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 #elif defined(__WXMAC__)
 120     #define wxVsnprintfA       vsnprintf
 121 #else   // !Windows
 122     #ifdef HAVE_VSNPRINTF
 123         #define wxVsnprintfA       vsnprintf
 124     #endif
 125 #endif  // Windows/!Windows
 126
 127 #ifndef wxVsnprintfA
 128     // in this case we'll use vsprintf() (which is ANSI and thus should be
 129     // always available), but it's unsafe because it doesn't check for buffer
 130     // size - so give a warning
 131     #define wxVsnprintfA(buf, len, format, arg) vsprintf(buf, format, arg)
 132
 133     #if defined(__VISUALC__)
 134         #pragma message("Using sprintf() because no snprintf()-like function defined")
 135     #elif defined(__GNUG__)
 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 wxSTD istream& operator>>(wxSTD 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 wxSTD ostream& operator<<(wxSTD 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(), len);
 205         buf[len-1] = wxT('\0');
 206     }
 207
 208     return iLen;
 209 #else // ANSI
 210     // vsnprintf() will not terminate the string with '\0' if there is not
 211     // enough place, but we want the string to always be NUL terminated
 212     int rc = wxVsnprintfA(buf, len - 1, format, argptr);
 213     if ( rc == -1 )
 214     {
 215         buf[len] = 0;
 216     }
 217
 218     return rc;
 219 #endif // Unicode/ANSI
 220 }
 221
 222 extern int WXDLLEXPORT wxSnprintf(wxChar *buf, size_t len,
 223                                   const wxChar *format, ...)
 224 {
 225     va_list argptr;
 226     va_start(argptr, format);
 227
 228     int iLen = wxVsnprintf(buf, len, format, argptr);
 229
 230     va_end(argptr);
 231
 232     return iLen;
 233 }
 234
 235 // ----------------------------------------------------------------------------
 236 // private classes
 237 // ----------------------------------------------------------------------------
 238
 239 // this small class is used to gather statistics for performance tuning
 240 //#define WXSTRING_STATISTICS
 241 #ifdef  WXSTRING_STATISTICS
 242   class Averager
 243   {
 244   public:
 245     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 246    ~Averager()
 247    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 248
 249     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 250
 251   private:
 252     size_t m_nCount, m_nTotal;
 253     const char *m_sz;
 254   } g_averageLength("allocation size"),
 255     g_averageSummandLength("summand length"),
 256     g_averageConcatHit("hit probability in concat"),
 257     g_averageInitialLength("initial string length");
 258
 259   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 260 #else
 261   #define STATISTICS_ADD(av, val)
 262 #endif // WXSTRING_STATISTICS
 263
 264 // ===========================================================================
 265 // wxString class core
 266 // ===========================================================================
 267
 268 // ---------------------------------------------------------------------------
 269 // construction
 270 // ---------------------------------------------------------------------------
 271
 272 // constructs string of <nLength> copies of character <ch>
 273 wxString::wxString(wxChar ch, size_t nLength)
 274 {
 275   Init();
 276
 277   if ( nLength > 0 ) {
 278     AllocBuffer(nLength);
 279
 280 #if wxUSE_UNICODE
 281     // memset only works on char
 282     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 283 #else
 284     memset(m_pchData, ch, nLength);
 285 #endif
 286   }
 287 }
 288
 289 // takes nLength elements of psz starting at nPos
 290 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 291 {
 292   Init();
 293
 294   // if the length is not given, assume the string to be NUL terminated
 295   if ( nLength == wxSTRING_MAXLEN ) {
 296     wxASSERT_MSG( nPos <= wxStrlen(psz), _T("index out of bounds") );
 297
 298     nLength = wxStrlen(psz + nPos);
 299   }
 300
 301   STATISTICS_ADD(InitialLength, nLength);
 302
 303   if ( nLength > 0 ) {
 304     // trailing '\0' is written in AllocBuffer()
 305     AllocBuffer(nLength);
 306     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 307   }
 308 }
 309
 310 #ifdef  wxSTD_STRING_COMPATIBILITY
 311
 312 // poor man's iterators are "void *" pointers
 313 wxString::wxString(const void *pStart, const void *pEnd)
 314 {
 315   InitWith((const wxChar *)pStart, 0,
 316            (const wxChar *)pEnd - (const wxChar *)pStart);
 317 }
 318
 319 #endif  //std::string compatibility
 320
 321 #if wxUSE_UNICODE
 322
 323 // from multibyte string
 324 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 325 {
 326   // first get necessary size
 327   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 328
 329   // nLength is number of *Unicode* characters here!
 330   if ((nLen != (size_t)-1) && (nLen > nLength))
 331     nLen = nLength;
 332
 333   // empty?
 334   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 335     AllocBuffer(nLen);
 336     conv.MB2WC(m_pchData, psz, nLen);
 337   }
 338   else {
 339     Init();
 340   }
 341 }
 342
 343 #else // ANSI
 344
 345 #if wxUSE_WCHAR_T
 346 // from wide string
 347 wxString::wxString(const wchar_t *pwz, wxMBConv& conv)
 348 {
 349   // first get necessary size
 350   size_t nLen = pwz ? conv.WC2MB((char *) NULL, pwz, 0) : 0;
 351
 352   // empty?
 353   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 354     AllocBuffer(nLen);
 355     conv.WC2MB(m_pchData, pwz, nLen);
 356   }
 357   else {
 358     Init();
 359   }
 360 }
 361 #endif // wxUSE_WCHAR_T
 362
 363 #endif // Unicode/ANSI
 364
 365 // ---------------------------------------------------------------------------
 366 // memory allocation
 367 // ---------------------------------------------------------------------------
 368
 369 // allocates memory needed to store a C string of length nLen
 370 void wxString::AllocBuffer(size_t nLen)
 371 {
 372   // allocating 0 sized buffer doesn't make sense, all empty strings should
 373   // reuse g_strEmpty
 374   wxASSERT( nLen >  0 );
 375
 376   // make sure that we don't overflow
 377   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 378                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 379
 380   STATISTICS_ADD(Length, nLen);
 381
 382   // allocate memory:
 383   // 1) one extra character for '\0' termination
 384   // 2) sizeof(wxStringData) for housekeeping info
 385   wxStringData* pData = (wxStringData*)
 386     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 387   pData->nRefs        = 1;
 388   pData->nDataLength  = nLen;
 389   pData->nAllocLength = nLen + EXTRA_ALLOC;
 390   m_pchData           = pData->data();  // data starts after wxStringData
 391   m_pchData[nLen]     = wxT('\0');
 392 }
 393
 394 // must be called before changing this string
 395 void wxString::CopyBeforeWrite()
 396 {
 397   wxStringData* pData = GetStringData();
 398
 399   if ( pData->IsShared() ) {
 400     pData->Unlock();                // memory not freed because shared
 401     size_t nLen = pData->nDataLength;
 402     AllocBuffer(nLen);
 403     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 404   }
 405
 406   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 407 }
 408
 409 // must be called before replacing contents of this string
 410 void wxString::AllocBeforeWrite(size_t nLen)
 411 {
 412   wxASSERT( nLen != 0 );  // doesn't make any sense
 413
 414   // must not share string and must have enough space
 415   wxStringData* pData = GetStringData();
 416   if ( pData->IsShared() || pData->IsEmpty() ) {
 417     // can't work with old buffer, get new one
 418     pData->Unlock();
 419     AllocBuffer(nLen);
 420   }
 421   else {
 422     if ( nLen > pData->nAllocLength ) {
 423       // realloc the buffer instead of calling malloc() again, this is more
 424       // efficient
 425       STATISTICS_ADD(Length, nLen);
 426
 427       nLen += EXTRA_ALLOC;
 428
 429       wxStringData *pDataOld = pData;
 430       pData = (wxStringData*)
 431           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 432       if ( !pData ) {
 433         // out of memory
 434         free(pDataOld);
 435
 436         // FIXME we're going to crash...
