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 wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 962 {
 963   size_t uiCount = 0;   // count of replacements made
 964
 965   size_t uiOldLen = wxStrlen(szOld);
 966
 967   wxString strTemp;
 968   const wxChar *pCurrent = m_pchData;
 969   const wxChar *pSubstr;
 970   while ( *pCurrent != wxT('\0') ) {
 971     pSubstr = wxStrstr(pCurrent, szOld);
 972     if ( pSubstr == NULL ) {
 973       // strTemp is unused if no replacements were made, so avoid the copy
 974       if ( uiCount == 0 )
 975         return 0;
 976
 977       strTemp += pCurrent;    // copy the rest
 978       break;                  // exit the loop
 979     }
 980     else {
 981       // take chars before match
 982       if ( !strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent) ) {
 983         wxFAIL_MSG( _T("out of memory in wxString::Replace") );
 984         return 0;
 985       }
 986       strTemp += szNew;
 987       pCurrent = pSubstr + uiOldLen;  // restart after match
 988
 989       uiCount++;
 990
 991       // stop now?
 992       if ( !bReplaceAll ) {
 993         strTemp += pCurrent;    // copy the rest
 994         break;                  // exit the loop
 995       }
 996     }
 997   }
 998
 999   // only done if there were replacements, otherwise would have returned above
1000   *this = strTemp;
1001
1002   return uiCount;
1003 }
1004
1005 bool wxString::IsAscii() const
1006 {
1007   const wxChar *s = (const wxChar*) *this;
1008   while(*s){
1009     if(!isascii(*s)) return(FALSE);
1010     s++;
1011   }
1012   return(TRUE);
1013 }
1014
1015 bool wxString::IsWord() const
1016 {
1017   const wxChar *s = (const wxChar*) *this;
1018   while(*s){
1019     if(!wxIsalpha(*s)) return(FALSE);
1020     s++;
1021   }
1022   return(TRUE);
1023 }
1024
1025 bool wxString::IsNumber() const
1026 {
1027   const wxChar *s = (const wxChar*) *this;
1028   if (wxStrlen(s))
1029      if ((s[0] == '-') || (s[0] == '+')) s++;
1030   while(*s){
1031     if(!wxIsdigit(*s)) return(FALSE);
1032     s++;
1033   }
1034   return(TRUE);
1035 }
1036
1037 wxString wxString::Strip(stripType w) const
1038 {
1039     wxString s = *this;
1040     if ( w & leading ) s.Trim(FALSE);
1041     if ( w & trailing ) s.Trim(TRUE);
1042     return s;
1043 }
1044
1045 // ---------------------------------------------------------------------------
1046 // case conversion
1047 // ---------------------------------------------------------------------------
1048
1049 wxString& wxString::MakeUpper()
1050 {
1051   if ( !CopyBeforeWrite() ) {
1052     wxFAIL_MSG( _T("out of memory in wxString::MakeUpper") );
1053     return *this;
1054   }
1055
1056   for ( wxChar *p = m_pchData; *p; p++ )
1057     *p = (wxChar)wxToupper(*p);
1058
1059   return *this;
1060 }
1061
1062 wxString& wxString::MakeLower()
1063 {
1064   if ( !CopyBeforeWrite() ) {
1065     wxFAIL_MSG( _T("out of memory in wxString::MakeLower") );
1066     return *this;
1067   }
1068
1069   for ( wxChar *p = m_pchData; *p; p++ )
1070     *p = (wxChar)wxTolower(*p);
1071
1072   return *this;
1073 }
1074
1075 // ---------------------------------------------------------------------------
1076 // trimming and padding
1077 // ---------------------------------------------------------------------------
1078
1079 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
1080 // isspace('ê') in the C locale which seems to be broken to me, but we have to
1081 // live with this by checking that the character is a 7 bit one - even if this
1082 // may fail to detect some spaces (I don't know if Unicode doesn't have
1083 // space-like symbols somewhere except in the first 128 chars), it is arguably
1084 // still better than trimming away accented letters
1085 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1086
1087 // trims spaces (in the sense of isspace) from left or right side
1088 wxString& wxString::Trim(bool bFromRight)
1089 {
1090   // first check if we're going to modify the string at all
1091   if ( !IsEmpty() &&
1092        (
1093         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1094         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1095        )
1096      )
1097   {
1098     // ok, there is at least one space to trim
1099     if ( !CopyBeforeWrite() ) {
1100       wxFAIL_MSG( _T("out of memory in wxString::Trim") );
1101       return *this;
1102     }
1103
1104     if ( bFromRight )
1105     {
1106       // find last non-space character
