src/common/string.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        string.cpp
   3 // Purpose:     wxString class
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifdef __GNUG__
  13   #pragma implementation "string.h"
  14 #endif
  15
  16 /*
  17  * About ref counting:
  18  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  19  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  20  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  21  */
  22
  23 // ===========================================================================
  24 // headers, declarations, constants
  25 // ===========================================================================
  26
  27 // For compilers that support precompilation, includes "wx.h".
  28 #include "wx/wxprec.h"
  29
  30 #ifdef __BORLANDC__
  31   #pragma hdrstop
  32 #endif
  33
  34 #ifndef WX_PRECOMP
  35   #include "wx/defs.h"
  36   #include "wx/string.h"
  37   #include "wx/intl.h"
  38 #if wxUSE_THREADS
  39   #include <wx/thread.h>
  40 #endif
  41 #endif
  42
  43 #include <ctype.h>
  44 #include <string.h>
  45 #include <stdlib.h>
  46
  47 #ifdef __SALFORDC__
  48   #include <clib.h>
  49 #endif
  50
  51 #if wxUSE_WCSRTOMBS
  52   #include <wchar.h>    // for wcsrtombs(), see comments where it's used
  53 #endif // GNU
  54
  55 #ifdef  WXSTRING_IS_WXOBJECT
  56   IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
  57 #endif  //WXSTRING_IS_WXOBJECT
  58
  59 #if wxUSE_UNICODE
  60 #undef wxUSE_EXPERIMENTAL_PRINTF
  61 #define wxUSE_EXPERIMENTAL_PRINTF 1
  62 #endif
  63
  64 // allocating extra space for each string consumes more memory but speeds up
  65 // the concatenation operations (nLen is the current string's length)
  66 // NB: EXTRA_ALLOC must be >= 0!
  67 #define EXTRA_ALLOC       (19 - nLen % 16)
  68
  69 // ---------------------------------------------------------------------------
  70 // static class variables definition
  71 // ---------------------------------------------------------------------------
  72
  73 #ifdef  wxSTD_STRING_COMPATIBILITY
  74   const size_t wxString::npos = wxSTRING_MAXLEN;
  75 #endif // wxSTD_STRING_COMPATIBILITY
  76
  77 // ----------------------------------------------------------------------------
  78 // static data
  79 // ----------------------------------------------------------------------------
  80
  81 // for an empty string, GetStringData() will return this address: this
  82 // structure has the same layout as wxStringData and it's data() method will
  83 // return the empty string (dummy pointer)
  84 static const struct
  85 {
  86   wxStringData data;
  87   wxChar dummy;
  88 } g_strEmpty = { {-1, 0, 0}, _T('\0') };
  89
  90 // empty C style string: points to 'string data' byte of g_strEmpty
  91 extern const wxChar WXDLLEXPORT *g_szNul = &g_strEmpty.dummy;
  92
  93 // ----------------------------------------------------------------------------
  94 // conditional compilation
  95 // ----------------------------------------------------------------------------
  96
  97 #if !defined(__WXSW__) && wxUSE_UNICODE
  98   #ifdef wxUSE_EXPERIMENTAL_PRINTF
  99     #undef wxUSE_EXPERIMENTAL_PRINTF
 100   #endif
 101   #define wxUSE_EXPERIMENTAL_PRINTF 1
 102 #endif
 103
 104 // we want to find out if the current platform supports vsnprintf()-like
 105 // function: for Unix this is done with configure, for Windows we test the
 106 // compiler explicitly.
 107 #ifdef __WXMSW__
 108     #ifdef __VISUALC__
 109         #define wxVsnprintf     _vsnprintf
 110     #endif
 111 #else   // !Windows
 112     #ifdef HAVE_VSNPRINTF
 113         #define wxVsnprintf       vsnprintf
 114     #endif
 115 #endif  // Windows/!Windows
 116
 117 #ifndef wxVsnprintf
 118     // in this case we'll use vsprintf() (which is ANSI and thus should be
 119     // always available), but it's unsafe because it doesn't check for buffer
 120     // size - so give a warning
 121     #define wxVsnprintf(buffer,len,format,argptr) vsprintf(buffer,format, argptr)
 122
 123     #if defined(__VISUALC__)
 124         #pragma message("Using sprintf() because no snprintf()-like function defined")
 125     #elif defined(__GNUG__) && !defined(__UNIX__)
 126         #warning "Using sprintf() because no snprintf()-like function defined"
 127     #elif defined(__MWERKS__)
 128              #warning "Using sprintf() because no snprintf()-like function defined"
 129     #endif //compiler
 130 #endif // no vsnprintf
 131
 132 #ifdef _AIX
 133   // AIX has vsnprintf, but there's no prototype in the system headers.
 134   extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
 135 #endif
 136
 137 // ----------------------------------------------------------------------------
 138 // global functions
 139 // ----------------------------------------------------------------------------
 140
 141 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
 142
 143 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
 144 // iostream ones.
 145 //
 146 // ATTN: you can _not_ use both of these in the same program!
 147
 148 istream& operator>>(istream& is, wxString& WXUNUSED(str))
 149 {
 150 #if 0
 151   int w = is.width(0);
 152   if ( is.ipfx(0) ) {
 153     streambuf *sb = is.rdbuf();
 154     str.erase();
 155     while ( true ) {
 156       int ch = sb->sbumpc ();
 157       if ( ch == EOF ) {
 158         is.setstate(ios::eofbit);
 159         break;
 160       }
 161       else if ( isspace(ch) ) {
 162         sb->sungetc();
 163         break;
 164       }
 165
 166       str += ch;
 167       if ( --w == 1 )
 168         break;
 169     }
 170   }
 171
 172   is.isfx();
 173   if ( str.length() == 0 )
 174     is.setstate(ios::failbit);
 175 #endif
 176   return is;
 177 }
 178
 179 #endif  //std::string compatibility
 180
 181 // ----------------------------------------------------------------------------
 182 // private classes
 183 // ----------------------------------------------------------------------------
 184
 185 // this small class is used to gather statistics for performance tuning
 186 //#define WXSTRING_STATISTICS
 187 #ifdef  WXSTRING_STATISTICS
 188   class Averager
 189   {
 190   public:
 191     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 192    ~Averager()
 193    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 194
 195     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 196
 197   private:
 198     size_t m_nCount, m_nTotal;
 199     const char *m_sz;
 200   } g_averageLength("allocation size"),
 201     g_averageSummandLength("summand length"),
 202     g_averageConcatHit("hit probability in concat"),
 203     g_averageInitialLength("initial string length");
 204
 205   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 206 #else
 207   #define STATISTICS_ADD(av, val)
 208 #endif // WXSTRING_STATISTICS
 209
 210 // ===========================================================================
 211 // wxString class core
 212 // ===========================================================================
 213
 214 // ---------------------------------------------------------------------------
 215 // construction
 216 // ---------------------------------------------------------------------------
 217
 218 // constructs string of <nLength> copies of character <ch>
 219 wxString::wxString(wxChar ch, size_t nLength)
 220 {
 221   Init();
 222
 223   if ( nLength > 0 ) {
 224     AllocBuffer(nLength);
 225
 226 #if wxUSE_UNICODE
 227     // memset only works on char
 228     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 229 #else
 230     memset(m_pchData, ch, nLength);
 231 #endif
 232   }
 233 }
 234
 235 // takes nLength elements of psz starting at nPos
 236 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 237 {
 238   Init();
 239
 240   wxASSERT( nPos <= wxStrlen(psz) );
 241
 242   if ( nLength == wxSTRING_MAXLEN )
 243     nLength = wxStrlen(psz + nPos);
 244
 245   STATISTICS_ADD(InitialLength, nLength);
 246
 247   if ( nLength > 0 ) {
 248     // trailing '\0' is written in AllocBuffer()
 249     AllocBuffer(nLength);
 250     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 251   }
 252 }
 253
 254 #ifdef  wxSTD_STRING_COMPATIBILITY
 255
 256 // poor man's iterators are "void *" pointers
 257 wxString::wxString(const void *pStart, const void *pEnd)
 258 {
 259   InitWith((const wxChar *)pStart, 0,
 260            (const wxChar *)pEnd - (const wxChar *)pStart);
 261 }
 262
 263 #endif  //std::string compatibility
 264
 265 #if wxUSE_UNICODE
 266
 267 // from multibyte string
 268 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 269 {
 270   // first get necessary size
 271   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 272
 273   // nLength is number of *Unicode* characters here!
 274   if ((nLen != (size_t)-1) && (nLen > nLength))
 275     nLen = nLength;
 276
 277   // empty?
 278   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 279     AllocBuffer(nLen);
 280     conv.MB2WC(m_pchData, psz, nLen);
 281   }
 282   else {
 283     Init();
 284   }
 285 }
 286
 287 #else
 288
 289 #if wxUSE_WCHAR_T
 290 // from wide string
 291 wxString::wxString(const wchar_t *pwz)
 292 {
 293   // first get necessary size
 294   size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0;
 295
 296   // empty?