 437         return;
 438       }
 439
 440       pData->nAllocLength = nLen;
 441       m_pchData = pData->data();
 442     }
 443
 444     // now we have enough space, just update the string length
 445     pData->nDataLength = nLen;
 446   }
 447
 448   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 449 }
 450
 451 // allocate enough memory for nLen characters
 452 void wxString::Alloc(size_t nLen)
 453 {
 454   wxStringData *pData = GetStringData();
 455   if ( pData->nAllocLength <= nLen ) {
 456     if ( pData->IsEmpty() ) {
 457       nLen += EXTRA_ALLOC;
 458
 459       wxStringData* pData = (wxStringData*)
 460         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 461       pData->nRefs = 1;
 462       pData->nDataLength = 0;
 463       pData->nAllocLength = nLen;
 464       m_pchData = pData->data();  // data starts after wxStringData
 465       m_pchData[0u] = wxT('\0');
 466     }
 467     else if ( pData->IsShared() ) {
 468       pData->Unlock();                // memory not freed because shared
 469       size_t nOldLen = pData->nDataLength;
 470       AllocBuffer(nLen);
 471       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 472     }
 473     else {
 474       nLen += EXTRA_ALLOC;
 475
 476       wxStringData *pDataOld = pData;
 477       wxStringData *p = (wxStringData *)
 478         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 479
 480       if ( p == NULL ) {
 481         // don't leak memory
 482         free(pDataOld);
 483
 484         // FIXME what to do on memory error?
 485         return;
 486       }
 487
 488       // it's not important if the pointer changed or not (the check for this
 489       // is not faster than assigning to m_pchData in all cases)
 490       p->nAllocLength = nLen;
 491       m_pchData = p->data();
 492     }
 493   }
 494   //else: we've already got enough
 495 }
 496
 497 // shrink to minimal size (releasing extra memory)
 498 void wxString::Shrink()
 499 {
 500   wxStringData *pData = GetStringData();
 501
 502   // this variable is unused in release build, so avoid the compiler warning
 503   // by just not declaring it
 504 #ifdef __WXDEBUG__
 505   void *p =
 506 #endif
 507   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 508
 509   // we rely on a reasonable realloc() implementation here - so far I haven't
 510   // seen any which wouldn't behave like this
 511
 512   wxASSERT( p != NULL );  // can't free memory?
 513   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 514 }
 515
 516 // get the pointer to writable buffer of (at least) nLen bytes
 517 wxChar *wxString::GetWriteBuf(size_t nLen)
 518 {
 519   AllocBeforeWrite(nLen);
 520
 521   wxASSERT( GetStringData()->nRefs == 1 );
 522   GetStringData()->Validate(FALSE);
 523
 524   return m_pchData;
 525 }
 526
 527 // put string back in a reasonable state after GetWriteBuf
 528 void wxString::UngetWriteBuf()
 529 {
 530   GetStringData()->nDataLength = wxStrlen(m_pchData);
 531   GetStringData()->Validate(TRUE);
 532 }
 533
 534 void wxString::UngetWriteBuf(size_t nLen)
 535 {
 536   GetStringData()->nDataLength = nLen;
 537   GetStringData()->Validate(TRUE);
 538 }
 539
 540 // ---------------------------------------------------------------------------
 541 // data access
 542 // ---------------------------------------------------------------------------
 543
 544 // all functions are inline in string.h
 545
 546 // ---------------------------------------------------------------------------
 547 // assignment operators
 548 // ---------------------------------------------------------------------------
 549
 550 // helper function: does real copy
 551 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 552 {
 553   if ( nSrcLen == 0 ) {
 554     Reinit();
 555   }
 556   else {
 557     AllocBeforeWrite(nSrcLen);
 558     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 559     GetStringData()->nDataLength = nSrcLen;
 560     m_pchData[nSrcLen] = wxT('\0');
 561   }
 562 }
 563
 564 // assigns one string to another
 565 wxString& wxString::operator=(const wxString& stringSrc)
 566 {
 567   wxASSERT( stringSrc.GetStringData()->IsValid() );
 568
 569   // don't copy string over itself
 570   if ( m_pchData != stringSrc.m_pchData ) {
 571     if ( stringSrc.GetStringData()->IsEmpty() ) {
 572       Reinit();
 573     }
 574     else {
 575       // adjust references
 576       GetStringData()->Unlock();
 577       m_pchData = stringSrc.m_pchData;
 578       GetStringData()->Lock();
 579     }
 580   }
 581
 582   return *this;
 583 }
 584
 585 // assigns a single character
 586 wxString& wxString::operator=(wxChar ch)
 587 {
 588   AssignCopy(1, &ch);
 589   return *this;
 590 }
 591
 592 // assigns C string
 593 wxString& wxString::operator=(const wxChar *psz)
 594 {
 595   AssignCopy(wxStrlen(psz), psz);
 596   return *this;
 597 }
 598
 599 #if !wxUSE_UNICODE
 600
 601 // same as 'signed char' variant
 602 wxString& wxString::operator=(const unsigned char* psz)
 603 {
 604   *this = (const char *)psz;
 605   return *this;
 606 }
 607
 608 #if wxUSE_WCHAR_T
 609 wxString& wxString::operator=(const wchar_t *pwz)
 610 {
 611   wxString str(pwz);
 612   *this = str;
 613   return *this;
 614 }
 615 #endif
 616
 617 #endif
 618
 619 // ---------------------------------------------------------------------------
 620 // string concatenation
 621 // ---------------------------------------------------------------------------
 622
 623 // add something to this string
 624 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 625 {
 626   STATISTICS_ADD(SummandLength, nSrcLen);
 627
 628   // concatenating an empty string is a NOP
 629   if ( nSrcLen > 0 ) {
 630     wxStringData *pData = GetStringData();
 631     size_t nLen = pData->nDataLength;
 632     size_t nNewLen = nLen + nSrcLen;
 633
 634     // alloc new buffer if current is too small
 635     if ( pData->IsShared() ) {
 636       STATISTICS_ADD(ConcatHit, 0);
 637
 638       // we have to allocate another buffer
 639       wxStringData* pOldData = GetStringData();
 640       AllocBuffer(nNewLen);
 641       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 642       pOldData->Unlock();
 643     }
 644     else if ( nNewLen > pData->nAllocLength ) {
 645       STATISTICS_ADD(ConcatHit, 0);
 646
 647       // we have to grow the buffer
 648       Alloc(nNewLen);
 649     }
 650     else {
 651       STATISTICS_ADD(ConcatHit, 1);
 652
 653       // the buffer is already big enough
 654     }
 655
 656     // should be enough space
 657     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 658
 659     // fast concatenation - all is done in our buffer
 660     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 661
 662     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 663     GetStringData()->nDataLength = nNewLen; // and fix the length
 664   }
 665   //else: the string to append was empty
