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 WXDLLIMPEXP_BASE *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         }
 264         else
 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                 m_pchData[nLen] = 0;
 271                 return;
 272             }
 273             //else: the conversion failed -- leave the string empty (what else?)
 274         }
 275     }
 276
 277     Init();
 278 }
 279
 280 #else // ANSI
 281
 282 #if wxUSE_WCHAR_T
 283 // from wide string
 284 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 285 {
 286     // first get the size of the buffer we need
 287     size_t nLen;
 288     if ( pwz )
 289     {
 290         // calculate the needed size ourselves or use a provide one
 291         nLen = nLength == wxSTRING_MAXLEN ? conv.WC2MB(NULL, pwz, 0) : nLength;
 292     }
 293     else
 294     {
 295         // nothing to convert
 296         nLen = 0;
 297     }
 298
 299     // anything to do?
 300     if ( (nLen != 0) && (nLen != (size_t)-1) )
 301     {
 302         if ( !AllocBuffer(nLen) )
 303         {
 304             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 305         }
 306         else
 307         {
 308             // WC2MB wants the buffer size, not the string length
 309             if ( conv.WC2MB(m_pchData, pwz, nLen + 1) != (size_t)-1 )
 310             {
 311                 // initialized ok
 312                 return;
 313             }
 314             //else: the conversion failed -- leave the string empty (what else?)
 315         }
 316     }
 317
 318     Init();
 319 }
 320 #endif // wxUSE_WCHAR_T
 321
 322 #endif // Unicode/ANSI
 323
 324 // ---------------------------------------------------------------------------
 325 // memory allocation
 326 // ---------------------------------------------------------------------------
 327
 328 // allocates memory needed to store a C string of length nLen
 329 bool wxString::AllocBuffer(size_t nLen)
 330 {
 331   // allocating 0 sized buffer doesn't make sense, all empty strings should
 332   // reuse g_strEmpty
 333   wxASSERT( nLen >  0 );
 334
 335   // make sure that we don't overflow
 336   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 337                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 338
 339   STATISTICS_ADD(Length, nLen);
 340
 341   // allocate memory:
 342   // 1) one extra character for '\0' termination
 343   // 2) sizeof(wxStringData) for housekeeping info
 344   wxStringData* pData = (wxStringData*)
 345     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 346
 347   if ( pData == NULL ) {
 348     // allocation failures are handled by the caller
 349     return FALSE;
 350   }
 351
 352   pData->nRefs        = 1;
 353   pData->nDataLength  = nLen;
 354   pData->nAllocLength = nLen + EXTRA_ALLOC;
 355   m_pchData           = pData->data();  // data starts after wxStringData
 356   m_pchData[nLen]     = wxT('\0');
 357   return TRUE;
 358 }
 359
 360 // must be called before changing this string
 361 bool wxString::CopyBeforeWrite()
 362 {
 363   wxStringData* pData = GetStringData();
 364
 365   if ( pData->IsShared() ) {
 366     pData->Unlock();                // memory not freed because shared
 367     size_t nLen = pData->nDataLength;
 368     if ( !AllocBuffer(nLen) ) {
 369       // allocation failures are handled by the caller
 370       return FALSE;
 371     }
 372     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 373   }
 374
 375   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 376
 377   return TRUE;
 378 }
 379
 380 // must be called before replacing contents of this string
 381 bool wxString::AllocBeforeWrite(size_t nLen)
 382 {
 383   wxASSERT( nLen != 0 );  // doesn't make any sense
 384
 385   // must not share string and must have enough space
 386   wxStringData* pData = GetStringData();
 387   if ( pData->IsShared() || pData->IsEmpty() ) {
 388     // can't work with old buffer, get new one
 389     pData->Unlock();
 390     if ( !AllocBuffer(nLen) ) {
 391       // allocation failures are handled by the caller
 392       return FALSE;
 393     }
 394   }
 395   else {
 396     if ( nLen > pData->nAllocLength ) {
 397       // realloc the buffer instead of calling malloc() again, this is more
 398       // efficient
 399       STATISTICS_ADD(Length, nLen);
 400
 401       nLen += EXTRA_ALLOC;
 402
 403       pData = (wxStringData*)
 404           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 405
 406       if ( pData == NULL ) {
 407         // allocation failures are handled by the caller
 408         // keep previous data since reallocation failed
 409         return FALSE;
 410       }
 411
 412       pData->nAllocLength = nLen;
 413       m_pchData = pData->data();
 414     }
 415
 416     // now we have enough space, just update the string length
 417     pData->nDataLength = nLen;
 418   }
 419
 420   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 421
 422   return TRUE;
 423 }
 424
 425 // allocate enough memory for nLen characters
 426 bool wxString::Alloc(size_t nLen)
 427 {
 428   wxStringData *pData = GetStringData();
 429   if ( pData->nAllocLength <= nLen ) {
 430     if ( pData->IsEmpty() ) {
 431       nLen += EXTRA_ALLOC;
 432
 433       wxStringData* pData = (wxStringData*)
 434           malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 435
 436       if ( pData == NULL ) {
 437         // allocation failure handled by caller
 438         return FALSE;
 439       }
 440
 441       pData->nRefs = 1;
 442       pData->nDataLength = 0;
 443       pData->nAllocLength = nLen;
 444       m_pchData = pData->data();  // data starts after wxStringData
 445       m_pchData[0u] = wxT('\0');
 446     }
 447     else if ( pData->IsShared() ) {
 448       pData->Unlock();                // memory not freed because shared
 449       size_t nOldLen = pData->nDataLength;
 450       if ( !AllocBuffer(nLen) ) {
 451         // allocation failure handled by caller
 452         return FALSE;
 453       }
 454       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 455     }
 456     else {
 457       nLen += EXTRA_ALLOC;
 458
 459       pData = (wxStringData *)
 460         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 461
 462       if ( pData == NULL ) {
 463         // allocation failure handled by caller
 464         // keep previous data since reallocation failed
 465         return FALSE;
 466       }
 467
 468       // it's not important if the pointer changed or not (the check for this
 469       // is not faster than assigning to m_pchData in all cases)
 470       pData->nAllocLength = nLen;
 471       m_pchData = pData->data();
 472     }
 473   }
 474   //else: we've already got enough
 475   return TRUE;
 476 }
 477
 478 // shrink to minimal size (releasing extra memory)
 479 bool wxString::Shrink()
 480 {
 481   wxStringData *pData = GetStringData();
 482
 483   size_t nLen = pData->nDataLength;
 484   void *p = realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 485
 486   if ( p == NULL) {
 487       wxFAIL_MSG( _T("out of memory reallocating wxString data") );
 488       // keep previous data since reallocation failed
 489       return FALSE;
 490   }
 491
 492   if ( p != pData )
 493   {
 494       // contrary to what one might believe, some realloc() implementation do
 495       // move the memory block even when its size is reduced
 496       pData = (wxStringData *)p;
 497
 498       m_pchData = pData->data();
 499   }
 500
