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 license
  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 #if wxUSE_WCSRTOMBS
  50   #include <wchar.h>    // for wcsrtombs(), see comments where it's used
  51 #endif // GNU
  52
  53 #ifdef  WXSTRING_IS_WXOBJECT
  54   IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
  55 #endif  //WXSTRING_IS_WXOBJECT
  56
  57 #if wxUSE_UNICODE
  58 #undef wxUSE_EXPERIMENTAL_PRINTF
  59 #define wxUSE_EXPERIMENTAL_PRINTF 1
  60 #endif
  61
  62 // allocating extra space for each string consumes more memory but speeds up
  63 // the concatenation operations (nLen is the current string's length)
  64 // NB: EXTRA_ALLOC must be >= 0!
  65 #define EXTRA_ALLOC       (19 - nLen % 16)
  66
  67 // ---------------------------------------------------------------------------
  68 // static class variables definition
  69 // ---------------------------------------------------------------------------
  70
  71 #ifdef  wxSTD_STRING_COMPATIBILITY
  72   const size_t wxString::npos = wxSTRING_MAXLEN;
  73 #endif // wxSTD_STRING_COMPATIBILITY
  74
  75 // ----------------------------------------------------------------------------
  76 // static data
  77 // ----------------------------------------------------------------------------
  78
  79 // for an empty string, GetStringData() will return this address: this
  80 // structure has the same layout as wxStringData and it's data() method will
  81 // return the empty string (dummy pointer)
  82 static const struct
  83 {
  84   wxStringData data;
  85   wxChar dummy;
  86 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  87
  88 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  89 // must define this static for VA or else you get multiply defined symbols everywhere
  90 const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100;
  91
  92 #endif
  93
  94 // empty C style string: points to 'string data' byte of g_strEmpty
  95 extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy;
  96
  97 // ----------------------------------------------------------------------------
  98 // conditional compilation
  99 // ----------------------------------------------------------------------------
 100
 101 #if !defined(__WXSW__) && wxUSE_UNICODE
 102   #ifdef wxUSE_EXPERIMENTAL_PRINTF
 103     #undef wxUSE_EXPERIMENTAL_PRINTF
 104   #endif
 105   #define wxUSE_EXPERIMENTAL_PRINTF 1
 106 #endif
 107
 108 // we want to find out if the current platform supports vsnprintf()-like
 109 // function: for Unix this is done with configure, for Windows we test the
 110 // compiler explicitly.
 111 //
 112 // FIXME currently, this is only for ANSI (!Unicode) strings, so we call this
 113 //       function wxVsnprintfA (A for ANSI), should also find one for Unicode
 114 //       strings in Unicode build
 115 #ifdef __WXMSW__
 116     #if (defined(__VISUALC__) || defined(wxUSE_NORLANDER_HEADERS)) && !defined(__MINGW32__)
 117         #define wxVsnprintfA     _vsnprintf
 118     #endif
 119 #else   // !Windows
 120     #ifdef HAVE_VSNPRINTF
 121         #define wxVsnprintfA       vsnprintf
 122     #endif
 123 #endif  // Windows/!Windows
 124
 125 #ifndef wxVsnprintfA
 126     // in this case we'll use vsprintf() (which is ANSI and thus should be
 127     // always available), but it's unsafe because it doesn't check for buffer
 128     // size - so give a warning
 129     #define wxVsnprintfA(buf, len, format, arg) vsprintf(buf, format, arg)
 130
 131     #if defined(__VISUALC__)
 132         #pragma message("Using sprintf() because no snprintf()-like function defined")
 133     #elif defined(__GNUG__) && !defined(__UNIX__)
 134         #warning "Using sprintf() because no snprintf()-like function defined"
 135     #elif defined(__MWERKS__)
 136         #warning "Using sprintf() because no snprintf()-like function defined"
 137     #endif //compiler
 138 #endif // no vsnprintf
 139
 140 #ifdef _AIX
 141   // AIX has vsnprintf, but there's no prototype in the system headers.
 142   extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
 143 #endif
 144
 145 // ----------------------------------------------------------------------------
 146 // global functions
 147 // ----------------------------------------------------------------------------
 148
 149 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
 150
 151 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
 152 // iostream ones.
 153 //
 154 // ATTN: you can _not_ use both of these in the same program!
 155
 156 istream& operator>>(istream& is, wxString& WXUNUSED(str))
 157 {
 158 #if 0
 159   int w = is.width(0);
 160   if ( is.ipfx(0) ) {
 161     streambuf *sb = is.rdbuf();
 162     str.erase();
 163     while ( true ) {
 164       int ch = sb->sbumpc ();
 165       if ( ch == EOF ) {
 166         is.setstate(ios::eofbit);
 167         break;
 168       }
 169       else if ( isspace(ch) ) {
 170         sb->sungetc();
 171         break;
 172       }
 173
 174       str += ch;
 175       if ( --w == 1 )
 176         break;
 177     }
 178   }
 179
 180   is.isfx();
 181   if ( str.length() == 0 )
 182     is.setstate(ios::failbit);
 183 #endif
 184   return is;
 185 }
 186
 187 ostream& operator<<(ostream& os, const wxString& str)
 188 {
 189   os << str.c_str();
 190   return os;
 191 }
 192
 193 #endif  //std::string compatibility
 194
 195 extern int WXDLLEXPORT wxVsnprintf(wxChar *buf, size_t len,
 196                                    const wxChar *format, va_list argptr)
 197 {
 198 #if wxUSE_UNICODE
 199     // FIXME should use wvsnprintf() or whatever if it's available
 200     wxString s;
 201     int iLen = s.PrintfV(format, argptr);
 202     if ( iLen != -1 )
 203     {
 204         wxStrncpy(buf, s.c_str(), iLen);
 205     }
 206
 207     return iLen;
 208 #else // ANSI
 209     // vsnprintf() will not terminate the string with '\0' if there is not
 210     // enough place, but we want the string to always be NUL terminated
 211     int rc = wxVsnprintfA(buf, len - 1, format, argptr);
 212     if ( rc == -1 )
 213     {
 214         buf[len] = 0;
 215     }
 216
 217     return rc;
 218 #endif // Unicode/ANSI
 219 }
 220
 221 extern int WXDLLEXPORT wxSnprintf(wxChar *buf, size_t len,
 222                                   const wxChar *format, ...)
