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 // some compilers (VC++ 6.0 not to name them) return TRUE for a call to
 990 // isspace('ê') in the C locale which seems to be broken to me, but we have to
 991 // live with this by checking that the character is a 7 bit one - even if this
 992 // may fail to detect some spaces (I don't know if Unicode doesn't have
 993 // space-like symbols somewhere except in the first 128 chars), it is arguably
 994 // still better than trimming away accented letters
 995 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
 996
 997 // trims spaces (in the sense of isspace) from left or right side
 998 wxString& wxString::Trim(bool bFromRight)
 999 {
1000   // first check if we're going to modify the string at all
1001   if ( !IsEmpty() &&
1002        (
1003         (bFromRight && wxSafeIsspace(GetChar(Len() - 1))) ||
1004         (!bFromRight && wxSafeIsspace(GetChar(0u)))
1005        )
1006      )
1007   {
1008     // ok, there is at least one space to trim
1009     CopyBeforeWrite();
1010
1011     if ( bFromRight )
1012     {
1013       // find last non-space character
1014       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
1015       while ( wxSafeIsspace(*psz) && (psz >= m_pchData) )
1016         psz--;
1017
1018       // truncate at trailing space start
1019       *++psz = wxT('\0');
1020       GetStringData()->nDataLength = psz - m_pchData;
1021     }
1022     else
1023     {
1024       // find first non-space character
1025       const wxChar *psz = m_pchData;
1026       while ( wxSafeIsspace(*psz) )
1027         psz++;
1028
1029       // fix up data and length
1030       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
1031       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
1032       GetStringData()->nDataLength = nDataLength;
1033     }
1034   }
1035
1036   return *this;
1037 }
1038
1039 // adds nCount characters chPad to the string from either side
1040 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
1041 {
1042   wxString s(chPad, nCount);
1043
1044   if ( bFromRight )
1045     *this += s;
1046   else
1047   {
1048     s += *this;
1049     *this = s;
1050   }
1051
1052   return *this;
1053 }
1054
1055 // truncate the string
1056 wxString& wxString::Truncate(size_t uiLen)
1057 {
1058   if ( uiLen < Len() ) {
1059     CopyBeforeWrite();
1060
1061     *(m_pchData + uiLen) = wxT('\0');
1062     GetStringData()->nDataLength = uiLen;
1063   }
1064   //else: nothing to do, string is already short enough
1065
1066   return *this;
1067 }
1068
1069 // ---------------------------------------------------------------------------
1070 // finding (return wxNOT_FOUND if not found and index otherwise)
1071 // ---------------------------------------------------------------------------
1072
1073 // find a character
1074 int wxString::Find(wxChar ch, bool bFromEnd) const
1075 {
1076   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1077
1078   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1079 }
1080
1081 // find a sub-string (like strstr)
1082 int wxString::Find(const wxChar *pszSub) const
1083 {
1084   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1085
1086   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1087 }
1088
1089 // ----------------------------------------------------------------------------
1090 // conversion to numbers
1091 // ----------------------------------------------------------------------------
1092
1093 bool wxString::ToLong(long *val) const
1094 {
1095     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1096
1097     const wxChar *start = c_str();
1098     wxChar *end;
1099     *val = wxStrtol(start, &end, 10);
1100
1101     // return TRUE only if scan was stopped by the terminating NUL and if the
1102     // string was not empty to start with
1103     return !*end && (end != start);
1104 }
1105
1106 bool wxString::ToULong(unsigned long *val) const
1107 {
1108     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1109
1110     const wxChar *start = c_str();
1111     wxChar *end;
1112     *val = wxStrtoul(start, &end, 10);
1113
1114     // return TRUE only if scan was stopped by the terminating NUL and if the
1115     // string was not empty to start with
1116     return !*end && (end != start);
1117 }
1118
1119 bool wxString::ToDouble(double *val) const
1120 {
1121     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1122
1123     const wxChar *start = c_str();
1124     wxChar *end;
1125     *val = wxStrtod(start, &end);
1126
1127     // return TRUE only if scan was stopped by the terminating NUL and if the
1128     // string was not empty to start with
1129     return !*end && (end != start);
1130 }
1131
1132 // ---------------------------------------------------------------------------
1133 // formatted output
1134 // ---------------------------------------------------------------------------
1135
1136 /* static */
1137 wxString wxString::Format(const wxChar *pszFormat, ...)
