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 #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
  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 #if !wxUSE_STL
  65   const size_t wxStringBase::npos = wxSTRING_MAXLEN;
  66 #endif
  67
  68 // ----------------------------------------------------------------------------
  69 // static data
  70 // ----------------------------------------------------------------------------
  71
  72 #if wxUSE_STL
  73
  74 extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T("");
  75
  76 #else
  77
  78 // for an empty string, GetStringData() will return this address: this
  79 // structure has the same layout as wxStringData and it's data() method will
  80 // return the empty string (dummy pointer)
  81 static const struct
  82 {
  83   wxStringData data;
  84   wxChar dummy;
  85 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  86
  87 // empty C style string: points to 'string data' byte of g_strEmpty
  88 extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy;
  89
  90 #endif
  91
  92 // ----------------------------------------------------------------------------
  93 // global functions
  94 // ----------------------------------------------------------------------------
  95
  96 #if wxUSE_STD_IOSTREAM
  97
  98 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
  99 // iostream ones.
 100 //
 101 // ATTN: you can _not_ use both of these in the same program!
 102
 103 wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str))
 104 {
 105 #if 0
 106   int w = is.width(0);
 107   if ( is.ipfx(0) ) {
 108     streambuf *sb = is.rdbuf();
 109     str.erase();
 110     while ( true ) {
 111       int ch = sb->sbumpc ();
 112       if ( ch == EOF ) {
 113         is.setstate(ios::eofbit);
 114         break;
 115       }
 116       else if ( isspace(ch) ) {
 117         sb->sungetc();
 118         break;
 119       }
 120
 121       str += ch;
 122       if ( --w == 1 )
 123         break;
 124     }
 125   }
 126
 127   is.isfx();
 128   if ( str.length() == 0 )
 129     is.setstate(ios::failbit);
 130 #endif
 131   return is;
 132 }
 133
 134 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
 135 {
 136   os << str.c_str();
 137   return os;
 138 }
 139
 140 #endif // wxUSE_STD_IOSTREAM
 141
 142 // ----------------------------------------------------------------------------
 143 // private classes
 144 // ----------------------------------------------------------------------------
 145
 146 // this small class is used to gather statistics for performance tuning
 147 //#define WXSTRING_STATISTICS
 148 #ifdef  WXSTRING_STATISTICS
 149   class Averager
 150   {
 151   public:
 152     Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 153    ~Averager()
 154    { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 155
 156     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 157
 158   private:
 159     size_t m_nCount, m_nTotal;
 160     const wxChar *m_sz;
 161   } g_averageLength("allocation size"),
 162     g_averageSummandLength("summand length"),
 163     g_averageConcatHit("hit probability in concat"),
 164     g_averageInitialLength("initial string length");
 165
 166   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 167 #else
 168   #define STATISTICS_ADD(av, val)
 169 #endif // WXSTRING_STATISTICS
 170
 171 // ===========================================================================
 172 // wxStringData class deallocation
 173 // ===========================================================================
 174
 175 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 176 #  pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
 177 void wxStringData::Free()
 178 {
 179     free(this);
 180 }
 181 #endif
 182
 183 #if !wxUSE_STL
 184
 185 // ===========================================================================
 186 // wxStringBase
 187 // ===========================================================================
 188
 189 // takes nLength elements of psz starting at nPos
 190 void wxStringBase::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 191 {
 192   Init();
 193
 194   // if the length is not given, assume the string to be NUL terminated
 195   if ( nLength == npos ) {
 196     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 197
 198     nLength = wxStrlen(psz + nPos);
 199   }
 200
 201   STATISTICS_ADD(InitialLength, nLength);
 202
 203   if ( nLength > 0 ) {
 204     // trailing '\0' is written in AllocBuffer()
 205     if ( !AllocBuffer(nLength) ) {
 206       wxFAIL_MSG( _T("out of memory in wxStringBase::InitWith") );
 207       return;
 208     }
 209     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 210   }
 211 }
 212
 213 // poor man's iterators are "void *" pointers
 214 wxStringBase::wxStringBase(const void *pStart, const void *pEnd)
 215 {
 216   InitWith((const wxChar *)pStart, 0,
 217            (const wxChar *)pEnd - (const wxChar *)pStart);
 218 }
 219
 220 wxStringBase::wxStringBase(size_type n, wxChar ch)
 221 {
 222   Init();
 223   append(n, ch);
 224 }
 225
 226 // ---------------------------------------------------------------------------
 227 // memory allocation
 228 // ---------------------------------------------------------------------------
 229
 230 // allocates memory needed to store a C string of length nLen
 231 bool wxStringBase::AllocBuffer(size_t nLen)
 232 {
 233   // allocating 0 sized buffer doesn't make sense, all empty strings should
 234   // reuse g_strEmpty
 235   wxASSERT( nLen >  0 );
 236
 237   // make sure that we don't overflow
 238   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 239                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 240
 241   STATISTICS_ADD(Length, nLen);
 242
 243   // allocate memory:
 244   // 1) one extra character for '\0' termination
 245   // 2) sizeof(wxStringData) for housekeeping info
 246   wxStringData* pData = (wxStringData*)
 247     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 248
 249   if ( pData == NULL ) {
 250     // allocation failures are handled by the caller
 251     return FALSE;
 252   }
 253
 254   pData->nRefs        = 1;
 255   pData->nDataLength  = nLen;
 256   pData->nAllocLength = nLen + EXTRA_ALLOC;
 257   m_pchData           = pData->data();  // data starts after wxStringData
 258   m_pchData[nLen]     = wxT('\0');
 259   return TRUE;
 260 }
 261
 262 // must be called before changing this string
 263 bool wxStringBase::CopyBeforeWrite()
 264 {
 265   wxStringData* pData = GetStringData();
 266
 267   if ( pData->IsShared() ) {
 268     pData->Unlock();                // memory not freed because shared
 269     size_t nLen = pData->nDataLength;
 270     if ( !AllocBuffer(nLen) ) {
 271       // allocation failures are handled by the caller
 272       return FALSE;
 273     }
 274     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 275   }
 276
 277   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 278
 279   return TRUE;
 280 }
 281
 282 // must be called before replacing contents of this string
 283 bool wxStringBase::AllocBeforeWrite(size_t nLen)
 284 {
 285   wxASSERT( nLen != 0 );  // doesn't make any sense
 286
 287   // must not share string and must have enough space
 288   wxStringData* pData = GetStringData();
 289   if ( pData->IsShared() || pData->IsEmpty() ) {
 290     // can't work with old buffer, get new one
 291     pData->Unlock();
 292     if ( !AllocBuffer(nLen) ) {
 293       // allocation failures are handled by the caller
 294       return FALSE;
 295     }
 296   }
 297   else {
 298     if ( nLen > pData->nAllocLength ) {
 299       // realloc the buffer instead of calling malloc() again, this is more
 300       // efficient
 301       STATISTICS_ADD(Length, nLen);
 302
 303       nLen += EXTRA_ALLOC;
 304
 305       pData = (wxStringData*)
 306           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 307
 308       if ( pData == NULL ) {
 309         // allocation failures are handled by the caller
 310         // keep previous data since reallocation failed
 311         return FALSE;
 312       }
 313
 314       pData->nAllocLength = nLen;
 315       m_pchData = pData->data();
 316     }
 317
 318     // now we have enough space, just update the string length
 319     pData->nDataLength = nLen;
 320   }
 321
 322   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 323
 324   return TRUE;
 325 }
 326
 327 wxStringBase& wxStringBase::append(size_t n, wxChar ch)
 328 {
 329     size_type len = length();
 330
 331     if ( !CopyBeforeWrite() || !Alloc(len + n) ) {
 332       wxFAIL_MSG( _T("out of memory in wxStringBase::append") );
 333     }
 334     GetStringData()->nDataLength = len + n;