 223 {
 224     va_list argptr;
 225     va_start(argptr, format);
 226
 227     int iLen = wxVsnprintf(buf, len, format, argptr);
 228
 229     va_end(argptr);
 230
 231     return iLen;
 232 }
 233
 234 // ----------------------------------------------------------------------------
 235 // private classes
 236 // ----------------------------------------------------------------------------
 237
 238 // this small class is used to gather statistics for performance tuning
 239 //#define WXSTRING_STATISTICS
 240 #ifdef  WXSTRING_STATISTICS
 241   class Averager
 242   {
 243   public:
 244     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 245    ~Averager()
 246    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 247
 248     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 249
 250   private:
 251     size_t m_nCount, m_nTotal;
 252     const char *m_sz;
 253   } g_averageLength("allocation size"),
 254     g_averageSummandLength("summand length"),
 255     g_averageConcatHit("hit probability in concat"),
 256     g_averageInitialLength("initial string length");
 257
 258   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 259 #else
 260   #define STATISTICS_ADD(av, val)
 261 #endif // WXSTRING_STATISTICS
 262
 263 // ===========================================================================
 264 // wxString class core
 265 // ===========================================================================
 266
 267 // ---------------------------------------------------------------------------
 268 // construction
 269 // ---------------------------------------------------------------------------
 270
 271 // constructs string of <nLength> copies of character <ch>
 272 wxString::wxString(wxChar ch, size_t nLength)
 273 {
 274   Init();
 275
 276   if ( nLength > 0 ) {
 277     AllocBuffer(nLength);
 278
 279 #if wxUSE_UNICODE
 280     // memset only works on char
 281     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 282 #else
 283     memset(m_pchData, ch, nLength);
 284 #endif
 285   }
 286 }
 287
 288 // takes nLength elements of psz starting at nPos
 289 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 290 {
 291   Init();
 292
 293   if ( nLength == wxSTRING_MAXLEN )
 294     nLength = wxStrlen(psz + nPos);
 295
 296   wxASSERT_MSG( nPos + nLength <= wxStrlen(psz), _T("index out of bounds") );
 297
 298   STATISTICS_ADD(InitialLength, nLength);
 299
 300   if ( nLength > 0 ) {
 301     // trailing '\0' is written in AllocBuffer()
 302     AllocBuffer(nLength);
 303     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 304   }
 305 }
 306
 307 #ifdef  wxSTD_STRING_COMPATIBILITY
 308
 309 // poor man's iterators are "void *" pointers
 310 wxString::wxString(const void *pStart, const void *pEnd)
 311 {
 312   InitWith((const wxChar *)pStart, 0,
 313            (const wxChar *)pEnd - (const wxChar *)pStart);
 314 }
 315
 316 #endif  //std::string compatibility
 317
 318 #if wxUSE_UNICODE
 319
 320 // from multibyte string
 321 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 322 {
 323   // first get necessary size
 324   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 325
 326   // nLength is number of *Unicode* characters here!
 327   if ((nLen != (size_t)-1) && (nLen > nLength))
 328     nLen = nLength;
 329
 330   // empty?
 331   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 332     AllocBuffer(nLen);
 333     conv.MB2WC(m_pchData, psz, nLen);
 334   }
 335   else {
 336     Init();
 337   }
 338 }
 339
 340 #else // ANSI
 341
 342 #if wxUSE_WCHAR_T
 343 // from wide string
 344 wxString::wxString(const wchar_t *pwz)
 345 {
 346   // first get necessary size
 347   size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0;
 348
 349   // empty?
 350   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 351     AllocBuffer(nLen);
 352     wxWC2MB(m_pchData, pwz, nLen);
 353   }
 354   else {
 355     Init();
 356   }
 357 }
 358 #endif // wxUSE_WCHAR_T
 359
 360 #endif // Unicode/ANSI
 361
 362 // ---------------------------------------------------------------------------
 363 // memory allocation
 364 // ---------------------------------------------------------------------------
 365
 366 // allocates memory needed to store a C string of length nLen
 367 void wxString::AllocBuffer(size_t nLen)
 368 {
 369   // allocating 0 sized buffer doesn't make sense, all empty strings should
 370   // reuse g_strEmpty
 371   wxASSERT( nLen >  0 );
 372
 373   // make sure that we don't overflow
 374   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 375                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 376
 377   STATISTICS_ADD(Length, nLen);
 378
 379   // allocate memory:
 380   // 1) one extra character for '\0' termination
 381   // 2) sizeof(wxStringData) for housekeeping info
 382   wxStringData* pData = (wxStringData*)
 383     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 384   pData->nRefs        = 1;
 385   pData->nDataLength  = nLen;
 386   pData->nAllocLength = nLen + EXTRA_ALLOC;
 387   m_pchData           = pData->data();  // data starts after wxStringData
 388   m_pchData[nLen]     = wxT('\0');
 389 }
 390
 391 // must be called before changing this string
 392 void wxString::CopyBeforeWrite()
 393 {
 394   wxStringData* pData = GetStringData();
 395
 396   if ( pData->IsShared() ) {
 397     pData->Unlock();                // memory not freed because shared
 398     size_t nLen = pData->nDataLength;
 399     AllocBuffer(nLen);
 400     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 401   }
 402
 403   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 404 }
 405
 406 // must be called before replacing contents of this string
 407 void wxString::AllocBeforeWrite(size_t nLen)
 408 {
 409   wxASSERT( nLen != 0 );  // doesn't make any sense
 410
 411   // must not share string and must have enough space
 412   wxStringData* pData = GetStringData();
 413   if ( pData->IsShared() || pData->IsEmpty() ) {
 414     // can't work with old buffer, get new one
 415     pData->Unlock();
 416     AllocBuffer(nLen);
 417   }
 418   else {
 419     if ( nLen > pData->nAllocLength ) {
 420       // realloc the buffer instead of calling malloc() again, this is more
 421       // efficient
 422       STATISTICS_ADD(Length, nLen);
 423
 424       nLen += EXTRA_ALLOC;
 425
 426       wxStringData *pDataOld = pData;
 427       pData = (wxStringData*)
 428           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 429       if ( !pData ) {
 430         // out of memory
 431         free(pDataOld);
 432
 433         // FIXME we're going to crash...
 434         return;
 435       }
 436
 437       pData->nAllocLength = nLen;
 438       m_pchData = pData->data();
 439     }
 440
 441     // now we have enough space, just update the string length
 442     pData->nDataLength = nLen;
 443   }
 444
 445   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 446 }
 447
 448 // allocate enough memory for nLen characters
 449 void wxString::Alloc(size_t nLen)
 450 {
 451   wxStringData *pData = GetStringData();
 452   if ( pData->nAllocLength <= nLen ) {
 453     if ( pData->IsEmpty() ) {
 454       nLen += EXTRA_ALLOC;
 455
 456       wxStringData* pData = (wxStringData*)
 457         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 458       pData->nRefs = 1;
 459       pData->nDataLength = 0;
 460       pData->nAllocLength = nLen;
 461       m_pchData = pData->data();  // data starts after wxStringData
 462       m_pchData[0u] = wxT('\0');
 463     }
 464     else if ( pData->IsShared() ) {
 465       pData->Unlock();                // memory not freed because shared
 466       size_t nOldLen = pData->nDataLength;
 467       AllocBuffer(nLen);
 468       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 469     }
 470     else {
 471       nLen += EXTRA_ALLOC;
 472
 473       wxStringData *pDataOld = pData;
 474       wxStringData *p = (wxStringData *)
 475         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 476
 477       if ( p == NULL ) {
 478         // don't leak memory
 479         free(pDataOld);
 480
 481         // FIXME what to do on memory error?
 482         return;
 483       }
 484
 485       // it's not important if the pointer changed or not (the check for this
 486       // is not faster than assigning to m_pchData in all cases)
 487       p->nAllocLength = nLen;
 488       m_pchData = p->data();
 489     }
 490   }
 491   //else: we've already got enough
 492 }
 493
 494 // shrink to minimal size (releasing extra memory)
 495 void wxString::Shrink()
 496 {
 497   wxStringData *pData = GetStringData();
 498
 499   // this variable is unused in release build, so avoid the compiler warning
 500   // by just not declaring it
 501 #ifdef __WXDEBUG__
 502   void *p =
 503 #endif
 504   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 505
 506   // we rely on a reasonable realloc() implementation here - so far I haven't
 507   // seen any which wouldn't behave like this
 508
 509   wxASSERT( p != NULL );  // can't free memory?