1138 {
1139     va_list argptr;
1140     va_start(argptr, pszFormat);
1141
1142     wxString s;
1143     s.PrintfV(pszFormat, argptr);
1144
1145     va_end(argptr);
1146
1147     return s;
1148 }
1149
1150 /* static */
1151 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1152 {
1153     wxString s;
1154     s.Printf(pszFormat, argptr);
1155     return s;
1156 }
1157
1158 int wxString::Printf(const wxChar *pszFormat, ...)
1159 {
1160   va_list argptr;
1161   va_start(argptr, pszFormat);
1162
1163   int iLen = PrintfV(pszFormat, argptr);
1164
1165   va_end(argptr);
1166
1167   return iLen;
1168 }
1169
1170 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1171 {
1172 #if wxUSE_EXPERIMENTAL_PRINTF
1173   // the new implementation
1174
1175   // buffer to avoid dynamic memory allocation each time for small strings
1176   char szScratch[1024];
1177
1178   Reinit();
1179   for (size_t n = 0; pszFormat[n]; n++)
1180     if (pszFormat[n] == wxT('%')) {
1181       static char s_szFlags[256] = "%";
1182       size_t flagofs = 1;
1183       bool adj_left = FALSE, in_prec = FALSE,
1184            prec_dot = FALSE, done = FALSE;
1185       int ilen = 0;
1186       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1187       do {
1188 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1189         switch (pszFormat[++n]) {
1190         case wxT('\0'):
1191           done = TRUE;
1192           break;
1193         case wxT('%'):
1194           *this += wxT('%');
1195           done = TRUE;
1196           break;
1197         case wxT('#'):
1198         case wxT('0'):
1199         case wxT(' '):
1200         case wxT('+'):
1201         case wxT('\''):
1202           CHECK_PREC
1203           s_szFlags[flagofs++] = pszFormat[n];
1204           break;
1205         case wxT('-'):
1206           CHECK_PREC
1207           adj_left = TRUE;
1208           s_szFlags[flagofs++] = pszFormat[n];
1209           break;
1210         case wxT('.'):
1211           CHECK_PREC
1212           in_prec = TRUE;
1213           prec_dot = FALSE;
1214           max_width = 0;
1215           // dot will be auto-added to s_szFlags if non-negative number follows
1216           break;
1217         case wxT('h'):
1218           ilen = -1;
1219           CHECK_PREC
1220           s_szFlags[flagofs++] = pszFormat[n];
1221           break;
1222         case wxT('l'):
1223           ilen = 1;
1224           CHECK_PREC
1225           s_szFlags[flagofs++] = pszFormat[n];
1226           break;
1227         case wxT('q'):
1228         case wxT('L'):
1229           ilen = 2;
1230           CHECK_PREC
1231           s_szFlags[flagofs++] = pszFormat[n];
1232           break;
1233         case wxT('Z'):
1234           ilen = 3;
1235           CHECK_PREC
1236           s_szFlags[flagofs++] = pszFormat[n];
1237           break;
1238         case wxT('*'):
1239           {
1240             int len = va_arg(argptr, int);
1241             if (in_prec) {
1242               if (len<0) break;
1243               CHECK_PREC
1244               max_width = len;
1245             } else {
1246               if (len<0) {
1247                 adj_left = !adj_left;
1248                 s_szFlags[flagofs++] = '-';
1249                 len = -len;
1250               }
1251               min_width = len;
1252             }
1253             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1254           }
1255           break;
1256         case wxT('1'): case wxT('2'): case wxT('3'):
1257         case wxT('4'): case wxT('5'): case wxT('6'):
1258         case wxT('7'): case wxT('8'): case wxT('9'):
1259           {
1260             int len = 0;
1261             CHECK_PREC
1262             while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) {
1263               s_szFlags[flagofs++] = pszFormat[n];
1264               len = len*10 + (pszFormat[n] - wxT('0'));
1265               n++;
1266             }
1267             if (in_prec) max_width = len;
1268             else min_width = len;
1269             n--; // the main loop pre-increments n again
1270           }
1271           break;
1272         case wxT('d'):
1273         case wxT('i'):
1274         case wxT('o'):
1275         case wxT('u'):
1276         case wxT('x'):
1277         case wxT('X'):
1278           CHECK_PREC
1279           s_szFlags[flagofs++] = pszFormat[n];
1280           s_szFlags[flagofs] = '\0';
