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