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 #ifdef __WXPM__
  17 #define LINKAGECPP _Optlink
  18 #else
  19 #define LINKAGECPP
  20 #endif
  21
  22 /*
  23  * About ref counting:
  24  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  25  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  26  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  27  */
  28
  29 // ===========================================================================
  30 // headers, declarations, constants
  31 // ===========================================================================
  32
  33 // For compilers that support precompilation, includes "wx.h".
  34 #include "wx/wxprec.h"
  35
  36 #ifdef __BORLANDC__
  37   #pragma hdrstop
  38 #endif
  39
  40 #ifndef WX_PRECOMP
  41   #include "wx/defs.h"
  42   #include "wx/string.h"
  43   #include "wx/intl.h"
  44 #if wxUSE_THREADS
  45   #include <wx/thread.h>
  46 #endif
  47 #endif
  48
  49 #include <ctype.h>
  50 #include <string.h>
  51 #include <stdlib.h>
  52
  53 #ifdef __SALFORDC__
  54   #include <clib.h>
  55 #endif
  56
  57 #if wxUSE_WCSRTOMBS
  58   #include <wchar.h>    // for wcsrtombs(), see comments where it's used
  59 #endif // GNU
  60
  61 #ifdef  WXSTRING_IS_WXOBJECT
  62   IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
  63 #endif  //WXSTRING_IS_WXOBJECT
  64
  65 #if wxUSE_UNICODE
  66 #undef wxUSE_EXPERIMENTAL_PRINTF
  67 #define wxUSE_EXPERIMENTAL_PRINTF 1
  68 #endif
  69
  70 // allocating extra space for each string consumes more memory but speeds up
  71 // the concatenation operations (nLen is the current string's length)
  72 // NB: EXTRA_ALLOC must be >= 0!
  73 #define EXTRA_ALLOC       (19 - nLen % 16)
  74
  75 // ---------------------------------------------------------------------------
  76 // static class variables definition
  77 // ---------------------------------------------------------------------------
  78
  79 #ifdef  wxSTD_STRING_COMPATIBILITY
  80   const size_t wxString::npos = wxSTRING_MAXLEN;
  81 #endif // wxSTD_STRING_COMPATIBILITY
  82
  83 // ----------------------------------------------------------------------------
  84 // static data
  85 // ----------------------------------------------------------------------------
  86
  87 // for an empty string, GetStringData() will return this address: this
  88 // structure has the same layout as wxStringData and it's data() method will
  89 // return the empty string (dummy pointer)
  90 static const struct
  91 {
  92   wxStringData data;
  93   wxChar dummy;
  94 } g_strEmpty = { {-1, 0, 0}, _T('\0') };
  95
  96 // empty C style string: points to 'string data' byte of g_strEmpty
  97 extern const wxChar WXDLLEXPORT *g_szNul = &g_strEmpty.dummy;
  98
  99 // ----------------------------------------------------------------------------
 100 // conditional compilation
 101 // ----------------------------------------------------------------------------
 102
 103 #if !defined(__WXSW__) && wxUSE_UNICODE
 104   #ifdef wxUSE_EXPERIMENTAL_PRINTF
 105     #undef wxUSE_EXPERIMENTAL_PRINTF
 106   #endif
 107   #define wxUSE_EXPERIMENTAL_PRINTF 1
 108 #endif
 109
 110 // we want to find out if the current platform supports vsnprintf()-like
 111 // function: for Unix this is done with configure, for Windows we test the
 112 // compiler explicitly.
 113 #ifdef __WXMSW__
 114     #ifdef __VISUALC__
 115         #define wxVsnprintf     _vsnprintf
 116     #endif
 117 #else   // !Windows
 118     #ifdef HAVE_VSNPRINTF
 119         #define wxVsnprintf       vsnprintf
 120     #endif
 121 #endif  // Windows/!Windows
 122
 123 #ifndef wxVsnprintf
 124     // in this case we'll use vsprintf() (which is ANSI and thus should be
 125     // always available), but it's unsafe because it doesn't check for buffer
 126     // size - so give a warning
 127     #define wxVsnprintf(buffer,len,format,argptr) vsprintf(buffer,format, argptr)
 128
 129     #if defined(__VISUALC__)
 130         #pragma message("Using sprintf() because no snprintf()-like function defined")
 131     #elif defined(__GNUG__) && !defined(__UNIX__)
 132         #warning "Using sprintf() because no snprintf()-like function defined"
 133     #elif defined(__MWERKS__)
 134              #warning "Using sprintf() because no snprintf()-like function defined"
 135     #endif //compiler
 136 #endif // no vsnprintf
 137
 138 #ifdef _AIX
 139   // AIX has vsnprintf, but there's no prototype in the system headers.
 140   extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
 141 #endif
 142
 143 // ----------------------------------------------------------------------------
 144 // global functions
 145 // ----------------------------------------------------------------------------
 146
 147 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
 148
 149 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
 150 // iostream ones.
 151 //
 152 // ATTN: you can _not_ use both of these in the same program!
 153
 154 istream& operator>>(istream& is, wxString& WXUNUSED(str))
 155 {
 156 #if 0
 157   int w = is.width(0);
 158   if ( is.ipfx(0) ) {
 159     streambuf *sb = is.rdbuf();
 160     str.erase();
 161     while ( true ) {
 162       int ch = sb->sbumpc ();
 163       if ( ch == EOF ) {
 164         is.setstate(ios::eofbit);
 165         break;
 166       }
 167       else if ( isspace(ch) ) {
 168         sb->sungetc();
 169         break;
 170       }
 171
 172       str += ch;
 173       if ( --w == 1 )
 174         break;
 175     }
 176   }
 177
 178   is.isfx();
 179   if ( str.length() == 0 )
 180     is.setstate(ios::failbit);
 181 #endif
 182   return is;
 183 }
 184
 185 #endif  //std::string compatibility
 186
 187 // ----------------------------------------------------------------------------
 188 // private classes
 189 // ----------------------------------------------------------------------------
 190
 191 // this small class is used to gather statistics for performance tuning
 192 //#define WXSTRING_STATISTICS
 193 #ifdef  WXSTRING_STATISTICS
 194   class Averager
 195   {
 196   public:
 197     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 198    ~Averager()
 199    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 200
 201     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 202
 203   private:
 204     size_t m_nCount, m_nTotal;
 205     const char *m_sz;
 206   } g_averageLength("allocation size"),
 207     g_averageSummandLength("summand length"),
 208     g_averageConcatHit("hit probability in concat"),
 209     g_averageInitialLength("initial string length");
 210
 211   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 212 #else
 213   #define STATISTICS_ADD(av, val)
 214 #endif // WXSTRING_STATISTICS
 215
 216 // ===========================================================================
 217 // wxString class core
 218 // ===========================================================================
 219
 220 // ---------------------------------------------------------------------------
 221 // construction
 222 // ---------------------------------------------------------------------------
 223
 224 // constructs string of <nLength> copies of character <ch>
 225 wxString::wxString(wxChar ch, size_t nLength)
 226 {
 227   Init();
 228
 229   if ( nLength > 0 ) {
 230     AllocBuffer(nLength);
 231
 232 #if wxUSE_UNICODE
 233     // memset only works on char
 234     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 235 #else
 236     memset(m_pchData, ch, nLength);
 237 #endif
 238   }
 239 }
 240
 241 // takes nLength elements of psz starting at nPos
 242 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 243 {
 244   Init();
 245
 246   wxASSERT( nPos <= wxStrlen(psz) );
 247
 248   if ( nLength == wxSTRING_MAXLEN )
 249     nLength = wxStrlen(psz + nPos);
 250
 251   STATISTICS_ADD(InitialLength, nLength);
 252
 253   if ( nLength > 0 ) {
 254     // trailing '\0' is written in AllocBuffer()
 255     AllocBuffer(nLength);
 256     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 257   }
 258 }
 259
 260 #ifdef  wxSTD_STRING_COMPATIBILITY
 261
 262 // poor man's iterators are "void *" pointers
 263 wxString::wxString(const void *pStart, const void *pEnd)
 264 {
 265   InitWith((const wxChar *)pStart, 0,
 266            (const wxChar *)pEnd - (const wxChar *)pStart);
 267 }
 268
 269 #endif  //std::string compatibility
 270
 271 #if wxUSE_UNICODE
 272
 273 // from multibyte string
 274 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 275 {
 276   // first get necessary size
 277   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 278
 279   // nLength is number of *Unicode* characters here!
 280   if ((nLen != (size_t)-1) && (nLen > nLength))
 281     nLen = nLength;
 282
 283   // empty?
 284   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 285     AllocBuffer(nLen);
 286     conv.MB2WC(m_pchData, psz, nLen);
 287   }
 288   else {
 289     Init();
 290   }
 291 }
 292
 293 #else
 294
 295 #if wxUSE_WCHAR_T
 296 // from wide string
 297 wxString::wxString(const wchar_t *pwz)
 298 {
 299   // first get necessary size
 300   size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0;
 301
 302   // empty?
