src/common/string.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        src/common/string.cpp
   3 // Purpose:     wxString class
   4 // Author:      Vadim Zeitlin, Ryan Norton
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 //              (c) 2004 Ryan Norton <wxprojects@comcast.net>
  10 // Licence:     wxWindows licence
  11 /////////////////////////////////////////////////////////////////////////////
  12
  13 /*
  14  * About ref counting:
  15  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  16  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  17  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  18  */
  19
  20 // ===========================================================================
  21 // headers, declarations, constants
  22 // ===========================================================================
  23
  24 // For compilers that support precompilation, includes "wx.h".
  25 #include "wx/wxprec.h"
  26
  27 #ifdef __BORLANDC__
  28     #pragma hdrstop
  29 #endif
  30
  31 #ifndef WX_PRECOMP
  32     #include "wx/string.h"
  33     #include "wx/intl.h"
  34     #include "wx/thread.h"
  35 #endif
  36
  37 #include <ctype.h>
  38
  39 #ifndef __WXWINCE__
  40     #include <errno.h>
  41 #endif
  42
  43 #include <string.h>
  44 #include <stdlib.h>
  45
  46 #ifdef __SALFORDC__
  47     #include <clib.h>
  48 #endif
  49
  50 // allocating extra space for each string consumes more memory but speeds up
  51 // the concatenation operations (nLen is the current string's length)
  52 // NB: EXTRA_ALLOC must be >= 0!
  53 #define EXTRA_ALLOC       (19 - nLen % 16)
  54
  55
  56 // string handling functions used by wxString:
  57 #if wxUSE_UNICODE_UTF8
  58     #define wxStringMemcpy   memcpy
  59     #define wxStringMemcmp   memcmp
  60     #define wxStringMemchr   memchr
  61     #define wxStringStrlen   strlen
  62 #else
  63     #define wxStringMemcpy   wxTmemcpy
  64     #define wxStringMemcmp   wxTmemcmp
  65     #define wxStringMemchr   wxTmemchr
  66     #define wxStringStrlen   wxStrlen
  67 #endif
  68
  69
  70 // ---------------------------------------------------------------------------
  71 // static class variables definition
  72 // ---------------------------------------------------------------------------
  73
  74 #if !wxUSE_STL_BASED_WXSTRING
  75   //According to STL _must_ be a -1 size_t
  76   const size_t wxStringImpl::npos = (size_t) -1;
  77 #endif
  78 const size_t wxString::npos = (size_t) -1;
  79
  80 // ----------------------------------------------------------------------------
  81 // static data
  82 // ----------------------------------------------------------------------------
  83
  84 #if wxUSE_STL_BASED_WXSTRING
  85
  86 extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = _T("");
  87
  88 #else
  89
  90 // for an empty string, GetStringData() will return this address: this
  91 // structure has the same layout as wxStringData and it's data() method will
  92 // return the empty string (dummy pointer)
  93 static const struct
  94 {
  95   wxStringData data;
  96   wxChar dummy;
  97 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  98
  99 // empty C style string: points to 'string data' byte of g_strEmpty
 100 extern const wxChar WXDLLIMPEXP_BASE *wxEmptyString = &g_strEmpty.dummy;
 101
 102 #endif
 103
 104 // ----------------------------------------------------------------------------
 105 // global functions
 106 // ----------------------------------------------------------------------------
 107
 108 #if wxUSE_STD_IOSTREAM
 109
 110 #include <iostream>
 111
 112 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
 113 {
 114     return os << str.c_str();
 115 }
 116
 117 wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
 118 {
 119 #if wxUSE_UNICODE && !defined(__BORLANDC__)
 120     return os << str.AsWChar();
 121 #else
 122     return os << str.AsChar();
 123 #endif
 124 }
 125
 126 #endif // wxUSE_STD_IOSTREAM
 127
 128 // ----------------------------------------------------------------------------
 129 // private classes
 130 // ----------------------------------------------------------------------------
 131
 132 // this small class is used to gather statistics for performance tuning
 133 //#define WXSTRING_STATISTICS
 134 #ifdef  WXSTRING_STATISTICS
 135   class Averager
 136   {
 137   public:
 138     Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 139    ~Averager()
 140    { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 141
 142     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 143
 144   private:
 145     size_t m_nCount, m_nTotal;
 146     const wxChar *m_sz;
 147   } g_averageLength("allocation size"),
 148     g_averageSummandLength("summand length"),
 149     g_averageConcatHit("hit probability in concat"),
 150     g_averageInitialLength("initial string length");
 151
 152   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 153 #else
 154   #define STATISTICS_ADD(av, val)
 155 #endif // WXSTRING_STATISTICS
 156
 157 #if !wxUSE_STL_BASED_WXSTRING
 158
 159 // ===========================================================================
 160 // wxStringData class deallocation
 161 // ===========================================================================
 162
 163 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 164 #  pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
 165 void wxStringData::Free()
 166 {
 167     free(this);
 168 }
 169 #endif
 170
 171 // ===========================================================================
 172 // wxStringImpl
 173 // ===========================================================================
 174
 175 // takes nLength elements of psz starting at nPos
 176 void wxStringImpl::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 177 {
 178   Init();
 179
 180   // if the length is not given, assume the string to be NUL terminated
 181   if ( nLength == npos ) {
 182     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 183
 184     nLength = wxStrlen(psz + nPos);
 185   }
 186
 187   STATISTICS_ADD(InitialLength, nLength);
 188
 189   if ( nLength > 0 ) {
 190     // trailing '\0' is written in AllocBuffer()
 191     if ( !AllocBuffer(nLength) ) {
 192       wxFAIL_MSG( _T("out of memory in wxStringImpl::InitWith") );
 193       return;
 194     }
 195     wxStringMemcpy(m_pchData, psz + nPos, nLength);
 196   }
 197 }
 198
 199 // poor man's iterators are "void *" pointers
 200 wxStringImpl::wxStringImpl(const void *pStart, const void *pEnd)
 201 {
 202   if ( pEnd >= pStart )
 203   {
 204     InitWith((const wxChar *)pStart, 0,
 205              (const wxChar *)pEnd - (const wxChar *)pStart);
 206   }
 207   else
 208   {
 209     wxFAIL_MSG( _T("pStart is not before pEnd") );
 210     Init();
 211   }
 212 }
 213
 214 wxStringImpl::wxStringImpl(size_type n, wxStringCharType ch)
 215 {
 216   Init();
 217   append(n, ch);
 218 }
 219
 220 // ---------------------------------------------------------------------------
 221 // memory allocation
 222 // ---------------------------------------------------------------------------
 223
 224 // allocates memory needed to store a C string of length nLen
 225 bool wxStringImpl::AllocBuffer(size_t nLen)
 226 {
 227   // allocating 0 sized buffer doesn't make sense, all empty strings should
 228   // reuse g_strEmpty
 229   wxASSERT( nLen >  0 );
 230
 231   // make sure that we don't overflow
 232   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 233                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 234
 235   STATISTICS_ADD(Length, nLen);
 236
 237   // allocate memory:
 238   // 1) one extra character for '\0' termination
 239   // 2) sizeof(wxStringData) for housekeeping info
 240   wxStringData* pData = (wxStringData*)
 241     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 242
 243   if ( pData == NULL ) {
 244     // allocation failures are handled by the caller
 245     return false;
 246   }
 247
 248   pData->nRefs        = 1;
 249   pData->nDataLength  = nLen;
 250   pData->nAllocLength = nLen + EXTRA_ALLOC;
 251   m_pchData           = pData->data();  // data starts after wxStringData
 252   m_pchData[nLen]     = wxT('\0');
 253   return true;
 254 }
 255
 256 // must be called before changing this string
 257 bool wxStringImpl::CopyBeforeWrite()
 258 {
 259   wxStringData* pData = GetStringData();
 260
 261   if ( pData->IsShared() ) {
 262     pData->Unlock();                // memory not freed because shared
 263     size_t nLen = pData->nDataLength;
 264     if ( !AllocBuffer(nLen) ) {
 265       // allocation failures are handled by the caller
 266       return false;
 267     }
 268     wxStringMemcpy(m_pchData, pData->data(), nLen);
 269   }
 270
 271   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 272
 273   return true;
 274 }
 275
 276 // must be called before replacing contents of this string
 277 bool wxStringImpl::AllocBeforeWrite(size_t nLen)
 278 {
 279   wxASSERT( nLen != 0 );  // doesn't make any sense
 280
 281   // must not share string and must have enough space
 282   wxStringData* pData = GetStringData();
 283   if ( pData->IsShared() || pData->IsEmpty() ) {
 284     // can't work with old buffer, get new one
 285     pData->Unlock();
 286     if ( !AllocBuffer(nLen) ) {
 287       // allocation failures are handled by the caller
 288       return false;
 289     }
 290   }
 291   else {
 292     if ( nLen > pData->nAllocLength ) {