1107       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1108       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1109         psz--;
1110
1111       // truncate at trailing space start
1112       *++psz = wxT('\0');
1113       GetStringData()->nDataLength = psz - m_pchData;
1114     }
1115     else
1116     {
1117       // find first non-space character
1118       const wxChar *psz = m_pchData;
1119       while ( wxSafeIsspace(*psz) )
1120         psz++;
1121
1122       // fix up data and length
1123       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1124       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1125       GetStringData()->nDataLength = nDataLength;
1126     }
1127   }
1128
1129   return *this;
1130 }
1131
1132 // adds nCount characters chPad to the string from either side
1133 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1134 {
1135   wxString s(chPad, nCount);
1136
1137   if ( bFromRight )
1138     *this += s;
1139   else
1140   {
1141     s += *this;
1142     *this = s;
1143   }
1144
1145   return *this;
1146 }
1147
1148 // truncate the string
1149 wxString& wxString::Truncate(size_t uiLen)
1150 {
1151   if ( uiLen < Len() ) {
1152     if ( !CopyBeforeWrite() ) {
1153       wxFAIL_MSG( _T("out of memory in wxString::Truncate") );
1154       return *this;
1155     }
1156
1157     *(m_pchData + uiLen) = wxT('\0');
1158     GetStringData()->nDataLength = uiLen;
1159   }
1160   //else: nothing to do, string is already short enough
1161
1162   return *this;
1163 }
1164
1165 // ---------------------------------------------------------------------------
1166 // finding (return wxNOT_FOUND if not found and index otherwise)
1167 // ---------------------------------------------------------------------------
1168
1169 // find a character
1170 int wxString::Find(wxChar ch, bool bFromEnd) const
1171 {
1172   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1173
1174   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1175 }
1176
1177 // find a sub-string (like strstr)
1178 int wxString::Find(const wxChar *pszSub) const
1179 {
1180   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1181
1182   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1183 }
1184
1185 // ----------------------------------------------------------------------------
1186 // conversion to numbers
1187 // ----------------------------------------------------------------------------
1188
1189 bool wxString::ToLong(long *val, int base) const
1190 {
1191     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1192     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1193
1194     const wxChar *start = c_str();
1195     wxChar *end;
1196     *val = wxStrtol(start, &end, base);
1197
1198     // return TRUE only if scan was stopped by the terminating NUL and if the
1199     // string was not empty to start with
1200     return !*end && (end != start);
1201 }
1202
1203 bool wxString::ToULong(unsigned long *val, int base) const
1204 {
1205     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1206     wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") );
1207
1208     const wxChar *start = c_str();
1209     wxChar *end;
1210     *val = wxStrtoul(start, &end, base);
1211
1212     // return TRUE only if scan was stopped by the terminating NUL and if the
1213     // string was not empty to start with
1214     return !*end && (end != start);
1215 }
1216
1217 bool wxString::ToDouble(double *val) const
1218 {
1219     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1220
1221     const wxChar *start = c_str();
1222     wxChar *end;
1223     *val = wxStrtod(start, &end);
1224
1225     // return TRUE only if scan was stopped by the terminating NUL and if the
1226     // string was not empty to start with
1227     return !*end && (end != start);
1228 }
1229
1230 // ---------------------------------------------------------------------------
1231 // formatted output
1232 // ---------------------------------------------------------------------------
1233
1234 /* static */
1235 wxString wxString::Format(const wxChar *pszFormat, ...)
1236 {
1237     va_list argptr;
1238     va_start(argptr, pszFormat);
1239
1240     wxString s;
1241     s.PrintfV(pszFormat, argptr);
1242
1243     va_end(argptr);
1244
1245     return s;
1246 }
1247
1248 /* static */
1249 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1250 {
1251     wxString s;
1252     s.PrintfV(pszFormat, argptr);
1253     return s;
1254 }
1255
1256 int wxString::Printf(const wxChar *pszFormat, ...)
