src/common/string.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        string.cpp
   3 // Purpose:     wxString class
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows licence
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifdef __GNUG__
  13   #pragma implementation "string.h"
  14 #endif
  15
  16 /*
  17  * About ref counting:
  18  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  19  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  20  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  21  */
  22
  23 // ===========================================================================
  24 // headers, declarations, constants
  25 // ===========================================================================
  26
  27 // For compilers that support precompilation, includes "wx.h".
  28 #include "wx/wxprec.h"
  29
  30 #ifdef __BORLANDC__
  31   #pragma hdrstop
  32 #endif
  33
  34 #ifndef WX_PRECOMP
  35   #include "wx/defs.h"
  36   #include "wx/string.h"
  37   #include "wx/intl.h"
  38   #include "wx/thread.h"
  39 #endif
  40
  41 #include <ctype.h>
  42 #include <string.h>
  43 #include <stdlib.h>
  44
  45 #ifdef __SALFORDC__
  46   #include <clib.h>
  47 #endif
  48
  49 // allocating extra space for each string consumes more memory but speeds up
  50 // the concatenation operations (nLen is the current string's length)
  51 // NB: EXTRA_ALLOC must be >= 0!
  52 #define EXTRA_ALLOC       (19 - nLen % 16)
  53
  54 // ---------------------------------------------------------------------------
  55 // static class variables definition
  56 // ---------------------------------------------------------------------------
  57
  58 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  59 // must define this static for VA or else you get multiply defined symbols
  60 // everywhere
  61 const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100;
  62 #endif // Visual Age
  63
  64 #ifdef  wxSTD_STRING_COMPATIBILITY
  65   const size_t wxString::npos = wxSTRING_MAXLEN;
  66 #endif // wxSTD_STRING_COMPATIBILITY
  67
  68 // ----------------------------------------------------------------------------
  69 // static data
  70 // ----------------------------------------------------------------------------
  71
  72 // for an empty string, GetStringData() will return this address: this
  73 // structure has the same layout as wxStringData and it's data() method will
  74 // return the empty string (dummy pointer)
  75 static const struct
  76 {
  77   wxStringData data;
  78   wxChar dummy;
  79 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  80
  81 // empty C style string: points to 'string data' byte of g_strEmpty
  82 extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy;
  83
  84 // ----------------------------------------------------------------------------
  85 // global functions
  86 // ----------------------------------------------------------------------------
  87
  88 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
  89
  90 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
  91 // iostream ones.
  92 //
  93 // ATTN: you can _not_ use both of these in the same program!
  94
  95 wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str))
  96 {
  97 #if 0
  98   int w = is.width(0);
  99   if ( is.ipfx(0) ) {
 100     streambuf *sb = is.rdbuf();
 101     str.erase();
 102     while ( true ) {
 103       int ch = sb->sbumpc ();
 104       if ( ch == EOF ) {
 105         is.setstate(ios::eofbit);
 106         break;
 107       }
 108       else if ( isspace(ch) ) {
 109         sb->sungetc();
 110         break;
 111       }
 112
 113       str += ch;
 114       if ( --w == 1 )
 115         break;
 116     }
 117   }
 118
 119   is.isfx();
 120   if ( str.length() == 0 )
 121     is.setstate(ios::failbit);
 122 #endif
 123   return is;
 124 }
 125
 126 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
 127 {
 128   os << str.c_str();
 129   return os;
 130 }
 131
 132 #endif  //std::string compatibility
 133
 134 // ----------------------------------------------------------------------------
 135 // private classes
 136 // ----------------------------------------------------------------------------
 137
 138 // this small class is used to gather statistics for performance tuning
 139 //#define WXSTRING_STATISTICS
 140 #ifdef  WXSTRING_STATISTICS
 141   class Averager
 142   {
 143   public:
 144     Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 145    ~Averager()
 146    { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 147
 148     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 149
 150   private:
 151     size_t m_nCount, m_nTotal;
 152     const wxChar *m_sz;
 153   } g_averageLength("allocation size"),
 154     g_averageSummandLength("summand length"),
 155     g_averageConcatHit("hit probability in concat"),
 156     g_averageInitialLength("initial string length");
 157
 158   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 159 #else
 160   #define STATISTICS_ADD(av, val)
 161 #endif // WXSTRING_STATISTICS
 162
 163 // ===========================================================================
 164 // wxStringData class deallocation
 165 // ===========================================================================
 166
 167 void wxStringData::Free()
 168 {
 169     free(this);
 170 }
 171
 172 // ===========================================================================
 173 // wxString class core
 174 // ===========================================================================
 175
 176 // ---------------------------------------------------------------------------
 177 // construction
 178 // ---------------------------------------------------------------------------
 179
 180 // constructs string of <nLength> copies of character <ch>
 181 wxString::wxString(wxChar ch, size_t nLength)
 182 {
 183   Init();
 184
 185   if ( nLength > 0 ) {
 186     if ( !AllocBuffer(nLength) ) {
 187       wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 188       return;
 189     }
 190
 191 #if wxUSE_UNICODE
 192     // memset only works on chars
 193     for ( size_t n = 0; n < nLength; n++ )
 194         m_pchData[n] = ch;
 195 #else
 196     memset(m_pchData, ch, nLength);
 197 #endif
 198   }
 199 }
 200
 201 // takes nLength elements of psz starting at nPos
 202 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 203 {
 204   Init();
 205
 206   // if the length is not given, assume the string to be NUL terminated
 207   if ( nLength == wxSTRING_MAXLEN ) {
 208     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 209
 210     nLength = wxStrlen(psz + nPos);
 211   }
 212
 213   STATISTICS_ADD(InitialLength, nLength);
 214
 215   if ( nLength > 0 ) {
 216     // trailing '\0' is written in AllocBuffer()
 217     if ( !AllocBuffer(nLength) ) {
 218       wxFAIL_MSG( _T("out of memory in wxString::InitWith") );
 219       return;
 220     }
 221     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 222   }
 223 }
 224
 225 #ifdef  wxSTD_STRING_COMPATIBILITY
 226
 227 // poor man's iterators are "void *" pointers
 228 wxString::wxString(const void *pStart, const void *pEnd)
 229 {
 230   InitWith((const wxChar *)pStart, 0,
 231            (const wxChar *)pEnd - (const wxChar *)pStart);
 232 }
 233
 234 #endif  //std::string compatibility
 235
 236 #if wxUSE_UNICODE
 237
 238 // from multibyte string
 239 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 240 {
 241     // first get the size of the buffer we need
 242     size_t nLen;
 243     if ( psz )
 244     {
 245         // calculate the needed size ourselves or use a provide one
 246         nLen = nLength == wxSTRING_MAXLEN ? conv.MB2WC(NULL, psz, 0) : nLength;
 247     }
 248     else
 249     {
 250         // nothing to convert
 251         nLen = 0;
 252     }
 253
 254     // anything to do?
 255     if ( (nLen != 0) && (nLen != (size_t)-1) )
 256     {
 257         if ( !AllocBuffer(nLen) )
 258         {
 259             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 260             return;
 261         }
 262
 263         // MB2WC wants the buffer size, not the string length
 264         if ( conv.MB2WC(m_pchData, psz, nLen + 1) != (size_t)-1 )
 265         {
 266             // initialized ok
 267             return;
 268         }
 269         //else: the conversion failed -- leave the string empty (what else?)