 501   pData->nAllocLength = nLen;
 502
 503   return TRUE;
 504 }
 505
 506 // get the pointer to writable buffer of (at least) nLen bytes
 507 wxChar *wxString::GetWriteBuf(size_t nLen)
 508 {
 509   if ( !AllocBeforeWrite(nLen) ) {
 510     // allocation failure handled by caller
 511     return NULL;
 512   }
 513
 514   wxASSERT( GetStringData()->nRefs == 1 );
 515   GetStringData()->Validate(FALSE);
 516
 517   return m_pchData;
 518 }
 519
 520 // put string back in a reasonable state after GetWriteBuf
 521 void wxString::UngetWriteBuf()
 522 {
 523   GetStringData()->nDataLength = wxStrlen(m_pchData);
 524   GetStringData()->Validate(TRUE);
 525 }
 526
 527 void wxString::UngetWriteBuf(size_t nLen)
 528 {
 529   GetStringData()->nDataLength = nLen;
 530   GetStringData()->Validate(TRUE);
 531 }
 532
 533 // ---------------------------------------------------------------------------
 534 // data access
 535 // ---------------------------------------------------------------------------
 536
 537 // all functions are inline in string.h
 538
 539 // ---------------------------------------------------------------------------
 540 // assignment operators
 541 // ---------------------------------------------------------------------------
 542
 543 // helper function: does real copy
 544 bool wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 545 {
 546   if ( nSrcLen == 0 ) {
 547     Reinit();
 548   }
 549   else {
 550     if ( !AllocBeforeWrite(nSrcLen) ) {
 551       // allocation failure handled by caller
 552       return FALSE;
 553     }
 554     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 555     GetStringData()->nDataLength = nSrcLen;
 556     m_pchData[nSrcLen] = wxT('\0');
 557   }
 558   return TRUE;
 559 }
 560
 561 // assigns one string to another
 562 wxString& wxString::operator=(const wxString& stringSrc)
 563 {
 564   wxASSERT( stringSrc.GetStringData()->IsValid() );
 565
 566   // don't copy string over itself
 567   if ( m_pchData != stringSrc.m_pchData ) {
 568     if ( stringSrc.GetStringData()->IsEmpty() ) {
 569       Reinit();
 570     }
 571     else {
 572       // adjust references
 573       GetStringData()->Unlock();
 574       m_pchData = stringSrc.m_pchData;
 575       GetStringData()->Lock();
 576     }
 577   }
 578
 579   return *this;
 580 }
 581
 582 // assigns a single character
 583 wxString& wxString::operator=(wxChar ch)
 584 {
 585   if ( !AssignCopy(1, &ch) ) {
 586     wxFAIL_MSG( _T("out of memory in wxString::operator=(wxChar)") );
 587   }
 588   return *this;
 589 }
 590
 591
 592 // assigns C string
 593 wxString& wxString::operator=(const wxChar *psz)
 594 {
 595   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 596     wxFAIL_MSG( _T("out of memory in wxString::operator=(const wxChar *)") );
 597   }
 598   return *this;
 599 }
 600
 601 #if !wxUSE_UNICODE
 602
 603 // same as 'signed char' variant
 604 wxString& wxString::operator=(const unsigned char* psz)
 605 {
 606   *this = (const char *)psz;
 607   return *this;
 608 }
 609
 610 #if wxUSE_WCHAR_T
 611 wxString& wxString::operator=(const wchar_t *pwz)
 612 {
 613   wxString str(pwz);
 614   *this = str;
 615   return *this;
 616 }
 617 #endif
 618
 619 #endif
 620
 621 // ---------------------------------------------------------------------------
 622 // string concatenation
 623 // ---------------------------------------------------------------------------
 624
 625 // add something to this string
 626 bool wxString::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData)
 627 {
 628   STATISTICS_ADD(SummandLength, nSrcLen);
 629
 630   // concatenating an empty string is a NOP
 631   if ( nSrcLen > 0 ) {
 632     wxStringData *pData = GetStringData();
 633     size_t nLen = pData->nDataLength;
 634     size_t nNewLen = nLen + nSrcLen;
 635
 636     // alloc new buffer if current is too small
 637     if ( pData->IsShared() ) {
 638       STATISTICS_ADD(ConcatHit, 0);
 639
 640       // we have to allocate another buffer
 641       wxStringData* pOldData = GetStringData();
 642       if ( !AllocBuffer(nNewLen) ) {
 643           // allocation failure handled by caller
 644           return FALSE;
 645       }
 646       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 647       pOldData->Unlock();
 648     }
 649     else if ( nNewLen > pData->nAllocLength ) {
 650       STATISTICS_ADD(ConcatHit, 0);
 651
 652       // we have to grow the buffer
 653       if ( !Alloc(nNewLen) ) {
 654           // allocation failure handled by caller
 655           return FALSE;
 656       }
 657     }
 658     else {
 659       STATISTICS_ADD(ConcatHit, 1);
 660
 661       // the buffer is already big enough
 662     }
 663
 664     // should be enough space
 665     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 666
 667     // fast concatenation - all is done in our buffer
 668     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 669
 670     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 671     GetStringData()->nDataLength = nNewLen; // and fix the length
 672   }
 673   //else: the string to append was empty
 674   return TRUE;
 675 }
 676
 677 /*
 678  * concatenation functions come in 5 flavours:
 679  *  string + string
 680  *  char   + string      and      string + char
 681  *  C str  + string      and      string + C str
 682  */
 683
 684 wxString operator+(const wxString& str1, const wxString& str2)
 685 {
 686   wxASSERT( str1.GetStringData()->IsValid() );
 687   wxASSERT( str2.GetStringData()->IsValid() );
 688
 689   wxString s = str1;
 690   s += str2;
 691
 692   return s;
 693 }
 694
 695 wxString operator+(const wxString& str, wxChar ch)
 696 {
 697   wxASSERT( str.GetStringData()->IsValid() );
 698
 699   wxString s = str;
 700   s += ch;
 701
 702   return s;
 703 }
 704
 705 wxString operator+(wxChar ch, const wxString& str)
 706 {
 707   wxASSERT( str.GetStringData()->IsValid() );
 708
 709   wxString s = ch;
 710   s += str;
 711
 712   return s;
 713 }
 714
 715 wxString operator+(const wxString& str, const wxChar *psz)
 716 {
 717   wxASSERT( str.GetStringData()->IsValid() );
 718
 719   wxString s;
 720   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 721     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 722   }
 723   s = str;
 724   s += psz;
 725
 726   return s;
 727 }
 728
 729 wxString operator+(const wxChar *psz, const wxString& str)
 730 {
 731   wxASSERT( str.GetStringData()->IsValid() );
 732
 733   wxString s;
 734   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 735     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 736   }
 737   s = psz;
 738   s += str;
 739
 740   return s;
 741 }
 742
 743 // ===========================================================================
 744 // other common string functions
 745 // ===========================================================================
 746
 747 #if wxUSE_UNICODE
 748
 749 wxString wxString::FromAscii(const char *ascii)
 750 {
 751     if (!ascii)
 752        return wxEmptyString;
 753
 754     size_t len = strlen( ascii );
 755     wxString res;
 756
 757     if ( len )
 758     {
 759         wxStringBuffer buf(res, len);
 760
 761         wchar_t *dest = buf;
 762
 763         for ( ;; )
 764         {
 765            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
 766                break;
 767         }
 768     }
 769
 770     return res;
 771 }
 772
 773 wxString wxString::FromAscii(const char ascii)
 774 {
 775     // What do we do with '\0' ?