 335     m_pchData[len + n] = '\0';
 336     for ( size_t i = 0; i < n; ++i )
 337         m_pchData[len + i] = ch;
 338     return *this;
 339 }
 340
 341 void wxStringBase::resize(size_t nSize, wxChar ch)
 342 {
 343     size_t len = length();
 344
 345     if ( nSize < len )
 346     {
 347         erase(begin() + nSize, end());
 348     }
 349     else if ( nSize > len )
 350     {
 351         append(nSize - len, ch);
 352     }
 353     //else: we have exactly the specified length, nothing to do
 354 }
 355
 356 // allocate enough memory for nLen characters
 357 bool wxStringBase::Alloc(size_t nLen)
 358 {
 359   wxStringData *pData = GetStringData();
 360   if ( pData->nAllocLength <= nLen ) {
 361     if ( pData->IsEmpty() ) {
 362       nLen += EXTRA_ALLOC;
 363
 364       wxStringData* pData = (wxStringData*)
 365           malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 366
 367       if ( pData == NULL ) {
 368         // allocation failure handled by caller
 369         return FALSE;
 370       }
 371
 372       pData->nRefs = 1;
 373       pData->nDataLength = 0;
 374       pData->nAllocLength = nLen;
 375       m_pchData = pData->data();  // data starts after wxStringData
 376       m_pchData[0u] = wxT('\0');
 377     }
 378     else if ( pData->IsShared() ) {
 379       pData->Unlock();                // memory not freed because shared
 380       size_t nOldLen = pData->nDataLength;
 381       if ( !AllocBuffer(nLen) ) {
 382         // allocation failure handled by caller
 383         return FALSE;
 384       }
 385       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 386     }
 387     else {
 388       nLen += EXTRA_ALLOC;
 389
 390       pData = (wxStringData *)
 391         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 392
 393       if ( pData == NULL ) {
 394         // allocation failure handled by caller
 395         // keep previous data since reallocation failed
 396         return FALSE;
 397       }
 398
 399       // it's not important if the pointer changed or not (the check for this
 400       // is not faster than assigning to m_pchData in all cases)
 401       pData->nAllocLength = nLen;
 402       m_pchData = pData->data();
 403     }
 404   }
 405   //else: we've already got enough
 406   return TRUE;
 407 }
 408
 409 wxStringBase::iterator wxStringBase::erase(iterator it)
 410 {
 411     size_type idx = it - begin();
 412     erase(idx, 1);
 413     return begin() + idx;
 414 }
 415
 416 wxStringBase& wxStringBase::erase(size_t nStart, size_t nLen)
 417 {
 418   wxASSERT(nStart <= length());
 419   size_t strLen = length() - nStart;
 420   // delete nLen or up to the end of the string characters
 421   nLen = strLen < nLen ? strLen : nLen;
 422   wxString strTmp(c_str(), nStart);
 423   strTmp.append(c_str() + nStart + nLen, length() - nStart - nLen);
 424
 425   swap(strTmp);
 426   return *this;
 427 }
 428
 429 wxStringBase& wxStringBase::insert(size_t nPos, const wxChar *sz, size_t n)
 430 {
 431   wxASSERT( nPos <= length() );
 432
 433   if ( n == npos ) n = wxStrlen(sz);
 434   if ( n == 0 ) return *this;
 435
 436   if ( !CopyBeforeWrite() || !Alloc(length() + n) ) {
 437     wxFAIL_MSG( _T("out of memory in wxStringBase::insert") );
 438   }
 439
 440   memmove(m_pchData + nPos + n, m_pchData + nPos,
 441           (length() - nPos) * sizeof(wxChar));
 442   memcpy(m_pchData + nPos, sz, n * sizeof(wxChar));
 443   GetStringData()->nDataLength = length() + n;
 444   m_pchData[length()] = '\0';
 445
 446   return *this;
 447 }
 448
 449 void wxStringBase::swap(wxStringBase& str)
 450 {
 451     wxChar* tmp = str.m_pchData;
 452     str.m_pchData = m_pchData;
 453     m_pchData = tmp;
 454 }
 455
 456 size_t wxStringBase::find(const wxStringBase& str, size_t nStart) const
 457 {
 458   wxASSERT( str.GetStringData()->IsValid() );
 459   wxASSERT( nStart <= length() );
 460
 461   const wxChar *p = wxStrstr(c_str() + nStart, str.c_str());
 462
 463   return p == NULL ? npos : p - c_str();
 464 }
 465
 466 size_t wxStringBase::find(const wxChar* sz, size_t nStart, size_t n) const
 467 {
 468   return find(wxStringBase(sz, n), nStart);
 469 }
 470
 471 size_t wxStringBase::find(wxChar ch, size_t nStart) const
 472 {
 473   wxASSERT( nStart <= length() );
 474
 475   const wxChar *p = wxStrchr(c_str() + nStart, ch);
 476
 477   return p == NULL ? npos : p - c_str();
 478 }
 479
 480 size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const
 481 {
 482   wxASSERT( str.GetStringData()->IsValid() );
 483   wxASSERT( nStart == npos || nStart <= length() );
 484
 485   // TODO could be made much quicker than that
 486   const wxChar *p = c_str() + (nStart == npos ? length() : nStart);
 487   while ( p >= c_str() + str.length() ) {
 488     if ( wxStrncmp(p - str.length(), str.c_str(), str.length()) == 0 )
 489       return p - str.length() - c_str();
 490     p--;
 491   }
 492
 493   return npos;
 494 }
 495
 496 size_t wxStringBase::rfind(const wxChar* sz, size_t nStart, size_t n) const
 497 {
 498     return rfind(wxStringBase(sz, n), nStart);
 499 }
 500
 501 size_t wxStringBase::rfind(wxChar ch, size_t nStart) const
 502 {
 503     if ( nStart == npos )
 504     {
 505         nStart = length();
 506     }
 507     else
 508     {
 509         wxASSERT( nStart <= length() );
 510     }
 511
 512     const wxChar *p = wxStrrchr(c_str(), ch);
 513
 514     if ( p == NULL )
 515         return npos;
 516
 517     size_t result = p - c_str();
 518     return ( result > nStart ) ? npos : result;
 519 }
 520
 521 size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart) const
 522 {
 523     const wxChar *start = c_str() + nStart;
 524     const wxChar *firstOf = wxStrpbrk(start, sz);
 525     if ( firstOf )
 526         return firstOf - c_str();
 527     else
 528         return npos;
 529 }
 530
 531 size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart) const
 532 {
 533     if ( nStart == npos )
 534     {
 535         nStart = length();
 536     }
 537     else
 538     {
 539         wxASSERT_MSG( nStart <= length(),
 540                         _T("invalid index in find_last_of()") );
 541     }
 542
 543     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
 544     {
 545         if ( wxStrchr(sz, *p) )
 546             return p - c_str();
 547     }
 548
 549     return npos;
 550 }
 551
 552 size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart) const
 553 {
 554     if ( nStart == npos )
 555     {
 556         nStart = length();
 557     }
 558     else
 559     {
 560         wxASSERT( nStart <= length() );
 561     }
 562
 563     size_t nAccept = wxStrspn(c_str() + nStart, sz);
 564     if ( nAccept >= length() - nStart )
 565         return npos;
 566     else
 567         return nAccept;
 568 }
 569
 570 size_t wxStringBase::find_first_not_of(wxChar ch, size_t nStart) const
 571 {
 572     wxASSERT( nStart <= length() );
 573
 574     for ( const wxChar *p = c_str() + nStart; *p; p++ )
 575     {
 576         if ( *p != ch )
 577             return p - c_str();
 578     }
 579
 580     return npos;
 581 }
 582
 583 size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart) const
 584 {
 585     if ( nStart == npos )
 586     {
 587         nStart = length();
 588     }
 589     else
 590     {
 591         wxASSERT( nStart <= length() );
 592     }
 593
 594     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
 595     {
 596         if ( !wxStrchr(sz, *p) )
 597             return p - c_str();
 598     }
 599
 600     return npos;
 601 }
 602
 603 size_t wxStringBase::find_last_not_of(wxChar ch, size_t nStart) const
 604 {
 605     if ( nStart == npos )
 606     {
 607         nStart = length();
 608     }
 609     else
 610     {
 611         wxASSERT( nStart <= length() );
 612     }
 613
 614     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
 615     {
 616         if ( *p != ch )
 617             return p - c_str();
 618     }
 619
 620     return npos;
 621 }
 622
 623 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 624                                     const wxChar *sz)
 625 {
 626   wxASSERT_MSG( nStart <= length(),
 627                 _T("index out of bounds in wxStringBase::replace") );
 628   size_t strLen = length() - nStart;
 629   nLen = strLen < nLen ? strLen : nLen;
 630
 631   wxStringBase strTmp;
 632   strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs
 633
 634   if ( nStart != 0 )
 635     strTmp.append(c_str(), nStart);
 636   strTmp.append(sz);
 637   strTmp.append(c_str() + nStart + nLen);
 638
 639   swap(strTmp);
 640   return *this;
 641 }
 642
 643 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 644                                     size_t nCount, wxChar ch)