 270     }
 271
 272     Init();
 273 }
 274
 275 #else // ANSI
 276
 277 #if wxUSE_WCHAR_T
 278 // from wide string
 279 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 280 {
 281     // first get the size of the buffer we need
 282     size_t nLen;
 283     if ( pwz )
 284     {
 285         // calculate the needed size ourselves or use a provide one
 286         nLen = nLength == wxSTRING_MAXLEN ? conv.WC2MB(NULL, pwz, 0) : nLength;
 287     }
 288     else
 289     {
 290         // nothing to convert
 291         nLen = 0;
 292     }
 293
 294     // anything to do?
 295     if ( (nLen != 0) && (nLen != (size_t)-1) )
 296     {
 297         if ( !AllocBuffer(nLen) )
 298         {
 299             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 300             return;
 301         }
 302
 303         // WC2MB wants the buffer size, not the string length
 304         if ( conv.WC2MB(m_pchData, pwz, nLen + 1) != (size_t)-1 )
 305         {
 306             // initialized ok
 307             return;
 308         }
 309         //else: the conversion failed -- leave the string empty (what else?)
 310     }
 311
 312     Init();
 313 }
 314 #endif // wxUSE_WCHAR_T
 315
 316 #endif // Unicode/ANSI
 317
 318 // ---------------------------------------------------------------------------
 319 // memory allocation
 320 // ---------------------------------------------------------------------------
 321
 322 // allocates memory needed to store a C string of length nLen
 323 bool wxString::AllocBuffer(size_t nLen)
 324 {
 325   // allocating 0 sized buffer doesn't make sense, all empty strings should
 326   // reuse g_strEmpty
 327   wxASSERT( nLen >  0 );
 328
 329   // make sure that we don't overflow
 330   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 331                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 332
 333   STATISTICS_ADD(Length, nLen);
 334
 335   // allocate memory:
 336   // 1) one extra character for '\0' termination
 337   // 2) sizeof(wxStringData) for housekeeping info
 338   wxStringData* pData = (wxStringData*)
 339     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 340
 341   if ( pData == NULL ) {
 342     // allocation failures are handled by the caller
 343     return FALSE;
 344   }
 345
 346   pData->nRefs        = 1;
 347   pData->nDataLength  = nLen;
 348   pData->nAllocLength = nLen + EXTRA_ALLOC;
 349   m_pchData           = pData->data();  // data starts after wxStringData
 350   m_pchData[nLen]     = wxT('\0');
 351   return TRUE;
 352 }
 353
 354 // must be called before changing this string
 355 bool wxString::CopyBeforeWrite()
 356 {
 357   wxStringData* pData = GetStringData();
 358
 359   if ( pData->IsShared() ) {
 360     pData->Unlock();                // memory not freed because shared
 361     size_t nLen = pData->nDataLength;
 362     if ( !AllocBuffer(nLen) ) {
 363       // allocation failures are handled by the caller
 364       return FALSE;
 365     }
 366     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 367   }
 368
 369   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 370
 371   return TRUE;
 372 }
 373
 374 // must be called before replacing contents of this string
 375 bool wxString::AllocBeforeWrite(size_t nLen)
 376 {
 377   wxASSERT( nLen != 0 );  // doesn't make any sense
 378
 379   // must not share string and must have enough space
 380   wxStringData* pData = GetStringData();
 381   if ( pData->IsShared() || pData->IsEmpty() ) {
 382     // can't work with old buffer, get new one
 383     pData->Unlock();
 384     if ( !AllocBuffer(nLen) ) {
 385       // allocation failures are handled by the caller
 386       return FALSE;
 387     }
 388   }
 389   else {
 390     if ( nLen > pData->nAllocLength ) {
 391       // realloc the buffer instead of calling malloc() again, this is more
 392       // efficient
 393       STATISTICS_ADD(Length, nLen);
 394
 395       nLen += EXTRA_ALLOC;
 396
 397       pData = (wxStringData*)
 398           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 399
 400       if ( pData == NULL ) {
 401         // allocation failures are handled by the caller
 402         // keep previous data since reallocation failed
 403         return FALSE;
 404       }
 405
 406       pData->nAllocLength = nLen;
 407       m_pchData = pData->data();
 408     }
 409
 410     // now we have enough space, just update the string length
 411     pData->nDataLength = nLen;
 412   }
 413
 414   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 415
 416   return TRUE;
 417 }
 418
 419 // allocate enough memory for nLen characters
 420 bool wxString::Alloc(size_t nLen)
 421 {
 422   wxStringData *pData = GetStringData();
 423   if ( pData->nAllocLength <= nLen ) {
 424     if ( pData->IsEmpty() ) {
 425       nLen += EXTRA_ALLOC;
 426
 427       wxStringData* pData = (wxStringData*)
 428           malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 429
 430       if ( pData == NULL ) {
 431         // allocation failure handled by caller
 432         return FALSE;
 433       }
 434
 435       pData->nRefs = 1;
 436       pData->nDataLength = 0;
 437       pData->nAllocLength = nLen;
 438       m_pchData = pData->data();  // data starts after wxStringData
 439       m_pchData[0u] = wxT('\0');
 440     }
 441     else if ( pData->IsShared() ) {
 442       pData->Unlock();                // memory not freed because shared
 443       size_t nOldLen = pData->nDataLength;
 444       if ( !AllocBuffer(nLen) ) {
 445         // allocation failure handled by caller
 446         return FALSE;
 447       }
 448       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 449     }
 450     else {
 451       nLen += EXTRA_ALLOC;
 452
 453       pData = (wxStringData *)
 454         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 455
 456       if ( pData == NULL ) {
 457         // allocation failure handled by caller
 458         // keep previous data since reallocation failed
 459         return FALSE;
 460       }
 461
 462       // it's not important if the pointer changed or not (the check for this
 463       // is not faster than assigning to m_pchData in all cases)
 464       pData->nAllocLength = nLen;
 465       m_pchData = pData->data();
 466     }
 467   }
 468   //else: we've already got enough
 469   return TRUE;
 470 }
 471
 472 // shrink to minimal size (releasing extra memory)
 473 bool wxString::Shrink()
 474 {
 475   wxStringData *pData = GetStringData();
 476
 477   size_t nLen = pData->nDataLength;
 478   void *p = realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 479
 480   if ( p == NULL) {
 481       wxFAIL_MSG( _T("out of memory reallocating wxString data") );
 482       // keep previous data since reallocation failed
 483       return FALSE;
 484   }
 485
 486   if ( p != pData )
 487   {
 488       // contrary to what one might believe, some realloc() implementation do
 489       // move the memory block even when its size is reduced
 490       pData = (wxStringData *)p;
 491
 492       m_pchData = pData->data();
 493   }
 494
 495   pData->nAllocLength = nLen;
 496
 497   return TRUE;
 498 }
 499
 500 // get the pointer to writable buffer of (at least) nLen bytes
 501 wxChar *wxString::GetWriteBuf(size_t nLen)
 502 {
 503   if ( !AllocBeforeWrite(nLen) ) {