 293       // realloc the buffer instead of calling malloc() again, this is more
 294       // efficient
 295       STATISTICS_ADD(Length, nLen);
 296
 297       nLen += EXTRA_ALLOC;
 298
 299       pData = (wxStringData*)
 300           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 301
 302       if ( pData == NULL ) {
 303         // allocation failures are handled by the caller
 304         // keep previous data since reallocation failed
 305         return false;
 306       }
 307
 308       pData->nAllocLength = nLen;
 309       m_pchData = pData->data();
 310     }
 311   }
 312
 313   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 314
 315   // it doesn't really matter what the string length is as it's going to be
 316   // overwritten later but, for extra safety, set it to 0 for now as we may
 317   // have some junk in m_pchData
 318   GetStringData()->nDataLength = 0;
 319
 320   return true;
 321 }
 322
 323 wxStringImpl& wxStringImpl::append(size_t n, wxStringCharType ch)
 324 {
 325     size_type len = length();
 326
 327     if ( !Alloc(len + n) || !CopyBeforeWrite() ) {
 328       wxFAIL_MSG( _T("out of memory in wxStringImpl::append") );
 329     }
 330     GetStringData()->nDataLength = len + n;
 331     m_pchData[len + n] = '\0';
 332     for ( size_t i = 0; i < n; ++i )
 333         m_pchData[len + i] = ch;
 334     return *this;
 335 }
 336
 337 void wxStringImpl::resize(size_t nSize, wxStringCharType ch)
 338 {
 339     size_t len = length();
 340
 341     if ( nSize < len )
 342     {
 343         erase(begin() + nSize, end());
 344     }
 345     else if ( nSize > len )
 346     {
 347         append(nSize - len, ch);
 348     }
 349     //else: we have exactly the specified length, nothing to do
 350 }
 351
 352 // allocate enough memory for nLen characters
 353 bool wxStringImpl::Alloc(size_t nLen)
 354 {
 355   wxStringData *pData = GetStringData();
 356   if ( pData->nAllocLength <= nLen ) {
 357     if ( pData->IsEmpty() ) {
 358       nLen += EXTRA_ALLOC;
 359
 360       pData = (wxStringData *)
 361                 malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 362
 363       if ( pData == NULL ) {
 364         // allocation failure handled by caller
 365         return false;
 366       }
 367
 368       pData->nRefs = 1;
 369       pData->nDataLength = 0;
 370       pData->nAllocLength = nLen;
 371       m_pchData = pData->data();  // data starts after wxStringData
 372       m_pchData[0u] = wxT('\0');
 373     }
 374     else if ( pData->IsShared() ) {
 375       pData->Unlock();                // memory not freed because shared
 376       size_t nOldLen = pData->nDataLength;
 377       if ( !AllocBuffer(nLen) ) {
 378         // allocation failure handled by caller
 379         return false;
 380       }
 381       // +1 to copy the terminator, too
 382       memcpy(m_pchData, pData->data(), (nOldLen+1)*sizeof(wxChar));
 383       GetStringData()->nDataLength = nOldLen;
 384     }
 385     else {
 386       nLen += EXTRA_ALLOC;
 387
 388       pData = (wxStringData *)
 389         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 390
 391       if ( pData == NULL ) {
 392         // allocation failure handled by caller
 393         // keep previous data since reallocation failed
 394         return false;
 395       }
 396
 397       // it's not important if the pointer changed or not (the check for this
 398       // is not faster than assigning to m_pchData in all cases)
 399       pData->nAllocLength = nLen;
 400       m_pchData = pData->data();
 401     }
 402   }
 403   //else: we've already got enough
 404   return true;
 405 }
 406
 407 wxStringImpl::iterator wxStringImpl::begin()
 408 {
 409     if (length() > 0)
 410         CopyBeforeWrite();
 411     return m_pchData;
 412 }
 413
 414 wxStringImpl::iterator wxStringImpl::end()
 415 {
 416     if (length() > 0)
 417         CopyBeforeWrite();
 418     return m_pchData + length();
 419 }
 420
 421 wxStringImpl::iterator wxStringImpl::erase(iterator it)
 422 {
 423     size_type idx = it - begin();
 424     erase(idx, 1);
 425     return begin() + idx;
 426 }
 427
 428 wxStringImpl& wxStringImpl::erase(size_t nStart, size_t nLen)
 429 {
 430     wxASSERT(nStart <= length());
 431     size_t strLen = length() - nStart;
 432     // delete nLen or up to the end of the string characters
 433     nLen = strLen < nLen ? strLen : nLen;
 434     wxStringImpl strTmp(c_str(), nStart);
 435     strTmp.append(c_str() + nStart + nLen, length() - nStart - nLen);
 436
 437     swap(strTmp);
 438     return *this;
 439 }
 440
 441 wxStringImpl& wxStringImpl::insert(size_t nPos, const wxChar *sz, size_t n)
 442 {
 443     wxASSERT( nPos <= length() );
 444
 445     if ( n == npos ) n = wxStrlen(sz);
 446     if ( n == 0 ) return *this;
 447
 448     if ( !Alloc(length() + n) || !CopyBeforeWrite() ) {
 449         wxFAIL_MSG( _T("out of memory in wxStringImpl::insert") );
 450     }
 451
 452     memmove(m_pchData + nPos + n, m_pchData + nPos,
 453             (length() - nPos) * sizeof(wxChar));
 454     memcpy(m_pchData + nPos, sz, n * sizeof(wxChar));
 455     GetStringData()->nDataLength = length() + n;
 456     m_pchData[length()] = '\0';
 457
 458     return *this;
 459 }
 460
 461 void wxStringImpl::swap(wxStringImpl& str)
 462 {
 463     wxStringCharType* tmp = str.m_pchData;
 464     str.m_pchData = m_pchData;
 465     m_pchData = tmp;
 466 }
 467
 468 size_t wxStringImpl::find(const wxStringImpl& str, size_t nStart) const
 469 {
 470     // deal with the special case of empty string first
 471     const size_t nLen = length();
 472     const size_t nLenOther = str.length();
 473
 474     if ( !nLenOther )
 475     {
 476         // empty string is a substring of anything
 477         return 0;
 478     }
 479
 480     if ( !nLen )
 481     {
 482         // the other string is non empty so can't be our substring
 483         return npos;
 484     }
 485
 486     wxASSERT( str.GetStringData()->IsValid() );
 487     wxASSERT( nStart <= nLen );
 488
 489     const wxStringCharType * const other = str.c_str();
 490
 491     // anchor
 492     const wxStringCharType* p =
 493         (const wxStringCharType*)wxStringMemchr(c_str() + nStart,
 494                                                 *other,
 495                                                 nLen - nStart);
 496
 497     if ( !p )
 498         return npos;
 499
 500     while ( p - c_str() + nLenOther <= nLen &&
 501             wxStringMemcmp(p, other, nLenOther) )
 502     {
 503         p++;
 504
 505         // anchor again
 506         p = (const wxStringCharType*)
 507                 wxStringMemchr(p, *other, nLen - (p - c_str()));
 508
 509         if ( !p )
 510             return npos;
 511     }
 512
 513     return p - c_str() + nLenOther <= nLen ? p - c_str() : npos;
 514 }
 515
 516 size_t wxStringImpl::find(const wxChar* sz, size_t nStart, size_t n) const
 517 {
 518     return find(wxStringImpl(sz, n), nStart);
 519 }
 520
 521 size_t wxStringImpl::find(wxStringCharType ch, size_t nStart) const
 522 {
 523     wxASSERT( nStart <= length() );
 524
 525     const wxStringCharType *p = (const wxStringCharType*)
 526         wxStringMemchr(c_str() + nStart, ch, length() - nStart);
 527
 528     return p == NULL ? npos : p - c_str();
 529 }
 530
 531 size_t wxStringImpl::rfind(const wxStringImpl& str, size_t nStart) const
 532 {
 533     wxASSERT( str.GetStringData()->IsValid() );
 534     wxASSERT( nStart == npos || nStart <= length() );
 535
 536     if ( length() >= str.length() )
 537     {
 538         // avoids a corner case later
 539         if ( length() == 0 && str.length() == 0 )
 540             return 0;
 541
 542         // "top" is the point where search starts from
 543         size_t top = length() - str.length();
 544
 545         if ( nStart == npos )
 546             nStart = length() - 1;
 547         if ( nStart < top )
 548             top = nStart;
 549
 550         const wxStringCharType *cursor = c_str() + top;
 551         do
 552         {
 553             if ( wxStringMemcmp(cursor, str.c_str(), str.length()) == 0 )
 554             {
 555                 return cursor - c_str();
 556             }
 557         } while ( cursor-- > c_str() );
 558     }
 559
 560     return npos;
 561 }
 562
 563 size_t wxStringImpl::rfind(const wxChar* sz, size_t nStart, size_t n) const
 564 {
 565     return rfind(wxStringImpl(sz, n), nStart);
 566 }
 567
 568 size_t wxStringImpl::rfind(wxStringCharType ch, size_t nStart) const
 569 {
 570     if ( nStart == npos )
 571     {
 572         nStart = length();
 573     }
 574     else
 575     {
 576         wxASSERT( nStart <= length() );
 577     }
 578
 579     const wxStringCharType *actual;
 580     for ( actual = c_str() + ( nStart == npos ? length() : nStart + 1 );
 581           actual > c_str(); --actual )
 582     {
 583         if ( *(actual - 1) == ch )
 584             return (actual - 1) - c_str();
 585     }
 586
 587     return npos;
 588 }
 589
 590 wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen,
 591                                     const wxChar *sz)
 592 {
 593   wxASSERT_MSG( nStart <= length(),
 594                 _T("index out of bounds in wxStringImpl::replace") );
 595   size_t strLen = length() - nStart;
 596   nLen = strLen < nLen ? strLen : nLen;
 597
 598   wxStringImpl strTmp;
 599   strTmp.reserve(length()); // micro optimisation to avoid multiple mem allocs
 600
 601   //This is kind of inefficient, but its pretty good considering...
