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 #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 // Gives a duplicate symbol (presumably a case-insensitivity problem)
 472 #if !defined(__BORLANDC__)
 473 size_t wxStringBase::find(wxChar ch, size_t nStart) const
 474 {
 475   wxASSERT( nStart <= length() );
 476
 477   const wxChar *p = wxStrchr(c_str() + nStart, ch);
 478
 479   return p == NULL ? npos : p - c_str();
 480 }
 481 #endif
 482
 483 size_t wxStringBase::rfind(const wxStringBase& str, size_t nStart) const
 484 {
 485   wxASSERT( str.GetStringData()->IsValid() );
 486   wxASSERT( nStart == npos || nStart <= length() );
 487
 488   // TODO could be made much quicker than that
 489   const wxChar *p = c_str() + (nStart == npos ? length() : nStart);
 490   while ( p >= c_str() + str.length() ) {
 491     if ( wxStrncmp(p - str.length(), str.c_str(), str.length()) == 0 )
 492       return p - str.length() - c_str();
 493     p--;
 494   }
 495
 496   return npos;
 497 }
 498
 499 size_t wxStringBase::rfind(const wxChar* sz, size_t nStart, size_t n) const
 500 {
 501     return rfind(wxStringBase(sz, n), nStart);
 502 }
 503
 504 size_t wxStringBase::rfind(wxChar ch, size_t nStart) const
 505 {
 506     if ( nStart == npos )
 507     {
 508         nStart = length();
 509     }
 510     else
 511     {
 512         wxASSERT( nStart <= length() );
 513     }
 514
 515     const wxChar *p = wxStrrchr(c_str(), ch);
 516
 517     if ( p == NULL )
 518         return npos;
 519
 520     size_t result = p - c_str();
 521     return ( result > nStart ) ? npos : result;
 522 }
 523
 524 size_t wxStringBase::find_first_of(const wxChar* sz, size_t nStart) const
 525 {
 526     const wxChar *start = c_str() + nStart;
 527     const wxChar *firstOf = wxStrpbrk(start, sz);
 528     if ( firstOf )
 529         return firstOf - c_str();
 530     else
 531         return npos;
 532 }
 533
 534 size_t wxStringBase::find_last_of(const wxChar* sz, size_t nStart) const
 535 {
 536     if ( nStart == npos )
 537     {
 538         nStart = length();
 539     }
 540     else
 541     {
 542         wxASSERT( nStart <= length() );
 543     }
 544
 545     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
 546     {
 547         if ( wxStrchr(sz, *p) )
 548             return p - c_str();
 549     }
 550
 551     return npos;
 552 }
 553
 554 size_t wxStringBase::find_first_not_of(const wxChar* sz, size_t nStart) const
 555 {
 556     if ( nStart == npos )
 557     {
 558         nStart = length();
 559     }
 560     else
 561     {
 562         wxASSERT( nStart <= length() );
 563     }
 564
 565     size_t nAccept = wxStrspn(c_str() + nStart, sz);
 566     if ( nAccept >= length() - nStart )
 567         return npos;
 568     else
 569         return nAccept;
 570 }
 571
 572 size_t wxStringBase::find_first_not_of(wxChar ch, size_t nStart) const
 573 {
 574     wxASSERT( nStart <= length() );
 575
 576     for ( const wxChar *p = c_str() + nStart; *p; p++ )
 577     {
 578         if ( *p != ch )
 579             return p - c_str();
 580     }
 581
 582     return npos;
 583 }
 584
 585 size_t wxStringBase::find_last_not_of(const wxChar* sz, size_t nStart) const
 586 {
 587     if ( nStart == npos )
 588     {
 589         nStart = length();
 590     }
 591     else
 592     {
 593         wxASSERT( nStart <= length() );
 594     }
 595
 596     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
 597     {
 598         if ( !wxStrchr(sz, *p) )
 599             return p - c_str();
 600     }
 601
 602     return npos;
 603 }
 604
 605 size_t wxStringBase::find_last_not_of(wxChar ch, size_t nStart) const
 606 {
 607     if ( nStart == npos )
 608     {
 609         nStart = length();
 610     }
 611     else
 612     {
 613         wxASSERT( nStart <= length() );
 614     }
 615
 616     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
 617     {
 618         if ( *p != ch )
 619             return p - c_str();
 620     }
 621
 622     return npos;
 623 }
 624
 625 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 626                                     const wxChar *sz)
 627 {
 628   wxASSERT_MSG( nStart <= length(),
 629                 _T("index out of bounds in wxStringBase::replace") );
 630   size_t strLen = length() - nStart;
 631   nLen = strLen < nLen ? strLen : nLen;
 632
 633   wxStringBase strTmp;
 634   strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs
 635
 636   if ( nStart != 0 )
 637     strTmp.append(c_str(), nStart);
 638   strTmp.append(sz);
 639   strTmp.append(c_str() + nStart + nLen);
 640
 641   swap(strTmp);
 642   return *this;
 643 }
 644
 645 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 646                                     size_t nCount, wxChar ch)
 647 {
 648   return replace(nStart, nLen, wxStringBase(ch, nCount).c_str());
 649 }
 650
 651 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 652                                     const wxStringBase& str,
 653                                     size_t nStart2, size_t nLen2)
 654 {
 655   return replace(nStart, nLen, str.substr(nStart2, nLen2));
 656 }
 657
 658 wxStringBase& wxStringBase::replace(size_t nStart, size_t nLen,
 659                                     const wxChar* sz, size_t nCount)
 660 {
 661   return replace(nStart, nLen, wxStringBase(sz, nCount).c_str());
 662 }
 663
 664 wxStringBase wxStringBase::substr(size_t nStart, size_t nLen) const
 665 {
 666   if ( nLen == npos )
 667     nLen = length() - nStart;
 668   return wxStringBase(*this, nStart, nLen);
 669 }
 670
 671 // assigns one string to another
 672 wxStringBase& wxStringBase::operator=(const wxStringBase& stringSrc)
 673 {
 674   wxASSERT( stringSrc.GetStringData()->IsValid() );
 675
 676   // don't copy string over itself
 677   if ( m_pchData != stringSrc.m_pchData ) {
 678     if ( stringSrc.GetStringData()->IsEmpty() ) {
 679       Reinit();
 680     }
 681     else {
 682       // adjust references
 683       GetStringData()->Unlock();
 684       m_pchData = stringSrc.m_pchData;
 685       GetStringData()->Lock();
 686     }
 687   }
 688
 689   return *this;
 690 }
 691
 692 // assigns a single character
 693 wxStringBase& wxStringBase::operator=(wxChar ch)
 694 {
 695   if ( !AssignCopy(1, &ch) ) {
 696     wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(wxChar)") );
 697   }
 698   return *this;
 699 }
 700
 701 // assigns C string
 702 wxStringBase& wxStringBase::operator=(const wxChar *psz)
 703 {
 704   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 705     wxFAIL_MSG( _T("out of memory in wxStringBase::operator=(const wxChar *)") );
 706   }
 707   return *this;
 708 }
 709
 710 // helper function: does real copy