 303   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 304     AllocBuffer(nLen);
 305     wxWC2MB(m_pchData, pwz, nLen);
 306   }
 307   else {
 308     Init();
 309   }
 310 }
 311 #endif
 312
 313 #endif
 314
 315 // ---------------------------------------------------------------------------
 316 // memory allocation
 317 // ---------------------------------------------------------------------------
 318
 319 // allocates memory needed to store a C string of length nLen
 320 void wxString::AllocBuffer(size_t nLen)
 321 {
 322   wxASSERT( nLen >  0         );    //
 323   wxASSERT( nLen <= INT_MAX-1 );    // max size (enough room for 1 extra)
 324
 325   STATISTICS_ADD(Length, nLen);
 326
 327   // allocate memory:
 328   // 1) one extra character for '\0' termination
 329   // 2) sizeof(wxStringData) for housekeeping info
 330   wxStringData* pData = (wxStringData*)
 331     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 332   pData->nRefs        = 1;
 333   pData->nDataLength  = nLen;
 334   pData->nAllocLength = nLen + EXTRA_ALLOC;
 335   m_pchData           = pData->data();  // data starts after wxStringData
 336   m_pchData[nLen]     = _T('\0');
 337 }
 338
 339 // must be called before changing this string
 340 void wxString::CopyBeforeWrite()
 341 {
 342   wxStringData* pData = GetStringData();
 343
 344   if ( pData->IsShared() ) {
 345     pData->Unlock();                // memory not freed because shared
 346     size_t nLen = pData->nDataLength;
 347     AllocBuffer(nLen);
 348     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 349   }
 350
 351   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 352 }
 353
 354 // must be called before replacing contents of this string
 355 void wxString::AllocBeforeWrite(size_t nLen)
 356 {
 357   wxASSERT( nLen != 0 );  // doesn't make any sense
 358
 359   // must not share string and must have enough space
 360   wxStringData* pData = GetStringData();
 361   if ( pData->IsShared() || (nLen > pData->nAllocLength) ) {
 362     // can't work with old buffer, get new one
 363     pData->Unlock();
 364     AllocBuffer(nLen);
 365   }
 366   else {
 367     // update the string length
 368     pData->nDataLength = nLen;
 369   }
 370
 371   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 372 }
 373
 374 // allocate enough memory for nLen characters
 375 void wxString::Alloc(size_t nLen)
 376 {
 377   wxStringData *pData = GetStringData();
 378   if ( pData->nAllocLength <= nLen ) {
 379     if ( pData->IsEmpty() ) {
 380       nLen += EXTRA_ALLOC;
 381
 382       wxStringData* pData = (wxStringData*)
 383         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 384       pData->nRefs = 1;
 385       pData->nDataLength = 0;
 386       pData->nAllocLength = nLen;
 387       m_pchData = pData->data();  // data starts after wxStringData
 388       m_pchData[0u] = _T('\0');
 389     }
 390     else if ( pData->IsShared() ) {
 391       pData->Unlock();                // memory not freed because shared
 392       size_t nOldLen = pData->nDataLength;
 393       AllocBuffer(nLen);
 394       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 395     }
 396     else {
 397       nLen += EXTRA_ALLOC;
 398
 399       wxStringData *p = (wxStringData *)
 400         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 401
 402       if ( p == NULL ) {
 403         // @@@ what to do on memory error?
 404         return;
 405       }
 406
 407       // it's not important if the pointer changed or not (the check for this
 408       // is not faster than assigning to m_pchData in all cases)
 409       p->nAllocLength = nLen;
 410       m_pchData = p->data();
 411     }
 412   }
 413   //else: we've already got enough
 414 }
 415
 416 // shrink to minimal size (releasing extra memory)
 417 void wxString::Shrink()
 418 {
 419   wxStringData *pData = GetStringData();
 420
 421   // this variable is unused in release build, so avoid the compiler warning by
 422   // just not declaring it
 423 #ifdef __WXDEBUG__
 424   void *p =
 425 #endif
 426   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 427
 428   wxASSERT( p != NULL );  // can't free memory?
 429   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 430 }
 431
 432 // get the pointer to writable buffer of (at least) nLen bytes
 433 wxChar *wxString::GetWriteBuf(size_t nLen)
 434 {
 435   AllocBeforeWrite(nLen);
 436
 437   wxASSERT( GetStringData()->nRefs == 1 );
 438   GetStringData()->Validate(FALSE);
 439
 440   return m_pchData;
 441 }
 442
 443 // put string back in a reasonable state after GetWriteBuf
 444 void wxString::UngetWriteBuf()
 445 {
 446   GetStringData()->nDataLength = wxStrlen(m_pchData);
 447   GetStringData()->Validate(TRUE);
 448 }
 449
 450 // ---------------------------------------------------------------------------
 451 // data access
 452 // ---------------------------------------------------------------------------
 453
 454 // all functions are inline in string.h
 455
 456 // ---------------------------------------------------------------------------
 457 // assignment operators
 458 // ---------------------------------------------------------------------------
 459
 460 // helper function: does real copy
 461 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 462 {
 463   if ( nSrcLen == 0 ) {
 464     Reinit();
 465   }
 466   else {
 467     AllocBeforeWrite(nSrcLen);
 468     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 469     GetStringData()->nDataLength = nSrcLen;
 470     m_pchData[nSrcLen] = _T('\0');
 471   }
 472 }
 473
 474 // assigns one string to another
 475 wxString& wxString::operator=(const wxString& stringSrc)
 476 {
 477   wxASSERT( stringSrc.GetStringData()->IsValid() );
 478
 479   // don't copy string over itself
 480   if ( m_pchData != stringSrc.m_pchData ) {
 481     if ( stringSrc.GetStringData()->IsEmpty() ) {
 482       Reinit();
 483     }
 484     else {
 485       // adjust references
 486       GetStringData()->Unlock();
 487       m_pchData = stringSrc.m_pchData;
 488       GetStringData()->Lock();
 489     }
 490   }
 491
 492   return *this;
 493 }
 494
 495 // assigns a single character
 496 wxString& wxString::operator=(wxChar ch)
 497 {
 498   AssignCopy(1, &ch);
 499   return *this;
 500 }
 501
 502 // assigns C string
 503 wxString& wxString::operator=(const wxChar *psz)
 504 {
 505   AssignCopy(wxStrlen(psz), psz);
 506   return *this;
 507 }
 508
 509 #if !wxUSE_UNICODE
 510
 511 // same as 'signed char' variant
 512 wxString& wxString::operator=(const unsigned char* psz)
 513 {
 514   *this = (const char *)psz;
 515   return *this;
 516 }
 517
 518 #if wxUSE_WCHAR_T
 519 wxString& wxString::operator=(const wchar_t *pwz)
 520 {
 521   wxString str(pwz);
 522   *this = str;
 523   return *this;
 524 }
 525 #endif
 526
 527 #endif
 528
 529 // ---------------------------------------------------------------------------
 530 // string concatenation
 531 // ---------------------------------------------------------------------------
 532
 533 // add something to this string
 534 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 535 {
 536   STATISTICS_ADD(SummandLength, nSrcLen);
 537
 538   // concatenating an empty string is a NOP
 539   if ( nSrcLen > 0 ) {
 540     wxStringData *pData = GetStringData();
 541     size_t nLen = pData->nDataLength;
 542     size_t nNewLen = nLen + nSrcLen;
 543
 544     // alloc new buffer if current is too small
 545     if ( pData->IsShared() ) {
 546       STATISTICS_ADD(ConcatHit, 0);
 547
 548       // we have to allocate another buffer
 549       wxStringData* pOldData = GetStringData();
 550       AllocBuffer(nNewLen);
 551       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 552       pOldData->Unlock();
 553     }
 554     else if ( nNewLen > pData->nAllocLength ) {
 555       STATISTICS_ADD(ConcatHit, 0);
 556
 557       // we have to grow the buffer
 558       Alloc(nNewLen);
 559     }
 560     else {
 561       STATISTICS_ADD(ConcatHit, 1);
 562
 563       // the buffer is already big enough
 564     }
 565
 566     // should be enough space
 567     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 568
 569     // fast concatenation - all is done in our buffer
 570     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 571
 572     m_pchData[nNewLen] = _T('\0');          // put terminating '\0'
 573     GetStringData()->nDataLength = nNewLen; // and fix the length
 574   }
 575   //else: the string to append was empty
 576 }
 577
 578 /*
 579  * concatenation functions come in 5 flavours:
 580  *  string + string
 581  *  char   + string      and      string + char
 582  *  C str  + string      and      string + C str
 583  */
 584
 585 wxString operator+(const wxString& string1, const wxString& string2)
 586 {
 587   wxASSERT( string1.GetStringData()->IsValid() );
 588   wxASSERT( string2.GetStringData()->IsValid() );
 589
 590   wxString s = string1;
 591   s += string2;
 592
 593   return s;
 594 }
 595
 596 wxString operator+(const wxString& string, wxChar ch)
 597 {
 598   wxASSERT( string.GetStringData()->IsValid() );
 599
 600   wxString s = string;
 601   s += ch;
 602
 603   return s;
 604 }
 605
 606 wxString operator+(wxChar ch, const wxString& string)
 607 {
 608   wxASSERT( string.GetStringData()->IsValid() );
 609
 610   wxString s = ch;
 611   s += string;
 612
 613   return s;
 614 }
 615
 616 wxString operator+(const wxString& string, const wxChar *psz)
 617 {
 618   wxASSERT( string.GetStringData()->IsValid() );
 619
 620   wxString s;
 621   s.Alloc(wxStrlen(psz) + string.Len());
 622   s = string;
 623   s += psz;
 624
 625   return s;
 626 }
 627
 628 wxString operator+(const wxChar *psz, const wxString& string)
 629 {
 630   wxASSERT( string.GetStringData()->IsValid() );
 631
 632   wxString s;
 633   s.Alloc(wxStrlen(psz) + string.Len());
 634   s = psz;
 635   s += string;
 636
 637   return s;
 638 }
 639