 666 }
 667
 668 /*
 669  * concatenation functions come in 5 flavours:
 670  *  string + string
 671  *  char   + string      and      string + char
 672  *  C str  + string      and      string + C str
 673  */
 674
 675 wxString operator+(const wxString& string1, const wxString& string2)
 676 {
 677   wxASSERT( string1.GetStringData()->IsValid() );
 678   wxASSERT( string2.GetStringData()->IsValid() );
 679
 680   wxString s = string1;
 681   s += string2;
 682
 683   return s;
 684 }
 685
 686 wxString operator+(const wxString& string, wxChar ch)
 687 {
 688   wxASSERT( string.GetStringData()->IsValid() );
 689
 690   wxString s = string;
 691   s += ch;
 692
 693   return s;
 694 }
 695
 696 wxString operator+(wxChar ch, const wxString& string)
 697 {
 698   wxASSERT( string.GetStringData()->IsValid() );
 699
 700   wxString s = ch;
 701   s += string;
 702
 703   return s;
 704 }
 705
 706 wxString operator+(const wxString& string, const wxChar *psz)
 707 {
 708   wxASSERT( string.GetStringData()->IsValid() );
 709
 710   wxString s;
 711   s.Alloc(wxStrlen(psz) + string.Len());
 712   s = string;
 713   s += psz;
 714
 715   return s;
 716 }
 717
 718 wxString operator+(const wxChar *psz, const wxString& string)
 719 {
 720   wxASSERT( string.GetStringData()->IsValid() );
 721
 722   wxString s;
 723   s.Alloc(wxStrlen(psz) + string.Len());
 724   s = psz;
 725   s += string;
 726
 727   return s;
 728 }
 729
 730 // ===========================================================================
 731 // other common string functions
 732 // ===========================================================================
 733
 734 // ---------------------------------------------------------------------------
 735 // simple sub-string extraction
 736 // ---------------------------------------------------------------------------
 737
 738 // helper function: clone the data attached to this string
 739 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 740 {
 741   if ( nCopyLen == 0 ) {
 742     dest.Init();
 743   }
 744   else {
 745     dest.AllocBuffer(nCopyLen);
 746     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 747   }
 748 }
 749
 750 // extract string of length nCount starting at nFirst
 751 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 752 {
 753   wxStringData *pData = GetStringData();
 754   size_t nLen = pData->nDataLength;
 755
 756   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 757   if ( nCount == wxSTRING_MAXLEN )
 758   {
 759     nCount = nLen - nFirst;
 760   }
 761
 762   // out-of-bounds requests return sensible things
 763   if ( nFirst + nCount > nLen )
 764   {
 765     nCount = nLen - nFirst;
 766   }
 767
 768   if ( nFirst > nLen )
 769   {
 770     // AllocCopy() will return empty string
 771     nCount = 0;
 772   }
 773
 774   wxString dest;
 775   AllocCopy(dest, nCount, nFirst);
 776
 777   return dest;
 778 }
 779
 780 // check that the tring starts with prefix and return the rest of the string
 781 // in the provided pointer if it is not NULL, otherwise return FALSE
 782 bool wxString::StartsWith(const wxChar *prefix, wxString *rest) const
 783 {
 784     wxASSERT_MSG( prefix, _T("invalid parameter in wxString::StartsWith") );
 785
 786     // first check if the beginning of the string matches the prefix: note
 787     // that we don't have to check that we don't run out of this string as
 788     // when we reach the terminating NUL, either prefix string ends too (and
 789     // then it's ok) or we break out of the loop because there is no match
 790     const wxChar *p = c_str();
 791     while ( *prefix )
 792     {
 793         if ( *prefix++ != *p++ )
 794         {
 795             // no match
 796             return FALSE;
 797         }
 798     }
 799
 800     if ( rest )
 801     {
 802         // put the rest of the string into provided pointer
 803         *rest = p;
 804     }
 805
 806     return TRUE;
 807 }
 808
 809 // extract nCount last (rightmost) characters
 810 wxString wxString::Right(size_t nCount) const
 811 {
 812   if ( nCount > (size_t)GetStringData()->nDataLength )
 813     nCount = GetStringData()->nDataLength;
 814
 815   wxString dest;
 816   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 817   return dest;
 818 }
 819
 820 // get all characters after the last occurence of ch
 821 // (returns the whole string if ch not found)
 822 wxString wxString::AfterLast(wxChar ch) const
 823 {
 824   wxString str;
 825   int iPos = Find(ch, TRUE);
 826   if ( iPos == wxNOT_FOUND )
 827     str = *this;
 828   else
 829     str = c_str() + iPos + 1;
 830
 831   return str;
 832 }
 833
 834 // extract nCount first (leftmost) characters
 835 wxString wxString::Left(size_t nCount) const
 836 {
 837   if ( nCount > (size_t)GetStringData()->nDataLength )
 838     nCount = GetStringData()->nDataLength;
 839
 840   wxString dest;
 841   AllocCopy(dest, nCount, 0);
 842   return dest;
 843 }
 844
 845 // get all characters before the first occurence of ch
 846 // (returns the whole string if ch not found)
 847 wxString wxString::BeforeFirst(wxChar ch) const
 848 {
 849   wxString str;
 850   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 851     str += *pc;
 852
 853   return str;
 854 }
 855
 856 /// get all characters before the last occurence of ch
 857 /// (returns empty string if ch not found)
 858 wxString wxString::BeforeLast(wxChar ch) const
 859 {
 860   wxString str;
 861   int iPos = Find(ch, TRUE);
 862   if ( iPos != wxNOT_FOUND && iPos != 0 )
 863     str = wxString(c_str(), iPos);
 864
 865   return str;
 866 }
 867
 868 /// get all characters after the first occurence of ch
 869 /// (returns empty string if ch not found)
 870 wxString wxString::AfterFirst(wxChar ch) const
 871 {
 872   wxString str;
 873   int iPos = Find(ch);
 874   if ( iPos != wxNOT_FOUND )
 875     str = c_str() + iPos + 1;
 876
 877   return str;
 878 }
 879
 880 // replace first (or all) occurences of some substring with another one
 881 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 882 {
 883   size_t uiCount = 0;   // count of replacements made
 884
 885   size_t uiOldLen = wxStrlen(szOld);
 886
 887   wxString strTemp;
 888   const wxChar *pCurrent = m_pchData;
 889   const wxChar *pSubstr;
 890   while ( *pCurrent != wxT('\0') ) {
 891     pSubstr = wxStrstr(pCurrent, szOld);
 892     if ( pSubstr == NULL ) {
 893       // strTemp is unused if no replacements were made, so avoid the copy
 894       if ( uiCount == 0 )
 895         return 0;
 896
 897       strTemp += pCurrent;    // copy the rest
 898       break;                  // exit the loop
 899     }
 900     else {
 901       // take chars before match
 902       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 903       strTemp += szNew;
 904       pCurrent = pSubstr + uiOldLen;  // restart after match
 905
 906       uiCount++;
 907
 908       // stop now?