 602   //we don't want to use character access operators here because on STL
 603   //it will freeze the reference count of strTmp, which means a deep copy
 604   //at the end when swap is called
 605   //
 606   //Also, we can't use append with the full character pointer and must
 607   //do it manually because this string can contain null characters
 608   for(size_t i1 = 0; i1 < nStart; ++i1)
 609       strTmp.append(1, this->c_str()[i1]);
 610
 611   //its safe to do the full version here because
 612   //sz must be a normal c string
 613   strTmp.append(sz);
 614
 615   for(size_t i2 = nStart + nLen; i2 < length(); ++i2)
 616       strTmp.append(1, this->c_str()[i2]);
 617
 618   swap(strTmp);
 619   return *this;
 620 }
 621
 622 wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen,
 623                                     size_t nCount, wxStringCharType ch)
 624 {
 625   return replace(nStart, nLen, wxStringImpl(nCount, ch).c_str());
 626 }
 627
 628 wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen,
 629                                     const wxStringImpl& str,
 630                                     size_t nStart2, size_t nLen2)
 631 {
 632   return replace(nStart, nLen, str.substr(nStart2, nLen2));
 633 }
 634
 635 wxStringImpl& wxStringImpl::replace(size_t nStart, size_t nLen,
 636                                     const wxChar* sz, size_t nCount)
 637 {
 638   return replace(nStart, nLen, wxStringImpl(sz, nCount).c_str());
 639 }
 640
 641 wxStringImpl wxStringImpl::substr(size_t nStart, size_t nLen) const
 642 {
 643   if ( nLen == npos )
 644     nLen = length() - nStart;
 645   return wxStringImpl(*this, nStart, nLen);
 646 }
 647
 648 // assigns one string to another
 649 wxStringImpl& wxStringImpl::operator=(const wxStringImpl& stringSrc)
 650 {
 651   wxASSERT( stringSrc.GetStringData()->IsValid() );
 652
 653   // don't copy string over itself
 654   if ( m_pchData != stringSrc.m_pchData ) {
 655     if ( stringSrc.GetStringData()->IsEmpty() ) {
 656       Reinit();
 657     }
 658     else {
 659       // adjust references
 660       GetStringData()->Unlock();
 661       m_pchData = stringSrc.m_pchData;
 662       GetStringData()->Lock();
 663     }
 664   }
 665
 666   return *this;
 667 }
 668
 669 // assigns a single character
 670 wxStringImpl& wxStringImpl::operator=(wxStringCharType ch)
 671 {
 672   wxChar c(ch);
 673   if ( !AssignCopy(1, &c) ) {
 674     wxFAIL_MSG( _T("out of memory in wxStringImpl::operator=(wxChar)") );
 675   }
 676   return *this;
 677 }
 678
 679 // assigns C string
 680 wxStringImpl& wxStringImpl::operator=(const wxChar *psz)
 681 {
 682   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 683     wxFAIL_MSG( _T("out of memory in wxStringImpl::operator=(const wxChar *)") );
 684   }
 685   return *this;
 686 }
 687
 688 // helper function: does real copy
 689 bool wxStringImpl::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 690 {
 691   if ( nSrcLen == 0 ) {
 692     Reinit();
 693   }
 694   else {
 695     if ( !AllocBeforeWrite(nSrcLen) ) {
 696       // allocation failure handled by caller
 697       return false;
 698     }
 699     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 700     GetStringData()->nDataLength = nSrcLen;
 701     m_pchData[nSrcLen] = wxT('\0');
 702   }
 703   return true;
 704 }
 705
 706 // ---------------------------------------------------------------------------
 707 // string concatenation
 708 // ---------------------------------------------------------------------------
 709
 710 // add something to this string
 711 bool wxStringImpl::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData,
 712                               size_t nMaxLen)
 713 {
 714   STATISTICS_ADD(SummandLength, nSrcLen);
 715
 716   nSrcLen = nSrcLen < nMaxLen ? nSrcLen : nMaxLen;
 717
 718   // concatenating an empty string is a NOP
 719   if ( nSrcLen > 0 ) {
 720     wxStringData *pData = GetStringData();
 721     size_t nLen = pData->nDataLength;
 722     size_t nNewLen = nLen + nSrcLen;
 723
 724     // alloc new buffer if current is too small
 725     if ( pData->IsShared() ) {
 726       STATISTICS_ADD(ConcatHit, 0);
 727
 728       // we have to allocate another buffer
 729       wxStringData* pOldData = GetStringData();
 730       if ( !AllocBuffer(nNewLen) ) {
 731           // allocation failure handled by caller
 732           return false;
 733       }
 734       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 735       pOldData->Unlock();
 736     }
 737     else if ( nNewLen > pData->nAllocLength ) {
 738       STATISTICS_ADD(ConcatHit, 0);
 739
 740       reserve(nNewLen);
 741       // we have to grow the buffer
 742       if ( capacity() < nNewLen ) {
 743           // allocation failure handled by caller
 744           return false;
 745       }
 746     }
 747     else {
 748       STATISTICS_ADD(ConcatHit, 1);
 749
 750       // the buffer is already big enough
 751     }
 752
 753     // should be enough space
 754     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 755
 756     // fast concatenation - all is done in our buffer
 757     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 758
 759     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 760     GetStringData()->nDataLength = nNewLen; // and fix the length
 761   }
 762   //else: the string to append was empty
 763   return true;
 764 }
 765
 766 #endif // !wxUSE_STL_BASED_WXSTRING
 767
 768 #ifdef HAVE_STD_STRING_COMPARE
 769
 770 // NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with
 771 //     UTF-8 encoded strings too, thanks to UTF-8's design which allows us to
 772 //     sort strings in characters code point order by sorting the byte sequence
 773 //     in byte values order (i.e. what strcmp() and memcmp() do).
 774
 775 int wxString::compare(const wxString& str) const
 776 {
 777     return m_impl.compare(str.m_impl);
 778 }
 779
 780 int wxString::compare(size_t nStart, size_t nLen,
 781                       const wxString& str) const
 782 {
 783     size_t pos, len;
 784     PosLenToImpl(nStart, nLen, &pos, &len);
 785     return m_impl.compare(pos, len, str.m_impl);
 786 }
 787
 788 int wxString::compare(size_t nStart, size_t nLen,
 789                       const wxString& str,
 790                       size_t nStart2, size_t nLen2) const
 791 {
 792     size_t pos, len;
 793     PosLenToImpl(nStart, nLen, &pos, &len);
 794
 795     size_t pos2, len2;
 796     str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
 797
 798     return m_impl.compare(pos, len, str.m_impl, pos2, len2);
 799 }
 800
 801 int wxString::compare(const char* sz) const
 802 {
 803     return m_impl.compare(ImplStr(sz));
 804 }
 805
 806 int wxString::compare(const wchar_t* sz) const
 807 {
 808     return m_impl.compare(ImplStr(sz));
 809 }
 810
 811 int wxString::compare(size_t nStart, size_t nLen,
 812                       const char* sz, size_t nCount) const
 813 {
 814     size_t pos, len;
 815     PosLenToImpl(nStart, nLen, &pos, &len);
 816
 817     SubstrBufFromMB str(sz, nCount);
 818
 819     return m_impl.compare(pos, len, str.data, str.len);
 820 }
 821
 822 int wxString::compare(size_t nStart, size_t nLen,
 823                       const wchar_t* sz, size_t nCount) const
 824 {
 825     size_t pos, len;
 826     PosLenToImpl(nStart, nLen, &pos, &len);
 827
 828     SubstrBufFromWC str(sz, nCount);
 829
 830     return m_impl.compare(pos, len, str.data, str.len);
 831 }
 832
 833 #else // !HAVE_STD_STRING_COMPARE
 834
 835 static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
 836                           const wxStringCharType* s2, size_t l2)
 837 {
 838     if( l1 == l2 )
 839         return wxStringMemcmp(s1, s2, l1);
 840     else if( l1 < l2 )
 841     {
 842         int ret = wxStringMemcmp(s1, s2, l1);
 843         return ret == 0 ? -1 : ret;
 844     }
 845     else
 846     {
 847         int ret = wxStringMemcmp(s1, s2, l2);
 848         return ret == 0 ? +1 : ret;
 849     }
 850 }
 851
 852 int wxString::compare(const wxString& str) const
 853 {
 854     return ::wxDoCmp(m_impl.data(), m_impl.length(),
 855                      str.m_impl.data(), str.m_impl.length());
 856 }
 857
 858 int wxString::compare(size_t nStart, size_t nLen,
 859                       const wxString& str) const
 860 {
 861     wxASSERT(nStart <= length());
 862     size_type strLen = length() - nStart;
 863     nLen = strLen < nLen ? strLen : nLen;
 864
 865     size_t pos, len;
 866     PosLenToImpl(nStart, nLen, &pos, &len);
 867
 868     return ::wxDoCmp(m_impl.data() + pos,  len,
 869                      str.m_impl.data(), str.m_impl.length());
 870 }
 871
 872 int wxString::compare(size_t nStart, size_t nLen,
 873                       const wxString& str,
 874                       size_t nStart2, size_t nLen2) const
 875 {
 876     wxASSERT(nStart <= length());
 877     wxASSERT(nStart2 <= str.length());
 878     size_type strLen  =     length() - nStart,
 879               strLen2 = str.length() - nStart2;
 880     nLen  = strLen  < nLen  ? strLen  : nLen;
 881     nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
 882
 883     size_t pos, len;
 884     PosLenToImpl(nStart, nLen, &pos, &len);
 885     size_t pos2, len2;
 886     str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
 887
 888     return ::wxDoCmp(m_impl.data() + pos, len,
 889                      str.m_impl.data() + pos2, len2);
 890 }
 891
 892 int wxString::compare(const char* sz) const
 893 {
 894     SubstrBufFromMB str(ImplStr(sz, npos));
 895     if ( str.len == npos )
 896         str.len = wxStringStrlen(str.data);
 897     return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
 898 }
 899
 900 int wxString::compare(const wchar_t* sz) const
 901 {
 902     SubstrBufFromWC str(ImplStr(sz, npos));
 903     if ( str.len == npos )
 904         str.len = wxStringStrlen(str.data);
 905     return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
 906 }
 907
 908 int wxString::compare(size_t nStart, size_t nLen,
 909                       const char* sz, size_t nCount) const
 910 {
 911     wxASSERT(nStart <= length());
 912     size_type strLen = length() - nStart;
 913     nLen = strLen < nLen ? strLen : nLen;
 914
 915     size_t pos, len;
 916     PosLenToImpl(nStart, nLen, &pos, &len);
 917
 918     SubstrBufFromMB str(ImplStr(sz, nCount));
 919     if ( str.len == npos )
 920         str.len = wxStringStrlen(str.data);
 921
 922     return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
 923 }
 924
 925 int wxString::compare(size_t nStart, size_t nLen,
 926                       const wchar_t* sz, size_t nCount) const
 927 {
 928     wxASSERT(nStart <= length());
 929     size_type strLen = length() - nStart;
 930     nLen = strLen < nLen ? strLen : nLen;
 931
 932     size_t pos, len;
 933     PosLenToImpl(nStart, nLen, &pos, &len);
 934
 935     SubstrBufFromWC str(ImplStr(sz, nCount));
 936     if ( str.len == npos )
 937         str.len = wxStringStrlen(str.data);
 938
 939     return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
 940 }
 941
 942 #endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
 943
 944 // ===========================================================================
 945 // wxString class core
 946 // ===========================================================================
 947
 948 // ---------------------------------------------------------------------------
 949 // construction and conversion
 950 // ---------------------------------------------------------------------------
 951
 952 #if wxUSE_UNICODE
 953 /* static */
 954 wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
 955                                            const wxMBConv& conv)
 956 {
 957     // anything to do?
 958     if ( !psz || nLength == 0 )
 959         return SubstrBufFromMB();
 960
 961     if ( nLength == npos )
 962         nLength = wxNO_LEN;
 963
 964     size_t wcLen;
 965     wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
 966     if ( !wcLen )
 967         return SubstrBufFromMB();
 968     else
 969         return SubstrBufFromMB(wcBuf, wcLen);
 970 }
 971 #else
 972 /* static */
 973 wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
 974                                            const wxMBConv& conv)
 975 {
 976     // anything to do?
 977     if ( !pwz || nLength == 0 )
 978         return SubstrBufFromWC();
 979
 980     if ( nLength == npos )
 981         nLength = wxNO_LEN;
 982
 983     size_t mbLen;
 984     wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
 985     if ( !mbLen )
 986         return SubstrBufFromWC();
 987     else
 988         return SubstrBufFromWC(mbBuf, mbLen);
 989 }
 990 #endif
 991
 992
 993 #if wxUSE_UNICODE
 994
 995 // from multibyte string
 996 wxString::wxString(const char *psz, const wxMBConv& conv, size_t nLength)
 997 {
 998     // FIXME-UTF8: this will need changes
 999
1000     // anything to do?