 297   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 298     AllocBuffer(nLen);
 299     wxWC2MB(m_pchData, pwz, nLen);
 300   }
 301   else {
 302     Init();
 303   }
 304 }
 305 #endif
 306
 307 #endif
 308
 309 // ---------------------------------------------------------------------------
 310 // memory allocation
 311 // ---------------------------------------------------------------------------
 312
 313 // allocates memory needed to store a C string of length nLen
 314 void wxString::AllocBuffer(size_t nLen)
 315 {
 316   wxASSERT( nLen >  0         );    //
 317   wxASSERT( nLen <= INT_MAX-1 );    // max size (enough room for 1 extra)
 318
 319   STATISTICS_ADD(Length, nLen);
 320
 321   // allocate memory:
 322   // 1) one extra character for '\0' termination
 323   // 2) sizeof(wxStringData) for housekeeping info
 324   wxStringData* pData = (wxStringData*)
 325     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 326   pData->nRefs        = 1;
 327   pData->nDataLength  = nLen;
 328   pData->nAllocLength = nLen + EXTRA_ALLOC;
 329   m_pchData           = pData->data();  // data starts after wxStringData
 330   m_pchData[nLen]     = _T('\0');
 331 }
 332
 333 // must be called before changing this string
 334 void wxString::CopyBeforeWrite()
 335 {
 336   wxStringData* pData = GetStringData();
 337
 338   if ( pData->IsShared() ) {
 339     pData->Unlock();                // memory not freed because shared
 340     size_t nLen = pData->nDataLength;
 341     AllocBuffer(nLen);
 342     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 343   }
 344
 345   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 346 }
 347
 348 // must be called before replacing contents of this string
 349 void wxString::AllocBeforeWrite(size_t nLen)
 350 {
 351   wxASSERT( nLen != 0 );  // doesn't make any sense
 352
 353   // must not share string and must have enough space
 354   wxStringData* pData = GetStringData();
 355   if ( pData->IsShared() || (nLen > pData->nAllocLength) ) {
 356     // can't work with old buffer, get new one
 357     pData->Unlock();
 358     AllocBuffer(nLen);
 359   }
 360   else {
 361     // update the string length
 362     pData->nDataLength = nLen;
 363   }
 364
 365   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 366 }
 367
 368 // allocate enough memory for nLen characters
 369 void wxString::Alloc(size_t nLen)
 370 {
 371   wxStringData *pData = GetStringData();
 372   if ( pData->nAllocLength <= nLen ) {
 373     if ( pData->IsEmpty() ) {
 374       nLen += EXTRA_ALLOC;
 375
 376       wxStringData* pData = (wxStringData*)
 377         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 378       pData->nRefs = 1;
 379       pData->nDataLength = 0;
 380       pData->nAllocLength = nLen;
 381       m_pchData = pData->data();  // data starts after wxStringData
 382       m_pchData[0u] = _T('\0');
 383     }
 384     else if ( pData->IsShared() ) {
 385       pData->Unlock();                // memory not freed because shared
 386       size_t nOldLen = pData->nDataLength;
 387       AllocBuffer(nLen);
 388       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 389     }
 390     else {
 391       nLen += EXTRA_ALLOC;
 392
 393       wxStringData *p = (wxStringData *)
 394         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 395
 396       if ( p == NULL ) {
 397         // @@@ what to do on memory error?
 398         return;
 399       }
 400
 401       // it's not important if the pointer changed or not (the check for this
 402       // is not faster than assigning to m_pchData in all cases)
 403       p->nAllocLength = nLen;
 404       m_pchData = p->data();
 405     }
 406   }
 407   //else: we've already got enough
 408 }
 409
 410 // shrink to minimal size (releasing extra memory)
 411 void wxString::Shrink()
 412 {
 413   wxStringData *pData = GetStringData();
 414
 415   // this variable is unused in release build, so avoid the compiler warning by
 416   // just not declaring it
 417 #ifdef __WXDEBUG__
 418   void *p =
 419 #endif
 420   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 421
 422   wxASSERT( p != NULL );  // can't free memory?
 423   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 424 }
 425
 426 // get the pointer to writable buffer of (at least) nLen bytes
 427 wxChar *wxString::GetWriteBuf(size_t nLen)
 428 {
 429   AllocBeforeWrite(nLen);
 430
 431   wxASSERT( GetStringData()->nRefs == 1 );
 432   GetStringData()->Validate(FALSE);
 433
 434   return m_pchData;
 435 }
 436
 437 // put string back in a reasonable state after GetWriteBuf
 438 void wxString::UngetWriteBuf()
 439 {
 440   GetStringData()->nDataLength = wxStrlen(m_pchData);
 441   GetStringData()->Validate(TRUE);
 442 }
 443
 444 // ---------------------------------------------------------------------------
 445 // data access
 446 // ---------------------------------------------------------------------------
 447
 448 // all functions are inline in string.h
 449
 450 // ---------------------------------------------------------------------------
 451 // assignment operators
 452 // ---------------------------------------------------------------------------
 453
 454 // helper function: does real copy
 455 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 456 {
 457   if ( nSrcLen == 0 ) {
 458     Reinit();
 459   }
 460   else {
 461     AllocBeforeWrite(nSrcLen);
 462     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 463     GetStringData()->nDataLength = nSrcLen;
 464     m_pchData[nSrcLen] = _T('\0');
 465   }
 466 }
 467
 468 // assigns one string to another
 469 wxString& wxString::operator=(const wxString& stringSrc)
 470 {
 471   wxASSERT( stringSrc.GetStringData()->IsValid() );
 472
 473   // don't copy string over itself
 474   if ( m_pchData != stringSrc.m_pchData ) {
 475     if ( stringSrc.GetStringData()->IsEmpty() ) {
 476       Reinit();
 477     }
 478     else {
 479       // adjust references
 480       GetStringData()->Unlock();
 481       m_pchData = stringSrc.m_pchData;
 482       GetStringData()->Lock();
 483     }
 484   }
 485
 486   return *this;
 487 }
 488
 489 // assigns a single character
 490 wxString& wxString::operator=(wxChar ch)
 491 {
 492   AssignCopy(1, &ch);
 493   return *this;
 494 }
 495
 496 // assigns C string
 497 wxString& wxString::operator=(const wxChar *psz)
 498 {
 499   AssignCopy(wxStrlen(psz), psz);
 500   return *this;
 501 }
 502
 503 #if !wxUSE_UNICODE
 504
 505 // same as 'signed char' variant
 506 wxString& wxString::operator=(const unsigned char* psz)
 507 {
 508   *this = (const char *)psz;
 509   return *this;
 510 }
 511
 512 #if wxUSE_WCHAR_T
 513 wxString& wxString::operator=(const wchar_t *pwz)
 514 {
 515   wxString str(pwz);
 516   *this = str;
 517   return *this;
 518 }
 519 #endif
 520
 521 #endif
 522
 523 // ---------------------------------------------------------------------------
 524 // string concatenation
 525 // ---------------------------------------------------------------------------
 526
 527 // add something to this string
 528 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 529 {
 530   STATISTICS_ADD(SummandLength, nSrcLen);
 531
 532   // concatenating an empty string is a NOP
 533   if ( nSrcLen > 0 ) {
 534     wxStringData *pData = GetStringData();
 535     size_t nLen = pData->nDataLength;
 536     size_t nNewLen = nLen + nSrcLen;
 537
 538     // alloc new buffer if current is too small
 539     if ( pData->IsShared() ) {
 540       STATISTICS_ADD(ConcatHit, 0);
 541
 542       // we have to allocate another buffer
 543       wxStringData* pOldData = GetStringData();
 544       AllocBuffer(nNewLen);
 545       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 546       pOldData->Unlock();
 547     }
 548     else if ( nNewLen > pData->nAllocLength ) {
 549       STATISTICS_ADD(ConcatHit, 0);
 550
 551       // we have to grow the buffer
 552       Alloc(nNewLen);
 553     }
 554     else {
 555       STATISTICS_ADD(ConcatHit, 1);
 556
 557       // the buffer is already big enough
 558     }
 559
 560     // should be enough space
 561     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 562
 563     // fast concatenation - all is done in our buffer
 564     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 565
 566     m_pchData[nNewLen] = _T('\0');          // put terminating '\0'
 567     GetStringData()->nDataLength = nNewLen; // and fix the length
 568   }
 569   //else: the string to append was empty
 570 }
 571
 572 /*
 573  * concatenation functions come in 5 flavours:
 574  *  string + string
 575  *  char   + string      and      string + char
 576  *  C str  + string      and      string + C str
 577  */
 578
 579 wxString operator+(const wxString& string1, const wxString& string2)
 580 {
 581   wxASSERT( string1.GetStringData()->IsValid() );
 582   wxASSERT( string2.GetStringData()->IsValid() );
 583
 584   wxString s = string1;
 585   s += string2;
 586
 587   return s;
 588 }
 589
 590 wxString operator+(const wxString& string, wxChar ch)
 591 {
 592   wxASSERT( string.GetStringData()->IsValid() );
 593
 594   wxString s = string;
 595   s += ch;
 596
 597   return s;
 598 }
 599
 600 wxString operator+(wxChar ch, const wxString& string)
 601 {
 602   wxASSERT( string.GetStringData()->IsValid() );
 603
 604   wxString s = ch;
 605   s += string;
 606
 607   return s;
 608 }
 609
 610 wxString operator+(const wxString& string, const wxChar *psz)
 611 {
 612   wxASSERT( string.GetStringData()->IsValid() );
 613
 614   wxString s;
 615   s.Alloc(wxStrlen(psz) + string.Len());
 616   s = string;
 617   s += psz;
 618
 619   return s;
 620 }
 621
 622 wxString operator+(const wxChar *psz, const wxString& string)
 623 {
 624   wxASSERT( string.GetStringData()->IsValid() );
 625
 626   wxString s;
 627   s.Alloc(wxStrlen(psz) + string.Len());
 628   s = psz;
 629   s += string;
 630
 631   return s;
 632 }
 633
