src/common/string.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        string.cpp
   3 // Purpose:     wxString class
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows licence
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifdef __GNUG__
  13   #pragma implementation "string.h"
  14 #endif
  15
  16 /*
  17  * About ref counting:
  18  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  19  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  20  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  21  */
  22
  23 // ===========================================================================
  24 // headers, declarations, constants
  25 // ===========================================================================
  26
  27 // For compilers that support precompilation, includes "wx.h".
  28 #include "wx/wxprec.h"
  29
  30 #ifdef __BORLANDC__
  31   #pragma hdrstop
  32 #endif
  33
  34 #ifndef WX_PRECOMP
  35   #include "wx/defs.h"
  36   #include "wx/string.h"
  37   #include "wx/intl.h"
  38   #include "wx/thread.h"
  39 #endif
  40
  41 #include <ctype.h>
  42 #include <string.h>
  43 #include <stdlib.h>
  44
  45 #ifdef __SALFORDC__
  46   #include <clib.h>
  47 #endif
  48
  49 // allocating extra space for each string consumes more memory but speeds up
  50 // the concatenation operations (nLen is the current string's length)
  51 // NB: EXTRA_ALLOC must be >= 0!
  52 #define EXTRA_ALLOC       (19 - nLen % 16)
  53
  54 // ---------------------------------------------------------------------------
  55 // static class variables definition
  56 // ---------------------------------------------------------------------------
  57
  58 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  59 // must define this static for VA or else you get multiply defined symbols
  60 // everywhere
  61 const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100;
  62 #endif // Visual Age
  63
  64 #ifdef  wxSTD_STRING_COMPATIBILITY
  65   const size_t wxString::npos = wxSTRING_MAXLEN;
  66 #endif // wxSTD_STRING_COMPATIBILITY
  67
  68 // ----------------------------------------------------------------------------
  69 // static data
  70 // ----------------------------------------------------------------------------
  71
  72 // for an empty string, GetStringData() will return this address: this
  73 // structure has the same layout as wxStringData and it's data() method will
  74 // return the empty string (dummy pointer)
  75 static const struct
  76 {
  77   wxStringData data;
  78   wxChar dummy;
  79 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  80
  81 // empty C style string: points to 'string data' byte of g_strEmpty
  82 extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy;
  83
  84 // ----------------------------------------------------------------------------
  85 // global functions
  86 // ----------------------------------------------------------------------------
  87
  88 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
  89
  90 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
  91 // iostream ones.
  92 //
  93 // ATTN: you can _not_ use both of these in the same program!
  94
  95 wxSTD istream& operator>>(wxSTD istream& is, wxString& WXUNUSED(str))
  96 {
  97 #if 0
  98   int w = is.width(0);
  99   if ( is.ipfx(0) ) {
 100     streambuf *sb = is.rdbuf();
 101     str.erase();
 102     while ( true ) {
 103       int ch = sb->sbumpc ();
 104       if ( ch == EOF ) {
 105         is.setstate(ios::eofbit);
 106         break;
 107       }
 108       else if ( isspace(ch) ) {
 109         sb->sungetc();
 110         break;
 111       }
 112
 113       str += ch;
 114       if ( --w == 1 )
 115         break;
 116     }
 117   }
 118
 119   is.isfx();
 120   if ( str.length() == 0 )
 121     is.setstate(ios::failbit);
 122 #endif
 123   return is;
 124 }
 125
 126 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
 127 {
 128   os << str.c_str();
 129   return os;
 130 }
 131
 132 #endif  //std::string compatibility
 133
 134 // ----------------------------------------------------------------------------
 135 // private classes
 136 // ----------------------------------------------------------------------------
 137
 138 // this small class is used to gather statistics for performance tuning
 139 //#define WXSTRING_STATISTICS
 140 #ifdef  WXSTRING_STATISTICS
 141   class Averager
 142   {
 143   public:
 144     Averager(const wxChar *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 145    ~Averager()
 146    { wxPrintf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 147
 148     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 149
 150   private:
 151     size_t m_nCount, m_nTotal;
 152     const wxChar *m_sz;
 153   } g_averageLength("allocation size"),
 154     g_averageSummandLength("summand length"),
 155     g_averageConcatHit("hit probability in concat"),
 156     g_averageInitialLength("initial string length");
 157
 158   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 159 #else
 160   #define STATISTICS_ADD(av, val)
 161 #endif // WXSTRING_STATISTICS
 162
 163 // ===========================================================================
 164 // wxStringData class deallocation
 165 // ===========================================================================
 166
 167 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 168 #  pragma message (__FILE__ ": building with Multithreaded non DLL runtime has a performance impact on wxString!")
 169 void wxStringData::Free()
 170 {
 171     free(this);
 172 }
 173 #endif
 174
 175 // ===========================================================================
 176 // wxString class core
 177 // ===========================================================================
 178
 179 // ---------------------------------------------------------------------------
 180 // construction
 181 // ---------------------------------------------------------------------------
 182
 183 // constructs string of <nLength> copies of character <ch>
 184 wxString::wxString(wxChar ch, size_t nLength)
 185 {
 186   Init();
 187
 188   if ( nLength > 0 ) {
 189     if ( !AllocBuffer(nLength) ) {
 190       wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 191       return;
 192     }
 193
 194 #if wxUSE_UNICODE
 195     // memset only works on chars
 196     for ( size_t n = 0; n < nLength; n++ )
 197         m_pchData[n] = ch;
 198 #else
 199     memset(m_pchData, ch, nLength);
 200 #endif
 201   }
 202 }
 203
 204 // takes nLength elements of psz starting at nPos
 205 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 206 {
 207   Init();
 208
 209   // if the length is not given, assume the string to be NUL terminated
 210   if ( nLength == wxSTRING_MAXLEN ) {
 211     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 212
 213     nLength = wxStrlen(psz + nPos);
 214   }
 215
 216   STATISTICS_ADD(InitialLength, nLength);
 217
 218   if ( nLength > 0 ) {
 219     // trailing '\0' is written in AllocBuffer()
 220     if ( !AllocBuffer(nLength) ) {
 221       wxFAIL_MSG( _T("out of memory in wxString::InitWith") );
 222       return;
 223     }
 224     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 225   }
 226 }
 227
 228 #ifdef  wxSTD_STRING_COMPATIBILITY
 229
 230 // poor man's iterators are "void *" pointers
 231 wxString::wxString(const void *pStart, const void *pEnd)
 232 {
 233   InitWith((const wxChar *)pStart, 0,
 234            (const wxChar *)pEnd - (const wxChar *)pStart);
 235 }
 236
 237 #endif  //std::string compatibility
 238
 239 #if wxUSE_UNICODE
 240
 241 // from multibyte string
 242 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 243 {
 244     // first get the size of the buffer we need
 245     size_t nLen;
 246     if ( psz )
 247     {
 248         // calculate the needed size ourselves or use a provide one
 249         nLen = nLength == wxSTRING_MAXLEN ? conv.MB2WC(NULL, psz, 0) : nLength;
 250     }
 251     else
 252     {
 253         // nothing to convert
 254         nLen = 0;
 255     }
 256
 257     // anything to do?
 258     if ( (nLen != 0) && (nLen != (size_t)-1) )
 259     {
 260         if ( !AllocBuffer(nLen) )
 261         {
 262             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 263             return;
 264         }
 265
 266         // MB2WC wants the buffer size, not the string length
 267         if ( conv.MB2WC(m_pchData, psz, nLen + 1) != (size_t)-1 )
 268         {
 269             // initialized ok
 270             m_pchData[nLen] = 0;
 271             return;
 272         }
 273         //else: the conversion failed -- leave the string empty (what else?)
