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(__MINGW32__) && wxUSE_NORLANDER_HEADERS)
 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 the length is not given, assume the string to be NUL terminated
 294   if ( nLength == wxSTRING_MAXLEN ) {
 295     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 296
 297     nLength = wxStrlen(psz + nPos);
 298   }
 299
 300   STATISTICS_ADD(InitialLength, nLength);
 301
 302   if ( nLength > 0 ) {
 303     // trailing '\0' is written in AllocBuffer()
 304     AllocBuffer(nLength);
 305     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 306   }
 307 }
 308
 309 #ifdef  wxSTD_STRING_COMPATIBILITY
 310
 311 // poor man's iterators are "void *" pointers
 312 wxString::wxString(const void *pStart, const void *pEnd)
 313 {
 314   InitWith((const wxChar *)pStart, 0,
 315            (const wxChar *)pEnd - (const wxChar *)pStart);
 316 }
 317
 318 #endif  //std::string compatibility
 319
 320 #if wxUSE_UNICODE
 321
 322 // from multibyte string
 323 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 324 {
 325   // first get necessary size
 326   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 327
 328   // nLength is number of *Unicode* characters here!
 329   if ((nLen != (size_t)-1) && (nLen > nLength))
 330     nLen = nLength;
 331
 332   // empty?
 333   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 334     AllocBuffer(nLen);
 335     conv.MB2WC(m_pchData, psz, nLen);
 336   }
 337   else {
 338     Init();
 339   }
 340 }
 341
 342 #else // ANSI
 343
 344 #if wxUSE_WCHAR_T
 345 // from wide string
 346 wxString::wxString(const wchar_t *pwz, wxMBConv& conv)
 347 {
 348   // first get necessary size
 349   size_t nLen = pwz ? conv.WC2MB((char *) NULL, pwz, 0) : 0;
 350
 351   // empty?
 352   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 353     AllocBuffer(nLen);
 354     conv.WC2MB(m_pchData, pwz, nLen);
 355   }
 356   else {
 357     Init();
 358   }
 359 }
 360 #endif // wxUSE_WCHAR_T
 361
 362 #endif // Unicode/ANSI
 363
 364 // ---------------------------------------------------------------------------
 365 // memory allocation
 366 // ---------------------------------------------------------------------------
 367
 368 // allocates memory needed to store a C string of length nLen
 369 void wxString::AllocBuffer(size_t nLen)
 370 {
 371   // allocating 0 sized buffer doesn't make sense, all empty strings should
 372   // reuse g_strEmpty
 373   wxASSERT( nLen >  0 );
 374
 375   // make sure that we don't overflow
 376   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 377                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 378
 379   STATISTICS_ADD(Length, nLen);
 380
 381   // allocate memory:
 382   // 1) one extra character for '\0' termination
 383   // 2) sizeof(wxStringData) for housekeeping info
 384   wxStringData* pData = (wxStringData*)
 385     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 386   pData->nRefs        = 1;
 387   pData->nDataLength  = nLen;
 388   pData->nAllocLength = nLen + EXTRA_ALLOC;
 389   m_pchData           = pData->data();  // data starts after wxStringData
 390   m_pchData[nLen]     = wxT('\0');
 391 }
 392
 393 // must be called before changing this string
 394 void wxString::CopyBeforeWrite()
 395 {
 396   wxStringData* pData = GetStringData();
 397
 398   if ( pData->IsShared() ) {
 399     pData->Unlock();                // memory not freed because shared
 400     size_t nLen = pData->nDataLength;
 401     AllocBuffer(nLen);
 402     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 403   }
 404
 405   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 406 }
 407
 408 // must be called before replacing contents of this string
 409 void wxString::AllocBeforeWrite(size_t nLen)
 410 {
 411   wxASSERT( nLen != 0 );  // doesn't make any sense
 412
 413   // must not share string and must have enough space
 414   wxStringData* pData = GetStringData();
 415   if ( pData->IsShared() || pData->IsEmpty() ) {
 416     // can't work with old buffer, get new one
 417     pData->Unlock();
 418     AllocBuffer(nLen);
 419   }
 420   else {
 421     if ( nLen > pData->nAllocLength ) {
 422       // realloc the buffer instead of calling malloc() again, this is more
 423       // efficient
 424       STATISTICS_ADD(Length, nLen);
 425
 426       nLen += EXTRA_ALLOC;
 427
 428       wxStringData *pDataOld = pData;
 429       pData = (wxStringData*)
 430           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 431       if ( !pData ) {
 432         // out of memory
 433         free(pDataOld);
 434
 435         // FIXME we're going to crash...
 436         return;
 437       }
 438
 439       pData->nAllocLength = nLen;
 440       m_pchData = pData->data();
 441     }
 442
 443     // now we have enough space, just update the string length
 444     pData->nDataLength = nLen;
 445   }
 446
 447   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 448 }
 449
 450 // allocate enough memory for nLen characters
 451 void wxString::Alloc(size_t nLen)
 452 {
 453   wxStringData *pData = GetStringData();
 454   if ( pData->nAllocLength <= nLen ) {
 455     if ( pData->IsEmpty() ) {
 456       nLen += EXTRA_ALLOC;
 457
 458       wxStringData* pData = (wxStringData*)
 459         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 460       pData->nRefs = 1;
 461       pData->nDataLength = 0;
 462       pData->nAllocLength = nLen;
 463       m_pchData = pData->data();  // data starts after wxStringData
 464       m_pchData[0u] = wxT('\0');
 465     }
 466     else if ( pData->IsShared() ) {
 467       pData->Unlock();                // memory not freed because shared
 468       size_t nOldLen = pData->nDataLength;
 469       AllocBuffer(nLen);
 470       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 471     }
 472     else {
 473       nLen += EXTRA_ALLOC;
 474
 475       wxStringData *pDataOld = pData;
 476       wxStringData *p = (wxStringData *)
 477         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 478
 479       if ( p == NULL ) {
 480         // don't leak memory
 481         free(pDataOld);
 482
 483         // FIXME what to do on memory error?
 484         return;
 485       }
 486
 487       // it's not important if the pointer changed or not (the check for this
 488       // is not faster than assigning to m_pchData in all cases)
 489       p->nAllocLength = nLen;
 490       m_pchData = p->data();
 491     }
 492   }
 493   //else: we've already got enough
 494 }
 495
 496 // shrink to minimal size (releasing extra memory)
 497 void wxString::Shrink()
 498 {
 499   wxStringData *pData = GetStringData();
 500
 501   // this variable is unused in release build, so avoid the compiler warning
 502   // by just not declaring it
 503 #ifdef __WXDEBUG__
 504   void *p =
 505 #endif
 506   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 507
 508   // we rely on a reasonable realloc() implementation here - so far I haven't
 509   // seen any which wouldn't behave like this
 510
 511   wxASSERT( p != NULL );  // can't free memory?