1257 {
1258   va_list argptr;
1259   va_start(argptr, pszFormat);
1260
1261   int iLen = PrintfV(pszFormat, argptr);
1262
1263   va_end(argptr);
1264
1265   return iLen;
1266 }
1267
1268 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1269 {
1270     int size = 1024;
1271     for ( ;; )
1272     {
1273         wxChar *buf = GetWriteBuf(size + 1);
1274         if ( !buf )
1275         {
1276             // out of memory
1277             return -1;
1278         }
1279
1280         int len = wxVsnprintf(buf, size, pszFormat, argptr);
1281
1282         // some implementations of vsnprintf() don't NUL terminate the string
1283         // if there is not enough space for it so always do it manually
1284         buf[size] = _T('\0');
1285
1286         UngetWriteBuf();
1287
1288         if ( len >= 0 )
1289         {
1290             // ok, there was enough space
1291             break;
1292         }
1293
1294         // still not enough, double it again
1295         size *= 2;
1296     }
1297
1298     // we could have overshot
1299     Shrink();
1300
1301     return Len();
1302 }
1303
1304 // ----------------------------------------------------------------------------
1305 // misc other operations
1306 // ----------------------------------------------------------------------------
1307
1308 // returns TRUE if the string matches the pattern which may contain '*' and
1309 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1310 // of them)
1311 bool wxString::Matches(const wxChar *pszMask) const
1312 {
1313     // I disable this code as it doesn't seem to be faster (in fact, it seems
1314     // to be much slower) than the old, hand-written code below and using it
1315     // here requires always linking with libregex even if the user code doesn't
1316     // use it
1317 #if 0 // wxUSE_REGEX
1318     // first translate the shell-like mask into a regex
1319     wxString pattern;
1320     pattern.reserve(wxStrlen(pszMask));
1321
1322     pattern += _T('^');
1323     while ( *pszMask )
1324     {
1325         switch ( *pszMask )
1326         {
1327             case _T('?'):
1328                 pattern += _T('.');
1329                 break;
1330
1331             case _T('*'):
1332                 pattern += _T(".*");
1333                 break;
1334
1335             case _T('^'):
1336             case _T('.'):
1337             case _T('$'):
1338             case _T('('):
1339             case _T(')'):
1340             case _T('|'):
1341             case _T('+'):
1342             case _T('\\'):
1343                 // these characters are special in a RE, quote them
1344                 // (however note that we don't quote '[' and ']' to allow
1345                 // using them for Unix shell like matching)
1346                 pattern += _T('\\');
1347                 // fall through
1348
1349             default:
1350                 pattern += *pszMask;
1351         }
1352
1353         pszMask++;
1354     }
1355     pattern += _T('$');
1356
1357     // and now use it
1358     return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1359 #else // !wxUSE_REGEX
1360   // TODO: this is, of course, awfully inefficient...
1361
1362   // the char currently being checked
1363   const wxChar *pszTxt = c_str();
1364
1365   // the last location where '*' matched
1366   const wxChar *pszLastStarInText = NULL;
1367   const wxChar *pszLastStarInMask = NULL;
1368
1369 match:
1370   for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1371     switch ( *pszMask ) {
1372       case wxT('?'):
1373         if ( *pszTxt == wxT('\0') )
1374           return FALSE;
1375
1376         // pszTxt and pszMask will be incremented in the loop statement
1377
1378         break;
1379
1380       case wxT('*'):
1381         {
1382           // remember where we started to be able to backtrack later
1383           pszLastStarInText = pszTxt;
1384           pszLastStarInMask = pszMask;
1385
1386           // ignore special chars immediately following this one
1387           // (should this be an error?)
1388           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1389             pszMask++;
1390
1391           // if there is nothing more, match
1392           if ( *pszMask == wxT('\0') )
1393             return TRUE;
1394
1395           // are there any other metacharacters in the mask?
