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 licence
  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   #include "wx/thread.h"
  39 #endif
  40
  41 #include <ctype.h>
  42 #include <string.h>
  43 #include <stdlib.h>
  44
  45 #ifdef __SALFORDC__
  46   #include <clib.h>
  47 #endif
  48
  49 // allocating extra space for each string consumes more memory but speeds up
  50 // the concatenation operations (nLen is the current string's length)
  51 // NB: EXTRA_ALLOC must be >= 0!
  52 #define EXTRA_ALLOC       (19 - nLen % 16)
  53
  54 // ---------------------------------------------------------------------------
  55 // static class variables definition
  56 // ---------------------------------------------------------------------------
  57
  58 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  59 // must define this static for VA or else you get multiply defined symbols
  60 // everywhere
  61 const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100;
  62 #endif // Visual Age
  63
  64 #ifdef  wxSTD_STRING_COMPATIBILITY
  65   const size_t wxString::npos = wxSTRING_MAXLEN;
  66 #endif // wxSTD_STRING_COMPATIBILITY
  67
  68 // ----------------------------------------------------------------------------
  69 // static data
  70 // ----------------------------------------------------------------------------
  71
  72 // for an empty string, GetStringData() will return this address: this
  73 // structure has the same layout as wxStringData and it's data() method will
  74 // return the empty string (dummy pointer)
  75 static const struct
  76 {
  77   wxStringData data;
  78   wxChar dummy;
  79 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  80
  81 // empty C style string: points to 'string data' byte of g_strEmpty
  82 extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy;
  83
  84 // ----------------------------------------------------------------------------
  85 // global functions
  86 // ----------------------------------------------------------------------------
  87
  88 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
  89
  90 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
  91 // iostream ones.
  92 //
  93 // ATTN: you can _not_ use both of these in the same program!
  94
  95 wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str))
  96 {
  97 #if 0
  98   int w = is.width(0);
  99   if ( is.ipfx(0) ) {
 100     streambuf *sb = is.rdbuf();
 101     str.erase();
 102     while ( true ) {
 103       int ch = sb->sbumpc ();
 104       if ( ch == EOF ) {
 105         is.setstate(ios::eofbit);
 106         break;
 107       }
 108       else if ( isspace(ch) ) {
 109         sb->sungetc();
 110         break;
 111       }
 112
 113       str += ch;
 114       if ( --w == 1 )
 115         break;
 116     }
 117   }
 118
 119   is.isfx();
 120   if ( str.length() == 0 )
 121     is.setstate(ios::failbit);
 122 #endif
 123   return is;
 124 }
 125
 126 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
 127 {
 128   os << str.c_str();
 129   return os;
 130 }
 131
 132 #endif  //std::string compatibility
 133
 134 // ----------------------------------------------------------------------------
 135 // private classes
 136 // ----------------------------------------------------------------------------
 137
 138 // this small class is used to gather statistics for performance tuning
 139 //#define WXSTRING_STATISTICS
 140 #ifdef  WXSTRING_STATISTICS
 141   class Averager
 142   {
 143   public:
 144     Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 145    ~Averager()
 146    { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 147
 148     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 149
 150   private:
 151     size_t m_nCount, m_nTotal;
 152     const wxChar *m_sz;
 153   } g_averageLength("allocation size"),
 154     g_averageSummandLength("summand length"),
 155     g_averageConcatHit("hit probability in concat"),
 156     g_averageInitialLength("initial string length");
 157
 158   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 159 #else
 160   #define STATISTICS_ADD(av, val)
 161 #endif // WXSTRING_STATISTICS
 162
 163 // ===========================================================================
 164 // wxStringData class deallocation
 165 // ===========================================================================
 166
 167 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 168 #  pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
 169 void wxStringData::Free()
 170 {
 171     free(this);
 172 }
 173 #endif
 174
 175 // ===========================================================================
 176 // wxString class core
 177 // ===========================================================================
 178
 179 // ---------------------------------------------------------------------------
 180 // construction
 181 // ---------------------------------------------------------------------------
 182
 183 // constructs string of <nLength> copies of character <ch>
 184 wxString::wxString(wxChar ch, size_t nLength)
 185 {
 186   Init();
 187
 188   if ( nLength > 0 ) {
 189     if ( !AllocBuffer(nLength) ) {
 190       wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 191       return;
 192     }
 193
 194 #if wxUSE_UNICODE
 195     // memset only works on chars
 196     for ( size_t n = 0; n < nLength; n++ )
 197         m_pchData[n] = ch;
 198 #else
 199     memset(m_pchData, ch, nLength);
 200 #endif
 201   }
 202 }
 203
 204 // takes nLength elements of psz starting at nPos
 205 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 206 {
 207   Init();
 208
 209   // if the length is not given, assume the string to be NUL terminated
 210   if ( nLength == wxSTRING_MAXLEN ) {
 211     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 212
 213     nLength = wxStrlen(psz + nPos);
 214   }
 215
 216   STATISTICS_ADD(InitialLength, nLength);
 217
 218   if ( nLength > 0 ) {
 219     // trailing '\0' is written in AllocBuffer()
 220     if ( !AllocBuffer(nLength) ) {
 221       wxFAIL_MSG( _T("out of memory in wxString::InitWith") );
 222       return;
 223     }
 224     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 225   }
 226 }
 227
 228 #ifdef  wxSTD_STRING_COMPATIBILITY
 229
 230 // poor man's iterators are "void *" pointers
 231 wxString::wxString(const void *pStart, const void *pEnd)
 232 {
 233   InitWith((const wxChar *)pStart, 0,
 234            (const wxChar *)pEnd - (const wxChar *)pStart);
 235 }
 236
 237 #endif  //std::string compatibility
 238
 239 #if wxUSE_UNICODE
 240
 241 // from multibyte string
 242 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 243 {
 244     // first get the size of the buffer we need
 245     size_t nLen;
 246     if ( psz )
 247     {
 248         // calculate the needed size ourselves or use a provide one
 249         nLen = nLength == wxSTRING_MAXLEN ? conv.MB2WC(NULL, psz, 0) : nLength;
 250     }
 251     else
 252     {
 253         // nothing to convert
 254         nLen = 0;
 255     }
 256
 257     // anything to do?
 258     if ( (nLen != 0) && (nLen != (size_t)-1) )
 259     {
 260         if ( !AllocBuffer(nLen) )
 261         {
 262             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 263             return;
 264         }
 265
 266         // MB2WC wants the buffer size, not the string length
 267         if ( conv.MB2WC(m_pchData, psz, nLen + 1) != (size_t)-1 )
 268         {
 269             // initialized ok
 270             return;
 271         }
 272         //else: the conversion failed -- leave the string empty (what else?)
 273     }
 274
 275     Init();
 276 }
 277
 278 #else // ANSI
 279
 280 #if wxUSE_WCHAR_T
 281 // from wide string
 282 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 283 {
 284     // first get the size of the buffer we need
 285     size_t nLen;
 286     if ( pwz )
 287     {
 288         // calculate the needed size ourselves or use a provide one
 289         nLen = nLength == wxSTRING_MAXLEN ? conv.WC2MB(NULL, pwz, 0) : nLength;
 290     }
 291     else
 292     {
 293         // nothing to convert
 294         nLen = 0;
 295     }
 296
 297     // anything to do?