1281           if (ilen == 0 ) {
1282             int val = va_arg(argptr, int);
1283             ::sprintf(szScratch, s_szFlags, val);
1284           }
1285           else if (ilen == -1) {
1286             short int val = va_arg(argptr, short int);
1287             ::sprintf(szScratch, s_szFlags, val);
1288           }
1289           else if (ilen == 1) {
1290             long int val = va_arg(argptr, long int);
1291             ::sprintf(szScratch, s_szFlags, val);
1292           }
1293           else if (ilen == 2) {
1294 #if SIZEOF_LONG_LONG
1295             long long int val = va_arg(argptr, long long int);
1296             ::sprintf(szScratch, s_szFlags, val);
1297 #else
1298             long int val = va_arg(argptr, long int);
1299             ::sprintf(szScratch, s_szFlags, val);
1300 #endif
1301           }
1302           else if (ilen == 3) {
1303             size_t val = va_arg(argptr, size_t);
1304             ::sprintf(szScratch, s_szFlags, val);
1305           }
1306           *this += wxString(szScratch);
1307           done = TRUE;
1308           break;
1309         case wxT('e'):
1310         case wxT('E'):
1311         case wxT('f'):
1312         case wxT('g'):
1313         case wxT('G'):
1314           CHECK_PREC
1315           s_szFlags[flagofs++] = pszFormat[n];
1316           s_szFlags[flagofs] = '\0';
1317           if (ilen == 2) {
1318             long double val = va_arg(argptr, long double);
1319             ::sprintf(szScratch, s_szFlags, val);
1320           } else {
1321             double val = va_arg(argptr, double);
1322             ::sprintf(szScratch, s_szFlags, val);
1323           }
1324           *this += wxString(szScratch);
1325           done = TRUE;
1326           break;
1327         case wxT('p'):
1328           {
1329             void *val = va_arg(argptr, void *);
1330             CHECK_PREC
1331             s_szFlags[flagofs++] = pszFormat[n];
1332             s_szFlags[flagofs] = '\0';
1333             ::sprintf(szScratch, s_szFlags, val);
1334             *this += wxString(szScratch);
1335             done = TRUE;
1336           }
1337           break;
1338         case wxT('c'):
1339           {
1340             wxChar val = va_arg(argptr, int);
1341             // we don't need to honor padding here, do we?
1342             *this += val;
1343             done = TRUE;
1344           }
1345           break;
1346         case wxT('s'):
1347           if (ilen == -1) {
1348             // wx extension: we'll let %hs mean non-Unicode strings
1349             char *val = va_arg(argptr, char *);
1350 #if wxUSE_UNICODE
1351             // ASCII->Unicode constructor handles max_width right
1352             wxString s(val, wxConvLibc, max_width);
1353 #else
1354             size_t len = wxSTRING_MAXLEN;
1355             if (val) {
1356               for (len = 0; val[len] && (len<max_width); len++);
1357             } else val = wxT("(null)");
1358             wxString s(val, len);
1359 #endif
1360             if (s.Len() < min_width)
1361               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1362             *this += s;
1363           } else {
1364             wxChar *val = va_arg(argptr, wxChar *);
1365             size_t len = wxSTRING_MAXLEN;
1366             if (val) {
1367               for (len = 0; val[len] && (len<max_width); len++);
1368             } else val = wxT("(null)");
1369             wxString s(val, len);
1370             if (s.Len() < min_width)
1371               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1372             *this += s;
1373           }
1374           done = TRUE;
1375           break;
1376         case wxT('n'):
1377           if (ilen == 0) {
1378             int *val = va_arg(argptr, int *);
1379             *val = Len();
1380           }
1381           else if (ilen == -1) {
1382             short int *val = va_arg(argptr, short int *);
1383             *val = Len();
1384           }
1385           else if (ilen >= 1) {
1386             long int *val = va_arg(argptr, long int *);
1387             *val = Len();
1388           }
1389           done = TRUE;
1390           break;
1391         default:
1392           if (wxIsalpha(pszFormat[n]))
1393             // probably some flag not taken care of here yet
1394             s_szFlags[flagofs++] = pszFormat[n];
1395           else {
1396             // bad format