 711 bool wxStringBase::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 712 {
 713   if ( nSrcLen == 0 ) {
 714     Reinit();
 715   }
 716   else {
 717     if ( !AllocBeforeWrite(nSrcLen) ) {
 718       // allocation failure handled by caller
 719       return FALSE;
 720     }
 721     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 722     GetStringData()->nDataLength = nSrcLen;
 723     m_pchData[nSrcLen] = wxT('\0');
 724   }
 725   return TRUE;
 726 }
 727
 728 // ---------------------------------------------------------------------------
 729 // string concatenation
 730 // ---------------------------------------------------------------------------
 731
 732 // add something to this string
 733 bool wxStringBase::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData,
 734                               size_t nMaxLen)
 735 {
 736   STATISTICS_ADD(SummandLength, nSrcLen);
 737
 738   nSrcLen = nSrcLen < nMaxLen ? nSrcLen : nMaxLen;
 739
 740   // concatenating an empty string is a NOP
 741   if ( nSrcLen > 0 ) {
 742     wxStringData *pData = GetStringData();
 743     size_t nLen = pData->nDataLength;
 744     size_t nNewLen = nLen + nSrcLen;
 745
 746     // alloc new buffer if current is too small
 747     if ( pData->IsShared() ) {
 748       STATISTICS_ADD(ConcatHit, 0);
 749
 750       // we have to allocate another buffer
 751       wxStringData* pOldData = GetStringData();
 752       if ( !AllocBuffer(nNewLen) ) {
 753           // allocation failure handled by caller
 754           return FALSE;
 755       }
 756       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 757       pOldData->Unlock();
 758     }
 759     else if ( nNewLen > pData->nAllocLength ) {
 760       STATISTICS_ADD(ConcatHit, 0);
 761
 762       reserve(nNewLen);
 763       // we have to grow the buffer
 764       if ( capacity() < nNewLen ) {
 765           // allocation failure handled by caller
 766           return FALSE;
 767       }
 768     }
 769     else {
 770       STATISTICS_ADD(ConcatHit, 1);
 771
 772       // the buffer is already big enough
 773     }
 774
 775     // should be enough space
 776     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 777
 778     // fast concatenation - all is done in our buffer
 779     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 780
 781     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 782     GetStringData()->nDataLength = nNewLen; // and fix the length
 783   }
 784   //else: the string to append was empty
 785   return TRUE;
 786 }
 787
 788 // ---------------------------------------------------------------------------
 789 // simple sub-string extraction
 790 // ---------------------------------------------------------------------------
 791
 792 // helper function: clone the data attached to this string
 793 bool wxStringBase::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 794 {
 795   if ( nCopyLen == 0 ) {
 796     dest.Init();
 797   }
 798   else {
 799     if ( !dest.AllocBuffer(nCopyLen) ) {
 800       // allocation failure handled by caller
 801       return FALSE;
 802     }
 803     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 804   }
 805   return TRUE;
 806 }
 807
 808 #endif // !wxUSE_STL
 809
 810 #if !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
 811
 812 #if !wxUSE_STL
 813     #define STRINGCLASS wxStringBase
 814 #else
 815     #define STRINGCLASS wxString
 816 #endif
 817
 818 static inline int wxDoCmp(const wxChar* s1, size_t l1,
 819                           const wxChar* s2, size_t l2)
 820 {
 821     if( l1 == l2 )
 822         return wxStrncmp(s1, s2, l1);
 823     else if( l1 < l2 )
 824     {
 825         int ret = wxStrncmp(s1, s2, l1);
 826         return ret == 0 ? -1 : ret;
 827     }
 828     else if( l1 > l2 )
 829     {
 830         int ret = wxStrncmp(s1, s2, l2);
 831         return ret == 0 ? +1 : ret;
 832     }
 833
 834     wxFAIL;   // must never get there
 835     return 0; // quiet compilers
 836 }
 837
 838 #if wxUSE_STL
 839
 840 int STRINGCLASS::compare(const wxStringBase& str) const
 841 {
 842     return ::wxDoCmp(data(), length(), str.data(), str.length());
 843 }
 844
 845 #endif
 846
 847 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 848                          const wxStringBase& str) const
 849 {
 850     wxASSERT(nStart <= length());
 851     size_type strLen = length() - nStart;
 852     nLen = strLen < nLen ? strLen : nLen;
 853     return ::wxDoCmp(data() + nStart, nLen, str.data(), str.length());
 854 }
 855
 856 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 857                          const wxStringBase& str,
 858                          size_t nStart2, size_t nLen2) const
 859 {
 860     wxASSERT(nStart <= length());
 861     wxASSERT(nStart2 <= str.length());
 862     size_type strLen  =     length() - nStart,
 863               strLen2 = str.length() - nStart2;
 864     nLen  = strLen  < nLen  ? strLen  : nLen;
 865     nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
 866     return ::wxDoCmp(data() + nStart, nLen, str.data() + nStart2, nLen2);
 867 }
 868
 869 #if wxUSE_STL
 870
 871 int STRINGCLASS::compare(const wxChar* sz) const
 872 {
 873     size_t nLen = wxStrlen(sz);
 874     return ::wxDoCmp(data(), length(), sz, nLen);
 875 }
 876
 877 #endif
 878
 879 int STRINGCLASS::compare(size_t nStart, size_t nLen,
 880                          const wxChar* sz, size_t nCount) const
 881 {
 882     wxASSERT(nStart <= length());
 883     size_type strLen = length() - nStart;
 884     nLen = strLen < nLen ? strLen : nLen;
 885     if( nCount == npos )
 886         nCount = wxStrlen(sz);
 887
 888     return ::wxDoCmp(data() + nStart, nLen, sz, nCount);
 889 }
 890
 891 #undef STRINGCLASS
 892
 893 #endif // !wxUSE_STL || !defined(HAVE_STD_STRING_COMPARE)
 894
 895 // ===========================================================================
 896 // wxString class core
 897 // ===========================================================================
 898
 899 // ---------------------------------------------------------------------------
 900 // construction
 901 // ---------------------------------------------------------------------------
 902
 903 #if wxUSE_UNICODE
 904
 905 // from multibyte string
 906 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 907 {
 908     // first get the size of the buffer we need
 909     size_t nLen;
 910     if ( psz )
 911     {
 912         // calculate the needed size ourselves or use the provided one
 913         nLen = nLength == npos ? conv.MB2WC(NULL, psz, 0) : nLength;
 914     }
 915     else
 916     {
 917         // nothing to convert
 918         nLen = 0;
 919     }
 920
 921     // anything to do?