 776
 777     wxString res;
 778     res += (wchar_t)(unsigned char) ascii;
 779
 780     return res;
 781 }
 782
 783 const wxCharBuffer wxString::ToAscii() const
 784 {
 785     // this will allocate enough space for the terminating NUL too
 786     wxCharBuffer buffer(length());
 787
 788     signed char *dest = (signed char *)buffer.data();
 789
 790     const wchar_t *pwc = c_str();
 791     for ( ;; )
 792     {
 793         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
 794
 795         // the output string can't have embedded NULs anyhow, so we can safely
 796         // stop at first of them even if we do have any
 797         if ( !*pwc++ )
 798             break;
 799     }
 800
 801     return buffer;
 802 }
 803
 804 #endif // Unicode
 805
 806 // ---------------------------------------------------------------------------
 807 // simple sub-string extraction
 808 // ---------------------------------------------------------------------------
 809
 810 // helper function: clone the data attached to this string
 811 bool wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 812 {
 813   if ( nCopyLen == 0 ) {
 814     dest.Init();
 815   }
 816   else {
 817     if ( !dest.AllocBuffer(nCopyLen) ) {
 818       // allocation failure handled by caller
 819       return FALSE;
 820     }
 821     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 822   }
 823   return TRUE;
 824 }
 825
 826 // extract string of length nCount starting at nFirst
 827 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 828 {
 829   wxStringData *pData = GetStringData();
 830   size_t nLen = pData->nDataLength;
 831
 832   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 833   if ( nCount == wxSTRING_MAXLEN )
 834   {
 835     nCount = nLen - nFirst;
 836   }
 837
 838   // out-of-bounds requests return sensible things
 839   if ( nFirst + nCount > nLen )
 840   {
 841     nCount = nLen - nFirst;
 842   }
 843
 844   if ( nFirst > nLen )
 845   {
 846     // AllocCopy() will return empty string
 847     nCount = 0;
 848   }
 849
 850   wxString dest;
 851   if ( !AllocCopy(dest, nCount, nFirst) ) {
 852       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
 853   }
 854
 855   return dest;
 856 }
 857
 858 // check that the tring starts with prefix and return the rest of the string
 859 // in the provided pointer if it is not NULL, otherwise return FALSE
 860 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 861 {
 862     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 863
 864     // first check if the beginning of the string matches the prefix: note
 865     // that we don't have to check that we don't run out of this string as
 866     // when we reach the terminating NUL, either prefix string ends too (and
 867     // then it's ok) or we break out of the loop because there is no match
 868     const wxChar *p = c_str();
 869     while ( *prefix )
 870     {
 871         if ( *prefix++ != *p++ )
 872         {
 873             // no match
 874             return FALSE;
 875         }
 876     }
 877
 878     if ( rest )
 879     {
 880         // put the rest of the string into provided pointer
 881         *rest = p;
 882     }
 883
 884     return TRUE;
 885 }
 886
 887 // extract nCount last (rightmost) characters
 888 wxString wxString::Right(size_t nCount) const
 889 {
 890   if ( nCount > (size_t)GetStringData()->nDataLength )
 891     nCount = GetStringData()->nDataLength;
 892
 893   wxString dest;
 894   if ( !AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount) ) {
 895     wxFAIL_MSG( _T("out of memory in wxString::Right") );
 896   }
 897   return dest;
 898 }
 899
 900 // get all characters after the last occurence of ch
 901 // (returns the whole string if ch not found)
 902 wxString wxString::AfterLast(wxChar ch) const
 903 {
 904   wxString str;
 905   int iPos = Find(ch, TRUE);
 906   if ( iPos == wxNOT_FOUND )
 907     str = *this;
 908   else
 909     str = c_str() + iPos + 1;
 910
 911   return str;
 912 }
 913
 914 // extract nCount first (leftmost) characters
 915 wxString wxString::Left(size_t nCount) const
 916 {
 917   if ( nCount > (size_t)GetStringData()->nDataLength )
 918     nCount = GetStringData()->nDataLength;
 919
 920   wxString dest;
 921   if ( !AllocCopy(dest, nCount, 0) ) {
 922     wxFAIL_MSG( _T("out of memory in wxString::Left") );
 923   }
 924   return dest;
 925 }
 926
 927 // get all characters before the first occurence of ch
 928 // (returns the whole string if ch not found)
 929 wxString wxString::BeforeFirst(wxChar ch) const
 930 {
 931   wxString str;
 932   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 933     str += *pc;
 934
 935   return str;
 936 }
 937
 938 /// get all characters before the last occurence of ch
 939 /// (returns empty string if ch not found)
 940 wxString wxString::BeforeLast(wxChar ch) const
 941 {
 942   wxString str;
 943   int iPos = Find(ch, TRUE);
 944   if ( iPos != wxNOT_FOUND && iPos != 0 )
 945     str = wxString(c_str(), iPos);
 946
 947   return str;
 948 }
 949
 950 /// get all characters after the first occurence of ch
 951 /// (returns empty string if ch not found)
 952 wxString wxString::AfterFirst(wxChar ch) const
 953 {
 954   wxString str;
 955   int iPos = Find(ch);
 956   if ( iPos != wxNOT_FOUND )
 957     str = c_str() + iPos + 1;
 958
 959   return str;
 960 }
 961
 962 // replace first (or all) occurences of some substring with another one
 963 size_t
 964 wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 965 {
 966     // if we tried to replace an empty string we'd enter an infinite loop below
 967     wxCHECK_MSG( szOld && *szOld && szNew, 0,
 968                  _T("wxString::Replace(): invalid parameter") );
 969
 970   size_t uiCount = 0;   // count of replacements made
 971
 972   size_t uiOldLen = wxStrlen(szOld);
 973
 974   wxString strTemp;
 975   const wxChar *pCurrent = m_pchData;
 976   const wxChar *pSubstr;
 977   while ( *pCurrent != wxT('\0') ) {
 978     pSubstr = wxStrstr(pCurrent, szOld);
 979     if ( pSubstr == NULL ) {
 980       // strTemp is unused if no replacements were made, so avoid the copy
 981       if ( uiCount == 0 )
 982         return 0;
 983
 984       strTemp += pCurrent;    // copy the rest
 985       break;                  // exit the loop
 986     }
 987     else {
 988       // take chars before match
 989       if ( !strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent) ) {
 990         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
 991         return 0;
 992       }
 993       strTemp += szNew;
 994       pCurrent = pSubstr + uiOldLen;  // restart after match
 995
 996       uiCount++;
 997
 998       // stop now?