 909       if ( !bReplaceAll ) {
 910         strTemp += pCurrent;    // copy the rest
 911         break;                  // exit the loop
 912       }
 913     }
 914   }
 915
 916   // only done if there were replacements, otherwise would have returned above
 917   *this = strTemp;
 918
 919   return uiCount;
 920 }
 921
 922 bool wxString::IsAscii() const
 923 {
 924   const wxChar *s = (const wxChar*) *this;
 925   while(*s){
 926     if(!isascii(*s)) return(FALSE);
 927     s++;
 928   }
 929   return(TRUE);
 930 }
 931
 932 bool wxString::IsWord() const
 933 {
 934   const wxChar *s = (const wxChar*) *this;
 935   while(*s){
 936     if(!wxIsalpha(*s)) return(FALSE);
 937     s++;
 938   }
 939   return(TRUE);
 940 }
 941
 942 bool wxString::IsNumber() const
 943 {
 944   const wxChar *s = (const wxChar*) *this;
 945   if (wxStrlen(s))
 946      if ((s[0] == '-') || (s[0] == '+')) s++;
 947   while(*s){
 948     if(!wxIsdigit(*s)) return(FALSE);
 949     s++;
 950   }
 951   return(TRUE);
 952 }
 953
 954 wxString wxString::Strip(stripType w) const
 955 {
 956     wxString s = *this;
 957     if ( w & leading ) s.Trim(FALSE);
 958     if ( w & trailing ) s.Trim(TRUE);
 959     return s;
 960 }
 961
 962 // ---------------------------------------------------------------------------
 963 // case conversion
 964 // ---------------------------------------------------------------------------
 965
 966 wxString& wxString::MakeUpper()
 967 {
 968   CopyBeforeWrite();
 969
 970   for ( wxChar *p = m_pchData; *p; p++ )
 971     *p = (wxChar)wxToupper(*p);
 972
 973   return *this;
 974 }
 975
 976 wxString& wxString::MakeLower()
 977 {
 978   CopyBeforeWrite();
 979
 980   for ( wxChar *p = m_pchData; *p; p++ )
 981     *p = (wxChar)wxTolower(*p);
 982
 983   return *this;
 984 }
 985
 986 // ---------------------------------------------------------------------------
 987 // trimming and padding
 988 // ---------------------------------------------------------------------------
 989
 990 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
 991 // isspace('ê') in the C locale which seems to be broken to me, but we have to
 992 // live with this by checking that the character is a 7 bit one - even if this
 993 // may fail to detect some spaces (I don't know if Unicode doesn't have
 994 // space-like symbols somewhere except in the first 128 chars), it is arguably
 995 // still better than trimming away accented letters
 996 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
 997
 998 // trims spaces (in the sense of isspace) from left or right side
 999 wxString& wxString::Trim(bool bFromRight)
1000 {
1001   // first check if we're going to modify the string at all
1002   if ( !IsEmpty() &&
1003        (
1004         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1005         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1006        )
1007      )
1008   {
1009     // ok, there is at least one space to trim
1010     CopyBeforeWrite();
1011
1012     if ( bFromRight )
1013     {
1014       // find last non-space character
1015       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1016       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1017         psz--;
1018
1019       // truncate at trailing space start
1020       *++psz = wxT('\0');
1021       GetStringData()->nDataLength = psz - m_pchData;
1022     }
1023     else
1024     {
1025       // find first non-space character
1026       const wxChar *psz = m_pchData;
1027       while ( wxSafeIsspace(*psz) )
1028         psz++;
1029
1030       // fix up data and length
1031       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1032       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1033       GetStringData()->nDataLength = nDataLength;
1034     }
1035   }
1036
1037   return *this;
1038 }
1039
1040 // adds nCount characters chPad to the string from either side
1041 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1042 {
1043   wxString s(chPad, nCount);
1044
1045   if ( bFromRight )
1046     *this += s;
1047   else
1048   {
1049     s += *this;
1050     *this = s;
1051   }
1052
1053   return *this;
1054 }
1055
1056 // truncate the string
1057 wxString& wxString::Truncate(size_t uiLen)
1058 {
1059   if ( uiLen < Len() ) {
1060     CopyBeforeWrite();
1061
1062     *(m_pchData + uiLen) = wxT('\0');
1063     GetStringData()->nDataLength = uiLen;
1064   }
1065   //else: nothing to do, string is already short enough
1066
1067   return *this;
1068 }
1069
1070 // ---------------------------------------------------------------------------
1071 // finding (return wxNOT_FOUND if not found and index otherwise)
1072 // ---------------------------------------------------------------------------
1073
1074 // find a character
1075 int wxString::Find(wxChar ch, bool bFromEnd) const
1076 {
1077   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1078
1079   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1080 }
1081
1082 // find a sub-string (like strstr)
1083 int wxString::Find(const wxChar *pszSub) const
1084 {
1085   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1086
1087   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1088 }
1089
1090 // ----------------------------------------------------------------------------
1091 // conversion to numbers
1092 // ----------------------------------------------------------------------------
1093
1094 bool wxString::ToLong(long *val) const
1095 {
1096     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1097
1098     const wxChar *start = c_str();
1099     wxChar *end;
1100     *val = wxStrtol(start, &end, 10);
1101
1102     // return TRUE only if scan was stopped by the terminating NUL and if the
1103     // string was not empty to start with
1104     return !*end && (end != start);
1105 }
1106
1107 bool wxString::ToULong(unsigned long *val) const
1108 {
1109     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1110
1111     const wxChar *start = c_str();
1112     wxChar *end;
1113     *val = wxStrtoul(start, &end, 10);
1114
1115     // return TRUE only if scan was stopped by the terminating NUL and if the
1116     // string was not empty to start with
1117     return !*end && (end != start);
1118 }
1119
1120 bool wxString::ToDouble(double *val) const
1121 {
1122     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1123
1124     const wxChar *start = c_str();
1125     wxChar *end;
1126     *val = wxStrtod(start, &end);
1127
1128     // return TRUE only if scan was stopped by the terminating NUL and if the
1129     // string was not empty to start with
1130     return !*end && (end != start);
1131 }
1132
1133 // ---------------------------------------------------------------------------
1134 // formatted output
1135 // ---------------------------------------------------------------------------
1136
1137 /* static */
1138 wxString wxString::Format(const wxChar *pszFormat, ...)
1139 {
1140     va_list argptr;
1141     va_start(argptr, pszFormat);
1142
1143     wxString s;
1144     s.PrintfV(pszFormat, argptr);
1145
1146     va_end(argptr);
1147
1148     return s;
1149 }
1150
1151 /* static */
1152 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1153 {
1154     wxString s;
1155     s.PrintfV(pszFormat, argptr);
1156     return s;
1157 }
1158
1159 int wxString::Printf(const wxChar *pszFormat, ...)