 510   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 511 }
 512
 513 // get the pointer to writable buffer of (at least) nLen bytes
 514 wxChar *wxString::GetWriteBuf(size_t nLen)
 515 {
 516   AllocBeforeWrite(nLen);
 517
 518   wxASSERT( GetStringData()->nRefs == 1 );
 519   GetStringData()->Validate(FALSE);
 520
 521   return m_pchData;
 522 }
 523
 524 // put string back in a reasonable state after GetWriteBuf
 525 void wxString::UngetWriteBuf()
 526 {
 527   GetStringData()->nDataLength = wxStrlen(m_pchData);
 528   GetStringData()->Validate(TRUE);
 529 }
 530
 531 void wxString::UngetWriteBuf(size_t nLen)
 532 {
 533   GetStringData()->nDataLength = nLen;
 534   GetStringData()->Validate(TRUE);
 535 }
 536
 537 // ---------------------------------------------------------------------------
 538 // data access
 539 // ---------------------------------------------------------------------------
 540
 541 // all functions are inline in string.h
 542
 543 // ---------------------------------------------------------------------------
 544 // assignment operators
 545 // ---------------------------------------------------------------------------
 546
 547 // helper function: does real copy
 548 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 549 {
 550   if ( nSrcLen == 0 ) {
 551     Reinit();
 552   }
 553   else {
 554     AllocBeforeWrite(nSrcLen);
 555     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 556     GetStringData()->nDataLength = nSrcLen;
 557     m_pchData[nSrcLen] = wxT('\0');
 558   }
 559 }
 560
 561 // assigns one string to another
 562 wxString& wxString::operator=(const wxString& stringSrc)
 563 {
 564   wxASSERT( stringSrc.GetStringData()->IsValid() );
 565
 566   // don't copy string over itself
 567   if ( m_pchData != stringSrc.m_pchData ) {
 568     if ( stringSrc.GetStringData()->IsEmpty() ) {
 569       Reinit();
 570     }
 571     else {
 572       // adjust references
 573       GetStringData()->Unlock();
 574       m_pchData = stringSrc.m_pchData;
 575       GetStringData()->Lock();
 576     }
 577   }
 578
 579   return *this;
 580 }
 581
 582 // assigns a single character
 583 wxString& wxString::operator=(wxChar ch)
 584 {
 585   AssignCopy(1, &ch);
 586   return *this;
 587 }
 588
 589 // assigns C string
 590 wxString& wxString::operator=(const wxChar *psz)
 591 {
 592   AssignCopy(wxStrlen(psz), psz);
 593   return *this;
 594 }
 595
 596 #if !wxUSE_UNICODE
 597
 598 // same as 'signed char' variant
 599 wxString& wxString::operator=(const unsigned char* psz)
 600 {
 601   *this = (const char *)psz;
 602   return *this;
 603 }
 604
 605 #if wxUSE_WCHAR_T
 606 wxString& wxString::operator=(const wchar_t *pwz)
 607 {
 608   wxString str(pwz);
 609   *this = str;
 610   return *this;
 611 }
 612 #endif
 613
 614 #endif
 615
 616 // ---------------------------------------------------------------------------
 617 // string concatenation
 618 // ---------------------------------------------------------------------------
 619
 620 // add something to this string
 621 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 622 {
 623   STATISTICS_ADD(SummandLength, nSrcLen);
 624
 625   // concatenating an empty string is a NOP
 626   if ( nSrcLen > 0 ) {
 627     wxStringData *pData = GetStringData();
 628     size_t nLen = pData->nDataLength;
 629     size_t nNewLen = nLen + nSrcLen;
 630
 631     // alloc new buffer if current is too small
 632     if ( pData->IsShared() ) {
 633       STATISTICS_ADD(ConcatHit, 0);
 634
 635       // we have to allocate another buffer
 636       wxStringData* pOldData = GetStringData();
 637       AllocBuffer(nNewLen);
 638       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 639       pOldData->Unlock();
 640     }
 641     else if ( nNewLen > pData->nAllocLength ) {
 642       STATISTICS_ADD(ConcatHit, 0);
 643
 644       // we have to grow the buffer
 645       Alloc(nNewLen);
 646     }
 647     else {
 648       STATISTICS_ADD(ConcatHit, 1);
 649
 650       // the buffer is already big enough
 651     }
 652
 653     // should be enough space
 654     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 655
 656     // fast concatenation - all is done in our buffer
 657     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 658
 659     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 660     GetStringData()->nDataLength = nNewLen; // and fix the length
 661   }
 662   //else: the string to append was empty
 663 }
 664
 665 /*
 666  * concatenation functions come in 5 flavours:
 667  *  string + string
 668  *  char   + string      and      string + char
 669  *  C str  + string      and      string + C str
 670  */
 671
 672 wxString operator+(const wxString& string1, const wxString& string2)
 673 {
 674   wxASSERT( string1.GetStringData()->IsValid() );
 675   wxASSERT( string2.GetStringData()->IsValid() );
 676
 677   wxString s = string1;
 678   s += string2;
 679
 680   return s;
 681 }
 682
 683 wxString operator+(const wxString& string, wxChar ch)
 684 {
 685   wxASSERT( string.GetStringData()->IsValid() );
 686
 687   wxString s = string;
 688   s += ch;
 689
 690   return s;
 691 }
 692
 693 wxString operator+(wxChar ch, const wxString& string)
 694 {
 695   wxASSERT( string.GetStringData()->IsValid() );
 696
 697   wxString s = ch;
 698   s += string;
 699
 700   return s;
 701 }
 702
 703 wxString operator+(const wxString& string, const wxChar *psz)
 704 {
 705   wxASSERT( string.GetStringData()->IsValid() );
 706
 707   wxString s;
 708   s.Alloc(wxStrlen(psz) + string.Len());
 709   s = string;
 710   s += psz;
 711
 712   return s;
 713 }
 714
 715 wxString operator+(const wxChar *psz, const wxString& string)
 716 {
 717   wxASSERT( string.GetStringData()->IsValid() );
 718
 719   wxString s;
 720   s.Alloc(wxStrlen(psz) + string.Len());
 721   s = psz;
 722   s += string;
 723
 724   return s;
 725 }
 726
 727 // ===========================================================================
 728 // other common string functions
 729 // ===========================================================================
 730
 731 // ---------------------------------------------------------------------------
 732 // simple sub-string extraction
 733 // ---------------------------------------------------------------------------
 734
 735 // helper function: clone the data attached to this string
 736 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 737 {
 738   if ( nCopyLen == 0 ) {
 739     dest.Init();
 740   }
 741   else {
 742     dest.AllocBuffer(nCopyLen);
 743     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 744   }
 745 }
 746
 747 // extract string of length nCount starting at nFirst
 748 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 749 {
 750   wxStringData *pData = GetStringData();
 751   size_t nLen = pData->nDataLength;
 752
 753   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 754   if ( nCount == wxSTRING_MAXLEN )
 755   {
 756     nCount = nLen - nFirst;
 757   }
 758
 759   // out-of-bounds requests return sensible things
 760   if ( nFirst + nCount > nLen )
 761   {
 762     nCount = nLen - nFirst;
 763   }
 764
 765   if ( nFirst > nLen )
 766   {
 767     // AllocCopy() will return empty string
 768     nCount = 0;
 769   }
 770
 771   wxString dest;
 772   AllocCopy(dest, nCount, nFirst);
 773
 774   return dest;
 775 }
 776
 777 // extract nCount last (rightmost) characters
 778 wxString wxString::Right(size_t nCount) const
 779 {
 780   if ( nCount > (size_t)GetStringData()->nDataLength )
 781     nCount = GetStringData()->nDataLength;
 782
 783   wxString dest;
 784   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 785   return dest;
 786 }
 787
 788 // get all characters after the last occurence of ch
 789 // (returns the whole string if ch not found)
 790 wxString wxString::AfterLast(wxChar ch) const
 791 {
 792   wxString str;
 793   int iPos = Find(ch, TRUE);
 794   if ( iPos == wxNOT_FOUND )
 795     str = *this;
 796   else
 797     str = c_str() + iPos + 1;
 798
 799   return str;
 800 }
 801
 802 // extract nCount first (leftmost) characters
 803 wxString wxString::Left(size_t nCount) const
 804 {
 805   if ( nCount > (size_t)GetStringData()->nDataLength )
 806     nCount = GetStringData()->nDataLength;
 807
 808   wxString dest;