 512   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 513 }
 514
 515 // get the pointer to writable buffer of (at least) nLen bytes
 516 wxChar *wxString::GetWriteBuf(size_t nLen)
 517 {
 518   AllocBeforeWrite(nLen);
 519
 520   wxASSERT( GetStringData()->nRefs == 1 );
 521   GetStringData()->Validate(FALSE);
 522
 523   return m_pchData;
 524 }
 525
 526 // put string back in a reasonable state after GetWriteBuf
 527 void wxString::UngetWriteBuf()
 528 {
 529   GetStringData()->nDataLength = wxStrlen(m_pchData);
 530   GetStringData()->Validate(TRUE);
 531 }
 532
 533 void wxString::UngetWriteBuf(size_t nLen)
 534 {
 535   GetStringData()->nDataLength = nLen;
 536   GetStringData()->Validate(TRUE);
 537 }
 538
 539 // ---------------------------------------------------------------------------
 540 // data access
 541 // ---------------------------------------------------------------------------
 542
 543 // all functions are inline in string.h
 544
 545 // ---------------------------------------------------------------------------
 546 // assignment operators
 547 // ---------------------------------------------------------------------------
 548
 549 // helper function: does real copy
 550 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 551 {
 552   if ( nSrcLen == 0 ) {
 553     Reinit();
 554   }
 555   else {
 556     AllocBeforeWrite(nSrcLen);
 557     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 558     GetStringData()->nDataLength = nSrcLen;
 559     m_pchData[nSrcLen] = wxT('\0');
 560   }
 561 }
 562
 563 // assigns one string to another
 564 wxString& wxString::operator=(const wxString& stringSrc)
 565 {
 566   wxASSERT( stringSrc.GetStringData()->IsValid() );
 567
 568   // don't copy string over itself
 569   if ( m_pchData != stringSrc.m_pchData ) {
 570     if ( stringSrc.GetStringData()->IsEmpty() ) {
 571       Reinit();
 572     }
 573     else {
 574       // adjust references
 575       GetStringData()->Unlock();
 576       m_pchData = stringSrc.m_pchData;
 577       GetStringData()->Lock();
 578     }
 579   }
 580
 581   return *this;
 582 }
 583
 584 // assigns a single character
 585 wxString& wxString::operator=(wxChar ch)
 586 {
 587   AssignCopy(1, &ch);
 588   return *this;
 589 }
 590
 591 // assigns C string
 592 wxString& wxString::operator=(const wxChar *psz)
 593 {
 594   AssignCopy(wxStrlen(psz), psz);
 595   return *this;
 596 }
 597
 598 #if !wxUSE_UNICODE
 599
 600 // same as 'signed char' variant
 601 wxString& wxString::operator=(const unsigned char* psz)
 602 {
 603   *this = (const char *)psz;
 604   return *this;
 605 }
 606
 607 #if wxUSE_WCHAR_T
 608 wxString& wxString::operator=(const wchar_t *pwz)
 609 {
 610   wxString str(pwz);
 611   *this = str;
 612   return *this;
 613 }
 614 #endif
 615
 616 #endif
 617
 618 // ---------------------------------------------------------------------------
 619 // string concatenation
 620 // ---------------------------------------------------------------------------
 621
 622 // add something to this string
 623 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 624 {
 625   STATISTICS_ADD(SummandLength, nSrcLen);
 626
 627   // concatenating an empty string is a NOP
 628   if ( nSrcLen > 0 ) {
 629     wxStringData *pData = GetStringData();
 630     size_t nLen = pData->nDataLength;
 631     size_t nNewLen = nLen + nSrcLen;
 632
 633     // alloc new buffer if current is too small
 634     if ( pData->IsShared() ) {
 635       STATISTICS_ADD(ConcatHit, 0);
 636
 637       // we have to allocate another buffer
 638       wxStringData* pOldData = GetStringData();
 639       AllocBuffer(nNewLen);
 640       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 641       pOldData->Unlock();
 642     }
 643     else if ( nNewLen > pData->nAllocLength ) {
 644       STATISTICS_ADD(ConcatHit, 0);
 645
 646       // we have to grow the buffer
 647       Alloc(nNewLen);
 648     }
 649     else {
 650       STATISTICS_ADD(ConcatHit, 1);
 651
 652       // the buffer is already big enough
 653     }
 654
 655     // should be enough space
 656     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 657
 658     // fast concatenation - all is done in our buffer
 659     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 660
 661     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 662     GetStringData()->nDataLength = nNewLen; // and fix the length
 663   }
 664   //else: the string to append was empty
 665 }
 666
 667 /*
 668  * concatenation functions come in 5 flavours:
 669  *  string + string
 670  *  char   + string      and      string + char
 671  *  C str  + string      and      string + C str
 672  */
 673
 674 wxString operator+(const wxString& string1, const wxString& string2)
 675 {
 676   wxASSERT( string1.GetStringData()->IsValid() );
 677   wxASSERT( string2.GetStringData()->IsValid() );
 678
 679   wxString s = string1;
 680   s += string2;
 681
 682   return s;
 683 }
 684
 685 wxString operator+(const wxString& string, wxChar ch)
 686 {
 687   wxASSERT( string.GetStringData()->IsValid() );
 688
 689   wxString s = string;
 690   s += ch;
 691
 692   return s;
 693 }
 694
 695 wxString operator+(wxChar ch, const wxString& string)
 696 {
 697   wxASSERT( string.GetStringData()->IsValid() );
 698
 699   wxString s = ch;
 700   s += string;
 701
 702   return s;
 703 }
 704
 705 wxString operator+(const wxString& string, const wxChar *psz)
 706 {
 707   wxASSERT( string.GetStringData()->IsValid() );
 708
 709   wxString s;
 710   s.Alloc(wxStrlen(psz) + string.Len());
 711   s = string;
 712   s += psz;
 713
 714   return s;
 715 }
 716
 717 wxString operator+(const wxChar *psz, const wxString& string)
 718 {
 719   wxASSERT( string.GetStringData()->IsValid() );
 720
 721   wxString s;
 722   s.Alloc(wxStrlen(psz) + string.Len());
 723   s = psz;
 724   s += string;
 725
 726   return s;
 727 }
 728
 729 // ===========================================================================
 730 // other common string functions
 731 // ===========================================================================
 732
 733 // ---------------------------------------------------------------------------
 734 // simple sub-string extraction
 735 // ---------------------------------------------------------------------------
 736
 737 // helper function: clone the data attached to this string
 738 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 739 {
 740   if ( nCopyLen == 0 ) {
 741     dest.Init();
 742   }
 743   else {
 744     dest.AllocBuffer(nCopyLen);
 745     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 746   }
 747 }
 748
 749 // extract string of length nCount starting at nFirst
 750 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 751 {
 752   wxStringData *pData = GetStringData();
 753   size_t nLen = pData->nDataLength;
 754
 755   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 756   if ( nCount == wxSTRING_MAXLEN )
 757   {
 758     nCount = nLen - nFirst;
 759   }
 760
 761   // out-of-bounds requests return sensible things
 762   if ( nFirst + nCount > nLen )
 763   {
 764     nCount = nLen - nFirst;
 765   }
 766
 767   if ( nFirst > nLen )
 768   {
 769     // AllocCopy() will return empty string
 770     nCount = 0;
 771   }
 772
 773   wxString dest;