1160 {
1161   va_list argptr;
1162   va_start(argptr, pszFormat);
1163
1164   int iLen = PrintfV(pszFormat, argptr);
1165
1166   va_end(argptr);
1167
1168   return iLen;
1169 }
1170
1171 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1172 {
1173 #if wxUSE_EXPERIMENTAL_PRINTF
1174   // the new implementation
1175
1176   // buffer to avoid dynamic memory allocation each time for small strings
1177   char szScratch[1024];
1178
1179   Reinit();
1180   for (size_t n = 0; pszFormat[n]; n++)
1181     if (pszFormat[n] == wxT('%')) {
1182       static char s_szFlags[256] = "%";
1183       size_t flagofs = 1;
1184       bool adj_left = FALSE, in_prec = FALSE,
1185            prec_dot = FALSE, done = FALSE;
1186       int ilen = 0;
1187       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1188       do {
1189 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1190         switch (pszFormat[++n]) {
1191         case wxT('\0'):
1192           done = TRUE;
1193           break;
1194         case wxT('%'):
1195           *this += wxT('%');
1196           done = TRUE;
1197           break;
1198         case wxT('#'):
1199         case wxT('0'):
1200         case wxT(' '):
1201         case wxT('+'):
1202         case wxT('\''):
1203           CHECK_PREC
1204           s_szFlags[flagofs++] = pszFormat[n];
1205           break;
1206         case wxT('-'):
1207           CHECK_PREC
1208           adj_left = TRUE;
1209           s_szFlags[flagofs++] = pszFormat[n];
1210           break;
1211         case wxT('.'):
1212           CHECK_PREC
1213           in_prec = TRUE;
1214           prec_dot = FALSE;
1215           max_width = 0;
1216           // dot will be auto-added to s_szFlags if non-negative number follows
1217           break;
1218         case wxT('h'):
1219           ilen = -1;
1220           CHECK_PREC
1221           s_szFlags[flagofs++] = pszFormat[n];
1222           break;
1223         case wxT('l'):
1224           ilen = 1;
1225           CHECK_PREC
1226           s_szFlags[flagofs++] = pszFormat[n];
1227           break;
1228         case wxT('q'):
1229         case wxT('L'):
1230           ilen = 2;
1231           CHECK_PREC
1232           s_szFlags[flagofs++] = pszFormat[n];
1233           break;
1234         case wxT('Z'):
1235           ilen = 3;
1236           CHECK_PREC
1237           s_szFlags[flagofs++] = pszFormat[n];
1238           break;
1239         case wxT('*'):
1240           {
1241             int len = va_arg(argptr, int);
1242             if (in_prec) {
1243               if (len<0) break;
1244               CHECK_PREC
1245               max_width = len;
1246             } else {
1247               if (len<0) {
1248                 adj_left = !adj_left;
1249                 s_szFlags[flagofs++] = '-';
1250                 len = -len;
1251               }
1252               min_width = len;
1253             }
1254             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1255           }
1256           break;
1257         case wxT('1'): case wxT('2'): case wxT('3'):
1258         case wxT('4'): case wxT('5'): case wxT('6'):
1259         case wxT('7'): case wxT('8'): case wxT('9'):
1260           {
1261             int len = 0;
1262             CHECK_PREC
1263             while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) {
1264               s_szFlags[flagofs++] = pszFormat[n];
1265               len = len*10 + (pszFormat[n] - wxT('0'));
1266               n++;
1267             }
1268             if (in_prec) max_width = len;
1269             else min_width = len;
1270             n--; // the main loop pre-increments n again
1271           }
1272           break;
1273         case wxT('d'):
1274         case wxT('i'):
1275         case wxT('o'):
1276         case wxT('u'):
1277         case wxT('x'):
1278         case wxT('X'):
1279           CHECK_PREC
1280           s_szFlags[flagofs++] = pszFormat[n];
1281           s_szFlags[flagofs] = '\0';
1282           if (ilen == 0 ) {
1283             int val = va_arg(argptr, int);
1284             ::sprintf(szScratch, s_szFlags, val);
1285           }
1286           else if (ilen == -1) {
1287             short int val = va_arg(argptr, short int);
1288             ::sprintf(szScratch, s_szFlags, val);
1289           }
1290           else if (ilen == 1) {
1291             long int val = va_arg(argptr, long int);
1292             ::sprintf(szScratch, s_szFlags, val);
1293           }
1294           else if (ilen == 2) {
1295 #if SIZEOF_LONG_LONG
1296             long long int val = va_arg(argptr, long long int);
1297             ::sprintf(szScratch, s_szFlags, val);
1298 #else
1299             long int val = va_arg(argptr, long int);
1300             ::sprintf(szScratch, s_szFlags, val);
1301 #endif
1302           }
1303           else if (ilen == 3) {
1304             size_t val = va_arg(argptr, size_t);
1305             ::sprintf(szScratch, s_szFlags, val);
1306           }
1307           *this += wxString(szScratch);
1308           done = TRUE;
1309           break;
1310         case wxT('e'):
1311         case wxT('E'):
1312         case wxT('f'):
1313         case wxT('g'):
1314         case wxT('G'):
1315           CHECK_PREC
1316           s_szFlags[flagofs++] = pszFormat[n];
1317           s_szFlags[flagofs] = '\0';
1318           if (ilen == 2) {
1319             long double val = va_arg(argptr, long double);
1320             ::sprintf(szScratch, s_szFlags, val);
1321           } else {
1322             double val = va_arg(argptr, double);
1323             ::sprintf(szScratch, s_szFlags, val);
1324           }
1325           *this += wxString(szScratch);
1326           done = TRUE;
1327           break;
1328         case wxT('p'):
1329           {
1330             void *val = va_arg(argptr, void *);
1331             CHECK_PREC
1332             s_szFlags[flagofs++] = pszFormat[n];
1333             s_szFlags[flagofs] = '\0';
1334             ::sprintf(szScratch, s_szFlags, val);
1335             *this += wxString(szScratch);
1336             done = TRUE;
1337           }
1338           break;
1339         case wxT('c'):
1340           {
1341             wxChar val = va_arg(argptr, int);
1342             // we don't need to honor padding here, do we?