 809   AllocCopy(dest, nCount, 0);
 810   return dest;
 811 }
 812
 813 // get all characters before the first occurence of ch
 814 // (returns the whole string if ch not found)
 815 wxString wxString::BeforeFirst(wxChar ch) const
 816 {
 817   wxString str;
 818   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 819     str += *pc;
 820
 821   return str;
 822 }
 823
 824 /// get all characters before the last occurence of ch
 825 /// (returns empty string if ch not found)
 826 wxString wxString::BeforeLast(wxChar ch) const
 827 {
 828   wxString str;
 829   int iPos = Find(ch, TRUE);
 830   if ( iPos != wxNOT_FOUND && iPos != 0 )
 831     str = wxString(c_str(), iPos);
 832
 833   return str;
 834 }
 835
 836 /// get all characters after the first occurence of ch
 837 /// (returns empty string if ch not found)
 838 wxString wxString::AfterFirst(wxChar ch) const
 839 {
 840   wxString str;
 841   int iPos = Find(ch);
 842   if ( iPos != wxNOT_FOUND )
 843     str = c_str() + iPos + 1;
 844
 845   return str;
 846 }
 847
 848 // replace first (or all) occurences of some substring with another one
 849 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 850 {
 851   size_t uiCount = 0;   // count of replacements made
 852
 853   size_t uiOldLen = wxStrlen(szOld);
 854
 855   wxString strTemp;
 856   const wxChar *pCurrent = m_pchData;
 857   const wxChar *pSubstr;
 858   while ( *pCurrent != wxT('\0') ) {
 859     pSubstr = wxStrstr(pCurrent, szOld);
 860     if ( pSubstr == NULL ) {
 861       // strTemp is unused if no replacements were made, so avoid the copy
 862       if ( uiCount == 0 )
 863         return 0;
 864
 865       strTemp += pCurrent;    // copy the rest
 866       break;                  // exit the loop
 867     }
 868     else {
 869       // take chars before match
 870       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 871       strTemp += szNew;
 872       pCurrent = pSubstr + uiOldLen;  // restart after match
 873
 874       uiCount++;
 875
 876       // stop now?
 877       if ( !bReplaceAll ) {
 878         strTemp += pCurrent;    // copy the rest
 879         break;                  // exit the loop
 880       }
 881     }
 882   }
 883
 884   // only done if there were replacements, otherwise would have returned above
 885   *this = strTemp;
 886
 887   return uiCount;
 888 }
 889
 890 bool wxString::IsAscii() const
 891 {
 892   const wxChar *s = (const wxChar*) *this;
 893   while(*s){
 894     if(!isascii(*s)) return(FALSE);
 895     s++;
 896   }
 897   return(TRUE);
 898 }
 899
 900 bool wxString::IsWord() const
 901 {
 902   const wxChar *s = (const wxChar*) *this;
 903   while(*s){
 904     if(!wxIsalpha(*s)) return(FALSE);
 905     s++;
 906   }
 907   return(TRUE);
 908 }
 909
 910 bool wxString::IsNumber() const
 911 {
 912   const wxChar *s = (const wxChar*) *this;
 913   while(*s){
 914     if(!wxIsdigit(*s)) return(FALSE);
 915     s++;
 916   }
 917   return(TRUE);
 918 }
 919
 920 wxString wxString::Strip(stripType w) const
 921 {
 922     wxString s = *this;
 923     if ( w & leading ) s.Trim(FALSE);
 924     if ( w & trailing ) s.Trim(TRUE);
 925     return s;
 926 }
 927
 928 // ---------------------------------------------------------------------------
 929 // case conversion
 930 // ---------------------------------------------------------------------------
 931
 932 wxString& wxString::MakeUpper()
 933 {
 934   CopyBeforeWrite();
 935
 936   for ( wxChar *p = m_pchData; *p; p++ )
 937     *p = (wxChar)wxToupper(*p);
 938
 939   return *this;
 940 }
 941
 942 wxString& wxString::MakeLower()
 943 {
 944   CopyBeforeWrite();
 945
 946   for ( wxChar *p = m_pchData; *p; p++ )
 947     *p = (wxChar)wxTolower(*p);
 948
 949   return *this;
 950 }
 951
 952 // ---------------------------------------------------------------------------
 953 // trimming and padding
 954 // ---------------------------------------------------------------------------
 955
 956 // trims spaces (in the sense of isspace) from left or right side
 957 wxString& wxString::Trim(bool bFromRight)
 958 {
 959   // first check if we're going to modify the string at all
 960   if ( !IsEmpty() &&
 961        (
 962         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 963         (!bFromRight && wxIsspace(GetChar(0u)))
 964        )
 965      )
 966   {
 967     // ok, there is at least one space to trim
 968     CopyBeforeWrite();
 969
 970     if ( bFromRight )
 971     {
 972       // find last non-space character
 973       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
 974       while ( wxIsspace(*psz) && (psz >= m_pchData) )
 975         psz--;
 976
 977       // truncate at trailing space start
 978       *++psz = wxT('\0');
 979       GetStringData()->nDataLength = psz - m_pchData;
 980     }
 981     else
 982     {
 983       // find first non-space character
 984       const wxChar *psz = m_pchData;
 985       while ( wxIsspace(*psz) )
 986         psz++;
 987
 988       // fix up data and length
 989       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
 990       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
 991       GetStringData()->nDataLength = nDataLength;
 992     }
 993   }
 994
 995   return *this;
 996 }
 997
 998 // adds nCount characters chPad to the string from either side
 999 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1000 {
1001   wxString s(chPad, nCount);
1002
1003   if ( bFromRight )
1004     *this += s;
1005   else
1006   {
1007     s += *this;
1008     *this = s;
1009   }
1010
1011   return *this;
1012 }
1013
1014 // truncate the string
1015 wxString& wxString::Truncate(size_t uiLen)
1016 {
1017   if ( uiLen < Len() ) {
1018     CopyBeforeWrite();
1019
1020     *(m_pchData + uiLen) = wxT('\0');
1021     GetStringData()->nDataLength = uiLen;
1022   }
1023   //else: nothing to do, string is already short enough
1024
1025   return *this;
1026 }
1027
1028 // ---------------------------------------------------------------------------
1029 // finding (return wxNOT_FOUND if not found and index otherwise)
1030 // ---------------------------------------------------------------------------
1031
1032 // find a character
1033 int wxString::Find(wxChar ch, bool bFromEnd) const
1034 {
1035   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1036
1037   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1038 }
1039
1040 // find a sub-string (like strstr)
1041 int wxString::Find(const wxChar *pszSub) const
1042 {
1043   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1044
1045   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1046 }
1047
1048 // ----------------------------------------------------------------------------
1049 // conversion to numbers
1050 // ----------------------------------------------------------------------------
1051
1052 bool wxString::ToLong(long *val) const
1053 {
1054     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1055
1056     const wxChar *start = c_str();
1057     wxChar *end;
1058     *val = wxStrtol(start, &end, 10);
1059
1060     // return TRUE only if scan was stopped by the terminating NUL and if the
1061     // string was not empty to start with
1062     return !*end && (end != start);
1063 }
1064
1065 bool wxString::ToULong(unsigned long *val) const
1066 {
1067     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1068
1069     const wxChar *start = c_str();
1070     wxChar *end;
1071     *val = wxStrtoul(start, &end, 10);
1072
1073     // return TRUE only if scan was stopped by the terminating NUL and if the
1074     // string was not empty to start with
1075     return !*end && (end != start);
1076 }
1077
1078 bool wxString::ToDouble(double *val) const
1079 {
1080     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1081
1082     const wxChar *start = c_str();
1083     wxChar *end;
1084     *val = wxStrtod(start, &end);
1085
1086     // return TRUE only if scan was stopped by the terminating NUL and if the
1087     // string was not empty to start with
1088     return !*end && (end != start);
1089 }
1090
1091 // ---------------------------------------------------------------------------
1092 // formatted output
1093 // ---------------------------------------------------------------------------
1094
1095 /* static */
1096 wxString wxString::Format(const wxChar *pszFormat, ...)