 922     if ( (nLen != 0) && (nLen != (size_t)-1) )
 923     {
 924         if ( !Alloc(nLen) )
 925         {
 926             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 927         }
 928         else
 929         {
 930             wxWCharBuffer buf(nLen + 1);
 931             // MB2WC wants the buffer size, not the string length hence +1
 932             nLen = conv.MB2WC(buf.data(), psz, nLen + 1);
 933
 934             if ( nLen != (size_t)-1 )
 935             {
 936                 // initialized ok, set the real length as nLength specified by
 937                 // the caller could be greater than the real string length
 938                 assign(buf.data(), nLen);
 939                 return;
 940             }
 941             //else: the conversion failed -- leave the string empty (what else?)
 942         }
 943     }
 944 }
 945
 946 #else // ANSI
 947
 948 #if wxUSE_WCHAR_T
 949 // from wide string
 950 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 951 {
 952     // first get the size of the buffer we need
 953     size_t nLen;
 954     if ( pwz )
 955     {
 956         // calculate the needed size ourselves or use the provided one
 957         nLen = nLength == npos ? conv.WC2MB(NULL, pwz, 0) : nLength;
 958     }
 959     else
 960     {
 961         // nothing to convert
 962         nLen = 0;
 963     }
 964
 965     // anything to do?
 966     if ( (nLen != 0) && (nLen != (size_t)-1) )
 967     {
 968         if ( !Alloc(nLen) )
 969         {
 970             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 971         }
 972         else
 973         {
 974             wxCharBuffer buf(nLen);
 975             // WC2MB wants the buffer size, not the string length
 976             if ( conv.WC2MB(buf.data(), pwz, nLen + 1) != (size_t)-1 )
 977             {
 978                 // initialized ok
 979                 assign(buf.data(), nLen);
 980                 return;
 981             }
 982             //else: the conversion failed -- leave the string empty (what else?)
 983         }
 984     }
 985
 986     // leave empty
 987 }
 988 #endif // wxUSE_WCHAR_T
 989
 990 #endif // Unicode/ANSI
 991
 992 // shrink to minimal size (releasing extra memory)
 993 bool wxString::Shrink()
 994 {
 995   wxString tmp(begin(), end());
 996   swap(tmp);
 997   return tmp.length() == length();
 998 }
 999
1000 #if !wxUSE_STL
1001 // get the pointer to writable buffer of (at least) nLen bytes
1002 wxChar *wxString::GetWriteBuf(size_t nLen)
1003 {
1004   if ( !AllocBeforeWrite(nLen) ) {
1005     // allocation failure handled by caller
1006     return NULL;
1007   }
1008
1009   wxASSERT( GetStringData()->nRefs == 1 );
1010   GetStringData()->Validate(FALSE);
1011
1012   return m_pchData;
1013 }
1014
1015 // put string back in a reasonable state after GetWriteBuf
1016 void wxString::UngetWriteBuf()
1017 {
1018   GetStringData()->nDataLength = wxStrlen(m_pchData);
1019   GetStringData()->Validate(TRUE);
1020 }
1021
1022 void wxString::UngetWriteBuf(size_t nLen)
1023 {
1024   GetStringData()->nDataLength = nLen;
1025   GetStringData()->Validate(TRUE);
1026 }
1027 #endif
1028
1029 // ---------------------------------------------------------------------------
1030 // data access
1031 // ---------------------------------------------------------------------------
1032
1033 // all functions are inline in string.h
1034
1035 // ---------------------------------------------------------------------------
1036 // assignment operators
1037 // ---------------------------------------------------------------------------
1038
1039 #if !wxUSE_UNICODE
1040
1041 // same as 'signed char' variant
1042 wxString& wxString::operator=(const unsigned char* psz)
1043 {
1044   *this = (const char *)psz;
1045   return *this;
1046 }
1047
1048 #if wxUSE_WCHAR_T
1049 wxString& wxString::operator=(const wchar_t *pwz)
1050 {
1051   wxString str(pwz);
1052   swap(str);
1053   return *this;
1054 }
1055 #endif
1056
1057 #endif
1058
1059 /*
1060  * concatenation functions come in 5 flavours:
1061  *  string + string
1062  *  char   + string      and      string + char
1063  *  C str  + string      and      string + C str
1064  */
1065
1066 wxString operator+(const wxString& str1, const wxString& str2)
1067 {
1068 #if !wxUSE_STL
1069   wxASSERT( str1.GetStringData()->IsValid() );
1070   wxASSERT( str2.GetStringData()->IsValid() );
1071 #endif
1072
1073   wxString s = str1;
1074   s += str2;
1075
1076   return s;
1077 }
1078
1079 wxString operator+(const wxString& str, wxChar ch)
1080 {
1081 #if !wxUSE_STL
1082   wxASSERT( str.GetStringData()->IsValid() );
1083 #endif
1084
1085   wxString s = str;
1086   s += ch;
1087
1088   return s;
1089 }
1090
1091 wxString operator+(wxChar ch, const wxString& str)
1092 {
1093 #if !wxUSE_STL
1094   wxASSERT( str.GetStringData()->IsValid() );
1095 #endif
1096
1097   wxString s = ch;
1098   s += str;
1099
1100   return s;
1101 }
1102
1103 wxString operator+(const wxString& str, const wxChar *psz)
1104 {
1105 #if !wxUSE_STL
1106   wxASSERT( str.GetStringData()->IsValid() );
1107 #endif
1108
1109   wxString s;
1110   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
1111     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1112   }
1113   s = str;
1114   s += psz;
1115
1116   return s;
1117 }
1118
1119 wxString operator+(const wxChar *psz, const wxString& str)
1120 {
1121 #if !wxUSE_STL
1122   wxASSERT( str.GetStringData()->IsValid() );
1123 #endif
1124
1125   wxString s;
1126   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
1127     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1128   }
1129   s = psz;
1130   s += str;
1131
1132   return s;
1133 }
1134
1135 // ===========================================================================
1136 // other common string functions
1137 // ===========================================================================
1138
1139 #if wxUSE_UNICODE
1140
1141 wxString wxString::FromAscii(const char *ascii)
1142 {
1143     if (!ascii)
1144        return wxEmptyString;
1145
1146     size_t len = strlen( ascii );
1147     wxString res;
1148
1149     if ( len )
1150     {
1151         wxStringBuffer buf(res, len);
1152
1153         wchar_t *dest = buf;
1154
1155         for ( ;; )
1156         {
1157            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
1158                break;
1159         }
1160     }
1161
1162     return res;
1163 }
1164
1165 wxString wxString::FromAscii(const char ascii)
1166 {
1167     // What do we do with '\0' ?