1343             *this += val;
1344             done = TRUE;
1345           }
1346           break;
1347         case wxT('s'):
1348           if (ilen == -1) {
1349             // wx extension: we'll let %hs mean non-Unicode strings
1350             char *val = va_arg(argptr, char *);
1351 #if wxUSE_UNICODE
1352             // ASCII->Unicode constructor handles max_width right
1353             wxString s(val, wxConvLibc, max_width);
1354 #else
1355             size_t len = wxSTRING_MAXLEN;
1356             if (val) {
1357               for (len = 0; val[len] && (len<max_width); len++);
1358             } else val = wxT("(null)");
1359             wxString s(val, len);
1360 #endif
1361             if (s.Len() < min_width)
1362               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1363             *this += s;
1364           } else {
1365             wxChar *val = va_arg(argptr, wxChar *);
1366             size_t len = wxSTRING_MAXLEN;
1367             if (val) {
1368               for (len = 0; val[len] && (len<max_width); len++);
1369             } else val = wxT("(null)");
1370             wxString s(val, len);
1371             if (s.Len() < min_width)
1372               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1373             *this += s;
1374           }
1375           done = TRUE;
1376           break;
1377         case wxT('n'):
1378           if (ilen == 0) {
1379             int *val = va_arg(argptr, int *);
1380             *val = Len();
1381           }
1382           else if (ilen == -1) {
1383             short int *val = va_arg(argptr, short int *);
1384             *val = Len();
1385           }
1386           else if (ilen >= 1) {
1387             long int *val = va_arg(argptr, long int *);
1388             *val = Len();
1389           }
1390           done = TRUE;
1391           break;
1392         default:
1393           if (wxIsalpha(pszFormat[n]))
1394             // probably some flag not taken care of here yet
1395             s_szFlags[flagofs++] = pszFormat[n];
1396           else {
1397             // bad format
1398             *this += wxT('%'); // just to pass the glibc tst-printf.c
1399             n--;
1400             done = TRUE;
1401           }
1402           break;
1403         }
1404 #undef CHECK_PREC
1405       } while (!done);
1406     } else *this += pszFormat[n];
1407
1408 #else
1409   // buffer to avoid dynamic memory allocation each time for small strings
1410   char szScratch[1024];
1411
1412   // NB: wxVsnprintf() may return either less than the buffer size or -1 if
1413   //     there is not enough place depending on implementation
1414   int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), (char *)pszFormat, argptr);
1415   if ( iLen != -1 ) {
1416     // the whole string is in szScratch
1417     *this = szScratch;
1418   }
1419   else {
1420       bool outOfMemory = FALSE;
1421       int size = 2*WXSIZEOF(szScratch);
1422       while ( !outOfMemory ) {
1423           char *buf = GetWriteBuf(size);
1424           if ( buf )
1425             iLen = wxVsnprintfA(buf, size, pszFormat, argptr);
1426           else
1427             outOfMemory = TRUE;
1428
1429           UngetWriteBuf();
1430
1431           if ( iLen != -1 ) {
1432               // ok, there was enough space
1433               break;
1434           }
1435
1436           // still not enough, double it again
1437           size *= 2;
1438       }
1439
1440       if ( outOfMemory ) {
1441           // out of memory
1442           return -1;
1443       }
1444   }
1445 #endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF
1446
1447   return Len();
1448 }
1449
1450 // ----------------------------------------------------------------------------
1451 // misc other operations
1452 // ----------------------------------------------------------------------------
1453
1454 // returns TRUE if the string matches the pattern which may contain '*' and
1455 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1456 // of them)
1457 bool wxString::Matches(const wxChar *pszMask) const
1458 {
1459   // check char by char
1460   const wxChar *pszTxt;
1461   for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1462     switch ( *pszMask ) {
1463       case wxT('?'):
1464         if ( *pszTxt == wxT('\0') )
1465           return FALSE;
1466
1467         // pszText and pszMask will be incremented in the loop statement
1468
1469         break;
1470
1471       case wxT('*'):
1472         {
1473           // ignore special chars immediately following this one
1474           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1475             pszMask++;
1476
1477           // if there is nothing more, match
1478           if ( *pszMask == wxT('\0') )
1479             return TRUE;
1480
1481           // are there any other metacharacters in the mask?
1482           size_t uiLenMask;
1483           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1484
1485           if ( pEndMask != NULL ) {
1486             // we have to match the string between two metachars
1487             uiLenMask = pEndMask - pszMask;
1488           }
1489           else {
1490             // we have to match the remainder of the string
1491             uiLenMask = wxStrlen(pszMask);
1492           }
1493
1494           wxString strToMatch(pszMask, uiLenMask);
1495           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1496           if ( pMatch == NULL )
1497             return FALSE;
1498
1499           // -1 to compensate "++" in the loop
1500           pszTxt = pMatch + uiLenMask - 1;
1501           pszMask += uiLenMask - 1;
1502         }
1503         break;
1504
1505       default:
1506         if ( *pszMask != *pszTxt )
1507           return FALSE;
1508         break;
1509     }
1510   }
1511
1512   // match only if nothing left
1513   return *pszTxt == wxT('\0');
1514 }
1515
1516 // Count the number of chars
1517 int wxString::Freq(wxChar ch) const
1518 {
1519     int count = 0;
1520     int len = Len();
1521     for (int i = 0; i < len; i++)
1522     {
1523         if (GetChar(i) == ch)
1524             count ++;
1525     }
1526     return count;
1527 }
1528
1529 // convert to upper case, return the copy of the string
1530 wxString wxString::Upper() const
1531 { wxString s(*this); return s.MakeUpper(); }
1532
1533 // convert to lower case, return the copy of the string
1534 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1535
1536 int wxString::sprintf(const wxChar *pszFormat, ...)
1537   {
1538     va_list argptr;
1539     va_start(argptr, pszFormat);
1540     int iLen = PrintfV(pszFormat, argptr);
1541     va_end(argptr);
1542     return iLen;
1543   }
1544
1545 // ---------------------------------------------------------------------------
1546 // standard C++ library string functions
1547 // ---------------------------------------------------------------------------
1548
1549 #ifdef  wxSTD_STRING_COMPATIBILITY
1550
1551 void wxString::resize(size_t nSize, wxChar ch)
1552 {
1553     size_t len = length();
1554
1555     if ( nSize < len )
1556     {
1557         Truncate(nSize);
1558     }
1559     else if ( nSize > len )
1560     {
1561         *this += wxString(ch, len - nSize);
1562     }
1563     //else: we have exactly the specified length, nothing to do
1564 }
1565
1566 void wxString::swap(wxString& str)
1567 {
1568     // this is slightly less efficient than fiddling with m_pchData directly,
1569     // but it is still quite efficient as we don't copy the string here because
1570     // ref count always stays positive
1571     wxString tmp = str;
1572     str = *this;
1573     *this = str;
1574 }
1575
1576 wxString& wxString::insert(size_t nPos, const wxString& str)
1577 {
1578   wxASSERT( str.GetStringData()->IsValid() );
1579   wxASSERT( nPos <= Len() );
1580
1581   if ( !str.IsEmpty() ) {
1582     wxString strTmp;
1583     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1584     wxStrncpy(pc, c_str(), nPos);
1585     wxStrcpy(pc + nPos, str);
1586     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1587     strTmp.UngetWriteBuf();
1588     *this = strTmp;
1589   }
1590
1591   return *this;
1592 }
1593
1594 size_t wxString::find(const wxString& str, size_t nStart) const
1595 {
1596   wxASSERT( str.GetStringData()->IsValid() );
1597   wxASSERT( nStart <= Len() );
1598
1599   const wxChar *p = wxStrstr(c_str() + nStart, str);
1600
1601   return p == NULL ? npos : p - c_str();
1602 }
1603
1604 // VC++ 1.5 can't cope with the default argument in the header.
1605 #if !defined(__VISUALC__) || defined(__WIN32__)
1606 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1607 {
1608   return find(wxString(sz, n), nStart);
1609 }
1610 #endif // VC++ 1.5
1611
1612 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1613 #if !defined(__BORLANDC__)
1614 size_t wxString::find(wxChar ch, size_t nStart) const
1615 {
1616   wxASSERT( nStart <= Len() );
1617
1618   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1619
1620   return p == NULL ? npos : p - c_str();
1621 }
1622 #endif
1623
1624 size_t wxString::rfind(const wxString& str, size_t nStart) const
1625 {
1626   wxASSERT( str.GetStringData()->IsValid() );
1627   wxASSERT( nStart <= Len() );
1628
1629   // TODO could be made much quicker than that
1630   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1631   while ( p >= c_str() + str.Len() ) {
1632     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1633       return p - str.Len() - c_str();
1634     p--;
1635   }
1636
1637   return npos;
1638 }
1639
1640 // VC++ 1.5 can't cope with the default argument in the header.