1397             *this += wxT('%'); // just to pass the glibc tst-printf.c
1398             n--;
1399             done = TRUE;
1400           }
1401           break;
1402         }
1403 #undef CHECK_PREC
1404       } while (!done);
1405     } else *this += pszFormat[n];
1406
1407 #else
1408   // buffer to avoid dynamic memory allocation each time for small strings
1409   char szScratch[1024];
1410
1411   // NB: wxVsnprintf() may return either less than the buffer size or -1 if
1412   //     there is not enough place depending on implementation
1413   int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), pszFormat, argptr);
1414   if ( iLen != -1 ) {
1415     // the whole string is in szScratch
1416     *this = szScratch;
1417   }
1418   else {
1419       bool outOfMemory = FALSE;
1420       int size = 2*WXSIZEOF(szScratch);
1421       while ( !outOfMemory ) {
1422           char *buf = GetWriteBuf(size);
1423           if ( buf )
1424             iLen = wxVsnprintfA(buf, size, pszFormat, argptr);
1425           else
1426             outOfMemory = TRUE;
1427
1428           UngetWriteBuf();
1429
1430           if ( iLen != -1 ) {
1431               // ok, there was enough space
1432               break;
1433           }
1434
1435           // still not enough, double it again
1436           size *= 2;
1437       }
1438
1439       if ( outOfMemory ) {
1440           // out of memory
1441           return -1;
1442       }
1443   }
1444 #endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF
1445
1446   return Len();
1447 }
1448
1449 // ----------------------------------------------------------------------------
1450 // misc other operations
1451 // ----------------------------------------------------------------------------
1452
1453 // returns TRUE if the string matches the pattern which may contain '*' and
1454 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1455 // of them)
1456 bool wxString::Matches(const wxChar *pszMask) const
1457 {
1458   // check char by char
1459   const wxChar *pszTxt;
1460   for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1461     switch ( *pszMask ) {
1462       case wxT('?'):
1463         if ( *pszTxt == wxT('\0') )
1464           return FALSE;
1465
1466         // pszText and pszMask will be incremented in the loop statement
1467
1468         break;
1469
1470       case wxT('*'):
1471         {
1472           // ignore special chars immediately following this one
1473           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1474             pszMask++;
1475
1476           // if there is nothing more, match
1477           if ( *pszMask == wxT('\0') )
1478             return TRUE;
1479
1480           // are there any other metacharacters in the mask?
1481           size_t uiLenMask;
1482           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1483
1484           if ( pEndMask != NULL ) {
1485             // we have to match the string between two metachars
1486             uiLenMask = pEndMask - pszMask;
1487           }
1488           else {
1489             // we have to match the remainder of the string
1490             uiLenMask = wxStrlen(pszMask);
1491           }
1492
1493           wxString strToMatch(pszMask, uiLenMask);
1494           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1495           if ( pMatch == NULL )
1496             return FALSE;
1497
1498           // -1 to compensate "++" in the loop
1499           pszTxt = pMatch + uiLenMask - 1;
1500           pszMask += uiLenMask - 1;
1501         }
1502         break;
1503
1504       default:
1505         if ( *pszMask != *pszTxt )
1506           return FALSE;
1507         break;
1508     }
1509   }
1510
1511   // match only if nothing left
1512   return *pszTxt == wxT('\0');
1513 }
1514
1515 // Count the number of chars
1516 int wxString::Freq(wxChar ch) const
1517 {
1518     int count = 0;
1519     int len = Len();
1520     for (int i = 0; i < len; i++)
1521     {
1522         if (GetChar(i) == ch)
1523             count ++;
1524     }
1525     return count;
1526 }
1527
1528 // convert to upper case, return the copy of the string
1529 wxString wxString::Upper() const
1530 { wxString s(*this); return s.MakeUpper(); }
1531
1532 // convert to lower case, return the copy of the string
1533 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1534
1535 int wxString::sprintf(const wxChar *pszFormat, ...)