 274     }
 275
 276     Init();
 277 }
 278
 279 #else // ANSI
 280
 281 #if wxUSE_WCHAR_T
 282 // from wide string
 283 wxString::wxString(const wchar_t *pwz, wxMBConv& conv, size_t nLength)
 284 {
 285     // first get the size of the buffer we need
 286     size_t nLen;
 287     if ( pwz )
 288     {
 289         // calculate the needed size ourselves or use a provide one
 290         nLen = nLength == wxSTRING_MAXLEN ? conv.WC2MB(NULL, pwz, 0) : nLength;
 291     }
 292     else
 293     {
 294         // nothing to convert
 295         nLen = 0;
 296     }
 297
 298     // anything to do?
 299     if ( (nLen != 0) && (nLen != (size_t)-1) )
 300     {
 301         if ( !AllocBuffer(nLen) )
 302         {
 303             wxFAIL_MSG( _T("out of memory in wxString::wxString") );
 304             return;
 305         }
 306
 307         // WC2MB wants the buffer size, not the string length
 308         if ( conv.WC2MB(m_pchData, pwz, nLen + 1) != (size_t)-1 )
 309         {
 310             // initialized ok
 311             return;
 312         }
 313         //else: the conversion failed -- leave the string empty (what else?)
 314     }
 315
 316     Init();
 317 }
 318 #endif // wxUSE_WCHAR_T
 319
 320 #endif // Unicode/ANSI
 321
 322 // ---------------------------------------------------------------------------
 323 // memory allocation
 324 // ---------------------------------------------------------------------------
 325
 326 // allocates memory needed to store a C string of length nLen
 327 bool wxString::AllocBuffer(size_t nLen)
 328 {
 329   // allocating 0 sized buffer doesn't make sense, all empty strings should
 330   // reuse g_strEmpty
 331   wxASSERT( nLen >  0 );
 332
 333   // make sure that we don't overflow
 334   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 335                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 336
 337   STATISTICS_ADD(Length, nLen);
 338
 339   // allocate memory:
 340   // 1) one extra character for '\0' termination
 341   // 2) sizeof(wxStringData) for housekeeping info
 342   wxStringData* pData = (wxStringData*)
 343     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 344
 345   if ( pData == NULL ) {
 346     // allocation failures are handled by the caller
 347     return FALSE;
 348   }
 349
 350   pData->nRefs        = 1;
 351   pData->nDataLength  = nLen;
 352   pData->nAllocLength = nLen + EXTRA_ALLOC;
 353   m_pchData           = pData->data();  // data starts after wxStringData
 354   m_pchData[nLen]     = wxT('\0');
 355   return TRUE;
 356 }
 357
 358 // must be called before changing this string
 359 bool wxString::CopyBeforeWrite()
 360 {
 361   wxStringData* pData = GetStringData();
 362
 363   if ( pData->IsShared() ) {
 364     pData->Unlock();                // memory not freed because shared
 365     size_t nLen = pData->nDataLength;
 366     if ( !AllocBuffer(nLen) ) {
 367       // allocation failures are handled by the caller
 368       return FALSE;
 369     }
 370     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 371   }
 372
 373   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 374
 375   return TRUE;
 376 }
 377
 378 // must be called before replacing contents of this string
 379 bool wxString::AllocBeforeWrite(size_t nLen)
 380 {
 381   wxASSERT( nLen != 0 );  // doesn't make any sense
 382
 383   // must not share string and must have enough space
 384   wxStringData* pData = GetStringData();
 385   if ( pData->IsShared() || pData->IsEmpty() ) {
 386     // can't work with old buffer, get new one
 387     pData->Unlock();
 388     if ( !AllocBuffer(nLen) ) {
 389       // allocation failures are handled by the caller
 390       return FALSE;
 391     }
 392   }
 393   else {
 394     if ( nLen > pData->nAllocLength ) {
 395       // realloc the buffer instead of calling malloc() again, this is more
 396       // efficient
 397       STATISTICS_ADD(Length, nLen);
 398
 399       nLen += EXTRA_ALLOC;
 400
 401       pData = (wxStringData*)
 402           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 403
 404       if ( pData == NULL ) {
 405         // allocation failures are handled by the caller
 406         // keep previous data since reallocation failed
 407         return FALSE;
 408       }
 409
 410       pData->nAllocLength = nLen;
 411       m_pchData = pData->data();
 412     }
 413
 414     // now we have enough space, just update the string length
 415     pData->nDataLength = nLen;
 416   }
 417
 418   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 419
 420   return TRUE;
 421 }
 422
 423 // allocate enough memory for nLen characters
 424 bool wxString::Alloc(size_t nLen)
 425 {
 426   wxStringData *pData = GetStringData();
 427   if ( pData->nAllocLength <= nLen ) {
 428     if ( pData->IsEmpty() ) {
 429       nLen += EXTRA_ALLOC;
 430
 431       wxStringData* pData = (wxStringData*)
 432           malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 433
 434       if ( pData == NULL ) {
 435         // allocation failure handled by caller
 436         return FALSE;
 437       }
 438
 439       pData->nRefs = 1;
 440       pData->nDataLength = 0;
 441       pData->nAllocLength = nLen;
 442       m_pchData = pData->data();  // data starts after wxStringData
 443       m_pchData[0u] = wxT('\0');
 444     }
 445     else if ( pData->IsShared() ) {
 446       pData->Unlock();                // memory not freed because shared
 447       size_t nOldLen = pData->nDataLength;
 448       if ( !AllocBuffer(nLen) ) {
 449         // allocation failure handled by caller
 450         return FALSE;
 451       }
 452       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 453     }
 454     else {
 455       nLen += EXTRA_ALLOC;
 456
 457       pData = (wxStringData *)
 458         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 459
 460       if ( pData == NULL ) {
 461         // allocation failure handled by caller
 462         // keep previous data since reallocation failed
 463         return FALSE;
 464       }
 465
 466       // it's not important if the pointer changed or not (the check for this
 467       // is not faster than assigning to m_pchData in all cases)
 468       pData->nAllocLength = nLen;
 469       m_pchData = pData->data();
 470     }
 471   }
 472   //else: we've already got enough
 473   return TRUE;
 474 }
 475
 476 // shrink to minimal size (releasing extra memory)
 477 bool wxString::Shrink()
 478 {
 479   wxStringData *pData = GetStringData();
 480
 481   size_t nLen = pData->nDataLength;
 482   void *p = realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 483
 484   if ( p == NULL) {
 485       wxFAIL_MSG( _T("out of memory reallocating wxString data") );
 486       // keep previous data since reallocation failed
 487       return FALSE;
 488   }
 489
 490   if ( p != pData )
 491   {
 492       // contrary to what one might believe, some realloc() implementation do
 493       // move the memory block even when its size is reduced
 494       pData = (wxStringData *)p;
 495
 496       m_pchData = pData->data();
 497   }
 498
 499   pData->nAllocLength = nLen;
 500
 501   return TRUE;
 502 }
 503
 504 // get the pointer to writable buffer of (at least) nLen bytes
 505 wxChar *wxString::GetWriteBuf(size_t nLen)
 506 {