1168
1169     wxString res;
1170     res += (wchar_t)(unsigned char) ascii;
1171
1172     return res;
1173 }
1174
1175 const wxCharBuffer wxString::ToAscii() const
1176 {
1177     // this will allocate enough space for the terminating NUL too
1178     wxCharBuffer buffer(length());
1179
1180     signed char *dest = (signed char *)buffer.data();
1181
1182     const wchar_t *pwc = c_str();
1183     for ( ;; )
1184     {
1185         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
1186
1187         // the output string can't have embedded NULs anyhow, so we can safely
1188         // stop at first of them even if we do have any
1189         if ( !*pwc++ )
1190             break;
1191     }
1192
1193     return buffer;
1194 }
1195
1196 #endif // Unicode
1197
1198 // extract string of length nCount starting at nFirst
1199 wxString wxString::Mid(size_t nFirst, size_t nCount) const
1200 {
1201   size_t nLen = length();
1202
1203   // default value of nCount is npos and means "till the end"
1204   if ( nCount == npos )
1205   {
1206     nCount = nLen - nFirst;
1207   }
1208
1209   // out-of-bounds requests return sensible things
1210   if ( nFirst + nCount > nLen )
1211   {
1212     nCount = nLen - nFirst;
1213   }
1214
1215   if ( nFirst > nLen )
1216   {
1217     // AllocCopy() will return empty string
1218     nCount = 0;
1219   }
1220
1221   wxString dest(*this, nFirst, nCount);
1222   if ( dest.length() != nCount ) {
1223       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
1224   }
1225
1226   return dest;
1227 }
1228
1229 // check that the string starts with prefix and return the rest of the string
1230 // in the provided pointer if it is not NULL, otherwise return FALSE
1231 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
1232 {
1233     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
1234
1235     // first check if the beginning of the string matches the prefix: note
1236     // that we don't have to check that we don't run out of this string as
1237     // when we reach the terminating NUL, either prefix string ends too (and
1238     // then it's ok) or we break out of the loop because there is no match
1239     const wxChar *p = c_str();
1240     while ( *prefix )
1241     {
1242         if ( *prefix++ != *p++ )
1243         {
1244             // no match
1245             return FALSE;
1246         }
1247     }
1248
1249     if ( rest )
1250     {
1251         // put the rest of the string into provided pointer
1252         *rest = p;
1253     }
1254
1255     return TRUE;
1256 }
1257
1258 // extract nCount last (rightmost) characters
1259 wxString wxString::Right(size_t nCount) const
1260 {
1261   if ( nCount > length() )
1262     nCount = length();
1263
1264   wxString dest(*this, length() - nCount, nCount);
1265   if ( dest.length() != nCount ) {
1266     wxFAIL_MSG( _T("out of memory in wxString::Right") );
1267   }
1268   return dest;
1269 }
1270
1271 // get all characters after the last occurence of ch
1272 // (returns the whole string if ch not found)
1273 wxString wxString::AfterLast(wxChar ch) const
1274 {
1275   wxString str;
1276   int iPos = Find(ch, TRUE);
1277   if ( iPos == wxNOT_FOUND )
1278     str = *this;
1279   else
1280     str = c_str() + iPos + 1;
1281
1282   return str;
1283 }
1284
1285 // extract nCount first (leftmost) characters
1286 wxString wxString::Left(size_t nCount) const
1287 {
1288   if ( nCount > length() )
1289     nCount = length();
1290
1291   wxString dest(*this, 0, nCount);
1292   if ( dest.length() != nCount ) {
1293     wxFAIL_MSG( _T("out of memory in wxString::Left") );
1294   }
1295   return dest;
1296 }
1297
1298 // get all characters before the first occurence of ch
1299 // (returns the whole string if ch not found)
1300 wxString wxString::BeforeFirst(wxChar ch) const
1301 {
1302   int iPos = Find(ch);
1303   if ( iPos == wxNOT_FOUND ) iPos = length();
1304   return wxString(*this, 0, iPos);
1305 }
1306
1307 /// get all characters before the last occurence of ch
1308 /// (returns empty string if ch not found)
1309 wxString wxString::BeforeLast(wxChar ch) const
1310 {
1311   wxString str;
1312   int iPos = Find(ch, TRUE);
1313   if ( iPos != wxNOT_FOUND && iPos != 0 )
1314     str = wxString(c_str(), iPos);
1315
1316   return str;
1317 }
1318
1319 /// get all characters after the first occurence of ch
1320 /// (returns empty string if ch not found)
1321 wxString wxString::AfterFirst(wxChar ch) const
1322 {
1323   wxString str;
1324   int iPos = Find(ch);
1325   if ( iPos != wxNOT_FOUND )
1326     str = c_str() + iPos + 1;
1327
1328   return str;
1329 }
1330
1331 // replace first (or all) occurences of some substring with another one
1332 size_t
1333 wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
1334 {
1335     // if we tried to replace an empty string we'd enter an infinite loop below
1336     wxCHECK_MSG( szOld && *szOld && szNew, 0,
1337                  _T("wxString::Replace(): invalid parameter") );
1338
1339   size_t uiCount = 0;   // count of replacements made
1340
1341   size_t uiOldLen = wxStrlen(szOld);
1342
1343   wxString strTemp;
1344   const wxChar *pCurrent = c_str();
1345   const wxChar *pSubstr;
1346   while ( *pCurrent != wxT('\0') ) {
1347     pSubstr = wxStrstr(pCurrent, szOld);
1348     if ( pSubstr == NULL ) {
1349       // strTemp is unused if no replacements were made, so avoid the copy
1350       if ( uiCount == 0 )
1351         return 0;
1352
1353       strTemp += pCurrent;    // copy the rest
1354       break;                  // exit the loop
1355     }
1356     else {
1357       // take chars before match
1358       size_type len = strTemp.length();
1359       strTemp.append(pCurrent, pSubstr - pCurrent);
1360       if ( strTemp.length() != (size_t)(len + pSubstr - pCurrent) ) {
1361         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
1362         return 0;
1363       }
1364       strTemp += szNew;
1365       pCurrent = pSubstr + uiOldLen;  // restart after match
1366
1367       uiCount++;
1368
1369       // stop now?