1536   {
1537     va_list argptr;
1538     va_start(argptr, pszFormat);
1539     int iLen = PrintfV(pszFormat, argptr);
1540     va_end(argptr);
1541     return iLen;
1542   }
1543
1544 // ---------------------------------------------------------------------------
1545 // standard C++ library string functions
1546 // ---------------------------------------------------------------------------
1547
1548 #ifdef  wxSTD_STRING_COMPATIBILITY
1549
1550 void wxString::resize(size_t nSize, wxChar ch)
1551 {
1552     size_t len = length();
1553
1554     if ( nSize < len )
1555     {
1556         Truncate(nSize);
1557     }
1558     else if ( nSize > len )
1559     {
1560         *this += wxString(ch, len - nSize);
1561     }
1562     //else: we have exactly the specified length, nothing to do
1563 }
1564
1565 void wxString::swap(wxString& str)
1566 {
1567     // this is slightly less efficient than fiddling with m_pchData directly,
1568     // but it is still quite efficient as we don't copy the string here because
1569     // ref count always stays positive
1570     wxString tmp = str;
1571     str = *this;
1572     *this = str;
1573 }
1574
1575 wxString& wxString::insert(size_t nPos, const wxString& str)
1576 {
1577   wxASSERT( str.GetStringData()->IsValid() );
1578   wxASSERT( nPos <= Len() );
1579
1580   if ( !str.IsEmpty() ) {
1581     wxString strTmp;
1582     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1583     wxStrncpy(pc, c_str(), nPos);
1584     wxStrcpy(pc + nPos, str);
1585     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1586     strTmp.UngetWriteBuf();
1587     *this = strTmp;
1588   }
1589
1590   return *this;
1591 }
1592
1593 size_t wxString::find(const wxString& str, size_t nStart) const
1594 {
1595   wxASSERT( str.GetStringData()->IsValid() );
1596   wxASSERT( nStart <= Len() );
1597
1598   const wxChar *p = wxStrstr(c_str() + nStart, str);
1599
1600   return p == NULL ? npos : p - c_str();
1601 }
1602
1603 // VC++ 1.5 can't cope with the default argument in the header.
1604 #if !defined(__VISUALC__) || defined(__WIN32__)
1605 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1606 {
1607   return find(wxString(sz, n), nStart);
1608 }
1609 #endif // VC++ 1.5
1610
1611 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1612 #if !defined(__BORLANDC__)
1613 size_t wxString::find(wxChar ch, size_t nStart) const
1614 {
1615   wxASSERT( nStart <= Len() );
1616
1617   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1618
1619   return p == NULL ? npos : p - c_str();
1620 }
1621 #endif
1622
1623 size_t wxString::rfind(const wxString& str, size_t nStart) const
1624 {
1625   wxASSERT( str.GetStringData()->IsValid() );
1626   wxASSERT( nStart <= Len() );
1627
1628   // TODO could be made much quicker than that
1629   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1630   while ( p >= c_str() + str.Len() ) {
1631     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1632       return p - str.Len() - c_str();
1633     p--;
1634   }
1635
1636   return npos;
1637 }
1638
1639 // VC++ 1.5 can't cope with the default argument in the header.
1640 #if !defined(__VISUALC__) || defined(__WIN32__)
1641 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1642 {
1643     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1644 }
1645
1646 size_t wxString::rfind(wxChar ch, size_t nStart) const
1647 {
1648     if ( nStart == npos )
1649     {
1650         nStart = Len();
1651     }
1652     else
1653     {
1654         wxASSERT( nStart <= Len() );
1655     }
1656
1657     const wxChar *p = wxStrrchr(c_str(), ch);
1658
1659     if ( p == NULL )
1660         return npos;
1661
1662     size_t result = p - c_str();
1663     return ( result > nStart ) ? npos : result;
1664 }
1665 #endif // VC++ 1.5
1666
1667 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1668 {
1669     const wxChar *start = c_str() + nStart;
1670     const wxChar *firstOf = wxStrpbrk(start, sz);
1671     if ( firstOf )
1672         return firstOf - c_str();
1673     else
1674         return npos;
1675 }
1676
1677 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1678 {
1679     if ( nStart == npos )
1680     {
1681         nStart = Len();
1682     }
1683     else
1684     {
1685         wxASSERT( nStart <= Len() );
1686     }
1687
1688     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1689     {
1690         if ( wxStrchr(sz, *p) )
1691             return p - c_str();
1692     }
1693
1694     return npos;
1695 }
1696
1697 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1698 {
1699     if ( nStart == npos )
1700     {
1701         nStart = Len();
1702     }
1703     else
1704     {
1705         wxASSERT( nStart <= Len() );
1706     }
1707
1708     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1709     if ( nAccept >= length() - nStart )
1710         return npos;
1711     else
1712         return nAccept;
1713 }
1714
1715 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1716 {
1717     wxASSERT( nStart <= Len() );
1718