 298     if ( (nLen != 0) && (nLen != (size_t)-1) )
 299     {
 300         if ( !AllocBuffer(nLen) )
 301         {
 302             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 303             return;
 304         }
 305
 306         // WC2MB wants the buffer size, not the string length
 307         if ( conv.WC2MB(m_pchData, pwz, nLen + 1) != (size_t)-1 )
 308         {
 309             // initialized ok
 310             return;
 311         }
 312         //else: the conversion failed -- leave the string empty (what else?)
 313     }
 314
 315     Init();
 316 }
 317 #endif // wxUSE_WCHAR_T
 318
 319 #endif // Unicode/ANSI
 320
 321 // ---------------------------------------------------------------------------
 322 // memory allocation
 323 // ---------------------------------------------------------------------------
 324
 325 // allocates memory needed to store a C string of length nLen
 326 bool wxString::AllocBuffer(size_t nLen)
 327 {
 328   // allocating 0 sized buffer doesn't make sense, all empty strings should
 329   // reuse g_strEmpty
 330   wxASSERT( nLen >  0 );
 331
 332   // make sure that we don't overflow
 333   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 334                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 335
 336   STATISTICS_ADD(Length, nLen);
 337
 338   // allocate memory:
 339   // 1) one extra character for '\0' termination
 340   // 2) sizeof(wxStringData) for housekeeping info
 341   wxStringData* pData = (wxStringData*)
 342     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 343
 344   if ( pData == NULL ) {
 345     // allocation failures are handled by the caller
 346     return FALSE;
 347   }
 348
 349   pData->nRefs        = 1;
 350   pData->nDataLength  = nLen;
 351   pData->nAllocLength = nLen + EXTRA_ALLOC;
 352   m_pchData           = pData->data();  // data starts after wxStringData
 353   m_pchData[nLen]     = wxT('\0');
 354   return TRUE;
 355 }
 356
 357 // must be called before changing this string
 358 bool wxString::CopyBeforeWrite()
 359 {
 360   wxStringData* pData = GetStringData();
 361
 362   if ( pData->IsShared() ) {
 363     pData->Unlock();                // memory not freed because shared
 364     size_t nLen = pData->nDataLength;
 365     if ( !AllocBuffer(nLen) ) {
 366       // allocation failures are handled by the caller
 367       return FALSE;
 368     }
 369     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 370   }
 371
 372   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 373
 374   return TRUE;
 375 }
 376
 377 // must be called before replacing contents of this string
 378 bool wxString::AllocBeforeWrite(size_t nLen)
 379 {
 380   wxASSERT( nLen != 0 );  // doesn't make any sense
 381
 382   // must not share string and must have enough space
 383   wxStringData* pData = GetStringData();
 384   if ( pData->IsShared() || pData->IsEmpty() ) {
 385     // can't work with old buffer, get new one
 386     pData->Unlock();
 387     if ( !AllocBuffer(nLen) ) {
 388       // allocation failures are handled by the caller
 389       return FALSE;
 390     }
 391   }
 392   else {
 393     if ( nLen > pData->nAllocLength ) {
 394       // realloc the buffer instead of calling malloc() again, this is more
 395       // efficient
 396       STATISTICS_ADD(Length, nLen);
 397
 398       nLen += EXTRA_ALLOC;
 399
 400       pData = (wxStringData*)
 401           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 402
 403       if ( pData == NULL ) {
 404         // allocation failures are handled by the caller
 405         // keep previous data since reallocation failed
 406         return FALSE;
 407       }
 408
 409       pData->nAllocLength = nLen;
 410       m_pchData = pData->data();
 411     }
 412
 413     // now we have enough space, just update the string length
 414     pData->nDataLength = nLen;
 415   }
 416
 417   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 418
 419   return TRUE;
 420 }
 421
 422 // allocate enough memory for nLen characters
 423 bool wxString::Alloc(size_t nLen)
 424 {
 425   wxStringData *pData = GetStringData();
 426   if ( pData->nAllocLength <= nLen ) {
 427     if ( pData->IsEmpty() ) {
 428       nLen += EXTRA_ALLOC;
 429
 430       wxStringData* pData = (wxStringData*)
 431           malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 432
 433       if ( pData == NULL ) {
 434         // allocation failure handled by caller
 435         return FALSE;
 436       }
 437
 438       pData->nRefs = 1;
 439       pData->nDataLength = 0;
 440       pData->nAllocLength = nLen;
 441       m_pchData = pData->data();  // data starts after wxStringData
 442       m_pchData[0u] = wxT('\0');
 443     }
 444     else if ( pData->IsShared() ) {
 445       pData->Unlock();                // memory not freed because shared
 446       size_t nOldLen = pData->nDataLength;
 447       if ( !AllocBuffer(nLen) ) {
 448         // allocation failure handled by caller
 449         return FALSE;
 450       }
 451       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 452     }
 453     else {
 454       nLen += EXTRA_ALLOC;
 455
 456       pData = (wxStringData *)
 457         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 458
 459       if ( pData == NULL ) {
 460         // allocation failure handled by caller
 461         // keep previous data since reallocation failed
 462         return FALSE;
 463       }
 464
 465       // it's not important if the pointer changed or not (the check for this
 466       // is not faster than assigning to m_pchData in all cases)
 467       pData->nAllocLength = nLen;
 468       m_pchData = pData->data();
 469     }
 470   }
 471   //else: we've already got enough
 472   return TRUE;
 473 }
 474
 475 // shrink to minimal size (releasing extra memory)
 476 bool wxString::Shrink()
 477 {
 478   wxStringData *pData = GetStringData();
 479
 480   size_t nLen = pData->nDataLength;
 481   void *p = realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 482
 483   if ( p == NULL) {
 484       wxFAIL_MSG( _T("out of memory reallocating wxString data") );
 485       // keep previous data since reallocation failed
 486       return FALSE;
 487   }
 488
 489   if ( p != pData )
 490   {
 491       // contrary to what one might believe, some realloc() implementation do
 492       // move the memory block even when its size is reduced
 493       pData = (wxStringData *)p;
 494
 495       m_pchData = pData->data();
 496   }
 497
 498   pData->nAllocLength = nLen;
 499
 500   return TRUE;
 501 }
 502
 503 // get the pointer to writable buffer of (at least) nLen bytes
 504 wxChar *wxString::GetWriteBuf(size_t nLen)
 505 {
 506   if ( !AllocBeforeWrite(nLen) ) {
 507     // allocation failure handled by caller
 508     return NULL;
 509   }
 510
 511   wxASSERT( GetStringData()->nRefs == 1 );
 512   GetStringData()->Validate(FALSE);
 513
 514   return m_pchData;
 515 }
 516
 517 // put string back in a reasonable state after GetWriteBuf