1641 #if !defined(__VISUALC__) || defined(__WIN32__)
1642 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1643 {
1644     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1645 }
1646
1647 size_t wxString::rfind(wxChar ch, size_t nStart) const
1648 {
1649     if ( nStart == npos )
1650     {
1651         nStart = Len();
1652     }
1653     else
1654     {
1655         wxASSERT( nStart <= Len() );
1656     }
1657
1658     const wxChar *p = wxStrrchr(c_str(), ch);
1659
1660     if ( p == NULL )
1661         return npos;
1662
1663     size_t result = p - c_str();
1664     return ( result > nStart ) ? npos : result;
1665 }
1666 #endif // VC++ 1.5
1667
1668 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1669 {
1670     const wxChar *start = c_str() + nStart;
1671     const wxChar *firstOf = wxStrpbrk(start, sz);
1672     if ( firstOf )
1673         return firstOf - c_str();
1674     else
1675         return npos;
1676 }
1677
1678 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1679 {
1680     if ( nStart == npos )
1681     {
1682         nStart = Len();
1683     }
1684     else
1685     {
1686         wxASSERT( nStart <= Len() );
1687     }
1688
1689     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1690     {
1691         if ( wxStrchr(sz, *p) )
1692             return p - c_str();
1693     }
1694
1695     return npos;
1696 }
1697
1698 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1699 {
1700     if ( nStart == npos )
1701     {
1702         nStart = Len();
1703     }
1704     else
1705     {
1706         wxASSERT( nStart <= Len() );
1707     }
1708
1709     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1710     if ( nAccept >= length() - nStart )
1711         return npos;
1712     else
1713         return nAccept;
1714 }
1715
1716 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1717 {
1718     wxASSERT( nStart <= Len() );
1719
1720     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1721     {
1722         if ( *p != ch )
1723             return p - c_str();
1724     }
1725
1726     return npos;
1727 }
1728
1729 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1730 {
1731     if ( nStart == npos )
1732     {
1733         nStart = Len();
1734     }
1735     else
1736     {
1737         wxASSERT( nStart <= Len() );
1738     }
1739
1740     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1741     {
1742         if ( !wxStrchr(sz, *p) )
1743             return p - c_str();
1744     }
1745
1746     return npos;
1747 }
1748
1749 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1750 {
1751     if ( nStart == npos )
1752     {
1753         nStart = Len();
1754     }
1755     else
1756     {
1757         wxASSERT( nStart <= Len() );
1758     }
1759
1760     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1761     {
1762         if ( *p != ch )
1763             return p - c_str();
1764     }
1765
1766     return npos;
1767 }
1768
1769 wxString& wxString::erase(size_t nStart, size_t nLen)
1770 {
1771   wxString strTmp(c_str(), nStart);
1772   if ( nLen != npos ) {
1773     wxASSERT( nStart + nLen <= Len() );
1774
1775     strTmp.append(c_str() + nStart + nLen);
1776   }
1777
1778   *this = strTmp;
1779   return *this;
1780 }
1781
1782 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1783 {
1784   wxASSERT_MSG( nStart + nLen <= Len(),
1785                 _T("index out of bounds in wxString::replace") );
1786
1787   wxString strTmp;
1788   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1789
1790   if ( nStart != 0 )
1791     strTmp.append(c_str(), nStart);
1792   strTmp << sz << c_str() + nStart + nLen;
1793
1794   *this = strTmp;
1795   return *this;
1796 }
1797
1798 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1799 {
1800   return replace(nStart, nLen, wxString(ch, nCount));
1801 }
1802
1803 wxString& wxString::replace(size_t nStart, size_t nLen,
1804                             const wxString& str, size_t nStart2, size_t nLen2)
1805 {
1806   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1807 }
1808
1809 wxString& wxString::replace(size_t nStart, size_t nLen,
1810                         const wxChar* sz, size_t nCount)
1811 {
1812   return replace(nStart, nLen, wxString(sz, nCount));
1813 }
1814
1815 #endif  //std::string compatibility
1816
1817 // ============================================================================
1818 // ArrayString
1819 // ============================================================================
1820
1821 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1822 #define   ARRAY_MAXSIZE_INCREMENT       4096
1823 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1824   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1825 #endif
1826
1827 #define   STRING(p)   ((wxString *)(&(p)))
1828
1829 // ctor
1830 wxArrayString::wxArrayString(bool autoSort)
1831 {
1832   m_nSize  =
1833   m_nCount = 0;
1834   m_pItems = (wxChar **) NULL;
1835   m_autoSort = autoSort;
1836 }
1837
1838 // copy ctor
1839 wxArrayString::wxArrayString(const wxArrayString& src)
1840 {
1841   m_nSize  =
1842   m_nCount = 0;
1843   m_pItems = (wxChar **) NULL;
1844   m_autoSort = src.m_autoSort;
1845
1846   *this = src;
1847 }
1848
1849 // assignment operator
1850 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1851 {
1852   if ( m_nSize > 0 )
1853     Clear();
1854
1855   Copy(src);
1856
1857   return *this;
1858 }
1859
1860 void wxArrayString::Copy(const wxArrayString& src)
1861 {
1862   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1863     Alloc(src.m_nCount);
1864
1865   for ( size_t n = 0; n < src.m_nCount; n++ )
1866     Add(src[n]);
1867 }
1868
1869 // grow the array
1870 void wxArrayString::Grow()
1871 {
1872   // only do it if no more place
1873   if( m_nCount == m_nSize ) {
1874     if( m_nSize == 0 ) {
1875       // was empty, alloc some memory
1876       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1877       m_pItems = new wxChar *[m_nSize];
1878     }
1879     else {
1880       // otherwise when it's called for the first time, nIncrement would be 0
1881       // and the array would never be expanded
1882 #if defined(__VISAGECPP__) && defined(__WXDEBUG__)
1883       int array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1884       wxASSERT( array_size != 0 );
1885 #else
1886       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1887 #endif
1888
1889       // add 50% but not too much
1890       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1891                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1892       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1893         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1894       m_nSize += nIncrement;
1895       wxChar **pNew = new wxChar *[m_nSize];
1896
1897       // copy data to new location
1898       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1899
1900       // delete old memory (but do not release the strings!)
1901       wxDELETEA(m_pItems);
1902
1903       m_pItems = pNew;
1904     }
1905   }
1906 }
1907
1908 void wxArrayString::Free()
1909 {
1910   for ( size_t n = 0; n < m_nCount; n++ ) {
1911     STRING(m_pItems[n])->GetStringData()->Unlock();
1912   }
1913 }
1914
1915 // deletes all the strings from the list
1916 void wxArrayString::Empty()
1917 {
1918   Free();
1919
1920   m_nCount = 0;
1921 }
1922
1923 // as Empty, but also frees memory
1924 void wxArrayString::Clear()
1925 {
1926   Free();
1927
1928   m_nSize  =
1929   m_nCount = 0;
1930
1931   wxDELETEA(m_pItems);
1932 }
1933
1934 // dtor
1935 wxArrayString::~wxArrayString()
1936 {
1937   Free();
1938
1939   wxDELETEA(m_pItems);
1940 }
1941
1942 // pre-allocates memory (frees the previous data!)