 634 // ===========================================================================
 635 // other common string functions
 636 // ===========================================================================
 637
 638 // ---------------------------------------------------------------------------
 639 // simple sub-string extraction
 640 // ---------------------------------------------------------------------------
 641
 642 // helper function: clone the data attached to this string
 643 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 644 {
 645   if ( nCopyLen == 0 ) {
 646     dest.Init();
 647   }
 648   else {
 649     dest.AllocBuffer(nCopyLen);
 650     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 651   }
 652 }
 653
 654 // extract string of length nCount starting at nFirst
 655 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 656 {
 657   wxStringData *pData = GetStringData();
 658   size_t nLen = pData->nDataLength;
 659
 660   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 661   if ( nCount == wxSTRING_MAXLEN )
 662   {
 663     nCount = nLen - nFirst;
 664   }
 665
 666   // out-of-bounds requests return sensible things
 667   if ( nFirst + nCount > nLen )
 668   {
 669     nCount = nLen - nFirst;
 670   }
 671
 672   if ( nFirst > nLen )
 673   {
 674     // AllocCopy() will return empty string
 675     nCount = 0;
 676   }
 677
 678   wxString dest;
 679   AllocCopy(dest, nCount, nFirst);
 680
 681   return dest;
 682 }
 683
 684 // extract nCount last (rightmost) characters
 685 wxString wxString::Right(size_t nCount) const
 686 {
 687   if ( nCount > (size_t)GetStringData()->nDataLength )
 688     nCount = GetStringData()->nDataLength;
 689
 690   wxString dest;
 691   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 692   return dest;
 693 }
 694
 695 // get all characters after the last occurence of ch
 696 // (returns the whole string if ch not found)
 697 wxString wxString::AfterLast(wxChar ch) const
 698 {
 699   wxString str;
 700   int iPos = Find(ch, TRUE);
 701   if ( iPos == wxNOT_FOUND )
 702     str = *this;
 703   else
 704     str = c_str() + iPos + 1;
 705
 706   return str;
 707 }
 708
 709 // extract nCount first (leftmost) characters
 710 wxString wxString::Left(size_t nCount) const
 711 {
 712   if ( nCount > (size_t)GetStringData()->nDataLength )
 713     nCount = GetStringData()->nDataLength;
 714
 715   wxString dest;
 716   AllocCopy(dest, nCount, 0);
 717   return dest;
 718 }
 719
 720 // get all characters before the first occurence of ch
 721 // (returns the whole string if ch not found)
 722 wxString wxString::BeforeFirst(wxChar ch) const
 723 {
 724   wxString str;
 725   for ( const wxChar *pc = m_pchData; *pc != _T('\0') && *pc != ch; pc++ )
 726     str += *pc;
 727
 728   return str;
 729 }
 730
 731 /// get all characters before the last occurence of ch
 732 /// (returns empty string if ch not found)
 733 wxString wxString::BeforeLast(wxChar ch) const
 734 {
 735   wxString str;
 736   int iPos = Find(ch, TRUE);
 737   if ( iPos != wxNOT_FOUND && iPos != 0 )
 738     str = wxString(c_str(), iPos);
 739
 740   return str;
 741 }
 742
 743 /// get all characters after the first occurence of ch
 744 /// (returns empty string if ch not found)
 745 wxString wxString::AfterFirst(wxChar ch) const
 746 {
 747   wxString str;
 748   int iPos = Find(ch);
 749   if ( iPos != wxNOT_FOUND )
 750     str = c_str() + iPos + 1;
 751
 752   return str;
 753 }
 754
 755 // replace first (or all) occurences of some substring with another one
 756 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 757 {
 758   size_t uiCount = 0;   // count of replacements made
 759
 760   size_t uiOldLen = wxStrlen(szOld);
 761
 762   wxString strTemp;
 763   const wxChar *pCurrent = m_pchData;
 764   const wxChar *pSubstr;
 765   while ( *pCurrent != _T('\0') ) {
 766     pSubstr = wxStrstr(pCurrent, szOld);
 767     if ( pSubstr == NULL ) {
 768       // strTemp is unused if no replacements were made, so avoid the copy
 769       if ( uiCount == 0 )
 770         return 0;
 771
 772       strTemp += pCurrent;    // copy the rest
 773       break;                  // exit the loop
 774     }
 775     else {
 776       // take chars before match
 777       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 778       strTemp += szNew;
 779       pCurrent = pSubstr + uiOldLen;  // restart after match
 780
 781       uiCount++;
 782
 783       // stop now?
 784       if ( !bReplaceAll ) {
 785         strTemp += pCurrent;    // copy the rest
 786         break;                  // exit the loop
 787       }
 788     }
 789   }
 790
 791   // only done if there were replacements, otherwise would have returned above
 792   *this = strTemp;
 793
 794   return uiCount;
 795 }
 796
 797 bool wxString::IsAscii() const
 798 {
 799   const wxChar *s = (const wxChar*) *this;
 800   while(*s){
 801     if(!isascii(*s)) return(FALSE);
 802     s++;
 803   }
 804   return(TRUE);
 805 }
 806
 807 bool wxString::IsWord() const
 808 {
 809   const wxChar *s = (const wxChar*) *this;
 810   while(*s){
 811     if(!wxIsalpha(*s)) return(FALSE);
 812     s++;
 813   }
 814   return(TRUE);
 815 }
 816
 817 bool wxString::IsNumber() const
 818 {
 819   const wxChar *s = (const wxChar*) *this;
 820   while(*s){
 821     if(!wxIsdigit(*s)) return(FALSE);
 822     s++;
 823   }
 824   return(TRUE);
 825 }
 826
 827 wxString wxString::Strip(stripType w) const
 828 {
 829     wxString s = *this;
 830     if ( w & leading ) s.Trim(FALSE);
 831     if ( w & trailing ) s.Trim(TRUE);
 832     return s;
 833 }
 834
 835 // ---------------------------------------------------------------------------
 836 // case conversion
 837 // ---------------------------------------------------------------------------
 838
 839 wxString& wxString::MakeUpper()
 840 {
 841   CopyBeforeWrite();
 842
 843   for ( wxChar *p = m_pchData; *p; p++ )
 844     *p = (wxChar)wxToupper(*p);
 845
 846   return *this;
 847 }
 848
 849 wxString& wxString::MakeLower()
 850 {
 851   CopyBeforeWrite();
 852
 853   for ( wxChar *p = m_pchData; *p; p++ )
 854     *p = (wxChar)wxTolower(*p);
 855
 856   return *this;
 857 }
 858
 859 // ---------------------------------------------------------------------------
 860 // trimming and padding
 861 // ---------------------------------------------------------------------------
 862
 863 // trims spaces (in the sense of isspace) from left or right side
 864 wxString& wxString::Trim(bool bFromRight)
 865 {
 866   // first check if we're going to modify the string at all
 867   if ( !IsEmpty() &&
 868        (
 869         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 870         (!bFromRight && wxIsspace(GetChar(0u)))
 871        )
 872      )
 873   {
 874     // ok, there is at least one space to trim
 875     CopyBeforeWrite();
 876
 877     if ( bFromRight )
 878     {
 879       // find last non-space character
 880       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
 881       while ( wxIsspace(*psz) && (psz >= m_pchData) )
 882         psz--;
 883
 884       // truncate at trailing space start
 885       *++psz = _T('\0');
 886       GetStringData()->nDataLength = psz - m_pchData;
 887     }
 888     else
 889     {
 890       // find first non-space character
 891       const wxChar *psz = m_pchData;
 892       while ( wxIsspace(*psz) )
 893         psz++;
 894
 895       // fix up data and length
 896       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
 897       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
 898       GetStringData()->nDataLength = nDataLength;
 899     }
 900   }
 901
 902   return *this;
 903 }
 904
 905 // adds nCount characters chPad to the string from either side
 906 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
 907 {
 908   wxString s(chPad, nCount);
 909
 910   if ( bFromRight )
 911     *this += s;
 912   else
 913   {
 914     s += *this;
 915     *this = s;
 916   }
 917
 918   return *this;
 919 }
 920
 921 // truncate the string
 922 wxString& wxString::Truncate(size_t uiLen)
 923 {
 924   if ( uiLen < Len() ) {
 925     CopyBeforeWrite();
 926
 927     *(m_pchData + uiLen) = _T('\0');
 928     GetStringData()->nDataLength = uiLen;
 929   }
 930   //else: nothing to do, string is already short enough
 931
 932   return *this;
 933 }
 934
 935 // ---------------------------------------------------------------------------
 936 // finding (return wxNOT_FOUND if not found and index otherwise)
 937 // ---------------------------------------------------------------------------
 938
 939 // find a character
 940 int wxString::Find(wxChar ch, bool bFromEnd) const
 941 {
 942   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
 943
 944   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 945 }
 946
 947 // find a sub-string (like strstr)
 948 int wxString::Find(const wxChar *pszSub) const
 949 {
 950   const wxChar *psz = wxStrstr(m_pchData, pszSub);
 951
 952   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 953 }
 954
 955 // ---------------------------------------------------------------------------
 956 // stream-like operators
 957 // ---------------------------------------------------------------------------
 958 wxString& wxString::operator<<(int i)
 959 {
 960     wxString res;
 961     res.Printf(_T("%d"), i);
 962
 963     return (*this) << res;
 964 }
 965
 966 wxString& wxString::operator<<(float f)
 967 {
 968     wxString res;
 969     res.Printf(_T("%f"), f);
 970
 971     return (*this) << res;
 972 }
 973
 974 wxString& wxString::operator<<(double d)
 975 {
 976     wxString res;
 977     res.Printf(_T("%g"), d);
 978
 979     return (*this) << res;
 980 }
 981
 982 // ---------------------------------------------------------------------------
 983 // formatted output
 984 // ---------------------------------------------------------------------------
 985 int wxString::Printf(const wxChar *pszFormat, ...)