1370       if ( !bReplaceAll ) {
1371         strTemp += pCurrent;    // copy the rest
1372         break;                  // exit the loop
1373       }
1374     }
1375   }
1376
1377   // only done if there were replacements, otherwise would have returned above
1378   swap(strTemp);
1379
1380   return uiCount;
1381 }
1382
1383 bool wxString::IsAscii() const
1384 {
1385   const wxChar *s = (const wxChar*) *this;
1386   while(*s){
1387     if(!isascii(*s)) return(FALSE);
1388     s++;
1389   }
1390   return(TRUE);
1391 }
1392
1393 bool wxString::IsWord() const
1394 {
1395   const wxChar *s = (const wxChar*) *this;
1396   while(*s){
1397     if(!wxIsalpha(*s)) return(FALSE);
1398     s++;
1399   }
1400   return(TRUE);
1401 }
1402
1403 bool wxString::IsNumber() const
1404 {
1405   const wxChar *s = (const wxChar*) *this;
1406   if (wxStrlen(s))
1407      if ((s[0] == '-') || (s[0] == '+')) s++;
1408   while(*s){
1409     if(!wxIsdigit(*s)) return(FALSE);
1410     s++;
1411   }
1412   return(TRUE);
1413 }
1414
1415 wxString wxString::Strip(stripType w) const
1416 {
1417     wxString s = *this;
1418     if ( w & leading ) s.Trim(FALSE);
1419     if ( w & trailing ) s.Trim(TRUE);
1420     return s;
1421 }
1422
1423 // ---------------------------------------------------------------------------
1424 // case conversion
1425 // ---------------------------------------------------------------------------
1426
1427 wxString& wxString::MakeUpper()
1428 {
1429   for ( iterator it = begin(), en = end(); it != en; ++it )
1430     *it = (wxChar)wxToupper(*it);
1431
1432   return *this;
1433 }
1434
1435 wxString& wxString::MakeLower()
1436 {
1437   for ( iterator it = begin(), en = end(); it != en; ++it )
1438     *it = (wxChar)wxTolower(*it);
1439
1440   return *this;
1441 }
1442
1443 // ---------------------------------------------------------------------------
1444 // trimming and padding
1445 // ---------------------------------------------------------------------------
1446
1447 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1448 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1449 // live with this by checking that the character is a 7 bit one - even if this
1450 // may fail to detect some spaces (I don't know if Unicode doesn't have
1451 // space-like symbols somewhere except in the first 128 chars), it is arguably
1452 // still better than trimming away accented letters
1453 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1454
1455 // trims spaces (in the sense of isspace) from left or right side
1456 wxString& wxString::Trim(bool bFromRight)
1457 {
1458   // first check if we're going to modify the string at all
1459   if ( !IsEmpty() &&
1460        (
1461         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1462         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1463        )
1464      )
1465   {
1466     if ( bFromRight )
1467     {
1468       // find last non-space character
1469       iterator psz = begin() + length() - 1;
1470       while ( wxSafeIsspace(*psz) && (psz >= begin()) )
1471         psz--;
1472
1473       // truncate at trailing space start
1474       *++psz = wxT('\0');
1475       erase(psz, end());
1476     }
1477     else
1478     {
1479       // find first non-space character
1480       iterator psz = begin();
1481       while ( wxSafeIsspace(*psz) )
1482         psz++;
1483
1484       // fix up data and length
1485       erase(begin(), psz);
1486     }
1487   }
1488
1489   return *this;
1490 }
1491
1492 // adds nCount characters chPad to the string from either side
1493 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1494 {
1495   wxString s(chPad, nCount);
1496
1497   if ( bFromRight )
1498     *this += s;
1499   else
1500   {
1501     s += *this;
1502     swap(s);
1503   }
1504
1505   return *this;
1506 }
1507
1508 // truncate the string
1509 wxString& wxString::Truncate(size_t uiLen)
1510 {
1511   if ( uiLen < Len() ) {
1512     erase(begin() + uiLen, end());
1513   }
1514   //else: nothing to do, string is already short enough
1515
1516   return *this;
1517 }
1518
1519 // ---------------------------------------------------------------------------
1520 // finding (return wxNOT_FOUND if not found and index otherwise)
1521 // ---------------------------------------------------------------------------
1522
1523 // find a character
1524 int wxString::Find(wxChar ch, bool bFromEnd) const
1525 {
1526   size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
1527
1528   return (idx == npos) ? wxNOT_FOUND : (int)idx;
1529 }
1530
1531 // find a sub-string (like strstr)
1532 int wxString::Find(const wxChar *pszSub) const
1533 {
1534   size_type idx = find(pszSub);
1535
1536   return (idx == npos) ? wxNOT_FOUND : (int)idx;
1537 }
1538
1539 // ----------------------------------------------------------------------------
1540 // conversion to numbers
1541 // ----------------------------------------------------------------------------
1542
1543 bool wxString::ToLong(long *val, int base) const
1544 {
1545     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1546     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1547
1548     const wxChar *start = c_str();
1549     wxChar *end;
1550     *val = wxStrtol(start, &end, base);
1551
1552     // return TRUE only if scan was stopped by the terminating NUL and if the
1553     // string was not empty to start with
1554     return !*end && (end != start);
1555 }
1556
1557 bool wxString::ToULong(unsigned long *val, int base) const
1558 {
1559     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1560     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1561
1562     const wxChar *start = c_str();
1563     wxChar *end;
1564     *val = wxStrtoul(start, &end, base);
1565
1566     // return TRUE only if scan was stopped by the terminating NUL and if the
1567     // string was not empty to start with
1568     return !*end && (end != start);
1569 }
1570
1571 bool wxString::ToDouble(double *val) const
1572 {
1573     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1574
1575     const wxChar *start = c_str();
1576     wxChar *end;
1577     *val = wxStrtod(start, &end);
1578
1579     // return TRUE only if scan was stopped by the terminating NUL and if the
1580     // string was not empty to start with
1581     return !*end && (end != start);
1582 }
1583
1584 // ---------------------------------------------------------------------------
1585 // formatted output
1586 // ---------------------------------------------------------------------------
1587
1588 /* static */
1589 wxString wxString::Format(const wxChar *pszFormat, ...)