1001     if ( psz && nLength != 0 )
1002     {
1003         if ( nLength == npos )
1004         {
1005             nLength = wxNO_LEN;
1006         }
1007
1008         size_t nLenWide;
1009         wxWCharBuffer wbuf = conv.cMB2WC(psz, nLength, &nLenWide);
1010
1011         if ( nLenWide )
1012             assign(wbuf, nLenWide);
1013     }
1014 }
1015
1016 wxString::wxString(const char *psz, size_t nLength)
1017 {
1018     assign(psz, nLength);
1019 }
1020
1021 //Convert wxString in Unicode mode to a multi-byte string
1022 const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
1023 {
1024     return conv.cWC2MB(c_str(), length() + 1 /* size, not length */, NULL);
1025 }
1026
1027 #else // ANSI
1028
1029 #if wxUSE_WCHAR_T
1030
1031 // from wide string
1032 wxString::wxString(const wchar_t *pwz, const wxMBConv& conv, size_t nLength)
1033 {
1034     // FIXME-UTF8: this will need changes
1035
1036     // anything to do?
1037     if ( pwz && nLength != 0 )
1038     {
1039         if ( nLength == npos )
1040         {
1041             nLength = wxNO_LEN;
1042         }
1043
1044         size_t nLenMB;
1045         wxCharBuffer buf = conv.cWC2MB(pwz, nLength, &nLenMB);
1046
1047         if ( nLenMB )
1048             assign(buf, nLenMB);
1049     }
1050
1051 }
1052
1053 wxString::wxString(const wchar_t *pwz, size_t nLength)
1054 {
1055     assign(pwz, nLength);
1056 }
1057
1058 //Converts this string to a wide character string if unicode
1059 //mode is not enabled and wxUSE_WCHAR_T is enabled
1060 const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const
1061 {
1062     return conv.cMB2WC(c_str(), length() + 1 /* size, not length */, NULL);
1063 }
1064
1065 #endif // wxUSE_WCHAR_T
1066
1067 #endif // Unicode/ANSI
1068
1069 // shrink to minimal size (releasing extra memory)
1070 bool wxString::Shrink()
1071 {
1072   wxString tmp(begin(), end());
1073   swap(tmp);
1074   return tmp.length() == length();
1075 }
1076
1077 #if !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
1078 // get the pointer to writable buffer of (at least) nLen bytes
1079 wxChar *wxStringImpl::DoGetWriteBuf(size_t nLen)
1080 {
1081   if ( !AllocBeforeWrite(nLen) ) {
1082     // allocation failure handled by caller
1083     return NULL;
1084   }
1085
1086   wxASSERT( GetStringData()->nRefs == 1 );
1087   GetStringData()->Validate(false);
1088
1089   return m_pchData;
1090 }
1091
1092 // put string back in a reasonable state after GetWriteBuf
1093 void wxStringImpl::DoUngetWriteBuf()
1094 {
1095   DoUngetWriteBuf(wxStrlen(m_pchData));
1096 }
1097
1098 void wxStringImpl::DoUngetWriteBuf(size_t nLen)
1099 {
1100   wxStringData * const pData = GetStringData();
1101
1102   wxASSERT_MSG( nLen < pData->nAllocLength, _T("buffer overrun") );
1103
1104   // the strings we store are always NUL-terminated
1105   pData->data()[nLen] = _T('\0');
1106   pData->nDataLength = nLen;
1107   pData->Validate(true);
1108 }
1109
1110 // deprecated compatibility code:
1111 #if WXWIN_COMPATIBILITY_2_8
1112 wxChar *wxString::GetWriteBuf(size_t nLen)
1113 {
1114     return DoGetWriteBuf(nLen);
1115 }
1116
1117 void wxString::UngetWriteBuf()
1118 {
1119     DoUngetWriteBuf();
1120 }
1121
1122 void wxString::UngetWriteBuf(size_t nLen)
1123 {
1124     DoUngetWriteBuf(nLen);
1125 }
1126 #endif // WXWIN_COMPATIBILITY_2_8
1127
1128 #endif // !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
1129
1130
1131 // ---------------------------------------------------------------------------
1132 // data access
1133 // ---------------------------------------------------------------------------
1134
1135 // all functions are inline in string.h
1136
1137 // ---------------------------------------------------------------------------
1138 // assignment operators
1139 // ---------------------------------------------------------------------------
1140
1141 #if !wxUSE_UNICODE
1142
1143 // same as 'signed char' variant
1144 wxString& wxString::operator=(const unsigned char* psz)
1145 {
1146   *this = (const char *)psz;
1147   return *this;
1148 }
1149
1150 #if wxUSE_WCHAR_T
1151 wxString& wxString::operator=(const wchar_t *pwz)
1152 {
1153   wxString str(pwz);
1154   swap(str);
1155   return *this;
1156 }
1157 #endif
1158
1159 #endif
1160
1161 /*
1162  * concatenation functions come in 5 flavours:
1163  *  string + string
1164  *  char   + string      and      string + char
1165  *  C str  + string      and      string + C str
1166  */
1167
1168 wxString operator+(const wxString& str1, const wxString& str2)
1169 {
1170 #if !wxUSE_STL_BASED_WXSTRING
1171     wxASSERT( str1.IsValid() );
1172     wxASSERT( str2.IsValid() );
1173 #endif
1174
1175     wxString s = str1;
1176     s += str2;
1177
1178     return s;
1179 }
1180
1181 wxString operator+(const wxString& str, wxUniChar ch)
1182 {
1183 #if !wxUSE_STL_BASED_WXSTRING
1184     wxASSERT( str.IsValid() );
1185 #endif
1186
1187     wxString s = str;
1188     s += ch;
1189
1190     return s;
1191 }
1192
1193 wxString operator+(wxUniChar ch, const wxString& str)
1194 {
1195 #if !wxUSE_STL_BASED_WXSTRING
1196     wxASSERT( str.IsValid() );
1197 #endif
1198
1199     wxString s = ch;
1200     s += str;
1201
1202     return s;
1203 }
1204
1205 wxString operator+(const wxString& str, const char *psz)
1206 {
1207 #if !wxUSE_STL_BASED_WXSTRING
1208     wxASSERT( str.IsValid() );
1209 #endif
1210
1211     wxString s;
1212     if ( !s.Alloc(strlen(psz) + str.length()) ) {
1213         wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1214     }
1215     s += str;
1216     s += psz;
1217
1218     return s;
1219 }
1220
1221 wxString operator+(const wxString& str, const wchar_t *pwz)
1222 {
1223 #if !wxUSE_STL_BASED_WXSTRING
1224     wxASSERT( str.IsValid() );
1225 #endif
1226
1227     wxString s;
1228     if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
1229         wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1230     }
1231     s += str;
1232     s += pwz;
1233
1234     return s;
1235 }
1236
1237 wxString operator+(const char *psz, const wxString& str)
1238 {
1239 #if !wxUSE_STL_BASED_WXSTRING
1240     wxASSERT( str.IsValid() );
1241 #endif
1242
1243     wxString s;
1244     if ( !s.Alloc(strlen(psz) + str.length()) ) {
1245         wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1246     }
1247     s = psz;
1248     s += str;
1249
1250     return s;
1251 }
1252
1253 wxString operator+(const wchar_t *pwz, const wxString& str)
1254 {
1255 #if !wxUSE_STL_BASED_WXSTRING
1256     wxASSERT( str.IsValid() );
1257 #endif
1258
1259     wxString s;
1260     if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
1261         wxFAIL_MSG( _T("out of memory in wxString::operator+") );
1262     }
1263     s = pwz;
1264     s += str;
1265
1266     return s;
1267 }
1268
1269 // ---------------------------------------------------------------------------
1270 // find_{first,last}_[not]_of functions
1271 // ---------------------------------------------------------------------------
1272
1273 #if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
1274
1275 // NB: All these functions are implemented  with the argument being wxChar*,
1276 //     i.e. widechar string in any Unicode build, even though native string
1277 //     representation is char* in the UTF-8 build. This is because we couldn't
1278 //     use memchr() to determine if a character is in a set encoded as UTF-8.