 986 {
 987   va_list argptr;
 988   va_start(argptr, pszFormat);
 989
 990   int iLen = PrintfV(pszFormat, argptr);
 991
 992   va_end(argptr);
 993
 994   return iLen;
 995 }
 996
 997 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
 998 {
 999   // static buffer to avoid dynamic memory allocation each time
1000   char s_szScratch[1024]; // using static buffer causes internal compiler err
1001 #if 0
1002 #if wxUSE_THREADS
1003   // protect the static buffer
1004   static wxCriticalSection critsect;
1005   wxCriticalSectionLocker lock(critsect);
1006 #endif
1007 #endif
1008
1009 #if wxUSE_EXPERIMENTAL_PRINTF
1010 // the new implementation
1011
1012   Reinit();
1013   for (size_t n = 0; pszFormat[n]; n++)
1014     if (pszFormat[n] == _T('%')) {
1015       static char s_szFlags[256] = "%";
1016       size_t flagofs = 1;
1017       bool adj_left = FALSE, in_prec = FALSE,
1018            prec_dot = FALSE, done = FALSE;
1019       int ilen = 0;
1020       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1021       do {
1022 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1023         switch (pszFormat[++n]) {
1024         case _T('\0'):
1025           done = TRUE;
1026           break;
1027         case _T('%'):
1028           *this += _T('%');
1029           done = TRUE;
1030           break;
1031         case _T('#'):
1032         case _T('0'):
1033         case _T(' '):
1034         case _T('+'):
1035         case _T('\''):
1036           CHECK_PREC
1037           s_szFlags[flagofs++] = pszFormat[n];
1038           break;
1039         case _T('-'):
1040           CHECK_PREC
1041           adj_left = TRUE;
1042           s_szFlags[flagofs++] = pszFormat[n];
1043           break;
1044         case _T('.'):
1045           CHECK_PREC
1046           in_prec = TRUE;
1047           prec_dot = FALSE;
1048           max_width = 0;
1049           // dot will be auto-added to s_szFlags if non-negative number follows
1050           break;
1051         case _T('h'):
1052           ilen = -1;
1053           CHECK_PREC
1054           s_szFlags[flagofs++] = pszFormat[n];
1055           break;
1056         case _T('l'):
1057           ilen = 1;
1058           CHECK_PREC
1059           s_szFlags[flagofs++] = pszFormat[n];
1060           break;
1061         case _T('q'):
1062         case _T('L'):
1063           ilen = 2;
1064           CHECK_PREC
1065           s_szFlags[flagofs++] = pszFormat[n];
1066           break;
1067         case _T('Z'):
1068           ilen = 3;
1069           CHECK_PREC
1070           s_szFlags[flagofs++] = pszFormat[n];
1071           break;
1072         case _T('*'):
1073           {
1074             int len = va_arg(argptr, int);
1075             if (in_prec) {
1076               if (len<0) break;
1077               CHECK_PREC
1078               max_width = len;
1079             } else {
1080               if (len<0) {
1081                 adj_left = !adj_left;
1082                 s_szFlags[flagofs++] = '-';
1083                 len = -len;
1084               }
1085               min_width = len;
1086             }
1087             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1088           }
1089           break;
1090         case _T('1'): case _T('2'): case _T('3'):
1091         case _T('4'): case _T('5'): case _T('6'):
1092         case _T('7'): case _T('8'): case _T('9'):
1093           {
1094             int len = 0;
1095             CHECK_PREC
1096             while ((pszFormat[n]>=_T('0')) && (pszFormat[n]<=_T('9'))) {
1097               s_szFlags[flagofs++] = pszFormat[n];
1098               len = len*10 + (pszFormat[n] - _T('0'));
1099               n++;
1100             }
1101             if (in_prec) max_width = len;
1102             else min_width = len;
1103             n--; // the main loop pre-increments n again
1104           }
1105           break;
1106         case _T('d'):
1107         case _T('i'):
1108         case _T('o'):
1109         case _T('u'):
1110         case _T('x'):
1111         case _T('X'):
1112           CHECK_PREC
1113           s_szFlags[flagofs++] = pszFormat[n];
1114           s_szFlags[flagofs] = '\0';
1115           if (ilen == 0 ) {
1116             int val = va_arg(argptr, int);
1117             ::sprintf(s_szScratch, s_szFlags, val);
1118           }
1119           else if (ilen == -1) {
1120             short int val = va_arg(argptr, short int);
1121             ::sprintf(s_szScratch, s_szFlags, val);
1122           }
1123           else if (ilen == 1) {
1124             long int val = va_arg(argptr, long int);
1125             ::sprintf(s_szScratch, s_szFlags, val);
1126           }
1127           else if (ilen == 2) {
1128 #if SIZEOF_LONG_LONG
1129             long long int val = va_arg(argptr, long long int);
1130             ::sprintf(s_szScratch, s_szFlags, val);
1131 #else
1132             long int val = va_arg(argptr, long int);
1133             ::sprintf(s_szScratch, s_szFlags, val);
1134 #endif
1135           }
1136           else if (ilen == 3) {
1137             size_t val = va_arg(argptr, size_t);
1138             ::sprintf(s_szScratch, s_szFlags, val);
1139           }
1140           *this += wxString(s_szScratch);
1141           done = TRUE;
1142           break;
1143         case _T('e'):
1144         case _T('E'):
1145         case _T('f'):
1146         case _T('g'):
1147         case _T('G'):
1148           CHECK_PREC
1149           s_szFlags[flagofs++] = pszFormat[n];
1150           s_szFlags[flagofs] = '\0';
1151           if (ilen == 2) {
1152             long double val = va_arg(argptr, long double);
1153             ::sprintf(s_szScratch, s_szFlags, val);
1154           } else {
1155             double val = va_arg(argptr, double);
1156             ::sprintf(s_szScratch, s_szFlags, val);
1157           }
1158           *this += wxString(s_szScratch);
1159           done = TRUE;
1160           break;
1161         case _T('p'):
1162           {
1163             void *val = va_arg(argptr, void *);
1164             CHECK_PREC
1165             s_szFlags[flagofs++] = pszFormat[n];
1166             s_szFlags[flagofs] = '\0';
1167             ::sprintf(s_szScratch, s_szFlags, val);
1168             *this += wxString(s_szScratch);
1169             done = TRUE;
1170           }
1171           break;
1172         case _T('c'):
1173           {
1174             wxChar val = va_arg(argptr, int);
1175             // we don't need to honor padding here, do we?