1719     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1720     {
1721         if ( *p != ch )
1722             return p - c_str();
1723     }
1724
1725     return npos;
1726 }
1727
1728 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1729 {
1730     if ( nStart == npos )
1731     {
1732         nStart = Len();
1733     }
1734     else
1735     {
1736         wxASSERT( nStart <= Len() );
1737     }
1738
1739     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1740     {
1741         if ( !wxStrchr(sz, *p) )
1742             return p - c_str();
1743     }
1744
1745     return npos;
1746 }
1747
1748 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1749 {
1750     if ( nStart == npos )
1751     {
1752         nStart = Len();
1753     }
1754     else
1755     {
1756         wxASSERT( nStart <= Len() );
1757     }
1758
1759     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1760     {
1761         if ( *p != ch )
1762             return p - c_str();
1763     }
1764
1765     return npos;
1766 }
1767
1768 wxString& wxString::erase(size_t nStart, size_t nLen)
1769 {
1770   wxString strTmp(c_str(), nStart);
1771   if ( nLen != npos ) {
1772     wxASSERT( nStart + nLen <= Len() );
1773
1774     strTmp.append(c_str() + nStart + nLen);
1775   }
1776
1777   *this = strTmp;
1778   return *this;
1779 }
1780
1781 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1782 {
1783   wxASSERT_MSG( nStart + nLen <= Len(),
1784                 _T("index out of bounds in wxString::replace") );
1785
1786   wxString strTmp;
1787   strTmp.Alloc(Len());      // micro optimisation to avoid multiple mem allocs
1788
1789   if ( nStart != 0 )
1790     strTmp.append(c_str(), nStart);
1791   strTmp << sz << c_str() + nStart + nLen;
1792
1793   *this = strTmp;
1794   return *this;
1795 }
1796
1797 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1798 {
1799   return replace(nStart, nLen, wxString(ch, nCount));
1800 }
1801
1802 wxString& wxString::replace(size_t nStart, size_t nLen,
1803                             const wxString& str, size_t nStart2, size_t nLen2)
1804 {
1805   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1806 }
1807
1808 wxString& wxString::replace(size_t nStart, size_t nLen,
1809                         const wxChar* sz, size_t nCount)
1810 {
1811   return replace(nStart, nLen, wxString(sz, nCount));
1812 }
1813
1814 #endif  //std::string compatibility
1815
1816 // ============================================================================
1817 // ArrayString
1818 // ============================================================================
1819
1820 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1821 #define   ARRAY_MAXSIZE_INCREMENT       4096
1822 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1823   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1824 #endif
1825
1826 #define   STRING(p)   ((wxString *)(&(p)))
1827
1828 // ctor
1829 wxArrayString::wxArrayString(bool autoSort)
1830 {
1831   m_nSize  =
1832   m_nCount = 0;
1833   m_pItems = (wxChar **) NULL;
1834   m_autoSort = autoSort;
1835 }
1836
1837 // copy ctor
1838 wxArrayString::wxArrayString(const wxArrayString& src)
1839 {
1840   m_nSize  =
1841   m_nCount = 0;
1842   m_pItems = (wxChar **) NULL;
1843   m_autoSort = src.m_autoSort;
1844
1845   *this = src;
1846 }
1847
1848 // assignment operator
1849 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1850 {
1851   if ( m_nSize > 0 )
1852     Clear();
1853
1854   Copy(src);
1855
1856   return *this;
1857 }
1858
1859 void wxArrayString::Copy(const wxArrayString& src)
1860 {
1861   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1862     Alloc(src.m_nCount);
1863
1864   for ( size_t n = 0; n < src.m_nCount; n++ )
1865     Add(src[n]);
1866 }
1867
1868 // grow the array
1869 void wxArrayString::Grow()
1870 {
1871   // only do it if no more place
1872   if( m_nCount == m_nSize ) {
1873     if( m_nSize == 0 ) {
1874       // was empty, alloc some memory
1875       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1876       m_pItems = new wxChar *[m_nSize];
1877     }
1878     else {
1879       // otherwise when it's called for the first time, nIncrement would be 0
1880       // and the array would never be expanded
1881 #if defined(__VISAGECPP__) && defined(__WXDEBUG__)
1882       int array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1883       wxASSERT( array_size != 0 );
1884 #else
1885       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1886 #endif
1887
1888       // add 50% but not too much
1889       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1890                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1891       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1892         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1893       m_nSize += nIncrement;
1894       wxChar **pNew = new wxChar *[m_nSize];
1895
1896       // copy data to new location
1897       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1898
1899       // delete old memory (but do not release the strings!)