 507   if ( !AllocBeforeWrite(nLen) ) {
 508     // allocation failure handled by caller
 509     return NULL;
 510   }
 511
 512   wxASSERT( GetStringData()->nRefs == 1 );
 513   GetStringData()->Validate(FALSE);
 514
 515   return m_pchData;
 516 }
 517
 518 // put string back in a reasonable state after GetWriteBuf
 519 void wxString::UngetWriteBuf()
 520 {
 521   GetStringData()->nDataLength = wxStrlen(m_pchData);
 522   GetStringData()->Validate(TRUE);
 523 }
 524
 525 void wxString::UngetWriteBuf(size_t nLen)
 526 {
 527   GetStringData()->nDataLength = nLen;
 528   GetStringData()->Validate(TRUE);
 529 }
 530
 531 // ---------------------------------------------------------------------------
 532 // data access
 533 // ---------------------------------------------------------------------------
 534
 535 // all functions are inline in string.h
 536
 537 // ---------------------------------------------------------------------------
 538 // assignment operators
 539 // ---------------------------------------------------------------------------
 540
 541 // helper function: does real copy
 542 bool wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 543 {
 544   if ( nSrcLen == 0 ) {
 545     Reinit();
 546   }
 547   else {
 548     if ( !AllocBeforeWrite(nSrcLen) ) {
 549       // allocation failure handled by caller
 550       return FALSE;
 551     }
 552     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 553     GetStringData()->nDataLength = nSrcLen;
 554     m_pchData[nSrcLen] = wxT('\0');
 555   }
 556   return TRUE;
 557 }
 558
 559 // assigns one string to another
 560 wxString& wxString::operator=(const wxString& stringSrc)
 561 {
 562   wxASSERT( stringSrc.GetStringData()->IsValid() );
 563
 564   // don't copy string over itself
 565   if ( m_pchData != stringSrc.m_pchData ) {
 566     if ( stringSrc.GetStringData()->IsEmpty() ) {
 567       Reinit();
 568     }
 569     else {
 570       // adjust references
 571       GetStringData()->Unlock();
 572       m_pchData = stringSrc.m_pchData;
 573       GetStringData()->Lock();
 574     }
 575   }
 576
 577   return *this;
 578 }
 579
 580 // assigns a single character
 581 wxString& wxString::operator=(wxChar ch)
 582 {
 583   if ( !AssignCopy(1, &ch) ) {
 584     wxFAIL_MSG( _T("out of memory in wxString::operator=(wxChar)") );
 585   }
 586   return *this;
 587 }
 588
 589
 590 // assigns C string
 591 wxString& wxString::operator=(const wxChar *psz)
 592 {
 593   if ( !AssignCopy(wxStrlen(psz), psz) ) {
 594     wxFAIL_MSG( _T("out of memory in wxString::operator=(const wxChar *)") );
 595   }
 596   return *this;
 597 }
 598
 599 #if !wxUSE_UNICODE
 600
 601 // same as 'signed char' variant
 602 wxString& wxString::operator=(const unsigned char* psz)
 603 {
 604   *this = (const char *)psz;
 605   return *this;
 606 }
 607
 608 #if wxUSE_WCHAR_T
 609 wxString& wxString::operator=(const wchar_t *pwz)
 610 {
 611   wxString str(pwz);
 612   *this = str;
 613   return *this;
 614 }
 615 #endif
 616
 617 #endif
 618
 619 // ---------------------------------------------------------------------------
 620 // string concatenation
 621 // ---------------------------------------------------------------------------
 622
 623 // add something to this string
 624 bool wxString::ConcatSelf(size_t nSrcLen, const wxChar *pszSrcData)
 625 {
 626   STATISTICS_ADD(SummandLength, nSrcLen);
 627
 628   // concatenating an empty string is a NOP
 629   if ( nSrcLen > 0 ) {
 630     wxStringData *pData = GetStringData();
 631     size_t nLen = pData->nDataLength;
 632     size_t nNewLen = nLen + nSrcLen;
 633
 634     // alloc new buffer if current is too small
 635     if ( pData->IsShared() ) {
 636       STATISTICS_ADD(ConcatHit, 0);
 637
 638       // we have to allocate another buffer
 639       wxStringData* pOldData = GetStringData();
 640       if ( !AllocBuffer(nNewLen) ) {
 641           // allocation failure handled by caller
 642           return FALSE;
 643       }
 644       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 645       pOldData->Unlock();
 646     }
 647     else if ( nNewLen > pData->nAllocLength ) {
 648       STATISTICS_ADD(ConcatHit, 0);
 649
 650       // we have to grow the buffer
 651       if ( !Alloc(nNewLen) ) {
 652           // allocation failure handled by caller
 653           return FALSE;
 654       }
 655     }
 656     else {
 657       STATISTICS_ADD(ConcatHit, 1);
 658
 659       // the buffer is already big enough
 660     }
 661
 662     // should be enough space
 663     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 664
 665     // fast concatenation - all is done in our buffer
 666     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 667
 668     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 669     GetStringData()->nDataLength = nNewLen; // and fix the length
 670   }
 671   //else: the string to append was empty
 672   return TRUE;
 673 }
 674
 675 /*
 676  * concatenation functions come in 5 flavours:
 677  *  string + string
 678  *  char   + string      and      string + char
 679  *  C str  + string      and      string + C str
 680  */
 681
 682 wxString operator+(const wxString& str1, const wxString& str2)
 683 {
 684   wxASSERT( str1.GetStringData()->IsValid() );
 685   wxASSERT( str2.GetStringData()->IsValid() );
 686
 687   wxString s = str1;
 688   s += str2;
 689
 690   return s;
 691 }
 692
 693 wxString operator+(const wxString& str, wxChar ch)
 694 {
 695   wxASSERT( str.GetStringData()->IsValid() );
 696
 697   wxString s = str;
 698   s += ch;
 699
 700   return s;
 701 }
 702
 703 wxString operator+(wxChar ch, const wxString& str)
 704 {
 705   wxASSERT( str.GetStringData()->IsValid() );
 706
 707   wxString s = ch;
 708   s += str;
 709
 710   return s;
 711 }
 712
 713 wxString operator+(const wxString& str, const wxChar *psz)
 714 {
 715   wxASSERT( str.GetStringData()->IsValid() );
 716
 717   wxString s;
 718   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 719     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 720   }
 721   s = str;
 722   s += psz;
 723
 724   return s;
 725 }
 726
 727 wxString operator+(const wxChar *psz, const wxString& str)
 728 {
 729   wxASSERT( str.GetStringData()->IsValid() );
 730
 731   wxString s;
 732   if ( !s.Alloc(wxStrlen(psz) + str.Len()) ) {
 733     wxFAIL_MSG( _T("out of memory in wxString::operator+") );
 734   }
 735   s = psz;
 736   s += str;
 737
 738   return s;
 739 }
 740
 741 // ===========================================================================
 742 // other common string functions
 743 // ===========================================================================
 744
 745 #if wxUSE_UNICODE
 746
 747 wxString wxString::FromAscii(const char *ascii)
 748 {
 749     if (!ascii)
 750        return wxEmptyString;
 751
 752     size_t len = strlen( ascii );
 753     wxString res;
 754
 755     if ( len )
 756     {
 757         wxStringBuffer buf(res, len);
 758
 759         wchar_t *dest = buf;
 760
 761         for ( ;; )
 762         {
 763            if ( (*dest++ = (wchar_t)(unsigned char)*ascii++) == L'\0' )
 764                break;
 765         }
 766     }
 767
 768     return res;
 769 }
 770
 771 wxString wxString::FromAscii(const char ascii)
 772 {
 773     // What do we do with '\0' ?