1279
1280 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1281 {
1282     return find_first_of(sz, nStart, wxStrlen(sz));
1283 }
1284
1285 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1286 {
1287     return find_first_not_of(sz, nStart, wxStrlen(sz));
1288 }
1289
1290 size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
1291 {
1292     wxASSERT_MSG( nStart <= length(),  _T("invalid index") );
1293
1294     size_t idx = nStart;
1295     for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
1296     {
1297         if ( wxTmemchr(sz, *i, n) )
1298             return idx;
1299     }
1300
1301     return npos;
1302 }
1303
1304 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
1305 {
1306     wxASSERT_MSG( nStart <= length(),  _T("invalid index") );
1307
1308     size_t idx = nStart;
1309     for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
1310     {
1311         if ( !wxTmemchr(sz, *i, n) )
1312             return idx;
1313     }
1314
1315     return npos;
1316 }
1317
1318
1319 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1320 {
1321     return find_last_of(sz, nStart, wxStrlen(sz));
1322 }
1323
1324 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1325 {
1326     return find_last_not_of(sz, nStart, wxStrlen(sz));
1327 }
1328
1329 size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const
1330 {
1331     size_t len = length();
1332
1333     if ( nStart == npos )
1334     {
1335         nStart = len - 1;
1336     }
1337     else
1338     {
1339         wxASSERT_MSG( nStart <= len, _T("invalid index") );
1340     }
1341
1342     size_t idx = nStart;
1343     for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1344           i != rend(); --idx, ++i )
1345     {
1346         if ( wxTmemchr(sz, *i, n) )
1347             return idx;
1348     }
1349
1350     return npos;
1351 }
1352
1353 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const
1354 {
1355     size_t len = length();
1356
1357     if ( nStart == npos )
1358     {
1359         nStart = len - 1;
1360     }
1361     else
1362     {
1363         wxASSERT_MSG( nStart <= len, _T("invalid index") );
1364     }
1365
1366     size_t idx = nStart;
1367     for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1368           i != rend(); --idx, ++i )
1369     {
1370         if ( !wxTmemchr(sz, *i, n) )
1371             return idx;
1372     }
1373
1374     return npos;
1375 }
1376
1377 size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const
1378 {
1379     wxASSERT_MSG( nStart <= length(),  _T("invalid index") );
1380
1381     size_t idx = nStart;
1382     for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
1383     {
1384         if ( *i != ch )
1385             return idx;
1386     }
1387
1388     return npos;
1389 }
1390
1391 size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const
1392 {
1393     size_t len = length();
1394
1395     if ( nStart == npos )
1396     {
1397         nStart = len - 1;
1398     }
1399     else
1400     {
1401         wxASSERT_MSG( nStart <= len, _T("invalid index") );
1402     }
1403
1404     size_t idx = nStart;
1405     for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1406           i != rend(); --idx, ++i )
1407     {
1408         if ( *i != ch )
1409             return idx;
1410     }
1411
1412     return npos;
1413 }
1414
1415 // the functions above were implemented for wchar_t* arguments in Unicode
1416 // build and char* in ANSI build; below are implementations for the other
1417 // version:
1418 #if wxUSE_UNICODE
1419     #define wxOtherCharType char
1420     #define STRCONV         (const wxChar*)wxConvLibc.cMB2WC
1421 #else
1422     #define wxOtherCharType wchar_t
1423     #define STRCONV         (const wxChar*)wxConvLibc.cWC2MB
1424 #endif
1425
1426 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const
1427     { return find_first_of(STRCONV(sz), nStart); }
1428
1429 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart,
1430                                size_t n) const
1431     { return find_first_of(STRCONV(sz, n, NULL), nStart, n); }
1432 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const
1433     { return find_last_of(STRCONV(sz), nStart); }
1434 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart,
1435                               size_t n) const
1436     { return find_last_of(STRCONV(sz, n, NULL), nStart, n); }
1437 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const
1438     { return find_first_not_of(STRCONV(sz), nStart); }
1439 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart,
1440                                    size_t n) const
1441     { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); }
1442 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const
1443     { return find_last_not_of(STRCONV(sz), nStart); }
1444 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart,
1445                                   size_t n) const
1446     { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); }
1447
1448 #undef wxOtherCharType
1449 #undef STRCONV
1450
1451 #endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
1452
1453 // ===========================================================================
1454 // other common string functions
1455 // ===========================================================================
1456
1457 int wxString::CmpNoCase(const wxString& s) const
1458 {
1459     // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
1460
1461     size_t idx = 0;
1462     const_iterator i1 = begin();
1463     const_iterator end1 = end();
1464     const_iterator i2 = s.begin();
1465     const_iterator end2 = s.end();
1466
1467     for ( ; i1 != end1 && i2 != end2; ++idx, ++i1, ++i2 )
1468     {
1469         wxUniChar lower1 = (wxChar)wxTolower(*i1);
1470         wxUniChar lower2 = (wxChar)wxTolower(*i2);
1471         if ( lower1 != lower2 )
1472             return lower1 < lower2 ? -1 : 1;
1473     }
1474
1475     size_t len1 = length();
1476     size_t len2 = s.length();
1477
1478     if ( len1 < len2 )
1479         return -1;
1480     else if ( len1 > len2 )
1481         return 1;
1482     return 0;
1483 }
1484
1485
1486 #if wxUSE_UNICODE
1487
1488 #ifdef __MWERKS__
1489 #ifndef __SCHAR_MAX__
1490 #define __SCHAR_MAX__ 127
1491 #endif
1492 #endif
1493
1494 wxString wxString::FromAscii(const char *ascii)
1495 {
1496     if (!ascii)
1497        return wxEmptyString;
1498
1499     size_t len = strlen( ascii );
1500     wxString res;
1501
1502     if ( len )
1503     {
1504         wxStringBuffer buf(res, len);
1505
1506         wchar_t *dest = buf;
1507
1508         for ( ;; )
1509         {
1510            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
1511                break;
1512         }
1513     }
1514
1515     return res;
1516 }
1517
1518 wxString wxString::FromAscii(const char ascii)
1519 {
1520     // What do we do with '\0' ?
1521
1522     wxString res;
1523     res += (wchar_t)(unsigned char) ascii;
1524
1525     return res;
1526 }
1527
1528 const wxCharBuffer wxString::ToAscii() const
1529 {
1530     // this will allocate enough space for the terminating NUL too
1531     wxCharBuffer buffer(length());
1532
1533
1534     char *dest = buffer.data();
1535
1536     const wchar_t *pwc = c_str();
1537     for ( ;; )
1538     {
1539         *dest++ = (char)(*pwc > SCHAR_MAX ? wxT('_') : *pwc);
1540
1541         // the output string can't have embedded NULs anyhow, so we can safely
1542         // stop at first of them even if we do have any
1543         if ( !*pwc++ )
1544             break;
1545     }
1546
1547     return buffer;
1548 }
1549
1550 #endif // Unicode
1551
1552 // extract string of length nCount starting at nFirst
1553 wxString wxString::Mid(size_t nFirst, size_t nCount) const
1554 {
1555     size_t nLen = length();
1556
1557     // default value of nCount is npos and means "till the end"
1558     if ( nCount == npos )
1559     {
1560         nCount = nLen - nFirst;
1561     }
1562
1563     // out-of-bounds requests return sensible things
1564     if ( nFirst + nCount > nLen )
1565     {
1566         nCount = nLen - nFirst;
1567     }
1568
1569     if ( nFirst > nLen )
1570     {
1571         // AllocCopy() will return empty string
1572         return wxEmptyString;
1573     }
1574
1575     wxString dest(*this, nFirst, nCount);
1576     if ( dest.length() != nCount )
1577     {
1578         wxFAIL_MSG( _T("out of memory in wxString::Mid") );
1579     }
1580
1581     return dest;
1582 }
1583
1584 // check that the string starts with prefix and return the rest of the string
1585 // in the provided pointer if it is not NULL, otherwise return false
1586 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
1587 {
1588     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
1589
1590     // first check if the beginning of the string matches the prefix: note
1591     // that we don't have to check that we don't run out of this string as
1592     // when we reach the terminating NUL, either prefix string ends too (and
1593     // then it's ok) or we break out of the loop because there is no match
1594     const wxChar *p = c_str();
1595     while ( *prefix )
1596     {
1597         if ( *prefix++ != *p++ )
1598         {
1599             // no match
1600             return false;
1601         }
1602     }
1603
1604     if ( rest )
1605     {
1606         // put the rest of the string into provided pointer
1607         *rest = p;
1608     }
1609
1610     return true;
1611 }
1612
1613
1614 // check that the string ends with suffix and return the rest of it in the
1615 // provided pointer if it is not NULL, otherwise return false
1616 bool wxString::EndsWith(const wxChar *suffix, wxString *rest) const
1617 {
1618     wxASSERT_MSG( suffix, _T("invalid parameter in wxString::EndssWith") );
1619
1620     int start = length() - wxStrlen(suffix);
1621     if ( start < 0 || wxStrcmp(wx_str() + start, suffix) != 0 )
1622         return false;
1623
1624     if ( rest )
1625     {
1626         // put the rest of the string into provided pointer
1627         rest->assign(*this, 0, start);
1628     }
1629
1630     return true;
1631 }
1632
1633
1634 // extract nCount last (rightmost) characters
1635 wxString wxString::Right(size_t nCount) const
1636 {
1637   if ( nCount > length() )
1638     nCount = length();
1639
1640   wxString dest(*this, length() - nCount, nCount);
1641   if ( dest.length() != nCount ) {
1642     wxFAIL_MSG( _T("out of memory in wxString::Right") );
1643   }
1644   return dest;
1645 }
1646
1647 // get all characters after the last occurence of ch
1648 // (returns the whole string if ch not found)
1649 wxString wxString::AfterLast(wxUniChar ch) const
1650 {
1651   wxString str;
1652   int iPos = Find(ch, true);
1653   if ( iPos == wxNOT_FOUND )
1654     str = *this;
1655   else
1656     str = wx_str() + iPos + 1;
1657
1658   return str;
1659 }
1660
1661 // extract nCount first (leftmost) characters
1662 wxString wxString::Left(size_t nCount) const
1663 {
1664   if ( nCount > length() )
1665     nCount = length();
1666
1667   wxString dest(*this, 0, nCount);
1668   if ( dest.length() != nCount ) {
1669     wxFAIL_MSG( _T("out of memory in wxString::Left") );
1670   }
1671   return dest;
1672 }
1673
1674 // get all characters before the first occurence of ch
1675 // (returns the whole string if ch not found)
1676 wxString wxString::BeforeFirst(wxUniChar ch) const
1677 {
1678   int iPos = Find(ch);
1679   if ( iPos == wxNOT_FOUND ) iPos = length();
1680   return wxString(*this, 0, iPos);
1681 }
1682
1683 /// get all characters before the last occurence of ch
1684 /// (returns empty string if ch not found)
1685 wxString wxString::BeforeLast(wxUniChar ch) const
1686 {
1687   wxString str;
1688   int iPos = Find(ch, true);
1689   if ( iPos != wxNOT_FOUND && iPos != 0 )
1690     str = wxString(c_str(), iPos);
1691
1692   return str;
1693 }
1694
1695 /// get all characters after the first occurence of ch
1696 /// (returns empty string if ch not found)
1697 wxString wxString::AfterFirst(wxUniChar ch) const
1698 {
1699   wxString str;
1700   int iPos = Find(ch);
1701   if ( iPos != wxNOT_FOUND )
1702     str = wx_str() + iPos + 1;
1703
1704   return str;
1705 }
1706
1707 // replace first (or all) occurences of some substring with another one
1708 size_t wxString::Replace(const wxChar *szOld,
1709                   const wxChar *szNew, bool bReplaceAll)
1710 {
1711     // if we tried to replace an empty string we'd enter an infinite loop below
1712     wxCHECK_MSG( szOld && *szOld && szNew, 0,
1713                  _T("wxString::Replace(): invalid parameter") );
1714
1715     size_t uiCount = 0;   // count of replacements made
1716
1717     size_t uiOldLen = wxStrlen(szOld);
1718     size_t uiNewLen = wxStrlen(szNew);
1719
1720     size_t dwPos = 0;
1721
1722     while ( this->c_str()[dwPos] != wxT('\0') )
1723     {
1724         //DO NOT USE STRSTR HERE
1725         //this string can contain embedded null characters,
1726         //so strstr will function incorrectly
1727         dwPos = find(szOld, dwPos);
1728         if ( dwPos == npos )
1729             break;                  // exit the loop
1730         else
1731         {
1732             //replace this occurance of the old string with the new one
1733             replace(dwPos, uiOldLen, szNew, uiNewLen);
1734
1735             //move up pos past the string that was replaced
1736             dwPos += uiNewLen;
1737
1738             //increase replace count
1739             ++uiCount;
1740
1741             // stop now?