1943 void wxArrayString::Alloc(size_t nSize)
1944 {
1945   wxASSERT( nSize > 0 );
1946
1947   // only if old buffer was not big enough
1948   if ( nSize > m_nSize ) {
1949     Free();
1950     wxDELETEA(m_pItems);
1951     m_pItems = new wxChar *[nSize];
1952     m_nSize  = nSize;
1953   }
1954
1955   m_nCount = 0;
1956 }
1957
1958 // minimizes the memory usage by freeing unused memory
1959 void wxArrayString::Shrink()
1960 {
1961   // only do it if we have some memory to free
1962   if( m_nCount < m_nSize ) {
1963     // allocates exactly as much memory as we need
1964     wxChar **pNew = new wxChar *[m_nCount];
1965
1966     // copy data to new location
1967     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1968     delete [] m_pItems;
1969     m_pItems = pNew;
1970   }
1971 }
1972
1973 // searches the array for an item (forward or backwards)
1974 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1975 {
1976   if ( m_autoSort ) {
1977     // use binary search in the sorted array
1978     wxASSERT_MSG( bCase && !bFromEnd,
1979                   wxT("search parameters ignored for auto sorted array") );
1980
1981     size_t i,
1982            lo = 0,
1983            hi = m_nCount;
1984     int res;
1985     while ( lo < hi ) {
1986       i = (lo + hi)/2;
1987
1988       res = wxStrcmp(sz, m_pItems[i]);
1989       if ( res < 0 )
1990         hi = i;
1991       else if ( res > 0 )
1992         lo = i + 1;
1993       else
1994         return i;
1995     }
1996
1997     return wxNOT_FOUND;
1998   }
1999   else {
2000     // use linear search in unsorted array
2001     if ( bFromEnd ) {
2002       if ( m_nCount > 0 ) {
2003         size_t ui = m_nCount;
2004         do {
2005           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
2006             return ui;
2007         }
2008         while ( ui != 0 );
2009       }
2010     }
2011     else {
2012       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2013         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
2014           return ui;
2015       }
2016     }
2017   }
2018
2019   return wxNOT_FOUND;
2020 }
2021
2022 // add item at the end
2023 size_t wxArrayString::Add(const wxString& str)
2024 {
2025   if ( m_autoSort ) {
2026     // insert the string at the correct position to keep the array sorted
2027     size_t i,
2028            lo = 0,
2029            hi = m_nCount;
2030     int res;
2031     while ( lo < hi ) {
2032       i = (lo + hi)/2;
2033
2034       res = wxStrcmp(str, m_pItems[i]);
2035       if ( res < 0 )
2036         hi = i;
2037       else if ( res > 0 )
2038         lo = i + 1;
2039       else {
2040         lo = hi = i;
2041         break;
2042       }
2043     }
2044
2045     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2046
2047     Insert(str, lo);
2048
2049     return (size_t)lo;
2050   }
2051   else {
2052     wxASSERT( str.GetStringData()->IsValid() );
2053
2054     Grow();
2055
2056     // the string data must not be deleted!
2057     str.GetStringData()->Lock();
2058
2059     // just append
2060     m_pItems[m_nCount] = (wxChar *)str.c_str(); // const_cast
2061
2062     return m_nCount++;
2063   }
2064 }
2065
2066 // add item at the given position
2067 void wxArrayString::Insert(const wxString& str, size_t nIndex)
2068 {
2069   wxASSERT( str.GetStringData()->IsValid() );
2070
2071   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2072
2073   Grow();
2074
2075   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
2076           (m_nCount - nIndex)*sizeof(wxChar *));
2077
2078   str.GetStringData()->Lock();
2079   m_pItems[nIndex] = (wxChar *)str.c_str();
2080
2081   m_nCount++;
2082 }
2083
2084 // removes item from array (by index)
2085 void wxArrayString::Remove(size_t nIndex)
2086 {
2087   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Remove") );
2088
2089   // release our lock
2090   Item(nIndex).GetStringData()->Unlock();
2091
2092   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
2093           (m_nCount - nIndex - 1)*sizeof(wxChar *));
2094   m_nCount--;
2095 }
2096
2097 // removes item from array (by value)
2098 void wxArrayString::Remove(const wxChar *sz)
2099 {
2100   int iIndex = Index(sz);
2101
2102   wxCHECK_RET( iIndex != wxNOT_FOUND,
2103                wxT("removing inexistent element in wxArrayString::Remove") );
2104
2105   Remove(iIndex);
2106 }
2107
2108 // ----------------------------------------------------------------------------
2109 // sorting
2110 // ----------------------------------------------------------------------------
2111
2112 // we can only sort one array at a time with the quick-sort based
2113 // implementation
2114 #if wxUSE_THREADS
2115   // need a critical section to protect access to gs_compareFunction and
2116   // gs_sortAscending variables
2117   static wxCriticalSection *gs_critsectStringSort = NULL;
2118
2119   // call this before the value of the global sort vars is changed/after
2120   // you're finished with them
2121   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2122                            gs_critsectStringSort = new wxCriticalSection;     \
2123                            gs_critsectStringSort->Enter()
2124   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2125                            delete gs_critsectStringSort;                      \
2126                            gs_critsectStringSort = NULL
2127 #else // !threads
2128   #define START_SORT()
2129   #define END_SORT()
2130 #endif // wxUSE_THREADS
2131
2132 // function to use for string comparaison
2133 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2134
2135 // if we don't use the compare function, this flag tells us if we sort the
2136 // array in ascending or descending order
2137 static bool gs_sortAscending = TRUE;
2138
2139 // function which is called by quick sort
2140 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
2141 {
2142   wxString *strFirst = (wxString *)first;
2143   wxString *strSecond = (wxString *)second;
2144
2145   if ( gs_compareFunction ) {
2146     return gs_compareFunction(*strFirst, *strSecond);
2147   }
2148   else {
2149     // maybe we should use wxStrcoll
2150     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2151
2152     return gs_sortAscending ? result : -result;
2153   }
2154 }
2155
2156 // sort array elements using passed comparaison function
2157 void wxArrayString::Sort(CompareFunction compareFunction)
2158 {
2159   START_SORT();
2160
2161   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2162   gs_compareFunction = compareFunction;
2163
2164   DoSort();
2165
2166   // reset it to NULL so that Sort(bool) will work the next time
2167   gs_compareFunction = NULL;
2168
2169   END_SORT();
2170 }
2171
2172 void wxArrayString::Sort(bool reverseOrder)
2173 {
2174   START_SORT();
2175
2176   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2177   gs_sortAscending = !reverseOrder;
2178
2179   DoSort();
2180
2181   END_SORT();
2182 }
2183
2184 void wxArrayString::DoSort()
2185 {
2186   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2187
2188   // just sort the pointers using qsort() - of course it only works because
2189   // wxString() *is* a pointer to its data
2190   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2191 }
2192
2193 bool wxArrayString::operator==(const wxArrayString& a) const
2194 {
2195     if ( m_nCount != a.m_nCount )
2196         return FALSE;
2197
2198     for ( size_t n = 0; n < m_nCount; n++ )
2199     {
2200         if ( Item(n) != a[n] )
2201             return FALSE;
2202     }
2203
2204     return TRUE;
2205 }
2206