1097 {
1098     va_list argptr;
1099     va_start(argptr, pszFormat);
1100
1101     wxString s;
1102     s.PrintfV(pszFormat, argptr);
1103
1104     va_end(argptr);
1105
1106     return s;
1107 }
1108
1109 /* static */
1110 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1111 {
1112     wxString s;
1113     s.Printf(pszFormat, argptr);
1114     return s;
1115 }
1116
1117 int wxString::Printf(const wxChar *pszFormat, ...)
1118 {
1119   va_list argptr;
1120   va_start(argptr, pszFormat);
1121
1122   int iLen = PrintfV(pszFormat, argptr);
1123
1124   va_end(argptr);
1125
1126   return iLen;
1127 }
1128
1129 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1130 {
1131 #if wxUSE_EXPERIMENTAL_PRINTF
1132   // the new implementation
1133
1134   // buffer to avoid dynamic memory allocation each time for small strings
1135   char szScratch[1024];
1136
1137   Reinit();
1138   for (size_t n = 0; pszFormat[n]; n++)
1139     if (pszFormat[n] == wxT('%')) {
1140       static char s_szFlags[256] = "%";
1141       size_t flagofs = 1;
1142       bool adj_left = FALSE, in_prec = FALSE,
1143            prec_dot = FALSE, done = FALSE;
1144       int ilen = 0;
1145       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1146       do {
1147 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1148         switch (pszFormat[++n]) {
1149         case wxT('\0'):
1150           done = TRUE;
1151           break;
1152         case wxT('%'):
1153           *this += wxT('%');
1154           done = TRUE;
1155           break;
1156         case wxT('#'):
1157         case wxT('0'):
1158         case wxT(' '):
1159         case wxT('+'):
1160         case wxT('\''):
1161           CHECK_PREC
1162           s_szFlags[flagofs++] = pszFormat[n];
1163           break;
1164         case wxT('-'):
1165           CHECK_PREC
1166           adj_left = TRUE;
1167           s_szFlags[flagofs++] = pszFormat[n];
1168           break;
1169         case wxT('.'):
1170           CHECK_PREC
1171           in_prec = TRUE;
1172           prec_dot = FALSE;
1173           max_width = 0;
1174           // dot will be auto-added to s_szFlags if non-negative number follows
1175           break;
1176         case wxT('h'):
1177           ilen = -1;
1178           CHECK_PREC
1179           s_szFlags[flagofs++] = pszFormat[n];
1180           break;
1181         case wxT('l'):
1182           ilen = 1;
1183           CHECK_PREC
1184           s_szFlags[flagofs++] = pszFormat[n];
1185           break;
1186         case wxT('q'):
1187         case wxT('L'):
1188           ilen = 2;
1189           CHECK_PREC
1190           s_szFlags[flagofs++] = pszFormat[n];
1191           break;
1192         case wxT('Z'):
1193           ilen = 3;
1194           CHECK_PREC
1195           s_szFlags[flagofs++] = pszFormat[n];
1196           break;
1197         case wxT('*'):
1198           {
1199             int len = va_arg(argptr, int);
1200             if (in_prec) {
1201               if (len<0) break;
1202               CHECK_PREC
1203               max_width = len;
1204             } else {
1205               if (len<0) {
1206                 adj_left = !adj_left;
1207                 s_szFlags[flagofs++] = '-';
1208                 len = -len;
1209               }
1210               min_width = len;
1211             }
1212             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1213           }
1214           break;
1215         case wxT('1'): case wxT('2'): case wxT('3'):
1216         case wxT('4'): case wxT('5'): case wxT('6'):
1217         case wxT('7'): case wxT('8'): case wxT('9'):
1218           {
1219             int len = 0;
1220             CHECK_PREC
1221             while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) {
1222               s_szFlags[flagofs++] = pszFormat[n];
1223               len = len*10 + (pszFormat[n] - wxT('0'));
1224               n++;
1225             }
1226             if (in_prec) max_width = len;
1227             else min_width = len;
1228             n--; // the main loop pre-increments n again
1229           }
1230           break;
1231         case wxT('d'):
1232         case wxT('i'):
1233         case wxT('o'):
1234         case wxT('u'):
1235         case wxT('x'):
1236         case wxT('X'):
1237           CHECK_PREC
1238           s_szFlags[flagofs++] = pszFormat[n];
1239           s_szFlags[flagofs] = '\0';
1240           if (ilen == 0 ) {
1241             int val = va_arg(argptr, int);
1242             ::sprintf(szScratch, s_szFlags, val);
1243           }
1244           else if (ilen == -1) {
1245             short int val = va_arg(argptr, short int);
1246             ::sprintf(szScratch, s_szFlags, val);
1247           }
1248           else if (ilen == 1) {
1249             long int val = va_arg(argptr, long int);
1250             ::sprintf(szScratch, s_szFlags, val);
1251           }
1252           else if (ilen == 2) {
1253 #if SIZEOF_LONG_LONG
1254             long long int val = va_arg(argptr, long long int);
1255             ::sprintf(szScratch, s_szFlags, val);
1256 #else
1257             long int val = va_arg(argptr, long int);
1258             ::sprintf(szScratch, s_szFlags, val);
1259 #endif
1260           }
1261           else if (ilen == 3) {
1262             size_t val = va_arg(argptr, size_t);
1263             ::sprintf(szScratch, s_szFlags, val);
1264           }
1265           *this += wxString(szScratch);
1266           done = TRUE;
1267           break;
1268         case wxT('e'):
1269         case wxT('E'):
1270         case wxT('f'):
1271         case wxT('g'):
1272         case wxT('G'):
1273           CHECK_PREC
1274           s_szFlags[flagofs++] = pszFormat[n];
1275           s_szFlags[flagofs] = '\0';
1276           if (ilen == 2) {
1277             long double val = va_arg(argptr, long double);
1278             ::sprintf(szScratch, s_szFlags, val);
1279           } else {
1280             double val = va_arg(argptr, double);
1281             ::sprintf(szScratch, s_szFlags, val);
1282           }
1283           *this += wxString(szScratch);
1284           done = TRUE;
1285           break;
1286         case wxT('p'):
1287           {
1288             void *val = va_arg(argptr, void *);
1289             CHECK_PREC
1290             s_szFlags[flagofs++] = pszFormat[n];
1291             s_szFlags[flagofs] = '\0';
1292             ::sprintf(szScratch, s_szFlags, val);
1293             *this += wxString(szScratch);
1294             done = TRUE;
1295           }
1296           break;
1297         case wxT('c'):
1298           {
1299             wxChar val = va_arg(argptr, int);
1300             // we don't need to honor padding here, do we?
