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 #define IGNORE_LIBC 0
 190
 191 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 192 {
 193 #if IGNORE_LIBC
 194     if (buf)
 195     {
 196         for (size_t i = 0; i < strlen( psz )+1; i++)
 197             buf[i] = (wchar_t) psz[i];
 198         // printf( "libc %s\n", buf );
 199         return strlen( psz );
 200     }
 201     else
 202     {
 203         return strlen( psz );
 204     }
 205 #else
 206     return wxMB2WC(buf, psz, n);
 207 #endif
 208 }
 209
 210 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 211 {
 212 #if IGNORE_LIBC
 213     if (buf)
 214     {
 215         for (size_t i = 0; i < wxStrlen( psz )+1; i++)
 216             buf[i] = (char) psz[i];
 217         // printf( "libc %s\n", buf );
 218         return wxStrlen( psz );
 219     }
 220     else
 221     {
 222         return wxStrlen( psz );
 223     }
 224 #else
 225     return wxWC2MB(buf, psz, n);
 226 #endif
 227 }
 228
 229 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
 230 {
 231     if (psz)
 232     {
 233         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);  // return value excludes /0
 234         if (nLen == (size_t)-1)
 235             return wxWCharBuffer((wchar_t *) NULL);
 236         wxWCharBuffer buf(nLen);                        // this allocates nLen1+
 237         MB2WC((wchar_t *)(const wchar_t *) buf, psz, nLen+1);
 238         return buf;
 239     }
 240     else
 241         return wxWCharBuffer((wchar_t *) NULL);
 242 }
 243
 244 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
 245 {
 246     if (psz)
 247     {
 248         size_t nLen = WC2MB((char *) NULL, psz, 0);  // return value excludes /0
 249         if (nLen == (size_t)-1)
 250             return wxCharBuffer((char *) NULL);
 251         wxCharBuffer buf(nLen);                      // this allocates nLen+1
 252         WC2MB((char *)(const char *) buf, psz, nLen+1);
 253         // printf( "str %s\n", (const char*) buf );
 254         return buf;
 255     }
 256     else
 257         return wxCharBuffer((char *) NULL);
 258 }
 259
 260 // ----------------------------------------------------------------------------
 261 // standard gdk conversion
 262 // ----------------------------------------------------------------------------
 263
 264 #ifdef __WXGTK12__
 265
 266 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
 267
 268 #include <gdk/gdk.h>
 269
 270 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 271 {
 272     if (buf)
 273     {
 274         return gdk_mbstowcs((GdkWChar *)buf, psz, n);
 275     }
 276     else
 277     {
 278         GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
 279         size_t len = gdk_mbstowcs(nbuf, psz, n);
 280         delete[] nbuf;
 281         return len;
 282     }
 283 }
 284
 285 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 286 {
 287     char *mbstr = gdk_wcstombs((GdkWChar *)psz);
 288     size_t len = mbstr ? strlen(mbstr) : 0;
 289     if (buf)
 290     {
 291         if (len > n)
 292             len = n;
 293         memcpy(buf, psz, len);
 294         if (len < n)
 295             buf[len] = 0;
 296     }
 297     return len;
 298 }
 299
 300 #endif // GTK > 1.0
 301
 302 // ----------------------------------------------------------------------------
 303 // UTF-7
 304 // ----------------------------------------------------------------------------
 305
 306 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
 307
 308 #if 0
 309 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 310                         "abcdefghijklmnopqrstuvwxyz"
 311                         "0123456789'(),-./:?";
 312 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
 313 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 314                         "abcdefghijklmnopqrstuvwxyz"
 315                         "0123456789+/";
 316 #endif
 317
 318 // TODO: write actual implementations of UTF-7 here
 319 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
 320                            const char * WXUNUSED(psz),
 321                            size_t WXUNUSED(n)) const
 322 {
 323   return 0;
 324 }
 325
 326 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
 327                            const wchar_t * WXUNUSED(psz),
 328                            size_t WXUNUSED(n)) const
 329 {
 330   return 0;
 331 }
 332
 333 // ----------------------------------------------------------------------------
 334 // UTF-8
 335 // ----------------------------------------------------------------------------
 336
 337 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
 338
 339 static wxUint32 utf8_max[]=
 340     { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff, 0xffffffff };
 341
 342 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 343 {
 344     size_t len = 0;
 345
 346     while (*psz && ((!buf) || (len < n)))
 347     {
 348         unsigned char cc = *psz++, fc = cc;
 349         unsigned cnt;
 350         for (cnt = 0; fc & 0x80; cnt++)
 351             fc <<= 1;