 774   AllocCopy(dest, nCount, nFirst);
 775
 776   return dest;
 777 }
 778
 779 // check that the tring starts with prefix and return the rest of the string
 780 // in the provided pointer if it is not NULL, otherwise return FALSE
 781 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 782 {
 783     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 784
 785     // first check if the beginning of the string matches the prefix: note
 786     // that we don't have to check that we don't run out of this string as
 787     // when we reach the terminating NUL, either prefix string ends too (and
 788     // then it's ok) or we break out of the loop because there is no match
 789     const wxChar *p = c_str();
 790     while ( *prefix )
 791     {
 792         if ( *prefix++ != *p++ )
 793         {
 794             // no match
 795             return FALSE;
 796         }
 797     }
 798
 799     if ( rest )
 800     {
 801         // put the rest of the string into provided pointer
 802         *rest = p;
 803     }
 804
 805     return TRUE;
 806 }
 807
 808 // extract nCount last (rightmost) characters
 809 wxString wxString::Right(size_t nCount) const
 810 {
 811   if ( nCount > (size_t)GetStringData()->nDataLength )
 812     nCount = GetStringData()->nDataLength;
 813
 814   wxString dest;
 815   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 816   return dest;
 817 }
 818
 819 // get all characters after the last occurence of ch
 820 // (returns the whole string if ch not found)
 821 wxString wxString::AfterLast(wxChar ch) const
 822 {
 823   wxString str;
 824   int iPos = Find(ch, TRUE);
 825   if ( iPos == wxNOT_FOUND )
 826     str = *this;
 827   else
 828     str = c_str() + iPos + 1;
 829
 830   return str;
 831 }
 832
 833 // extract nCount first (leftmost) characters
 834 wxString wxString::Left(size_t nCount) const
 835 {
 836   if ( nCount > (size_t)GetStringData()->nDataLength )
 837     nCount = GetStringData()->nDataLength;
 838
 839   wxString dest;
 840   AllocCopy(dest, nCount, 0);
 841   return dest;
 842 }
 843
 844 // get all characters before the first occurence of ch
 845 // (returns the whole string if ch not found)
 846 wxString wxString::BeforeFirst(wxChar ch) const
 847 {
 848   wxString str;
 849   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 850     str += *pc;
 851
 852   return str;
 853 }
 854
 855 /// get all characters before the last occurence of ch
 856 /// (returns empty string if ch not found)
 857 wxString wxString::BeforeLast(wxChar ch) const
 858 {
 859   wxString str;
 860   int iPos = Find(ch, TRUE);
 861   if ( iPos != wxNOT_FOUND && iPos != 0 )
 862     str = wxString(c_str(), iPos);
 863
 864   return str;
 865 }
 866
 867 /// get all characters after the first occurence of ch
 868 /// (returns empty string if ch not found)
 869 wxString wxString::AfterFirst(wxChar ch) const
 870 {
 871   wxString str;
 872   int iPos = Find(ch);
 873   if ( iPos != wxNOT_FOUND )
 874     str = c_str() + iPos + 1;
 875
 876   return str;
 877 }
 878
 879 // replace first (or all) occurences of some substring with another one
 880 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 881 {
 882   size_t uiCount = 0;   // count of replacements made
 883
 884   size_t uiOldLen = wxStrlen(szOld);
 885
 886   wxString strTemp;
 887   const wxChar *pCurrent = m_pchData;
 888   const wxChar *pSubstr;
 889   while ( *pCurrent != wxT('\0') ) {
 890     pSubstr = wxStrstr(pCurrent, szOld);
 891     if ( pSubstr == NULL ) {
 892       // strTemp is unused if no replacements were made, so avoid the copy
 893       if ( uiCount == 0 )
 894         return 0;
 895
 896       strTemp += pCurrent;    // copy the rest
 897       break;                  // exit the loop
 898     }
 899     else {
 900       // take chars before match
 901       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 902       strTemp += szNew;
 903       pCurrent = pSubstr + uiOldLen;  // restart after match
 904
 905       uiCount++;
 906
 907       // stop now?
 908       if ( !bReplaceAll ) {
 909         strTemp += pCurrent;    // copy the rest
 910         break;                  // exit the loop
 911       }
 912     }
 913   }
 914
 915   // only done if there were replacements, otherwise would have returned above
 916   *this = strTemp;
 917
 918   return uiCount;
 919 }
 920
 921 bool wxString::IsAscii() const
 922 {
 923   const wxChar *s = (const wxChar*) *this;
 924   while(*s){
 925     if(!isascii(*s)) return(FALSE);
 926     s++;
 927   }
 928   return(TRUE);
 929 }
 930
 931 bool wxString::IsWord() const
 932 {
 933   const wxChar *s = (const wxChar*) *this;
 934   while(*s){
 935     if(!wxIsalpha(*s)) return(FALSE);
 936     s++;
 937   }
 938   return(TRUE);
 939 }
 940
 941 bool wxString::IsNumber() const
 942 {
 943   const wxChar *s = (const wxChar*) *this;
 944   if (wxStrlen(s))
 945      if ((s[0] == '-') || (s[0] == '+')) s++;
 946   while(*s){
 947     if(!wxIsdigit(*s)) return(FALSE);
 948     s++;
 949   }
 950   return(TRUE);
 951 }
 952
 953 wxString wxString::Strip(stripType w) const
 954 {
 955     wxString s = *this;
 956     if ( w & leading ) s.Trim(FALSE);
 957     if ( w & trailing ) s.Trim(TRUE);
 958     return s;
 959 }
 960
 961 // ---------------------------------------------------------------------------
 962 // case conversion
 963 // ---------------------------------------------------------------------------
 964
 965 wxString& wxString::MakeUpper()
 966 {
 967   CopyBeforeWrite();
 968
 969   for ( wxChar *p = m_pchData; *p; p++ )
 970     *p = (wxChar)wxToupper(*p);
 971
 972   return *this;
 973 }
 974
 975 wxString& wxString::MakeLower()
 976 {
 977   CopyBeforeWrite();
 978
 979   for ( wxChar *p = m_pchData; *p; p++ )
 980     *p = (wxChar)wxTolower(*p);
 981
 982   return *this;
 983 }
 984
 985 // ---------------------------------------------------------------------------
 986 // trimming and padding
 987 // ---------------------------------------------------------------------------
 988
 989 // trims spaces (in the sense of isspace) from left or right side
 990 wxString& wxString::Trim(bool bFromRight)
 991 {
 992   // first check if we're going to modify the string at all
 993   if ( !IsEmpty() &&
 994        (
 995         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 996         (!bFromRight && wxIsspace(GetChar(0u)))
 997        )
 998      )
 999   {
1000     // ok, there is at least one space to trim
1001     CopyBeforeWrite();
1002
1003     if ( bFromRight )
1004     {
1005       // find last non-space character
1006       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1007       while ( wxIsspace(*psz) && (psz >= m_pchData) )
1008         psz--;
1009
1010       // truncate at trailing space start
1011       *++psz = wxT('\0');
1012       GetStringData()->nDataLength = psz - m_pchData;
1013     }
1014     else
1015     {
1016       // find first non-space character
1017       const wxChar *psz = m_pchData;
1018       while ( wxIsspace(*psz) )
1019         psz++;
1020
1021       // fix up data and length
1022       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1023       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1024       GetStringData()->nDataLength = nDataLength;
1025     }
1026   }