 645 {
 646   return replace(nStart, nLen, wxStringBase(ch, nCount).c_str());
 647 }
 648
 649 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 650                                     const wxStringBase& str,
 651                                     size_t nStart2, size_t nLen2)
 652 {
 653   return replace(nStart, nLen, str.substr(nStart2, nLen2));
 654 }
 655
 656 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 657                                     const wxChar* sz, size_t nCount)
 658 {
 659   return replace(nStart, nLen, wxStringBase(sz, nCount).c_str());
 660 }
 661
 662 wxStringBase wxStringBase::substr(size_t nStart, size_t nLen) const
 663 {
 664   if ( nLen == npos )
 665     nLen = length() - nStart;
 666   return wxStringBase(*this, nStart, nLen);
 667 }
 668
 669 // assigns one string to another
 670 wxStringBase& wxStringBase::operator=(const wxStringBase& stringSrc)
 671 {
 672   wxASSERT( stringSrc.GetStringData()->IsValid() );
 673
 674   // don't copy string over itself
 675   if ( m_pchData != stringSrc.m_pchData ) {
 676     if ( stringSrc.GetStringData()->IsEmpty() ) {
 677       Reinit();
 678     }
 679     else {
 680       // adjust references
 681       GetStringData()->Unlock();
 682       m_pchData = stringSrc.m_pchData;
 683       GetStringData()->Lock();
 684     }
 685   }
 686
 687   return *this;
 688 }
 689
 690 // assigns a single character
 691 wxStringBase& wxStringBase::operator=(wxChar ch)
 692 {
 693   if ( !AssignCopy(1, &ch) ) {
 694     wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(wxChar)") );
 695   }
 696   return *this;
 697 }
 698
 699 // assigns C string
 700 wxStringBase& wxStringBase::operator=(const wxChar *psz)
 701 {
 702   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 703     wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(const wxChar *)") );
 704   }
 705   return *this;
 706 }
 707
 708 // helper function: does real copy
 709 bool wxStringBase::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 710 {
 711   if ( nSrcLen == 0 ) {
 712     Reinit();
 713   }
 714   else {
 715     if ( !AllocBeforeWrite(nSrcLen) ) {
 716       // allocation failure handled by caller
 717       return FALSE;
 718     }
 719     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 720     GetStringData()->nDataLength = nSrcLen;
 721     m_pchData[nSrcLen] = wxT('\0');
 722   }
 723   return TRUE;
 724 }
 725
 726 // ---------------------------------------------------------------------------
 727 // string concatenation
 728 // ---------------------------------------------------------------------------
 729
 730 // add something to this string
 731 bool wxStringBase::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData,
 732                               size_t nMaxLen)
 733 {
 734   STATISTICS_ADD(SummandLength, nSrcLen);
 735
 736   nSrcLen = nSrcLen < nMaxLen ? nSrcLen : nMaxLen;
 737
 738   // concatenating an empty string is a NOP
 739   if ( nSrcLen > 0 ) {
 740     wxStringData *pData = GetStringData();
 741     size_t nLen = pData->nDataLength;
 742     size_t nNewLen = nLen + nSrcLen;
 743
 744     // alloc new buffer if current is too small
 745     if ( pData->IsShared() ) {
 746       STATISTICS_ADD(ConcatHit, 0);
 747
 748       // we have to allocate another buffer
 749       wxStringData* pOldData = GetStringData();
 750       if ( !AllocBuffer(nNewLen) ) {
 751           // allocation failure handled by caller
 752           return FALSE;
 753       }
 754       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 755       pOldData->Unlock();
 756     }
 757     else if ( nNewLen > pData->nAllocLength ) {
 758       STATISTICS_ADD(ConcatHit, 0);
 759
 760       reserve(nNewLen);
 761       // we have to grow the buffer
 762       if ( capacity() < nNewLen ) {
 763           // allocation failure handled by caller
 764           return FALSE;
 765       }
 766     }
 767     else {
 768       STATISTICS_ADD(ConcatHit, 1);
 769
 770       // the buffer is already big enough
 771     }
 772
 773     // should be enough space
 774     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 775
 776     // fast concatenation - all is done in our buffer
 777     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 778
 779     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 780     GetStringData()->nDataLength = nNewLen; // and fix the length
 781   }
 782   //else: the string to append was empty
 783   return TRUE;
 784 }
 785
 786 // ---------------------------------------------------------------------------
 787 // simple sub-string extraction
 788 // ---------------------------------------------------------------------------
 789
 790 // helper function: clone the data attached to this string
 791 bool wxStringBase::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 792 {
 793   if ( nCopyLen == 0 ) {
 794     dest.Init();
 795   }
 796   else {
 797     if ( !dest.AllocBuffer(nCopyLen) ) {
 798       // allocation failure handled by caller
 799       return FALSE;
 800     }
 801     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 802   }
 803   return TRUE;
 804 }
 805
 806 #endif // !wxUSE_STL
 807
 808 #if !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
 809
 810 #if !wxUSE_STL
 811     #define STRINGCLASS wxStringBase
 812 #else
 813     #define STRINGCLASS wxString
 814 #endif
 815
 816 static inline int wxDoCmp(const wxChar* s1, size_t l1,
 817                           const wxChar* s2, size_t l2)
 818 {
 819     if( l1 == l2 )
 820         return wxStrncmp(s1, s2, l1);
 821     else if( l1 < l2 )
 822     {
 823         int ret = wxStrncmp(s1, s2, l1);
 824         return ret == 0 ? -1 : ret;
 825     }
 826     else if( l1 > l2 )
 827     {
 828         int ret = wxStrncmp(s1, s2, l2);
 829         return ret == 0 ? +1 : ret;
 830     }
 831
 832     wxFAIL;   // must never get there
 833     return 0; // quiet compilers
 834 }
 835
 836 #if wxUSE_STL
 837
 838 int STRINGCLASS::compare(const wxStringBase& str) const
 839 {
 840     return ::wxDoCmp(data(), length(), str.data(), str.length());
 841 }
 842
 843 #endif
 844
 845 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 846                          const wxStringBase& str) const
 847 {
 848     wxASSERT(nStart <= length());
 849     size_type strLen = length() - nStart;
 850     nLen = strLen < nLen ? strLen : nLen;
 851     return ::wxDoCmp(data() + nStart, nLen, str.data(), str.length());
 852 }
 853
 854 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 855                          const wxStringBase& str,
 856                          size_t nStart2, size_t nLen2) const
 857 {
 858     wxASSERT(nStart <= length());
 859     wxASSERT(nStart2 <= str.length());
 860     size_type strLen  =     length() - nStart,
 861               strLen2 = str.length() - nStart2;
 862     nLen  = strLen  < nLen  ? strLen  : nLen;
 863     nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
 864     return ::wxDoCmp(data() + nStart, nLen, str.data() + nStart2, nLen2);
 865 }
 866
 867 #if wxUSE_STL
 868
 869 int STRINGCLASS::compare(const wxChar* sz) const
 870 {
 871     size_t nLen = wxStrlen(sz);
 872     return ::wxDoCmp(data(), length(), sz, nLen);
 873 }
 874
 875 #endif
 876
 877 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 878                          const wxChar* sz, size_t nCount) const
 879 {
 880     wxASSERT(nStart <= length());
 881     size_type strLen = length() - nStart;
 882     nLen = strLen < nLen ? strLen : nLen;
 883     if( nCount == npos )
 884         nCount = wxStrlen(sz);
 885
 886     return ::wxDoCmp(data() + nStart, nLen, sz, nCount);
 887 }
 888
 889 #undef STRINGCLASS
 890
 891 #endif // !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
 892
 893 // ===========================================================================
 894 // wxString class core
 895 // ===========================================================================
 896
 897 // ---------------------------------------------------------------------------
 898 // construction
 899 // ---------------------------------------------------------------------------
 900
 901 #if wxUSE_UNICODE
 902
 903 // from multibyte string
 904 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 905 {
 906     // first get the size of the buffer we need
 907     size_t nLen;
 908     if ( psz )
 909     {
 910         // calculate the needed size ourselves or use the provided one
 911         nLen = nLength == npos ? conv.MB2WC(NULL, psz, 0) : nLength;
 912     }
 913     else
 914     {
 915         // nothing to convert
 916         nLen = 0;
 917     }
 918
 919     // anything to do?