 504     // allocation failure handled by caller
 505     return NULL;
 506   }
 507
 508   wxASSERT( GetStringData()->nRefs == 1 );
 509   GetStringData()->Validate(FALSE);
 510
 511   return m_pchData;
 512 }
 513
 514 // put string back in a reasonable state after GetWriteBuf
 515 void wxString::UngetWriteBuf()
 516 {
 517   GetStringData()->nDataLength = wxStrlen(m_pchData);
 518   GetStringData()->Validate(TRUE);
 519 }
 520
 521 void wxString::UngetWriteBuf(size_t nLen)
 522 {
 523   GetStringData()->nDataLength = nLen;
 524   GetStringData()->Validate(TRUE);
 525 }
 526
 527 // ---------------------------------------------------------------------------
 528 // data access
 529 // ---------------------------------------------------------------------------
 530
 531 // all functions are inline in string.h
 532
 533 // ---------------------------------------------------------------------------
 534 // assignment operators
 535 // ---------------------------------------------------------------------------
 536
 537 // helper function: does real copy
 538 bool wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 539 {
 540   if ( nSrcLen == 0 ) {
 541     Reinit();
 542   }
 543   else {
 544     if ( !AllocBeforeWrite(nSrcLen) ) {
 545       // allocation failure handled by caller
 546       return FALSE;
 547     }
 548     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 549     GetStringData()->nDataLength = nSrcLen;
 550     m_pchData[nSrcLen] = wxT('\0');
 551   }
 552   return TRUE;
 553 }
 554
 555 // assigns one string to another
 556 wxString& wxString::operator=(const wxString& stringSrc)
 557 {
 558   wxASSERT( stringSrc.GetStringData()->IsValid() );
 559
 560   // don't copy string over itself
 561   if ( m_pchData != stringSrc.m_pchData ) {
 562     if ( stringSrc.GetStringData()->IsEmpty() ) {
 563       Reinit();
 564     }
 565     else {
 566       // adjust references
 567       GetStringData()->Unlock();
 568       m_pchData = stringSrc.m_pchData;
 569       GetStringData()->Lock();
 570     }
 571   }
 572
 573   return *this;
 574 }
 575
 576 // assigns a single character
 577 wxString& wxString::operator=(wxChar ch)
 578 {
 579   if ( !AssignCopy(1, &ch) ) {
 580     wxFAIL_MSG( _T("out of memory in wxString::operator=(wxChar)") );
 581   }
 582   return *this;
 583 }
 584
 585
 586 // assigns C string
 587 wxString& wxString::operator=(const wxChar *psz)
 588 {
 589   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 590     wxFAIL_MSG( _T("out of memory in wxString::operator=(const wxChar *)") );
 591   }
 592   return *this;
 593 }
 594
 595 #if !wxUSE_UNICODE
 596
 597 // same as 'signed char' variant
 598 wxString& wxString::operator=(const unsigned char* psz)
 599 {
 600   *this = (const char *)psz;
 601   return *this;
 602 }
 603
 604 #if wxUSE_WCHAR_T
 605 wxString& wxString::operator=(const wchar_t *pwz)
 606 {
 607   wxString str(pwz);
 608   *this = str;
 609   return *this;
 610 }
 611 #endif
 612
 613 #endif
 614
 615 // ---------------------------------------------------------------------------
 616 // string concatenation
 617 // ---------------------------------------------------------------------------
 618
 619 // add something to this string
 620 bool wxString::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData)
 621 {
 622   STATISTICS_ADD(SummandLength, nSrcLen);
 623
 624   // concatenating an empty string is a NOP
 625   if ( nSrcLen > 0 ) {
 626     wxStringData *pData = GetStringData();
 627     size_t nLen = pData->nDataLength;
 628     size_t nNewLen = nLen + nSrcLen;
 629
 630     // alloc new buffer if current is too small
 631     if ( pData->IsShared() ) {
 632       STATISTICS_ADD(ConcatHit, 0);
 633
 634       // we have to allocate another buffer
 635       wxStringData* pOldData = GetStringData();
 636       if ( !AllocBuffer(nNewLen) ) {
 637           // allocation failure handled by caller
 638           return FALSE;
 639       }
 640       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 641       pOldData->Unlock();
 642     }
 643     else if ( nNewLen > pData->nAllocLength ) {
 644       STATISTICS_ADD(ConcatHit, 0);
 645
 646       // we have to grow the buffer
 647       if ( !Alloc(nNewLen) ) {
 648           // allocation failure handled by caller
 649           return FALSE;
 650       }
 651     }
 652     else {
 653       STATISTICS_ADD(ConcatHit, 1);
 654
 655       // the buffer is already big enough
 656     }
 657
 658     // should be enough space
 659     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 660
 661     // fast concatenation - all is done in our buffer
 662     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 663
 664     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 665     GetStringData()->nDataLength = nNewLen; // and fix the length
 666   }
 667   //else: the string to append was empty
 668   return TRUE;
 669 }
 670
 671 /*
 672  * concatenation functions come in 5 flavours:
 673  *  string + string
 674  *  char   + string      and      string + char
 675  *  C str  + string      and      string + C str
 676  */
 677
 678 wxString operator+(const wxString& str1, const wxString& str2)
 679 {
 680   wxASSERT( str1.GetStringData()->IsValid() );
 681   wxASSERT( str2.GetStringData()->IsValid() );
 682
 683   wxString s = str1;
 684   s += str2;
 685
 686   return s;
 687 }
 688
 689 wxString operator+(const wxString& str, wxChar ch)
 690 {
 691   wxASSERT( str.GetStringData()->IsValid() );
 692
 693   wxString s = str;
 694   s += ch;
 695
 696   return s;
 697 }
 698
 699 wxString operator+(wxChar ch, const wxString& str)
 700 {
 701   wxASSERT( str.GetStringData()->IsValid() );
 702
 703   wxString s = ch;
 704   s += str;
 705
 706   return s;
 707 }
 708
 709 wxString operator+(const wxString& str, const wxChar *psz)
 710 {
 711   wxASSERT( str.GetStringData()->IsValid() );
 712
 713   wxString s;
 714   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 715     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 716   }
 717   s = str;
 718   s += psz;
 719
 720   return s;
 721 }
 722
 723 wxString operator+(const wxChar *psz, const wxString& str)
 724 {
 725   wxASSERT( str.GetStringData()->IsValid() );
 726
 727   wxString s;
 728   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 729     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 730   }
 731   s = psz;
 732   s += str;
 733
 734   return s;
 735 }
 736
 737 // ===========================================================================
 738 // other common string functions
 739 // ===========================================================================
 740
 741 #if wxUSE_UNICODE
 742
 743 wxString wxString::FromAscii(const char *ascii)
 744 {
 745     if (!ascii)
 746        return wxEmptyString;
 747
 748     size_t len = strlen( ascii );
 749     wxString res;
 750
 751     if ( len )
 752     {
 753         wxStringBuffer buf(res, len);
 754
 755         wchar_t *dest = buf;
 756
 757         for ( ;; )
 758         {
 759            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
 760                break;
 761         }
 762     }
 763
 764     return res;
 765 }
 766
 767 wxString wxString::FromAscii(const char ascii)
 768 {
 769     // What do we do with '\0' ?