1900       wxDELETEA(m_pItems);
1901
1902       m_pItems = pNew;
1903     }
1904   }
1905 }
1906
1907 void wxArrayString::Free()
1908 {
1909   for ( size_t n = 0; n < m_nCount; n++ ) {
1910     STRING(m_pItems[n])->GetStringData()->Unlock();
1911   }
1912 }
1913
1914 // deletes all the strings from the list
1915 void wxArrayString::Empty()
1916 {
1917   Free();
1918
1919   m_nCount = 0;
1920 }
1921
1922 // as Empty, but also frees memory
1923 void wxArrayString::Clear()
1924 {
1925   Free();
1926
1927   m_nSize  =
1928   m_nCount = 0;
1929
1930   wxDELETEA(m_pItems);
1931 }
1932
1933 // dtor
1934 wxArrayString::~wxArrayString()
1935 {
1936   Free();
1937
1938   wxDELETEA(m_pItems);
1939 }
1940
1941 // pre-allocates memory (frees the previous data!)
1942 void wxArrayString::Alloc(size_t nSize)
1943 {
1944   wxASSERT( nSize > 0 );
1945
1946   // only if old buffer was not big enough
1947   if ( nSize > m_nSize ) {
1948     Free();
1949     wxDELETEA(m_pItems);
1950     m_pItems = new wxChar *[nSize];
1951     m_nSize  = nSize;
1952   }
1953
1954   m_nCount = 0;
1955 }
1956
1957 // minimizes the memory usage by freeing unused memory
1958 void wxArrayString::Shrink()
1959 {
1960   // only do it if we have some memory to free
1961   if( m_nCount < m_nSize ) {
1962     // allocates exactly as much memory as we need
1963     wxChar **pNew = new wxChar *[m_nCount];
1964
1965     // copy data to new location
1966     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1967     delete [] m_pItems;
1968     m_pItems = pNew;
1969   }
1970 }
1971
1972 // searches the array for an item (forward or backwards)
1973 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1974 {
1975   if ( m_autoSort ) {
1976     // use binary search in the sorted array
1977     wxASSERT_MSG( bCase && !bFromEnd,
1978                   wxT("search parameters ignored for auto sorted array") );
1979
1980     size_t i,
1981            lo = 0,
1982            hi = m_nCount;
1983     int res;
1984     while ( lo < hi ) {
1985       i = (lo + hi)/2;
1986
1987       res = wxStrcmp(sz, m_pItems[i]);
1988       if ( res < 0 )
1989         hi = i;
1990       else if ( res > 0 )
1991         lo = i + 1;
1992       else
1993         return i;
1994     }
1995
1996     return wxNOT_FOUND;
1997   }
1998   else {
1999     // use linear search in unsorted array
2000     if ( bFromEnd ) {
2001       if ( m_nCount > 0 ) {
2002         size_t ui = m_nCount;
2003         do {
2004           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
2005             return ui;
2006         }
2007         while ( ui != 0 );
2008       }
2009     }
2010     else {
2011       for( size_t ui = 0; ui < m_nCount; ui++ ) {
2012         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
2013           return ui;
2014       }
2015     }
2016   }
2017
2018   return wxNOT_FOUND;
2019 }
2020
2021 // add item at the end
2022 size_t wxArrayString::Add(const wxString& str)
2023 {
2024   if ( m_autoSort ) {
2025     // insert the string at the correct position to keep the array sorted
2026     size_t i,
2027            lo = 0,
2028            hi = m_nCount;
2029     int res;
2030     while ( lo < hi ) {
2031       i = (lo + hi)/2;
2032
2033       res = wxStrcmp(str, m_pItems[i]);
2034       if ( res < 0 )
2035         hi = i;
2036       else if ( res > 0 )
2037         lo = i + 1;
2038       else {
2039         lo = hi = i;
2040         break;
2041       }
2042     }
2043
2044     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
2045
2046     Insert(str, lo);
2047
2048     return (size_t)lo;
2049   }
2050   else {
2051     wxASSERT( str.GetStringData()->IsValid() );
2052
2053     Grow();
2054
2055     // the string data must not be deleted!