1742             if ( !bReplaceAll )
1743                 break;                  // exit the loop
1744         }
1745     }
1746
1747     return uiCount;
1748 }
1749
1750 bool wxString::IsAscii() const
1751 {
1752   const wxChar *s = (const wxChar*) *this;
1753   while(*s){
1754     if(!isascii(*s)) return(false);
1755     s++;
1756   }
1757   return(true);
1758 }
1759
1760 bool wxString::IsWord() const
1761 {
1762   const wxChar *s = (const wxChar*) *this;
1763   while(*s){
1764     if(!wxIsalpha(*s)) return(false);
1765     s++;
1766   }
1767   return(true);
1768 }
1769
1770 bool wxString::IsNumber() const
1771 {
1772   const wxChar *s = (const wxChar*) *this;
1773   if (wxStrlen(s))
1774      if ((s[0] == wxT('-')) || (s[0] == wxT('+'))) s++;
1775   while(*s){
1776     if(!wxIsdigit(*s)) return(false);
1777     s++;
1778   }
1779   return(true);
1780 }
1781
1782 wxString wxString::Strip(stripType w) const
1783 {
1784     wxString s = *this;
1785     if ( w & leading ) s.Trim(false);
1786     if ( w & trailing ) s.Trim(true);
1787     return s;
1788 }
1789
1790 // ---------------------------------------------------------------------------
1791 // case conversion
1792 // ---------------------------------------------------------------------------
1793
1794 wxString& wxString::MakeUpper()
1795 {
1796   for ( iterator it = begin(), en = end(); it != en; ++it )
1797     *it = (wxChar)wxToupper(*it);
1798
1799   return *this;
1800 }
1801
1802 wxString& wxString::MakeLower()
1803 {
1804   for ( iterator it = begin(), en = end(); it != en; ++it )
1805     *it = (wxChar)wxTolower(*it);
1806
1807   return *this;
1808 }
1809
1810 // ---------------------------------------------------------------------------
1811 // trimming and padding
1812 // ---------------------------------------------------------------------------
1813
1814 // some compilers (VC++ 6.0 not to name them) return true for a call to
1815 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1816 // live with this by checking that the character is a 7 bit one - even if this
1817 // may fail to detect some spaces (I don't know if Unicode doesn't have
1818 // space-like symbols somewhere except in the first 128 chars), it is arguably
1819 // still better than trimming away accented letters
1820 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1821
1822 // trims spaces (in the sense of isspace) from left or right side
1823 wxString& wxString::Trim(bool bFromRight)
1824 {
1825     // first check if we're going to modify the string at all
1826     if ( !empty() &&
1827          (
1828           (bFromRight && wxSafeIsspace(GetChar(length() - 1))) ||
1829           (!bFromRight && wxSafeIsspace(GetChar(0u)))
1830          )
1831        )
1832     {
1833         if ( bFromRight )
1834         {
1835             // find last non-space character
1836             reverse_iterator psz = rbegin();
1837             while ( (psz != rend()) && wxSafeIsspace(*psz) )
1838                 psz++;
1839
1840             // truncate at trailing space start
1841             erase(psz.base(), end());
1842         }
1843         else
1844         {
1845             // find first non-space character
1846             iterator psz = begin();
1847             while ( (psz != end()) && wxSafeIsspace(*psz) )
1848                 psz++;
1849
1850             // fix up data and length
1851             erase(begin(), psz);
1852         }
1853     }
1854
1855     return *this;
1856 }
1857
1858 // adds nCount characters chPad to the string from either side
1859 wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight)
1860 {
1861     wxString s(chPad, nCount);
1862
1863     if ( bFromRight )
1864         *this += s;
1865     else
1866     {
1867         s += *this;
1868         swap(s);
1869     }
1870
1871     return *this;
1872 }
1873
1874 // truncate the string
1875 wxString& wxString::Truncate(size_t uiLen)
1876 {
1877     if ( uiLen < length() )
1878     {
1879         erase(begin() + uiLen, end());
1880     }
1881     //else: nothing to do, string is already short enough
1882
1883     return *this;
1884 }
1885
1886 // ---------------------------------------------------------------------------
1887 // finding (return wxNOT_FOUND if not found and index otherwise)
1888 // ---------------------------------------------------------------------------
1889
1890 // find a character
1891 int wxString::Find(wxUniChar ch, bool bFromEnd) const
1892 {
1893     size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
1894
1895     return (idx == npos) ? wxNOT_FOUND : (int)idx;
1896 }
1897
1898 // find a sub-string (like strstr)
1899 int wxString::Find(const wxChar *pszSub) const
1900 {
1901     size_type idx = find(pszSub);
1902
1903     return (idx == npos) ? wxNOT_FOUND : (int)idx;
1904 }
1905
1906 // ----------------------------------------------------------------------------
1907 // conversion to numbers
1908 // ----------------------------------------------------------------------------
1909
1910 // the implementation of all the functions below is exactly the same so factor
1911 // it out
1912
1913 template <typename T, typename F>
1914 bool wxStringToIntType(const wxChar *start,
1915                        T *val,
1916                        int base,
1917                        F func)
1918 {
1919     wxCHECK_MSG( val, false, _T("NULL output pointer") );
1920     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1921
1922 #ifndef __WXWINCE__
1923     errno = 0;
1924 #endif
1925
1926     wxChar *end;
1927     *val = (*func)(start, &end, base);
1928
1929     // return true only if scan was stopped by the terminating NUL and if the
1930     // string was not empty to start with and no under/overflow occurred
1931     return !*end && (end != start)
1932 #ifndef __WXWINCE__
1933         && (errno != ERANGE)
1934 #endif
1935     ;
1936 }
1937
1938 bool wxString::ToLong(long *val, int base) const
1939 {
1940     return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtol);
1941 }
1942
1943 bool wxString::ToULong(unsigned long *val, int base) const
1944 {
1945     return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoul);
1946 }
1947
1948 bool wxString::ToLongLong(wxLongLong_t *val, int base) const
1949 {
1950 #ifdef wxHAS_STRTOLL
1951     return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoll);
1952 #else
1953     // TODO: implement this ourselves
1954     wxUnusedVar(val);
1955     wxUnusedVar(base);
1956     return false;
1957 #endif // wxHAS_STRTOLL
1958 }
1959
1960 bool wxString::ToULongLong(wxULongLong_t *val, int base) const
1961 {
1962 #ifdef wxHAS_STRTOLL
1963     return wxStringToIntType((const wxChar*)c_str(), val, base, wxStrtoull);
1964 #else
1965     // TODO: implement this ourselves
1966     wxUnusedVar(val);
1967     wxUnusedVar(base);
1968     return false;
1969 #endif
1970 }
1971
1972 bool wxString::ToDouble(double *val) const
1973 {
1974     wxCHECK_MSG( val, false, _T("NULL pointer in wxString::ToDouble") );
1975
1976 #ifndef __WXWINCE__
1977     errno = 0;
1978 #endif
1979
1980     const wxChar *start = c_str();
1981     wxChar *end;
1982     *val = wxStrtod(start, &end);
1983
1984     // return true only if scan was stopped by the terminating NUL and if the
1985     // string was not empty to start with and no under/overflow occurred
1986     return !*end && (end != start)
1987 #ifndef __WXWINCE__
1988         && (errno != ERANGE)
1989 #endif
1990     ;
1991 }
1992
1993 // ---------------------------------------------------------------------------
1994 // formatted output
1995 // ---------------------------------------------------------------------------
1996
1997 /* static */
1998 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1999 wxString wxStringPrintfMixinBase::DoFormat(const wxChar *format, ...)
2000 #else
2001 wxString wxString::DoFormat(const wxChar *format, ...)
2002 #endif
2003 {
2004     va_list argptr;
2005     va_start(argptr, format);
2006
2007     wxString s;
2008     s.PrintfV(format, argptr);
2009
2010     va_end(argptr);
2011
2012     return s;
2013 }
2014
2015 /* static */
2016 wxString wxString::FormatV(const wxString& format, va_list argptr)
2017 {
2018     wxString s;
2019     s.PrintfV(format, argptr);
2020     return s;
2021 }
2022
2023 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
2024 int wxStringPrintfMixinBase::DoPrintf(const wxChar *format, ...)
2025 #else
2026 int wxString::DoPrintf(const wxChar *format, ...)
2027 #endif
2028 {
2029     va_list argptr;
2030     va_start(argptr, format);
2031
2032 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
2033     // get a pointer to the wxString instance; we have to use dynamic_cast<>
2034     // because it's the only cast that works safely for downcasting when
2035     // multiple inheritance is used:
2036     wxString *str = static_cast<wxString*>(this);
2037 #else
2038     wxString *str = this;
2039 #endif
2040
2041     int iLen = str->PrintfV(format, argptr);
2042
2043     va_end(argptr);
2044
2045     return iLen;
2046 }
2047
2048 int wxString::PrintfV(const wxString& format, va_list argptr)
2049 {
2050     int size = 1024;
2051
2052     for ( ;; )
2053     {
2054         wxStringBuffer tmp(*this, size + 1);
2055         wxChar *buf = tmp;
2056
2057         if ( !buf )
2058         {
2059             // out of memory
2060             return -1;
2061         }
2062
2063         // wxVsnprintf() may modify the original arg pointer, so pass it
2064         // only a copy
2065         va_list argptrcopy;
2066         wxVaCopy(argptrcopy, argptr);
2067         int len = wxVsnprintf(buf, size, format, argptrcopy);
2068         va_end(argptrcopy);
2069
2070         // some implementations of vsnprintf() don't NUL terminate
2071         // the string if there is not enough space for it so
2072         // always do it manually
2073         buf[size] = _T('\0');
2074
2075         // vsnprintf() may return either -1 (traditional Unix behaviour) or the
2076         // total number of characters which would have been written if the
2077         // buffer were large enough (newer standards such as Unix98)
2078         if ( len < 0 )
2079         {
2080 #if wxUSE_WXVSNPRINTF
2081             // we know that our own implementation of wxVsnprintf() returns -1
2082             // only for a format error - thus there's something wrong with
2083             // the user's format string
2084             return -1;
2085 #else // assume that system version only returns error if not enough space
2086             // still not enough, as we don't know how much we need, double the
2087             // current size of the buffer
2088             size *= 2;
2089 #endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
2090         }
2091         else if ( len >= size )
2092         {
2093 #if wxUSE_WXVSNPRINTF
2094             // we know that our own implementation of wxVsnprintf() returns
2095             // size+1 when there's not enough space but that's not the size
2096             // of the required buffer!