1176             *this += val;
1177             done = TRUE;
1178           }
1179           break;
1180         case _T('s'):
1181           if (ilen == -1) {
1182             // wx extension: we'll let %hs mean non-Unicode strings
1183             char *val = va_arg(argptr, char *);
1184 #if wxUSE_UNICODE
1185             // ASCII->Unicode constructor handles max_width right
1186             wxString s(val, wxConvLibc, max_width);
1187 #else
1188             size_t len = wxSTRING_MAXLEN;
1189             if (val) {
1190               for (len = 0; val[len] && (len<max_width); len++);
1191             } else val = _T("(null)");
1192             wxString s(val, len);
1193 #endif
1194             if (s.Len() < min_width)
1195               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1196             *this += s;
1197           } else {
1198             wxChar *val = va_arg(argptr, wxChar *);
1199             size_t len = wxSTRING_MAXLEN;
1200             if (val) {
1201               for (len = 0; val[len] && (len<max_width); len++);
1202             } else val = _T("(null)");
1203             wxString s(val, len);
1204             if (s.Len() < min_width)
1205               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1206             *this += s;
1207           }
1208           done = TRUE;
1209           break;
1210         case _T('n'):
1211           if (ilen == 0) {
1212             int *val = va_arg(argptr, int *);
1213             *val = Len();
1214           }
1215           else if (ilen == -1) {
1216             short int *val = va_arg(argptr, short int *);
1217             *val = Len();
1218           }
1219           else if (ilen >= 1) {
1220             long int *val = va_arg(argptr, long int *);
1221             *val = Len();
1222           }
1223           done = TRUE;
1224           break;
1225         default:
1226           if (wxIsalpha(pszFormat[n]))
1227             // probably some flag not taken care of here yet
1228             s_szFlags[flagofs++] = pszFormat[n];
1229           else {
1230             // bad format
1231             *this += _T('%'); // just to pass the glibc tst-printf.c
1232             n--;
1233             done = TRUE;
1234           }
1235           break;
1236         }
1237 #undef CHECK_PREC
1238       } while (!done);
1239     } else *this += pszFormat[n];
1240
1241 #else
1242   // NB: wxVsnprintf() may return either less than the buffer size or -1 if there
1243   //     is not enough place depending on implementation
1244   int iLen = wxVsnprintf(s_szScratch, WXSIZEOF(s_szScratch), pszFormat, argptr);
1245   char *buffer;
1246   if ( iLen < (int)WXSIZEOF(s_szScratch) ) {
1247     buffer = s_szScratch;
1248   }
1249   else {
1250       int size = WXSIZEOF(s_szScratch) * 2;
1251       buffer = (char *)malloc(size);
1252       while ( buffer != NULL ) {
1253           iLen = wxVsnprintf(buffer, WXSIZEOF(s_szScratch), pszFormat, argptr);
1254           if ( iLen < size ) {
1255               // ok, there was enough space
1256               break;
1257           }
1258
1259           // still not enough, double it again
1260           buffer = (char *)realloc(buffer, size *= 2);
1261       }
1262
1263       if ( !buffer ) {
1264           // out of memory
1265           return -1;
1266       }
1267   }
1268
1269   wxString s(buffer);
1270   *this = s;
1271
1272   if ( buffer != s_szScratch )
1273       free(buffer);
1274 #endif
1275
1276   return Len();
1277 }
1278
1279 // ----------------------------------------------------------------------------
1280 // misc other operations
1281 // ----------------------------------------------------------------------------
1282 bool wxString::Matches(const wxChar *pszMask) const
1283 {
1284   // check char by char
1285   const wxChar *pszTxt;
1286   for ( pszTxt = c_str(); *pszMask != _T('\0'); pszMask++, pszTxt++ ) {
1287     switch ( *pszMask ) {
1288       case _T('?'):
1289         if ( *pszTxt == _T('\0') )
1290           return FALSE;
1291
1292         pszTxt++;
1293         pszMask++;
1294         break;
1295
1296       case _T('*'):
1297         {
1298           // ignore special chars immediately following this one
1299           while ( *pszMask == _T('*') || *pszMask == _T('?') )
1300             pszMask++;
1301
1302           // if there is nothing more, match
1303           if ( *pszMask == _T('\0') )
1304             return TRUE;
1305
1306           // are there any other metacharacters in the mask?
1307           size_t uiLenMask;
1308           const wxChar *pEndMask = wxStrpbrk(pszMask, _T("*?"));
1309
1310           if ( pEndMask != NULL ) {
1311             // we have to match the string between two metachars
1312             uiLenMask = pEndMask - pszMask;
1313           }
1314           else {
1315             // we have to match the remainder of the string
1316             uiLenMask = wxStrlen(pszMask);
1317           }
1318
1319           wxString strToMatch(pszMask, uiLenMask);
1320           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1321           if ( pMatch == NULL )
1322             return FALSE;
1323
1324           // -1 to compensate "++" in the loop
1325           pszTxt = pMatch + uiLenMask - 1;
1326           pszMask += uiLenMask - 1;
1327         }
1328         break;
1329
1330       default:
1331         if ( *pszMask != *pszTxt )
1332           return FALSE;
1333         break;
1334     }
1335   }
1336
1337   // match only if nothing left
1338   return *pszTxt == _T('\0');
1339 }
1340
1341 // Count the number of chars
1342 int wxString::Freq(wxChar ch) const
1343 {
1344     int count = 0;
1345     int len = Len();
1346     for (int i = 0; i < len; i++)
1347     {
1348         if (GetChar(i) == ch)
1349             count ++;
1350     }
1351     return count;
1352 }
1353
1354 // convert to upper case, return the copy of the string
1355 wxString wxString::Upper() const
1356 { wxString s(*this); return s.MakeUpper(); }
1357
1358 // convert to lower case, return the copy of the string
1359 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1360
1361 int wxString::sprintf(const wxChar *pszFormat, ...)
1362   {
1363     va_list argptr;
1364     va_start(argptr, pszFormat);
1365     int iLen = PrintfV(pszFormat, argptr);
1366     va_end(argptr);
1367     return iLen;
1368   }
1369
1370 // ---------------------------------------------------------------------------
1371 // standard C++ library string functions
1372 // ---------------------------------------------------------------------------
1373 #ifdef  wxSTD_STRING_COMPATIBILITY
1374
1375 wxString& wxString::insert(size_t nPos, const wxString& str)
1376 {
1377   wxASSERT( str.GetStringData()->IsValid() );
1378   wxASSERT( nPos <= Len() );
1379
1380   if ( !str.IsEmpty() ) {
1381     wxString strTmp;
1382     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1383     wxStrncpy(pc, c_str(), nPos);
1384     wxStrcpy(pc + nPos, str);
1385     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1386     strTmp.UngetWriteBuf();
1387     *this = strTmp;
1388   }
1389
1390   return *this;
1391 }
1392
1393 size_t wxString::find(const wxString& str, size_t nStart) const
1394 {
1395   wxASSERT( str.GetStringData()->IsValid() );
1396   wxASSERT( nStart <= Len() );
1397
1398   const wxChar *p = wxStrstr(c_str() + nStart, str);
1399
1400   return p == NULL ? npos : p - c_str();
1401 }
1402
1403 // VC++ 1.5 can't cope with the default argument in the header.
1404 #if !defined(__VISUALC__) || defined(__WIN32__)
1405 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1406 {
1407   return find(wxString(sz, n == npos ? 0 : n), nStart);
1408 }
1409 #endif // VC++ 1.5
1410
1411 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1412 #if !defined(__BORLANDC__)
1413 size_t wxString::find(wxChar ch, size_t nStart) const
1414 {
1415   wxASSERT( nStart <= Len() );
1416
1417   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1418
1419   return p == NULL ? npos : p - c_str();
1420 }
1421 #endif
1422
1423 size_t wxString::rfind(const wxString& str, size_t nStart) const
1424 {
1425   wxASSERT( str.GetStringData()->IsValid() );
1426   wxASSERT( nStart <= Len() );
1427
1428   // TODO could be made much quicker than that
1429   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1430   while ( p >= c_str() + str.Len() ) {
1431     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1432       return p - str.Len() - c_str();
1433     p--;
1434   }
1435
1436   return npos;
1437 }
1438
1439 // VC++ 1.5 can't cope with the default argument in the header.