1301             *this += val;
1302             done = TRUE;
1303           }
1304           break;
1305         case wxT('s'):
1306           if (ilen == -1) {
1307             // wx extension: we'll let %hs mean non-Unicode strings
1308             char *val = va_arg(argptr, char *);
1309 #if wxUSE_UNICODE
1310             // ASCII->Unicode constructor handles max_width right
1311             wxString s(val, wxConvLibc, max_width);
1312 #else
1313             size_t len = wxSTRING_MAXLEN;
1314             if (val) {
1315               for (len = 0; val[len] && (len<max_width); len++);
1316             } else val = wxT("(null)");
1317             wxString s(val, len);
1318 #endif
1319             if (s.Len() < min_width)
1320               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1321             *this += s;
1322           } else {
1323             wxChar *val = va_arg(argptr, wxChar *);
1324             size_t len = wxSTRING_MAXLEN;
1325             if (val) {
1326               for (len = 0; val[len] && (len<max_width); len++);
1327             } else val = wxT("(null)");
1328             wxString s(val, len);
1329             if (s.Len() < min_width)
1330               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1331             *this += s;
1332           }
1333           done = TRUE;
1334           break;
1335         case wxT('n'):
1336           if (ilen == 0) {
1337             int *val = va_arg(argptr, int *);
1338             *val = Len();
1339           }
1340           else if (ilen == -1) {
1341             short int *val = va_arg(argptr, short int *);
1342             *val = Len();
1343           }
1344           else if (ilen >= 1) {
1345             long int *val = va_arg(argptr, long int *);
1346             *val = Len();
1347           }
1348           done = TRUE;
1349           break;
1350         default:
1351           if (wxIsalpha(pszFormat[n]))
1352             // probably some flag not taken care of here yet
1353             s_szFlags[flagofs++] = pszFormat[n];
1354           else {
1355             // bad format
1356             *this += wxT('%'); // just to pass the glibc tst-printf.c
1357             n--;
1358             done = TRUE;
1359           }
1360           break;
1361         }
1362 #undef CHECK_PREC
1363       } while (!done);
1364     } else *this += pszFormat[n];
1365
1366 #else
1367   // buffer to avoid dynamic memory allocation each time for small strings
1368   char szScratch[1024];
1369
1370   // NB: wxVsnprintf() may return either less than the buffer size or -1 if
1371   //     there is not enough place depending on implementation
1372   int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), pszFormat, argptr);
1373   if ( iLen != -1 ) {
1374     // the whole string is in szScratch
1375     *this = szScratch;
1376   }
1377   else {
1378       bool outOfMemory = FALSE;
1379       int size = 2*WXSIZEOF(szScratch);
1380       while ( !outOfMemory ) {
1381           char *buf = GetWriteBuf(size);
1382           if ( buf )
1383             iLen = wxVsnprintfA(buf, size, pszFormat, argptr);
1384           else
1385             outOfMemory = TRUE;
1386
1387           UngetWriteBuf();
1388
1389           if ( iLen != -1 ) {
1390               // ok, there was enough space
1391               break;
1392           }
1393
1394           // still not enough, double it again
1395           size *= 2;
1396       }
1397
1398       if ( outOfMemory ) {
1399           // out of memory
1400           return -1;
1401       }
1402   }
1403 #endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF
1404
1405   return Len();
1406 }
1407
1408 // ----------------------------------------------------------------------------
1409 // misc other operations
1410 // ----------------------------------------------------------------------------
1411
1412 // returns TRUE if the string matches the pattern which may contain '*' and
1413 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1414 // of them)
1415 bool wxString::Matches(const wxChar *pszMask) const
1416 {
1417   // check char by char
1418   const wxChar *pszTxt;
1419   for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1420     switch ( *pszMask ) {
1421       case wxT('?'):
1422         if ( *pszTxt == wxT('\0') )
1423           return FALSE;
1424
1425         // pszText and pszMask will be incremented in the loop statement
1426
1427         break;
1428
1429       case wxT('*'):
1430         {
1431           // ignore special chars immediately following this one
1432           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1433             pszMask++;
1434
1435           // if there is nothing more, match
1436           if ( *pszMask == wxT('\0') )
1437             return TRUE;
1438
1439           // are there any other metacharacters in the mask?
1440           size_t uiLenMask;
1441           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1442
1443           if ( pEndMask != NULL ) {
1444             // we have to match the string between two metachars
1445             uiLenMask = pEndMask - pszMask;
1446           }
1447           else {
1448             // we have to match the remainder of the string
1449             uiLenMask = wxStrlen(pszMask);
1450           }
1451
1452           wxString strToMatch(pszMask, uiLenMask);
1453           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1454           if ( pMatch == NULL )
1455             return FALSE;
1456
1457           // -1 to compensate "++" in the loop
1458           pszTxt = pMatch + uiLenMask - 1;
1459           pszMask += uiLenMask - 1;
1460         }
1461         break;
1462
1463       default:
1464         if ( *pszMask != *pszTxt )
1465           return FALSE;
1466         break;
1467     }
1468   }
1469
1470   // match only if nothing left
1471   return *pszTxt == wxT('\0');
1472 }
1473
1474 // Count the number of chars
1475 int wxString::Freq(wxChar ch) const
1476 {
1477     int count = 0;
1478     int len = Len();
1479     for (int i = 0; i < len; i++)
1480     {
1481         if (GetChar(i) == ch)
1482             count ++;
1483     }
1484     return count;
1485 }
1486
1487 // convert to upper case, return the copy of the string
1488 wxString wxString::Upper() const
1489 { wxString s(*this); return s.MakeUpper(); }
1490
1491 // convert to lower case, return the copy of the string
1492 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1493
1494 int wxString::sprintf(const wxChar *pszFormat, ...)