 352         if (!cnt)
 353         {
 354             // plain ASCII char
 355             if (buf)
 356                 *buf++ = cc;
 357             len++;
 358         }
 359         else
 360         {
 361             cnt--;
 362             if (!cnt)
 363             {
 364                 // invalid UTF-8 sequence
 365                 return (size_t)-1;
 366             }
 367             else
 368             {
 369                 unsigned ocnt = cnt - 1;
 370                 wxUint32 res = cc & (0x3f >> cnt);
 371                 while (cnt--)
 372                 {
 373                     cc = *psz++;
 374                     if ((cc & 0xC0) != 0x80)
 375                     {
 376                         // invalid UTF-8 sequence
 377                         return (size_t)-1;
 378                     }
 379                     res = (res << 6) | (cc & 0x3f);
 380                 }
 381                 if (res <= utf8_max[ocnt])
 382                 {
 383                     // illegal UTF-8 encoding
 384                     return (size_t)-1;
 385                 }
 386 #ifdef WC_UTF16
 387                 size_t pa = encode_utf16(res, buf);
 388                 if (pa == (size_t)-1)
 389                   return (size_t)-1;
 390                 if (buf)
 391                     buf += pa;
 392                 len += pa;
 393 #else // !WC_UTF16
 394                 if (buf)
 395                     *buf++ = res;
 396                 len++;
 397 #endif // WC_UTF16/!WC_UTF16
 398             }
 399         }
 400     }
 401     if (buf && (len < n))
 402         *buf = 0;
 403     return len;
 404 }
 405
 406 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
 407 {
 408     size_t len = 0;
 409
 410     while (*psz && ((!buf) || (len < n)))
 411     {
 412         wxUint32 cc;
 413 #ifdef WC_UTF16
 414         size_t pa = decode_utf16(psz, cc);
 415         psz += (pa == (size_t)-1) ? 1 : pa;
 416 #else
 417         cc=(*psz++) & 0x7fffffff;
 418 #endif
 419         unsigned cnt;
 420         for (cnt = 0; cc > utf8_max[cnt]; cnt++) {}
 421         if (!cnt)
 422         {
 423             // plain ASCII char
 424             if (buf)
 425                 *buf++ = (char) cc;
 426             len++;
 427         }
 428
 429         else
 430         {
 431             len += cnt + 1;
 432             if (buf)
 433             {
 434                 *buf++ = (char) ((-128 >> cnt) | ((cc >> (cnt * 6)) & (0x3f >> cnt)));
 435                 while (cnt--)
 436                     *buf++ = (char) (0x80 | ((cc >> (cnt * 6)) & 0x3f));
 437             }
 438         }
 439     }
 440
 441     if (buf && (len<n)) *buf = 0;
 442
 443     return len;
 444 }
 445
 446 // ============================================================================
 447 // wxCharacterSet and derived classes
 448 // ============================================================================
 449
 450 // ----------------------------------------------------------------------------
 451 // wxCharacterSet is the ABC for the classes below
 452 // ----------------------------------------------------------------------------
 453
 454 class wxCharacterSet
 455 {
 456 public:
 457     wxCharacterSet(const wxChar*name) : cname(name) {}
 458     virtual ~wxCharacterSet() {}
 459     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n) = 0;
 460     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n) = 0;
 461     virtual bool usable() const = 0;
 462 public:
 463     const wxChar*cname;
 464 };
 465
 466 // ----------------------------------------------------------------------------
 467 // ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
 468 // ----------------------------------------------------------------------------
 469
 470 class ID_CharSet : public wxCharacterSet
 471 {
 472 public:
 473     ID_CharSet(const wxChar *name, wxMBConv *cnv)
 474         : wxCharacterSet(name), work(cnv) {}
 475
 476     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 477         { return work ? work->MB2WC(buf,psz,n) : (size_t)-1; }
 478
 479     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 480         { return work ? work->WC2MB(buf,psz,n) : (size_t)-1; }
 481
 482     bool usable() const
 483         { return work!=NULL; }
 484 public:
 485     wxMBConv*work;
 486 };
 487
 488
 489 // ============================================================================
 490 // The classes doing conversion using the iconv_xxx() functions
 491 // ============================================================================
 492
 493 #ifdef HAVE_ICONV
 494
 495 // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
 496 //     if output buffer is _exactly_ as big as needed. Such case is (unless there's
 497 //     yet another bug in glibc) the only case when iconv() returns with (size_t)-1
 498 //     (which means error) and says there are 0 bytes left in the input buffer --
 499 //     when _real_ error occurs, bytes-left-in-input buffer is non-zero. Hence,
 500 //     this alternative test for iconv() failure.