 774
 775     wxString res;
 776     res += (wchar_t)(unsigned char) ascii;
 777
 778     return res;
 779 }
 780
 781 const wxCharBuffer wxString::ToAscii() const
 782 {
 783     // this will allocate enough space for the terminating NUL too
 784     wxCharBuffer buffer(length());
 785
 786     signed char *dest = (signed char *)buffer.data();
 787
 788     const wchar_t *pwc = c_str();
 789     for ( ;; )
 790     {
 791         *dest++ = *pwc > SCHAR_MAX ? '_' : *pwc;
 792
 793         // the output string can't have embedded NULs anyhow, so we can safely
 794         // stop at first of them even if we do have any
 795         if ( !*pwc++ )
 796             break;
 797     }
 798
 799     return buffer;
 800 }
 801
 802 #endif // Unicode
 803
 804 // ---------------------------------------------------------------------------
 805 // simple sub-string extraction
 806 // ---------------------------------------------------------------------------
 807
 808 // helper function: clone the data attached to this string
 809 bool wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 810 {
 811   if ( nCopyLen == 0 ) {
 812     dest.Init();
 813   }
 814   else {
 815     if ( !dest.AllocBuffer(nCopyLen) ) {
 816       // allocation failure handled by caller
 817       return FALSE;
 818     }
 819     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 820   }
 821   return TRUE;
 822 }
 823
 824 // extract string of length nCount starting at nFirst
 825 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 826 {
 827   wxStringData *pData = GetStringData();
 828   size_t nLen = pData->nDataLength;
 829
 830   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 831   if ( nCount == wxSTRING_MAXLEN )
 832   {
 833     nCount = nLen - nFirst;
 834   }
 835
 836   // out-of-bounds requests return sensible things
 837   if ( nFirst + nCount > nLen )
 838   {
 839     nCount = nLen - nFirst;
 840   }
 841
 842   if ( nFirst > nLen )
 843   {
 844     // AllocCopy() will return empty string
 845     nCount = 0;
 846   }
 847
 848   wxString dest;
 849   if ( !AllocCopy(dest, nCount, nFirst) ) {
 850       wxFAIL_MSG( _T("out of memory in wxString::Mid") );
 851   }
 852
 853   return dest;
 854 }
 855
 856 // check that the tring starts with prefix and return the rest of the string
 857 // in the provided pointer if it is not NULL, otherwise return FALSE
 858 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 859 {
 860     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 861
 862     // first check if the beginning of the string matches the prefix: note
 863     // that we don't have to check that we don't run out of this string as
 864     // when we reach the terminating NUL, either prefix string ends too (and
 865     // then it's ok) or we break out of the loop because there is no match
 866     const wxChar *p = c_str();
 867     while ( *prefix )
 868     {
 869         if ( *prefix++ != *p++ )
 870         {
 871             // no match
 872             return FALSE;
 873         }
 874     }
 875
 876     if ( rest )
 877     {
 878         // put the rest of the string into provided pointer
 879         *rest = p;
 880     }
 881
 882     return TRUE;
 883 }
 884
 885 // extract nCount last (rightmost) characters
 886 wxString wxString::Right(size_t nCount) const
 887 {
 888   if ( nCount > (size_t)GetStringData()->nDataLength )
 889     nCount = GetStringData()->nDataLength;
 890
 891   wxString dest;
 892   if ( !AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount) ) {
 893     wxFAIL_MSG( _T("out of memory in wxString::Right") );
 894   }
 895   return dest;
 896 }
 897
 898 // get all characters after the last occurence of ch
 899 // (returns the whole string if ch not found)
 900 wxString wxString::AfterLast(wxChar ch) const
 901 {
 902   wxString str;
 903   int iPos = Find(ch, TRUE);
 904   if ( iPos == wxNOT_FOUND )
 905     str = *this;
 906   else
 907     str = c_str() + iPos + 1;
 908
 909   return str;
 910 }
 911
 912 // extract nCount first (leftmost) characters
 913 wxString wxString::Left(size_t nCount) const
 914 {
 915   if ( nCount > (size_t)GetStringData()->nDataLength )
 916     nCount = GetStringData()->nDataLength;
 917
 918   wxString dest;
 919   if ( !AllocCopy(dest, nCount, 0) ) {
 920     wxFAIL_MSG( _T("out of memory in wxString::Left") );
 921   }
 922   return dest;
 923 }
 924
 925 // get all characters before the first occurence of ch
 926 // (returns the whole string if ch not found)
 927 wxString wxString::BeforeFirst(wxChar ch) const
 928 {
 929   wxString str;
 930   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 931     str += *pc;
 932
 933   return str;
 934 }
 935
 936 /// get all characters before the last occurence of ch
 937 /// (returns empty string if ch not found)
 938 wxString wxString::BeforeLast(wxChar ch) const
 939 {
 940   wxString str;
 941   int iPos = Find(ch, TRUE);
 942   if ( iPos != wxNOT_FOUND && iPos != 0 )
 943     str = wxString(c_str(), iPos);
 944
 945   return str;
 946 }
 947
 948 /// get all characters after the first occurence of ch
 949 /// (returns empty string if ch not found)
 950 wxString wxString::AfterFirst(wxChar ch) const
 951 {
 952   wxString str;
 953   int iPos = Find(ch);
 954   if ( iPos != wxNOT_FOUND )
 955     str = c_str() + iPos + 1;
 956
 957   return str;
 958 }
 959
 960 // replace first (or all) occurences of some substring with another one
 961 size_t
 962 wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 963 {
 964     // if we tried to replace an empty string we'd enter an infinite loop below
 965     wxCHECK_MSG( szOld && *szOld && szNew, 0,
 966                  _T("wxString::Replace(): invalid parameter") );
 967
 968   size_t uiCount = 0;   // count of replacements made
 969
 970   size_t uiOldLen = wxStrlen(szOld);
 971
 972   wxString strTemp;
 973   const wxChar *pCurrent = m_pchData;
 974   const wxChar *pSubstr;
 975   while ( *pCurrent != wxT('\0') ) {
 976     pSubstr = wxStrstr(pCurrent, szOld);
 977     if ( pSubstr == NULL ) {
 978       // strTemp is unused if no replacements were made, so avoid the copy
 979       if ( uiCount == 0 )
 980         return 0;
 981
 982       strTemp += pCurrent;    // copy the rest
 983       break;                  // exit the loop
 984     }
 985     else {
 986       // take chars before match
 987       if ( !strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent) ) {
 988         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
 989         return 0;
 990       }
 991       strTemp += szNew;
 992       pCurrent = pSubstr + uiOldLen;  // restart after match
 993
 994       uiCount++;
 995
 996       // stop now?