 640 // ===========================================================================
 641 // other common string functions
 642 // ===========================================================================
 643
 644 // ---------------------------------------------------------------------------
 645 // simple sub-string extraction
 646 // ---------------------------------------------------------------------------
 647
 648 // helper function: clone the data attached to this string
 649 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 650 {
 651   if ( nCopyLen == 0 ) {
 652     dest.Init();
 653   }
 654   else {
 655     dest.AllocBuffer(nCopyLen);
 656     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 657   }
 658 }
 659
 660 // extract string of length nCount starting at nFirst
 661 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 662 {
 663   wxStringData *pData = GetStringData();
 664   size_t nLen = pData->nDataLength;
 665
 666   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 667   if ( nCount == wxSTRING_MAXLEN )
 668   {
 669     nCount = nLen - nFirst;
 670   }
 671
 672   // out-of-bounds requests return sensible things
 673   if ( nFirst + nCount > nLen )
 674   {
 675     nCount = nLen - nFirst;
 676   }
 677
 678   if ( nFirst > nLen )
 679   {
 680     // AllocCopy() will return empty string
 681     nCount = 0;
 682   }
 683
 684   wxString dest;
 685   AllocCopy(dest, nCount, nFirst);
 686
 687   return dest;
 688 }
 689
 690 // extract nCount last (rightmost) characters
 691 wxString wxString::Right(size_t nCount) const
 692 {
 693   if ( nCount > (size_t)GetStringData()->nDataLength )
 694     nCount = GetStringData()->nDataLength;
 695
 696   wxString dest;
 697   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 698   return dest;
 699 }
 700
 701 // get all characters after the last occurence of ch
 702 // (returns the whole string if ch not found)
 703 wxString wxString::AfterLast(wxChar ch) const
 704 {
 705   wxString str;
 706   int iPos = Find(ch, TRUE);
 707   if ( iPos == wxNOT_FOUND )
 708     str = *this;
 709   else
 710     str = c_str() + iPos + 1;
 711
 712   return str;
 713 }
 714
 715 // extract nCount first (leftmost) characters
 716 wxString wxString::Left(size_t nCount) const
 717 {
 718   if ( nCount > (size_t)GetStringData()->nDataLength )
 719     nCount = GetStringData()->nDataLength;
 720
 721   wxString dest;
 722   AllocCopy(dest, nCount, 0);
 723   return dest;
 724 }
 725
 726 // get all characters before the first occurence of ch
 727 // (returns the whole string if ch not found)
 728 wxString wxString::BeforeFirst(wxChar ch) const
 729 {
 730   wxString str;
 731   for ( const wxChar *pc = m_pchData; *pc != _T('\0') && *pc != ch; pc++ )
 732     str += *pc;
 733
 734   return str;
 735 }
 736
 737 /// get all characters before the last occurence of ch
 738 /// (returns empty string if ch not found)
 739 wxString wxString::BeforeLast(wxChar ch) const
 740 {
 741   wxString str;
 742   int iPos = Find(ch, TRUE);
 743   if ( iPos != wxNOT_FOUND && iPos != 0 )
 744     str = wxString(c_str(), iPos);
 745
 746   return str;
 747 }
 748
 749 /// get all characters after the first occurence of ch
 750 /// (returns empty string if ch not found)
 751 wxString wxString::AfterFirst(wxChar ch) const
 752 {
 753   wxString str;
 754   int iPos = Find(ch);
 755   if ( iPos != wxNOT_FOUND )
 756     str = c_str() + iPos + 1;
 757
 758   return str;
 759 }
 760
 761 // replace first (or all) occurences of some substring with another one
 762 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 763 {
 764   size_t uiCount = 0;   // count of replacements made
 765
 766   size_t uiOldLen = wxStrlen(szOld);
 767
 768   wxString strTemp;
 769   const wxChar *pCurrent = m_pchData;
 770   const wxChar *pSubstr;
 771   while ( *pCurrent != _T('\0') ) {
 772     pSubstr = wxStrstr(pCurrent, szOld);
 773     if ( pSubstr == NULL ) {
 774       // strTemp is unused if no replacements were made, so avoid the copy
 775       if ( uiCount == 0 )
 776         return 0;
 777
 778       strTemp += pCurrent;    // copy the rest
 779       break;                  // exit the loop
 780     }
 781     else {
 782       // take chars before match
 783       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 784       strTemp += szNew;
 785       pCurrent = pSubstr + uiOldLen;  // restart after match
 786
 787       uiCount++;
 788
 789       // stop now?
 790       if ( !bReplaceAll ) {
 791         strTemp += pCurrent;    // copy the rest
 792         break;                  // exit the loop
 793       }
 794     }
 795   }
 796
 797   // only done if there were replacements, otherwise would have returned above
 798   *this = strTemp;
 799
 800   return uiCount;
 801 }
 802
 803 bool wxString::IsAscii() const
 804 {
 805   const wxChar *s = (const wxChar*) *this;
 806   while(*s){
 807     if(!isascii(*s)) return(FALSE);
 808     s++;
 809   }
 810   return(TRUE);
 811 }
 812
 813 bool wxString::IsWord() const
 814 {
 815   const wxChar *s = (const wxChar*) *this;
 816   while(*s){
 817     if(!wxIsalpha(*s)) return(FALSE);
 818     s++;
 819   }
 820   return(TRUE);
 821 }
 822
 823 bool wxString::IsNumber() const
 824 {
 825   const wxChar *s = (const wxChar*) *this;
 826   while(*s){
 827     if(!wxIsdigit(*s)) return(FALSE);
 828     s++;
 829   }
 830   return(TRUE);
 831 }
 832
 833 wxString wxString::Strip(stripType w) const
 834 {
 835     wxString s = *this;
 836     if ( w & leading ) s.Trim(FALSE);
 837     if ( w & trailing ) s.Trim(TRUE);
 838     return s;
 839 }
 840
 841 // ---------------------------------------------------------------------------
 842 // case conversion
 843 // ---------------------------------------------------------------------------
 844
 845 wxString& wxString::MakeUpper()
 846 {
 847   CopyBeforeWrite();
 848
 849   for ( wxChar *p = m_pchData; *p; p++ )
 850     *p = (wxChar)wxToupper(*p);
 851
 852   return *this;
 853 }
 854
 855 wxString& wxString::MakeLower()
 856 {
 857   CopyBeforeWrite();
 858
 859   for ( wxChar *p = m_pchData; *p; p++ )
 860     *p = (wxChar)wxTolower(*p);
 861
 862   return *this;
 863 }
 864
 865 // ---------------------------------------------------------------------------
 866 // trimming and padding
 867 // ---------------------------------------------------------------------------
 868
 869 // trims spaces (in the sense of isspace) from left or right side
 870 wxString& wxString::Trim(bool bFromRight)
 871 {
 872   // first check if we're going to modify the string at all
 873   if ( !IsEmpty() &&
 874        (
 875         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 876         (!bFromRight && wxIsspace(GetChar(0u)))
 877        )
 878      )
 879   {
 880     // ok, there is at least one space to trim
 881     CopyBeforeWrite();
 882
 883     if ( bFromRight )
 884     {
 885       // find last non-space character
 886       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
 887       while ( wxIsspace(*psz) && (psz >= m_pchData) )
 888         psz--;
 889
 890       // truncate at trailing space start
 891       *++psz = _T('\0');
 892       GetStringData()->nDataLength = psz - m_pchData;
 893     }
 894     else
 895     {
 896       // find first non-space character
 897       const wxChar *psz = m_pchData;
 898       while ( wxIsspace(*psz) )
 899         psz++;
 900
 901       // fix up data and length
 902       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
 903       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
 904       GetStringData()->nDataLength = nDataLength;
 905     }
 906   }
 907
 908   return *this;
 909 }
 910
 911 // adds nCount characters chPad to the string from either side
 912 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
 913 {
 914   wxString s(chPad, nCount);
 915
 916   if ( bFromRight )
 917     *this += s;
 918   else
 919   {
 920     s += *this;
 921     *this = s;
 922   }
 923
 924   return *this;
 925 }
 926
 927 // truncate the string
 928 wxString& wxString::Truncate(size_t uiLen)
 929 {
 930   if ( uiLen < Len() ) {
 931     CopyBeforeWrite();
 932
 933     *(m_pchData + uiLen) = _T('\0');
 934     GetStringData()->nDataLength = uiLen;
 935   }
 936   //else: nothing to do, string is already short enough
 937
 938   return *this;
 939 }
 940
 941 // ---------------------------------------------------------------------------
 942 // finding (return wxNOT_FOUND if not found and index otherwise)
 943 // ---------------------------------------------------------------------------
 944
 945 // find a character
 946 int wxString::Find(wxChar ch, bool bFromEnd) const
 947 {
 948   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
 949
 950   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 951 }
 952
 953 // find a sub-string (like strstr)
 954 int wxString::Find(const wxChar *pszSub) const
 955 {
 956   const wxChar *psz = wxStrstr(m_pchData, pszSub);
 957
 958   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 959 }
 960
 961 // ---------------------------------------------------------------------------
 962 // stream-like operators
 963 // ---------------------------------------------------------------------------
 964 wxString& wxString::operator<<(int i)
 965 {
 966     wxString res;
 967     res.Printf(_T("%d"), i);
 968
 969     return (*this) << res;
 970 }
 971
 972 wxString& wxString::operator<<(float f)
 973 {
 974     wxString res;
 975     res.Printf(_T("%f"), f);
 976
 977     return (*this) << res;
 978 }
 979
 980 wxString& wxString::operator<<(double d)
 981 {
 982     wxString res;
 983     res.Printf(_T("%g"), d);
 984
 985     return (*this) << res;
 986 }
 987
 988 // ---------------------------------------------------------------------------
 989 // formatted output
 990 // ---------------------------------------------------------------------------
 991 int wxString::Printf(const wxChar *pszFormat, ...)
