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