 997       if ( !bReplaceAll ) {
 998         strTemp += pCurrent;    // copy the rest
 999         break;                  // exit the loop
1000       }
1001     }
1002   }
1003
1004   // only done if there were replacements, otherwise would have returned above
1005   *this = strTemp;
1006
1007   return uiCount;
1008 }
1009
1010 bool wxString::IsAscii() const
1011 {
1012   const wxChar *s = (const wxChar*) *this;
1013   while(*s){
1014     if(!isascii(*s)) return(FALSE);
1015     s++;
1016   }
1017   return(TRUE);
1018 }
1019
1020 bool wxString::IsWord() const
1021 {
1022   const wxChar *s = (const wxChar*) *this;
1023   while(*s){
1024     if(!wxIsalpha(*s)) return(FALSE);
1025     s++;
1026   }
1027   return(TRUE);
1028 }
1029
1030 bool wxString::IsNumber() const
1031 {
1032   const wxChar *s = (const wxChar*) *this;
1033   if (wxStrlen(s))
1034      if ((s[0] == '-') || (s[0] == '+')) s++;
1035   while(*s){
1036     if(!wxIsdigit(*s)) return(FALSE);
1037     s++;
1038   }
1039   return(TRUE);
1040 }
1041
1042 wxString wxString::Strip(stripType w) const
1043 {
1044     wxString s = *this;
1045     if ( w & leading ) s.Trim(FALSE);
1046     if ( w & trailing ) s.Trim(TRUE);
1047     return s;
1048 }
1049
1050 // ---------------------------------------------------------------------------
1051 // case conversion
1052 // ---------------------------------------------------------------------------
1053
1054 wxString& wxString::MakeUpper()
1055 {
1056   if ( !CopyBeforeWrite() ) {
1057     wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") );
1058     return *this;
1059   }
1060
1061   for ( wxChar *p = m_pchData; *p; p++ )
1062     *p = (wxChar)wxToupper(*p);
1063
1064   return *this;
1065 }
1066
1067 wxString& wxString::MakeLower()
1068 {
1069   if ( !CopyBeforeWrite() ) {
1070     wxFAIL_MSG( _T("out of memory in wxString::MakeLower") );
1071     return *this;
1072   }
1073
1074   for ( wxChar *p = m_pchData; *p; p++ )
1075     *p = (wxChar)wxTolower(*p);
1076
1077   return *this;
1078 }
1079
1080 // ---------------------------------------------------------------------------
1081 // trimming and padding
1082 // ---------------------------------------------------------------------------
1083
1084 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1085 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1086 // live with this by checking that the character is a 7 bit one - even if this
1087 // may fail to detect some spaces (I don't know if Unicode doesn't have
1088 // space-like symbols somewhere except in the first 128 chars), it is arguably
1089 // still better than trimming away accented letters
1090 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1091
1092 // trims spaces (in the sense of isspace) from left or right side
1093 wxString& wxString::Trim(bool bFromRight)
1094 {
1095   // first check if we're going to modify the string at all
1096   if ( !IsEmpty() &&
1097        (
1098         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1099         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1100        )
1101      )
1102   {
1103     // ok, there is at least one space to trim
1104     if ( !CopyBeforeWrite() ) {
1105       wxFAIL_MSG( _T("out of memory in wxString::Trim") );
1106       return *this;
1107     }
1108
1109     if ( bFromRight )
1110     {
1111       // find last non-space character
1112       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1113       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1114         psz--;
1115
1116       // truncate at trailing space start
1117       *++psz = wxT('\0');
1118       GetStringData()->nDataLength = psz - m_pchData;
1119     }
1120     else
1121     {
1122       // find first non-space character
1123       const wxChar *psz = m_pchData;
1124       while ( wxSafeIsspace(*psz) )
1125         psz++;
1126
1127       // fix up data and length
1128       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1129       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1130       GetStringData()->nDataLength = nDataLength;
1131     }
1132   }
1133
1134   return *this;
1135 }
1136
1137 // adds nCount characters chPad to the string from either side
1138 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1139 {
1140   wxString s(chPad, nCount);
1141
1142   if ( bFromRight )
1143     *this += s;
1144   else
1145   {
1146     s += *this;
1147     *this = s;
1148   }
1149
1150   return *this;
1151 }
1152
1153 // truncate the string
1154 wxString& wxString::Truncate(size_t uiLen)
1155 {
1156   if ( uiLen < Len() ) {
1157     if ( !CopyBeforeWrite() ) {
1158       wxFAIL_MSG( _T("out of memory in wxString::Truncate") );
1159       return *this;
1160     }
1161
1162     *(m_pchData + uiLen) = wxT('\0');
1163     GetStringData()->nDataLength = uiLen;
1164   }
1165   //else: nothing to do, string is already short enough
1166
1167   return *this;
1168 }
1169
1170 // ---------------------------------------------------------------------------
1171 // finding (return wxNOT_FOUND if not found and index otherwise)
1172 // ---------------------------------------------------------------------------
1173
1174 // find a character
1175 int wxString::Find(wxChar ch, bool bFromEnd) const
1176 {
1177   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1178
1179   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1180 }
1181
1182 // find a sub-string (like strstr)
1183 int wxString::Find(const wxChar *pszSub) const
1184 {
1185   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1186
1187   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1188 }
1189
1190 // ----------------------------------------------------------------------------
1191 // conversion to numbers
1192 // ----------------------------------------------------------------------------
1193
1194 bool wxString::ToLong(long *val, int base) const
1195 {
1196     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1197     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1198
1199     const wxChar *start = c_str();
1200     wxChar *end;
1201     *val = wxStrtol(start, &end, base);
1202
1203     // return TRUE only if scan was stopped by the terminating NUL and if the
1204     // string was not empty to start with
1205     return !*end && (end != start);
1206 }
1207
1208 bool wxString::ToULong(unsigned long *val, int base) const
1209 {
1210     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1211     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1212
1213     const wxChar *start = c_str();
1214     wxChar *end;
1215     *val = wxStrtoul(start, &end, base);
1216
1217     // return TRUE only if scan was stopped by the terminating NUL and if the
1218     // string was not empty to start with
1219     return !*end && (end != start);
1220 }
1221
1222 bool wxString::ToDouble(double *val) const
1223 {
1224     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1225
1226     const wxChar *start = c_str();
1227     wxChar *end;
1228     *val = wxStrtod(start, &end);
1229
1230     // return TRUE only if scan was stopped by the terminating NUL and if the
1231     // string was not empty to start with
1232     return !*end && (end != start);
1233 }
1234
1235 // ---------------------------------------------------------------------------
1236 // formatted output
1237 // ---------------------------------------------------------------------------
1238
1239 /* static */
1240 wxString wxString::Format(const wxChar *pszFormat, ...)
1241 {
1242     va_list argptr;
1243     va_start(argptr, pszFormat);
1244
1245     wxString s;
1246     s.PrintfV(pszFormat, argptr);
1247
1248     va_end(argptr);
1249
1250     return s;
1251 }
1252
1253 /* static */
1254 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1255 {
1256     wxString s;
1257     s.PrintfV(pszFormat, argptr);
1258     return s;
1259 }
1260
1261 int wxString::Printf(const wxChar *pszFormat, ...)
1262 {
1263   va_list argptr;
1264   va_start(argptr, pszFormat);
1265
1266   int iLen = PrintfV(pszFormat, argptr);
1267
1268   va_end(argptr);
1269
1270   return iLen;
1271 }
1272
1273 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1274 {
1275     int size = 1024;
1276     for ( ;; )
1277     {
1278         wxChar *buf = GetWriteBuf(size + 1);
1279         if ( !buf )
1280         {
1281             // out of memory
1282             return -1;
1283         }
1284
1285         int len = wxVsnprintf(buf, size, pszFormat, argptr);
1286
1287         // some implementations of vsnprintf() don't NUL terminate the string
1288         // if there is not enough space for it so always do it manually
1289         buf[size] = _T('\0');
1290
1291         UngetWriteBuf();
1292
1293         if ( len >= 0 )
1294         {
1295             // ok, there was enough space
1296             break;
1297         }
1298
1299         // still not enough, double it again
1300         size *= 2;
1301     }
1302
1303     // we could have overshot
1304     Shrink();
1305
1306     return Len();
1307 }
1308
1309 // ----------------------------------------------------------------------------
1310 // misc other operations
1311 // ----------------------------------------------------------------------------
1312
1313 // returns TRUE if the string matches the pattern which may contain '*' and
1314 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1315 // of them)
1316 bool wxString::Matches(const wxChar *pszMask) const
1317 {
1318     // I disable this code as it doesn't seem to be faster (in fact, it seems
1319     // to be much slower) than the old, hand-written code below and using it
1320     // here requires always linking with libregex even if the user code doesn't
1321     // use it
1322 #if 0 // wxUSE_REGEX
1323     // first translate the shell-like mask into a regex
1324     wxString pattern;
1325     pattern.reserve(wxStrlen(pszMask));
1326
1327     pattern += _T('^');
1328     while ( *pszMask )
1329     {
1330         switch ( *pszMask )
1331         {
1332             case _T('?'):
1333                 pattern += _T('.');
1334                 break;
1335
1336             case _T('*'):
1337                 pattern += _T(".*");
1338                 break;
1339
1340             case _T('^'):
1341             case _T('.'):
1342             case _T('$'):
1343             case _T('('):
1344             case _T(')'):
1345             case _T('|'):
1346             case _T('+'):
1347             case _T('\\'):
1348                 // these characters are special in a RE, quote them
1349                 // (however note that we don't quote '[' and ']' to allow
1350                 // using them for Unix shell like matching)
1351                 pattern += _T('\\');
1352                 // fall through
1353
1354             default:
1355                 pattern += *pszMask;
1356         }
1357
1358         pszMask++;
1359     }
1360     pattern += _T('$');
1361
1362     // and now use it
1363     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1364 #else // !wxUSE_REGEX
1365   // TODO: this is, of course, awfully inefficient...