1590 {
1591     va_list argptr;
1592     va_start(argptr, pszFormat);
1593
1594     wxString s;
1595     s.PrintfV(pszFormat, argptr);
1596
1597     va_end(argptr);
1598
1599     return s;
1600 }
1601
1602 /* static */
1603 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1604 {
1605     wxString s;
1606     s.PrintfV(pszFormat, argptr);
1607     return s;
1608 }
1609
1610 int wxString::Printf(const wxChar *pszFormat, ...)
1611 {
1612   va_list argptr;
1613   va_start(argptr, pszFormat);
1614
1615   int iLen = PrintfV(pszFormat, argptr);
1616
1617   va_end(argptr);
1618
1619   return iLen;
1620 }
1621
1622 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1623 {
1624     int size = 1024;
1625     int len;
1626
1627     for ( ;; )
1628     {
1629         {
1630             wxStringBuffer tmp(*this, size + 1);
1631             wxChar* buf = tmp;
1632
1633             if ( !buf )
1634             {
1635                 // out of memory
1636                 return -1;
1637             }
1638
1639             len = wxVsnprintf(buf, size, pszFormat, argptr);
1640
1641             // some implementations of vsnprintf() don't NUL terminate
1642             // the string if there is not enough space for it so
1643             // always do it manually
1644             buf[size] = _T('\0');
1645         }
1646
1647         if ( len >= 0 )
1648         {
1649             // ok, there was enough space
1650             break;
1651         }
1652
1653         // still not enough, double it again
1654         size *= 2;
1655     }
1656
1657     // we could have overshot
1658     Shrink();
1659
1660     return Len();
1661 }
1662
1663 // ----------------------------------------------------------------------------
1664 // misc other operations
1665 // ----------------------------------------------------------------------------
1666
1667 // returns TRUE if the string matches the pattern which may contain '*' and
1668 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1669 // of them)
1670 bool wxString::Matches(const wxChar *pszMask) const
1671 {
1672     // I disable this code as it doesn't seem to be faster (in fact, it seems
1673     // to be much slower) than the old, hand-written code below and using it
1674     // here requires always linking with libregex even if the user code doesn't
1675     // use it
1676 #if 0 // wxUSE_REGEX
1677     // first translate the shell-like mask into a regex
1678     wxString pattern;
1679     pattern.reserve(wxStrlen(pszMask));
1680
1681     pattern += _T('^');
1682     while ( *pszMask )
1683     {
1684         switch ( *pszMask )
1685         {
1686             case _T('?'):
1687                 pattern += _T('.');
1688                 break;
1689
1690             case _T('*'):
1691                 pattern += _T(".*");
1692                 break;
1693
1694             case _T('^'):
1695             case _T('.'):
1696             case _T('$'):
1697             case _T('('):
1698             case _T(')'):
1699             case _T('|'):
1700             case _T('+'):
1701             case _T('\\'):
1702                 // these characters are special in a RE, quote them
1703                 // (however note that we don't quote '[' and ']' to allow
1704                 // using them for Unix shell like matching)
1705                 pattern += _T('\\');
1706                 // fall through
1707
1708             default:
1709                 pattern += *pszMask;
1710         }
1711
1712         pszMask++;
1713     }
1714     pattern += _T('$');
1715
1716     // and now use it
1717     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1718 #else // !wxUSE_REGEX
1719   // TODO: this is, of course, awfully inefficient...
1720
1721   // the char currently being checked
1722   const wxChar *pszTxt = c_str();
1723
1724   // the last location where '*' matched
1725   const wxChar *pszLastStarInText = NULL;
1726   const wxChar *pszLastStarInMask = NULL;
1727
1728 match:
1729   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1730     switch ( *pszMask ) {
1731       case wxT('?'):
1732         if ( *pszTxt == wxT('\0') )
1733           return FALSE;
1734
1735         // pszTxt and pszMask will be incremented in the loop statement
1736
1737         break;
1738
1739       case wxT('*'):
1740         {
1741           // remember where we started to be able to backtrack later
1742           pszLastStarInText = pszTxt;
1743           pszLastStarInMask = pszMask;
1744
1745           // ignore special chars immediately following this one
1746           // (should this be an error?)
1747           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1748             pszMask++;
1749
1750           // if there is nothing more, match
1751           if ( *pszMask == wxT('\0') )
1752             return TRUE;
1753
1754           // are there any other metacharacters in the mask?
1755           size_t uiLenMask;
1756           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1757
1758           if ( pEndMask != NULL ) {
1759             // we have to match the string between two metachars
1760             uiLenMask = pEndMask - pszMask;
1761           }
1762           else {
1763             // we have to match the remainder of the string
1764             uiLenMask = wxStrlen(pszMask);
1765           }
1766
1767           wxString strToMatch(pszMask, uiLenMask);
1768           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1769           if ( pMatch == NULL )
1770             return FALSE;
1771
1772           // -1 to compensate "++" in the loop
1773           pszTxt = pMatch + uiLenMask - 1;
1774           pszMask += uiLenMask - 1;
1775         }
1776         break;
1777
1778       default:
1779         if ( *pszMask != *pszTxt )
1780           return FALSE;
1781         break;
1782     }
1783   }
1784
1785   // match only if nothing left
1786   if ( *pszTxt == wxT('\0') )
1787     return TRUE;
1788
1789   // if we failed to match, backtrack if we can
1790   if ( pszLastStarInText ) {
1791     pszTxt = pszLastStarInText + 1;
1792     pszMask = pszLastStarInMask;
1793
1794     pszLastStarInText = NULL;
1795
1796     // don't bother resetting pszLastStarInMask, it's unnecessary
1797
1798     goto match;
1799   }
1800
1801   return FALSE;
1802 #endif // wxUSE_REGEX/!wxUSE_REGEX
1803 }
1804
1805 // Count the number of chars
1806 int wxString::Freq(wxChar ch) const
1807 {
1808     int count = 0;
1809     int len = Len();
1810     for (int i = 0; i < len; i++)
1811     {
1812         if (GetChar(i) == ch)
1813             count ++;
1814     }
1815     return count;
1816 }
1817
1818 // convert to upper case, return the copy of the string
1819 wxString wxString::Upper() const
1820 { wxString s(*this); return s.MakeUpper(); }
1821
1822 // convert to lower case, return the copy of the string
1823 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1824
1825 int wxString::sprintf(const wxChar *pszFormat, ...)