1027
1028   return *this;
1029 }
1030
1031 // adds nCount characters chPad to the string from either side
1032 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1033 {
1034   wxString s(chPad, nCount);
1035
1036   if ( bFromRight )
1037     *this += s;
1038   else
1039   {
1040     s += *this;
1041     *this = s;
1042   }
1043
1044   return *this;
1045 }
1046
1047 // truncate the string
1048 wxString& wxString::Truncate(size_t uiLen)
1049 {
1050   if ( uiLen < Len() ) {
1051     CopyBeforeWrite();
1052
1053     *(m_pchData + uiLen) = wxT('\0');
1054     GetStringData()->nDataLength = uiLen;
1055   }
1056   //else: nothing to do, string is already short enough
1057
1058   return *this;
1059 }
1060
1061 // ---------------------------------------------------------------------------
1062 // finding (return wxNOT_FOUND if not found and index otherwise)
1063 // ---------------------------------------------------------------------------
1064
1065 // find a character
1066 int wxString::Find(wxChar ch, bool bFromEnd) const
1067 {
1068   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1069
1070   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1071 }
1072
1073 // find a sub-string (like strstr)
1074 int wxString::Find(const wxChar *pszSub) const
1075 {
1076   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1077
1078   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1079 }
1080
1081 // ----------------------------------------------------------------------------
1082 // conversion to numbers
1083 // ----------------------------------------------------------------------------
1084
1085 bool wxString::ToLong(long *val) const
1086 {
1087     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1088
1089     const wxChar *start = c_str();
1090     wxChar *end;
1091     *val = wxStrtol(start, &end, 10);
1092
1093     // return TRUE only if scan was stopped by the terminating NUL and if the
1094     // string was not empty to start with
1095     return !*end && (end != start);
1096 }
1097
1098 bool wxString::ToULong(unsigned long *val) const
1099 {
1100     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1101
1102     const wxChar *start = c_str();
1103     wxChar *end;
1104     *val = wxStrtoul(start, &end, 10);
1105
1106     // return TRUE only if scan was stopped by the terminating NUL and if the
1107     // string was not empty to start with
1108     return !*end && (end != start);
1109 }
1110
1111 bool wxString::ToDouble(double *val) const
1112 {
1113     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1114
1115     const wxChar *start = c_str();
1116     wxChar *end;
1117     *val = wxStrtod(start, &end);
1118
1119     // return TRUE only if scan was stopped by the terminating NUL and if the
1120     // string was not empty to start with
1121     return !*end && (end != start);
1122 }
1123
1124 // ---------------------------------------------------------------------------
1125 // formatted output
1126 // ---------------------------------------------------------------------------
1127
1128 /* static */
1129 wxString wxString::Format(const wxChar *pszFormat, ...)
1130 {
1131     va_list argptr;
1132     va_start(argptr, pszFormat);
1133
1134     wxString s;
1135     s.PrintfV(pszFormat, argptr);
1136
1137     va_end(argptr);
1138
1139     return s;
1140 }
1141
1142 /* static */
1143 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1144 {
1145     wxString s;
1146     s.Printf(pszFormat, argptr);
1147     return s;
1148 }
1149
1150 int wxString::Printf(const wxChar *pszFormat, ...)
1151 {
1152   va_list argptr;
1153   va_start(argptr, pszFormat);
1154
1155   int iLen = PrintfV(pszFormat, argptr);
1156
1157   va_end(argptr);
1158
1159   return iLen;
1160 }
1161
1162 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1163 {
1164 #if wxUSE_EXPERIMENTAL_PRINTF
1165   // the new implementation
1166
1167   // buffer to avoid dynamic memory allocation each time for small strings
1168   char szScratch[1024];
1169
1170   Reinit();
1171   for (size_t n = 0; pszFormat[n]; n++)
1172     if (pszFormat[n] == wxT('%')) {
1173       static char s_szFlags[256] = "%";
1174       size_t flagofs = 1;
1175       bool adj_left = FALSE, in_prec = FALSE,
1176            prec_dot = FALSE, done = FALSE;
1177       int ilen = 0;
1178       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1179       do {
1180 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1181         switch (pszFormat[++n]) {
1182         case wxT('\0'):
1183           done = TRUE;
1184           break;
1185         case wxT('%'):
1186           *this += wxT('%');
1187           done = TRUE;
1188           break;
1189         case wxT('#'):
1190         case wxT('0'):
1191         case wxT(' '):
1192         case wxT('+'):
1193         case wxT('\''):
1194           CHECK_PREC
1195           s_szFlags[flagofs++] = pszFormat[n];
1196           break;
1197         case wxT('-'):
1198           CHECK_PREC
1199           adj_left = TRUE;
1200           s_szFlags[flagofs++] = pszFormat[n];
1201           break;
1202         case wxT('.'):
1203           CHECK_PREC
1204           in_prec = TRUE;
1205           prec_dot = FALSE;
1206           max_width = 0;
1207           // dot will be auto-added to s_szFlags if non-negative number follows
1208           break;
1209         case wxT('h'):
1210           ilen = -1;
1211           CHECK_PREC
1212           s_szFlags[flagofs++] = pszFormat[n];
1213           break;
1214         case wxT('l'):
1215           ilen = 1;
1216           CHECK_PREC
1217           s_szFlags[flagofs++] = pszFormat[n];
1218           break;
1219         case wxT('q'):
1220         case wxT('L'):
1221           ilen = 2;
1222           CHECK_PREC
1223           s_szFlags[flagofs++] = pszFormat[n];
1224           break;
1225         case wxT('Z'):
1226           ilen = 3;
1227           CHECK_PREC
1228           s_szFlags[flagofs++] = pszFormat[n];
1229           break;
1230         case wxT('*'):
1231           {
1232             int len = va_arg(argptr, int);
1233             if (in_prec) {
1234               if (len<0) break;
1235               CHECK_PREC
1236               max_width = len;
1237             } else {
1238               if (len<0) {
1239                 adj_left = !adj_left;
1240                 s_szFlags[flagofs++] = '-';
1241                 len = -len;
1242               }
1243               min_width = len;
1244             }
1245             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1246           }
1247           break;
1248         case wxT('1'): case wxT('2'): case wxT('3'):
1249         case wxT('4'): case wxT('5'): case wxT('6'):
1250         case wxT('7'): case wxT('8'): case wxT('9'):
1251           {
1252             int len = 0;
1253             CHECK_PREC
1254             while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) {
1255               s_szFlags[flagofs++] = pszFormat[n];
1256               len = len*10 + (pszFormat[n] - wxT('0'));
1257               n++;
1258             }
1259             if (in_prec) max_width = len;
1260             else min_width = len;
1261             n--; // the main loop pre-increments n again
1262           }
1263           break;
1264         case wxT('d'):
1265         case wxT('i'):
1266         case wxT('o'):