1366
1367   // the char currently being checked
1368   const wxChar *pszTxt = c_str();
1369
1370   // the last location where '*' matched
1371   const wxChar *pszLastStarInText = NULL;
1372   const wxChar *pszLastStarInMask = NULL;
1373
1374 match:
1375   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1376     switch ( *pszMask ) {
1377       case wxT('?'):
1378         if ( *pszTxt == wxT('\0') )
1379           return FALSE;
1380
1381         // pszTxt and pszMask will be incremented in the loop statement
1382
1383         break;
1384
1385       case wxT('*'):
1386         {
1387           // remember where we started to be able to backtrack later
1388           pszLastStarInText = pszTxt;
1389           pszLastStarInMask = pszMask;
1390
1391           // ignore special chars immediately following this one
1392           // (should this be an error?)
1393           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1394             pszMask++;
1395
1396           // if there is nothing more, match
1397           if ( *pszMask == wxT('\0') )
1398             return TRUE;
1399
1400           // are there any other metacharacters in the mask?
1401           size_t uiLenMask;
1402           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1403
1404           if ( pEndMask != NULL ) {
1405             // we have to match the string between two metachars
1406             uiLenMask = pEndMask - pszMask;
1407           }
1408           else {
1409             // we have to match the remainder of the string
1410             uiLenMask = wxStrlen(pszMask);
1411           }
1412
1413           wxString strToMatch(pszMask, uiLenMask);
1414           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1415           if ( pMatch == NULL )
1416             return FALSE;
1417
1418           // -1 to compensate "++" in the loop
1419           pszTxt = pMatch + uiLenMask - 1;
1420           pszMask += uiLenMask - 1;
1421         }
1422         break;
1423
1424       default:
1425         if ( *pszMask != *pszTxt )
1426           return FALSE;
1427         break;
1428     }
1429   }
1430
1431   // match only if nothing left
1432   if ( *pszTxt == wxT('\0') )
1433     return TRUE;
1434
1435   // if we failed to match, backtrack if we can
1436   if ( pszLastStarInText ) {
1437     pszTxt = pszLastStarInText + 1;
1438     pszMask = pszLastStarInMask;
1439
1440     pszLastStarInText = NULL;
1441
1442     // don't bother resetting pszLastStarInMask, it's unnecessary
1443
1444     goto match;
1445   }
1446
1447   return FALSE;
1448 #endif // wxUSE_REGEX/!wxUSE_REGEX
1449 }
1450
1451 // Count the number of chars
1452 int wxString::Freq(wxChar ch) const
1453 {
1454     int count = 0;
1455     int len = Len();
1456     for (int i = 0; i < len; i++)
1457     {
1458         if (GetChar(i) == ch)
1459             count ++;
1460     }
1461     return count;
1462 }
1463
1464 // convert to upper case, return the copy of the string
1465 wxString wxString::Upper() const
1466 { wxString s(*this); return s.MakeUpper(); }
1467
1468 // convert to lower case, return the copy of the string
1469 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1470
1471 int wxString::sprintf(const wxChar *pszFormat, ...)
1472   {
1473     va_list argptr;
1474     va_start(argptr, pszFormat);
1475     int iLen = PrintfV(pszFormat, argptr);
1476     va_end(argptr);
1477     return iLen;
1478   }
1479
1480 // ---------------------------------------------------------------------------
1481 // standard C++ library string functions
1482 // ---------------------------------------------------------------------------
1483
1484 #ifdef  wxSTD_STRING_COMPATIBILITY
1485
1486 void wxString::resize(size_t nSize, wxChar ch)
1487 {
1488     size_t len = length();
1489
1490     if ( nSize < len )
1491     {
1492         Truncate(nSize);
1493     }
1494     else if ( nSize > len )
1495     {
1496         *this += wxString(ch, nSize - len);
1497     }
1498     //else: we have exactly the specified length, nothing to do
1499 }
1500
1501 void wxString::swap(wxString& str)
1502 {
1503     // this is slightly less efficient than fiddling with m_pchData directly,
1504     // but it is still quite efficient as we don't copy the string here because
1505     // ref count always stays positive
1506     wxString tmp = str;
1507     str = *this;
1508     *this = tmp;
1509 }
1510
1511 wxString& wxString::insert(size_t nPos, const wxString& str)
1512 {
1513   wxASSERT( str.GetStringData()->IsValid() );
1514   wxASSERT( nPos <= Len() );
1515
1516   if ( !str.IsEmpty() ) {
1517     wxString strTmp;
1518     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1519     wxStrncpy(pc, c_str(), nPos);
1520     wxStrcpy(pc + nPos, str);
1521     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1522     strTmp.UngetWriteBuf();
1523     *this = strTmp;
1524   }
1525
1526   return *this;
1527 }
1528
1529 size_t wxString::find(const wxString& str, size_t nStart) const
1530 {
1531   wxASSERT( str.GetStringData()->IsValid() );
1532   wxASSERT( nStart <= Len() );
1533
1534   const wxChar *p = wxStrstr(c_str() + nStart, str);
1535
1536   return p == NULL ? npos : p - c_str();
1537 }
1538
1539 // VC++ 1.5 can't cope with the default argument in the header.
1540 #if !defined(__VISUALC__) || defined(__WIN32__)
1541 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1542 {
1543   return find(wxString(sz, n), nStart);
1544 }
1545 #endif // VC++ 1.5
1546
1547 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1548 #if !defined(__BORLANDC__)
1549 size_t wxString::find(wxChar ch, size_t nStart) const
1550 {
1551   wxASSERT( nStart <= Len() );
1552
1553   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1554
1555   return p == NULL ? npos : p - c_str();
1556 }
1557 #endif
1558
1559 size_t wxString::rfind(const wxString& str, size_t nStart) const
1560 {
1561   wxASSERT( str.GetStringData()->IsValid() );
1562   wxASSERT( nStart == npos || nStart <= Len() );
1563
1564   // TODO could be made much quicker than that
1565   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1566   while ( p >= c_str() + str.Len() ) {
1567     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1568       return p - str.Len() - c_str();
1569     p--;
1570   }
1571
1572   return npos;
1573 }
1574
1575 // VC++ 1.5 can't cope with the default argument in the header.