 501 //     [This bug does not appear in glibc 2.2.]
 502 #if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ <= 1
 503 #define ICONV_FAILED(cres, bufLeft) ((cres == (size_t)-1) && \
 504                                      (errno != E2BIG || bufLeft != 0))
 505 #else
 506 #define ICONV_FAILED(cres, bufLeft)  (cres == (size_t)-1)
 507 #endif
 508
 509 #define ICONV_CHAR_CAST(x)  ((ICONV_CONST char **)(x))
 510
 511 // ----------------------------------------------------------------------------
 512 // IC_CharSet: encapsulates an iconv character set
 513 // ----------------------------------------------------------------------------
 514
 515 class IC_CharSet : public wxCharacterSet
 516 {
 517 public:
 518     IC_CharSet(const wxChar *name);
 519     virtual ~IC_CharSet();
 520
 521     virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
 522     virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
 523
 524     bool usable() const
 525         { return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
 526
 527 protected:
 528     // the iconv handlers used to translate from multibyte to wide char and in
 529     // the other direction
 530     iconv_t m2w,
 531             w2m;
 532
 533 private:
 534     // the name (for iconv_open()) of a wide char charset - if none is
 535     // available on this machine, it will remain NULL
 536     static const char *ms_wcCharsetName;
 537
 538     // true if the wide char encoding we use (i.e. ms_wcCharsetName) has
 539     // different endian-ness than the native one
 540     static bool ms_wcNeedsSwap;
 541 };
 542
 543 const char *IC_CharSet::ms_wcCharsetName = NULL;
 544 bool IC_CharSet::ms_wcNeedsSwap = FALSE;
 545
 546 IC_CharSet::IC_CharSet(const wxChar *name)
 547           : wxCharacterSet(name)
 548 {
 549     // Do it the hard way
 550     char cname[100];
 551     for (size_t i = 0; i < wxStrlen(name)+1; i++)
 552         cname[i] = (char) name[i];
 553
 554     // check for charset that represents wchar_t:
 555     if (ms_wcCharsetName == NULL)
 556     {
 557         ms_wcNeedsSwap = FALSE;
 558
 559         // try charset with explicit bytesex info (e.g. "UCS-4LE"):
 560         ms_wcCharsetName = WC_NAME_BEST;
 561         m2w = iconv_open(ms_wcCharsetName, cname);
 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             ms_wcCharsetName = WC_NAME;
 568             m2w = iconv_open(ms_wcCharsetName, cname);
 569
 570             // last bet, try if it knows WCHAR_T pseudo-charset
 571             if (m2w == (iconv_t)-1)
 572             {
 573                 ms_wcCharsetName = "WCHAR_T";
 574                 m2w = iconv_open(ms_wcCharsetName, cname);
 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                 res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
 593                             (char**)&wbufPtr, &outsz);
 594
 595                 if (ICONV_FAILED(res, insz))
 596                 {
 597                     ms_wcCharsetName = NULL;
 598                     wxLogLastError(wxT("iconv"));
 599                     wxLogError(_("Convertion to charset '%s' doesn't work."), name);
 600                 }
 601                 else
 602                 {
 603                     ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
 604                 }
 605             }
 606             else
 607             {
 608                 ms_wcCharsetName = NULL;
 609
 610                 // VS: we must not output an error here, since wxWindows will safely
 611                 //     fall back to using wxEncodingConverter.
 612                 wxLogTrace(wxT("strconv"), wxT("Impossible to convert to/from charset '%s' with iconv, falling back to wxEncodingConverter."), name);
 613                 //wxLogError(
 614             }
 615         }
 616         wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
 617     }
 618     else // we already have ms_wcCharsetName
 619     {
 620         m2w = iconv_open(ms_wcCharsetName, cname);
 621     }
 622
 623     // NB: don't ever pass NULL to iconv_open(), it may crash!