 920     if ( (nLen != 0) && (nLen != (size_t)-1) )
 921     {
 922         if ( !Alloc(nLen) )
 923         {
 924             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 925         }
 926         else
 927         {
 928             wxWCharBuffer buf(nLen + 1);
 929             // MB2WC wants the buffer size, not the string length hence +1
 930             nLen = conv.MB2WC(buf.data(), psz, nLen + 1);
 931
 932             if ( nLen != (size_t)-1 )
 933             {
 934                 // initialized ok, set the real length as nLength specified by
 935                 // the caller could be greater than the real string length
 936                 assign(buf.data(), nLen);
 937                 return;
 938             }
 939             //else: the conversion failed -- leave the string empty (what else?)
 940         }
 941     }
 942 }
 943
 944 #else // ANSI
 945
 946 #if wxUSE_WCHAR_T
 947 // from wide string
 948 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 949 {
 950     // first get the size of the buffer we need
 951     size_t nLen;
 952     if ( pwz )
 953     {
 954         // calculate the needed size ourselves or use the provided one
 955         nLen = nLength == npos ? conv.WC2MB(NULL, pwz, 0) : nLength;
 956     }
 957     else
 958     {
 959         // nothing to convert
 960         nLen = 0;
 961     }
 962
 963     // anything to do?
 964     if ( (nLen != 0) && (nLen != (size_t)-1) )
 965     {
 966         if ( !Alloc(nLen) )
 967         {
 968             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 969         }
 970         else
 971         {
 972             wxCharBuffer buf(nLen);
 973             // WC2MB wants the buffer size, not the string length
 974             if ( conv.WC2MB(buf.data(), pwz, nLen + 1) != (size_t)-1 )
 975             {
 976                 // initialized ok
 977                 assign(buf.data(), nLen);
 978                 return;
 979             }
 980             //else: the conversion failed -- leave the string empty (what else?)
 981         }
 982     }
 983
 984     // leave empty
 985 }
 986 #endif // wxUSE_WCHAR_T
 987
 988 #endif // Unicode/ANSI
 989
 990 // shrink to minimal size (releasing extra memory)
 991 bool wxString::Shrink()
 992 {
 993   wxString tmp(begin(), end());
 994   swap(tmp);
 995   return tmp.length() == length();
 996 }
 997
 998 #if !wxUSE_STL
 999 // get the pointer to writable buffer of (at least) nLen bytes
1000 wxChar *wxString::GetWriteBuf(size_t nLen)
1001 {
1002   if ( !AllocBeforeWrite(nLen) ) {
1003     // allocation failure handled by caller
1004     return NULL;
1005   }
1006
1007   wxASSERT( GetStringData()->nRefs == 1 );
1008   GetStringData()->Validate(FALSE);
1009
1010   return m_pchData;
1011 }
1012
1013 // put string back in a reasonable state after GetWriteBuf
1014 void wxString::UngetWriteBuf()
1015 {
1016   GetStringData()->nDataLength = wxStrlen(m_pchData);
1017   GetStringData()->Validate(TRUE);
1018 }
1019
1020 void wxString::UngetWriteBuf(size_t nLen)
1021 {
1022   GetStringData()->nDataLength = nLen;
1023   GetStringData()->Validate(TRUE);
1024 }
1025 #endif
1026
1027 // ---------------------------------------------------------------------------
1028 // data access
1029 // ---------------------------------------------------------------------------
1030
1031 // all functions are inline in string.h
1032
1033 // ---------------------------------------------------------------------------
1034 // assignment operators
1035 // ---------------------------------------------------------------------------
1036
1037 #if !wxUSE_UNICODE
1038
1039 // same as 'signed char' variant
1040 wxString& wxString::operator=(const unsigned char* psz)
1041 {
1042   *this = (const char *)psz;
1043   return *this;
1044 }
1045
1046 #if wxUSE_WCHAR_T
1047 wxString& wxString::operator=(const wchar_t *pwz)
1048 {
1049   wxString str(pwz);
1050   swap(str);
1051   return *this;
1052 }
1053 #endif
1054
1055 #endif
1056
1057 /*
1058  * concatenation functions come in 5 flavours:
1059  *  string + string
1060  *  char   + string      and      string + char
1061  *  C str  + string      and      string + C str
1062  */
1063
1064 wxString operator+(const wxString& str1, const wxString& str2)
1065 {
1066 #if !wxUSE_STL
1067   wxASSERT( str1.GetStringData()->IsValid() );
1068   wxASSERT( str2.GetStringData()->IsValid() );
1069 #endif
1070
1071   wxString s = str1;
1072   s += str2;
1073
1074   return s;
1075 }
1076
1077 wxString operator+(const wxString& str, wxChar ch)
1078 {
1079 #if !wxUSE_STL
1080   wxASSERT( str.GetStringData()->IsValid() );
1081 #endif
1082
1083   wxString s = str;
1084   s += ch;
1085
1086   return s;
1087 }
1088
1089 wxString operator+(wxChar ch, const wxString& str)
1090 {
1091 #if !wxUSE_STL
1092   wxASSERT( str.GetStringData()->IsValid() );
1093 #endif
1094
1095   wxString s = ch;
1096   s += str;
1097
1098   return s;
1099 }
1100
1101 wxString operator+(const wxString& str, const wxChar *psz)
1102 {
1103 #if !wxUSE_STL
1104   wxASSERT( str.GetStringData()->IsValid() );
1105 #endif
1106
1107   wxString s;
1108   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
1109     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1110   }
1111   s = str;
1112   s += psz;
1113
1114   return s;
1115 }
1116
1117 wxString operator+(const wxChar *psz, const wxString& str)
1118 {
1119 #if !wxUSE_STL
1120   wxASSERT( str.GetStringData()->IsValid() );
1121 #endif
1122
1123   wxString s;
1124   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
1125     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1126   }
1127   s = psz;
1128   s += str;
1129
1130   return s;
1131 }
1132
1133 // ===========================================================================
1134 // other common string functions
1135 // ===========================================================================
1136
1137 #if wxUSE_UNICODE
1138
1139 wxString wxString::FromAscii(const char *ascii)
1140 {
1141     if (!ascii)
1142        return wxEmptyString;
1143
1144     size_t len = strlen( ascii );
1145     wxString res;
1146
1147     if ( len )
1148     {
1149         wxStringBuffer buf(res, len);
1150
1151         wchar_t *dest = buf;
1152
1153         for ( ;; )
1154         {
1155            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
1156                break;
1157         }
1158     }
1159
1160     return res;
1161 }
1162
1163 wxString wxString::FromAscii(const char ascii)
1164 {
1165     // What do we do with '\0' ?