1576 #if !defined(__VISUALC__) || defined(__WIN32__)
1577 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1578 {
1579     return rfind(wxString(sz, n == npos ? wxSTRING_MAXLEN : n), nStart);
1580 }
1581
1582 size_t wxString::rfind(wxChar ch, size_t nStart) const
1583 {
1584     if ( nStart == npos )
1585     {
1586         nStart = Len();
1587     }
1588     else
1589     {
1590         wxASSERT( nStart <= Len() );
1591     }
1592
1593     const wxChar *p = wxStrrchr(c_str(), ch);
1594
1595     if ( p == NULL )
1596         return npos;
1597
1598     size_t result = p - c_str();
1599     return ( result > nStart ) ? npos : result;
1600 }
1601 #endif // VC++ 1.5
1602
1603 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1604 {
1605     const wxChar *start = c_str() + nStart;
1606     const wxChar *firstOf = wxStrpbrk(start, sz);
1607     if ( firstOf )
1608         return firstOf - c_str();
1609     else
1610         return npos;
1611 }
1612
1613 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1614 {
1615     if ( nStart == npos )
1616     {
1617         nStart = Len();
1618     }
1619     else
1620     {
1621         wxASSERT( nStart <= Len() );
1622     }
1623
1624     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1625     {
1626         if ( wxStrchr(sz, *p) )
1627             return p - c_str();
1628     }
1629
1630     return npos;
1631 }
1632
1633 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1634 {
1635     if ( nStart == npos )
1636     {
1637         nStart = Len();
1638     }
1639     else
1640     {
1641         wxASSERT( nStart <= Len() );
1642     }
1643
1644     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1645     if ( nAccept >= length() - nStart )
1646         return npos;
1647     else
1648         return nAccept;
1649 }
1650
1651 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1652 {
1653     wxASSERT( nStart <= Len() );
1654
1655     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1656     {
1657         if ( *p != ch )
1658             return p - c_str();
1659     }
1660
1661     return npos;
1662 }
1663
1664 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1665 {
1666     if ( nStart == npos )
1667     {
1668         nStart = Len();
1669     }
1670     else
1671     {
1672         wxASSERT( nStart <= Len() );
1673     }
1674
1675     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1676     {
1677         if ( !wxStrchr(sz, *p) )
1678             return p - c_str();
1679     }
1680
1681     return npos;
1682 }
1683
1684 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1685 {
1686     if ( nStart == npos )
1687     {
1688         nStart = Len();
1689     }
1690     else
1691     {
1692         wxASSERT( nStart <= Len() );
1693     }
1694
1695     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1696     {
1697         if ( *p != ch )
1698             return p - c_str();
1699     }
1700
1701     return npos;
1702 }
1703
1704 wxString& wxString::erase(size_t nStart, size_t nLen)
1705 {
1706   wxString strTmp(c_str(), nStart);
1707   if ( nLen != npos ) {
1708     wxASSERT( nStart + nLen <= Len() );
1709
1710     strTmp.append(c_str() + nStart + nLen);
1711   }
1712
1713   *this = strTmp;
1714   return *this;
1715 }
1716
1717 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1718 {
1719   wxASSERT_MSG( nStart + nLen <= Len(),
1720                 _T("index out of bounds in wxString::replace") );
1721
1722   wxString strTmp;
1723   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1724
1725   if ( nStart != 0 )
1726     strTmp.append(c_str(), nStart);
1727   strTmp << sz << c_str() + nStart + nLen;
1728
1729   *this = strTmp;
1730   return *this;
1731 }
1732
1733 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1734 {
1735   return replace(nStart, nLen, wxString(ch, nCount));
1736 }
1737
1738 wxString& wxString::replace(size_t nStart, size_t nLen,
1739                             const wxString& str, size_t nStart2, size_t nLen2)
1740 {
1741   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1742 }
1743
1744 wxString& wxString::replace(size_t nStart, size_t nLen,
1745                         const wxChar* sz, size_t nCount)
1746 {
1747   return replace(nStart, nLen, wxString(sz, nCount));
1748 }
1749
1750 #endif  //std::string compatibility
1751
1752 // ============================================================================
1753 // ArrayString
1754 // ============================================================================
1755
1756 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1757 #define   ARRAY_MAXSIZE_INCREMENT       4096
1758
1759 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1760 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1761 #endif
1762
1763 #define   STRING(p)   ((wxString *)(&(p)))
1764
1765 // ctor
1766 void wxArrayString::Init(bool autoSort)
1767 {
1768   m_nSize  =
1769   m_nCount = 0;
1770   m_pItems = (wxChar **) NULL;
1771   m_autoSort = autoSort;
1772 }
1773
1774 // copy ctor
1775 wxArrayString::wxArrayString(const wxArrayString& src)
1776 {
1777   Init(src.m_autoSort);
1778
1779   *this = src;
1780 }
1781
1782 // assignment operator
1783 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1784 {
1785   if ( m_nSize > 0 )
1786     Clear();
1787
1788   Copy(src);
1789
1790   m_autoSort = src.m_autoSort;
1791
1792   return *this;
1793 }
1794
1795 void wxArrayString::Copy(const wxArrayString& src)
1796 {
1797   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1798     Alloc(src.m_nCount);
1799
1800   for ( size_t n = 0; n < src.m_nCount; n++ )
1801     Add(src[n]);
1802 }
1803
1804 // grow the array
1805 void wxArrayString::Grow(size_t nIncrement)
1806 {
1807   // only do it if no more place
1808   if ( (m_nSize - m_nCount) < nIncrement ) {
1809     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1810     // be never resized!
1811     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1812       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1813     #endif
1814
1815     if ( m_nSize == 0 ) {
1816       // was empty, alloc some memory
1817       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1818       if (m_nSize < nIncrement)
1819           m_nSize = nIncrement;
1820       m_pItems = new wxChar *[m_nSize];
1821     }
1822     else {
1823       // otherwise when it's called for the first time, nIncrement would be 0
1824       // and the array would never be expanded
1825       // add 50% but not too much
1826       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1827                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1828       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1829         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1830       if ( nIncrement < ndefIncrement )
1831         nIncrement = ndefIncrement;
1832       m_nSize += nIncrement;
1833       wxChar **pNew = new wxChar *[m_nSize];
1834
1835       // copy data to new location
1836       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1837
1838       // delete old memory (but do not release the strings!)
1839       wxDELETEA(m_pItems);
1840
1841       m_pItems = pNew;
1842     }
1843   }
1844 }
1845
1846 void wxArrayString::Free()
1847 {
1848   for ( size_t n = 0; n < m_nCount; n++ ) {
1849     STRING(m_pItems[n])->GetStringData()->Unlock();
1850   }
1851 }
1852
1853 // deletes all the strings from the list
1854 void wxArrayString::Empty()
1855 {
1856   Free();
1857
1858   m_nCount = 0;
1859 }
1860
1861 // as Empty, but also frees memory
1862 void wxArrayString::Clear()
1863 {
1864   Free();
1865
1866   m_nSize  =
1867   m_nCount = 0;
1868
1869   wxDELETEA(m_pItems);
1870 }
1871
1872 // dtor
1873 wxArrayString::~wxArrayString()
1874 {
1875   Free();
1876
1877   wxDELETEA(m_pItems);
1878 }
1879
1880 // pre-allocates memory (frees the previous data!)
1881 void wxArrayString::Alloc(size_t nSize)
1882 {
1883   // only if old buffer was not big enough
1884   if ( nSize > m_nSize ) {
1885     Free();
1886     wxDELETEA(m_pItems);
1887     m_pItems = new wxChar *[nSize];
1888     m_nSize  = nSize;
1889   }
1890
1891   m_nCount = 0;
1892 }
1893
1894 // minimizes the memory usage by freeing unused memory
1895 void wxArrayString::Shrink()
1896 {
1897   // only do it if we have some memory to free
1898   if( m_nCount < m_nSize ) {
1899     // allocates exactly as much memory as we need
1900     wxChar **pNew = new wxChar *[m_nCount];
1901
1902     // copy data to new location
1903     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1904     delete [] m_pItems;
1905     m_pItems = pNew;
1906   }
1907 }
1908
1909 // return a wxString[] as required for some control ctors.