 992 {
 993   va_list argptr;
 994   va_start(argptr, pszFormat);
 995
 996   int iLen = PrintfV(pszFormat, argptr);
 997
 998   va_end(argptr);
 999
1000   return iLen;
1001 }
1002
1003 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1004 {
1005   // static buffer to avoid dynamic memory allocation each time
1006   char s_szScratch[1024]; // using static buffer causes internal compiler err
1007 #if 0
1008 #if wxUSE_THREADS
1009   // protect the static buffer
1010   static wxCriticalSection critsect;
1011   wxCriticalSectionLocker lock(critsect);
1012 #endif
1013 #endif
1014
1015 #if wxUSE_EXPERIMENTAL_PRINTF
1016 // the new implementation
1017
1018   Reinit();
1019   for (size_t n = 0; pszFormat[n]; n++)
1020     if (pszFormat[n] == _T('%')) {
1021       static char s_szFlags[256] = "%";
1022       size_t flagofs = 1;
1023       bool adj_left = FALSE, in_prec = FALSE,
1024            prec_dot = FALSE, done = FALSE;
1025       int ilen = 0;
1026       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1027       do {
1028 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1029         switch (pszFormat[++n]) {
1030         case _T('\0'):
1031           done = TRUE;
1032           break;
1033         case _T('%'):
1034           *this += _T('%');
1035           done = TRUE;
1036           break;
1037         case _T('#'):
1038         case _T('0'):
1039         case _T(' '):
1040         case _T('+'):
1041         case _T('\''):
1042           CHECK_PREC
1043           s_szFlags[flagofs++] = pszFormat[n];
1044           break;
1045         case _T('-'):
1046           CHECK_PREC
1047           adj_left = TRUE;
1048           s_szFlags[flagofs++] = pszFormat[n];
1049           break;
1050         case _T('.'):
1051           CHECK_PREC
1052           in_prec = TRUE;
1053           prec_dot = FALSE;
1054           max_width = 0;
1055           // dot will be auto-added to s_szFlags if non-negative number follows
1056           break;
1057         case _T('h'):
1058           ilen = -1;
1059           CHECK_PREC
1060           s_szFlags[flagofs++] = pszFormat[n];
1061           break;
1062         case _T('l'):
1063           ilen = 1;
1064           CHECK_PREC
1065           s_szFlags[flagofs++] = pszFormat[n];
1066           break;
1067         case _T('q'):
1068         case _T('L'):
1069           ilen = 2;
1070           CHECK_PREC
1071           s_szFlags[flagofs++] = pszFormat[n];
1072           break;
1073         case _T('Z'):
1074           ilen = 3;
1075           CHECK_PREC
1076           s_szFlags[flagofs++] = pszFormat[n];
1077           break;
1078         case _T('*'):
1079           {
1080             int len = va_arg(argptr, int);
1081             if (in_prec) {
1082               if (len<0) break;
1083               CHECK_PREC
1084               max_width = len;
1085             } else {
1086               if (len<0) {
1087                 adj_left = !adj_left;
1088                 s_szFlags[flagofs++] = '-';
1089                 len = -len;
1090               }
1091               min_width = len;
1092             }
1093             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1094           }
1095           break;
1096         case _T('1'): case _T('2'): case _T('3'):
1097         case _T('4'): case _T('5'): case _T('6'):
1098         case _T('7'): case _T('8'): case _T('9'):
1099           {
1100             int len = 0;
1101             CHECK_PREC
1102             while ((pszFormat[n]>=_T('0')) && (pszFormat[n]<=_T('9'))) {
1103               s_szFlags[flagofs++] = pszFormat[n];
1104               len = len*10 + (pszFormat[n] - _T('0'));
1105               n++;
1106             }
1107             if (in_prec) max_width = len;
1108             else min_width = len;
1109             n--; // the main loop pre-increments n again
1110           }
1111           break;
1112         case _T('d'):
1113         case _T('i'):
1114         case _T('o'):
1115         case _T('u'):
1116         case _T('x'):
1117         case _T('X'):
1118           CHECK_PREC
1119           s_szFlags[flagofs++] = pszFormat[n];
1120           s_szFlags[flagofs] = '\0';
1121           if (ilen == 0 ) {
1122             int val = va_arg(argptr, int);
1123             ::sprintf(s_szScratch, s_szFlags, val);
1124           }
1125           else if (ilen == -1) {
1126             short int val = va_arg(argptr, short int);
1127             ::sprintf(s_szScratch, s_szFlags, val);
1128           }
1129           else if (ilen == 1) {
1130             long int val = va_arg(argptr, long int);
1131             ::sprintf(s_szScratch, s_szFlags, val);
1132           }
1133           else if (ilen == 2) {
1134 #if SIZEOF_LONG_LONG
1135             long long int val = va_arg(argptr, long long int);
1136             ::sprintf(s_szScratch, s_szFlags, val);
1137 #else
1138             long int val = va_arg(argptr, long int);
1139             ::sprintf(s_szScratch, s_szFlags, val);
1140 #endif
1141           }
1142           else if (ilen == 3) {
1143             size_t val = va_arg(argptr, size_t);
1144             ::sprintf(s_szScratch, s_szFlags, val);
1145           }
1146           *this += wxString(s_szScratch);
1147           done = TRUE;
1148           break;
1149         case _T('e'):
1150         case _T('E'):
1151         case _T('f'):
1152         case _T('g'):
1153         case _T('G'):
1154           CHECK_PREC
1155           s_szFlags[flagofs++] = pszFormat[n];
1156           s_szFlags[flagofs] = '\0';
1157           if (ilen == 2) {
1158             long double val = va_arg(argptr, long double);
1159             ::sprintf(s_szScratch, s_szFlags, val);
1160           } else {
1161             double val = va_arg(argptr, double);
1162             ::sprintf(s_szScratch, s_szFlags, val);
1163           }
1164           *this += wxString(s_szScratch);
1165           done = TRUE;
1166           break;
1167         case _T('p'):
1168           {
1169             void *val = va_arg(argptr, void *);
1170             CHECK_PREC
1171             s_szFlags[flagofs++] = pszFormat[n];
1172             s_szFlags[flagofs] = '\0';
1173             ::sprintf(s_szScratch, s_szFlags, val);
1174             *this += wxString(s_szScratch);
1175             done = TRUE;
1176           }
1177           break;
1178         case _T('c'):
1179           {
1180             wxChar val = va_arg(argptr, int);
1181             // we don't need to honor padding here, do we?