 770
 771     wxString res;
 772     res += (wchar_t)(unsigned char) ascii;
 773
 774     return res;
 775 }
 776
 777 const wxCharBuffer wxString::ToAscii() const
 778 {
 779     // this will allocate enough space for the terminating NUL too
 780     wxCharBuffer buffer(length());
 781
 782     signed char *dest = (signed char *)buffer.data();
 783
 784     const wchar_t *pwc = c_str();
 785     for ( ;; )
 786     {
 787         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
 788
 789         // the output string can't have embedded NULs anyhow, so we can safely
 790         // stop at first of them even if we do have any
 791         if ( !*pwc++ )
 792             break;
 793     }
 794
 795     return buffer;
 796 }
 797
 798 #endif // Unicode
 799
 800 // ---------------------------------------------------------------------------
 801 // simple sub-string extraction
 802 // ---------------------------------------------------------------------------
 803
 804 // helper function: clone the data attached to this string
 805 bool wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 806 {
 807   if ( nCopyLen == 0 ) {
 808     dest.Init();
 809   }
 810   else {
 811     if ( !dest.AllocBuffer(nCopyLen) ) {
 812       // allocation failure handled by caller
 813       return FALSE;
 814     }
 815     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 816   }
 817   return TRUE;
 818 }
 819
 820 // extract string of length nCount starting at nFirst
 821 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 822 {
 823   wxStringData *pData = GetStringData();
 824   size_t nLen = pData->nDataLength;
 825
 826   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 827   if ( nCount == wxSTRING_MAXLEN )
 828   {
 829     nCount = nLen - nFirst;
 830   }
 831
 832   // out-of-bounds requests return sensible things
 833   if ( nFirst + nCount > nLen )
 834   {
 835     nCount = nLen - nFirst;
 836   }
 837
 838   if ( nFirst > nLen )
 839   {
 840     // AllocCopy() will return empty string
 841     nCount = 0;
 842   }
 843
 844   wxString dest;
 845   if ( !AllocCopy(dest, nCount, nFirst) ) {
 846       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
 847   }
 848
 849   return dest;
 850 }
 851
 852 // check that the tring starts with prefix and return the rest of the string
 853 // in the provided pointer if it is not NULL, otherwise return FALSE
 854 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 855 {
 856     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 857
 858     // first check if the beginning of the string matches the prefix: note
 859     // that we don't have to check that we don't run out of this string as
 860     // when we reach the terminating NUL, either prefix string ends too (and
 861     // then it's ok) or we break out of the loop because there is no match
 862     const wxChar *p = c_str();
 863     while ( *prefix )
 864     {
 865         if ( *prefix++ != *p++ )
 866         {
 867             // no match
 868             return FALSE;
 869         }
 870     }
 871
 872     if ( rest )
 873     {
 874         // put the rest of the string into provided pointer
 875         *rest = p;
 876     }
 877
 878     return TRUE;
 879 }
 880
 881 // extract nCount last (rightmost) characters
 882 wxString wxString::Right(size_t nCount) const
 883 {
 884   if ( nCount > (size_t)GetStringData()->nDataLength )
 885     nCount = GetStringData()->nDataLength;
 886
 887   wxString dest;
 888   if ( !AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount) ) {
 889     wxFAIL_MSG( _T("out of memory in wxString::Right") );
 890   }
 891   return dest;
 892 }
 893
 894 // get all characters after the last occurence of ch
 895 // (returns the whole string if ch not found)
 896 wxString wxString::AfterLast(wxChar ch) const
 897 {
 898   wxString str;
 899   int iPos = Find(ch, TRUE);
 900   if ( iPos == wxNOT_FOUND )
 901     str = *this;
 902   else
 903     str = c_str() + iPos + 1;
 904
 905   return str;
 906 }
 907
 908 // extract nCount first (leftmost) characters
 909 wxString wxString::Left(size_t nCount) const
 910 {
 911   if ( nCount > (size_t)GetStringData()->nDataLength )
 912     nCount = GetStringData()->nDataLength;
 913
 914   wxString dest;
 915   if ( !AllocCopy(dest, nCount, 0) ) {
 916     wxFAIL_MSG( _T("out of memory in wxString::Left") );
 917   }
 918   return dest;
 919 }
 920
 921 // get all characters before the first occurence of ch
 922 // (returns the whole string if ch not found)
 923 wxString wxString::BeforeFirst(wxChar ch) const
 924 {
 925   wxString str;
 926   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 927     str += *pc;
 928
 929   return str;
 930 }
 931
 932 /// get all characters before the last occurence of ch
 933 /// (returns empty string if ch not found)
 934 wxString wxString::BeforeLast(wxChar ch) const
 935 {
 936   wxString str;
 937   int iPos = Find(ch, TRUE);
 938   if ( iPos != wxNOT_FOUND && iPos != 0 )
 939     str = wxString(c_str(), iPos);
 940
 941   return str;
 942 }
 943
 944 /// get all characters after the first occurence of ch
 945 /// (returns empty string if ch not found)
 946 wxString wxString::AfterFirst(wxChar ch) const
 947 {
 948   wxString str;
 949   int iPos = Find(ch);
 950   if ( iPos != wxNOT_FOUND )
 951     str = c_str() + iPos + 1;
 952
 953   return str;
 954 }
 955
 956 // replace first (or all) occurences of some substring with another one
 957 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 958 {
 959   size_t uiCount = 0;   // count of replacements made
 960
 961   size_t uiOldLen = wxStrlen(szOld);
 962
 963   wxString strTemp;
 964   const wxChar *pCurrent = m_pchData;
 965   const wxChar *pSubstr;
 966   while ( *pCurrent != wxT('\0') ) {
 967     pSubstr = wxStrstr(pCurrent, szOld);
 968     if ( pSubstr == NULL ) {
 969       // strTemp is unused if no replacements were made, so avoid the copy
 970       if ( uiCount == 0 )
 971         return 0;
 972
 973       strTemp += pCurrent;    // copy the rest
 974       break;                  // exit the loop
 975     }
 976     else {
 977       // take chars before match
 978       if ( !strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent) ) {
 979         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
 980         return 0;
 981       }
 982       strTemp += szNew;
 983       pCurrent = pSubstr + uiOldLen;  // restart after match
 984
 985       uiCount++;
 986
 987       // stop now?