 999       if ( !bReplaceAll ) {
1000         strTemp += pCurrent;    // copy the rest
1001         break;                  // exit the loop
1002       }
1003     }
1004   }
1005
1006   // only done if there were replacements, otherwise would have returned above
1007   *this = strTemp;
1008
1009   return uiCount;
1010 }
1011
1012 bool wxString::IsAscii() const
1013 {
1014   const wxChar *s = (const wxChar*) *this;
1015   while(*s){
1016     if(!isascii(*s)) return(FALSE);
1017     s++;
1018   }
1019   return(TRUE);
1020 }
1021
1022 bool wxString::IsWord() const
1023 {
1024   const wxChar *s = (const wxChar*) *this;
1025   while(*s){
1026     if(!wxIsalpha(*s)) return(FALSE);
1027     s++;
1028   }
1029   return(TRUE);
1030 }
1031
1032 bool wxString::IsNumber() const
1033 {
1034   const wxChar *s = (const wxChar*) *this;
1035   if (wxStrlen(s))
1036      if ((s[0] == '-') || (s[0] == '+')) s++;
1037   while(*s){
1038     if(!wxIsdigit(*s)) return(FALSE);
1039     s++;
1040   }
1041   return(TRUE);
1042 }
1043
1044 wxString wxString::Strip(stripType w) const
1045 {
1046     wxString s = *this;
1047     if ( w & leading ) s.Trim(FALSE);
1048     if ( w & trailing ) s.Trim(TRUE);
1049     return s;
1050 }
1051
1052 // ---------------------------------------------------------------------------
1053 // case conversion
1054 // ---------------------------------------------------------------------------
1055
1056 wxString& wxString::MakeUpper()
1057 {
1058   if ( !CopyBeforeWrite() ) {
1059     wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") );
1060     return *this;
1061   }
1062
1063   for ( wxChar *p = m_pchData; *p; p++ )
1064     *p = (wxChar)wxToupper(*p);
1065
1066   return *this;
1067 }
1068
1069 wxString& wxString::MakeLower()
1070 {
1071   if ( !CopyBeforeWrite() ) {
1072     wxFAIL_MSG( _T("out of memory in wxString::MakeLower") );
1073     return *this;
1074   }
1075
1076   for ( wxChar *p = m_pchData; *p; p++ )
1077     *p = (wxChar)wxTolower(*p);
1078
1079   return *this;
1080 }
1081
1082 // ---------------------------------------------------------------------------
1083 // trimming and padding
1084 // ---------------------------------------------------------------------------
1085
1086 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1087 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1088 // live with this by checking that the character is a 7 bit one - even if this
1089 // may fail to detect some spaces (I don't know if Unicode doesn't have
1090 // space-like symbols somewhere except in the first 128 chars), it is arguably
1091 // still better than trimming away accented letters
1092 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1093
1094 // trims spaces (in the sense of isspace) from left or right side
1095 wxString& wxString::Trim(bool bFromRight)
1096 {
1097   // first check if we're going to modify the string at all
1098   if ( !IsEmpty() &&
1099        (
1100         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1101         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1102        )
1103      )
1104   {
1105     // ok, there is at least one space to trim
1106     if ( !CopyBeforeWrite() ) {
1107       wxFAIL_MSG( _T("out of memory in wxString::Trim") );
1108       return *this;
1109     }
1110
1111     if ( bFromRight )
1112     {
1113       // find last non-space character
1114       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1115       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1116         psz--;
1117
1118       // truncate at trailing space start
1119       *++psz = wxT('\0');
1120       GetStringData()->nDataLength = psz - m_pchData;
1121     }
1122     else
1123     {
1124       // find first non-space character
1125       const wxChar *psz = m_pchData;
1126       while ( wxSafeIsspace(*psz) )
1127         psz++;
1128
1129       // fix up data and length
1130       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1131       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1132       GetStringData()->nDataLength = nDataLength;
1133     }
1134   }
1135
1136   return *this;
1137 }
1138
1139 // adds nCount characters chPad to the string from either side
1140 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1141 {
1142   wxString s(chPad, nCount);
1143
1144   if ( bFromRight )
1145     *this += s;
1146   else
1147   {
1148     s += *this;
1149     *this = s;
1150   }
1151
1152   return *this;
1153 }
1154
1155 // truncate the string
1156 wxString& wxString::Truncate(size_t uiLen)
1157 {
1158   if ( uiLen < Len() ) {
1159     if ( !CopyBeforeWrite() ) {
1160       wxFAIL_MSG( _T("out of memory in wxString::Truncate") );
1161       return *this;
1162     }
1163
1164     *(m_pchData + uiLen) = wxT('\0');
1165     GetStringData()->nDataLength = uiLen;
1166   }
1167   //else: nothing to do, string is already short enough
1168
1169   return *this;
1170 }
1171
1172 // ---------------------------------------------------------------------------
1173 // finding (return wxNOT_FOUND if not found and index otherwise)
1174 // ---------------------------------------------------------------------------
1175
1176 // find a character
1177 int wxString::Find(wxChar ch, bool bFromEnd) const
1178 {
1179   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1180
1181   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1182 }
1183
1184 // find a sub-string (like strstr)
1185 int wxString::Find(const wxChar *pszSub) const
1186 {
1187   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1188
1189   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1190 }
1191
1192 // ----------------------------------------------------------------------------
1193 // conversion to numbers
1194 // ----------------------------------------------------------------------------
1195
1196 bool wxString::ToLong(long *val, int base) const
1197 {
1198     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1199     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1200
1201     const wxChar *start = c_str();
1202     wxChar *end;
1203     *val = wxStrtol(start, &end, base);
1204
1205     // return TRUE only if scan was stopped by the terminating NUL and if the
1206     // string was not empty to start with
1207     return !*end && (end != start);
1208 }
1209
1210 bool wxString::ToULong(unsigned long *val, int base) const
1211 {
1212     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1213     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1214
1215     const wxChar *start = c_str();
1216     wxChar *end;
1217     *val = wxStrtoul(start, &end, base);
1218
1219     // return TRUE only if scan was stopped by the terminating NUL and if the
1220     // string was not empty to start with
1221     return !*end && (end != start);
1222 }
1223
1224 bool wxString::ToDouble(double *val) const
1225 {
1226     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1227
1228     const wxChar *start = c_str();
1229     wxChar *end;
1230     *val = wxStrtod(start, &end);
1231
1232     // return TRUE only if scan was stopped by the terminating NUL and if the
1233     // string was not empty to start with
1234     return !*end && (end != start);
1235 }
1236
1237 // ---------------------------------------------------------------------------
1238 // formatted output
1239 // ---------------------------------------------------------------------------
1240
1241 /* static */
1242 wxString wxString::Format(const wxChar *pszFormat, ...)