 518 void wxString::UngetWriteBuf()
 519 {
 520   GetStringData()->nDataLength = wxStrlen(m_pchData);
 521   GetStringData()->Validate(TRUE);
 522 }
 523
 524 void wxString::UngetWriteBuf(size_t nLen)
 525 {
 526   GetStringData()->nDataLength = nLen;
 527   GetStringData()->Validate(TRUE);
 528 }
 529
 530 // ---------------------------------------------------------------------------
 531 // data access
 532 // ---------------------------------------------------------------------------
 533
 534 // all functions are inline in string.h
 535
 536 // ---------------------------------------------------------------------------
 537 // assignment operators
 538 // ---------------------------------------------------------------------------
 539
 540 // helper function: does real copy
 541 bool wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 542 {
 543   if ( nSrcLen == 0 ) {
 544     Reinit();
 545   }
 546   else {
 547     if ( !AllocBeforeWrite(nSrcLen) ) {
 548       // allocation failure handled by caller
 549       return FALSE;
 550     }
 551     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 552     GetStringData()->nDataLength = nSrcLen;
 553     m_pchData[nSrcLen] = wxT('\0');
 554   }
 555   return TRUE;
 556 }
 557
 558 // assigns one string to another
 559 wxString& wxString::operator=(const wxString& stringSrc)
 560 {
 561   wxASSERT( stringSrc.GetStringData()->IsValid() );
 562
 563   // don't copy string over itself
 564   if ( m_pchData != stringSrc.m_pchData ) {
 565     if ( stringSrc.GetStringData()->IsEmpty() ) {
 566       Reinit();
 567     }
 568     else {
 569       // adjust references
 570       GetStringData()->Unlock();
 571       m_pchData = stringSrc.m_pchData;
 572       GetStringData()->Lock();
 573     }
 574   }
 575
 576   return *this;
 577 }
 578
 579 // assigns a single character
 580 wxString& wxString::operator=(wxChar ch)
 581 {
 582   if ( !AssignCopy(1, &ch) ) {
 583     wxFAIL_MSG( _T("out of memory in wxString::operator=(wxChar)") );
 584   }
 585   return *this;
 586 }
 587
 588
 589 // assigns C string
 590 wxString& wxString::operator=(const wxChar *psz)
 591 {
 592   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 593     wxFAIL_MSG( _T("out of memory in wxString::operator=(const wxChar *)") );
 594   }
 595   return *this;
 596 }
 597
 598 #if !wxUSE_UNICODE
 599
 600 // same as 'signed char' variant
 601 wxString& wxString::operator=(const unsigned char* psz)
 602 {
 603   *this = (const char *)psz;
 604   return *this;
 605 }
 606
 607 #if wxUSE_WCHAR_T
 608 wxString& wxString::operator=(const wchar_t *pwz)
 609 {
 610   wxString str(pwz);
 611   *this = str;
 612   return *this;
 613 }
 614 #endif
 615
 616 #endif
 617
 618 // ---------------------------------------------------------------------------
 619 // string concatenation
 620 // ---------------------------------------------------------------------------
 621
 622 // add something to this string
 623 bool wxString::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData)
 624 {
 625   STATISTICS_ADD(SummandLength, nSrcLen);
 626
 627   // concatenating an empty string is a NOP
 628   if ( nSrcLen > 0 ) {
 629     wxStringData *pData = GetStringData();
 630     size_t nLen = pData->nDataLength;
 631     size_t nNewLen = nLen + nSrcLen;
 632
 633     // alloc new buffer if current is too small
 634     if ( pData->IsShared() ) {
 635       STATISTICS_ADD(ConcatHit, 0);
 636
 637       // we have to allocate another buffer
 638       wxStringData* pOldData = GetStringData();
 639       if ( !AllocBuffer(nNewLen) ) {
 640           // allocation failure handled by caller
 641           return FALSE;
 642       }
 643       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 644       pOldData->Unlock();
 645     }
 646     else if ( nNewLen > pData->nAllocLength ) {
 647       STATISTICS_ADD(ConcatHit, 0);
 648
 649       // we have to grow the buffer
 650       if ( !Alloc(nNewLen) ) {
 651           // allocation failure handled by caller
 652           return FALSE;
 653       }
 654     }
 655     else {
 656       STATISTICS_ADD(ConcatHit, 1);
 657
 658       // the buffer is already big enough
 659     }
 660
 661     // should be enough space
 662     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 663
 664     // fast concatenation - all is done in our buffer
 665     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 666
 667     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 668     GetStringData()->nDataLength = nNewLen; // and fix the length
 669   }
 670   //else: the string to append was empty
 671   return TRUE;
 672 }
 673
 674 /*
 675  * concatenation functions come in 5 flavours:
 676  *  string + string
 677  *  char   + string      and      string + char
 678  *  C str  + string      and      string + C str
 679  */
 680
 681 wxString operator+(const wxString& str1, const wxString& str2)
 682 {
 683   wxASSERT( str1.GetStringData()->IsValid() );
 684   wxASSERT( str2.GetStringData()->IsValid() );
 685
 686   wxString s = str1;
 687   s += str2;
 688
 689   return s;
 690 }
 691
 692 wxString operator+(const wxString& str, wxChar ch)
 693 {
 694   wxASSERT( str.GetStringData()->IsValid() );
 695
 696   wxString s = str;
 697   s += ch;
 698
 699   return s;
 700 }
 701
 702 wxString operator+(wxChar ch, const wxString& str)
 703 {
 704   wxASSERT( str.GetStringData()->IsValid() );
 705
 706   wxString s = ch;
 707   s += str;
 708
 709   return s;
 710 }
 711
 712 wxString operator+(const wxString& str, const wxChar *psz)
 713 {
 714   wxASSERT( str.GetStringData()->IsValid() );
 715
 716   wxString s;
 717   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 718     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 719   }
 720   s = str;
 721   s += psz;
 722
 723   return s;
 724 }
 725
 726 wxString operator+(const wxChar *psz, const wxString& str)
 727 {
 728   wxASSERT( str.GetStringData()->IsValid() );
 729
 730   wxString s;
 731   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 732     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 733   }
 734   s = psz;
 735   s += str;
 736
 737   return s;
 738 }
 739
 740 // ===========================================================================
 741 // other common string functions
 742 // ===========================================================================
 743
 744 #if wxUSE_UNICODE
 745
 746 wxString wxString::FromAscii(const char *ascii)
 747 {
 748     if (!ascii)
 749        return wxEmptyString;
 750
 751     size_t len = strlen( ascii );
 752     wxString res;
 753
 754     if ( len )
 755     {
 756         wxStringBuffer buf(res, len);
 757
 758         wchar_t *dest = buf;
 759
 760         for ( ;; )
 761         {
 762            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
 763                break;
 764         }
 765     }
 766
 767     return res;
 768 }
 769
 770 wxString wxString::FromAscii(const char ascii)
 771 {
 772     // What do we do with '\0' ?
