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