1243 {
1244     va_list argptr;
1245     va_start(argptr, pszFormat);
1246
1247     wxString s;
1248     s.PrintfV(pszFormat, argptr);
1249
1250     va_end(argptr);
1251
1252     return s;
1253 }
1254
1255 /* static */
1256 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1257 {
1258     wxString s;
1259     s.PrintfV(pszFormat, argptr);
1260     return s;
1261 }
1262
1263 int wxString::Printf(const wxChar *pszFormat, ...)
1264 {
1265   va_list argptr;
1266   va_start(argptr, pszFormat);
1267
1268   int iLen = PrintfV(pszFormat, argptr);
1269
1270   va_end(argptr);
1271
1272   return iLen;
1273 }
1274
1275 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1276 {
1277     int size = 1024;
1278     for ( ;; )
1279     {
1280         wxChar *buf = GetWriteBuf(size + 1);
1281         if ( !buf )
1282         {
1283             // out of memory
1284             return -1;
1285         }
1286
1287         int len = wxVsnprintf(buf, size, pszFormat, argptr);
1288
1289         // some implementations of vsnprintf() don't NUL terminate the string
1290         // if there is not enough space for it so always do it manually
1291         buf[size] = _T('\0');
1292
1293         UngetWriteBuf();
1294
1295         if ( len >= 0 )
1296         {
1297             // ok, there was enough space
1298             break;
1299         }
1300
1301         // still not enough, double it again
1302         size *= 2;
1303     }
1304
1305     // we could have overshot
1306     Shrink();
1307
1308     return Len();
1309 }
1310
1311 // ----------------------------------------------------------------------------
1312 // misc other operations
1313 // ----------------------------------------------------------------------------
1314
1315 // returns TRUE if the string matches the pattern which may contain '*' and
1316 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1317 // of them)
1318 bool wxString::Matches(const wxChar *pszMask) const
1319 {
1320     // I disable this code as it doesn't seem to be faster (in fact, it seems
1321     // to be much slower) than the old, hand-written code below and using it
1322     // here requires always linking with libregex even if the user code doesn't
1323     // use it
1324 #if 0 // wxUSE_REGEX
1325     // first translate the shell-like mask into a regex
1326     wxString pattern;
1327     pattern.reserve(wxStrlen(pszMask));
1328
1329     pattern += _T('^');
1330     while ( *pszMask )
1331     {
1332         switch ( *pszMask )
1333         {
1334             case _T('?'):
1335                 pattern += _T('.');
1336                 break;
1337
1338             case _T('*'):
1339                 pattern += _T(".*");
1340                 break;
1341
1342             case _T('^'):
1343             case _T('.'):
1344             case _T('$'):
1345             case _T('('):
1346             case _T(')'):
1347             case _T('|'):
1348             case _T('+'):
1349             case _T('\\'):
1350                 // these characters are special in a RE, quote them
1351                 // (however note that we don't quote '[' and ']' to allow
1352                 // using them for Unix shell like matching)
1353                 pattern += _T('\\');
1354                 // fall through
1355
1356             default:
1357                 pattern += *pszMask;
1358         }
1359
1360         pszMask++;
1361     }
1362     pattern += _T('$');
1363
1364     // and now use it
1365     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1366 #else // !wxUSE_REGEX
1367   // TODO: this is, of course, awfully inefficient...
1368
1369   // the char currently being checked
1370   const wxChar *pszTxt = c_str();
1371
1372   // the last location where '*' matched
1373   const wxChar *pszLastStarInText = NULL;
1374   const wxChar *pszLastStarInMask = NULL;
1375
1376 match:
1377   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1378     switch ( *pszMask ) {
1379       case wxT('?'):
1380         if ( *pszTxt == wxT('\0') )
1381           return FALSE;
1382
1383         // pszTxt and pszMask will be incremented in the loop statement
1384
1385         break;
1386
1387       case wxT('*'):
1388         {
1389           // remember where we started to be able to backtrack later
1390           pszLastStarInText = pszTxt;
1391           pszLastStarInMask = pszMask;
1392
1393           // ignore special chars immediately following this one
1394           // (should this be an error?)
1395           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1396             pszMask++;
1397
1398           // if there is nothing more, match
1399           if ( *pszMask == wxT('\0') )
1400             return TRUE;
1401
1402           // are there any other metacharacters in the mask?
1403           size_t uiLenMask;
1404           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1405
1406           if ( pEndMask != NULL ) {
1407             // we have to match the string between two metachars
1408             uiLenMask = pEndMask - pszMask;
1409           }
1410           else {
1411             // we have to match the remainder of the string
1412             uiLenMask = wxStrlen(pszMask);
1413           }
1414
1415           wxString strToMatch(pszMask, uiLenMask);
1416           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1417           if ( pMatch == NULL )
1418             return FALSE;
1419
1420           // -1 to compensate "++" in the loop
1421           pszTxt = pMatch + uiLenMask - 1;
1422           pszMask += uiLenMask - 1;
1423         }
1424         break;
1425
1426       default:
1427         if ( *pszMask != *pszTxt )
1428           return FALSE;
1429         break;
1430     }
1431   }
1432
1433   // match only if nothing left
1434   if ( *pszTxt == wxT('\0') )
1435     return TRUE;
1436
1437   // if we failed to match, backtrack if we can
1438   if ( pszLastStarInText ) {
1439     pszTxt = pszLastStarInText + 1;
1440     pszMask = pszLastStarInMask;
1441
1442     pszLastStarInText = NULL;
1443
1444     // don't bother resetting pszLastStarInMask, it's unnecessary
1445
1446     goto match;
1447   }
1448
1449   return FALSE;
1450 #endif // wxUSE_REGEX/!wxUSE_REGEX
1451 }
1452
1453 // Count the number of chars
1454 int wxString::Freq(wxChar ch) const
1455 {
1456     int count = 0;
1457     int len = Len();
1458     for (int i = 0; i < len; i++)
1459     {
1460         if (GetChar(i) == ch)
1461             count ++;
1462     }
1463     return count;
1464 }
1465
1466 // convert to upper case, return the copy of the string
1467 wxString wxString::Upper() const
1468 { wxString s(*this); return s.MakeUpper(); }
1469
1470 // convert to lower case, return the copy of the string
1471 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1472
1473 int wxString::sprintf(const wxChar *pszFormat, ...)