2056     str.GetStringData()->Lock();
2057
2058     // just append
2059     m_pItems[m_nCount] = (wxChar *)str.c_str(); // const_cast
2060
2061     return m_nCount++;
2062   }
2063 }
2064
2065 // add item at the given position
2066 void wxArrayString::Insert(const wxString& str, size_t nIndex)
2067 {
2068   wxASSERT( str.GetStringData()->IsValid() );
2069
2070   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
2071
2072   Grow();
2073
2074   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
2075           (m_nCount - nIndex)*sizeof(wxChar *));
2076
2077   str.GetStringData()->Lock();
2078   m_pItems[nIndex] = (wxChar *)str.c_str();
2079
2080   m_nCount++;
2081 }
2082
2083 // removes item from array (by index)
2084 void wxArrayString::Remove(size_t nIndex)
2085 {
2086   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Remove") );
2087
2088   // release our lock
2089   Item(nIndex).GetStringData()->Unlock();
2090
2091   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
2092           (m_nCount - nIndex - 1)*sizeof(wxChar *));
2093   m_nCount--;
2094 }
2095
2096 // removes item from array (by value)
2097 void wxArrayString::Remove(const wxChar *sz)
2098 {
2099   int iIndex = Index(sz);
2100
2101   wxCHECK_RET( iIndex != wxNOT_FOUND,
2102                wxT("removing inexistent element in wxArrayString::Remove") );
2103
2104   Remove(iIndex);
2105 }
2106
2107 // ----------------------------------------------------------------------------
2108 // sorting
2109 // ----------------------------------------------------------------------------
2110
2111 // we can only sort one array at a time with the quick-sort based
2112 // implementation
2113 #if wxUSE_THREADS
2114   // need a critical section to protect access to gs_compareFunction and
2115   // gs_sortAscending variables
2116   static wxCriticalSection *gs_critsectStringSort = NULL;
2117
2118   // call this before the value of the global sort vars is changed/after
2119   // you're finished with them
2120   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2121                            gs_critsectStringSort = new wxCriticalSection;     \
2122                            gs_critsectStringSort->Enter()
2123   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2124                            delete gs_critsectStringSort;                      \
2125                            gs_critsectStringSort = NULL
2126 #else // !threads
2127   #define START_SORT()
2128   #define END_SORT()
2129 #endif // wxUSE_THREADS
2130
2131 // function to use for string comparaison
2132 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2133
2134 // if we don't use the compare function, this flag tells us if we sort the
2135 // array in ascending or descending order
2136 static bool gs_sortAscending = TRUE;
2137
2138 // function which is called by quick sort
2139 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
2140 {
2141   wxString *strFirst = (wxString *)first;
2142   wxString *strSecond = (wxString *)second;
2143
2144   if ( gs_compareFunction ) {
2145     return gs_compareFunction(*strFirst, *strSecond);
2146   }
2147   else {
2148     // maybe we should use wxStrcoll
2149     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2150
2151     return gs_sortAscending ? result : -result;
2152   }
2153 }
2154
2155 // sort array elements using passed comparaison function
2156 void wxArrayString::Sort(CompareFunction compareFunction)
2157 {
2158   START_SORT();
2159
2160   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2161   gs_compareFunction = compareFunction;
2162
2163   DoSort();
2164
2165   // reset it to NULL so that Sort(bool) will work the next time
2166   gs_compareFunction = NULL;
2167
2168   END_SORT();
2169 }
2170
2171 void wxArrayString::Sort(bool reverseOrder)
2172 {
2173   START_SORT();
2174
2175   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2176   gs_sortAscending = !reverseOrder;
2177
2178   DoSort();
2179
2180   END_SORT();
2181 }
2182
2183 void wxArrayString::DoSort()
2184 {
2185   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2186
2187   // just sort the pointers using qsort() - of course it only works because
2188   // wxString() *is* a pointer to its data
2189   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2190 }
2191
2192 bool wxArrayString::operator==(const wxArrayString& a) const
2193 {
2194     if ( m_nCount != a.m_nCount )
2195         return FALSE;
2196
2197     for ( size_t n = 0; n < m_nCount; n++ )
2198     {
2199         if ( Item(n) != a[n] )
2200             return FALSE;
2201     }
2202
2203     return TRUE;
2204 }
2205