1440 #if !defined(__VISUALC__) || defined(__WIN32__)
1441 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1442 {
1443     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1444 }
1445
1446 size_t wxString::rfind(wxChar ch, size_t nStart) const
1447 {
1448     if ( nStart == npos )
1449     {
1450         nStart = Len();
1451     }
1452     else
1453     {
1454         wxASSERT( nStart <= Len() );
1455     }
1456
1457     const wxChar *p = wxStrrchr(c_str(), ch);
1458
1459     if ( p == NULL )
1460         return npos;
1461
1462     size_t result = p - c_str();
1463     return ( result > nStart ) ? npos : result;
1464 }
1465 #endif // VC++ 1.5
1466
1467 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1468 {
1469     const wxChar *start = c_str() + nStart;
1470     const wxChar *firstOf = wxStrpbrk(start, sz);
1471     if ( firstOf )
1472         return firstOf - start;
1473     else
1474         return npos;
1475 }
1476
1477 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1478 {
1479     if ( nStart == npos )
1480     {
1481         nStart = Len();
1482     }
1483     else
1484     {
1485         wxASSERT( nStart <= Len() );
1486     }
1487
1488     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1489     {
1490         if ( wxStrchr(sz, *p) )
1491             return p - c_str();
1492     }
1493
1494     return npos;
1495 }
1496
1497 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1498 {
1499     if ( nStart == npos )
1500     {
1501         nStart = Len();
1502     }
1503     else
1504     {
1505         wxASSERT( nStart <= Len() );
1506     }
1507
1508     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1509     if ( nAccept >= length() - nStart )
1510         return npos;
1511     else
1512         return nAccept;
1513 }
1514
1515 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1516 {
1517     wxASSERT( nStart <= Len() );
1518
1519     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1520     {
1521         if ( *p != ch )
1522             return p - c_str();
1523     }
1524
1525     return npos;
1526 }
1527
1528 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1529 {
1530     if ( nStart == npos )
1531     {
1532         nStart = Len();
1533     }
1534     else
1535     {
1536         wxASSERT( nStart <= Len() );
1537     }
1538
1539     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1540     {
1541         if ( !wxStrchr(sz, *p) )
1542             return p - c_str();
1543     }
1544
1545     return npos;
1546 }
1547
1548 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1549 {
1550     if ( nStart == npos )
1551     {
1552         nStart = Len();
1553     }
1554     else
1555     {
1556         wxASSERT( nStart <= Len() );
1557     }
1558
1559     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1560     {
1561         if ( *p != ch )
1562             return p - c_str();
1563     }
1564
1565     return npos;
1566 }
1567
1568 wxString wxString::substr(size_t nStart, size_t nLen) const
1569 {
1570   // npos means 'take all'
1571   if ( nLen == npos )
1572     nLen = 0;
1573
1574   wxASSERT( nStart + nLen <= Len() );
1575
1576   return wxString(c_str() + nStart, nLen == npos ? 0 : nLen);
1577 }
1578
1579 wxString& wxString::erase(size_t nStart, size_t nLen)
1580 {
1581   wxString strTmp(c_str(), nStart);
1582   if ( nLen != npos ) {
1583     wxASSERT( nStart + nLen <= Len() );
1584
1585     strTmp.append(c_str() + nStart + nLen);
1586   }
1587
1588   *this = strTmp;
1589   return *this;
1590 }
1591
1592 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1593 {
1594   wxASSERT( nStart + nLen <= wxStrlen(sz) );
1595
1596   wxString strTmp;
1597   if ( nStart != 0 )
1598     strTmp.append(c_str(), nStart);
1599   strTmp += sz;
1600   strTmp.append(c_str() + nStart + nLen);
1601
1602   *this = strTmp;
1603   return *this;
1604 }
1605
1606 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1607 {
1608   return replace(nStart, nLen, wxString(ch, nCount));
1609 }
1610
1611 wxString& wxString::replace(size_t nStart, size_t nLen,
1612                             const wxString& str, size_t nStart2, size_t nLen2)
1613 {
1614   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1615 }
1616
1617 wxString& wxString::replace(size_t nStart, size_t nLen,
1618                         const wxChar* sz, size_t nCount)
1619 {
1620   return replace(nStart, nLen, wxString(sz, nCount));
1621 }
1622
1623 #endif  //std::string compatibility
1624
1625 // ============================================================================
1626 // ArrayString
1627 // ============================================================================
1628
1629 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1630 #define   ARRAY_MAXSIZE_INCREMENT       4096
1631 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1632   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1633 #endif
1634
1635 #define   STRING(p)   ((wxString *)(&(p)))
1636
1637 // ctor
1638 wxArrayString::wxArrayString()
1639 {
1640   m_nSize  =
1641   m_nCount = 0;
1642   m_pItems = (wxChar **) NULL;
1643 }
1644
1645 // copy ctor
1646 wxArrayString::wxArrayString(const wxArrayString& src)
1647 {
1648   m_nSize  =
1649   m_nCount = 0;
1650   m_pItems = (wxChar **) NULL;
1651
1652   *this = src;
1653 }
1654
1655 // assignment operator
1656 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1657 {
1658   if ( m_nSize > 0 )
1659     Clear();
1660
1661   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1662     Alloc(src.m_nCount);
1663
1664   // we can't just copy the pointers here because otherwise we would share
1665   // the strings with another array
1666   for ( size_t n = 0; n < src.m_nCount; n++ )
1667     Add(src[n]);
1668
1669   if ( m_nCount != 0 )
1670     memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(wxChar *));
1671
1672   return *this;
1673 }
1674
1675 // grow the array
1676 void wxArrayString::Grow()
1677 {
1678   // only do it if no more place
1679   if( m_nCount == m_nSize ) {
1680     if( m_nSize == 0 ) {
1681       // was empty, alloc some memory
1682       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1683       m_pItems = new wxChar *[m_nSize];
1684     }
1685     else {
1686       // otherwise when it's called for the first time, nIncrement would be 0
1687       // and the array would never be expanded
1688       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1689
1690       // add 50% but not too much
1691       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1692                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1693       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1694         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1695       m_nSize += nIncrement;
1696       wxChar **pNew = new wxChar *[m_nSize];
1697
1698       // copy data to new location
1699       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1700
1701       // delete old memory (but do not release the strings!)
1702       wxDELETEA(m_pItems);
1703
1704       m_pItems = pNew;
1705     }
1706   }
1707 }
1708
1709 void wxArrayString::Free()
1710 {
1711   for ( size_t n = 0; n < m_nCount; n++ ) {
1712     STRING(m_pItems[n])->GetStringData()->Unlock();
1713   }
1714 }
1715
1716 // deletes all the strings from the list
1717 void wxArrayString::Empty()
1718 {
1719   Free();
1720
1721   m_nCount = 0;
1722 }
1723
1724 // as Empty, but also frees memory
1725 void wxArrayString::Clear()
1726 {
1727   Free();
1728
1729   m_nSize  =
1730   m_nCount = 0;
1731
1732   wxDELETEA(m_pItems);
1733 }
1734
1735 // dtor
1736 wxArrayString::~wxArrayString()
1737 {
1738   Free();
1739
1740   wxDELETEA(m_pItems);
1741 }
1742
1743 // pre-allocates memory (frees the previous data!)
1744 void wxArrayString::Alloc(size_t nSize)
1745 {
1746   wxASSERT( nSize > 0 );
1747
1748   // only if old buffer was not big enough
1749   if ( nSize > m_nSize ) {
1750     Free();
1751     wxDELETEA(m_pItems);
1752     m_pItems = new wxChar *[nSize];
1753     m_nSize  = nSize;
1754   }
1755
1756   m_nCount = 0;
1757 }
1758
1759 // minimizes the memory usage by freeing unused memory
1760 void wxArrayString::Shrink()
1761 {
1762   // only do it if we have some memory to free
1763   if( m_nCount < m_nSize ) {
1764     // allocates exactly as much memory as we need
1765     wxChar **pNew = new wxChar *[m_nCount];
1766
1767     // copy data to new location
1768     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1769     delete [] m_pItems;
1770     m_pItems = pNew;
1771   }
1772 }
1773
1774 // searches the array for an item (forward or backwards)
1775 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1776 {
1777   if ( bFromEnd ) {
1778     if ( m_nCount > 0 ) {
1779       size_t ui = m_nCount;
1780       do {
1781         if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1782           return ui;
1783       }
1784       while ( ui != 0 );
1785     }
1786   }
1787   else {
1788     for( size_t ui = 0; ui < m_nCount; ui++ ) {
1789       if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1790         return ui;
1791     }
1792   }
1793
1794   return wxNOT_FOUND;
1795 }
1796
1797 // add item at the end
1798 void wxArrayString::Add(const wxString& str)
1799 {
1800   wxASSERT( str.GetStringData()->IsValid() );
1801
1802   Grow();
1803
1804   // the string data must not be deleted!