1474   {
1475     va_list argptr;
1476     va_start(argptr, pszFormat);
1477     int iLen = PrintfV(pszFormat, argptr);
1478     va_end(argptr);
1479     return iLen;
1480   }
1481
1482 // ---------------------------------------------------------------------------
1483 // standard C++ library string functions
1484 // ---------------------------------------------------------------------------
1485
1486 #ifdef  wxSTD_STRING_COMPATIBILITY
1487
1488 void wxString::resize(size_t nSize, wxChar ch)
1489 {
1490     size_t len = length();
1491
1492     if ( nSize < len )
1493     {
1494         Truncate(nSize);
1495     }
1496     else if ( nSize > len )
1497     {
1498         *this += wxString(ch, nSize - len);
1499     }
1500     //else: we have exactly the specified length, nothing to do
1501 }
1502
1503 void wxString::swap(wxString& str)
1504 {
1505     // this is slightly less efficient than fiddling with m_pchData directly,
1506     // but it is still quite efficient as we don't copy the string here because
1507     // ref count always stays positive
1508     wxString tmp = str;
1509     str = *this;
1510     *this = tmp;
1511 }
1512
1513 wxString& wxString::insert(size_t nPos, const wxString& str)
1514 {
1515   wxASSERT( str.GetStringData()->IsValid() );
1516   wxASSERT( nPos <= Len() );
1517
1518   if ( !str.IsEmpty() ) {
1519     wxString strTmp;
1520     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1521     wxStrncpy(pc, c_str(), nPos);
1522     wxStrcpy(pc + nPos, str);
1523     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1524     strTmp.UngetWriteBuf();
1525     *this = strTmp;
1526   }
1527
1528   return *this;
1529 }
1530
1531 size_t wxString::find(const wxString& str, size_t nStart) const
1532 {
1533   wxASSERT( str.GetStringData()->IsValid() );
1534   wxASSERT( nStart <= Len() );
1535
1536   const wxChar *p = wxStrstr(c_str() + nStart, str);
1537
1538   return p == NULL ? npos : p - c_str();
1539 }
1540
1541 // VC++ 1.5 can't cope with the default argument in the header.
1542 #if !defined(__VISUALC__) || defined(__WIN32__)
1543 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1544 {
1545   return find(wxString(sz, n), nStart);
1546 }
1547 #endif // VC++ 1.5
1548
1549 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1550 #if !defined(__BORLANDC__)
1551 size_t wxString::find(wxChar ch, size_t nStart) const
1552 {
1553   wxASSERT( nStart <= Len() );
1554
1555   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1556
1557   return p == NULL ? npos : p - c_str();
1558 }
1559 #endif
1560
1561 size_t wxString::rfind(const wxString& str, size_t nStart) const
1562 {
1563   wxASSERT( str.GetStringData()->IsValid() );
1564   wxASSERT( nStart == npos || nStart <= Len() );
1565
1566   // TODO could be made much quicker than that
1567   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1568   while ( p >= c_str() + str.Len() ) {
1569     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1570       return p - str.Len() - c_str();
1571     p--;
1572   }
1573
1574   return npos;
1575 }
1576
1577 // VC++ 1.5 can't cope with the default argument in the header.
1578 #if !defined(__VISUALC__) || defined(__WIN32__)
1579 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1580 {
1581     return rfind(wxString(sz, n == npos ? wxSTRING_MAXLEN : n), nStart);
1582 }
1583
1584 size_t wxString::rfind(wxChar ch, size_t nStart) const
1585 {
1586     if ( nStart == npos )
1587     {
1588         nStart = Len();
1589     }
1590     else
1591     {
1592         wxASSERT( nStart <= Len() );
1593     }
1594
1595     const wxChar *p = wxStrrchr(c_str(), ch);
1596
1597     if ( p == NULL )
1598         return npos;
1599
1600     size_t result = p - c_str();
1601     return ( result > nStart ) ? npos : result;
1602 }
1603 #endif // VC++ 1.5
1604
1605 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1606 {
1607     const wxChar *start = c_str() + nStart;
1608     const wxChar *firstOf = wxStrpbrk(start, sz);
1609     if ( firstOf )
1610         return firstOf - c_str();
1611     else
1612         return npos;
1613 }
1614
1615 size_t wxString::find_last_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     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1627     {
1628         if ( wxStrchr(sz, *p) )
1629             return p - c_str();
1630     }
1631
1632     return npos;
1633 }
1634
1635 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1636 {
1637     if ( nStart == npos )
1638     {
1639         nStart = Len();
1640     }
1641     else
1642     {
1643         wxASSERT( nStart <= Len() );
1644     }
1645
1646     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1647     if ( nAccept >= length() - nStart )
1648         return npos;
1649     else
1650         return nAccept;
1651 }
1652
1653 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1654 {
1655     wxASSERT( nStart <= Len() );
1656
1657     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1658     {
1659         if ( *p != ch )
1660             return p - c_str();
1661     }
1662
1663     return npos;
1664 }
1665
1666 size_t wxString::find_last_not_of(const wxChar* sz, 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 ( !wxStrchr(sz, *p) )
1680             return p - c_str();
1681     }
1682
1683     return npos;
1684 }
1685
1686 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1687 {
1688     if ( nStart == npos )
1689     {
1690         nStart = Len();
1691     }
1692     else
1693     {
1694         wxASSERT( nStart <= Len() );
1695     }
1696
1697     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1698     {
1699         if ( *p != ch )
1700             return p - c_str();
1701     }
1702
1703     return npos;
1704 }
1705
1706 wxString& wxString::erase(size_t nStart, size_t nLen)