1267         case wxT('u'):
1268         case wxT('x'):
1269         case wxT('X'):
1270           CHECK_PREC
1271           s_szFlags[flagofs++] = pszFormat[n];
1272           s_szFlags[flagofs] = '\0';
1273           if (ilen == 0 ) {
1274             int val = va_arg(argptr, int);
1275             ::sprintf(szScratch, s_szFlags, val);
1276           }
1277           else if (ilen == -1) {
1278             short int val = va_arg(argptr, short int);
1279             ::sprintf(szScratch, s_szFlags, val);
1280           }
1281           else if (ilen == 1) {
1282             long int val = va_arg(argptr, long int);
1283             ::sprintf(szScratch, s_szFlags, val);
1284           }
1285           else if (ilen == 2) {
1286 #if SIZEOF_LONG_LONG
1287             long long int val = va_arg(argptr, long long int);
1288             ::sprintf(szScratch, s_szFlags, val);
1289 #else
1290             long int val = va_arg(argptr, long int);
1291             ::sprintf(szScratch, s_szFlags, val);
1292 #endif
1293           }
1294           else if (ilen == 3) {
1295             size_t val = va_arg(argptr, size_t);
1296             ::sprintf(szScratch, s_szFlags, val);
1297           }
1298           *this += wxString(szScratch);
1299           done = TRUE;
1300           break;
1301         case wxT('e'):
1302         case wxT('E'):
1303         case wxT('f'):
1304         case wxT('g'):
1305         case wxT('G'):
1306           CHECK_PREC
1307           s_szFlags[flagofs++] = pszFormat[n];
1308           s_szFlags[flagofs] = '\0';
1309           if (ilen == 2) {
1310             long double val = va_arg(argptr, long double);
1311             ::sprintf(szScratch, s_szFlags, val);
1312           } else {
1313             double val = va_arg(argptr, double);
1314             ::sprintf(szScratch, s_szFlags, val);
1315           }
1316           *this += wxString(szScratch);
1317           done = TRUE;
1318           break;
1319         case wxT('p'):
1320           {
1321             void *val = va_arg(argptr, void *);
1322             CHECK_PREC
1323             s_szFlags[flagofs++] = pszFormat[n];
1324             s_szFlags[flagofs] = '\0';
1325             ::sprintf(szScratch, s_szFlags, val);
1326             *this += wxString(szScratch);
1327             done = TRUE;
1328           }
1329           break;
1330         case wxT('c'):
1331           {
1332             wxChar val = va_arg(argptr, int);
1333             // we don't need to honor padding here, do we?
1334             *this += val;
1335             done = TRUE;
1336           }
1337           break;
1338         case wxT('s'):
1339           if (ilen == -1) {
1340             // wx extension: we'll let %hs mean non-Unicode strings
1341             char *val = va_arg(argptr, char *);
1342 #if wxUSE_UNICODE
1343             // ASCII->Unicode constructor handles max_width right
1344             wxString s(val, wxConvLibc, max_width);
1345 #else
1346             size_t len = wxSTRING_MAXLEN;
1347             if (val) {
1348               for (len = 0; val[len] && (len<max_width); len++);
1349             } else val = wxT("(null)");
1350             wxString s(val, len);
1351 #endif
1352             if (s.Len() < min_width)
1353               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1354             *this += s;
1355           } else {
1356             wxChar *val = va_arg(argptr, wxChar *);
1357             size_t len = wxSTRING_MAXLEN;
1358             if (val) {
1359               for (len = 0; val[len] && (len<max_width); len++);
1360             } else val = wxT("(null)");
1361             wxString s(val, len);
1362             if (s.Len() < min_width)
1363               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1364             *this += s;
1365           }
1366           done = TRUE;
1367           break;
1368         case wxT('n'):
1369           if (ilen == 0) {
1370             int *val = va_arg(argptr, int *);
1371             *val = Len();
1372           }
1373           else if (ilen == -1) {
1374             short int *val = va_arg(argptr, short int *);
1375             *val = Len();
1376           }
1377           else if (ilen >= 1) {
1378             long int *val = va_arg(argptr, long int *);
1379             *val = Len();
1380           }
1381           done = TRUE;
1382           break;
1383         default:
1384           if (wxIsalpha(pszFormat[n]))
1385             // probably some flag not taken care of here yet
1386             s_szFlags[flagofs++] = pszFormat[n];
1387           else {
1388             // bad format
1389             *this += wxT('%'); // just to pass the glibc tst-printf.c
1390             n--;
1391             done = TRUE;
1392           }
1393           break;
1394         }
1395 #undef CHECK_PREC
1396       } while (!done);
1397     } else *this += pszFormat[n];
1398
1399 #else
1400   // buffer to avoid dynamic memory allocation each time for small strings
1401   char szScratch[1024];
1402
1403   // NB: wxVsnprintf() may return either less than the buffer size or -1 if
1404   //     there is not enough place depending on implementation
1405   int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), pszFormat, argptr);
1406   if ( iLen != -1 ) {
1407     // the whole string is in szScratch
1408     *this = szScratch;
1409   }
1410   else {
1411       bool outOfMemory = FALSE;
1412       int size = 2*WXSIZEOF(szScratch);
1413       while ( !outOfMemory ) {
1414           char *buf = GetWriteBuf(size);
1415           if ( buf )
1416             iLen = wxVsnprintfA(buf, size, pszFormat, argptr);
1417           else
1418             outOfMemory = TRUE;
1419
1420           UngetWriteBuf();
1421
1422           if ( iLen != -1 ) {
1423               // ok, there was enough space
1424               break;
1425           }
1426
1427           // still not enough, double it again
1428           size *= 2;
1429       }
1430
1431       if ( outOfMemory ) {
1432           // out of memory
1433           return -1;
1434       }
1435   }
1436 #endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF
1437
1438   return Len();
1439 }
1440
1441 // ----------------------------------------------------------------------------
1442 // misc other operations
1443 // ----------------------------------------------------------------------------
1444
1445 // returns TRUE if the string matches the pattern which may contain '*' and
1446 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1447 // of them)
1448 bool wxString::Matches(const wxChar *pszMask) const
1449 {
1450   // check char by char
1451   const wxChar *pszTxt;
1452   for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1453     switch ( *pszMask ) {
1454       case wxT('?'):
1455         if ( *pszTxt == wxT('\0') )
1456           return FALSE;
1457
1458         // pszText and pszMask will be incremented in the loop statement
1459
1460         break;
1461
1462       case wxT('*'):
1463         {
1464           // ignore special chars immediately following this one
1465           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1466             pszMask++;
1467
1468           // if there is nothing more, match
1469           if ( *pszMask == wxT('\0') )
1470             return TRUE;
1471
1472           // are there any other metacharacters in the mask?