1495   {
1496     va_list argptr;
1497     va_start(argptr, pszFormat);
1498     int iLen = PrintfV(pszFormat, argptr);
1499     va_end(argptr);
1500     return iLen;
1501   }
1502
1503 // ---------------------------------------------------------------------------
1504 // standard C++ library string functions
1505 // ---------------------------------------------------------------------------
1506 #ifdef  wxSTD_STRING_COMPATIBILITY
1507
1508 wxString& wxString::insert(size_t nPos, const wxString& str)
1509 {
1510   wxASSERT( str.GetStringData()->IsValid() );
1511   wxASSERT( nPos <= Len() );
1512
1513   if ( !str.IsEmpty() ) {
1514     wxString strTmp;
1515     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1516     wxStrncpy(pc, c_str(), nPos);
1517     wxStrcpy(pc + nPos, str);
1518     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1519     strTmp.UngetWriteBuf();
1520     *this = strTmp;
1521   }
1522
1523   return *this;
1524 }
1525
1526 size_t wxString::find(const wxString& str, size_t nStart) const
1527 {
1528   wxASSERT( str.GetStringData()->IsValid() );
1529   wxASSERT( nStart <= Len() );
1530
1531   const wxChar *p = wxStrstr(c_str() + nStart, str);
1532
1533   return p == NULL ? npos : p - c_str();
1534 }
1535
1536 // VC++ 1.5 can't cope with the default argument in the header.
1537 #if !defined(__VISUALC__) || defined(__WIN32__)
1538 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1539 {
1540   return find(wxString(sz, n), nStart);
1541 }
1542 #endif // VC++ 1.5
1543
1544 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1545 #if !defined(__BORLANDC__)
1546 size_t wxString::find(wxChar ch, size_t nStart) const
1547 {
1548   wxASSERT( nStart <= Len() );
1549
1550   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1551
1552   return p == NULL ? npos : p - c_str();
1553 }
1554 #endif
1555
1556 size_t wxString::rfind(const wxString& str, size_t nStart) const
1557 {
1558   wxASSERT( str.GetStringData()->IsValid() );
1559   wxASSERT( nStart <= Len() );
1560
1561   // TODO could be made much quicker than that
1562   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1563   while ( p >= c_str() + str.Len() ) {
1564     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1565       return p - str.Len() - c_str();
1566     p--;
1567   }
1568
1569   return npos;
1570 }
1571
1572 // VC++ 1.5 can't cope with the default argument in the header.
1573 #if !defined(__VISUALC__) || defined(__WIN32__)
1574 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1575 {
1576     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1577 }
1578
1579 size_t wxString::rfind(wxChar ch, size_t nStart) const
1580 {
1581     if ( nStart == npos )
1582     {
1583         nStart = Len();
1584     }
1585     else
1586     {
1587         wxASSERT( nStart <= Len() );
1588     }
1589
1590     const wxChar *p = wxStrrchr(c_str(), ch);
1591
1592     if ( p == NULL )
1593         return npos;
1594
1595     size_t result = p - c_str();
1596     return ( result > nStart ) ? npos : result;
1597 }
1598 #endif // VC++ 1.5
1599
1600 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1601 {
1602     const wxChar *start = c_str() + nStart;
1603     const wxChar *firstOf = wxStrpbrk(start, sz);
1604     if ( firstOf )
1605         return firstOf - c_str();
1606     else
1607         return npos;
1608 }
1609
1610 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1611 {
1612     if ( nStart == npos )
1613     {
1614         nStart = Len();
1615     }
1616     else
1617     {
1618         wxASSERT( nStart <= Len() );
1619     }
1620
1621     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1622     {
1623         if ( wxStrchr(sz, *p) )
1624             return p - c_str();
1625     }
1626
1627     return npos;
1628 }
1629
1630 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1631 {
1632     if ( nStart == npos )
1633     {
1634         nStart = Len();
1635     }
1636     else
1637     {
1638         wxASSERT( nStart <= Len() );
1639     }
1640
1641     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1642     if ( nAccept >= length() - nStart )
1643         return npos;
1644     else
1645         return nAccept;
1646 }
1647
1648 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1649 {
1650     wxASSERT( nStart <= Len() );
1651
1652     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1653     {
1654         if ( *p != ch )
1655             return p - c_str();
1656     }
1657
1658     return npos;
1659 }
1660
1661 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1662 {
1663     if ( nStart == npos )
1664     {
1665         nStart = Len();
1666     }
1667     else
1668     {
1669         wxASSERT( nStart <= Len() );
1670     }
1671
1672     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1673     {
1674         if ( !wxStrchr(sz, *p) )
1675             return p - c_str();
1676     }
1677
1678     return npos;
1679 }
1680
1681 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1682 {
1683     if ( nStart == npos )
1684     {
1685         nStart = Len();
1686     }
1687     else
1688     {
1689         wxASSERT( nStart <= Len() );
1690     }
1691
1692     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1693     {
1694         if ( *p != ch )
1695             return p - c_str();
1696     }
1697
1698     return npos;
1699 }
1700
1701 wxString& wxString::erase(size_t nStart, size_t nLen)
1702 {
1703   wxString strTmp(c_str(), nStart);
1704   if ( nLen != npos ) {
1705     wxASSERT( nStart + nLen <= Len() );
1706
1707     strTmp.append(c_str() + nStart + nLen);
1708   }
1709
1710   *this = strTmp;
1711   return *this;
1712 }
1713
1714 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1715 {
1716   wxASSERT( nStart + nLen <= wxStrlen(sz) );
1717
1718   wxString strTmp;
1719   if ( nStart != 0 )
1720     strTmp.append(c_str(), nStart);
1721   strTmp += sz;
1722   strTmp.append(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 = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1752 #define   ARRAY_MAXSIZE_INCREMENT       4096
1753 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1754   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1755 #endif
1756
1757 #define   STRING(p)   ((wxString *)(&(p)))
1758
1759 // ctor
1760 wxArrayString::wxArrayString(bool autoSort)
1761 {
1762   m_nSize  =
1763   m_nCount = 0;
1764   m_pItems = (wxChar **) NULL;
1765   m_autoSort = autoSort;
1766 }
1767
1768 // copy ctor
1769 wxArrayString::wxArrayString(const wxArrayString& src)
1770 {
1771   m_nSize  =
1772   m_nCount = 0;
1773   m_pItems = (wxChar **) NULL;
1774   m_autoSort = src.m_autoSort;
1775
1776   *this = src;
1777 }
1778
1779 // assignment operator
1780 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1781 {
1782   if ( m_nSize > 0 )
1783     Clear();
1784
1785   Copy(src);
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()
1801 {
1802   // only do it if no more place
1803   if( m_nCount == m_nSize ) {
1804     if( m_nSize == 0 ) {
1805       // was empty, alloc some memory
1806       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1807       m_pItems = new wxChar *[m_nSize];
1808     }
1809     else {
1810       // otherwise when it's called for the first time, nIncrement would be 0
1811       // and the array would never be expanded
1812 #if defined(__VISAGECPP__) && defined(__WXDEBUG__)
1813       int array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1814       wxASSERT( array_size != 0 );
1815 #else
1816       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1817 #endif
1818
1819       // add 50% but not too much
1820       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1821                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1822       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1823         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1824       m_nSize += nIncrement;
1825       wxChar **pNew = new wxChar *[m_nSize];
1826
1827       // copy data to new location
1828       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1829
1830       // delete old memory (but do not release the strings!)