1707 {
1708   wxString strTmp(c_str(), nStart);
1709   if ( nLen != npos ) {
1710     wxASSERT( nStart + nLen <= Len() );
1711
1712     strTmp.append(c_str() + nStart + nLen);
1713   }
1714
1715   *this = strTmp;
1716   return *this;
1717 }
1718
1719 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1720 {
1721   wxASSERT_MSG( nStart + nLen <= Len(),
1722                 _T("index out of bounds in wxString::replace") );
1723
1724   wxString strTmp;
1725   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1726
1727   if ( nStart != 0 )
1728     strTmp.append(c_str(), nStart);
1729   strTmp << sz << c_str() + nStart + nLen;
1730
1731   *this = strTmp;
1732   return *this;
1733 }
1734
1735 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1736 {
1737   return replace(nStart, nLen, wxString(ch, nCount));
1738 }
1739
1740 wxString& wxString::replace(size_t nStart, size_t nLen,
1741                             const wxString& str, size_t nStart2, size_t nLen2)
1742 {
1743   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1744 }
1745
1746 wxString& wxString::replace(size_t nStart, size_t nLen,
1747                         const wxChar* sz, size_t nCount)
1748 {
1749   return replace(nStart, nLen, wxString(sz, nCount));
1750 }
1751
1752 #endif  //std::string compatibility
1753
1754 // ============================================================================
1755 // ArrayString
1756 // ============================================================================
1757
1758 #if !wxUSE_STL
1759
1760 #include "wx/arrstr.h"
1761
1762 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1763 #define   ARRAY_MAXSIZE_INCREMENT       4096
1764
1765 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1766 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1767 #endif
1768
1769 #define   STRING(p)   ((wxString *)(&(p)))
1770
1771 // ctor
1772 void wxArrayString::Init(bool autoSort)
1773 {
1774   m_nSize  =
1775   m_nCount = 0;
1776   m_pItems = (wxChar **) NULL;
1777   m_autoSort = autoSort;
1778 }
1779
1780 // copy ctor
1781 wxArrayString::wxArrayString(const wxArrayString& src)
1782 {
1783   Init(src.m_autoSort);
1784
1785   *this = src;
1786 }
1787
1788 // assignment operator
1789 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1790 {
1791   if ( m_nSize > 0 )
1792     Clear();
1793
1794   Copy(src);
1795
1796   m_autoSort = src.m_autoSort;
1797
1798   return *this;
1799 }
1800
1801 void wxArrayString::Copy(const wxArrayString& src)
1802 {
1803   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1804     Alloc(src.m_nCount);
1805
1806   for ( size_t n = 0; n < src.m_nCount; n++ )
1807     Add(src[n]);
1808 }
1809
1810 // grow the array
1811 void wxArrayString::Grow(size_t nIncrement)
1812 {
1813   // only do it if no more place
1814   if ( (m_nSize - m_nCount) < nIncrement ) {
1815     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1816     // be never resized!
1817     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1818       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1819     #endif
1820
1821     if ( m_nSize == 0 ) {
1822       // was empty, alloc some memory
1823       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1824       if (m_nSize < nIncrement)
1825           m_nSize = nIncrement;
1826       m_pItems = new wxChar *[m_nSize];
1827     }
1828     else {
1829       // otherwise when it's called for the first time, nIncrement would be 0
1830       // and the array would never be expanded
1831       // add 50% but not too much
1832       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1833                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1834       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1835         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1836       if ( nIncrement < ndefIncrement )
1837         nIncrement = ndefIncrement;
1838       m_nSize += nIncrement;
1839       wxChar **pNew = new wxChar *[m_nSize];
1840
1841       // copy data to new location
1842       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1843
1844       // delete old memory (but do not release the strings!)
1845       wxDELETEA(m_pItems);
1846
1847       m_pItems = pNew;
1848     }
1849   }
1850 }
1851
1852 void wxArrayString::Free()
1853 {
1854   for ( size_t n = 0; n < m_nCount; n++ ) {
1855     STRING(m_pItems[n])->GetStringData()->Unlock();
1856   }
1857 }
1858
1859 // deletes all the strings from the list
1860 void wxArrayString::Empty()
1861 {
1862   Free();
1863
1864   m_nCount = 0;
1865 }
1866
1867 // as Empty, but also frees memory
1868 void wxArrayString::Clear()
1869 {
1870   Free();
1871
1872   m_nSize  =
1873   m_nCount = 0;
1874
1875   wxDELETEA(m_pItems);
1876 }
1877
1878 // dtor
1879 wxArrayString::~wxArrayString()
1880 {
1881   Free();
1882
1883   wxDELETEA(m_pItems);
1884 }
1885
1886 // pre-allocates memory (frees the previous data!)
1887 void wxArrayString::Alloc(size_t nSize)
1888 {
1889   // only if old buffer was not big enough
1890   if ( nSize > m_nSize ) {
1891     Free();
1892     wxDELETEA(m_pItems);
1893     m_pItems = new wxChar *[nSize];
1894     m_nSize  = nSize;
1895   }
1896
1897   m_nCount = 0;
1898 }
1899
1900 // minimizes the memory usage by freeing unused memory
1901 void wxArrayString::Shrink()
1902 {
1903   // only do it if we have some memory to free
1904   if( m_nCount < m_nSize ) {
1905     // allocates exactly as much memory as we need
1906     wxChar **pNew = new wxChar *[m_nCount];
1907
1908     // copy data to new location
1909     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1910     delete [] m_pItems;
1911     m_pItems = pNew;
1912   }
1913 }
1914
1915 #if WXWIN_COMPATIBILITY_2_4
1916
1917 // return a wxString[] as required for some control ctors.