1473           size_t uiLenMask;
1474           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1475
1476           if ( pEndMask != NULL ) {
1477             // we have to match the string between two metachars
1478             uiLenMask = pEndMask - pszMask;
1479           }
1480           else {
1481             // we have to match the remainder of the string
1482             uiLenMask = wxStrlen(pszMask);
1483           }
1484
1485           wxString strToMatch(pszMask, uiLenMask);
1486           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1487           if ( pMatch == NULL )
1488             return FALSE;
1489
1490           // -1 to compensate "++" in the loop
1491           pszTxt = pMatch + uiLenMask - 1;
1492           pszMask += uiLenMask - 1;
1493         }
1494         break;
1495
1496       default:
1497         if ( *pszMask != *pszTxt )
1498           return FALSE;
1499         break;
1500     }
1501   }
1502
1503   // match only if nothing left
1504   return *pszTxt == wxT('\0');
1505 }
1506
1507 // Count the number of chars
1508 int wxString::Freq(wxChar ch) const
1509 {
1510     int count = 0;
1511     int len = Len();
1512     for (int i = 0; i < len; i++)
1513     {
1514         if (GetChar(i) == ch)
1515             count ++;
1516     }
1517     return count;
1518 }
1519
1520 // convert to upper case, return the copy of the string
1521 wxString wxString::Upper() const
1522 { wxString s(*this); return s.MakeUpper(); }
1523
1524 // convert to lower case, return the copy of the string
1525 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1526
1527 int wxString::sprintf(const wxChar *pszFormat, ...)
1528   {
1529     va_list argptr;
1530     va_start(argptr, pszFormat);
1531     int iLen = PrintfV(pszFormat, argptr);
1532     va_end(argptr);
1533     return iLen;
1534   }
1535
1536 // ---------------------------------------------------------------------------
1537 // standard C++ library string functions
1538 // ---------------------------------------------------------------------------
1539
1540 #ifdef  wxSTD_STRING_COMPATIBILITY
1541
1542 void wxString::resize(size_t nSize, wxChar ch)
1543 {
1544     size_t len = length();
1545
1546     if ( nSize < len )
1547     {
1548         Truncate(nSize);
1549     }
1550     else if ( nSize > len )
1551     {
1552         *this += wxString(ch, len - nSize);
1553     }
1554     //else: we have exactly the specified length, nothing to do
1555 }
1556
1557 void wxString::swap(wxString& str)
1558 {
1559     // this is slightly less efficient than fiddling with m_pchData directly,
1560     // but it is still quite efficient as we don't copy the string here because
1561     // ref count always stays positive
1562     wxString tmp = str;
1563     str = *this;
1564     *this = str;
1565 }
1566
1567 wxString& wxString::insert(size_t nPos, const wxString& str)
1568 {
1569   wxASSERT( str.GetStringData()->IsValid() );
1570   wxASSERT( nPos <= Len() );
1571
1572   if ( !str.IsEmpty() ) {
1573     wxString strTmp;
1574     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1575     wxStrncpy(pc, c_str(), nPos);
1576     wxStrcpy(pc + nPos, str);
1577     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1578     strTmp.UngetWriteBuf();
1579     *this = strTmp;
1580   }
1581
1582   return *this;
1583 }
1584
1585 size_t wxString::find(const wxString& str, size_t nStart) const
1586 {
1587   wxASSERT( str.GetStringData()->IsValid() );
1588   wxASSERT( nStart <= Len() );
1589
1590   const wxChar *p = wxStrstr(c_str() + nStart, str);
1591
1592   return p == NULL ? npos : p - c_str();
1593 }
1594
1595 // VC++ 1.5 can't cope with the default argument in the header.
1596 #if !defined(__VISUALC__) || defined(__WIN32__)
1597 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1598 {
1599   return find(wxString(sz, n), nStart);
1600 }
1601 #endif // VC++ 1.5
1602
1603 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1604 #if !defined(__BORLANDC__)
1605 size_t wxString::find(wxChar ch, size_t nStart) const
1606 {
1607   wxASSERT( nStart <= Len() );
1608
1609   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1610
1611   return p == NULL ? npos : p - c_str();
1612 }
1613 #endif
1614
1615 size_t wxString::rfind(const wxString& str, size_t nStart) const
1616 {
1617   wxASSERT( str.GetStringData()->IsValid() );
1618   wxASSERT( nStart <= Len() );
1619
1620   // TODO could be made much quicker than that
1621   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1622   while ( p >= c_str() + str.Len() ) {
1623     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1624       return p - str.Len() - c_str();
1625     p--;
1626   }
1627
1628   return npos;
1629 }
1630
1631 // VC++ 1.5 can't cope with the default argument in the header.