 988       if ( !bReplaceAll ) {
 989         strTemp += pCurrent;    // copy the rest
 990         break;                  // exit the loop
 991       }
 992     }
 993   }
 994
 995   // only done if there were replacements, otherwise would have returned above
 996   *this = strTemp;
 997
 998   return uiCount;
 999 }
1000
1001 bool wxString::IsAscii() const
1002 {
1003   const wxChar *s = (const wxChar*) *this;
1004   while(*s){
1005     if(!isascii(*s)) return(FALSE);
1006     s++;
1007   }
1008   return(TRUE);
1009 }
1010
1011 bool wxString::IsWord() const
1012 {
1013   const wxChar *s = (const wxChar*) *this;
1014   while(*s){
1015     if(!wxIsalpha(*s)) return(FALSE);
1016     s++;
1017   }
1018   return(TRUE);
1019 }
1020
1021 bool wxString::IsNumber() const
1022 {
1023   const wxChar *s = (const wxChar*) *this;
1024   if (wxStrlen(s))
1025      if ((s[0] == '-') || (s[0] == '+')) s++;
1026   while(*s){
1027     if(!wxIsdigit(*s)) return(FALSE);
1028     s++;
1029   }
1030   return(TRUE);
1031 }
1032
1033 wxString wxString::Strip(stripType w) const
1034 {
1035     wxString s = *this;
1036     if ( w & leading ) s.Trim(FALSE);
1037     if ( w & trailing ) s.Trim(TRUE);
1038     return s;
1039 }
1040
1041 // ---------------------------------------------------------------------------
1042 // case conversion
1043 // ---------------------------------------------------------------------------
1044
1045 wxString& wxString::MakeUpper()
1046 {
1047   if ( !CopyBeforeWrite() ) {
1048     wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") );
1049     return *this;
1050   }
1051
1052   for ( wxChar *p = m_pchData; *p; p++ )
1053     *p = (wxChar)wxToupper(*p);
1054
1055   return *this;
1056 }
1057
1058 wxString& wxString::MakeLower()
1059 {
1060   if ( !CopyBeforeWrite() ) {
1061     wxFAIL_MSG( _T("out of memory in wxString::MakeLower") );
1062     return *this;
1063   }
1064
1065   for ( wxChar *p = m_pchData; *p; p++ )
1066     *p = (wxChar)wxTolower(*p);
1067
1068   return *this;
1069 }
1070
1071 // ---------------------------------------------------------------------------
1072 // trimming and padding
1073 // ---------------------------------------------------------------------------
1074
1075 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1076 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1077 // live with this by checking that the character is a 7 bit one - even if this
1078 // may fail to detect some spaces (I don't know if Unicode doesn't have
1079 // space-like symbols somewhere except in the first 128 chars), it is arguably
1080 // still better than trimming away accented letters
1081 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1082
1083 // trims spaces (in the sense of isspace) from left or right side
1084 wxString& wxString::Trim(bool bFromRight)
1085 {
1086   // first check if we're going to modify the string at all
1087   if ( !IsEmpty() &&
1088        (
1089         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1090         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1091        )
1092      )
1093   {
1094     // ok, there is at least one space to trim
1095     if ( !CopyBeforeWrite() ) {
1096       wxFAIL_MSG( _T("out of memory in wxString::Trim") );
1097       return *this;
1098     }
1099
1100     if ( bFromRight )
1101     {
1102       // find last non-space character
1103       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1104       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1105         psz--;
1106
1107       // truncate at trailing space start
1108       *++psz = wxT('\0');
1109       GetStringData()->nDataLength = psz - m_pchData;
1110     }
1111     else
1112     {
1113       // find first non-space character
1114       const wxChar *psz = m_pchData;
1115       while ( wxSafeIsspace(*psz) )
1116         psz++;
1117
1118       // fix up data and length
1119       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1120       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1121       GetStringData()->nDataLength = nDataLength;
1122     }
1123   }
1124
1125   return *this;
1126 }
1127
1128 // adds nCount characters chPad to the string from either side
1129 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1130 {
1131   wxString s(chPad, nCount);
1132
1133   if ( bFromRight )
1134     *this += s;
1135   else
1136   {
1137     s += *this;
1138     *this = s;
1139   }
1140
1141   return *this;
1142 }
1143
1144 // truncate the string
1145 wxString& wxString::Truncate(size_t uiLen)
1146 {
1147   if ( uiLen < Len() ) {
1148     if ( !CopyBeforeWrite() ) {
1149       wxFAIL_MSG( _T("out of memory in wxString::Truncate") );
1150       return *this;
1151     }
1152
1153     *(m_pchData + uiLen) = wxT('\0');
1154     GetStringData()->nDataLength = uiLen;
1155   }
1156   //else: nothing to do, string is already short enough
1157
1158   return *this;
1159 }
1160
1161 // ---------------------------------------------------------------------------
1162 // finding (return wxNOT_FOUND if not found and index otherwise)
1163 // ---------------------------------------------------------------------------
1164
1165 // find a character
1166 int wxString::Find(wxChar ch, bool bFromEnd) const
1167 {
1168   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1169
1170   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1171 }
1172
1173 // find a sub-string (like strstr)
1174 int wxString::Find(const wxChar *pszSub) const
1175 {
1176   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1177
1178   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1179 }
1180
1181 // ----------------------------------------------------------------------------
1182 // conversion to numbers
1183 // ----------------------------------------------------------------------------
1184
1185 bool wxString::ToLong(long *val, int base) const
1186 {
1187     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1188     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1189
1190     const wxChar *start = c_str();
1191     wxChar *end;
1192     *val = wxStrtol(start, &end, base);
1193
1194     // return TRUE only if scan was stopped by the terminating NUL and if the
1195     // string was not empty to start with
1196     return !*end && (end != start);
1197 }
1198
1199 bool wxString::ToULong(unsigned long *val, int base) const
1200 {
1201     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1202     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1203
1204     const wxChar *start = c_str();
1205     wxChar *end;
1206     *val = wxStrtoul(start, &end, base);
1207
1208     // return TRUE only if scan was stopped by the terminating NUL and if the
1209     // string was not empty to start with
1210     return !*end && (end != start);
1211 }
1212
1213 bool wxString::ToDouble(double *val) const
1214 {
1215     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1216
1217     const wxChar *start = c_str();
1218     wxChar *end;
1219     *val = wxStrtod(start, &end);
1220
1221     // return TRUE only if scan was stopped by the terminating NUL and if the
1222     // string was not empty to start with
1223     return !*end && (end != start);
1224 }
1225
1226 // ---------------------------------------------------------------------------
1227 // formatted output
1228 // ---------------------------------------------------------------------------
1229
1230 /* static */
1231 wxString wxString::Format(const wxChar *pszFormat, ...)
1232 {
1233     va_list argptr;
1234     va_start(argptr, pszFormat);
1235
1236     wxString s;
1237     s.PrintfV(pszFormat, argptr);
1238
1239     va_end(argptr);
1240
1241     return s;
1242 }
1243
1244 /* static */
1245 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1246 {
1247     wxString s;
1248     s.PrintfV(pszFormat, argptr);
1249     return s;
1250 }
1251
1252 int wxString::Printf(const wxChar *pszFormat, ...)
1253 {
1254   va_list argptr;
1255   va_start(argptr, pszFormat);
1256
1257   int iLen = PrintfV(pszFormat, argptr);
1258
1259   va_end(argptr);
1260
1261   return iLen;
1262 }
1263
1264 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1265 {
1266     int size = 1024;
1267     for ( ;; )
1268     {
1269         wxChar *buf = GetWriteBuf(size + 1);
1270         if ( !buf )
1271         {
1272             // out of memory
1273             return -1;
1274         }
1275
1276         int len = wxVsnprintf(buf, size, pszFormat, argptr);
1277
1278         // some implementations of vsnprintf() don't NUL terminate the string
1279         // if there is not enough space for it so always do it manually
1280         buf[size] = _T('\0');
1281
1282         UngetWriteBuf();
1283
1284         if ( len >= 0 )
1285         {
1286             // ok, there was enough space
1287             break;
1288         }
1289
1290         // still not enough, double it again
1291         size *= 2;
1292     }
1293
1294     // we could have overshot
1295     Shrink();
1296
1297     return Len();
1298 }
1299
1300 // ----------------------------------------------------------------------------
1301 // misc other operations
1302 // ----------------------------------------------------------------------------
1303
1304 // returns TRUE if the string matches the pattern which may contain '*' and
1305 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1306 // of them)
1307 bool wxString::Matches(const wxChar *pszMask) const
1308 {
1309     // I disable this code as it doesn't seem to be faster (in fact, it seems
1310     // to be much slower) than the old, hand-written code below and using it
1311     // here requires always linking with libregex even if the user code doesn't
1312     // use it
1313 #if 0 // wxUSE_REGEX
1314     // first translate the shell-like mask into a regex
1315     wxString pattern;
1316     pattern.reserve(wxStrlen(pszMask));
1317
1318     pattern += _T('^');
1319     while ( *pszMask )
1320     {
1321         switch ( *pszMask )
1322         {
1323             case _T('?'):
1324                 pattern += _T('.');
1325                 break;
1326
1327             case _T('*'):
1328                 pattern += _T(".*");
1329                 break;
1330
1331             case _T('^'):
1332             case _T('.'):
1333             case _T('$'):
1334             case _T('('):
1335             case _T(')'):
1336             case _T('|'):
1337             case _T('+'):
1338             case _T('\\'):
1339                 // these characters are special in a RE, quote them
1340                 // (however note that we don't quote '[' and ']' to allow
1341                 // using them for Unix shell like matching)
1342                 pattern += _T('\\');
1343                 // fall through
1344
1345             default:
1346                 pattern += *pszMask;
1347         }
1348
1349         pszMask++;
1350     }
1351     pattern += _T('$');
1352
1353     // and now use it
1354     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1355 #else // !wxUSE_REGEX
1356   // TODO: this is, of course, awfully inefficient...