1831       wxDELETEA(m_pItems);
1832
1833       m_pItems = pNew;
1834     }
1835   }
1836 }
1837
1838 void wxArrayString::Free()
1839 {
1840   for ( size_t n = 0; n < m_nCount; n++ ) {
1841     STRING(m_pItems[n])->GetStringData()->Unlock();
1842   }
1843 }
1844
1845 // deletes all the strings from the list
1846 void wxArrayString::Empty()
1847 {
1848   Free();
1849
1850   m_nCount = 0;
1851 }
1852
1853 // as Empty, but also frees memory
1854 void wxArrayString::Clear()
1855 {
1856   Free();
1857
1858   m_nSize  =
1859   m_nCount = 0;
1860
1861   wxDELETEA(m_pItems);
1862 }
1863
1864 // dtor
1865 wxArrayString::~wxArrayString()
1866 {
1867   Free();
1868
1869   wxDELETEA(m_pItems);
1870 }
1871
1872 // pre-allocates memory (frees the previous data!)
1873 void wxArrayString::Alloc(size_t nSize)
1874 {
1875   wxASSERT( nSize > 0 );
1876
1877   // only if old buffer was not big enough
1878   if ( nSize > m_nSize ) {
1879     Free();
1880     wxDELETEA(m_pItems);
1881     m_pItems = new wxChar *[nSize];
1882     m_nSize  = nSize;
1883   }
1884
1885   m_nCount = 0;
1886 }
1887
1888 // minimizes the memory usage by freeing unused memory
1889 void wxArrayString::Shrink()
1890 {
1891   // only do it if we have some memory to free
1892   if( m_nCount < m_nSize ) {
1893     // allocates exactly as much memory as we need
1894     wxChar **pNew = new wxChar *[m_nCount];
1895
1896     // copy data to new location
1897     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1898     delete [] m_pItems;
1899     m_pItems = pNew;
1900   }
1901 }
1902
1903 // searches the array for an item (forward or backwards)
1904 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1905 {
1906   if ( m_autoSort ) {
1907     // use binary search in the sorted array
1908     wxASSERT_MSG( bCase && !bFromEnd,
1909                   wxT("search parameters ignored for auto sorted array") );
1910
1911     size_t i,
1912            lo = 0,
1913            hi = m_nCount;
1914     int res;
1915     while ( lo < hi ) {
1916       i = (lo + hi)/2;
1917
1918       res = wxStrcmp(sz, m_pItems[i]);
1919       if ( res < 0 )
1920         hi = i;
1921       else if ( res > 0 )
1922         lo = i + 1;
1923       else
1924         return i;
1925     }
1926
1927     return wxNOT_FOUND;
1928   }
1929   else {
1930     // use linear search in unsorted array
1931     if ( bFromEnd ) {
1932       if ( m_nCount > 0 ) {
1933         size_t ui = m_nCount;
1934         do {
1935           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1936             return ui;
1937         }
1938         while ( ui != 0 );
1939       }
1940     }
1941     else {
1942       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1943         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1944           return ui;
1945       }
1946     }
1947   }
1948
1949   return wxNOT_FOUND;
1950 }
1951
1952 // add item at the end
1953 size_t wxArrayString::Add(const wxString& str)
1954 {
1955   if ( m_autoSort ) {
1956     // insert the string at the correct position to keep the array sorted
1957     size_t i,
1958            lo = 0,
1959            hi = m_nCount;
1960     int res;
1961     while ( lo < hi ) {
1962       i = (lo + hi)/2;
1963
1964       res = wxStrcmp(str, m_pItems[i]);
1965       if ( res < 0 )
1966         hi = i;
1967       else if ( res > 0 )
1968         lo = i + 1;
1969       else {
1970         lo = hi = i;
1971         break;
1972       }
1973     }
1974
1975     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1976
1977     Insert(str, lo);
1978
1979     return (size_t)lo;
1980   }
1981   else {
1982     wxASSERT( str.GetStringData()->IsValid() );
1983
1984     Grow();
1985
1986     // the string data must not be deleted!
1987     str.GetStringData()->Lock();
1988
1989     // just append
1990     m_pItems[m_nCount] = (wxChar *)str.c_str(); // const_cast
1991
1992     return m_nCount++;
1993   }
1994 }
1995
1996 // add item at the given position
1997 void wxArrayString::Insert(const wxString& str, size_t nIndex)
1998 {
1999   wxASSERT( str.GetStringData()->IsValid() );
2000
2001   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2002
2003   Grow();
2004
2005   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
2006           (m_nCount - nIndex)*sizeof(wxChar *));
2007
2008   str.GetStringData()->Lock();
2009   m_pItems[nIndex] = (wxChar *)str.c_str();
2010
2011   m_nCount++;
2012 }
2013
2014 // removes item from array (by index)
2015 void wxArrayString::Remove(size_t nIndex)
2016 {
2017   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Remove") );
2018
2019   // release our lock
2020   Item(nIndex).GetStringData()->Unlock();
2021
2022   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
2023           (m_nCount - nIndex - 1)*sizeof(wxChar *));
2024   m_nCount--;
2025 }
2026
2027 // removes item from array (by value)
2028 void wxArrayString::Remove(const wxChar *sz)
2029 {
2030   int iIndex = Index(sz);
2031
2032   wxCHECK_RET( iIndex != wxNOT_FOUND,
2033                wxT("removing inexistent element in wxArrayString::Remove") );
2034
2035   Remove(iIndex);
2036 }
2037
2038 // ----------------------------------------------------------------------------
2039 // sorting
2040 // ----------------------------------------------------------------------------
2041
2042 // we can only sort one array at a time with the quick-sort based
2043 // implementation
2044 #if wxUSE_THREADS
2045   // need a critical section to protect access to gs_compareFunction and
2046   // gs_sortAscending variables
2047   static wxCriticalSection *gs_critsectStringSort = NULL;
2048
2049   // call this before the value of the global sort vars is changed/after
2050   // you're finished with them
2051   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2052                            gs_critsectStringSort = new wxCriticalSection;     \
2053                            gs_critsectStringSort->Enter()
2054   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2055                            delete gs_critsectStringSort;                      \
2056                            gs_critsectStringSort = NULL
2057 #else // !threads
2058   #define START_SORT()
2059   #define END_SORT()
2060 #endif // wxUSE_THREADS
2061
2062 // function to use for string comparaison
2063 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2064
2065 // if we don't use the compare function, this flag tells us if we sort the
2066 // array in ascending or descending order
2067 static bool gs_sortAscending = TRUE;
2068
2069 // function which is called by quick sort
2070 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
2071 {
2072   wxString *strFirst = (wxString *)first;
2073   wxString *strSecond = (wxString *)second;
2074
2075   if ( gs_compareFunction ) {
2076     return gs_compareFunction(*strFirst, *strSecond);
2077   }
2078   else {
2079     // maybe we should use wxStrcoll
2080     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2081
2082     return gs_sortAscending ? result : -result;
2083   }
2084 }
2085
2086 // sort array elements using passed comparaison function
2087 void wxArrayString::Sort(CompareFunction compareFunction)
2088 {
2089   START_SORT();
2090
2091   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2092   gs_compareFunction = compareFunction;
2093
2094   DoSort();
2095
2096   END_SORT();
2097 }
2098
2099 void wxArrayString::Sort(bool reverseOrder)
2100 {
2101   START_SORT();
2102
2103   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2104   gs_sortAscending = !reverseOrder;
2105
2106   DoSort();
2107
2108   END_SORT();
2109 }
2110
2111 void wxArrayString::DoSort()
2112 {
2113   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2114
2115   // just sort the pointers using qsort() - of course it only works because
2116   // wxString() *is* a pointer to its data
2117   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2118 }
2119