1182             *this += val;
1183             done = TRUE;
1184           }
1185           break;
1186         case _T('s'):
1187           if (ilen == -1) {
1188             // wx extension: we'll let %hs mean non-Unicode strings
1189             char *val = va_arg(argptr, char *);
1190 #if wxUSE_UNICODE
1191             // ASCII->Unicode constructor handles max_width right
1192             wxString s(val, wxConvLibc, max_width);
1193 #else
1194             size_t len = wxSTRING_MAXLEN;
1195             if (val) {
1196               for (len = 0; val[len] && (len<max_width); len++);
1197             } else val = _T("(null)");
1198             wxString s(val, len);
1199 #endif
1200             if (s.Len() < min_width)
1201               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1202             *this += s;
1203           } else {
1204             wxChar *val = va_arg(argptr, wxChar *);
1205             size_t len = wxSTRING_MAXLEN;
1206             if (val) {
1207               for (len = 0; val[len] && (len<max_width); len++);
1208             } else val = _T("(null)");
1209             wxString s(val, len);
1210             if (s.Len() < min_width)
1211               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1212             *this += s;
1213           }
1214           done = TRUE;
1215           break;
1216         case _T('n'):
1217           if (ilen == 0) {
1218             int *val = va_arg(argptr, int *);
1219             *val = Len();
1220           }
1221           else if (ilen == -1) {
1222             short int *val = va_arg(argptr, short int *);
1223             *val = Len();
1224           }
1225           else if (ilen >= 1) {
1226             long int *val = va_arg(argptr, long int *);
1227             *val = Len();
1228           }
1229           done = TRUE;
1230           break;
1231         default:
1232           if (wxIsalpha(pszFormat[n]))
1233             // probably some flag not taken care of here yet
1234             s_szFlags[flagofs++] = pszFormat[n];
1235           else {
1236             // bad format
1237             *this += _T('%'); // just to pass the glibc tst-printf.c
1238             n--;
1239             done = TRUE;
1240           }
1241           break;
1242         }
1243 #undef CHECK_PREC
1244       } while (!done);
1245     } else *this += pszFormat[n];
1246
1247 #else
1248   // NB: wxVsnprintf() may return either less than the buffer size or -1 if there
1249   //     is not enough place depending on implementation
1250   int iLen = wxVsnprintf(s_szScratch, WXSIZEOF(s_szScratch), pszFormat, argptr);
1251   char *buffer;
1252   if ( iLen < (int)WXSIZEOF(s_szScratch) ) {
1253     buffer = s_szScratch;
1254   }
1255   else {
1256       int size = WXSIZEOF(s_szScratch) * 2;
1257       buffer = (char *)malloc(size);
1258       while ( buffer != NULL ) {
1259           iLen = wxVsnprintf(buffer, WXSIZEOF(s_szScratch), pszFormat, argptr);
1260           if ( iLen < size ) {
1261               // ok, there was enough space
1262               break;
1263           }
1264
1265           // still not enough, double it again
1266           buffer = (char *)realloc(buffer, size *= 2);
1267       }
1268
1269       if ( !buffer ) {
1270           // out of memory
1271           return -1;
1272       }
1273   }
1274
1275   wxString s(buffer);
1276   *this = s;
1277
1278   if ( buffer != s_szScratch )
1279       free(buffer);
1280 #endif
1281
1282   return Len();
1283 }
1284
1285 // ----------------------------------------------------------------------------
1286 // misc other operations
1287 // ----------------------------------------------------------------------------
1288 bool wxString::Matches(const wxChar *pszMask) const
1289 {
1290   // check char by char
1291   const wxChar *pszTxt;
1292   for ( pszTxt = c_str(); *pszMask != _T('\0'); pszMask++, pszTxt++ ) {
1293     switch ( *pszMask ) {
1294       case _T('?'):
1295         if ( *pszTxt == _T('\0') )
1296           return FALSE;
1297
1298         pszTxt++;
1299         pszMask++;
1300         break;
1301
1302       case _T('*'):
1303         {
1304           // ignore special chars immediately following this one
1305           while ( *pszMask == _T('*') || *pszMask == _T('?') )
1306             pszMask++;
1307
1308           // if there is nothing more, match
1309           if ( *pszMask == _T('\0') )
1310             return TRUE;
1311
1312           // are there any other metacharacters in the mask?
1313           size_t uiLenMask;
1314           const wxChar *pEndMask = wxStrpbrk(pszMask, _T("*?"));
1315
1316           if ( pEndMask != NULL ) {
1317             // we have to match the string between two metachars
1318             uiLenMask = pEndMask - pszMask;
1319           }
1320           else {
1321             // we have to match the remainder of the string
1322             uiLenMask = wxStrlen(pszMask);
1323           }
1324
1325           wxString strToMatch(pszMask, uiLenMask);
1326           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1327           if ( pMatch == NULL )
1328             return FALSE;
1329
1330           // -1 to compensate "++" in the loop
1331           pszTxt = pMatch + uiLenMask - 1;
1332           pszMask += uiLenMask - 1;
1333         }
1334         break;
1335
1336       default:
1337         if ( *pszMask != *pszTxt )
1338           return FALSE;
1339         break;
1340     }
1341   }
1342
1343   // match only if nothing left
1344   return *pszTxt == _T('\0');
1345 }
1346
1347 // Count the number of chars
1348 int wxString::Freq(wxChar ch) const
1349 {
1350     int count = 0;
1351     int len = Len();
1352     for (int i = 0; i < len; i++)
1353     {
1354         if (GetChar(i) == ch)
1355             count ++;
1356     }
1357     return count;
1358 }
1359
1360 // convert to upper case, return the copy of the string
1361 wxString wxString::Upper() const
1362 { wxString s(*this); return s.MakeUpper(); }
1363
1364 // convert to lower case, return the copy of the string
1365 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1366
1367 int wxString::sprintf(const wxChar *pszFormat, ...)
1368   {
1369     va_list argptr;
1370     va_start(argptr, pszFormat);
1371     int iLen = PrintfV(pszFormat, argptr);
1372     va_end(argptr);
1373     return iLen;
1374   }
1375
1376 // ---------------------------------------------------------------------------
1377 // standard C++ library string functions
1378 // ---------------------------------------------------------------------------
1379 #ifdef  wxSTD_STRING_COMPATIBILITY
1380
1381 wxString& wxString::insert(size_t nPos, const wxString& str)
1382 {
1383   wxASSERT( str.GetStringData()->IsValid() );
1384   wxASSERT( nPos <= Len() );
1385
1386   if ( !str.IsEmpty() ) {
1387     wxString strTmp;
1388     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1389     wxStrncpy(pc, c_str(), nPos);
1390     wxStrcpy(pc + nPos, str);
1391     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1392     strTmp.UngetWriteBuf();
1393     *this = strTmp;
1394   }
1395
1396   return *this;
1397 }
1398
1399 size_t wxString::find(const wxString& str, size_t nStart) const
1400 {
1401   wxASSERT( str.GetStringData()->IsValid() );
1402   wxASSERT( nStart <= Len() );
1403
1404   const wxChar *p = wxStrstr(c_str() + nStart, str);
1405
1406   return p == NULL ? npos : p - c_str();
1407 }
1408
1409 // VC++ 1.5 can't cope with the default argument in the header.
1410 #if !defined(__VISUALC__) || defined(__WIN32__)
1411 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1412 {
1413   return find(wxString(sz, n == npos ? 0 : n), nStart);
1414 }
1415 #endif // VC++ 1.5
1416
1417 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1418 #if !defined(__BORLANDC__)
1419 size_t wxString::find(wxChar ch, size_t nStart) const
1420 {
1421   wxASSERT( nStart <= Len() );
1422
1423   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1424
1425   return p == NULL ? npos : p - c_str();
1426 }
1427 #endif
1428
1429 size_t wxString::rfind(const wxString& str, size_t nStart) const
1430 {
1431   wxASSERT( str.GetStringData()->IsValid() );
1432   wxASSERT( nStart <= Len() );
1433
1434   // TODO could be made much quicker than that
1435   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1436   while ( p >= c_str() + str.Len() ) {
1437     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1438       return p - str.Len() - c_str();
1439     p--;
1440   }
1441
1442   return npos;
1443 }
1444
1445 // VC++ 1.5 can't cope with the default argument in the header.
