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