1396           size_t uiLenMask;
1397           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1398
1399           if ( pEndMask != NULL ) {
1400             // we have to match the string between two metachars
1401             uiLenMask = pEndMask - pszMask;
1402           }
1403           else {
1404             // we have to match the remainder of the string
1405             uiLenMask = wxStrlen(pszMask);
1406           }
1407
1408           wxString strToMatch(pszMask, uiLenMask);
1409           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1410           if ( pMatch == NULL )
1411             return FALSE;
1412
1413           // -1 to compensate "++" in the loop
1414           pszTxt = pMatch + uiLenMask - 1;
1415           pszMask += uiLenMask - 1;
1416         }
1417         break;
1418
1419       default:
1420         if ( *pszMask != *pszTxt )
1421           return FALSE;
1422         break;
1423     }
1424   }
1425
1426   // match only if nothing left
1427   if ( *pszTxt == wxT('\0') )
1428     return TRUE;
1429
1430   // if we failed to match, backtrack if we can
1431   if ( pszLastStarInText ) {
1432     pszTxt = pszLastStarInText + 1;
1433     pszMask = pszLastStarInMask;
1434
1435     pszLastStarInText = NULL;
1436
1437     // don't bother resetting pszLastStarInMask, it's unnecessary
1438
1439     goto match;
1440   }
1441
1442   return FALSE;
1443 #endif // wxUSE_REGEX/!wxUSE_REGEX
1444 }
1445
1446 // Count the number of chars
1447 int wxString::Freq(wxChar ch) const
1448 {
1449     int count = 0;
1450     int len = Len();
1451     for (int i = 0; i < len; i++)
1452     {
1453         if (GetChar(i) == ch)
1454             count ++;
1455     }
1456     return count;
1457 }
1458
1459 // convert to upper case, return the copy of the string
1460 wxString wxString::Upper() const
1461 { wxString s(*this); return s.MakeUpper(); }
1462
1463 // convert to lower case, return the copy of the string
1464 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1465
1466 int wxString::sprintf(const wxChar *pszFormat, ...)
1467   {
1468     va_list argptr;
1469     va_start(argptr, pszFormat);
1470     int iLen = PrintfV(pszFormat, argptr);
1471     va_end(argptr);
1472     return iLen;
1473   }
1474
1475 // ---------------------------------------------------------------------------
1476 // standard C++ library string functions
1477 // ---------------------------------------------------------------------------
1478
1479 #ifdef  wxSTD_STRING_COMPATIBILITY
1480
1481 void wxString::resize(size_t nSize, wxChar ch)
1482 {
1483     size_t len = length();
1484
1485     if ( nSize < len )
1486     {
1487         Truncate(nSize);
1488     }
1489     else if ( nSize > len )
1490     {
1491         *this += wxString(ch, nSize - len);
1492     }
1493     //else: we have exactly the specified length, nothing to do
1494 }
1495
1496 void wxString::swap(wxString& str)
1497 {
1498     // this is slightly less efficient than fiddling with m_pchData directly,
1499     // but it is still quite efficient as we don't copy the string here because
1500     // ref count always stays positive
1501     wxString tmp = str;
1502     str = *this;
1503     *this = tmp;
1504 }
1505
1506 wxString& wxString::insert(size_t nPos, const wxString& str)
1507 {
1508   wxASSERT( str.GetStringData()->IsValid() );
1509   wxASSERT( nPos <= Len() );
1510
1511   if ( !str.IsEmpty() ) {
1512     wxString strTmp;
1513     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1514     wxStrncpy(pc, c_str(), nPos);
1515     wxStrcpy(pc + nPos, str);
1516     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1517     strTmp.UngetWriteBuf();
1518     *this = strTmp;
1519   }
1520
1521   return *this;
1522 }
1523
1524 size_t wxString::find(const wxString& str, size_t nStart) const
1525 {
1526   wxASSERT( str.GetStringData()->IsValid() );
1527   wxASSERT( nStart <= Len() );
1528
1529   const wxChar *p = wxStrstr(c_str() + nStart, str);
1530
1531   return p == NULL ? npos : p - c_str();
1532 }
1533
1534 // VC++ 1.5 can't cope with the default argument in the header.
1535 #if !defined(__VISUALC__) || defined(__WIN32__)
1536 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1537 {
1538   return find(wxString(sz, n), nStart);
1539 }
1540 #endif // VC++ 1.5
1541
1542 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1543 #if !defined(__BORLANDC__)
1544 size_t wxString::find(wxChar ch, size_t nStart) const
1545 {
1546   wxASSERT( nStart <= Len() );
1547
1548   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1549
1550   return p == NULL ? npos : p - c_str();
1551 }
1552 #endif
1553
1554 size_t wxString::rfind(const wxString& str, size_t nStart) const
1555 {
1556   wxASSERT( str.GetStringData()->IsValid() );
1557   wxASSERT( nStart == npos || nStart <= Len() );
1558
1559   // TODO could be made much quicker than that
1560   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1561   while ( p >= c_str() + str.Len() ) {
1562     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1563       return p - str.Len() - c_str();
1564     p--;
1565   }
1566
1567   return npos;
1568 }
1569
1570 // VC++ 1.5 can't cope with the default argument in the header.