1166
1167     wxString res;
1168     res += (wchar_t)(unsigned char) ascii;
1169
1170     return res;
1171 }
1172
1173 const wxCharBuffer wxString::ToAscii() const
1174 {
1175     // this will allocate enough space for the terminating NUL too
1176     wxCharBuffer buffer(length());
1177
1178     signed char *dest = (signed char *)buffer.data();
1179
1180     const wchar_t *pwc = c_str();
1181     for ( ;; )
1182     {
1183         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
1184
1185         // the output string can't have embedded NULs anyhow, so we can safely
1186         // stop at first of them even if we do have any
1187         if ( !*pwc++ )
1188             break;
1189     }
1190
1191     return buffer;
1192 }
1193
1194 #endif // Unicode
1195
1196 // extract string of length nCount starting at nFirst
1197 wxString wxString::Mid(size_t nFirst, size_t nCount) const
1198 {
1199   size_t nLen = length();
1200
1201   // default value of nCount is npos and means "till the end"
1202   if ( nCount == npos )
1203   {
1204     nCount = nLen - nFirst;
1205   }
1206
1207   // out-of-bounds requests return sensible things
1208   if ( nFirst + nCount > nLen )
1209   {
1210     nCount = nLen - nFirst;
1211   }
1212
1213   if ( nFirst > nLen )
1214   {
1215     // AllocCopy() will return empty string
1216     nCount = 0;
1217   }
1218
1219   wxString dest(*this, nFirst, nCount);
1220   if ( dest.length() != nCount ) {
1221       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
1222   }
1223
1224   return dest;
1225 }
1226
1227 // check that the string starts with prefix and return the rest of the string
1228 // in the provided pointer if it is not NULL, otherwise return FALSE
1229 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
1230 {
1231     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
1232
1233     // first check if the beginning of the string matches the prefix: note
1234     // that we don't have to check that we don't run out of this string as
1235     // when we reach the terminating NUL, either prefix string ends too (and
1236     // then it's ok) or we break out of the loop because there is no match
1237     const wxChar *p = c_str();
1238     while ( *prefix )
1239     {
1240         if ( *prefix++ != *p++ )
1241         {
1242             // no match
1243             return FALSE;
1244         }
1245     }
1246
1247     if ( rest )
1248     {
1249         // put the rest of the string into provided pointer
1250         *rest = p;
1251     }
1252
1253     return TRUE;
1254 }
1255
1256 // extract nCount last (rightmost) characters
1257 wxString wxString::Right(size_t nCount) const
1258 {
1259   if ( nCount > length() )
1260     nCount = length();
1261
1262   wxString dest(*this, length() - nCount, nCount);
1263   if ( dest.length() != nCount ) {
1264     wxFAIL_MSG( _T("out of memory in wxString::Right") );
1265   }
1266   return dest;
1267 }
1268
1269 // get all characters after the last occurence of ch
1270 // (returns the whole string if ch not found)
1271 wxString wxString::AfterLast(wxChar ch) const
1272 {
1273   wxString str;
1274   int iPos = Find(ch, TRUE);
1275   if ( iPos == wxNOT_FOUND )
1276     str = *this;
1277   else
1278     str = c_str() + iPos + 1;
1279
1280   return str;
1281 }
1282
1283 // extract nCount first (leftmost) characters
1284 wxString wxString::Left(size_t nCount) const
1285 {
1286   if ( nCount > length() )
1287     nCount = length();
1288
1289   wxString dest(*this, 0, nCount);
1290   if ( dest.length() != nCount ) {
1291     wxFAIL_MSG( _T("out of memory in wxString::Left") );
1292   }
1293   return dest;
1294 }
1295
1296 // get all characters before the first occurence of ch
1297 // (returns the whole string if ch not found)
1298 wxString wxString::BeforeFirst(wxChar ch) const
1299 {
1300   int iPos = Find(ch);
1301   if ( iPos == wxNOT_FOUND ) iPos = length();
1302   return wxString(*this, 0, iPos);
1303 }
1304
1305 /// get all characters before the last occurence of ch
1306 /// (returns empty string if ch not found)
1307 wxString wxString::BeforeLast(wxChar ch) const
1308 {
1309   wxString str;
1310   int iPos = Find(ch, TRUE);
1311   if ( iPos != wxNOT_FOUND && iPos != 0 )
1312     str = wxString(c_str(), iPos);
1313
1314   return str;
1315 }
1316
1317 /// get all characters after the first occurence of ch
1318 /// (returns empty string if ch not found)
1319 wxString wxString::AfterFirst(wxChar ch) const
1320 {
1321   wxString str;
1322   int iPos = Find(ch);
1323   if ( iPos != wxNOT_FOUND )
1324     str = c_str() + iPos + 1;
1325
1326   return str;
1327 }
1328
1329 // replace first (or all) occurences of some substring with another one
1330 size_t
1331 wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
1332 {
1333     // if we tried to replace an empty string we'd enter an infinite loop below
1334     wxCHECK_MSG( szOld && *szOld && szNew, 0,
1335                  _T("wxString::Replace(): invalid parameter") );
1336
1337   size_t uiCount = 0;   // count of replacements made
1338
1339   size_t uiOldLen = wxStrlen(szOld);
1340
1341   wxString strTemp;
1342   const wxChar *pCurrent = c_str();
1343   const wxChar *pSubstr;
1344   while ( *pCurrent != wxT('\0') ) {
1345     pSubstr = wxStrstr(pCurrent, szOld);
1346     if ( pSubstr == NULL ) {
1347       // strTemp is unused if no replacements were made, so avoid the copy
1348       if ( uiCount == 0 )
1349         return 0;
1350
1351       strTemp += pCurrent;    // copy the rest
1352       break;                  // exit the loop
1353     }
1354     else {
1355       // take chars before match
1356       size_type len = strTemp.length();
1357       strTemp.append(pCurrent, pSubstr - pCurrent);
1358       if ( strTemp.length() != (size_t)(len + pSubstr - pCurrent) ) {
1359         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
1360         return 0;
1361       }
1362       strTemp += szNew;
1363       pCurrent = pSubstr + uiOldLen;  // restart after match
1364
1365       uiCount++;
1366
1367       // stop now?