1446 #if !defined(__VISUALC__) || defined(__WIN32__)
1447 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1448 {
1449     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1450 }
1451
1452 size_t wxString::rfind(wxChar ch, size_t nStart) const
1453 {
1454     if ( nStart == npos )
1455     {
1456         nStart = Len();
1457     }
1458     else
1459     {
1460         wxASSERT( nStart <= Len() );
1461     }
1462
1463     const wxChar *p = wxStrrchr(c_str(), ch);
1464
1465     if ( p == NULL )
1466         return npos;
1467
1468     size_t result = p - c_str();
1469     return ( result > nStart ) ? npos : result;
1470 }
1471 #endif // VC++ 1.5
1472
1473 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1474 {
1475     const wxChar *start = c_str() + nStart;
1476     const wxChar *firstOf = wxStrpbrk(start, sz);
1477     if ( firstOf )
1478         return firstOf - start;
1479     else
1480         return npos;
1481 }
1482
1483 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1484 {
1485     if ( nStart == npos )
1486     {
1487         nStart = Len();
1488     }
1489     else
1490     {
1491         wxASSERT( nStart <= Len() );
1492     }
1493
1494     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1495     {
1496         if ( wxStrchr(sz, *p) )
1497             return p - c_str();
1498     }
1499
1500     return npos;
1501 }
1502
1503 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1504 {
1505     if ( nStart == npos )
1506     {
1507         nStart = Len();
1508     }
1509     else
1510     {
1511         wxASSERT( nStart <= Len() );
1512     }
1513
1514     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1515     if ( nAccept >= length() - nStart )
1516         return npos;
1517     else
1518         return nAccept;
1519 }
1520
1521 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1522 {
1523     wxASSERT( nStart <= Len() );
1524
1525     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1526     {
1527         if ( *p != ch )
1528             return p - c_str();
1529     }
1530
1531     return npos;
1532 }
1533
1534 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1535 {
1536     if ( nStart == npos )
1537     {
1538         nStart = Len();
1539     }
1540     else
1541     {
1542         wxASSERT( nStart <= Len() );
1543     }
1544
1545     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1546     {
1547         if ( !wxStrchr(sz, *p) )
1548             return p - c_str();
1549     }
1550
1551     return npos;
1552 }
1553
1554 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1555 {
1556     if ( nStart == npos )
1557     {
1558         nStart = Len();
1559     }
1560     else
1561     {
1562         wxASSERT( nStart <= Len() );
1563     }
1564
1565     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1566     {
1567         if ( *p != ch )
1568             return p - c_str();
1569     }
1570
1571     return npos;
1572 }
1573
1574 wxString wxString::substr(size_t nStart, size_t nLen) const
1575 {
1576   // npos means 'take all'
1577   if ( nLen == npos )
1578     nLen = 0;
1579
1580   wxASSERT( nStart + nLen <= Len() );
1581
1582   return wxString(c_str() + nStart, nLen == npos ? 0 : nLen);
1583 }
1584
1585 wxString& wxString::erase(size_t nStart, size_t nLen)
1586 {
1587   wxString strTmp(c_str(), nStart);
1588   if ( nLen != npos ) {
1589     wxASSERT( nStart + nLen <= Len() );
1590
1591     strTmp.append(c_str() + nStart + nLen);
1592   }
1593
1594   *this = strTmp;
1595   return *this;
1596 }
1597
1598 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1599 {
1600   wxASSERT( nStart + nLen <= wxStrlen(sz) );
1601
1602   wxString strTmp;
1603   if ( nStart != 0 )
1604     strTmp.append(c_str(), nStart);
1605   strTmp += sz;
1606   strTmp.append(c_str() + nStart + nLen);
1607
1608   *this = strTmp;
1609   return *this;
1610 }
1611
1612 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1613 {
1614   return replace(nStart, nLen, wxString(ch, nCount));
1615 }
1616
1617 wxString& wxString::replace(size_t nStart, size_t nLen,
1618                             const wxString& str, size_t nStart2, size_t nLen2)
1619 {
1620   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1621 }
1622
1623 wxString& wxString::replace(size_t nStart, size_t nLen,
1624                         const wxChar* sz, size_t nCount)
1625 {
1626   return replace(nStart, nLen, wxString(sz, nCount));
1627 }
1628
1629 #endif  //std::string compatibility
1630
1631 // ============================================================================
1632 // ArrayString
1633 // ============================================================================
1634
1635 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1636 #define   ARRAY_MAXSIZE_INCREMENT       4096
1637 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1638   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1639 #endif
1640
1641 #define   STRING(p)   ((wxString *)(&(p)))
1642
1643 // ctor
1644 wxArrayString::wxArrayString()
1645 {
1646   m_nSize  =
1647   m_nCount = 0;
1648   m_pItems = (wxChar **) NULL;
1649 }
1650
1651 // copy ctor
1652 wxArrayString::wxArrayString(const wxArrayString& src)
1653 {
1654   m_nSize  =
1655   m_nCount = 0;
1656   m_pItems = (wxChar **) NULL;
1657
1658   *this = src;
1659 }
1660
1661 // assignment operator
1662 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1663 {
1664   if ( m_nSize > 0 )
1665     Clear();
1666
1667   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1668     Alloc(src.m_nCount);
1669
1670   // we can't just copy the pointers here because otherwise we would share
1671   // the strings with another array
1672   for ( size_t n = 0; n < src.m_nCount; n++ )
1673     Add(src[n]);
1674
1675   if ( m_nCount != 0 )
1676     memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(wxChar *));
1677
1678   return *this;
1679 }
1680
1681 // grow the array
1682 void wxArrayString::Grow()
1683 {
1684   // only do it if no more place
1685   if( m_nCount == m_nSize ) {
1686     if( m_nSize == 0 ) {
1687       // was empty, alloc some memory
1688       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1689       m_pItems = new wxChar *[m_nSize];
1690     }
1691     else {
1692       // otherwise when it's called for the first time, nIncrement would be 0
1693       // and the array would never be expanded
1694 #if defined(__VISAGECPP__)
1695       int                           array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1696       wxASSERT( array_size != 0 );
1697 #else
1698       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1699 #endif
1700
1701       // add 50% but not too much
1702       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1703                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1704       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1705         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1706       m_nSize += nIncrement;
1707       wxChar **pNew = new wxChar *[m_nSize];
1708
1709       // copy data to new location
1710       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1711
1712       // delete old memory (but do not release the strings!)
1713       wxDELETEA(m_pItems);
1714
1715       m_pItems = pNew;
1716     }
1717   }
1718 }
1719
1720 void wxArrayString::Free()
1721 {
1722   for ( size_t n = 0; n < m_nCount; n++ ) {
1723     STRING(m_pItems[n])->GetStringData()->Unlock();
1724   }
1725 }
1726
1727 // deletes all the strings from the list
1728 void wxArrayString::Empty()
1729 {
1730   Free();
1731
1732   m_nCount = 0;
1733 }
1734
1735 // as Empty, but also frees memory
1736 void wxArrayString::Clear()
1737 {
1738   Free();
1739
1740   m_nSize  =
1741   m_nCount = 0;
1742
1743   wxDELETEA(m_pItems);
1744 }
1745
1746 // dtor
1747 wxArrayString::~wxArrayString()
1748 {
1749   Free();
1750
1751   wxDELETEA(m_pItems);
1752 }
1753
1754 // pre-allocates memory (frees the previous data!)
1755 void wxArrayString::Alloc(size_t nSize)
1756 {
1757   wxASSERT( nSize > 0 );
1758
1759   // only if old buffer was not big enough
1760   if ( nSize > m_nSize ) {
1761     Free();
1762     wxDELETEA(m_pItems);
1763     m_pItems = new wxChar *[nSize];
1764     m_nSize  = nSize;
1765   }
1766
1767   m_nCount = 0;
1768 }
1769
1770 // minimizes the memory usage by freeing unused memory
1771 void wxArrayString::Shrink()
1772 {
1773   // only do it if we have some memory to free
1774   if( m_nCount < m_nSize ) {
1775     // allocates exactly as much memory as we need
1776     wxChar **pNew = new wxChar *[m_nCount];
1777
1778     // copy data to new location
1779     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1780     delete [] m_pItems;
1781     m_pItems = pNew;
1782   }
1783 }
1784
1785 // searches the array for an item (forward or backwards)
1786 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1787 {
1788   if ( bFromEnd ) {
1789     if ( m_nCount > 0 ) {
1790       size_t ui = m_nCount;
1791       do {
1792         if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1793           return ui;
1794       }
1795       while ( ui != 0 );
1796     }
1797   }
1798   else {
1799     for( size_t ui = 0; ui < m_nCount; ui++ ) {
1800       if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1801         return ui;
1802     }
1803   }
1804
1805   return wxNOT_FOUND;
1806 }
1807
1808 // add item at the end
1809 void wxArrayString::Add(const wxString& str)
1810 {
1811   wxASSERT( str.GetStringData()->IsValid() );
1812
1813   Grow();
1814
1815   // the string data must not be deleted!