1910 wxString* wxArrayString::GetStringArray() const
1911 {
1912     wxString *array = 0;
1913
1914     if( m_nCount > 0 )
1915     {
1916         array = new wxString[m_nCount];
1917         for( size_t i = 0; i < m_nCount; i++ )
1918             array[i] = m_pItems[i];
1919     }
1920
1921     return array;
1922 }
1923
1924 // searches the array for an item (forward or backwards)
1925 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1926 {
1927   if ( m_autoSort ) {
1928     // use binary search in the sorted array
1929     wxASSERT_MSG( bCase && !bFromEnd,
1930                   wxT("search parameters ignored for auto sorted array") );
1931
1932     size_t i,
1933            lo = 0,
1934            hi = m_nCount;
1935     int res;
1936     while ( lo < hi ) {
1937       i = (lo + hi)/2;
1938
1939       res = wxStrcmp(sz, m_pItems[i]);
1940       if ( res < 0 )
1941         hi = i;
1942       else if ( res > 0 )
1943         lo = i + 1;
1944       else
1945         return i;
1946     }
1947
1948     return wxNOT_FOUND;
1949   }
1950   else {
1951     // use linear search in unsorted array
1952     if ( bFromEnd ) {
1953       if ( m_nCount > 0 ) {
1954         size_t ui = m_nCount;
1955         do {
1956           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1957             return ui;
1958         }
1959         while ( ui != 0 );
1960       }
1961     }
1962     else {
1963       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1964         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1965           return ui;
1966       }
1967     }
1968   }
1969
1970   return wxNOT_FOUND;
1971 }
1972
1973 // add item at the end
1974 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
1975 {
1976   if ( m_autoSort ) {
1977     // insert the string at the correct position to keep the array sorted
1978     size_t i,
1979            lo = 0,
1980            hi = m_nCount;
1981     int res;
1982     while ( lo < hi ) {
1983       i = (lo + hi)/2;
1984
1985       res = wxStrcmp(str, m_pItems[i]);
1986       if ( res < 0 )
1987         hi = i;
1988       else if ( res > 0 )
1989         lo = i + 1;
1990       else {
1991         lo = hi = i;
1992         break;
1993       }
1994     }
1995
1996     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1997
1998     Insert(str, lo, nInsert);
1999
2000     return (size_t)lo;
2001   }
2002   else {
2003     wxASSERT( str.GetStringData()->IsValid() );
2004
2005     Grow(nInsert);
2006
2007     for (size_t i = 0; i < nInsert; i++)
2008     {
2009         // the string data must not be deleted!
2010         str.GetStringData()->Lock();
2011
2012         // just append
2013         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2014     }
2015     size_t ret = m_nCount;
2016     m_nCount += nInsert;
2017     return ret;
2018   }
2019 }
2020
2021 // add item at the given position
2022 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2023 {
2024   wxASSERT( str.GetStringData()->IsValid() );
2025
2026   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2027   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2028                wxT("array size overflow in wxArrayString::Insert") );
2029
2030   Grow(nInsert);
2031
2032   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2033           (m_nCount - nIndex)*sizeof(wxChar *));
2034
2035   for (size_t i = 0; i < nInsert; i++)
2036   {
2037       str.GetStringData()->Lock();
2038       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2039   }
2040   m_nCount += nInsert;
2041 }
2042
2043 // expand the array
2044 void wxArrayString::SetCount(size_t count)
2045 {
2046     Alloc(count);
2047
2048     wxString s;
2049     while ( m_nCount < count )
2050         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2051 }
2052
2053 // removes item from array (by index)
2054 void wxArrayString::Remove(size_t nIndex, size_t nRemove)
2055 {
2056   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2057   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2058                wxT("removing too many elements in wxArrayString::Remove") );
2059
2060   // release our lock
2061   for (size_t i = 0; i < nRemove; i++)
2062       Item(nIndex + i).GetStringData()->Unlock();
2063
2064   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2065           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2066   m_nCount -= nRemove;
2067 }
2068
2069 // removes item from array (by value)
2070 void wxArrayString::Remove(const wxChar *sz)
2071 {
2072   int iIndex = Index(sz);
2073
2074   wxCHECK_RET( iIndex != wxNOT_FOUND,
2075                wxT("removing inexistent element in wxArrayString::Remove") );
2076
2077   Remove(iIndex);
2078 }
2079
2080 // ----------------------------------------------------------------------------
2081 // sorting
2082 // ----------------------------------------------------------------------------
2083
2084 // we can only sort one array at a time with the quick-sort based
2085 // implementation
2086 #if wxUSE_THREADS
2087   // need a critical section to protect access to gs_compareFunction and
2088   // gs_sortAscending variables
2089   static wxCriticalSection *gs_critsectStringSort = NULL;
2090
2091   // call this before the value of the global sort vars is changed/after
2092   // you're finished with them
2093   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2094                            gs_critsectStringSort = new wxCriticalSection;     \
2095                            gs_critsectStringSort->Enter()
2096   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2097                            delete gs_critsectStringSort;                      \
2098                            gs_critsectStringSort = NULL
2099 #else // !threads
2100   #define START_SORT()
2101   #define END_SORT()
2102 #endif // wxUSE_THREADS
2103
2104 // function to use for string comparaison
2105 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2106
2107 // if we don't use the compare function, this flag tells us if we sort the
2108 // array in ascending or descending order
2109 static bool gs_sortAscending = TRUE;
2110
2111 // function which is called by quick sort
2112 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2113 wxStringCompareFunction(const void *first, const void *second)
2114 {
2115   wxString *strFirst = (wxString *)first;
2116   wxString *strSecond = (wxString *)second;
2117
2118   if ( gs_compareFunction ) {
2119     return gs_compareFunction(*strFirst, *strSecond);
2120   }
2121   else {
2122     // maybe we should use wxStrcoll
2123     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2124
2125     return gs_sortAscending ? result : -result;
2126   }
2127 }
2128
2129 // sort array elements using passed comparaison function
2130 void wxArrayString::Sort(CompareFunction compareFunction)
2131 {
2132   START_SORT();
2133
2134   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2135   gs_compareFunction = compareFunction;
2136
2137   DoSort();
2138
2139   // reset it to NULL so that Sort(bool) will work the next time
2140   gs_compareFunction = NULL;
2141
2142   END_SORT();
2143 }
2144
2145 void wxArrayString::Sort(bool reverseOrder)
2146 {
2147   START_SORT();
2148
2149   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2150   gs_sortAscending = !reverseOrder;
2151
2152   DoSort();
2153
2154   END_SORT();
2155 }
2156
2157 void wxArrayString::DoSort()
2158 {
2159   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2160
2161   // just sort the pointers using qsort() - of course it only works because
2162   // wxString() *is* a pointer to its data
2163   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2164 }
2165
2166 bool wxArrayString::operator==(const wxArrayString& a) const
2167 {
2168     if ( m_nCount != a.m_nCount )
2169         return FALSE;
2170
2171     for ( size_t n = 0; n < m_nCount; n++ )
2172     {
2173         if ( Item(n) != a[n] )
2174             return FALSE;
2175     }
2176
2177     return TRUE;
2178 }
2179