1571 #if !defined(__VISUALC__) || defined(__WIN32__)
1572 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1573 {
1574     return rfind(wxString(sz, n == npos ? wxSTRING_MAXLEN : n), nStart);
1575 }
1576
1577 size_t wxString::rfind(wxChar ch, size_t nStart) const
1578 {
1579     if ( nStart == npos )
1580     {
1581         nStart = Len();
1582     }
1583     else
1584     {
1585         wxASSERT( nStart <= Len() );
1586     }
1587
1588     const wxChar *p = wxStrrchr(c_str(), ch);
1589
1590     if ( p == NULL )
1591         return npos;
1592
1593     size_t result = p - c_str();
1594     return ( result > nStart ) ? npos : result;
1595 }
1596 #endif // VC++ 1.5
1597
1598 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1599 {
1600     const wxChar *start = c_str() + nStart;
1601     const wxChar *firstOf = wxStrpbrk(start, sz);
1602     if ( firstOf )
1603         return firstOf - c_str();
1604     else
1605         return npos;
1606 }
1607
1608 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1609 {
1610     if ( nStart == npos )
1611     {
1612         nStart = Len();
1613     }
1614     else
1615     {
1616         wxASSERT( nStart <= Len() );
1617     }
1618
1619     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1620     {
1621         if ( wxStrchr(sz, *p) )
1622             return p - c_str();
1623     }
1624
1625     return npos;
1626 }
1627
1628 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1629 {
1630     if ( nStart == npos )
1631     {
1632         nStart = Len();
1633     }
1634     else
1635     {
1636         wxASSERT( nStart <= Len() );
1637     }
1638
1639     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1640     if ( nAccept >= length() - nStart )
1641         return npos;
1642     else
1643         return nAccept;
1644 }
1645
1646 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1647 {
1648     wxASSERT( nStart <= Len() );
1649
1650     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1651     {
1652         if ( *p != ch )
1653             return p - c_str();
1654     }
1655
1656     return npos;
1657 }
1658
1659 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1660 {
1661     if ( nStart == npos )
1662     {
1663         nStart = Len();
1664     }
1665     else
1666     {
1667         wxASSERT( nStart <= Len() );
1668     }
1669
1670     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1671     {
1672         if ( !wxStrchr(sz, *p) )
1673             return p - c_str();
1674     }
1675
1676     return npos;
1677 }
1678
1679 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1680 {
1681     if ( nStart == npos )
1682     {
1683         nStart = Len();
1684     }
1685     else
1686     {
1687         wxASSERT( nStart <= Len() );
1688     }
1689
1690     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1691     {
1692         if ( *p != ch )
1693             return p - c_str();
1694     }
1695
1696     return npos;
1697 }
1698
1699 wxString& wxString::erase(size_t nStart, size_t nLen)
1700 {
1701   wxString strTmp(c_str(), nStart);
1702   if ( nLen != npos ) {
1703     wxASSERT( nStart + nLen <= Len() );
1704
1705     strTmp.append(c_str() + nStart + nLen);
1706   }
1707
1708   *this = strTmp;
1709   return *this;
1710 }
1711
1712 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1713 {
1714   wxASSERT_MSG( nStart + nLen <= Len(),
1715                 _T("index out of bounds in wxString::replace") );
1716
1717   wxString strTmp;
1718   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1719
1720   if ( nStart != 0 )
1721     strTmp.append(c_str(), nStart);
1722   strTmp << sz << c_str() + nStart + nLen;
1723
1724   *this = strTmp;
1725   return *this;
1726 }
1727
1728 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1729 {
1730   return replace(nStart, nLen, wxString(ch, nCount));
1731 }
1732
1733 wxString& wxString::replace(size_t nStart, size_t nLen,