1357
1358   // the char currently being checked
1359   const wxChar *pszTxt = c_str();
1360
1361   // the last location where '*' matched
1362   const wxChar *pszLastStarInText = NULL;
1363   const wxChar *pszLastStarInMask = NULL;
1364
1365 match:
1366   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1367     switch ( *pszMask ) {
1368       case wxT('?'):
1369         if ( *pszTxt == wxT('\0') )
1370           return FALSE;
1371
1372         // pszTxt and pszMask will be incremented in the loop statement
1373
1374         break;
1375
1376       case wxT('*'):
1377         {
1378           // remember where we started to be able to backtrack later
1379           pszLastStarInText = pszTxt;
1380           pszLastStarInMask = pszMask;
1381
1382           // ignore special chars immediately following this one
1383           // (should this be an error?)
1384           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1385             pszMask++;
1386
1387           // if there is nothing more, match
1388           if ( *pszMask == wxT('\0') )
1389             return TRUE;
1390
1391           // are there any other metacharacters in the mask?
1392           size_t uiLenMask;
1393           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1394
1395           if ( pEndMask != NULL ) {
1396             // we have to match the string between two metachars
1397             uiLenMask = pEndMask - pszMask;
1398           }
1399           else {
1400             // we have to match the remainder of the string
1401             uiLenMask = wxStrlen(pszMask);
1402           }
1403
1404           wxString strToMatch(pszMask, uiLenMask);
1405           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1406           if ( pMatch == NULL )
1407             return FALSE;
1408
1409           // -1 to compensate "++" in the loop
1410           pszTxt = pMatch + uiLenMask - 1;
1411           pszMask += uiLenMask - 1;
1412         }
1413         break;
1414
1415       default:
1416         if ( *pszMask != *pszTxt )
1417           return FALSE;
1418         break;
1419     }
1420   }
1421
1422   // match only if nothing left
1423   if ( *pszTxt == wxT('\0') )
1424     return TRUE;
1425
1426   // if we failed to match, backtrack if we can
1427   if ( pszLastStarInText ) {
1428     pszTxt = pszLastStarInText + 1;
1429     pszMask = pszLastStarInMask;
1430
1431     pszLastStarInText = NULL;
1432
1433     // don't bother resetting pszLastStarInMask, it's unnecessary
1434
1435     goto match;
1436   }
1437
1438   return FALSE;
1439 #endif // wxUSE_REGEX/!wxUSE_REGEX
1440 }
1441
1442 // Count the number of chars
1443 int wxString::Freq(wxChar ch) const
1444 {
1445     int count = 0;
1446     int len = Len();
1447     for (int i = 0; i < len; i++)
1448     {
1449         if (GetChar(i) == ch)
1450             count ++;
1451     }
1452     return count;
1453 }
1454
1455 // convert to upper case, return the copy of the string
1456 wxString wxString::Upper() const
1457 { wxString s(*this); return s.MakeUpper(); }
1458
1459 // convert to lower case, return the copy of the string
1460 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1461
1462 int wxString::sprintf(const wxChar *pszFormat, ...)
1463   {
1464     va_list argptr;
1465     va_start(argptr, pszFormat);
1466     int iLen = PrintfV(pszFormat, argptr);
1467     va_end(argptr);
1468     return iLen;
1469   }
1470
1471 // ---------------------------------------------------------------------------
1472 // standard C++ library string functions
1473 // ---------------------------------------------------------------------------
1474
1475 #ifdef  wxSTD_STRING_COMPATIBILITY
1476
1477 void wxString::resize(size_t nSize, wxChar ch)
1478 {
1479     size_t len = length();
1480
1481     if ( nSize < len )
1482     {
1483         Truncate(nSize);
1484     }
1485     else if ( nSize > len )
1486     {
1487         *this += wxString(ch, nSize - len);
1488     }
1489     //else: we have exactly the specified length, nothing to do
1490 }
1491
1492 void wxString::swap(wxString& str)
1493 {
1494     // this is slightly less efficient than fiddling with m_pchData directly,
1495     // but it is still quite efficient as we don't copy the string here because
1496     // ref count always stays positive
1497     wxString tmp = str;
1498     str = *this;
1499     *this = tmp;
1500 }
1501
1502 wxString& wxString::insert(size_t nPos, const wxString& str)
1503 {
1504   wxASSERT( str.GetStringData()->IsValid() );
1505   wxASSERT( nPos <= Len() );
1506
1507   if ( !str.IsEmpty() ) {
1508     wxString strTmp;
1509     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1510     wxStrncpy(pc, c_str(), nPos);
1511     wxStrcpy(pc + nPos, str);
1512     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1513     strTmp.UngetWriteBuf();
1514     *this = strTmp;
1515   }
1516
1517   return *this;
1518 }
1519
1520 size_t wxString::find(const wxString& str, size_t nStart) const
1521 {
1522   wxASSERT( str.GetStringData()->IsValid() );
1523   wxASSERT( nStart <= Len() );
1524
1525   const wxChar *p = wxStrstr(c_str() + nStart, str);
1526
1527   return p == NULL ? npos : p - c_str();
1528 }
1529
1530 // VC++ 1.5 can't cope with the default argument in the header.
1531 #if !defined(__VISUALC__) || defined(__WIN32__)
1532 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1533 {
1534   return find(wxString(sz, n), nStart);
1535 }
1536 #endif // VC++ 1.5
1537
1538 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1539 #if !defined(__BORLANDC__)
1540 size_t wxString::find(wxChar ch, size_t nStart) const
1541 {
1542   wxASSERT( nStart <= Len() );
1543
1544   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1545
1546   return p == NULL ? npos : p - c_str();
1547 }
1548 #endif
1549
1550 size_t wxString::rfind(const wxString& str, size_t nStart) const
1551 {
1552   wxASSERT( str.GetStringData()->IsValid() );
1553   wxASSERT( nStart == npos || nStart <= Len() );
1554
1555   // TODO could be made much quicker than that
1556   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1557   while ( p >= c_str() + str.Len() ) {
1558     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1559       return p - str.Len() - c_str();
1560     p--;
1561   }
1562
1563   return npos;
1564 }
1565
1566 // VC++ 1.5 can't cope with the default argument in the header.