1816   str.GetStringData()->Lock();
1817   m_pItems[m_nCount++] = (wxChar *)str.c_str();
1818 }
1819
1820 // add item at the given position
1821 void wxArrayString::Insert(const wxString& str, size_t nIndex)
1822 {
1823   wxASSERT( str.GetStringData()->IsValid() );
1824
1825   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Insert") );
1826
1827   Grow();
1828
1829   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
1830           (m_nCount - nIndex)*sizeof(wxChar *));
1831
1832   str.GetStringData()->Lock();
1833   m_pItems[nIndex] = (wxChar *)str.c_str();
1834
1835   m_nCount++;
1836 }
1837
1838 // removes item from array (by index)
1839 void wxArrayString::Remove(size_t nIndex)
1840 {
1841   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Remove") );
1842
1843   // release our lock
1844   Item(nIndex).GetStringData()->Unlock();
1845
1846   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
1847           (m_nCount - nIndex - 1)*sizeof(wxChar *));
1848   m_nCount--;
1849 }
1850
1851 // removes item from array (by value)
1852 void wxArrayString::Remove(const wxChar *sz)
1853 {
1854   int iIndex = Index(sz);
1855
1856   wxCHECK_RET( iIndex != wxNOT_FOUND,
1857                _("removing inexistent element in wxArrayString::Remove") );
1858
1859   Remove(iIndex);
1860 }
1861
1862 // ----------------------------------------------------------------------------
1863 // sorting
1864 // ----------------------------------------------------------------------------
1865
1866 // we can only sort one array at a time with the quick-sort based
1867 // implementation
1868 #if wxUSE_THREADS
1869   // need a critical section to protect access to gs_compareFunction and
1870   // gs_sortAscending variables
1871   static wxCriticalSection *gs_critsectStringSort = NULL;
1872
1873   // call this before the value of the global sort vars is changed/after
1874   // you're finished with them
1875   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
1876                            gs_critsectStringSort = new wxCriticalSection;     \
1877                            gs_critsectStringSort->Enter()
1878   #define END_SORT()       gs_critsectStringSort->Leave();                    \
1879                            delete gs_critsectStringSort;                      \
1880                            gs_critsectStringSort = NULL
1881 #else // !threads
1882   #define START_SORT()
1883   #define END_SORT()
1884 #endif // wxUSE_THREADS
1885
1886 // function to use for string comparaison
1887 static wxArrayString::CompareFunction gs_compareFunction = NULL;
1888
1889 // if we don't use the compare function, this flag tells us if we sort the
1890 // array in ascending or descending order
1891 static bool gs_sortAscending = TRUE;
1892
1893 // function which is called by quick sort
1894 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
1895 {
1896   wxString *strFirst = (wxString *)first;
1897   wxString *strSecond = (wxString *)second;
1898
1899   if ( gs_compareFunction ) {
1900     return gs_compareFunction(*strFirst, *strSecond);
1901   }
1902   else {
1903     // maybe we should use wxStrcoll
1904     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
1905
1906     return gs_sortAscending ? result : -result;
1907   }
1908 }
1909
1910 // sort array elements using passed comparaison function
1911 void wxArrayString::Sort(CompareFunction compareFunction)
1912 {
1913   START_SORT();
1914
1915   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1916   gs_compareFunction = compareFunction;
1917
1918   DoSort();
1919
1920   END_SORT();
1921 }
1922
1923 void wxArrayString::Sort(bool reverseOrder)
1924 {
1925   START_SORT();
1926
1927   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1928   gs_sortAscending = !reverseOrder;
1929
1930   DoSort();
1931
1932   END_SORT();
1933 }
1934
1935 void wxArrayString::DoSort()
1936 {
1937   // just sort the pointers using qsort() - of course it only works because
1938   // wxString() *is* a pointer to its data
1939   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
1940 }
1941
1942 // ============================================================================
1943 // MBConv
1944 // ============================================================================
1945
1946 WXDLLEXPORT_DATA(wxMBConv *) wxConvCurrent = &wxConvLibc;
1947 #if !wxUSE_WCHAR_T
1948 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc, wxConvFile;
1949 #endif
1950
1951 #if wxUSE_WCHAR_T
1952
1953 // ----------------------------------------------------------------------------
1954 // standard libc conversion
1955 // ----------------------------------------------------------------------------
1956
1957 WXDLLEXPORT_DATA(wxMBConv) wxConvLibc;
1958
1959 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1960 {
1961   return wxMB2WC(buf, psz, n);
1962 }
1963
1964 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1965 {
1966   return wxWC2MB(buf, psz, n);
1967 }
1968
1969 // ----------------------------------------------------------------------------
1970 // standard file conversion
1971 // ----------------------------------------------------------------------------
1972
1973 WXDLLEXPORT_DATA(wxMBConvFile) wxConvFile;
1974
1975 // just use the libc conversion for now
1976 size_t wxMBConvFile::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1977 {
1978   return wxMB2WC(buf, psz, n);
1979 }
1980
1981 size_t wxMBConvFile::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1982 {
1983   return wxWC2MB(buf, psz, n);
1984 }
1985
1986 // ----------------------------------------------------------------------------
1987 // standard gdk conversion
1988 // ----------------------------------------------------------------------------
1989
1990 #ifdef __WXGTK12__
1991 WXDLLEXPORT_DATA(wxMBConvGdk) wxConvGdk;
1992
1993 #include <gdk/gdk.h>
1994
1995 size_t wxMBConvGdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1996 {
1997   if (buf) {
1998     return gdk_mbstowcs((GdkWChar *)buf, psz, n);
1999   } else {
2000     GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
2001     size_t len = gdk_mbstowcs(nbuf, psz, n);
2002     delete [] nbuf;
2003     return len;
2004   }
2005 }
2006
2007 size_t wxMBConvGdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2008 {
2009   char *mbstr = gdk_wcstombs((GdkWChar *)psz);
2010   size_t len = mbstr ? strlen(mbstr) : 0;
2011   if (buf) {
2012     if (len > n) len = n;
2013     memcpy(buf, psz, len);
2014     if (len < n) buf[len] = 0;
2015   }
2016   return len;
2017 }
2018 #endif // GTK > 1.0
2019
2020 // ----------------------------------------------------------------------------
2021 // UTF-7
2022 // ----------------------------------------------------------------------------
2023
2024 WXDLLEXPORT_DATA(wxMBConvUTF7) wxConvUTF7;
2025
2026 #if 0
2027 static char utf7_setD[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
2028                         "abcdefghijklmnopqrstuvwxyz"
2029                         "0123456789'(),-./:?";
2030 static char utf7_setO[]="!\"#$%&*;<=>@[]^_`{|}";
2031 static char utf7_setB[]="ABCDEFGHIJKLMNOPQRSTUVWXYZ"
2032                         "abcdefghijklmnopqrstuvwxyz"
2033                         "0123456789+/";
2034 #endif
2035
2036 // TODO: write actual implementations of UTF-7 here
2037 size_t wxMBConvUTF7::MB2WC(wchar_t * WXUNUSED(buf),
2038                            const char * WXUNUSED(psz),
2039                            size_t WXUNUSED(n)) const
2040 {
2041   return 0;
2042 }
2043
2044 size_t wxMBConvUTF7::WC2MB(char * WXUNUSED(buf),
2045                            const wchar_t * WXUNUSED(psz),
2046                            size_t WXUNUSED(n)) const
2047 {
2048   return 0;
2049 }
2050
2051 // ----------------------------------------------------------------------------
2052 // UTF-8
2053 // ----------------------------------------------------------------------------
2054
2055 WXDLLEXPORT_DATA(wxMBConvUTF8) wxConvUTF8;
2056
2057 static unsigned long utf8_max[]={0x7f,0x7ff,0xffff,0x1fffff,0x3ffffff,0x7fffffff,0xffffffff};
2058
2059 size_t wxMBConvUTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2060 {
2061   size_t len = 0;
2062
2063   while (*psz && ((!buf) || (len<n))) {
2064     unsigned char cc=*psz++, fc=cc;
2065     unsigned cnt;
2066     for (cnt=0; fc&0x80; cnt++) fc<<=1;
2067     if (!cnt) {