2097             size *= 2;      // so we just double the current size of the buffer
2098 #else
2099             // some vsnprintf() implementations NUL-terminate the buffer and
2100             // some don't in len == size case, to be safe always add 1
2101             size = len + 1;
2102 #endif
2103         }
2104         else // ok, there was enough space
2105         {
2106             break;
2107         }
2108     }
2109
2110     // we could have overshot
2111     Shrink();
2112
2113     return length();
2114 }
2115
2116 // ----------------------------------------------------------------------------
2117 // misc other operations
2118 // ----------------------------------------------------------------------------
2119
2120 // returns true if the string matches the pattern which may contain '*' and
2121 // '?' metacharacters (as usual, '?' matches any character and '*' any number
2122 // of them)
2123 bool wxString::Matches(const wxChar *pszMask) const
2124 {
2125     // I disable this code as it doesn't seem to be faster (in fact, it seems
2126     // to be much slower) than the old, hand-written code below and using it
2127     // here requires always linking with libregex even if the user code doesn't
2128     // use it
2129 #if 0 // wxUSE_REGEX
2130     // first translate the shell-like mask into a regex
2131     wxString pattern;
2132     pattern.reserve(wxStrlen(pszMask));
2133
2134     pattern += _T('^');
2135     while ( *pszMask )
2136     {
2137         switch ( *pszMask )
2138         {
2139             case _T('?'):
2140                 pattern += _T('.');
2141                 break;
2142
2143             case _T('*'):
2144                 pattern += _T(".*");
2145                 break;
2146
2147             case _T('^'):
2148             case _T('.'):
2149             case _T('$'):
2150             case _T('('):
2151             case _T(')'):
2152             case _T('|'):
2153             case _T('+'):
2154             case _T('\\'):
2155                 // these characters are special in a RE, quote them
2156                 // (however note that we don't quote '[' and ']' to allow
2157                 // using them for Unix shell like matching)
2158                 pattern += _T('\\');
2159                 // fall through
2160
2161             default:
2162                 pattern += *pszMask;
2163         }
2164
2165         pszMask++;
2166     }
2167     pattern += _T('$');
2168
2169     // and now use it
2170     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
2171 #else // !wxUSE_REGEX
2172   // TODO: this is, of course, awfully inefficient...
2173
2174   // the char currently being checked
2175   const wxChar *pszTxt = c_str();
2176
2177   // the last location where '*' matched
2178   const wxChar *pszLastStarInText = NULL;
2179   const wxChar *pszLastStarInMask = NULL;
2180
2181 match:
2182   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
2183     switch ( *pszMask ) {
2184       case wxT('?'):
2185         if ( *pszTxt == wxT('\0') )
2186           return false;
2187
2188         // pszTxt and pszMask will be incremented in the loop statement
2189
2190         break;
2191
2192       case wxT('*'):
2193         {
2194           // remember where we started to be able to backtrack later
2195           pszLastStarInText = pszTxt;
2196           pszLastStarInMask = pszMask;
2197
2198           // ignore special chars immediately following this one
2199           // (should this be an error?)
2200           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
2201             pszMask++;
2202
2203           // if there is nothing more, match
2204           if ( *pszMask == wxT('\0') )
2205             return true;
2206
2207           // are there any other metacharacters in the mask?
2208           size_t uiLenMask;
2209           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
2210
2211           if ( pEndMask != NULL ) {
2212             // we have to match the string between two metachars
2213             uiLenMask = pEndMask - pszMask;
2214           }
2215           else {
2216             // we have to match the remainder of the string
2217             uiLenMask = wxStrlen(pszMask);
2218           }
2219
2220           wxString strToMatch(pszMask, uiLenMask);
2221           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
2222           if ( pMatch == NULL )
2223             return false;
2224
2225           // -1 to compensate "++" in the loop
2226           pszTxt = pMatch + uiLenMask - 1;
2227           pszMask += uiLenMask - 1;
2228         }
2229         break;
2230
2231       default:
2232         if ( *pszMask != *pszTxt )
2233           return false;
2234         break;
2235     }
2236   }
2237
2238   // match only if nothing left
2239   if ( *pszTxt == wxT('\0') )
2240     return true;
2241
2242   // if we failed to match, backtrack if we can
2243   if ( pszLastStarInText ) {
2244     pszTxt = pszLastStarInText + 1;
2245     pszMask = pszLastStarInMask;
2246
2247     pszLastStarInText = NULL;
2248
2249     // don't bother resetting pszLastStarInMask, it's unnecessary
2250
2251     goto match;
2252   }
2253
2254   return false;
2255 #endif // wxUSE_REGEX/!wxUSE_REGEX
2256 }
2257
2258 // Count the number of chars
2259 int wxString::Freq(wxUniChar ch) const
2260 {
2261     int count = 0;
2262     for ( const_iterator i = begin(); i != end(); ++i )
2263     {
2264         if ( *i == ch )
2265             count ++;
2266     }
2267     return count;
2268 }
2269
2270 // convert to upper case, return the copy of the string
2271 wxString wxString::Upper() const
2272 { wxString s(*this); return s.MakeUpper(); }
2273
2274 // convert to lower case, return the copy of the string
2275 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
2276
2277 // ============================================================================
2278 // ArrayString
2279 // ============================================================================
2280
2281 #include "wx/arrstr.h"
2282
2283 wxArrayString::wxArrayString(size_t sz, const wxChar** a)
2284 {
2285 #if !wxUSE_STL
2286     Init(false);
2287 #endif
2288     for (size_t i=0; i < sz; i++)
2289         Add(a[i]);
2290 }
2291
2292 wxArrayString::wxArrayString(size_t sz, const wxString* a)
2293 {
2294 #if !wxUSE_STL
2295     Init(false);
2296 #endif
2297     for (size_t i=0; i < sz; i++)
2298         Add(a[i]);
2299 }
2300
2301 #if !wxUSE_STL
2302
2303 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
2304 #define   ARRAY_MAXSIZE_INCREMENT       4096
2305
2306 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
2307 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
2308 #endif
2309
2310 // ctor
2311 void wxArrayString::Init(bool autoSort)
2312 {
2313   m_nSize  =
2314   m_nCount = 0;
2315   m_pItems = NULL;
2316   m_autoSort = autoSort;
2317 }
2318
2319 // copy ctor
2320 wxArrayString::wxArrayString(const wxArrayString& src)
2321 {
2322   Init(src.m_autoSort);
2323
2324   *this = src;
2325 }
2326
2327 // assignment operator
2328 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
2329 {
2330   if ( m_nSize > 0 )
2331     Clear();
2332
2333   Copy(src);
2334
2335   m_autoSort = src.m_autoSort;
2336
2337   return *this;
2338 }
2339
2340 void wxArrayString::Copy(const wxArrayString& src)
2341 {
2342   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
2343     Alloc(src.m_nCount);
2344
2345   for ( size_t n = 0; n < src.m_nCount; n++ )
2346     Add(src[n]);
2347 }
2348
2349 // grow the array
2350 void wxArrayString::Grow(size_t nIncrement)
2351 {
2352   // only do it if no more place
2353   if ( (m_nSize - m_nCount) < nIncrement ) {
2354     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
2355     // be never resized!
2356     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
2357       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
2358     #endif
2359
2360     if ( m_nSize == 0 ) {
2361       // was empty, alloc some memory
2362       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
2363       if (m_nSize < nIncrement)
2364           m_nSize = nIncrement;
2365       m_pItems = new wxString[m_nSize];
2366     }
2367     else {
2368       // otherwise when it's called for the first time, nIncrement would be 0
2369       // and the array would never be expanded
2370       // add 50% but not too much
2371       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
2372                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
2373       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
2374         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
2375       if ( nIncrement < ndefIncrement )
2376         nIncrement = ndefIncrement;
2377       m_nSize += nIncrement;
2378       wxString *pNew = new wxString[m_nSize];
2379
2380       // copy data to new location
2381       for ( size_t j = 0; j < m_nCount; j++ )
2382           pNew[j] = m_pItems[j];
2383
2384       // delete old memory (but do not release the strings!)
2385       wxDELETEA(m_pItems);
2386
2387       m_pItems = pNew;
2388     }
2389   }
2390 }
2391
2392 // deletes all the strings from the list
2393 void wxArrayString::Empty()
2394 {
2395   m_nCount = 0;
2396 }
2397
2398 // as Empty, but also frees memory
2399 void wxArrayString::Clear()
2400 {
2401   m_nSize  =
2402   m_nCount = 0;
2403
2404   wxDELETEA(m_pItems);
2405 }
2406
2407 // dtor
2408 wxArrayString::~wxArrayString()
2409 {
2410   wxDELETEA(m_pItems);
2411 }
2412
2413 void wxArrayString::reserve(size_t nSize)
2414 {
2415     Alloc(nSize);
2416 }
2417
2418 // pre-allocates memory (frees the previous data!)