1368       if ( !bReplaceAll ) {
1369         strTemp += pCurrent;    // copy the rest
1370         break;                  // exit the loop
1371       }
1372     }
1373   }
1374
1375   // only done if there were replacements, otherwise would have returned above
1376   swap(strTemp);
1377
1378   return uiCount;
1379 }
1380
1381 bool wxString::IsAscii() const
1382 {
1383   const wxChar *s = (const wxChar*) *this;
1384   while(*s){
1385     if(!isascii(*s)) return(FALSE);
1386     s++;
1387   }
1388   return(TRUE);
1389 }
1390
1391 bool wxString::IsWord() const
1392 {
1393   const wxChar *s = (const wxChar*) *this;
1394   while(*s){
1395     if(!wxIsalpha(*s)) return(FALSE);
1396     s++;
1397   }
1398   return(TRUE);
1399 }
1400
1401 bool wxString::IsNumber() const
1402 {
1403   const wxChar *s = (const wxChar*) *this;
1404   if (wxStrlen(s))
1405      if ((s[0] == '-') || (s[0] == '+')) s++;
1406   while(*s){
1407     if(!wxIsdigit(*s)) return(FALSE);
1408     s++;
1409   }
1410   return(TRUE);
1411 }
1412
1413 wxString wxString::Strip(stripType w) const
1414 {
1415     wxString s = *this;
1416     if ( w & leading ) s.Trim(FALSE);
1417     if ( w & trailing ) s.Trim(TRUE);
1418     return s;
1419 }
1420
1421 // ---------------------------------------------------------------------------
1422 // case conversion
1423 // ---------------------------------------------------------------------------
1424
1425 wxString& wxString::MakeUpper()
1426 {
1427   for ( iterator it = begin(), en = end(); it != en; ++it )
1428     *it = (wxChar)wxToupper(*it);
1429
1430   return *this;
1431 }
1432
1433 wxString& wxString::MakeLower()
1434 {
1435   for ( iterator it = begin(), en = end(); it != en; ++it )
1436     *it = (wxChar)wxTolower(*it);
1437
1438   return *this;
1439 }
1440
1441 // ---------------------------------------------------------------------------
1442 // trimming and padding
1443 // ---------------------------------------------------------------------------
1444
1445 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1446 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1447 // live with this by checking that the character is a 7 bit one - even if this
1448 // may fail to detect some spaces (I don't know if Unicode doesn't have
1449 // space-like symbols somewhere except in the first 128 chars), it is arguably
1450 // still better than trimming away accented letters
1451 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1452
1453 // trims spaces (in the sense of isspace) from left or right side
1454 wxString& wxString::Trim(bool bFromRight)
1455 {
1456   // first check if we're going to modify the string at all
1457   if ( !IsEmpty() &&
1458        (
1459         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1460         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1461        )
1462      )
1463   {
1464     if ( bFromRight )
1465     {
1466       // find last non-space character
1467       iterator psz = begin() + length() - 1;
1468       while ( wxSafeIsspace(*psz) && (psz >= begin()) )
1469         psz--;
1470
1471       // truncate at trailing space start
1472       *++psz = wxT('\0');
1473       erase(psz, end());
1474     }
1475     else
1476     {
1477       // find first non-space character
1478       iterator psz = begin();
1479       while ( wxSafeIsspace(*psz) )
1480         psz++;
1481
1482       // fix up data and length
1483       erase(begin(), psz);
1484     }
1485   }
1486
1487   return *this;
1488 }
1489
1490 // adds nCount characters chPad to the string from either side
1491 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1492 {
1493   wxString s(chPad, nCount);
1494
1495   if ( bFromRight )
1496     *this += s;
1497   else
1498   {
1499     s += *this;
1500     swap(s);
1501   }
1502
1503   return *this;
1504 }
1505
1506 // truncate the string
1507 wxString& wxString::Truncate(size_t uiLen)
1508 {
1509   if ( uiLen < Len() ) {
1510     erase(begin() + uiLen, end());
1511   }
1512   //else: nothing to do, string is already short enough
1513
1514   return *this;
1515 }
1516
1517 // ---------------------------------------------------------------------------
1518 // finding (return wxNOT_FOUND if not found and index otherwise)
1519 // ---------------------------------------------------------------------------
1520
1521 // find a character
1522 int wxString::Find(wxChar ch, bool bFromEnd) const
1523 {
1524   size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
1525
1526   return (idx == npos) ? wxNOT_FOUND : (int)idx;
1527 }
1528
1529 // find a sub-string (like strstr)
1530 int wxString::Find(const wxChar *pszSub) const
1531 {
1532   size_type idx = find(pszSub);
1533
1534   return (idx == npos) ? wxNOT_FOUND : (int)idx;
1535 }
1536
1537 // ----------------------------------------------------------------------------
1538 // conversion to numbers
1539 // ----------------------------------------------------------------------------
1540
1541 bool wxString::ToLong(long *val, int base) const
1542 {
1543     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1544     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1545
1546     const wxChar *start = c_str();
1547     wxChar *end;
1548     *val = wxStrtol(start, &end, base);
1549
1550     // return TRUE only if scan was stopped by the terminating NUL and if the
1551     // string was not empty to start with
1552     return !*end && (end != start);
1553 }
1554
1555 bool wxString::ToULong(unsigned long *val, int base) const
1556 {
1557     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1558     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1559
1560     const wxChar *start = c_str();
1561     wxChar *end;
1562     *val = wxStrtoul(start, &end, base);
1563
1564     // return TRUE only if scan was stopped by the terminating NUL and if the
1565     // string was not empty to start with
1566     return !*end && (end != start);
1567 }
1568
1569 bool wxString::ToDouble(double *val) const
1570 {
1571     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1572
1573     const wxChar *start = c_str();
1574     wxChar *end;
1575     *val = wxStrtod(start, &end);
1576
1577     // return TRUE only if scan was stopped by the terminating NUL and if the
1578     // string was not empty to start with
1579     return !*end && (end != start);
1580 }
1581
1582 // ---------------------------------------------------------------------------
1583 // formatted output
1584 // ---------------------------------------------------------------------------
1585
1586 /* static */
1587 wxString wxString::Format(const wxChar *pszFormat, ...)
1588 {
1589     va_list argptr;
1590     va_start(argptr, pszFormat);
1591
1592     wxString s;
1593     s.PrintfV(pszFormat, argptr);
1594
1595     va_end(argptr);
1596
1597     return s;
1598 }
1599
1600 /* static */
1601 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1602 {
1603     wxString s;
1604     s.PrintfV(pszFormat, argptr);
1605     return s;
1606 }
1607
1608 int wxString::Printf(const wxChar *pszFormat, ...)
1609 {
1610   va_list argptr;
1611   va_start(argptr, pszFormat);
1612
1613   int iLen = PrintfV(pszFormat, argptr);
1614
1615   va_end(argptr);
1616
1617   return iLen;
1618 }
1619
1620 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1621 {
1622     int size = 1024;
1623     int len;
1624
1625     for ( ;; )
1626     {
1627         {
1628             wxStringBuffer tmp(*this, size + 1);
1629             wxChar* buf = tmp;
1630
1631             if ( !buf )
1632             {
1633                 // out of memory
1634                 return -1;
1635             }
1636
1637             len = wxVsnprintf(buf, size, pszFormat, argptr);
1638
1639             // some implementations of vsnprintf() don't NUL terminate
1640             // the string if there is not enough space for it so
1641             // always do it manually
1642             buf[size] = _T('\0');
1643         }
1644
1645         if ( len >= 0 )
1646         {
1647             // ok, there was enough space
1648             break;
1649         }
1650
1651         // still not enough, double it again
1652         size *= 2;
1653     }
1654
1655     // we could have overshot
1656     Shrink();
1657
1658     return Len();
1659 }
1660
1661 // ----------------------------------------------------------------------------
1662 // misc other operations
1663 // ----------------------------------------------------------------------------
1664
1665 // returns TRUE if the string matches the pattern which may contain '*' and
1666 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1667 // of them)
1668 bool wxString::Matches(const wxChar *pszMask) const
1669 {
1670     // I disable this code as it doesn't seem to be faster (in fact, it seems
1671     // to be much slower) than the old, hand-written code below and using it
1672     // here requires always linking with libregex even if the user code doesn't
1673     // use it
1674 #if 0 // wxUSE_REGEX
1675     // first translate the shell-like mask into a regex
1676     wxString pattern;
1677     pattern.reserve(wxStrlen(pszMask));
1678
1679     pattern += _T('^');
1680     while ( *pszMask )
1681     {
1682         switch ( *pszMask )
1683         {
1684             case _T('?'):
1685                 pattern += _T('.');
1686                 break;
1687
1688             case _T('*'):
1689                 pattern += _T(".*");
1690                 break;
1691
1692             case _T('^'):
1693             case _T('.'):
1694             case _T('$'):
1695             case _T('('):
1696             case _T(')'):
1697             case _T('|'):
1698             case _T('+'):
1699             case _T('\\'):
1700                 // these characters are special in a RE, quote them
1701                 // (however note that we don't quote '[' and ']' to allow
1702                 // using them for Unix shell like matching)
1703                 pattern += _T('\\');
1704                 // fall through
1705
1706             default:
1707                 pattern += *pszMask;
1708         }
1709
1710         pszMask++;
1711     }
1712     pattern += _T('$');
1713
1714     // and now use it
1715     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1716 #else // !wxUSE_REGEX
1717   // TODO: this is, of course, awfully inefficient...