1632 #if !defined(__VISUALC__) || defined(__WIN32__)
1633 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1634 {
1635     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1636 }
1637
1638 size_t wxString::rfind(wxChar ch, size_t nStart) const
1639 {
1640     if ( nStart == npos )
1641     {
1642         nStart = Len();
1643     }
1644     else
1645     {
1646         wxASSERT( nStart <= Len() );
1647     }
1648
1649     const wxChar *p = wxStrrchr(c_str(), ch);
1650
1651     if ( p == NULL )
1652         return npos;
1653
1654     size_t result = p - c_str();
1655     return ( result > nStart ) ? npos : result;
1656 }
1657 #endif // VC++ 1.5
1658
1659 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1660 {
1661     const wxChar *start = c_str() + nStart;
1662     const wxChar *firstOf = wxStrpbrk(start, sz);
1663     if ( firstOf )
1664         return firstOf - c_str();
1665     else
1666         return npos;
1667 }
1668
1669 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1670 {
1671     if ( nStart == npos )
1672     {
1673         nStart = Len();
1674     }
1675     else
1676     {
1677         wxASSERT( nStart <= Len() );
1678     }
1679
1680     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1681     {
1682         if ( wxStrchr(sz, *p) )
1683             return p - c_str();
1684     }
1685
1686     return npos;
1687 }
1688
1689 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1690 {
1691     if ( nStart == npos )
1692     {
1693         nStart = Len();
1694     }
1695     else
1696     {
1697         wxASSERT( nStart <= Len() );
1698     }
1699
1700     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1701     if ( nAccept >= length() - nStart )
1702         return npos;
1703     else
1704         return nAccept;
1705 }
1706
1707 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1708 {
1709     wxASSERT( nStart <= Len() );
1710
1711     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1712     {
1713         if ( *p != ch )
1714             return p - c_str();
1715     }
1716
1717     return npos;
1718 }
1719
1720 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1721 {
1722     if ( nStart == npos )
1723     {
1724         nStart = Len();
1725     }
1726     else
1727     {
1728         wxASSERT( nStart <= Len() );
1729     }
1730
1731     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1732     {
1733         if ( !wxStrchr(sz, *p) )
1734             return p - c_str();
1735     }
1736
1737     return npos;
1738 }
1739
1740 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1741 {
1742     if ( nStart == npos )
1743     {
1744         nStart = Len();
1745     }
1746     else
1747     {
1748         wxASSERT( nStart <= Len() );
1749     }
1750
1751     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1752     {
1753         if ( *p != ch )
1754             return p - c_str();
1755     }
1756
1757     return npos;
1758 }
1759
1760 wxString& wxString::erase(size_t nStart, size_t nLen)
1761 {
1762   wxString strTmp(c_str(), nStart);
1763   if ( nLen != npos ) {
1764     wxASSERT( nStart + nLen <= Len() );
1765
1766     strTmp.append(c_str() + nStart + nLen);
1767   }
1768
1769   *this = strTmp;
1770   return *this;
1771 }
1772
1773 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1774 {
1775   wxASSERT_MSG( nStart + nLen <= Len(),
1776                 _T("index out of bounds in wxString::replace") );
1777
1778   wxString strTmp;
1779   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1780
1781   if ( nStart != 0 )
1782     strTmp.append(c_str(), nStart);
1783   strTmp << sz << c_str() + nStart + nLen;
1784
1785   *this = strTmp;
1786   return *this;
1787 }
1788
1789 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1790 {
1791   return replace(nStart, nLen, wxString(ch, nCount));
1792 }
1793
1794 wxString& wxString::replace(size_t nStart, size_t nLen,
1795                             const wxString& str, size_t nStart2, size_t nLen2)
1796 {
1797   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1798 }
1799
1800 wxString& wxString::replace(size_t nStart, size_t nLen,
1801                         const wxChar* sz, size_t nCount)
1802 {
1803   return replace(nStart, nLen, wxString(sz, nCount));
1804 }
1805
1806 #endif  //std::string compatibility
1807
1808 // ============================================================================
1809 // ArrayString
1810 // ============================================================================
1811
1812 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1813 #define   ARRAY_MAXSIZE_INCREMENT       4096
1814 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1815   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1816 #endif
1817
1818 #define   STRING(p)   ((wxString *)(&(p)))
1819
1820 // ctor
1821 wxArrayString::wxArrayString(bool autoSort)
1822 {
1823   m_nSize  =
1824   m_nCount = 0;
1825   m_pItems = (wxChar **) NULL;
1826   m_autoSort = autoSort;
1827 }
1828
1829 // copy ctor
1830 wxArrayString::wxArrayString(const wxArrayString& src)
1831 {
1832   m_nSize  =
1833   m_nCount = 0;
1834   m_pItems = (wxChar **) NULL;
1835   m_autoSort = src.m_autoSort;
1836
1837   *this = src;
1838 }
1839
1840 // assignment operator
1841 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1842 {
1843   if ( m_nSize > 0 )
1844     Clear();
1845
1846   Copy(src);
1847
1848   return *this;
1849 }
1850
1851 void wxArrayString::Copy(const wxArrayString& src)
1852 {
1853   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1854     Alloc(src.m_nCount);
1855
1856   for ( size_t n = 0; n < src.m_nCount; n++ )
1857     Add(src[n]);
1858 }
1859
1860 // grow the array
1861 void wxArrayString::Grow()
1862 {
1863   // only do it if no more place
1864   if( m_nCount == m_nSize ) {
1865     if( m_nSize == 0 ) {
1866       // was empty, alloc some memory
1867       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1868       m_pItems = new wxChar *[m_nSize];
1869     }
1870     else {
1871       // otherwise when it's called for the first time, nIncrement would be 0
1872       // and the array would never be expanded
1873 #if defined(__VISAGECPP__) && defined(__WXDEBUG__)
1874       int array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1875       wxASSERT( array_size != 0 );
1876 #else
1877       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1878 #endif
1879
1880       // add 50% but not too much
1881       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1882                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1883       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1884         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1885       m_nSize += nIncrement;
1886       wxChar **pNew = new wxChar *[m_nSize];
1887
1888       // copy data to new location
1889       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1890
1891       // delete old memory (but do not release the strings!)
1892       wxDELETEA(m_pItems);
1893
1894       m_pItems = pNew;
1895     }
1896   }
1897 }
1898
1899 void wxArrayString::Free()
1900 {
1901   for ( size_t n = 0; n < m_nCount; n++ ) {
1902     STRING(m_pItems[n])->GetStringData()->Unlock();
1903   }
1904 }
1905
1906 // deletes all the strings from the list
1907 void wxArrayString::Empty()
1908 {
1909   Free();
1910
1911   m_nCount = 0;
1912 }
1913
1914 // as Empty, but also frees memory
1915 void wxArrayString::Clear()
1916 {
1917   Free();
1918
1919   m_nSize  =
1920   m_nCount = 0;
1921
1922   wxDELETEA(m_pItems);
1923 }
1924
1925 // dtor
1926 wxArrayString::~wxArrayString()
1927 {
1928   Free();
1929
1930   wxDELETEA(m_pItems);
1931 }
1932
1933 // pre-allocates memory (frees the previous data!)