1567 #if !defined(__VISUALC__) || defined(__WIN32__)
1568 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1569 {
1570     return rfind(wxString(sz, n == npos ? wxSTRING_MAXLEN : n), nStart);
1571 }
1572
1573 size_t wxString::rfind(wxChar ch, size_t nStart) const
1574 {
1575     if ( nStart == npos )
1576     {
1577         nStart = Len();
1578     }
1579     else
1580     {
1581         wxASSERT( nStart <= Len() );
1582     }
1583
1584     const wxChar *p = wxStrrchr(c_str(), ch);
1585
1586     if ( p == NULL )
1587         return npos;
1588
1589     size_t result = p - c_str();
1590     return ( result > nStart ) ? npos : result;
1591 }
1592 #endif // VC++ 1.5
1593
1594 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1595 {
1596     const wxChar *start = c_str() + nStart;
1597     const wxChar *firstOf = wxStrpbrk(start, sz);
1598     if ( firstOf )
1599         return firstOf - c_str();
1600     else
1601         return npos;
1602 }
1603
1604 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1605 {
1606     if ( nStart == npos )
1607     {
1608         nStart = Len();
1609     }
1610     else
1611     {
1612         wxASSERT( nStart <= Len() );
1613     }
1614
1615     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1616     {
1617         if ( wxStrchr(sz, *p) )
1618             return p - c_str();
1619     }
1620
1621     return npos;
1622 }
1623
1624 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1625 {
1626     if ( nStart == npos )
1627     {
1628         nStart = Len();
1629     }
1630     else
1631     {
1632         wxASSERT( nStart <= Len() );
1633     }
1634
1635     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1636     if ( nAccept >= length() - nStart )
1637         return npos;
1638     else
1639         return nAccept;
1640 }
1641
1642 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1643 {
1644     wxASSERT( nStart <= Len() );
1645
1646     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1647     {
1648         if ( *p != ch )
1649             return p - c_str();
1650     }
1651
1652     return npos;
1653 }
1654
1655 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1656 {
1657     if ( nStart == npos )
1658     {
1659         nStart = Len();
1660     }
1661     else
1662     {
1663         wxASSERT( nStart <= Len() );
1664     }
1665
1666     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1667     {
1668         if ( !wxStrchr(sz, *p) )
1669             return p - c_str();
1670     }
1671
1672     return npos;
1673 }
1674
1675 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1676 {
1677     if ( nStart == npos )
1678     {
1679         nStart = Len();
1680     }
1681     else
1682     {
1683         wxASSERT( nStart <= Len() );
1684     }
1685
1686     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1687     {
1688         if ( *p != ch )
1689             return p - c_str();
1690     }
1691
1692     return npos;
1693 }
1694
1695 wxString& wxString::erase(size_t nStart, size_t nLen)
1696 {
1697   wxString strTmp(c_str(), nStart);
1698   if ( nLen != npos ) {
1699     wxASSERT( nStart + nLen <= Len() );
1700
1701     strTmp.append(c_str() + nStart + nLen);
1702   }
1703
1704   *this = strTmp;
1705   return *this;
1706 }
1707
1708 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1709 {
1710   wxASSERT_MSG( nStart + nLen <= Len(),
1711                 _T("index out of bounds in wxString::replace") );
1712
1713   wxString strTmp;
1714   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1715
1716   if ( nStart != 0 )
1717     strTmp.append(c_str(), nStart);
1718   strTmp << sz << c_str() + nStart + nLen;
1719
1720   *this = strTmp;
1721   return *this;
1722 }
1723
1724 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1725 {
1726   return replace(nStart, nLen, wxString(ch, nCount));
1727 }
1728
1729 wxString& wxString::replace(size_t nStart, size_t nLen,
1730                             const wxString& str, size_t nStart2, size_t nLen2)
1731 {
1732   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1733 }
1734
1735 wxString& wxString::replace(size_t nStart, size_t nLen,
1736                         const wxChar* sz, size_t nCount)
1737 {
1738   return replace(nStart, nLen, wxString(sz, nCount));
1739 }
1740
1741 #endif  //std::string compatibility
1742
1743 // ============================================================================
1744 // ArrayString
1745 // ============================================================================
1746
1747 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1748 #define   ARRAY_MAXSIZE_INCREMENT       4096
1749
1750 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1751 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1752 #endif
1753
1754 #define   STRING(p)   ((wxString *)(&(p)))
1755
1756 // ctor
1757 void wxArrayString::Init(bool autoSort)
1758 {
1759   m_nSize  =
1760   m_nCount = 0;
1761   m_pItems = (wxChar **) NULL;
1762   m_autoSort = autoSort;
1763 }
1764
1765 // copy ctor
1766 wxArrayString::wxArrayString(const wxArrayString& src)
1767 {
1768   Init(src.m_autoSort);
1769
1770   *this = src;
1771 }
1772
1773 // assignment operator
1774 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1775 {
1776   if ( m_nSize > 0 )
1777     Clear();
1778
1779   Copy(src);
1780
1781   m_autoSort = src.m_autoSort;
1782
1783   return *this;
1784 }
1785
1786 void wxArrayString::Copy(const wxArrayString& src)
1787 {
1788   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1789     Alloc(src.m_nCount);
1790
1791   for ( size_t n = 0; n < src.m_nCount; n++ )
1792     Add(src[n]);
1793 }
1794
1795 // grow the array
1796 void wxArrayString::Grow(size_t nIncrement)
1797 {
1798   // only do it if no more place
1799   if ( (m_nSize - m_nCount) < nIncrement ) {
1800     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1801     // be never resized!
1802     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1803       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1804     #endif
1805
1806     if ( m_nSize == 0 ) {
1807       // was empty, alloc some memory
1808       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1809       if (m_nSize < nIncrement)
1810           m_nSize = nIncrement;
1811       m_pItems = new wxChar *[m_nSize];
1812     }
1813     else {
1814       // otherwise when it's called for the first time, nIncrement would be 0
1815       // and the array would never be expanded
1816       // add 50% but not too much
1817       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1818                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1819       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1820         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1821       if ( nIncrement < ndefIncrement )
1822         nIncrement = ndefIncrement;
1823       m_nSize += nIncrement;
1824       wxChar **pNew = new wxChar *[m_nSize];
1825
1826       // copy data to new location
1827       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1828
1829       // delete old memory (but do not release the strings!)
1830       wxDELETEA(m_pItems);
1831
1832       m_pItems = pNew;
1833     }
1834   }
1835 }
1836
1837 void wxArrayString::Free()
1838 {
1839   for ( size_t n = 0; n < m_nCount; n++ ) {
1840     STRING(m_pItems[n])->GetStringData()->Unlock();
1841   }
1842 }
1843
1844 // deletes all the strings from the list
1845 void wxArrayString::Empty()
1846 {
1847   Free();
1848
1849   m_nCount = 0;
1850 }
1851
1852 // as Empty, but also frees memory
1853 void wxArrayString::Clear()
1854 {
1855   Free();
1856
1857   m_nSize  =
1858   m_nCount = 0;
1859
1860   wxDELETEA(m_pItems);
1861 }
1862
1863 // dtor
1864 wxArrayString::~wxArrayString()
1865 {
1866   Free();
1867
1868   wxDELETEA(m_pItems);
1869 }
1870
1871 // pre-allocates memory (frees the previous data!)
1872 void wxArrayString::Alloc(size_t nSize)
1873 {
1874   // only if old buffer was not big enough
1875   if ( nSize > m_nSize ) {
1876     Free();
1877     wxDELETEA(m_pItems);
1878     m_pItems = new wxChar *[nSize];
1879     m_nSize  = nSize;
1880   }
1881
1882   m_nCount = 0;
1883 }
1884
1885 // minimizes the memory usage by freeing unused memory
1886 void wxArrayString::Shrink()
1887 {
1888   // only do it if we have some memory to free
1889   if( m_nCount < m_nSize ) {
1890     // allocates exactly as much memory as we need
1891     wxChar **pNew = new wxChar *[m_nCount];
1892
1893     // copy data to new location
1894     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1895     delete [] m_pItems;
1896     m_pItems = pNew;
1897   }
1898 }
1899
1900 // return a wxString[] as required for some control ctors.