 773
 774     wxString res;
 775     res += (wchar_t)(unsigned char) ascii;
 776
 777     return res;
 778 }
 779
 780 const wxCharBuffer wxString::ToAscii() const
 781 {
 782     // this will allocate enough space for the terminating NUL too
 783     wxCharBuffer buffer(length());
 784
 785     signed char *dest = (signed char *)buffer.data();
 786
 787     const wchar_t *pwc = c_str();
 788     for ( ;; )
 789     {
 790         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
 791
 792         // the output string can't have embedded NULs anyhow, so we can safely
 793         // stop at first of them even if we do have any
 794         if ( !*pwc++ )
 795             break;
 796     }
 797
 798     return buffer;
 799 }
 800
 801 #endif // Unicode
 802
 803 // ---------------------------------------------------------------------------
 804 // simple sub-string extraction
 805 // ---------------------------------------------------------------------------
 806
 807 // helper function: clone the data attached to this string
 808 bool wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 809 {
 810   if ( nCopyLen == 0 ) {
 811     dest.Init();
 812   }
 813   else {
 814     if ( !dest.AllocBuffer(nCopyLen) ) {
 815       // allocation failure handled by caller
 816       return FALSE;
 817     }
 818     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 819   }
 820   return TRUE;
 821 }
 822
 823 // extract string of length nCount starting at nFirst
 824 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 825 {
 826   wxStringData *pData = GetStringData();
 827   size_t nLen = pData->nDataLength;
 828
 829   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 830   if ( nCount == wxSTRING_MAXLEN )
 831   {
 832     nCount = nLen - nFirst;
 833   }
 834
 835   // out-of-bounds requests return sensible things
 836   if ( nFirst + nCount > nLen )
 837   {
 838     nCount = nLen - nFirst;
 839   }
 840
 841   if ( nFirst > nLen )
 842   {
 843     // AllocCopy() will return empty string
 844     nCount = 0;
 845   }
 846
 847   wxString dest;
 848   if ( !AllocCopy(dest, nCount, nFirst) ) {
 849       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
 850   }
 851
 852   return dest;
 853 }
 854
 855 // check that the tring starts with prefix and return the rest of the string
 856 // in the provided pointer if it is not NULL, otherwise return FALSE
 857 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 858 {
 859     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 860
 861     // first check if the beginning of the string matches the prefix: note
 862     // that we don't have to check that we don't run out of this string as
 863     // when we reach the terminating NUL, either prefix string ends too (and
 864     // then it's ok) or we break out of the loop because there is no match
 865     const wxChar *p = c_str();
 866     while ( *prefix )
 867     {
 868         if ( *prefix++ != *p++ )
 869         {
 870             // no match
 871             return FALSE;
 872         }
 873     }
 874
 875     if ( rest )
 876     {
 877         // put the rest of the string into provided pointer
 878         *rest = p;
 879     }
 880
 881     return TRUE;
 882 }
 883
 884 // extract nCount last (rightmost) characters
 885 wxString wxString::Right(size_t nCount) const
 886 {
 887   if ( nCount > (size_t)GetStringData()->nDataLength )
 888     nCount = GetStringData()->nDataLength;
 889
 890   wxString dest;
 891   if ( !AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount) ) {
 892     wxFAIL_MSG( _T("out of memory in wxString::Right") );
 893   }
 894   return dest;
 895 }
 896
 897 // get all characters after the last occurence of ch
 898 // (returns the whole string if ch not found)
 899 wxString wxString::AfterLast(wxChar ch) const
 900 {
 901   wxString str;
 902   int iPos = Find(ch, TRUE);
 903   if ( iPos == wxNOT_FOUND )
 904     str = *this;
 905   else
 906     str = c_str() + iPos + 1;
 907
 908   return str;
 909 }
 910
 911 // extract nCount first (leftmost) characters
 912 wxString wxString::Left(size_t nCount) const
 913 {
 914   if ( nCount > (size_t)GetStringData()->nDataLength )
 915     nCount = GetStringData()->nDataLength;
 916
 917   wxString dest;
 918   if ( !AllocCopy(dest, nCount, 0) ) {
 919     wxFAIL_MSG( _T("out of memory in wxString::Left") );
 920   }
 921   return dest;
 922 }
 923
 924 // get all characters before the first occurence of ch
 925 // (returns the whole string if ch not found)
 926 wxString wxString::BeforeFirst(wxChar ch) const
 927 {
 928   wxString str;
 929   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 930     str += *pc;
 931
 932   return str;
 933 }
 934
 935 /// get all characters before the last occurence of ch
 936 /// (returns empty string if ch not found)
 937 wxString wxString::BeforeLast(wxChar ch) const
 938 {
 939   wxString str;
 940   int iPos = Find(ch, TRUE);
 941   if ( iPos != wxNOT_FOUND && iPos != 0 )
 942     str = wxString(c_str(), iPos);
 943
 944   return str;
 945 }
 946
 947 /// get all characters after the first occurence of ch
 948 /// (returns empty string if ch not found)
 949 wxString wxString::AfterFirst(wxChar ch) const
 950 {
 951   wxString str;
 952   int iPos = Find(ch);
 953   if ( iPos != wxNOT_FOUND )
 954     str = c_str() + iPos + 1;
 955
 956   return str;
 957 }
 958
 959 // replace first (or all) occurences of some substring with another one
 960 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 961 {
 962   size_t uiCount = 0;   // count of replacements made
 963
 964   size_t uiOldLen = wxStrlen(szOld);
 965
 966   wxString strTemp;
 967   const wxChar *pCurrent = m_pchData;
 968   const wxChar *pSubstr;
 969   while ( *pCurrent != wxT('\0') ) {
 970     pSubstr = wxStrstr(pCurrent, szOld);
 971     if ( pSubstr == NULL ) {
 972       // strTemp is unused if no replacements were made, so avoid the copy
 973       if ( uiCount == 0 )
 974         return 0;
 975
 976       strTemp += pCurrent;    // copy the rest
 977       break;                  // exit the loop
 978     }
 979     else {
 980       // take chars before match
 981       if ( !strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent) ) {
 982         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
 983         return 0;
 984       }
 985       strTemp += szNew;
 986       pCurrent = pSubstr + uiOldLen;  // restart after match
 987
 988       uiCount++;
 989
 990       // stop now?