1805   str.GetStringData()->Lock();
1806   m_pItems[m_nCount++] = (wxChar *)str.c_str();
1807 }
1808
1809 // add item at the given position
1810 void wxArrayString::Insert(const wxString& str, size_t nIndex)
1811 {
1812   wxASSERT( str.GetStringData()->IsValid() );
1813
1814   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Insert") );
1815
1816   Grow();
1817
1818   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
1819           (m_nCount - nIndex)*sizeof(wxChar *));
1820
1821   str.GetStringData()->Lock();
1822   m_pItems[nIndex] = (wxChar *)str.c_str();
1823
1824   m_nCount++;
1825 }
1826
1827 // removes item from array (by index)
1828 void wxArrayString::Remove(size_t nIndex)
1829 {
1830   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Remove") );
1831
1832   // release our lock
1833   Item(nIndex).GetStringData()->Unlock();
1834
1835   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
1836           (m_nCount - nIndex - 1)*sizeof(wxChar *));
1837   m_nCount--;
1838 }
1839
1840 // removes item from array (by value)
1841 void wxArrayString::Remove(const wxChar *sz)
1842 {
1843   int iIndex = Index(sz);
1844
1845   wxCHECK_RET( iIndex != wxNOT_FOUND,
1846                _("removing inexistent element in wxArrayString::Remove") );
1847
1848   Remove(iIndex);
1849 }
1850
1851 // ----------------------------------------------------------------------------
1852 // sorting
1853 // ----------------------------------------------------------------------------
1854
1855 // we can only sort one array at a time with the quick-sort based
1856 // implementation
1857 #if wxUSE_THREADS
1858   // need a critical section to protect access to gs_compareFunction and
1859   // gs_sortAscending variables
1860   static wxCriticalSection *gs_critsectStringSort = NULL;
1861
1862   // call this before the value of the global sort vars is changed/after
1863   // you're finished with them
1864   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
1865                            gs_critsectStringSort = new wxCriticalSection;     \
1866                            gs_critsectStringSort->Enter()
1867   #define END_SORT()       gs_critsectStringSort->Leave();                    \
1868                            delete gs_critsectStringSort;                      \
1869                            gs_critsectStringSort = NULL
1870 #else // !threads
1871   #define START_SORT()
1872   #define END_SORT()
1873 #endif // wxUSE_THREADS
1874
1875 // function to use for string comparaison
1876 static wxArrayString::CompareFunction gs_compareFunction = NULL;
1877
1878 // if we don't use the compare function, this flag tells us if we sort the
1879 // array in ascending or descending order
1880 static bool gs_sortAscending = TRUE;
1881
1882 // function which is called by quick sort
1883 static int wxStringCompareFunction(const void *first, const void *second)
1884 {
1885   wxString *strFirst = (wxString *)first;
1886   wxString *strSecond = (wxString *)second;
1887
1888   if ( gs_compareFunction ) {
1889     return gs_compareFunction(*strFirst, *strSecond);
1890   }
1891   else {
1892     // maybe we should use wxStrcoll
1893     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
1894
1895     return gs_sortAscending ? result : -result;
1896   }
1897 }
1898
1899 // sort array elements using passed comparaison function
1900 void wxArrayString::Sort(CompareFunction compareFunction)
1901 {
1902   START_SORT();
1903
1904   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1905   gs_compareFunction = compareFunction;
1906
1907   DoSort();
1908
1909   END_SORT();
1910 }
1911
1912 void wxArrayString::Sort(bool reverseOrder)
1913 {
1914   START_SORT();
1915
1916   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1917   gs_sortAscending = !reverseOrder;
1918
1919   DoSort();
1920
1921   END_SORT();
1922 }
1923
1924 void wxArrayString::DoSort()
1925 {
1926   // just sort the pointers using qsort() - of course it only works because
1927   // wxString() *is* a pointer to its data
1928   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
1929 }
1930
1931 // ============================================================================
1932 // MBConv
1933 // ============================================================================
1934
1935 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
1936 #if !wxUSE_WCHAR_T
1937 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
1938 #endif
1939
1940 #if wxUSE_WCHAR_T
1941
1942 // ----------------------------------------------------------------------------
1943 // standard libc conversion
1944 // ----------------------------------------------------------------------------
1945
1946 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
1947
1948 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1949 {
1950   return wxMB2WC(buf, psz, n);
1951 }
1952
1953 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1954 {
1955   return wxWC2MB(buf, psz, n);
1956 }
1957
1958 // ----------------------------------------------------------------------------
1959 // standard file conversion
1960 // ----------------------------------------------------------------------------
1961
1962 WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
1963
1964 // just use the libc conversion for now
1965 size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1966 {
1967   return wxMB2WC(buf, psz, n);
1968 }
1969
1970 size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1971 {
1972   return wxWC2MB(buf, psz, n);
1973 }
1974
1975 // ----------------------------------------------------------------------------
1976 // standard gdk conversion
1977 // ----------------------------------------------------------------------------
1978
1979 #ifdef __WXGTK12__
1980 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
1981
1982 #include <gdk/gdk.h>
1983
1984 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1985 {
1986   if (buf) {
1987     return gdk_mbstowcs((GdkWChar *)buf, psz, n);
1988   } else {
1989     GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
1990     size_t len = gdk_mbstowcs(nbuf, psz, n);
1991     delete [] nbuf;
1992     return len;
1993   }
1994 }
1995
1996 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1997 {
1998   char *mbstr = gdk_wcstombs((GdkWChar *)psz);
1999   size_t len = mbstr ? strlen(mbstr) : 0;
2000   if (buf) {
2001     if (len > n) len = n;
2002     memcpy(buf, psz, len);
2003     if (len < n) buf[len] = 0;
2004   }
2005   return len;
2006 }
2007 #endif // GTK > 1.0
2008
2009 // ----------------------------------------------------------------------------
2010 // UTF-7
2011 // ----------------------------------------------------------------------------
2012
2013 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
2014
2015 #if 0
2016 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
2017                         "abcdefghijklmnopqrstuvwxyz"
2018                         "0123456789'(),-./:?";
2019 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
2020 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
2021                         "abcdefghijklmnopqrstuvwxyz"
2022                         "0123456789+/";
2023 #endif
2024
2025 // TODO: write actual implementations of UTF-7 here
2026 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
2027                            const char * WXUNUSED(psz),
2028                            size_t WXUNUSED(n)) const
2029 {
2030   return 0;
2031 }
2032
2033 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
2034                            const wchar_t * WXUNUSED(psz),
2035                            size_t WXUNUSED(n)) const
2036 {
2037   return 0;
2038 }
2039
2040 // ----------------------------------------------------------------------------
2041 // UTF-8
2042 // ----------------------------------------------------------------------------
2043
2044 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
2045
2046 static unsigned long utf8_max[]={0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff,0xffffffff};
2047
2048 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2049 {
2050   size_t len = 0;
2051
2052   while (*psz && ((!buf) || (len<n))) {
2053     unsigned char cc=*psz++, fc=cc;
2054     unsigned cnt;
2055     for (cnt=0; fc&0x80; cnt++) fc<<=1;
2056     if (!cnt) {
2057       // plain ASCII char
2058       if (buf) *buf++=cc;
2059       len++;
2060     } else {
2061       cnt--;
2062       if (!cnt) {
2063         // invalid UTF-8 sequence
2064         return (size_t)-1;
2065       } else {
2066         unsigned ocnt=cnt-1;
2067         unsigned long res=cc&(0x3f>>cnt);
2068         while (cnt--) {
2069           cc = *psz++;
2070           if ((cc&0xC0)!=0x80) {
2071             // invalid UTF-8 sequence
2072             return (size_t)-1;
2073           }
2074           res=(res<<6)|(cc&0x3f);