2419 void wxArrayString::Alloc(size_t nSize)
2420 {
2421   // only if old buffer was not big enough
2422   if ( nSize > m_nSize ) {
2423     wxString *pNew = new wxString[nSize];
2424     if ( !pNew )
2425         return;
2426
2427     for ( size_t j = 0; j < m_nCount; j++ )
2428         pNew[j] = m_pItems[j];
2429     delete [] m_pItems;
2430
2431     m_pItems = pNew;
2432     m_nSize  = nSize;
2433   }
2434 }
2435
2436 // minimizes the memory usage by freeing unused memory
2437 void wxArrayString::Shrink()
2438 {
2439   // only do it if we have some memory to free
2440   if( m_nCount < m_nSize ) {
2441     // allocates exactly as much memory as we need
2442     wxString *pNew = new wxString[m_nCount];
2443
2444     // copy data to new location
2445     for ( size_t j = 0; j < m_nCount; j++ )
2446         pNew[j] = m_pItems[j];
2447     delete [] m_pItems;
2448     m_pItems = pNew;
2449   }
2450 }
2451
2452 // searches the array for an item (forward or backwards)
2453 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
2454 {
2455   if ( m_autoSort ) {
2456     // use binary search in the sorted array
2457     wxASSERT_MSG( bCase && !bFromEnd,
2458                   wxT("search parameters ignored for auto sorted array") );
2459
2460     size_t i,
2461            lo = 0,
2462            hi = m_nCount;
2463     int res;
2464     while ( lo < hi ) {
2465       i = (lo + hi)/2;
2466
2467       res = wxStrcmp(sz, m_pItems[i]);
2468       if ( res < 0 )
2469         hi = i;
2470       else if ( res > 0 )
2471         lo = i + 1;
2472       else
2473         return i;
2474     }
2475
2476     return wxNOT_FOUND;
2477   }
2478   else {
2479     // use linear search in unsorted array
2480     if ( bFromEnd ) {
2481       if ( m_nCount > 0 ) {
2482         size_t ui = m_nCount;
2483         do {
2484           if ( m_pItems[--ui].IsSameAs(sz, bCase) )
2485             return ui;
2486         }
2487         while ( ui != 0 );
2488       }
2489     }
2490     else {
2491       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2492         if( m_pItems[ui].IsSameAs(sz, bCase) )
2493           return ui;
2494       }
2495     }
2496   }
2497
2498   return wxNOT_FOUND;
2499 }
2500
2501 // add item at the end
2502 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
2503 {
2504   if ( m_autoSort ) {
2505     // insert the string at the correct position to keep the array sorted
2506     size_t i,
2507            lo = 0,
2508            hi = m_nCount;
2509     int res;
2510     while ( lo < hi ) {
2511       i = (lo + hi)/2;
2512
2513       res = str.Cmp(m_pItems[i]);
2514       if ( res < 0 )
2515         hi = i;
2516       else if ( res > 0 )
2517         lo = i + 1;
2518       else {
2519         lo = hi = i;
2520         break;
2521       }
2522     }
2523
2524     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2525
2526     Insert(str, lo, nInsert);
2527
2528     return (size_t)lo;
2529   }
2530   else {
2531     Grow(nInsert);
2532
2533     for (size_t i = 0; i < nInsert; i++)
2534     {
2535         // just append
2536         m_pItems[m_nCount + i] = str;
2537     }
2538     size_t ret = m_nCount;
2539     m_nCount += nInsert;
2540     return ret;
2541   }
2542 }
2543
2544 // add item at the given position
2545 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2546 {
2547   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2548   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2549                wxT("array size overflow in wxArrayString::Insert") );
2550
2551   Grow(nInsert);
2552
2553   for (int j = m_nCount - nIndex - 1; j >= 0; j--)
2554       m_pItems[nIndex + nInsert + j] = m_pItems[nIndex + j];
2555
2556   for (size_t i = 0; i < nInsert; i++)
2557   {
2558       m_pItems[nIndex + i] = str;
2559   }
2560   m_nCount += nInsert;
2561 }
2562
2563 // range insert (STL 23.2.4.3)
2564 void
2565 wxArrayString::insert(iterator it, const_iterator first, const_iterator last)
2566 {
2567     const int idx = it - begin();
2568
2569     // grow it once
2570     Grow(last - first);
2571
2572     // reset "it" since it can change inside Grow()
2573     it = begin() + idx;
2574
2575     while ( first != last )
2576     {
2577         it = insert(it, *first);
2578
2579         // insert returns an iterator to the last element inserted but we need
2580         // insert the next after this one, that is before the next one
2581         ++it;
2582
2583         ++first;
2584     }
2585 }
2586
2587 // expand the array
2588 void wxArrayString::SetCount(size_t count)
2589 {
2590     Alloc(count);
2591
2592     wxString s;
2593     while ( m_nCount < count )
2594         m_pItems[m_nCount++] = s;
2595 }
2596
2597 // removes item from array (by index)
2598 void wxArrayString::RemoveAt(size_t nIndex, size_t nRemove)
2599 {
2600   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2601   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2602                wxT("removing too many elements in wxArrayString::Remove") );
2603
2604   for ( size_t j =  0; j < m_nCount - nIndex -nRemove; j++)
2605       m_pItems[nIndex + j] = m_pItems[nIndex + nRemove + j];
2606
2607   m_nCount -= nRemove;
2608 }
2609
2610 // removes item from array (by value)
2611 void wxArrayString::Remove(const wxChar *sz)
2612 {
2613   int iIndex = Index(sz);
2614
2615   wxCHECK_RET( iIndex != wxNOT_FOUND,
2616                wxT("removing inexistent element in wxArrayString::Remove") );
2617
2618   RemoveAt(iIndex);
2619 }
2620
2621 void wxArrayString::assign(const_iterator first, const_iterator last)
2622 {
2623     reserve(last - first);
2624     for(; first != last; ++first)
2625         push_back(*first);
2626 }
2627
2628 // ----------------------------------------------------------------------------
2629 // sorting
2630 // ----------------------------------------------------------------------------
2631
2632 // we can only sort one array at a time with the quick-sort based
2633 // implementation
2634 #if wxUSE_THREADS
2635   // need a critical section to protect access to gs_compareFunction and
2636   // gs_sortAscending variables
2637   static wxCriticalSection gs_critsectStringSort;
2638 #endif // wxUSE_THREADS
2639
2640 // function to use for string comparaison
2641 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2642
2643 // if we don't use the compare function, this flag tells us if we sort the
2644 // array in ascending or descending order
2645 static bool gs_sortAscending = true;
2646
2647 // function which is called by quick sort
2648 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2649 wxStringCompareFunction(const void *first, const void *second)
2650 {
2651   wxString *strFirst = (wxString *)first;
2652   wxString *strSecond = (wxString *)second;
2653
2654   if ( gs_compareFunction ) {
2655     return gs_compareFunction(*strFirst, *strSecond);
2656   }
2657   else {
2658     // maybe we should use wxStrcoll
2659     int result = strFirst->Cmp(*strSecond);
2660
2661     return gs_sortAscending ? result : -result;
2662   }
2663 }
2664
2665 // sort array elements using passed comparaison function
2666 void wxArrayString::Sort(CompareFunction compareFunction)
2667 {
2668   wxCRIT_SECT_LOCKER(lockCmpFunc, gs_critsectStringSort);
2669
2670   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2671   gs_compareFunction = compareFunction;
2672
2673   DoSort();
2674
2675   // reset it to NULL so that Sort(bool) will work the next time
2676   gs_compareFunction = NULL;
2677 }
2678
2679 extern "C"
2680 {
2681     typedef int (wxC_CALLING_CONV * wxStringCompareFn)(const void *first,
2682                                                        const void *second);
2683 }
2684
2685 void wxArrayString::Sort(CompareFunction2 compareFunction)
2686 {
2687   qsort(m_pItems, m_nCount, sizeof(wxString), (wxStringCompareFn)compareFunction);
2688 }
2689
2690 void wxArrayString::Sort(bool reverseOrder)
2691 {
2692   Sort(reverseOrder ? wxStringSortDescending : wxStringSortAscending);
2693 }
2694
2695 void wxArrayString::DoSort()
2696 {
2697   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2698
2699   qsort(m_pItems, m_nCount, sizeof(wxString), wxStringCompareFunction);
2700 }
2701
2702 bool wxArrayString::operator==(const wxArrayString& a) const
2703 {
2704     if ( m_nCount != a.m_nCount )
2705         return false;
2706
2707     for ( size_t n = 0; n < m_nCount; n++ )
2708     {
2709         if ( Item(n) != a[n] )
2710             return false;
2711     }
2712
2713     return true;
2714 }
2715
2716 #endif // !wxUSE_STL
2717
2718 int wxCMPFUNC_CONV wxStringSortAscending(wxString* s1, wxString* s2)
2719 {
2720     return  s1->Cmp(*s2);
2721 }
2722
2723 int wxCMPFUNC_CONV wxStringSortDescending(wxString* s1, wxString* s2)
2724 {
2725     return -s1->Cmp(*s2);
2726 }
2727
2728
2729
2730 // ===========================================================================
2731 // wxJoin and wxSplit
2732 // ===========================================================================
2733
2734 #include "wx/tokenzr.h"
2735
2736 wxString wxJoin(const wxArrayString& arr, const wxChar sep, const wxChar escape)
2737 {
2738     size_t count = arr.size();
2739     if ( count == 0 )
2740         return wxEmptyString;
2741
2742     wxString str;
2743
2744     // pre-allocate memory using the estimation of the average length of the
2745     // strings in the given array: this is very imprecise, of course, but
2746     // better than nothing
2747     str.reserve(count*(arr[0].length() + arr[count-1].length()) / 2);
2748
2749     if ( escape == wxT('\0') )
2750     {
2751         // escaping is disabled:
2752         for ( size_t i = 0; i < count; i++ )
2753         {
2754             if ( i )
2755                 str += sep;
2756             str += arr[i];
2757         }
2758     }
2759     else // use escape character
2760     {
2761         for ( size_t n = 0; n < count; n++ )
2762         {
2763             if ( n )
2764                 str += sep;
2765
2766             for ( wxString::const_iterator i = arr[n].begin(),
2767                                          end = arr[n].end();
2768                   i != end;
2769                   ++i )
2770             {
2771                 const wxChar ch = *i;
2772                 if ( ch == sep )
2773                     str += escape;      // escape this separator
2774                 str += ch;
2775             }
2776         }
2777     }
2778
2779     str.Shrink(); // release extra memory if we allocated too much
2780     return str;
2781 }
2782
2783 wxArrayString wxSplit(const wxString& str, const wxChar sep, const wxChar escape)
2784 {
2785     if ( escape == wxT('\0') )
2786     {
2787         // simple case: we don't need to honour the escape character
2788         return wxStringTokenize(str, sep, wxTOKEN_RET_EMPTY_ALL);
2789     }
2790
2791     wxArrayString ret;
2792     wxString curr;
2793     wxChar prev = wxT('\0');
2794
2795     for ( wxString::const_iterator i = str.begin(),
2796                                  end = str.end();
2797           i != end;
2798           ++i )
2799     {
2800         const wxChar ch = *i;
2801
2802         if ( ch == sep )
2803         {
2804             if ( prev == escape )
2805             {
2806                 // remove the escape character and don't consider this
2807                 // occurrence of 'sep' as a real separator
2808                 *curr.rbegin() = sep;
2809             }
2810             else // real separator
2811             {
2812                 ret.push_back(curr);
2813                 curr.clear();
2814             }
2815         }
2816         else // normal character
2817         {
2818             curr += ch;
2819         }
2820
2821         prev = ch;
2822     }
2823
2824     // add the last token
2825     if ( !curr.empty() || prev == sep )
2826         ret.Add(curr);
2827
2828     return ret;
2829 }