 991       if ( !bReplaceAll ) {
 992         strTemp += pCurrent;    // copy the rest
 993         break;                  // exit the loop
 994       }
 995     }
 996   }
 997
 998   // only done if there were replacements, otherwise would have returned above
 999   *this = strTemp;
1000
1001   return uiCount;
1002 }
1003
1004 bool wxString::IsAscii() const
1005 {
1006   const wxChar *s = (const wxChar*) *this;
1007   while(*s){
1008     if(!isascii(*s)) return(FALSE);
1009     s++;
1010   }
1011   return(TRUE);
1012 }
1013
1014 bool wxString::IsWord() const
1015 {
1016   const wxChar *s = (const wxChar*) *this;
1017   while(*s){
1018     if(!wxIsalpha(*s)) return(FALSE);
1019     s++;
1020   }
1021   return(TRUE);
1022 }
1023
1024 bool wxString::IsNumber() const
1025 {
1026   const wxChar *s = (const wxChar*) *this;
1027   if (wxStrlen(s))
1028      if ((s[0] == '-') || (s[0] == '+')) s++;
1029   while(*s){
1030     if(!wxIsdigit(*s)) return(FALSE);
1031     s++;
1032   }
1033   return(TRUE);
1034 }
1035
1036 wxString wxString::Strip(stripType w) const
1037 {
1038     wxString s = *this;
1039     if ( w & leading ) s.Trim(FALSE);
1040     if ( w & trailing ) s.Trim(TRUE);
1041     return s;
1042 }
1043
1044 // ---------------------------------------------------------------------------
1045 // case conversion
1046 // ---------------------------------------------------------------------------
1047
1048 wxString& wxString::MakeUpper()
1049 {
1050   if ( !CopyBeforeWrite() ) {
1051     wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") );
1052     return *this;
1053   }
1054
1055   for ( wxChar *p = m_pchData; *p; p++ )
1056     *p = (wxChar)wxToupper(*p);
1057
1058   return *this;
1059 }
1060
1061 wxString& wxString::MakeLower()
1062 {
1063   if ( !CopyBeforeWrite() ) {
1064     wxFAIL_MSG( _T("out of memory in wxString::MakeLower") );
1065     return *this;
1066   }
1067
1068   for ( wxChar *p = m_pchData; *p; p++ )
1069     *p = (wxChar)wxTolower(*p);
1070
1071   return *this;
1072 }
1073
1074 // ---------------------------------------------------------------------------
1075 // trimming and padding
1076 // ---------------------------------------------------------------------------
1077
1078 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1079 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1080 // live with this by checking that the character is a 7 bit one - even if this
1081 // may fail to detect some spaces (I don't know if Unicode doesn't have
1082 // space-like symbols somewhere except in the first 128 chars), it is arguably
1083 // still better than trimming away accented letters
1084 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1085
1086 // trims spaces (in the sense of isspace) from left or right side
1087 wxString& wxString::Trim(bool bFromRight)
1088 {
1089   // first check if we're going to modify the string at all
1090   if ( !IsEmpty() &&
1091        (
1092         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1093         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1094        )
1095      )
1096   {
1097     // ok, there is at least one space to trim
1098     if ( !CopyBeforeWrite() ) {
1099       wxFAIL_MSG( _T("out of memory in wxString::Trim") );
1100       return *this;
1101     }
1102
1103     if ( bFromRight )
1104     {
1105       // find last non-space character
1106       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1107       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1108         psz--;
1109
1110       // truncate at trailing space start
1111       *++psz = wxT('\0');
1112       GetStringData()->nDataLength = psz - m_pchData;
1113     }
1114     else
1115     {
1116       // find first non-space character
1117       const wxChar *psz = m_pchData;
1118       while ( wxSafeIsspace(*psz) )
1119         psz++;
1120
1121       // fix up data and length
1122       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1123       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1124       GetStringData()->nDataLength = nDataLength;
1125     }
1126   }
1127
1128   return *this;
1129 }
1130
1131 // adds nCount characters chPad to the string from either side
1132 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1133 {
1134   wxString s(chPad, nCount);
1135
1136   if ( bFromRight )
1137     *this += s;
1138   else
1139   {
1140     s += *this;
1141     *this = s;
1142   }
1143
1144   return *this;
1145 }
1146
1147 // truncate the string
1148 wxString& wxString::Truncate(size_t uiLen)
1149 {
1150   if ( uiLen < Len() ) {
1151     if ( !CopyBeforeWrite() ) {
1152       wxFAIL_MSG( _T("out of memory in wxString::Truncate") );
1153       return *this;
1154     }
1155
1156     *(m_pchData + uiLen) = wxT('\0');
1157     GetStringData()->nDataLength = uiLen;
1158   }
1159   //else: nothing to do, string is already short enough
1160
1161   return *this;
1162 }
1163
1164 // ---------------------------------------------------------------------------
1165 // finding (return wxNOT_FOUND if not found and index otherwise)
1166 // ---------------------------------------------------------------------------
1167
1168 // find a character
1169 int wxString::Find(wxChar ch, bool bFromEnd) const
1170 {
1171   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1172
1173   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1174 }
1175
1176 // find a sub-string (like strstr)
1177 int wxString::Find(const wxChar *pszSub) const
1178 {
1179   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1180
1181   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1182 }
1183
1184 // ----------------------------------------------------------------------------
1185 // conversion to numbers
1186 // ----------------------------------------------------------------------------
1187
1188 bool wxString::ToLong(long *val, int base) const
1189 {
1190     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1191     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1192
1193     const wxChar *start = c_str();
1194     wxChar *end;
1195     *val = wxStrtol(start, &end, base);
1196
1197     // return TRUE only if scan was stopped by the terminating NUL and if the
1198     // string was not empty to start with
1199     return !*end && (end != start);
1200 }
1201
1202 bool wxString::ToULong(unsigned long *val, int base) const
1203 {
1204     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1205     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1206
1207     const wxChar *start = c_str();
1208     wxChar *end;
1209     *val = wxStrtoul(start, &end, base);
1210
1211     // return TRUE only if scan was stopped by the terminating NUL and if the
1212     // string was not empty to start with
1213     return !*end && (end != start);
1214 }
1215
1216 bool wxString::ToDouble(double *val) const
1217 {
1218     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1219
1220     const wxChar *start = c_str();
1221     wxChar *end;
1222     *val = wxStrtod(start, &end);
1223
1224     // return TRUE only if scan was stopped by the terminating NUL and if the
1225     // string was not empty to start with
1226     return !*end && (end != start);
1227 }
1228
1229 // ---------------------------------------------------------------------------
1230 // formatted output
1231 // ---------------------------------------------------------------------------
1232
1233 /* static */
1234 wxString wxString::Format(const wxChar *pszFormat, ...)
1235 {
1236     va_list argptr;
1237     va_start(argptr, pszFormat);
1238
1239     wxString s;
1240     s.PrintfV(pszFormat, argptr);
1241
1242     va_end(argptr);
1243
1244     return s;
1245 }
1246
1247 /* static */
1248 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1249 {
1250     wxString s;
1251     s.PrintfV(pszFormat, argptr);
1252     return s;
1253 }
1254
1255 int wxString::Printf(const wxChar *pszFormat, ...)
1256 {
1257   va_list argptr;
1258   va_start(argptr, pszFormat);
1259
1260   int iLen = PrintfV(pszFormat, argptr);
1261
1262   va_end(argptr);
1263
1264   return iLen;
1265 }
1266
1267 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1268 {
1269     int size = 1024;
1270     for ( ;; )
1271     {
1272         wxChar *buf = GetWriteBuf(size + 1);
1273         if ( !buf )
1274         {
1275             // out of memory
1276             return -1;
1277         }
1278
1279         int len = wxVsnprintf(buf, size, pszFormat, argptr);
1280
1281         // some implementations of vsnprintf() don't NUL terminate the string
1282         // if there is not enough space for it so always do it manually
1283         buf[size] = _T('\0');
1284
1285         UngetWriteBuf();
1286
1287         if ( len >= 0 )
1288         {
1289             // ok, there was enough space
1290             break;
1291         }
1292
1293         // still not enough, double it again
1294         size *= 2;
1295     }
1296
1297     // we could have overshot
1298     Shrink();
1299
1300     return Len();
1301 }
1302
1303 // ----------------------------------------------------------------------------
1304 // misc other operations
1305 // ----------------------------------------------------------------------------
1306
1307 // returns TRUE if the string matches the pattern which may contain '*' and
1308 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1309 // of them)
1310 bool wxString::Matches(const wxChar *pszMask) const
1311 {
1312     // I disable this code as it doesn't seem to be faster (in fact, it seems
1313     // to be much slower) than the old, hand-written code below and using it
1314     // here requires always linking with libregex even if the user code doesn't
1315     // use it
1316 #if 0 // wxUSE_REGEX
1317     // first translate the shell-like mask into a regex
1318     wxString pattern;
1319     pattern.reserve(wxStrlen(pszMask));
1320
1321     pattern += _T('^');
1322     while ( *pszMask )
1323     {
1324         switch ( *pszMask )
1325         {
1326             case _T('?'):
1327                 pattern += _T('.');
1328                 break;
1329
1330             case _T('*'):
1331                 pattern += _T(".*");
1332                 break;
1333
1334             case _T('^'):
1335             case _T('.'):
1336             case _T('$'):
1337             case _T('('):
1338             case _T(')'):
1339             case _T('|'):
1340             case _T('+'):
1341             case _T('\\'):
1342                 // these characters are special in a RE, quote them
1343                 // (however note that we don't quote '[' and ']' to allow
1344                 // using them for Unix shell like matching)
1345                 pattern += _T('\\');
1346                 // fall through
1347
1348             default:
1349                 pattern += *pszMask;
1350         }
1351
1352         pszMask++;
1353     }
1354     pattern += _T('$');
1355
1356     // and now use it
1357     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1358 #else // !wxUSE_REGEX
1359   // TODO: this is, of course, awfully inefficient...