1734                             const wxString& str, size_t nStart2, size_t nLen2)
1735 {
1736   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1737 }
1738
1739 wxString& wxString::replace(size_t nStart, size_t nLen,
1740                         const wxChar* sz, size_t nCount)
1741 {
1742   return replace(nStart, nLen, wxString(sz, nCount));
1743 }
1744
1745 #endif  //std::string compatibility
1746
1747 // ============================================================================
1748 // ArrayString
1749 // ============================================================================
1750
1751 // size increment = min(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1752 #define   ARRAY_MAXSIZE_INCREMENT       4096
1753
1754 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1755 #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1756 #endif
1757
1758 #define   STRING(p)   ((wxString *)(&(p)))
1759
1760 // ctor
1761 void wxArrayString::Init(bool autoSort)
1762 {
1763   m_nSize  =
1764   m_nCount = 0;
1765   m_pItems = (wxChar **) NULL;
1766   m_autoSort = autoSort;
1767 }
1768
1769 // copy ctor
1770 wxArrayString::wxArrayString(const wxArrayString& src)
1771 {
1772   Init(src.m_autoSort);
1773
1774   *this = src;
1775 }
1776
1777 // assignment operator
1778 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1779 {
1780   if ( m_nSize > 0 )
1781     Clear();
1782
1783   Copy(src);
1784
1785   m_autoSort = src.m_autoSort;
1786
1787   return *this;
1788 }
1789
1790 void wxArrayString::Copy(const wxArrayString& src)
1791 {
1792   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1793     Alloc(src.m_nCount);
1794
1795   for ( size_t n = 0; n < src.m_nCount; n++ )
1796     Add(src[n]);
1797 }
1798
1799 // grow the array
1800 void wxArrayString::Grow(size_t nIncrement)
1801 {
1802   // only do it if no more place
1803   if ( (m_nSize - m_nCount) < nIncrement ) {
1804     // if ARRAY_DEFAULT_INITIAL_SIZE were set to 0, the initially empty would
1805     // be never resized!
1806     #if ARRAY_DEFAULT_INITIAL_SIZE == 0
1807       #error "ARRAY_DEFAULT_INITIAL_SIZE must be > 0!"
1808     #endif
1809
1810     if ( m_nSize == 0 ) {
1811       // was empty, alloc some memory
1812       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1813       if (m_nSize < nIncrement)
1814           m_nSize = nIncrement;
1815       m_pItems = new wxChar *[m_nSize];
1816     }
1817     else {
1818       // otherwise when it's called for the first time, nIncrement would be 0
1819       // and the array would never be expanded
1820       // add 50% but not too much
1821       size_t ndefIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1822                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1823       if ( ndefIncrement > ARRAY_MAXSIZE_INCREMENT )
1824         ndefIncrement = ARRAY_MAXSIZE_INCREMENT;
1825       if ( nIncrement < ndefIncrement )
1826         nIncrement = ndefIncrement;
1827       m_nSize += nIncrement;
1828       wxChar **pNew = new wxChar *[m_nSize];
1829
1830       // copy data to new location
1831       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1832
1833       // delete old memory (but do not release the strings!)
1834       wxDELETEA(m_pItems);
1835
1836       m_pItems = pNew;
1837     }
1838   }
1839 }
1840
1841 void wxArrayString::Free()
1842 {
1843   for ( size_t n = 0; n < m_nCount; n++ ) {
1844     STRING(m_pItems[n])->GetStringData()->Unlock();
1845   }
1846 }
1847
1848 // deletes all the strings from the list
1849 void wxArrayString::Empty()
1850 {
1851   Free();
1852
1853   m_nCount = 0;
1854 }
1855
1856 // as Empty, but also frees memory
1857 void wxArrayString::Clear()
1858 {
1859   Free();
1860
1861   m_nSize  =
1862   m_nCount = 0;
1863
1864   wxDELETEA(m_pItems);
1865 }
1866
1867 // dtor
1868 wxArrayString::~wxArrayString()
1869 {
1870   Free();
1871
1872   wxDELETEA(m_pItems);
1873 }
1874
1875 // pre-allocates memory (frees the previous data!)