1901 wxString* wxArrayString::GetStringArray() const
1902 {
1903     wxString *array = 0;
1904
1905     if( m_nCount > 0 )
1906     {
1907         array = new wxString[m_nCount];
1908         for( size_t i = 0; i < m_nCount; i++ )
1909             array[i] = m_pItems[i];
1910     }
1911
1912     return array;
1913 }
1914
1915 // searches the array for an item (forward or backwards)
1916 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1917 {
1918   if ( m_autoSort ) {
1919     // use binary search in the sorted array
1920     wxASSERT_MSG( bCase && !bFromEnd,
1921                   wxT("search parameters ignored for auto sorted array") );
1922
1923     size_t i,
1924            lo = 0,
1925            hi = m_nCount;
1926     int res;
1927     while ( lo < hi ) {
1928       i = (lo + hi)/2;
1929
1930       res = wxStrcmp(sz, m_pItems[i]);
1931       if ( res < 0 )
1932         hi = i;
1933       else if ( res > 0 )
1934         lo = i + 1;
1935       else
1936         return i;
1937     }
1938
1939     return wxNOT_FOUND;
1940   }
1941   else {
1942     // use linear search in unsorted array
1943     if ( bFromEnd ) {
1944       if ( m_nCount > 0 ) {
1945         size_t ui = m_nCount;
1946         do {
1947           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1948             return ui;
1949         }
1950         while ( ui != 0 );
1951       }
1952     }
1953     else {
1954       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1955         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1956           return ui;
1957       }
1958     }
1959   }
1960
1961   return wxNOT_FOUND;
1962 }
1963
1964 // add item at the end
1965 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
1966 {
1967   if ( m_autoSort ) {
1968     // insert the string at the correct position to keep the array sorted
1969     size_t i,
1970            lo = 0,
1971            hi = m_nCount;
1972     int res;
1973     while ( lo < hi ) {
1974       i = (lo + hi)/2;
1975
1976       res = wxStrcmp(str, m_pItems[i]);
1977       if ( res < 0 )
1978         hi = i;
1979       else if ( res > 0 )
1980         lo = i + 1;
1981       else {
1982         lo = hi = i;
1983         break;
1984       }
1985     }
1986
1987     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1988
1989     Insert(str, lo, nInsert);
1990
1991     return (size_t)lo;
1992   }
1993   else {
1994     wxASSERT( str.GetStringData()->IsValid() );
1995
1996     Grow(nInsert);
1997
1998     for (size_t i = 0; i < nInsert; i++)
1999     {
2000         // the string data must not be deleted!
2001         str.GetStringData()->Lock();
2002
2003         // just append
2004         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2005     }
2006     size_t ret = m_nCount;
2007     m_nCount += nInsert;
2008     return ret;
2009   }
2010 }
2011
2012 // add item at the given position
2013 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2014 {
2015   wxASSERT( str.GetStringData()->IsValid() );
2016
2017   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2018   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2019                wxT("array size overflow in wxArrayString::Insert") );
2020
2021   Grow(nInsert);
2022
2023   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2024           (m_nCount - nIndex)*sizeof(wxChar *));
2025
2026   for (size_t i = 0; i < nInsert; i++)
2027   {
2028       str.GetStringData()->Lock();
2029       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2030   }
2031   m_nCount += nInsert;
2032 }
2033
2034 // expand the array
2035 void wxArrayString::SetCount(size_t count)
2036 {
2037     Alloc(count);
2038
2039     wxString s;
2040     while ( m_nCount < count )
2041         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2042 }
2043
2044 // removes item from array (by index)
2045 void wxArrayString::Remove(size_t nIndex, size_t nRemove)
2046 {
2047   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2048   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2049                wxT("removing too many elements in wxArrayString::Remove") );
2050
2051   // release our lock
2052   for (size_t i = 0; i < nRemove; i++)
2053       Item(nIndex + i).GetStringData()->Unlock();
2054
2055   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2056           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2057   m_nCount -= nRemove;
2058 }
2059
2060 // removes item from array (by value)
2061 void wxArrayString::Remove(const wxChar *sz)
2062 {
2063   int iIndex = Index(sz);
2064
2065   wxCHECK_RET( iIndex != wxNOT_FOUND,
2066                wxT("removing inexistent element in wxArrayString::Remove") );
2067
2068   Remove(iIndex);
2069 }
2070
2071 // ----------------------------------------------------------------------------
2072 // sorting
2073 // ----------------------------------------------------------------------------
2074
2075 // we can only sort one array at a time with the quick-sort based
2076 // implementation
2077 #if wxUSE_THREADS
2078   // need a critical section to protect access to gs_compareFunction and
2079   // gs_sortAscending variables
2080   static wxCriticalSection *gs_critsectStringSort = NULL;
2081
2082   // call this before the value of the global sort vars is changed/after
2083   // you're finished with them
2084   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2085                            gs_critsectStringSort = new wxCriticalSection;     \
2086                            gs_critsectStringSort->Enter()
2087   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2088                            delete gs_critsectStringSort;                      \
2089                            gs_critsectStringSort = NULL
2090 #else // !threads
2091   #define START_SORT()
2092   #define END_SORT()
2093 #endif // wxUSE_THREADS
2094
2095 // function to use for string comparaison
2096 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2097
2098 // if we don't use the compare function, this flag tells us if we sort the
2099 // array in ascending or descending order
2100 static bool gs_sortAscending = TRUE;
2101
2102 // function which is called by quick sort
2103 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2104 wxStringCompareFunction(const void *first, const void *second)
2105 {
2106   wxString *strFirst = (wxString *)first;
2107   wxString *strSecond = (wxString *)second;
2108
2109   if ( gs_compareFunction ) {
2110     return gs_compareFunction(*strFirst, *strSecond);
2111   }
2112   else {
2113     // maybe we should use wxStrcoll
2114     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2115
2116     return gs_sortAscending ? result : -result;
2117   }
2118 }
2119
2120 // sort array elements using passed comparaison function
2121 void wxArrayString::Sort(CompareFunction compareFunction)
2122 {
2123   START_SORT();
2124
2125   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2126   gs_compareFunction = compareFunction;
2127
2128   DoSort();
2129
2130   // reset it to NULL so that Sort(bool) will work the next time
2131   gs_compareFunction = NULL;
2132
2133   END_SORT();
2134 }
2135
2136 void wxArrayString::Sort(bool reverseOrder)
2137 {
2138   START_SORT();
2139
2140   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2141   gs_sortAscending = !reverseOrder;
2142
2143   DoSort();
2144
2145   END_SORT();
2146 }
2147
2148 void wxArrayString::DoSort()
2149 {
2150   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2151
2152   // just sort the pointers using qsort() - of course it only works because
2153   // wxString() *is* a pointer to its data
2154   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2155 }
2156
2157 bool wxArrayString::operator==(const wxArrayString& a) const
2158 {
2159     if ( m_nCount != a.m_nCount )
2160         return FALSE;
2161
2162     for ( size_t n = 0; n < m_nCount; n++ )
2163     {
2164         if ( Item(n) != a[n] )
2165             return FALSE;
2166     }
2167
2168     return TRUE;
2169 }
2170