1360
1361   // the char currently being checked
1362   const wxChar *pszTxt = c_str();
1363
1364   // the last location where '*' matched
1365   const wxChar *pszLastStarInText = NULL;
1366   const wxChar *pszLastStarInMask = NULL;
1367
1368 match:
1369   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1370     switch ( *pszMask ) {
1371       case wxT('?'):
1372         if ( *pszTxt == wxT('\0') )
1373           return FALSE;
1374
1375         // pszTxt and pszMask will be incremented in the loop statement
1376
1377         break;
1378
1379       case wxT('*'):
1380         {
1381           // remember where we started to be able to backtrack later
1382           pszLastStarInText = pszTxt;
1383           pszLastStarInMask = pszMask;
1384
1385           // ignore special chars immediately following this one
1386           // (should this be an error?)
1387           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1388             pszMask++;
1389
1390           // if there is nothing more, match
1391           if ( *pszMask == wxT('\0') )
1392             return TRUE;
1393
1394           // are there any other metacharacters in the mask?
1395           size_t uiLenMask;
1396           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1397
1398           if ( pEndMask != NULL ) {
1399             // we have to match the string between two metachars
1400             uiLenMask = pEndMask - pszMask;
1401           }
1402           else {
1403             // we have to match the remainder of the string
1404             uiLenMask = wxStrlen(pszMask);
1405           }
1406
1407           wxString strToMatch(pszMask, uiLenMask);
1408           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1409           if ( pMatch == NULL )
1410             return FALSE;
1411
1412           // -1 to compensate "++" in the loop
1413           pszTxt = pMatch + uiLenMask - 1;
1414           pszMask += uiLenMask - 1;
1415         }
1416         break;
1417
1418       default:
1419         if ( *pszMask != *pszTxt )
1420           return FALSE;
1421         break;
1422     }
1423   }
1424
1425   // match only if nothing left
1426   if ( *pszTxt == wxT('\0') )
1427     return TRUE;
1428
1429   // if we failed to match, backtrack if we can
1430   if ( pszLastStarInText ) {
1431     pszTxt = pszLastStarInText + 1;
1432     pszMask = pszLastStarInMask;
1433
1434     pszLastStarInText = NULL;
1435
1436     // don't bother resetting pszLastStarInMask, it's unnecessary
1437
1438     goto match;
1439   }
1440
1441   return FALSE;
1442 #endif // wxUSE_REGEX/!wxUSE_REGEX
1443 }
1444
1445 // Count the number of chars
1446 int wxString::Freq(wxChar ch) const
1447 {
1448     int count = 0;
1449     int len = Len();
1450     for (int i = 0; i < len; i++)
1451     {
1452         if (GetChar(i) == ch)
1453             count ++;
1454     }
1455     return count;
1456 }
1457
1458 // convert to upper case, return the copy of the string
1459 wxString wxString::Upper() const
1460 { wxString s(*this); return s.MakeUpper(); }
1461
1462 // convert to lower case, return the copy of the string
1463 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1464
1465 int wxString::sprintf(const wxChar *pszFormat, ...)
1466   {
1467     va_list argptr;
1468     va_start(argptr, pszFormat);
1469     int iLen = PrintfV(pszFormat, argptr);
1470     va_end(argptr);
1471     return iLen;
1472   }
1473
1474 // ---------------------------------------------------------------------------
1475 // standard C++ library string functions
1476 // ---------------------------------------------------------------------------
1477
1478 #ifdef  wxSTD_STRING_COMPATIBILITY
1479
1480 void wxString::resize(size_t nSize, wxChar ch)
1481 {
1482     size_t len = length();
1483
1484     if ( nSize < len )
1485     {
1486         Truncate(nSize);
1487     }
1488     else if ( nSize > len )
1489     {
1490         *this += wxString(ch, nSize - len);
1491     }
1492     //else: we have exactly the specified length, nothing to do
1493 }
1494
1495 void wxString::swap(wxString& str)
1496 {
1497     // this is slightly less efficient than fiddling with m_pchData directly,
1498     // but it is still quite efficient as we don't copy the string here because
1499     // ref count always stays positive
1500     wxString tmp = str;
1501     str = *this;
1502     *this = tmp;
1503 }
1504
1505 wxString& wxString::insert(size_t nPos, const wxString& str)
1506 {
1507   wxASSERT( str.GetStringData()->IsValid() );
1508   wxASSERT( nPos <= Len() );
1509
1510   if ( !str.IsEmpty() ) {
1511     wxString strTmp;
1512     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1513     wxStrncpy(pc, c_str(), nPos);
1514     wxStrcpy(pc + nPos, str);
1515     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1516     strTmp.UngetWriteBuf();
1517     *this = strTmp;
1518   }
1519
1520   return *this;
1521 }
1522
1523 size_t wxString::find(const wxString& str, size_t nStart) const
1524 {
1525   wxASSERT( str.GetStringData()->IsValid() );
1526   wxASSERT( nStart <= Len() );
1527
1528   const wxChar *p = wxStrstr(c_str() + nStart, str);
1529
1530   return p == NULL ? npos : p - c_str();
1531 }
1532
1533 // VC++ 1.5 can't cope with the default argument in the header.
1534 #if !defined(__VISUALC__) || defined(__WIN32__)
1535 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1536 {
1537   return find(wxString(sz, n), nStart);
1538 }
1539 #endif // VC++ 1.5
1540
1541 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1542 #if !defined(__BORLANDC__)
1543 size_t wxString::find(wxChar ch, size_t nStart) const
1544 {
1545   wxASSERT( nStart <= Len() );
1546
1547   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1548
1549   return p == NULL ? npos : p - c_str();
1550 }
1551 #endif
1552
1553 size_t wxString::rfind(const wxString& str, size_t nStart) const
1554 {
1555   wxASSERT( str.GetStringData()->IsValid() );
1556   wxASSERT( nStart == npos || nStart <= Len() );
1557
1558   // TODO could be made much quicker than that
1559   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1560   while ( p >= c_str() + str.Len() ) {
1561     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1562       return p - str.Len() - c_str();
1563     p--;
1564   }
1565
1566   return npos;
1567 }
1568
1569 // VC++ 1.5 can't cope with the default argument in the header.