 624     if ( ms_wcCharsetName )
 625     {
 626         w2m = iconv_open( cname, ms_wcCharsetName);
 627     }
 628     else
 629     {
 630         w2m = (iconv_t)-1;
 631     }
 632 }
 633
 634 IC_CharSet::~IC_CharSet()
 635 {
 636     if ( m2w != (iconv_t)-1 )
 637         iconv_close(m2w);
 638     if ( w2m != (iconv_t)-1 )
 639         iconv_close(w2m);
 640 }
 641
 642 size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
 643 {
 644     size_t inbuf = strlen(psz);
 645     size_t outbuf = n * SIZEOF_WCHAR_T;
 646     size_t res, cres;
 647     // VS: Use these instead of psz, buf because iconv() modifies its arguments:
 648     wchar_t *bufPtr = buf;
 649     const char *pszPtr = psz;
 650
 651     if (buf)
 652     {
 653         // have destination buffer, convert there
 654         cres = iconv(m2w,
 655                      ICONV_CHAR_CAST(&pszPtr), &inbuf,
 656                      (char**)&bufPtr, &outbuf);
 657         res = n - (outbuf / SIZEOF_WCHAR_T);
 658
 659         if (ms_wcNeedsSwap)
 660         {
 661             // convert to native endianness
 662             WC_BSWAP(buf /* _not_ bufPtr */, res)
 663         }
 664
 665         // iconv doesn't seem to set the trailing 0
 666         buf[res] = 0;
 667     }
 668     else
 669     {
 670         // no destination buffer... convert using temp buffer
 671         // to calculate destination buffer requirement
 672         wchar_t tbuf[8];
 673         res = 0;
 674         do {
 675             bufPtr = tbuf;
 676             outbuf = 8*SIZEOF_WCHAR_T;
 677
 678             cres = iconv(m2w,
 679                          ICONV_CHAR_CAST(&pszPtr), &inbuf,
 680                          (char**)&bufPtr, &outbuf );
 681
 682             res += 8-(outbuf/SIZEOF_WCHAR_T);
 683         } while ((cres==(size_t)-1) && (errno==E2BIG));
 684     }
 685
 686     if (ICONV_FAILED(cres, inbuf))
 687     {
 688         //VS: it is ok if iconv fails, hence trace only
 689         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 690         return (size_t)-1;
 691     }
 692
 693     return res;
 694 }
 695
 696 size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
 697 {
 698 #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
 699     size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
 700 #else
 701     size_t inbuf = ::wcslen(psz) * SIZEOF_WCHAR_T;
 702 #endif
 703     size_t outbuf = n;
 704     size_t res, cres;
 705
 706     wchar_t *tmpbuf = 0;
 707
 708     if (ms_wcNeedsSwap)
 709     {
 710         // need to copy to temp buffer to switch endianness
 711         // this absolutely doesn't rock!
 712         // (no, doing WC_BSWAP twice on the original buffer won't help, as it
 713         //  could be in read-only memory, or be accessed in some other thread)
 714         tmpbuf=(wchar_t*)malloc((inbuf+1)*SIZEOF_WCHAR_T);
 715         memcpy(tmpbuf,psz,(inbuf+1)*SIZEOF_WCHAR_T);
 716         WC_BSWAP(tmpbuf, inbuf)
 717         psz=tmpbuf;
 718     }
 719
 720     if (buf)
 721     {
 722         // have destination buffer, convert there
 723         cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 724
 725         res = n-outbuf;
 726     }
 727     else
 728     {
 729         // no destination buffer... convert using temp buffer
 730         // to calculate destination buffer requirement
 731         char tbuf[16];
 732         res = 0;
 733         do {
 734             buf = tbuf; outbuf = 16;
 735
 736             cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
 737
 738             res += 16 - outbuf;
 739         } while ((cres==(size_t)-1) && (errno==E2BIG));