1934 void wxArrayString::Alloc(size_t nSize)
1935 {
1936   wxASSERT( nSize > 0 );
1937
1938   // only if old buffer was not big enough
1939   if ( nSize > m_nSize ) {
1940     Free();
1941     wxDELETEA(m_pItems);
1942     m_pItems = new wxChar *[nSize];
1943     m_nSize  = nSize;
1944   }
1945
1946   m_nCount = 0;
1947 }
1948
1949 // minimizes the memory usage by freeing unused memory
1950 void wxArrayString::Shrink()
1951 {
1952   // only do it if we have some memory to free
1953   if( m_nCount < m_nSize ) {
1954     // allocates exactly as much memory as we need
1955     wxChar **pNew = new wxChar *[m_nCount];
1956
1957     // copy data to new location
1958     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1959     delete [] m_pItems;
1960     m_pItems = pNew;
1961   }
1962 }
1963
1964 // searches the array for an item (forward or backwards)
1965 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1966 {
1967   if ( m_autoSort ) {
1968     // use binary search in the sorted array
1969     wxASSERT_MSG( bCase && !bFromEnd,
1970                   wxT("search parameters ignored for auto sorted array") );
1971
1972     size_t i,
1973            lo = 0,
1974            hi = m_nCount;
1975     int res;
1976     while ( lo < hi ) {
1977       i = (lo + hi)/2;
1978
1979       res = wxStrcmp(sz, m_pItems[i]);
1980       if ( res < 0 )
1981         hi = i;
1982       else if ( res > 0 )
1983         lo = i + 1;
1984       else
1985         return i;
1986     }
1987
1988     return wxNOT_FOUND;
1989   }
1990   else {
1991     // use linear search in unsorted array
1992     if ( bFromEnd ) {
1993       if ( m_nCount > 0 ) {
1994         size_t ui = m_nCount;
1995         do {
1996           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1997             return ui;
1998         }
1999         while ( ui != 0 );
2000       }
2001     }
2002     else {
2003       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2004         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
2005           return ui;
2006       }
2007     }
2008   }
2009
2010   return wxNOT_FOUND;
2011 }
2012
2013 // add item at the end
2014 size_t wxArrayString::Add(const wxString& str)
2015 {
2016   if ( m_autoSort ) {
2017     // insert the string at the correct position to keep the array sorted
2018     size_t i,
2019            lo = 0,
2020            hi = m_nCount;
2021     int res;
2022     while ( lo < hi ) {
2023       i = (lo + hi)/2;
2024
2025       res = wxStrcmp(str, m_pItems[i]);
2026       if ( res < 0 )
2027         hi = i;
2028       else if ( res > 0 )
2029         lo = i + 1;
2030       else {
2031         lo = hi = i;
2032         break;
2033       }
2034     }
2035
2036     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2037
2038     Insert(str, lo);
2039
2040     return (size_t)lo;
2041   }
2042   else {
2043     wxASSERT( str.GetStringData()->IsValid() );
2044
2045     Grow();
2046
2047     // the string data must not be deleted!
2048     str.GetStringData()->Lock();
2049
2050     // just append
2051     m_pItems[m_nCount] = (wxChar *)str.c_str(); // const_cast
2052
2053     return m_nCount++;
2054   }
2055 }
2056
2057 // add item at the given position
2058 void wxArrayString::Insert(const wxString& str, size_t nIndex)
2059 {
2060   wxASSERT( str.GetStringData()->IsValid() );
2061
2062   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2063
2064   Grow();
2065
2066   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
2067           (m_nCount - nIndex)*sizeof(wxChar *));
2068
2069   str.GetStringData()->Lock();
2070   m_pItems[nIndex] = (wxChar *)str.c_str();
2071
2072   m_nCount++;
2073 }
2074
2075 // removes item from array (by index)
2076 void wxArrayString::Remove(size_t nIndex)
2077 {
2078   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Remove") );
2079
2080   // release our lock
2081   Item(nIndex).GetStringData()->Unlock();
2082
2083   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
2084           (m_nCount - nIndex - 1)*sizeof(wxChar *));
2085   m_nCount--;
2086 }
2087
2088 // removes item from array (by value)
2089 void wxArrayString::Remove(const wxChar *sz)
2090 {
2091   int iIndex = Index(sz);
2092
2093   wxCHECK_RET( iIndex != wxNOT_FOUND,
2094                wxT("removing inexistent element in wxArrayString::Remove") );
2095
2096   Remove(iIndex);
2097 }
2098
2099 // ----------------------------------------------------------------------------
2100 // sorting
2101 // ----------------------------------------------------------------------------
2102
2103 // we can only sort one array at a time with the quick-sort based
2104 // implementation
2105 #if wxUSE_THREADS
2106   // need a critical section to protect access to gs_compareFunction and
2107   // gs_sortAscending variables
2108   static wxCriticalSection *gs_critsectStringSort = NULL;
2109
2110   // call this before the value of the global sort vars is changed/after
2111   // you're finished with them
2112   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2113                            gs_critsectStringSort = new wxCriticalSection;     \
2114                            gs_critsectStringSort->Enter()
2115   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2116                            delete gs_critsectStringSort;                      \
2117                            gs_critsectStringSort = NULL
2118 #else // !threads
2119   #define START_SORT()
2120   #define END_SORT()
2121 #endif // wxUSE_THREADS
2122
2123 // function to use for string comparaison
2124 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2125
2126 // if we don't use the compare function, this flag tells us if we sort the
2127 // array in ascending or descending order
2128 static bool gs_sortAscending = TRUE;
2129
2130 // function which is called by quick sort
2131 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
2132 {
2133   wxString *strFirst = (wxString *)first;
2134   wxString *strSecond = (wxString *)second;
2135
2136   if ( gs_compareFunction ) {
2137     return gs_compareFunction(*strFirst, *strSecond);
2138   }
2139   else {
2140     // maybe we should use wxStrcoll
2141     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2142
2143     return gs_sortAscending ? result : -result;
2144   }
2145 }
2146
2147 // sort array elements using passed comparaison function
2148 void wxArrayString::Sort(CompareFunction compareFunction)
2149 {
2150   START_SORT();
2151
2152   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2153   gs_compareFunction = compareFunction;
2154
2155   DoSort();
2156
2157   // reset it to NULL so that Sort(bool) will work the next time
2158   gs_compareFunction = NULL;
2159
2160   END_SORT();
2161 }
2162
2163 void wxArrayString::Sort(bool reverseOrder)
2164 {
2165   START_SORT();
2166
2167   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2168   gs_sortAscending = !reverseOrder;
2169
2170   DoSort();
2171
2172   END_SORT();
2173 }
2174
2175 void wxArrayString::DoSort()
2176 {
2177   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2178
2179   // just sort the pointers using qsort() - of course it only works because
2180   // wxString() *is* a pointer to its data
2181   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2182 }
2183
2184 bool wxArrayString::operator==(const wxArrayString& a) const
2185 {
2186     if ( m_nCount != a.m_nCount )
2187         return FALSE;
2188
2189     for ( size_t n = 0; n < m_nCount; n++ )
2190     {
2191         if ( Item(n) != a[n] )
2192             return FALSE;
2193     }
2194
2195     return TRUE;
2196 }
2197