1876 void wxArrayString::Alloc(size_t nSize)
1877 {
1878   // only if old buffer was not big enough
1879   if ( nSize > m_nSize ) {
1880     Free();
1881     wxDELETEA(m_pItems);
1882     m_pItems = new wxChar *[nSize];
1883     m_nSize  = nSize;
1884   }
1885
1886   m_nCount = 0;
1887 }
1888
1889 // minimizes the memory usage by freeing unused memory
1890 void wxArrayString::Shrink()
1891 {
1892   // only do it if we have some memory to free
1893   if( m_nCount < m_nSize ) {
1894     // allocates exactly as much memory as we need
1895     wxChar **pNew = new wxChar *[m_nCount];
1896
1897     // copy data to new location
1898     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1899     delete [] m_pItems;
1900     m_pItems = pNew;
1901   }
1902 }
1903
1904 // return a wxString[] as required for some control ctors.
1905 wxString* wxArrayString::GetStringArray() const
1906 {
1907     wxString *array = 0;
1908
1909     if( m_nCount > 0 )
1910     {
1911         array = new wxString[m_nCount];
1912         for( size_t i = 0; i < m_nCount; i++ )
1913             array[i] = m_pItems[i];
1914     }
1915
1916     return array;
1917 }
1918
1919 // searches the array for an item (forward or backwards)
1920 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1921 {
1922   if ( m_autoSort ) {
1923     // use binary search in the sorted array
1924     wxASSERT_MSG( bCase && !bFromEnd,
1925                   wxT("search parameters ignored for auto sorted array") );
1926
1927     size_t i,
1928            lo = 0,
1929            hi = m_nCount;
1930     int res;
1931     while ( lo < hi ) {
1932       i = (lo + hi)/2;
1933
1934       res = wxStrcmp(sz, m_pItems[i]);
1935       if ( res < 0 )
1936         hi = i;
1937       else if ( res > 0 )
1938         lo = i + 1;
1939       else
1940         return i;
1941     }
1942
1943     return wxNOT_FOUND;
1944   }
1945   else {
1946     // use linear search in unsorted array
1947     if ( bFromEnd ) {
1948       if ( m_nCount > 0 ) {
1949         size_t ui = m_nCount;
1950         do {
1951           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1952             return ui;
1953         }
1954         while ( ui != 0 );
1955       }
1956     }
1957     else {
1958       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1959         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1960           return ui;
1961       }
1962     }
1963   }
1964
1965   return wxNOT_FOUND;
1966 }
1967
1968 // add item at the end
1969 size_t wxArrayString::Add(const wxString& str, size_t nInsert)
1970 {
1971   if ( m_autoSort ) {
1972     // insert the string at the correct position to keep the array sorted
1973     size_t i,
1974            lo = 0,
1975            hi = m_nCount;
1976     int res;
1977     while ( lo < hi ) {
1978       i = (lo + hi)/2;
1979
1980       res = wxStrcmp(str, m_pItems[i]);
1981       if ( res < 0 )
1982         hi = i;
1983       else if ( res > 0 )
1984         lo = i + 1;
1985       else {
1986         lo = hi = i;
1987         break;
1988       }
1989     }
1990
1991     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1992
1993     Insert(str, lo, nInsert);
1994
1995     return (size_t)lo;
1996   }
1997   else {
1998     wxASSERT( str.GetStringData()->IsValid() );
1999
2000     Grow(nInsert);
2001
2002     for (size_t i = 0; i < nInsert; i++)
2003     {
2004         // the string data must not be deleted!
