 852             cset = new IC_CharSet(name); // may not take NULL
 853 #endif
 854         }
 855     }
 856
 857     if (cset && cset->usable())
 858         return cset;
 859
 860     if (cset)
 861     {
 862         delete cset;
 863         cset = NULL;
 864     }
 865
 866 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 867     cset = new CP_CharSet(name); // may take NULL
 868     if (cset->usable())
 869         return cset;
 870
 871     delete cset;
 872 #endif // __WIN32__
 873
 874 #if wxUSE_FONTMAP
 875     cset = new EC_CharSet(name);
 876     if (cset->usable())
 877         return cset;
 878 #endif // wxUSE_FONTMAP
 879
 880     delete cset;
 881     wxLogError(_("Unknown encoding '%s'!"), name);
 882     return NULL;
 883 }
 884
 885 wxCSConv::wxCSConv(const wxChar *charset)
 886 {
 887     m_name = (wxChar *)NULL;
 888     m_cset = (wxCharacterSet *) NULL;
 889     m_deferred = TRUE;
 890
 891     SetName(charset);
 892 }
 893
 894 wxCSConv::~wxCSConv()
 895 {
 896     free(m_name);
 897     delete m_cset;
 898 }
 899
 900 void wxCSConv::SetName(const wxChar *charset)
 901 {
 902     if (charset)
 903     {
 904         m_name = wxStrdup(charset);
 905         m_deferred = TRUE;
 906     }
 907 }
 908
 909 void wxCSConv::LoadNow()
 910 {
 911     if (m_deferred)
 912     {
 913         if ( !m_name )
 914         {
 915             wxString name = wxLocale::GetSystemEncodingName();
 916             if ( !name.empty() )
 917                 SetName(name);
 918         }
 919
 920         // wxGetCharacterSet() complains about NULL name
 921         m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
 922         m_deferred = FALSE;
 923     }
 924 }
 925
 926 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 927 {
 928     ((wxCSConv *)this)->LoadNow(); // discard constness
 929
 930     if (m_cset)
 931         return m_cset->MB2WC(buf, psz, n);
 932
 933     // latin-1 (direct)
 934     size_t len = strlen(psz);
 935
 936     if (buf)
 937     {
 938         for (size_t c = 0; c <= len; c++)
 939             buf[c] = (unsigned char)(psz[c]);
 940     }
 941
 942     return len;
 943 }
 944
 945 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 946 {
 947     ((wxCSConv *)this)->LoadNow(); // discard constness
 948
 949     if (m_cset)
 950         return m_cset->WC2MB(buf, psz, n);
 951
 952     // latin-1 (direct)
 953 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
 954     || ( defined(__MWERKS__) && defined(__WXMSW__) )
 955     size_t len=std::wcslen(psz);
 956 #else
 957     size_t len=::wcslen(psz);
 958 #endif
 959     if (buf)
 960     {
 961         for (size_t c = 0; c <= len; c++)
 962             buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
 963     }
 964
 965     return len;
 966 }
 967
 968 #ifdef HAVE_ICONV_H
 969
 970 class IC_CharSetConverter
 971 {
 972 public:
 973     IC_CharSetConverter(IC_CharSet *from, IC_CharSet *to)
 974     {
 975         cnv = iconv_open(wxConvLibc.cWX2MB(to->cname),
 976                          wxConvLibc.cWX2MB(from->cname));
 977     }
 978
 979     ~IC_CharSetConverter()
 980     {
 981         if (cnv != (iconv_t)-1)
 982             iconv_close(cnv);
 983     }
 984
 985     size_t Convert(char *buf, const char *psz, size_t n)
 986     {
 987         size_t inbuf = strlen(psz);
 988         size_t outbuf = n;
 989 #ifdef WX_ICONV_TAKES_CHAR
 990         size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf );
 991 #else
 992         size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf );
 993 #endif
 994         if (res == (size_t)-1)
 995             return (size_t)-1;
 996         return (n - outbuf);
 997     }
 998
 999 public:
1000     iconv_t cnv;
1001 };
1002
1003 #endif // HAVE_ICONV_H
1004
1005 class EC_CharSetConverter
1006 {
1007 public:
1008     EC_CharSetConverter(EC_CharSet* from,EC_CharSet* to)
1009         { cnv.Init(from->enc,to->enc); }
1010
1011     size_t Convert(char* buf, const char* psz, size_t n)
1012     {
1013         size_t inbuf = strlen(psz);
1014         if (buf) cnv.Convert(psz,buf);
1015         return inbuf;
1016     }
1017
1018 public:
1019     wxEncodingConverter cnv;
1020 };
1021
1022 #else // !wxUSE_WCHAR_T
1023
1024 // ----------------------------------------------------------------------------
1025 // stand-ins in absence of wchar_t
1026 // ----------------------------------------------------------------------------
1027
1028 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
1029
1030 #endif // wxUSE_WCHAR_T
1031
1032