 740     }
 741
 742     if (ms_wcNeedsSwap)
 743     {
 744         free(tmpbuf);
 745     }
 746
 747     if (ICONV_FAILED(cres, inbuf))
 748     {
 749         //VS: it is ok if iconv fails, hence trace only
 750         wxLogTrace(wxT("strconv"), wxT("iconv failed: %s"), wxSysErrorMsg(wxSysErrorCode()));
 751         return (size_t)-1;
 752     }
 753
 754     return res;
 755 }
 756
 757 #endif // HAVE_ICONV
 758
 759 // ============================================================================
 760 // Win32 conversion classes
 761 // ============================================================================
 762
 763 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 764
 765 extern long wxCharsetToCodepage(const wxChar *charset); // from utils.cpp
 766
 767 class CP_CharSet : public wxCharacterSet
 768 {
 769 public:
 770     CP_CharSet(const wxChar* name)
 771         : wxCharacterSet(name)
 772         {
 773             m_CodePage = wxCharsetToCodepage(name);
 774         }
 775
 776     size_t MB2WC(wchar_t *buf, const char *psz, size_t n)
 777     {
 778         size_t len =
 779             MultiByteToWideChar(m_CodePage, 0, psz, -1, buf, buf ? n : 0);
 780         //VS: returns # of written chars for buf!=NULL and *size*
 781         //    needed buffer for buf==NULL
 782         return len ? (buf ? len : len-1) : (size_t)-1;
 783     }
 784
 785     size_t WC2MB(char *buf, const wchar_t *psz, size_t n)
 786     {
 787         size_t len = WideCharToMultiByte(m_CodePage, 0, psz, -1, buf,
 788                                          buf ? n : 0, NULL, NULL);
 789         //VS: returns # of written chars for buf!=NULL and *size*
 790         //    needed buffer for buf==NULL
 791         return len ? (buf ? len : len-1) : (size_t)-1;
 792     }
 793
 794     bool usable() const
 795         { return m_CodePage != -1; }
 796
 797 public:
 798     long m_CodePage;
 799 };
 800 #endif // __WIN32__
 801
 802 // ============================================================================
 803 // wxEncodingConverter based conversion classes
 804 // ============================================================================
 805
 806 #if wxUSE_FONTMAP
 807
 808 class EC_CharSet : public wxCharacterSet
 809 {
 810 public:
 811     // temporarily just use wxEncodingConverter stuff,
 812     // so that it works while a better implementation is built
 813     EC_CharSet(const wxChar* name) : wxCharacterSet(name),
 814                                      enc(wxFONTENCODING_SYSTEM)
 815     {
 816         if (name)
 817             enc = wxFontMapper::Get()->CharsetToEncoding(name, FALSE);
 818
 819         m_ok = m2w.Init(enc, wxFONTENCODING_UNICODE) &&
 820                w2m.Init(wxFONTENCODING_UNICODE, enc);
 821     }
 822
 823     size_t MB2WC(wchar_t *buf, const char *psz, size_t WXUNUSED(n))
 824     {
 825         size_t inbuf = strlen(psz);
 826         if (buf)
 827             m2w.Convert(psz,buf);
 828         return inbuf;
 829     }
 830
 831     size_t WC2MB(char *buf, const wchar_t *psz, size_t WXUNUSED(n))
 832     {
 833 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
 834     || ( defined(__MWERKS__) && defined(__WXMSW__) )
 835         size_t inbuf = std::wcslen(psz);
 836 #else
 837         size_t inbuf = ::wcslen(psz);
 838 #endif
 839         if (buf)
 840             w2m.Convert(psz,buf);
 841
 842         return inbuf;
 843     }
 844
 845     bool usable() const { return m_ok; }
 846
 847 public:
 848     wxFontEncoding enc;
 849     wxEncodingConverter m2w, w2m;
 850
 851     // were we initialized successfully?