1718
1719   // the char currently being checked
1720   const wxChar *pszTxt = c_str();
1721
1722   // the last location where '*' matched
1723   const wxChar *pszLastStarInText = NULL;
1724   const wxChar *pszLastStarInMask = NULL;
1725
1726 match:
1727   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1728     switch ( *pszMask ) {
1729       case wxT('?'):
1730         if ( *pszTxt == wxT('\0') )
1731           return FALSE;
1732
1733         // pszTxt and pszMask will be incremented in the loop statement
1734
1735         break;
1736
1737       case wxT('*'):
1738         {
1739           // remember where we started to be able to backtrack later
1740           pszLastStarInText = pszTxt;
1741           pszLastStarInMask = pszMask;
1742
1743           // ignore special chars immediately following this one
1744           // (should this be an error?)
1745           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1746             pszMask++;
1747
1748           // if there is nothing more, match
1749           if ( *pszMask == wxT('\0') )
1750             return TRUE;
1751
1752           // are there any other metacharacters in the mask?
1753           size_t uiLenMask;
1754           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1755
1756           if ( pEndMask != NULL ) {
1757             // we have to match the string between two metachars
1758             uiLenMask = pEndMask - pszMask;
1759           }
1760           else {
1761             // we have to match the remainder of the string
1762             uiLenMask = wxStrlen(pszMask);
1763           }
1764
1765           wxString strToMatch(pszMask, uiLenMask);
1766           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1767           if ( pMatch == NULL )
1768             return FALSE;
1769
1770           // -1 to compensate "++" in the loop
1771           pszTxt = pMatch + uiLenMask - 1;
1772           pszMask += uiLenMask - 1;
1773         }
1774         break;
1775
1776       default:
1777         if ( *pszMask != *pszTxt )
1778           return FALSE;
1779         break;
1780     }
1781   }
1782
1783   // match only if nothing left
1784   if ( *pszTxt == wxT('\0') )
1785     return TRUE;
1786
1787   // if we failed to match, backtrack if we can
1788   if ( pszLastStarInText ) {
1789     pszTxt = pszLastStarInText + 1;
1790     pszMask = pszLastStarInMask;
1791
1792     pszLastStarInText = NULL;
1793
1794     // don't bother resetting pszLastStarInMask, it's unnecessary
1795
1796     goto match;
1797   }
1798
1799   return FALSE;
1800 #endif // wxUSE_REGEX/!wxUSE_REGEX
1801 }
1802
1803 // Count the number of chars
1804 int wxString::Freq(wxChar ch) const
1805 {
1806     int count = 0;
1807     int len = Len();
1808     for (int i = 0; i < len; i++)
1809     {
1810         if (GetChar(i) == ch)
1811             count ++;
1812     }
1813     return count;
1814 }
1815
1816 // convert to upper case, return the copy of the string
1817 wxString wxString::Upper() const
1818 { wxString s(*this); return s.MakeUpper(); }
1819
1820 // convert to lower case, return the copy of the string
1821 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1822
1823 int wxString::sprintf(const wxChar *pszFormat, ...)
1824   {
1825     va_list argptr;
1826     va_start(argptr, pszFormat);
1827     int iLen = PrintfV(pszFormat, argptr);
1828     va_end(argptr);
1829     return iLen;
1830   }
1831
1832 // ============================================================================
1833 // ArrayString
1834 // ============================================================================
1835
1836 #include "wx/arrstr.h"
1837
1838 #if !wxUSE_STL
1839
1840 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1841 #define   ARRAY_MAXSIZE_INCREMENT       4096
1842
1843 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1844 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1845 #endif
1846
1847 #define   STRING(p)   ((wxString *)(&(p)))
1848
1849 // ctor
1850 void wxArrayString::Init(bool autoSort)
1851 {
1852   m_nSize  =
1853   m_nCount = 0;
1854   m_pItems = (wxChar **) NULL;
1855   m_autoSort = autoSort;
1856 }
1857
1858 // copy ctor
1859 wxArrayString::wxArrayString(const wxArrayString& src)
1860 {
1861   Init(src.m_autoSort);
1862
1863   *this = src;
1864 }
1865
1866 // assignment operator
1867 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1868 {
1869   if ( m_nSize > 0 )
1870     Clear();
1871
1872   Copy(src);
1873
1874   m_autoSort = src.m_autoSort;
1875
1876   return *this;
1877 }
1878
1879 void wxArrayString::Copy(const wxArrayString& src)
1880 {
1881   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1882     Alloc(src.m_nCount);
1883
1884   for ( size_t n = 0; n < src.m_nCount; n++ )
1885     Add(src[n]);
1886 }
1887
1888 // grow the array
1889 void wxArrayString::Grow(size_t nIncrement)
1890 {
1891   // only do it if no more place
1892   if ( (m_nSize - m_nCount) < nIncrement ) {
1893     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1894     // be never resized!
1895     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1896       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1897     #endif
1898
1899     if ( m_nSize == 0 ) {
1900       // was empty, alloc some memory
1901       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1902       if (m_nSize < nIncrement)
1903           m_nSize = nIncrement;
1904       m_pItems = new wxChar *[m_nSize];
1905     }
1906     else {
1907       // otherwise when it's called for the first time, nIncrement would be 0
1908       // and the array would never be expanded
1909       // add 50% but not too much
1910       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1911                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1912       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1913         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1914       if ( nIncrement < ndefIncrement )
1915         nIncrement = ndefIncrement;
1916       m_nSize += nIncrement;
1917       wxChar **pNew = new wxChar *[m_nSize];
1918
1919       // copy data to new location
1920       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1921
1922       // delete old memory (but do not release the strings!)
1923       wxDELETEA(m_pItems);
1924
1925       m_pItems = pNew;
1926     }
1927   }
1928 }
1929
1930 void wxArrayString::Free()
1931 {
1932   for ( size_t n = 0; n < m_nCount; n++ ) {
1933     STRING(m_pItems[n])->GetStringData()->Unlock();
1934   }
1935 }
1936
1937 // deletes all the strings from the list
1938 void wxArrayString::Empty()
1939 {
1940   Free();
1941
1942   m_nCount = 0;
1943 }
1944
1945 // as Empty, but also frees memory
1946 void wxArrayString::Clear()
1947 {
1948   Free();
1949
1950   m_nSize  =
1951   m_nCount = 0;
1952
1953   wxDELETEA(m_pItems);
1954 }
1955
1956 // dtor
1957 wxArrayString::~wxArrayString()
1958 {
1959   Free();
1960
1961   wxDELETEA(m_pItems);
1962 }
1963
1964 // pre-allocates memory (frees the previous data!)
1965 void wxArrayString::Alloc(size_t nSize)
1966 {
1967   // only if old buffer was not big enough
1968   if ( nSize > m_nSize ) {
1969     Free();
1970     wxDELETEA(m_pItems);
1971     m_pItems = new wxChar *[nSize];
1972     m_nSize  = nSize;
1973   }
1974
1975   m_nCount = 0;
1976 }
1977
1978 // minimizes the memory usage by freeing unused memory
1979 void wxArrayString::Shrink()
1980 {
1981   // only do it if we have some memory to free
1982   if( m_nCount < m_nSize ) {
1983     // allocates exactly as much memory as we need
1984     wxChar **pNew = new wxChar *[m_nCount];
1985
1986     // copy data to new location
1987     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1988     delete [] m_pItems;
1989     m_pItems = pNew;
1990   }
1991 }
1992
1993 #if WXWIN_COMPATIBILITY_2_4
1994
1995 // return a wxString[] as required for some control ctors.