2075         }
2076         if (res<=utf8_max[ocnt]) {
2077           // illegal UTF-8 encoding
2078           return (size_t)-1;
2079         }
2080         if (buf) *buf++=res;
2081         len++;
2082       }
2083     }
2084   }
2085   if (buf && (len<n)) *buf = 0;
2086   return len;
2087 }
2088
2089 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2090 {
2091   size_t len = 0;
2092
2093   while (*psz && ((!buf) || (len<n))) {
2094     unsigned long cc=(*psz++)&0x7fffffff;
2095     unsigned cnt;
2096     for (cnt=0; cc>utf8_max[cnt]; cnt++);
2097     if (!cnt) {
2098       // plain ASCII char
2099       if (buf) *buf++=cc;
2100       len++;
2101     } else {
2102       len+=cnt+1;
2103       if (buf) {
2104         *buf++=(-128>>cnt)|((cc>>(cnt*6))&(0x3f>>cnt));
2105         while (cnt--)
2106           *buf++=0x80|((cc>>(cnt*6))&0x3f);
2107       }
2108     }
2109   }
2110   if (buf && (len<n)) *buf = 0;
2111   return len;
2112 }
2113
2114 // ----------------------------------------------------------------------------
2115 // specified character set
2116 // ----------------------------------------------------------------------------
2117
2118 class wxCharacterSet
2119 {
2120 public:
2121   wxArrayString names;
2122   wchar_t *data;
2123 };
2124
2125 #ifndef WX_PRECOMP
2126   #include "wx/dynarray.h"
2127   #include "wx/filefn.h"
2128   #include "wx/textfile.h"
2129   #include "wx/tokenzr.h"
2130   #include "wx/utils.h"
2131 #endif
2132
2133 WX_DECLARE_OBJARRAY(wxCharacterSet, wxCSArray);
2134 #include "wx/arrimpl.cpp"
2135 WX_DEFINE_OBJARRAY(wxCSArray);
2136
2137 static wxCSArray wxCharsets;
2138
2139 static void wxLoadCharacterSets(void)
2140 {
2141   static bool already_loaded = FALSE;
2142
2143   if (already_loaded) return;
2144
2145   already_loaded = TRUE;
2146 #if defined(__UNIX__) && wxUSE_TEXTFILE
2147   // search through files in /usr/share/i18n/charmaps
2148   wxString fname;
2149   for (fname = ::wxFindFirstFile(_T("/usr/share/i18n/charmaps/*"));
2150        !fname.IsEmpty();
2151        fname = ::wxFindNextFile()) {
2152     wxTextFile cmap(fname);
2153     if (cmap.Open()) {
2154       wxCharacterSet *cset = new wxCharacterSet;
2155       wxString comchar,escchar;
2156       bool in_charset = FALSE;
2157
2158       // wxFprintf(stderr,_T("Loaded: %s\n"),fname.c_str());
2159
2160       wxString line;
2161       for (line = cmap.GetFirstLine();
2162            !cmap.Eof();
2163            line = cmap.GetNextLine()) {
2164         // wxFprintf(stderr,_T("line contents: %s\n"),line.c_str());
2165         wxStringTokenizer token(line);
2166         wxString cmd = token.GetNextToken();
2167         if (cmd == comchar) {
2168           if (token.GetNextToken() == _T("alias"))
2169             cset->names.Add(token.GetNextToken());
2170         }
2171         else if (cmd == _T("<code_set_name>"))
2172           cset->names.Add(token.GetNextToken());
2173         else if (cmd == _T("<comment_char>"))
2174           comchar = token.GetNextToken();
2175         else if (cmd == _T("<escape_char>"))
2176           escchar = token.GetNextToken();
2177         else if (cmd == _T("<mb_cur_min>")) {
2178           delete cset;
2179           cset = (wxCharacterSet *) NULL;
2180           break; // we don't support multibyte charsets ourselves (yet)
2181         }
2182         else if (cmd == _T("CHARMAP")) {
2183           cset->data = (wchar_t *)calloc(256, sizeof(wchar_t));
2184           in_charset = TRUE;
2185         }
2186         else if (cmd == _T("END")) {
2187           if (token.GetNextToken() == _T("CHARMAP"))
2188             in_charset = FALSE;
2189         }
2190         else if (in_charset) {
2191           // format: <NUL> /x00 <U0000> NULL (NUL)
2192           //         <A>   /x41 <U0041> LATIN CAPITAL LETTER A
2193           wxString hex = token.GetNextToken();
2194           // skip whitespace (why doesn't wxStringTokenizer do this?)
2195           while (wxIsEmpty(hex) && token.HasMoreTokens()) hex = token.GetNextToken();
2196           wxString uni = token.GetNextToken();
2197           // skip whitespace again
2198           while (wxIsEmpty(uni) && token.HasMoreTokens()) uni = token.GetNextToken();
2199           if ((hex.Len() > 2) && (wxString(hex.GetChar(0)) == escchar) && (hex.GetChar(1) == _T('x')) &&
2200               (uni.Left(2) == _T("<U"))) {
2201             hex.MakeUpper(); uni.MakeUpper();
2202             int pos = ::wxHexToDec(hex.Mid(2,2));
2203             if (pos>=0) {
2204               unsigned long uni1 = ::wxHexToDec(uni.Mid(2,2));
2205               unsigned long uni2 = ::wxHexToDec(uni.Mid(4,2));
2206               cset->data[pos] = (uni1 << 16) | uni2;
2207               // wxFprintf(stderr,_T("char %02x mapped to %04x (%c)\n"),pos,cset->data[pos],cset->data[pos]);
2208             }
2209           }
2210         }
2211       }
2212       if (cset) {
2213         cset->names.Shrink();
2214         wxCharsets.Add(cset);
2215       }
2216     }
2217   }
2218 #endif
2219   wxCharsets.Shrink();
2220 }
2221
2222 static wxCharacterSet *wxFindCharacterSet(const wxChar *charset)
2223 {
2224   if (!charset) return (wxCharacterSet *)NULL;
2225   wxLoadCharacterSets();
2226   for (size_t n=0; n<wxCharsets.GetCount(); n++)
2227     if (wxCharsets[n].names.Index(charset) != wxNOT_FOUND)
2228       return &(wxCharsets[n]);
2229   return (wxCharacterSet *)NULL;
2230 }
2231
2232 WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
2233
2234 wxCSConv::wxCSConv(const wxChar *charset)
2235 {
2236   m_name = (wxChar *) NULL;
2237   m_cset = (wxCharacterSet *) NULL;
2238   m_deferred = TRUE;
2239   SetName(charset);
2240 }
2241
2242 wxCSConv::~wxCSConv()
2243 {
2244   if (m_name) free(m_name);
2245 }
2246
2247 void wxCSConv::SetName(const wxChar *charset)
2248 {
2249   if (charset) {
2250 #ifdef __UNIX__
2251     // first, convert the character set name to standard form
2252     wxString codeset;
2253     if (wxString(charset,3).CmpNoCase(_T("ISO")) == 0) {
2254       // make sure it's represented in the standard form: ISO_8859-1
2255       codeset = _T("ISO_");
2256       charset += 3;
2257       if ((*charset == _T('-')) || (*charset == _T('_'))) charset++;
2258       if (wxStrlen(charset)>4) {
2259         if (wxString(charset,4) == _T("8859")) {
2260           codeset << _T("8859-");
2261           if (*charset == _T('-')) charset++;
2262         }
2263       }
2264     }
2265     codeset << charset;
2266     codeset.MakeUpper();
2267     m_name = wxStrdup(codeset.c_str());
2268     m_deferred = TRUE;
2269 #endif
2270   }
2271 }
2272
2273 void wxCSConv::LoadNow()
2274 {
2275 //  wxPrintf(_T("Conversion request\n"));
2276   if (m_deferred) {
2277     if (!m_name) {
2278 #ifdef __UNIX__
2279       wxChar *lang = wxGetenv(_T("LANG"));
2280       wxChar *dot = lang ? wxStrchr(lang, _T('.')) : (wxChar *)NULL;
2281       if (dot) SetName(dot+1);
2282 #endif
2283     }
2284     m_cset = wxFindCharacterSet(m_name);
2285     m_deferred = FALSE;
2286   }
2287 }
2288
2289 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2290 {
2291   ((wxCSConv *)this)->LoadNow(); // discard constness
2292   if (buf) {
2293     if (m_cset) {
2294       for (size_t c=0; c<n; c++)
2295         buf[c] = m_cset->data[(unsigned char)(psz[c])];
2296     } else {
2297       // latin-1 (direct)
2298       for (size_t c=0; c<n; c++)
2299         buf[c] = (unsigned char)(psz[c]);
2300     }
2301     return n;
2302   }
2303   return strlen(psz);
2304 }
2305
2306 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2307 {
2308   ((wxCSConv *)this)->LoadNow(); // discard constness
2309   if (buf) {
2310     if (m_cset) {
2311       for (size_t c=0; c<n; c++) {
2312         size_t n;
2313         for (n=0; (n<256) && (m_cset->data[n] != psz[c]); n++);
2314         buf[c] = (n>0xff) ? '?' : n;
2315       }
2316     } else {
2317       // latin-1 (direct)
2318       for (size_t c=0; c<n; c++)
2319         buf[c] = (psz[c]>0xff) ? '?' : psz[c];
2320     }
2321     return n;
2322   }
2323   return wcslen(psz);
2324 }
2325
2326 #endif//wxUSE_WCHAR_T
2327
2328 #if wxUSE_WCHAR_T
2329 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
2330     {
2331       if (psz) {
2332         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
2333         wxWCharBuffer buf(nLen);
2334         MB2WC(WCSTRINGCAST buf, psz, nLen);
2335         return buf;
2336       } else return wxWCharBuffer((wchar_t *) NULL);
2337     }
2338
2339 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
2340     {
2341       if (psz) {
2342         size_t nLen = WC2MB((char *) NULL, psz, 0);
2343         wxCharBuffer buf(nLen);
2344         WC2MB(MBSTRINGCAST buf, psz, nLen);
2345         return buf;
2346       } else return wxCharBuffer((char *) NULL);
2347     }
2348
2349 #endif//wxUSE_WCHAR_T
2350