1570 #if !defined(__VISUALC__) || defined(__WIN32__)
1571 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1572 {
1573     return rfind(wxString(sz, n == npos ? wxSTRING_MAXLEN : n), nStart);
1574 }
1575
1576 size_t wxString::rfind(wxChar ch, size_t nStart) const
1577 {
1578     if ( nStart == npos )
1579     {
1580         nStart = Len();
1581     }
1582     else
1583     {
1584         wxASSERT( nStart <= Len() );
1585     }
1586
1587     const wxChar *p = wxStrrchr(c_str(), ch);
1588
1589     if ( p == NULL )
1590         return npos;
1591
1592     size_t result = p - c_str();
1593     return ( result > nStart ) ? npos : result;
1594 }
1595 #endif // VC++ 1.5
1596
1597 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1598 {
1599     const wxChar *start = c_str() + nStart;
1600     const wxChar *firstOf = wxStrpbrk(start, sz);
1601     if ( firstOf )
1602         return firstOf - c_str();
1603     else
1604         return npos;
1605 }
1606
1607 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1608 {
1609     if ( nStart == npos )
1610     {
1611         nStart = Len();
1612     }
1613     else
1614     {
1615         wxASSERT( nStart <= Len() );
1616     }
1617
1618     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1619     {
1620         if ( wxStrchr(sz, *p) )
1621             return p - c_str();
1622     }
1623
1624     return npos;
1625 }
1626
1627 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1628 {
1629     if ( nStart == npos )
1630     {
1631         nStart = Len();
1632     }
1633     else
1634     {
1635         wxASSERT( nStart <= Len() );
1636     }
1637
1638     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1639     if ( nAccept >= length() - nStart )
1640         return npos;
1641     else
1642         return nAccept;
1643 }
1644
1645 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1646 {
1647     wxASSERT( nStart <= Len() );
1648
1649     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1650     {
1651         if ( *p != ch )
1652             return p - c_str();
1653     }
1654
1655     return npos;
1656 }
1657
1658 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1659 {
1660     if ( nStart == npos )
1661     {
1662         nStart = Len();
1663     }
1664     else
1665     {
1666         wxASSERT( nStart <= Len() );
1667     }
1668
1669     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1670     {
1671         if ( !wxStrchr(sz, *p) )
1672             return p - c_str();
1673     }
1674
1675     return npos;
1676 }
1677
1678 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1679 {
1680     if ( nStart == npos )
1681     {
1682         nStart = Len();
1683     }
1684     else
1685     {
1686         wxASSERT( nStart <= Len() );
1687     }
1688
1689     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1690     {
1691         if ( *p != ch )
1692             return p - c_str();
1693     }
1694
1695     return npos;
1696 }
1697
1698 wxString& wxString::erase(size_t nStart, size_t nLen)
1699 {
1700   wxString strTmp(c_str(), nStart);
1701   if ( nLen != npos ) {
1702     wxASSERT( nStart + nLen <= Len() );
1703
1704     strTmp.append(c_str() + nStart + nLen);
1705   }
1706
1707   *this = strTmp;
1708   return *this;
1709 }
1710
1711 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1712 {
1713   wxASSERT_MSG( nStart + nLen <= Len(),
1714                 _T("index out of bounds in wxString::replace") );
1715
1716   wxString strTmp;
1717   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1718
1719   if ( nStart != 0 )
1720     strTmp.append(c_str(), nStart);
1721   strTmp << sz << c_str() + nStart + nLen;
1722
1723   *this = strTmp;
1724   return *this;
1725 }
1726
1727 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1728 {
1729   return replace(nStart, nLen, wxString(ch, nCount));
1730 }
1731
1732 wxString& wxString::replace(size_t nStart, size_t nLen,
1733                             const wxString& str, size_t nStart2, size_t nLen2)
1734 {
1735   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1736 }
1737
1738 wxString& wxString::replace(size_t nStart, size_t nLen,
1739                         const wxChar* sz, size_t nCount)
1740 {
1741   return replace(nStart, nLen, wxString(sz, nCount));
1742 }
1743
1744 #endif  //std::string compatibility
1745
1746 // ============================================================================
1747 // ArrayString
1748 // ============================================================================
1749
1750 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1751 #define   ARRAY_MAXSIZE_INCREMENT       4096
1752
1753 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1754 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1755 #endif
1756
1757 #define   STRING(p)   ((wxString *)(&(p)))
1758
1759 // ctor
1760 void wxArrayString::Init(bool autoSort)
1761 {
1762   m_nSize  =
1763   m_nCount = 0;
1764   m_pItems = (wxChar **) NULL;
1765   m_autoSort = autoSort;
1766 }
1767
1768 // copy ctor
1769 wxArrayString::wxArrayString(const wxArrayString& src)
1770 {
1771   Init(src.m_autoSort);
1772
1773   *this = src;
1774 }
1775
1776 // assignment operator
1777 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1778 {
1779   if ( m_nSize > 0 )
1780     Clear();
1781
1782   Copy(src);
1783
1784   m_autoSort = src.m_autoSort;
1785
1786   return *this;
1787 }
1788
1789 void wxArrayString::Copy(const wxArrayString& src)
1790 {
1791   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1792     Alloc(src.m_nCount);
1793
1794   for ( size_t n = 0; n < src.m_nCount; n++ )
1795     Add(src[n]);
1796 }
1797
1798 // grow the array
1799 void wxArrayString::Grow(size_t nIncrement)
1800 {
1801   // only do it if no more place
1802   if ( (m_nSize - m_nCount) < nIncrement ) {
1803     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1804     // be never resized!
1805     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1806       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1807     #endif
1808
1809     if ( m_nSize == 0 ) {
1810       // was empty, alloc some memory
1811       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1812       if (m_nSize < nIncrement)
1813           m_nSize = nIncrement;
1814       m_pItems = new wxChar *[m_nSize];
1815     }
1816     else {
1817       // otherwise when it's called for the first time, nIncrement would be 0
1818       // and the array would never be expanded
1819       // add 50% but not too much
1820       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1821                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1822       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1823         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1824       if ( nIncrement < ndefIncrement )
1825         nIncrement = ndefIncrement;
1826       m_nSize += nIncrement;
1827       wxChar **pNew = new wxChar *[m_nSize];
1828
1829       // copy data to new location
1830       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1831
1832       // delete old memory (but do not release the strings!)
1833       wxDELETEA(m_pItems);
1834
1835       m_pItems = pNew;
1836     }
1837   }
1838 }
1839
1840 void wxArrayString::Free()
1841 {
1842   for ( size_t n = 0; n < m_nCount; n++ ) {
1843     STRING(m_pItems[n])->GetStringData()->Unlock();
1844   }
1845 }
1846
1847 // deletes all the strings from the list
1848 void wxArrayString::Empty()
1849 {
1850   Free();
1851
1852   m_nCount = 0;
1853 }
1854
1855 // as Empty, but also frees memory
1856 void wxArrayString::Clear()
1857 {
1858   Free();
1859
1860   m_nSize  =
1861   m_nCount = 0;
1862
1863   wxDELETEA(m_pItems);
1864 }
1865
1866 // dtor
1867 wxArrayString::~wxArrayString()
1868 {
1869   Free();
1870
1871   wxDELETEA(m_pItems);
1872 }
1873
1874 // pre-allocates memory (frees the previous data!)
1875 void wxArrayString::Alloc(size_t nSize)
1876 {
1877   // only if old buffer was not big enough
1878   if ( nSize > m_nSize ) {
1879     Free();
1880     wxDELETEA(m_pItems);
1881     m_pItems = new wxChar *[nSize];
1882     m_nSize  = nSize;
1883   }
1884
1885   m_nCount = 0;
1886 }
1887
1888 // minimizes the memory usage by freeing unused memory
1889 void wxArrayString::Shrink()
1890 {
1891   // only do it if we have some memory to free
1892   if( m_nCount < m_nSize ) {
1893     // allocates exactly as much memory as we need
1894     wxChar **pNew = new wxChar *[m_nCount];
1895
1896     // copy data to new location
1897     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1898     delete [] m_pItems;
1899     m_pItems = pNew;
1900   }
1901 }
1902
1903 // return a wxString[] as required for some control ctors.