1996 wxString* wxArrayString::GetStringArray() const
1997 {
1998     wxString *array = 0;
1999
2000     if( m_nCount > 0 )
2001     {
2002         array = new wxString[m_nCount];
2003         for( size_t i = 0; i < m_nCount; i++ )
2004             array[i] = m_pItems[i];
2005     }
2006
2007     return array;
2008 }
2009
2010 #endif // WXWIN_COMPATIBILITY_2_4
2011
2012 // searches the array for an item (forward or backwards)
2013 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
2014 {
2015   if ( m_autoSort ) {
2016     // use binary search in the sorted array
2017     wxASSERT_MSG( bCase && !bFromEnd,
2018                   wxT("search parameters ignored for auto sorted array") );
2019
2020     size_t i,
2021            lo = 0,
2022            hi = m_nCount;
2023     int res;
2024     while ( lo < hi ) {
2025       i = (lo + hi)/2;
2026
2027       res = wxStrcmp(sz, m_pItems[i]);
2028       if ( res < 0 )
2029         hi = i;
2030       else if ( res > 0 )
2031         lo = i + 1;
2032       else
2033         return i;
2034     }
2035
2036     return wxNOT_FOUND;
2037   }
2038   else {
2039     // use linear search in unsorted array
2040     if ( bFromEnd ) {
2041       if ( m_nCount > 0 ) {
2042         size_t ui = m_nCount;
2043         do {
2044           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
2045             return ui;
2046         }
2047         while ( ui != 0 );
2048       }
2049     }
2050     else {
2051       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2052         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
2053           return ui;
2054       }
2055     }
2056   }
2057
2058   return wxNOT_FOUND;
2059 }
2060
2061 // add item at the end
2062 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
2063 {
2064   if ( m_autoSort ) {
2065     // insert the string at the correct position to keep the array sorted
2066     size_t i,
2067            lo = 0,
2068            hi = m_nCount;
2069     int res;
2070     while ( lo < hi ) {
2071       i = (lo + hi)/2;
2072
2073       res = wxStrcmp(str, m_pItems[i]);
2074       if ( res < 0 )
2075         hi = i;
2076       else if ( res > 0 )
2077         lo = i + 1;
2078       else {
2079         lo = hi = i;
2080         break;
2081       }
2082     }
2083
2084     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2085
2086     Insert(str, lo, nInsert);
2087
2088     return (size_t)lo;
2089   }
2090   else {
2091     wxASSERT( str.GetStringData()->IsValid() );
2092
2093     Grow(nInsert);
2094
2095     for (size_t i = 0; i < nInsert; i++)
2096     {
2097         // the string data must not be deleted!
2098         str.GetStringData()->Lock();
2099
2100         // just append
2101         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2102     }
2103     size_t ret = m_nCount;
2104     m_nCount += nInsert;
2105     return ret;
2106   }
2107 }
2108
2109 // add item at the given position
2110 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2111 {
2112   wxASSERT( str.GetStringData()->IsValid() );
2113
2114   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2115   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2116                wxT("array size overflow in wxArrayString::Insert") );
2117
2118   Grow(nInsert);
2119
2120   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2121           (m_nCount - nIndex)*sizeof(wxChar *));
2122
2123   for (size_t i = 0; i < nInsert; i++)
2124   {
2125       str.GetStringData()->Lock();
2126       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2127   }
2128   m_nCount += nInsert;
2129 }
2130
2131 // expand the array
2132 void wxArrayString::SetCount(size_t count)
2133 {
2134     Alloc(count);
2135
2136     wxString s;
2137     while ( m_nCount < count )
2138         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2139 }
2140
2141 // removes item from array (by index)
2142 void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove)
2143 {
2144   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2145   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2146                wxT("removing too many elements in wxArrayString::Remove") );
2147
2148   // release our lock
2149   for (size_t i = 0; i < nRemove; i++)
2150       Item(nIndex + i).GetStringData()->Unlock();
2151
2152   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2153           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2154   m_nCount -= nRemove;
2155 }
2156
2157 // removes item from array (by value)
2158 void wxArrayString::Remove(const wxChar *sz)
2159 {
2160   int iIndex = Index(sz);
2161
2162   wxCHECK_RET( iIndex != wxNOT_FOUND,
2163                wxT("removing inexistent element in wxArrayString::Remove") );
2164
2165   RemoveAt(iIndex);
2166 }
2167
2168 // ----------------------------------------------------------------------------
2169 // sorting
2170 // ----------------------------------------------------------------------------
2171
2172 // we can only sort one array at a time with the quick-sort based
2173 // implementation
2174 #if wxUSE_THREADS
2175   // need a critical section to protect access to gs_compareFunction and
2176   // gs_sortAscending variables
2177   static wxCriticalSection *gs_critsectStringSort = NULL;
2178
2179   // call this before the value of the global sort vars is changed/after
2180   // you're finished with them
2181   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2182                            gs_critsectStringSort = new wxCriticalSection;     \
2183                            gs_critsectStringSort->Enter()
2184   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2185                            delete gs_critsectStringSort;                      \
2186                            gs_critsectStringSort = NULL
2187 #else // !threads
2188   #define START_SORT()
2189   #define END_SORT()
2190 #endif // wxUSE_THREADS
2191
2192 // function to use for string comparaison
2193 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2194
2195 // if we don't use the compare function, this flag tells us if we sort the
2196 // array in ascending or descending order
2197 static bool gs_sortAscending = TRUE;
2198
2199 // function which is called by quick sort
2200 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2201 wxStringCompareFunction(const void *first, const void *second)
2202 {
2203   wxString *strFirst = (wxString *)first;
2204   wxString *strSecond = (wxString *)second;
2205
2206   if ( gs_compareFunction ) {
2207     return gs_compareFunction(*strFirst, *strSecond);
2208   }
2209   else {
2210     // maybe we should use wxStrcoll
2211     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2212
2213     return gs_sortAscending ? result : -result;
2214   }
2215 }
2216
2217 // sort array elements using passed comparaison function
2218 void wxArrayString::Sort(CompareFunction compareFunction)
2219 {
2220   START_SORT();
2221
2222   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2223   gs_compareFunction = compareFunction;
2224
2225   DoSort();
2226
2227   // reset it to NULL so that Sort(bool) will work the next time
2228   gs_compareFunction = NULL;
2229
2230   END_SORT();
2231 }
2232
2233 typedef  int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first, const void *second);
2234
2235 void wxArrayString::Sort(CompareFunction2 compareFunction)
2236 {
2237   qsort(m_pItems, m_nCount, sizeof(wxChar *), (wxStringCompareFn)compareFunction);
2238 }
2239
2240 void wxArrayString::Sort(bool reverseOrder)
2241 {
2242   Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending);
2243 }
2244
2245 void wxArrayString::DoSort()
2246 {
2247   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2248
2249   // just sort the pointers using qsort() - of course it only works because
2250   // wxString() *is* a pointer to its data
2251   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2252 }
2253
2254 bool wxArrayString::operator==(const wxArrayString& a) const
2255 {
2256     if ( m_nCount != a.m_nCount )
2257         return FALSE;
2258
2259     for ( size_t n = 0; n < m_nCount; n++ )
2260     {
2261         if ( Item(n) != a[n] )
2262             return FALSE;
2263     }
2264
2265     return TRUE;
2266 }
2267
2268 #endif // !wxUSE_STL
2269
2270 int wxStringSortAscending(wxString* s1, wxString* s2)
2271 {
2272     return wxStrcmp(s1->c_str(), s2->c_str());
2273 }
2274
2275 int wxStringSortDescending(wxString* s1, wxString* s2)
2276 {
2277     return -wxStrcmp(s1->c_str(), s2->c_str());
2278 }