1826   {
1827     va_list argptr;
1828     va_start(argptr, pszFormat);
1829     int iLen = PrintfV(pszFormat, argptr);
1830     va_end(argptr);
1831     return iLen;
1832   }
1833
1834 // ============================================================================
1835 // ArrayString
1836 // ============================================================================
1837
1838 #include "wx/arrstr.h"
1839
1840 #if !wxUSE_STL
1841
1842 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1843 #define   ARRAY_MAXSIZE_INCREMENT       4096
1844
1845 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1846 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1847 #endif
1848
1849 #define   STRING(p)   ((wxString *)(&(p)))
1850
1851 // ctor
1852 void wxArrayString::Init(bool autoSort)
1853 {
1854   m_nSize  =
1855   m_nCount = 0;
1856   m_pItems = (wxChar **) NULL;
1857   m_autoSort = autoSort;
1858 }
1859
1860 // copy ctor
1861 wxArrayString::wxArrayString(const wxArrayString& src)
1862 {
1863   Init(src.m_autoSort);
1864
1865   *this = src;
1866 }
1867
1868 // assignment operator
1869 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1870 {
1871   if ( m_nSize > 0 )
1872     Clear();
1873
1874   Copy(src);
1875
1876   m_autoSort = src.m_autoSort;
1877
1878   return *this;
1879 }
1880
1881 void wxArrayString::Copy(const wxArrayString& src)
1882 {
1883   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1884     Alloc(src.m_nCount);
1885
1886   for ( size_t n = 0; n < src.m_nCount; n++ )
1887     Add(src[n]);
1888 }
1889
1890 // grow the array
1891 void wxArrayString::Grow(size_t nIncrement)
1892 {
1893   // only do it if no more place
1894   if ( (m_nSize - m_nCount) < nIncrement ) {
1895     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1896     // be never resized!
1897     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1898       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1899     #endif
1900
1901     if ( m_nSize == 0 ) {
1902       // was empty, alloc some memory
1903       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1904       if (m_nSize < nIncrement)
1905           m_nSize = nIncrement;
1906       m_pItems = new wxChar *[m_nSize];
1907     }
1908     else {
1909       // otherwise when it's called for the first time, nIncrement would be 0
1910       // and the array would never be expanded
1911       // add 50% but not too much
1912       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1913                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1914       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1915         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1916       if ( nIncrement < ndefIncrement )
1917         nIncrement = ndefIncrement;
1918       m_nSize += nIncrement;
1919       wxChar **pNew = new wxChar *[m_nSize];
1920
1921       // copy data to new location
1922       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1923
1924       // delete old memory (but do not release the strings!)
1925       wxDELETEA(m_pItems);
1926
1927       m_pItems = pNew;
1928     }
1929   }
1930 }
1931
1932 void wxArrayString::Free()
1933 {
1934   for ( size_t n = 0; n < m_nCount; n++ ) {
1935     STRING(m_pItems[n])->GetStringData()->Unlock();
1936   }
1937 }
1938
1939 // deletes all the strings from the list
1940 void wxArrayString::Empty()
1941 {
1942   Free();
1943
1944   m_nCount = 0;
1945 }
1946
1947 // as Empty, but also frees memory
1948 void wxArrayString::Clear()
1949 {
1950   Free();
1951
1952   m_nSize  =
1953   m_nCount = 0;
1954
1955   wxDELETEA(m_pItems);
1956 }
1957
1958 // dtor
1959 wxArrayString::~wxArrayString()
1960 {
1961   Free();
1962
1963   wxDELETEA(m_pItems);
1964 }
1965
1966 // pre-allocates memory (frees the previous data!)
1967 void wxArrayString::Alloc(size_t nSize)
1968 {
1969   // only if old buffer was not big enough
1970   if ( nSize > m_nSize ) {
1971     Free();
1972     wxDELETEA(m_pItems);
1973     m_pItems = new wxChar *[nSize];
1974     m_nSize  = nSize;
1975   }
1976
1977   m_nCount = 0;
1978 }
1979
1980 // minimizes the memory usage by freeing unused memory
1981 void wxArrayString::Shrink()
1982 {
1983   // only do it if we have some memory to free
1984   if( m_nCount < m_nSize ) {
1985     // allocates exactly as much memory as we need
1986     wxChar **pNew = new wxChar *[m_nCount];
1987
1988     // copy data to new location
1989     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1990     delete [] m_pItems;
1991     m_pItems = pNew;
1992   }
1993 }
1994
1995 #if WXWIN_COMPATIBILITY_2_4
1996
1997 // return a wxString[] as required for some control ctors.
1998 wxString* wxArrayString::GetStringArray() const
1999 {
2000     wxString *array = 0;
2001
2002     if( m_nCount > 0 )
2003     {
2004         array = new wxString[m_nCount];
2005         for( size_t i = 0; i < m_nCount; i++ )
2006             array[i] = m_pItems[i];
2007     }
2008
2009     return array;
2010 }
2011
2012 #endif // WXWIN_COMPATIBILITY_2_4
2013
2014 // searches the array for an item (forward or backwards)
2015 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
2016 {
2017   if ( m_autoSort ) {
2018     // use binary search in the sorted array
2019     wxASSERT_MSG( bCase && !bFromEnd,
2020                   wxT("search parameters ignored for auto sorted array") );
2021
2022     size_t i,
2023            lo = 0,
2024            hi = m_nCount;
2025     int res;
2026     while ( lo < hi ) {
2027       i = (lo + hi)/2;
2028
2029       res = wxStrcmp(sz, m_pItems[i]);
2030       if ( res < 0 )
2031         hi = i;
2032       else if ( res > 0 )
2033         lo = i + 1;
2034       else
2035         return i;
2036     }
2037
2038     return wxNOT_FOUND;
2039   }
2040   else {
2041     // use linear search in unsorted array
2042     if ( bFromEnd ) {
2043       if ( m_nCount > 0 ) {
2044         size_t ui = m_nCount;
2045         do {
2046           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
2047             return ui;
2048         }
2049         while ( ui != 0 );
2050       }
2051     }
2052     else {
2053       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2054         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
2055           return ui;
2056       }
2057     }
2058   }
2059
2060   return wxNOT_FOUND;
2061 }
2062
2063 // add item at the end
2064 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
2065 {
2066   if ( m_autoSort ) {
2067     // insert the string at the correct position to keep the array sorted
2068     size_t i,
2069            lo = 0,
2070            hi = m_nCount;
2071     int res;
2072     while ( lo < hi ) {
2073       i = (lo + hi)/2;
2074
2075       res = wxStrcmp(str, m_pItems[i]);
2076       if ( res < 0 )
2077         hi = i;
2078       else if ( res > 0 )
2079         lo = i + 1;
2080       else {
2081         lo = hi = i;
2082         break;
2083       }
2084     }
2085
2086     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2087
2088     Insert(str, lo, nInsert);
2089
2090     return (size_t)lo;
2091   }
2092   else {
2093     wxASSERT( str.GetStringData()->IsValid() );
2094
2095     Grow(nInsert);
2096
2097     for (size_t i = 0; i < nInsert; i++)
2098     {
2099         // the string data must not be deleted!