1918 wxString* wxArrayString::GetStringArray() const
1919 {
1920     wxString *array = 0;
1921
1922     if( m_nCount > 0 )
1923     {
1924         array = new wxString[m_nCount];
1925         for( size_t i = 0; i < m_nCount; i++ )
1926             array[i] = m_pItems[i];
1927     }
1928
1929     return array;
1930 }
1931
1932 #endif // WXWIN_COMPATIBILITY_2_4
1933
1934 // searches the array for an item (forward or backwards)
1935 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1936 {
1937   if ( m_autoSort ) {
1938     // use binary search in the sorted array
1939     wxASSERT_MSG( bCase && !bFromEnd,
1940                   wxT("search parameters ignored for auto sorted array") );
1941
1942     size_t i,
1943            lo = 0,
1944            hi = m_nCount;
1945     int res;
1946     while ( lo < hi ) {
1947       i = (lo + hi)/2;
1948
1949       res = wxStrcmp(sz, m_pItems[i]);
1950       if ( res < 0 )
1951         hi = i;
1952       else if ( res > 0 )
1953         lo = i + 1;
1954       else
1955         return i;
1956     }
1957
1958     return wxNOT_FOUND;
1959   }
1960   else {
1961     // use linear search in unsorted array
1962     if ( bFromEnd ) {
1963       if ( m_nCount > 0 ) {
1964         size_t ui = m_nCount;
1965         do {
1966           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1967             return ui;
1968         }
1969         while ( ui != 0 );
1970       }
1971     }
1972     else {
1973       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1974         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1975           return ui;
1976       }
1977     }
1978   }
1979
1980   return wxNOT_FOUND;
1981 }
1982
1983 // add item at the end
1984 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
1985 {
1986   if ( m_autoSort ) {
1987     // insert the string at the correct position to keep the array sorted
1988     size_t i,
1989            lo = 0,
1990            hi = m_nCount;
1991     int res;
1992     while ( lo < hi ) {
1993       i = (lo + hi)/2;
1994
1995       res = wxStrcmp(str, m_pItems[i]);
1996       if ( res < 0 )
1997         hi = i;
1998       else if ( res > 0 )
1999         lo = i + 1;
2000       else {
2001         lo = hi = i;
2002         break;
2003       }
2004     }
2005
2006     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2007
2008     Insert(str, lo, nInsert);
2009
2010     return (size_t)lo;
2011   }
2012   else {
2013     wxASSERT( str.GetStringData()->IsValid() );
2014
2015     Grow(nInsert);
2016
2017     for (size_t i = 0; i < nInsert; i++)
2018     {
2019         // the string data must not be deleted!
2020         str.GetStringData()->Lock();
2021
2022         // just append
2023         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2024     }
2025     size_t ret = m_nCount;
2026     m_nCount += nInsert;
2027     return ret;
2028   }
2029 }
2030
2031 // add item at the given position
2032 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2033 {
2034   wxASSERT( str.GetStringData()->IsValid() );
2035
2036   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2037   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2038                wxT("array size overflow in wxArrayString::Insert") );
2039
2040   Grow(nInsert);
2041
2042   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2043           (m_nCount - nIndex)*sizeof(wxChar *));
2044
2045   for (size_t i = 0; i < nInsert; i++)
2046   {
2047       str.GetStringData()->Lock();
2048       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2049   }
2050   m_nCount += nInsert;
2051 }
2052
2053 // expand the array
2054 void wxArrayString::SetCount(size_t count)
2055 {
2056     Alloc(count);
2057
2058     wxString s;
2059     while ( m_nCount < count )
2060         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2061 }
2062
2063 // removes item from array (by index)
2064 void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove)
2065 {
2066   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2067   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2068                wxT("removing too many elements in wxArrayString::Remove") );
2069
2070   // release our lock
2071   for (size_t i = 0; i < nRemove; i++)
2072       Item(nIndex + i).GetStringData()->Unlock();
2073
2074   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2075           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2076   m_nCount -= nRemove;
2077 }
2078
2079 // removes item from array (by value)
2080 void wxArrayString::Remove(const wxChar *sz)
2081 {
2082   int iIndex = Index(sz);
2083
2084   wxCHECK_RET( iIndex != wxNOT_FOUND,
2085                wxT("removing inexistent element in wxArrayString::Remove") );
2086
2087   RemoveAt(iIndex);
2088 }
2089
2090 // ----------------------------------------------------------------------------
2091 // sorting
2092 // ----------------------------------------------------------------------------
2093
2094 // we can only sort one array at a time with the quick-sort based
2095 // implementation
2096 #if wxUSE_THREADS
2097   // need a critical section to protect access to gs_compareFunction and
2098   // gs_sortAscending variables
2099   static wxCriticalSection *gs_critsectStringSort = NULL;
2100
2101   // call this before the value of the global sort vars is changed/after
2102   // you're finished with them
2103   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2104                            gs_critsectStringSort = new wxCriticalSection;     \
2105                            gs_critsectStringSort->Enter()
2106   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2107                            delete gs_critsectStringSort;                      \
2108                            gs_critsectStringSort = NULL
2109 #else // !threads
2110   #define START_SORT()
2111   #define END_SORT()
2112 #endif // wxUSE_THREADS
2113
2114 // function to use for string comparaison
2115 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2116
2117 // if we don't use the compare function, this flag tells us if we sort the
2118 // array in ascending or descending order
2119 static bool gs_sortAscending = TRUE;
2120
2121 // function which is called by quick sort
2122 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2123 wxStringCompareFunction(const void *first, const void *second)
2124 {
2125   wxString *strFirst = (wxString *)first;
2126   wxString *strSecond = (wxString *)second;
2127
2128   if ( gs_compareFunction ) {
2129     return gs_compareFunction(*strFirst, *strSecond);
2130   }
2131   else {
2132     // maybe we should use wxStrcoll
2133     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2134
2135     return gs_sortAscending ? result : -result;
2136   }
2137 }
2138
2139 // sort array elements using passed comparaison function
2140 void wxArrayString::Sort(CompareFunction compareFunction)
2141 {
2142   START_SORT();
2143
2144   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2145   gs_compareFunction = compareFunction;
2146
2147   DoSort();
2148
2149   // reset it to NULL so that Sort(bool) will work the next time
2150   gs_compareFunction = NULL;
2151
2152   END_SORT();
2153 }
2154
2155 typedef  int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first, const void *second);
2156
2157 void wxArrayString::Sort(CompareFunction2 compareFunction)
2158 {
2159   qsort(m_pItems, m_nCount, sizeof(wxChar *), (wxStringCompareFn)compareFunction);
2160 }
2161
2162 void wxArrayString::Sort(bool reverseOrder)
2163 {
2164   Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending);
2165 }
2166
2167 void wxArrayString::DoSort()
2168 {
2169   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2170
2171   // just sort the pointers using qsort() - of course it only works because
2172   // wxString() *is* a pointer to its data
2173   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2174 }
2175
2176 bool wxArrayString::operator==(const wxArrayString& a) const
2177 {
2178     if ( m_nCount != a.m_nCount )
2179         return FALSE;
2180
2181     for ( size_t n = 0; n < m_nCount; n++ )
2182     {
2183         if ( Item(n) != a[n] )
2184             return FALSE;
2185     }
2186
2187     return TRUE;
2188 }
2189
2190 #endif // !wxUSE_STL
2191
2192 int wxStringSortAscending(wxString* s1, wxString* s2)
2193 {
2194     return wxStrcmp(s1->c_str(), s2->c_str());
2195 }
2196
2197 int wxStringSortDescending(wxString* s1, wxString* s2)
2198 {
2199     return -wxStrcmp(s1->c_str(), s2->c_str());
2200 }