 852     bool m_ok;
 853 };
 854
 855 #endif // wxUSE_FONTMAP
 856
 857 // ----------------------------------------------------------------------------
 858 // the function creating the wxCharacterSet for the specified charset on the
 859 // current system, trying all possibilities
 860 // ----------------------------------------------------------------------------
 861
 862 static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
 863 {
 864     // check for the special case of ASCII charset
 865 #if wxUSE_FONTMAP
 866     if ( wxFontMapper::Get()->CharsetToEncoding(name) == wxFONTENCODING_DEFAULT )
 867 #else // wxUSE_FONTMAP
 868     if ( !name )
 869 #endif // wxUSE_FONTMAP/!wxUSE_FONTMAP
 870     {
 871         // don't convert at all
 872         return NULL;
 873     }
 874
 875     // the test above must have taken care of this case
 876     wxCHECK_MSG( name, NULL, _T("NULL name must be wxFONTENCODING_DEFAULT") );
 877
 878     wxCharacterSet *cset;
 879
 880     if ( wxStricmp(name, wxT("UTF8")) == 0 || wxStricmp(name, wxT("UTF-8")) == 0)
 881     {
 882         cset = new ID_CharSet(name, &wxConvUTF8);
 883     }
 884     else
 885     {
 886 #ifdef HAVE_ICONV
 887         cset = new IC_CharSet(name);
 888 #else // !HAVE_ICONV
 889         cset = NULL;
 890 #endif // HAVE_ICONV/!HAVE_ICONV
 891     }
 892
 893     // it can only be NULL in this case
 894 #ifndef HAVE_ICONV
 895     if ( cset )
 896 #endif // !HAVE_ICONV
 897     {
 898         if ( cset->usable() )
 899             return cset;
 900
 901         delete cset;
 902         cset = NULL;
 903     }
 904
 905 #if defined(__WIN32__) && !defined(__WXMICROWIN__)
 906     cset = new CP_CharSet(name);
 907     if ( cset->usable() )
 908         return cset;
 909
 910     delete cset;
 911     cset = NULL;
 912 #endif // __WIN32__
 913
 914 #if wxUSE_FONTMAP
 915     cset = new EC_CharSet(name);
 916     if ( cset->usable() )
 917         return cset;
 918
 919     delete cset;
 920     cset = NULL;
 921 #endif // wxUSE_FONTMAP
 922
 923     wxLogError(_("Cannot convert from encoding '%s'!"), name);
 924
 925     return NULL;
 926 }
 927
 928 // ============================================================================
 929 // wxCSConv implementation
 930 // ============================================================================
 931
 932 wxCSConv::wxCSConv(const wxChar *charset)
 933 {
 934     m_name = (wxChar *)NULL;
 935     m_cset = (wxCharacterSet *) NULL;
 936     m_deferred = TRUE;
 937
 938     SetName(charset);
 939 }
 940
 941 wxCSConv::~wxCSConv()
 942 {
 943     Clear();
 944 }
 945
 946 wxCSConv::wxCSConv(const wxCSConv& conv)
 947     : wxMBConv()
 948 {
 949     Clear();
 950     SetName(conv.m_name);
 951 }
 952
 953 wxCSConv& wxCSConv::operator=(const wxCSConv& conv)
 954 {
 955     Clear();
 956     SetName(conv.m_name);
 957     return *this;
 958 }
 959
 960 void wxCSConv::Clear()
 961 {
 962     if (m_name)
 963         free(m_name);
 964     if (m_cset)
 965         delete m_cset;
 966     m_name = NULL;
 967     m_cset = NULL;
 968 }
 969
 970 void wxCSConv::SetName(const wxChar *charset)
 971 {
 972     if (charset)
 973     {
 974         m_name = wxStrdup(charset);
 975         m_deferred = TRUE;
 976     }
 977 }
 978
 979 void wxCSConv::LoadNow()
 980 {
 981     if (m_deferred)
 982     {
 983         if ( !m_name )
 984         {
 985             wxString name = wxLocale::GetSystemEncodingName();
 986             if ( !name.empty() )
 987             {
 988                 SetName(name);
 989             }
 990         }
 991
 992         // wxGetCharacterSet() complains about NULL name
 993         m_cset = m_name ? wxGetCharacterSet(m_name) : NULL;
 994         m_deferred = FALSE;
 995     }
 996 }
 997
 998 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
 999 {
1000     ((wxCSConv *)this)->LoadNow(); // discard constness
1001
1002     if (m_cset)
1003         return m_cset->MB2WC(buf, psz, n);
1004
1005     // latin-1 (direct)
1006     size_t len = strlen(psz);
1007
1008     if (buf)
1009     {
1010         for (size_t c = 0; c <= len; c++)
1011             buf[c] = (unsigned char)(psz[c]);
1012     }
1013
1014     return len;
1015 }
1016
1017 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1018 {
1019     ((wxCSConv *)this)->LoadNow(); // discard constness
1020
1021     if (m_cset)
1022         return m_cset->WC2MB(buf, psz, n);
1023
1024     // latin-1 (direct)
1025 #if ( defined(__BORLANDC__) && (__BORLANDC__ > 0x530) ) \
1026     || ( defined(__MWERKS__) && defined(__WXMSW__) )
1027     size_t len=std::wcslen(psz);
1028 #else
1029     size_t len=::wcslen(psz);
1030 #endif
1031     if (buf)
1032     {
1033         for (size_t c = 0; c <= len; c++)
1034             buf[c] = (psz[c] > 0xff) ? '?' : psz[c];
1035     }
1036
1037     return len;
1038 }
1039
1040 #endif // wxUSE_WCHAR_T
1041
1042