1904 wxString* wxArrayString::GetStringArray() const
1905 {
1906     wxString *array = 0;
1907
1908     if( m_nCount > 0 )
1909     {
1910         array = new wxString[m_nCount];
1911         for( size_t i = 0; i < m_nCount; i++ )
1912             array[i] = m_pItems[i];
1913     }
1914
1915     return array;
1916 }
1917
1918 // searches the array for an item (forward or backwards)
1919 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1920 {
1921   if ( m_autoSort ) {
1922     // use binary search in the sorted array
1923     wxASSERT_MSG( bCase && !bFromEnd,
1924                   wxT("search parameters ignored for auto sorted array") );
1925
1926     size_t i,
1927            lo = 0,
1928            hi = m_nCount;
1929     int res;
1930     while ( lo < hi ) {
1931       i = (lo + hi)/2;
1932
1933       res = wxStrcmp(sz, m_pItems[i]);
1934       if ( res < 0 )
1935         hi = i;
1936       else if ( res > 0 )
1937         lo = i + 1;
1938       else
1939         return i;
1940     }
1941
1942     return wxNOT_FOUND;
1943   }
1944   else {
1945     // use linear search in unsorted array
1946     if ( bFromEnd ) {
1947       if ( m_nCount > 0 ) {
1948         size_t ui = m_nCount;
1949         do {
1950           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1951             return ui;
1952         }
1953         while ( ui != 0 );
1954       }
1955     }
1956     else {
1957       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1958         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1959           return ui;
1960       }
1961     }
1962   }
1963
1964   return wxNOT_FOUND;
1965 }
1966
1967 // add item at the end
1968 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
1969 {
1970   if ( m_autoSort ) {
1971     // insert the string at the correct position to keep the array sorted
1972     size_t i,
1973            lo = 0,
1974            hi = m_nCount;
1975     int res;
1976     while ( lo < hi ) {
1977       i = (lo + hi)/2;
1978
1979       res = wxStrcmp(str, m_pItems[i]);
1980       if ( res < 0 )
1981         hi = i;
1982       else if ( res > 0 )
1983         lo = i + 1;
1984       else {
1985         lo = hi = i;
1986         break;
1987       }
1988     }
1989
1990     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1991
1992     Insert(str, lo, nInsert);
1993
1994     return (size_t)lo;
1995   }
1996   else {
1997     wxASSERT( str.GetStringData()->IsValid() );
1998
1999     Grow(nInsert);
2000
2001     for (size_t i = 0; i < nInsert; i++)
2002     {
2003         // the string data must not be deleted!
2004         str.GetStringData()->Lock();
2005
2006         // just append
2007         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2008     }
2009     size_t ret = m_nCount;
2010     m_nCount += nInsert;
2011     return ret;
2012   }
2013 }
2014
2015 // add item at the given position
2016 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2017 {
2018   wxASSERT( str.GetStringData()->IsValid() );
2019
2020   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2021   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2022                wxT("array size overflow in wxArrayString::Insert") );
2023
2024   Grow(nInsert);
2025
2026   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2027           (m_nCount - nIndex)*sizeof(wxChar *));
2028
2029   for (size_t i = 0; i < nInsert; i++)
2030   {
2031       str.GetStringData()->Lock();
2032       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2033   }
2034   m_nCount += nInsert;
2035 }
2036
2037 // expand the array
2038 void wxArrayString::SetCount(size_t count)
2039 {
2040     Alloc(count);
2041
2042     wxString s;
2043     while ( m_nCount < count )
2044         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2045 }
2046
2047 // removes item from array (by index)
2048 void wxArrayString::Remove(size_t nIndex, size_t nRemove)
2049 {
2050   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2051   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2052                wxT("removing too many elements in wxArrayString::Remove") );
2053
2054   // release our lock
2055   for (size_t i = 0; i < nRemove; i++)
2056       Item(nIndex + i).GetStringData()->Unlock();
2057
2058   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2059           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2060   m_nCount -= nRemove;
2061 }
2062
2063 // removes item from array (by value)
2064 void wxArrayString::Remove(const wxChar *sz)
2065 {
2066   int iIndex = Index(sz);
2067
2068   wxCHECK_RET( iIndex != wxNOT_FOUND,
2069                wxT("removing inexistent element in wxArrayString::Remove") );
2070
2071   Remove(iIndex);
2072 }
2073
2074 // ----------------------------------------------------------------------------
2075 // sorting
2076 // ----------------------------------------------------------------------------
2077
2078 // we can only sort one array at a time with the quick-sort based
2079 // implementation
2080 #if wxUSE_THREADS
2081   // need a critical section to protect access to gs_compareFunction and
2082   // gs_sortAscending variables
2083   static wxCriticalSection *gs_critsectStringSort = NULL;
2084
2085   // call this before the value of the global sort vars is changed/after
2086   // you're finished with them
2087   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2088                            gs_critsectStringSort = new wxCriticalSection;     \
2089                            gs_critsectStringSort->Enter()
2090   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2091                            delete gs_critsectStringSort;                      \
2092                            gs_critsectStringSort = NULL
2093 #else // !threads
2094   #define START_SORT()
2095   #define END_SORT()
2096 #endif // wxUSE_THREADS
2097
2098 // function to use for string comparaison
2099 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2100
2101 // if we don't use the compare function, this flag tells us if we sort the
2102 // array in ascending or descending order
2103 static bool gs_sortAscending = TRUE;
2104
2105 // function which is called by quick sort
2106 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2107 wxStringCompareFunction(const void *first, const void *second)
2108 {
2109   wxString *strFirst = (wxString *)first;
2110   wxString *strSecond = (wxString *)second;
2111
2112   if ( gs_compareFunction ) {
2113     return gs_compareFunction(*strFirst, *strSecond);
2114   }
2115   else {
2116     // maybe we should use wxStrcoll
2117     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2118
2119     return gs_sortAscending ? result : -result;
2120   }
2121 }
2122
2123 // sort array elements using passed comparaison function
2124 void wxArrayString::Sort(CompareFunction compareFunction)
2125 {
2126   START_SORT();
2127
2128   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2129   gs_compareFunction = compareFunction;
2130
2131   DoSort();
2132
2133   // reset it to NULL so that Sort(bool) will work the next time
2134   gs_compareFunction = NULL;
2135
2136   END_SORT();
2137 }
2138
2139 void wxArrayString::Sort(bool reverseOrder)
2140 {
2141   START_SORT();
2142
2143   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2144   gs_sortAscending = !reverseOrder;
2145
2146   DoSort();
2147
2148   END_SORT();
2149 }
2150
2151 void wxArrayString::DoSort()
2152 {
2153   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2154
2155   // just sort the pointers using qsort() - of course it only works because
2156   // wxString() *is* a pointer to its data
2157   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2158 }
2159
2160 bool wxArrayString::operator==(const wxArrayString& a) const
2161 {
2162     if ( m_nCount != a.m_nCount )
2163         return FALSE;
2164
2165     for ( size_t n = 0; n < m_nCount; n++ )
2166     {
2167         if ( Item(n) != a[n] )
2168             return FALSE;
2169     }
2170
2171     return TRUE;
2172 }
2173