2100         str.GetStringData()->Lock();
2101
2102         // just append
2103         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2104     }
2105     size_t ret = m_nCount;
2106     m_nCount += nInsert;
2107     return ret;
2108   }
2109 }
2110
2111 // add item at the given position
2112 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2113 {
2114   wxASSERT( str.GetStringData()->IsValid() );
2115
2116   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2117   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2118                wxT("array size overflow in wxArrayString::Insert") );
2119
2120   Grow(nInsert);
2121
2122   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2123           (m_nCount - nIndex)*sizeof(wxChar *));
2124
2125   for (size_t i = 0; i < nInsert; i++)
2126   {
2127       str.GetStringData()->Lock();
2128       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2129   }
2130   m_nCount += nInsert;
2131 }
2132
2133 // expand the array
2134 void wxArrayString::SetCount(size_t count)
2135 {
2136     Alloc(count);
2137
2138     wxString s;
2139     while ( m_nCount < count )
2140         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2141 }
2142
2143 // removes item from array (by index)
2144 void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove)
2145 {
2146   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2147   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2148                wxT("removing too many elements in wxArrayString::Remove") );
2149
2150   // release our lock
2151   for (size_t i = 0; i < nRemove; i++)
2152       Item(nIndex + i).GetStringData()->Unlock();
2153
2154   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2155           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2156   m_nCount -= nRemove;
2157 }
2158
2159 // removes item from array (by value)
2160 void wxArrayString::Remove(const wxChar *sz)
2161 {
2162   int iIndex = Index(sz);
2163
2164   wxCHECK_RET( iIndex != wxNOT_FOUND,
2165                wxT("removing inexistent element in wxArrayString::Remove") );
2166
2167   RemoveAt(iIndex);
2168 }
2169
2170 // ----------------------------------------------------------------------------
2171 // sorting
2172 // ----------------------------------------------------------------------------
2173
2174 // we can only sort one array at a time with the quick-sort based
2175 // implementation
2176 #if wxUSE_THREADS
2177   // need a critical section to protect access to gs_compareFunction and
2178   // gs_sortAscending variables
2179   static wxCriticalSection *gs_critsectStringSort = NULL;
2180
2181   // call this before the value of the global sort vars is changed/after
2182   // you're finished with them
2183   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2184                            gs_critsectStringSort = new wxCriticalSection;     \
2185                            gs_critsectStringSort->Enter()
2186   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2187                            delete gs_critsectStringSort;                      \
2188                            gs_critsectStringSort = NULL
2189 #else // !threads
2190   #define START_SORT()
2191   #define END_SORT()
2192 #endif // wxUSE_THREADS
2193
2194 // function to use for string comparaison
2195 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2196
2197 // if we don't use the compare function, this flag tells us if we sort the
2198 // array in ascending or descending order
2199 static bool gs_sortAscending = TRUE;
2200
2201 // function which is called by quick sort
2202 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2203 wxStringCompareFunction(const void *first, const void *second)
2204 {
2205   wxString *strFirst = (wxString *)first;
2206   wxString *strSecond = (wxString *)second;
2207
2208   if ( gs_compareFunction ) {
2209     return gs_compareFunction(*strFirst, *strSecond);
2210   }
2211   else {
2212     // maybe we should use wxStrcoll
2213     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2214
2215     return gs_sortAscending ? result : -result;
2216   }
2217 }
2218
2219 // sort array elements using passed comparaison function
2220 void wxArrayString::Sort(CompareFunction compareFunction)
2221 {
2222   START_SORT();
2223
2224   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2225   gs_compareFunction = compareFunction;
2226
2227   DoSort();
2228
2229   // reset it to NULL so that Sort(bool) will work the next time
2230   gs_compareFunction = NULL;
2231
2232   END_SORT();
2233 }
2234
2235 typedef  int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first, const void *second);
2236
2237 void wxArrayString::Sort(CompareFunction2 compareFunction)
2238 {
2239   qsort(m_pItems, m_nCount, sizeof(wxChar *), (wxStringCompareFn)compareFunction);
2240 }
2241
2242 void wxArrayString::Sort(bool reverseOrder)
2243 {
2244   Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending);
2245 }
2246
2247 void wxArrayString::DoSort()
2248 {
2249   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2250
2251   // just sort the pointers using qsort() - of course it only works because
2252   // wxString() *is* a pointer to its data
2253   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2254 }
2255
2256 bool wxArrayString::operator==(const wxArrayString& a) const
2257 {
2258     if ( m_nCount != a.m_nCount )
2259         return FALSE;
2260
2261     for ( size_t n = 0; n < m_nCount; n++ )
2262     {
2263         if ( Item(n) != a[n] )
2264             return FALSE;
2265     }
2266
2267     return TRUE;
2268 }
2269
2270 #endif // !wxUSE_STL
2271
2272 int wxStringSortAscending(wxString* s1, wxString* s2)
2273 {
2274     return wxStrcmp(s1->c_str(), s2->c_str());
2275 }
2276
2277 int wxStringSortDescending(wxString* s1, wxString* s2)
2278 {
2279     return -wxStrcmp(s1->c_str(), s2->c_str());
2280 }