2005         str.GetStringData()->Lock();
2006
2007         // just append
2008         m_pItems[m_nCount + i] = (wxChar *)str.c_str(); // const_cast
2009     }
2010     size_t ret = m_nCount;
2011     m_nCount += nInsert;
2012     return ret;
2013   }
2014 }
2015
2016 // add item at the given position
2017 void wxArrayString::Insert(const wxString& str, size_t nIndex, size_t nInsert)
2018 {
2019   wxASSERT( str.GetStringData()->IsValid() );
2020
2021   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2022   wxCHECK_RET( m_nCount <= m_nCount + nInsert,
2023                wxT("array size overflow in wxArrayString::Insert") );
2024
2025   Grow(nInsert);
2026
2027   memmove(&m_pItems[nIndex + nInsert], &m_pItems[nIndex],
2028           (m_nCount - nIndex)*sizeof(wxChar *));
2029
2030   for (size_t i = 0; i < nInsert; i++)
2031   {
2032       str.GetStringData()->Lock();
2033       m_pItems[nIndex + i] = (wxChar *)str.c_str();
2034   }
2035   m_nCount += nInsert;
2036 }
2037
2038 // expand the array
2039 void wxArrayString::SetCount(size_t count)
2040 {
2041     Alloc(count);
2042
2043     wxString s;
2044     while ( m_nCount < count )
2045         m_pItems[m_nCount++] = (wxChar *)s.c_str();
2046 }
2047
2048 // removes item from array (by index)
2049 void wxArrayString::Remove(size_t nIndex, size_t nRemove)
2050 {
2051   wxCHECK_RET( nIndex < m_nCount, wxT("bad index in wxArrayString::Remove") );
2052   wxCHECK_RET( nIndex + nRemove <= m_nCount,
2053                wxT("removing too many elements in wxArrayString::Remove") );
2054
2055   // release our lock
2056   for (size_t i = 0; i < nRemove; i++)
2057       Item(nIndex + i).GetStringData()->Unlock();
2058
2059   memmove(&m_pItems[nIndex], &m_pItems[nIndex + nRemove],
2060           (m_nCount - nIndex - nRemove)*sizeof(wxChar *));
2061   m_nCount -= nRemove;
2062 }
2063
2064 // removes item from array (by value)
2065 void wxArrayString::Remove(const wxChar *sz)
2066 {
2067   int iIndex = Index(sz);
2068
2069   wxCHECK_RET( iIndex != wxNOT_FOUND,
2070                wxT("removing inexistent element in wxArrayString::Remove") );
2071
2072   Remove(iIndex);
2073 }
2074
2075 // ----------------------------------------------------------------------------
2076 // sorting
2077 // ----------------------------------------------------------------------------
2078
2079 // we can only sort one array at a time with the quick-sort based
2080 // implementation
2081 #if wxUSE_THREADS
2082   // need a critical section to protect access to gs_compareFunction and
2083   // gs_sortAscending variables
2084   static wxCriticalSection *gs_critsectStringSort = NULL;
2085
2086   // call this before the value of the global sort vars is changed/after
2087   // you're finished with them
2088   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2089                            gs_critsectStringSort = new wxCriticalSection;     \
2090                            gs_critsectStringSort->Enter()
2091   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2092                            delete gs_critsectStringSort;                      \
2093                            gs_critsectStringSort = NULL
2094 #else // !threads
2095   #define START_SORT()
2096   #define END_SORT()
2097 #endif // wxUSE_THREADS
2098
2099 // function to use for string comparaison
2100 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2101
2102 // if we don't use the compare function, this flag tells us if we sort the
2103 // array in ascending or descending order
2104 static bool gs_sortAscending = TRUE;
2105
2106 // function which is called by quick sort
2107 extern "C" int wxC_CALLING_CONV     // LINKAGEMODE
2108 wxStringCompareFunction(const void *first, const void *second)
2109 {
2110   wxString *strFirst = (wxString *)first;
2111   wxString *strSecond = (wxString *)second;
2112
2113   if ( gs_compareFunction ) {
2114     return gs_compareFunction(*strFirst, *strSecond);
2115   }
2116   else {
2117     // maybe we should use wxStrcoll
2118     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2119
2120     return gs_sortAscending ? result : -result;
2121   }
2122 }
2123
2124 // sort array elements using passed comparaison function
2125 void wxArrayString::Sort(CompareFunction compareFunction)
2126 {
2127   START_SORT();
2128
2129   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2130   gs_compareFunction = compareFunction;
2131
2132   DoSort();
2133
2134   // reset it to NULL so that Sort(bool) will work the next time
2135   gs_compareFunction = NULL;
2136
2137   END_SORT();
2138 }
2139
2140 void wxArrayString::Sort(bool reverseOrder)
2141 {
2142   START_SORT();
2143
2144   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2145   gs_sortAscending = !reverseOrder;
2146
2147   DoSort();
2148
2149   END_SORT();
2150 }
2151
2152 void wxArrayString::DoSort()
2153 {
2154   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2155
2156   // just sort the pointers using qsort() - of course it only works because
2157   // wxString() *is* a pointer to its data
2158   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2159 }
2160
2161 bool wxArrayString::operator==(const wxArrayString& a) const
2162 {
2163     if ( m_nCount != a.m_nCount )
2164         return FALSE;
2165
2166     for ( size_t n = 0; n < m_nCount; n++ )
2167     {
2168         if ( Item(n) != a[n] )
2169             return FALSE;
2170     }
2171
2172     return TRUE;
2173 }
2174