2068       // plain ASCII char
2069       if (buf) *buf++=cc;
2070       len++;
2071     } else {
2072       cnt--;
2073       if (!cnt) {
2074         // invalid UTF-8 sequence
2075         return (size_t)-1;
2076       } else {
2077         unsigned ocnt=cnt-1;
2078         unsigned long res=cc&(0x3f>>cnt);
2079         while (cnt--) {
2080           cc = *psz++;
2081           if ((cc&0xC0)!=0x80) {
2082             // invalid UTF-8 sequence
2083             return (size_t)-1;
2084           }
2085           res=(res<<6)|(cc&0x3f);
2086         }
2087         if (res<=utf8_max[ocnt]) {
2088           // illegal UTF-8 encoding
2089           return (size_t)-1;
2090         }
2091         if (buf) *buf++=res;
2092         len++;
2093       }
2094     }
2095   }
2096   if (buf && (len<n)) *buf = 0;
2097   return len;
2098 }
2099
2100 size_t wxMBConvUTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2101 {
2102   size_t len = 0;
2103
2104   while (*psz && ((!buf) || (len<n))) {
2105     unsigned long cc=(*psz++)&0x7fffffff;
2106     unsigned cnt;
2107     for (cnt=0; cc>utf8_max[cnt]; cnt++);
2108     if (!cnt) {
2109       // plain ASCII char
2110       if (buf) *buf++=cc;
2111       len++;
2112     } else {
2113       len+=cnt+1;
2114       if (buf) {
2115         *buf++=(-128>>cnt)|((cc>>(cnt*6))&(0x3f>>cnt));
2116         while (cnt--)
2117           *buf++=0x80|((cc>>(cnt*6))&0x3f);
2118       }
2119     }
2120   }
2121   if (buf && (len<n)) *buf = 0;
2122   return len;
2123 }
2124
2125 // ----------------------------------------------------------------------------
2126 // specified character set
2127 // ----------------------------------------------------------------------------
2128
2129 class wxCharacterSet
2130 {
2131 public:
2132   wxArrayString names;
2133   wchar_t *data;
2134 };
2135
2136 #ifndef WX_PRECOMP
2137   #include "wx/dynarray.h"
2138   #include "wx/filefn.h"
2139   #include "wx/textfile.h"
2140   #include "wx/tokenzr.h"
2141   #include "wx/utils.h"
2142 #endif
2143
2144 WX_DECLARE_OBJARRAY(wxCharacterSet, wxCSArray);
2145 #include "wx/arrimpl.cpp"
2146 WX_DEFINE_OBJARRAY(wxCSArray);
2147
2148 static wxCSArray wxCharsets;
2149
2150 static void wxLoadCharacterSets(void)
2151 {
2152   static bool already_loaded = FALSE;
2153
2154   if (already_loaded) return;
2155
2156   already_loaded = TRUE;
2157 #if defined(__UNIX__) && wxUSE_TEXTFILE
2158   // search through files in /usr/share/i18n/charmaps
2159   wxString fname;
2160   for (fname = ::wxFindFirstFile(_T("/usr/share/i18n/charmaps/*"));
2161        !fname.IsEmpty();
2162        fname = ::wxFindNextFile()) {
2163     wxTextFile cmap(fname);
2164     if (cmap.Open()) {
2165       wxCharacterSet *cset = new wxCharacterSet;
2166       wxString comchar,escchar;
2167       bool in_charset = FALSE;
2168
2169       // wxFprintf(stderr,_T("Loaded: %s\n"),fname.c_str());
2170
2171       wxString line;
2172       for (line = cmap.GetFirstLine();
2173            !cmap.Eof();
2174            line = cmap.GetNextLine()) {
2175         // wxFprintf(stderr,_T("line contents: %s\n"),line.c_str());
2176         wxStringTokenizer token(line);
2177         wxString cmd = token.GetNextToken();
2178         if (cmd == comchar) {
2179           if (token.GetNextToken() == _T("alias"))
2180             cset->names.Add(token.GetNextToken());
2181         }
2182         else if (cmd == _T("<code_set_name>"))
2183           cset->names.Add(token.GetNextToken());
2184         else if (cmd == _T("<comment_char>"))
2185           comchar = token.GetNextToken();
2186         else if (cmd == _T("<escape_char>"))
2187           escchar = token.GetNextToken();
2188         else if (cmd == _T("<mb_cur_min>")) {
2189           delete cset;
2190           cset = (wxCharacterSet *) NULL;
2191           break; // we don't support multibyte charsets ourselves (yet)
2192         }
2193         else if (cmd == _T("CHARMAP")) {
2194           cset->data = (wchar_t *)calloc(256, sizeof(wchar_t));
2195           in_charset = TRUE;
2196         }
2197         else if (cmd == _T("END")) {
2198           if (token.GetNextToken() == _T("CHARMAP"))
2199             in_charset = FALSE;
2200         }
2201         else if (in_charset) {
2202           // format: <NUL> /x00 <U0000> NULL (NUL)
2203           //         <A>   /x41 <U0041> LATIN CAPITAL LETTER A
2204           wxString hex = token.GetNextToken();
2205           // skip whitespace (why doesn't wxStringTokenizer do this?)
2206           while (wxIsEmpty(hex) && token.HasMoreTokens()) hex = token.GetNextToken();
2207           wxString uni = token.GetNextToken();
2208           // skip whitespace again
2209           while (wxIsEmpty(uni) && token.HasMoreTokens()) uni = token.GetNextToken();
2210           if ((hex.Len() > 2) && (wxString(hex.GetChar(0)) == escchar) && (hex.GetChar(1) == _T('x')) &&
2211               (uni.Left(2) == _T("<U"))) {
2212             hex.MakeUpper(); uni.MakeUpper();
2213             int pos = ::wxHexToDec(hex.Mid(2,2));
2214             if (pos>=0) {
2215               unsigned long uni1 = ::wxHexToDec(uni.Mid(2,2));
2216               unsigned long uni2 = ::wxHexToDec(uni.Mid(4,2));
2217               cset->data[pos] = (uni1 << 16) | uni2;
2218               // wxFprintf(stderr,_T("char %02x mapped to %04x (%c)\n"),pos,cset->data[pos],cset->data[pos]);
2219             }
2220           }
2221         }
2222       }
2223       if (cset) {
2224         cset->names.Shrink();
2225         wxCharsets.Add(cset);
2226       }
2227     }
2228   }
2229 #endif
2230   wxCharsets.Shrink();
2231 }
2232
2233 static wxCharacterSet *wxFindCharacterSet(const wxChar *charset)
2234 {
2235   if (!charset) return (wxCharacterSet *)NULL;
2236   wxLoadCharacterSets();
2237   for (size_t n=0; n<wxCharsets.GetCount(); n++)
2238     if (wxCharsets[n].names.Index(charset) != wxNOT_FOUND)
2239       return &(wxCharsets[n]);
2240   return (wxCharacterSet *)NULL;
2241 }
2242
2243 WXDLLEXPORT_DATA(wxCSConv) wxConvLocal((const wxChar *)NULL);
2244
2245 wxCSConv::wxCSConv(const wxChar *charset)
2246 {
2247   m_name = (wxChar *) NULL;
2248   m_cset = (wxCharacterSet *) NULL;
2249   m_deferred = TRUE;
2250   SetName(charset);
2251 }
2252
2253 wxCSConv::~wxCSConv()
2254 {
2255   if (m_name) free(m_name);
2256 }
2257
2258 void wxCSConv::SetName(const wxChar *charset)
2259 {
2260   if (charset) {
2261 #ifdef __UNIX__
2262     // first, convert the character set name to standard form
2263     wxString codeset;
2264     if (wxString(charset,3).CmpNoCase(_T("ISO")) == 0) {
2265       // make sure it's represented in the standard form: ISO_8859-1
2266       codeset = _T("ISO_");
2267       charset += 3;
2268       if ((*charset == _T('-')) || (*charset == _T('_'))) charset++;
2269       if (wxStrlen(charset)>4) {
2270         if (wxString(charset,4) == _T("8859")) {
2271           codeset << _T("8859-");
2272           if (*charset == _T('-')) charset++;
2273         }
2274       }
2275     }
2276     codeset << charset;
2277     codeset.MakeUpper();
2278     m_name = wxStrdup(codeset.c_str());
2279     m_deferred = TRUE;
2280 #endif
2281   }
2282 }
2283
2284 void wxCSConv::LoadNow()
2285 {
2286 //  wxPrintf(_T("Conversion request\n"));
2287   if (m_deferred) {
2288     if (!m_name) {
2289 #ifdef __UNIX__
2290       wxChar *lang = wxGetenv(_T("LANG"));
2291       wxChar *dot = lang ? wxStrchr(lang, _T('.')) : (wxChar *)NULL;
2292       if (dot) SetName(dot+1);
2293 #endif
2294     }
2295     m_cset = wxFindCharacterSet(m_name);
2296     m_deferred = FALSE;
2297   }
2298 }
2299
2300 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2301 {
2302   ((wxCSConv *)this)->LoadNow(); // discard constness
2303   if (buf) {
2304     if (m_cset) {
2305       for (size_t c=0; c<n; c++)
2306         buf[c] = m_cset->data[(unsigned char)(psz[c])];
2307     } else {
2308       // latin-1 (direct)
2309       for (size_t c=0; c<n; c++)
2310         buf[c] = (unsigned char)(psz[c]);
2311     }
2312     return n;
2313   }
2314   return strlen(psz);
2315 }
2316
2317 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2318 {
2319   ((wxCSConv *)this)->LoadNow(); // discard constness
2320   if (buf) {
2321     if (m_cset) {
2322       for (size_t c=0; c<n; c++) {
2323         size_t n;
2324         for (n=0; (n<256) && (m_cset->data[n] != psz[c]); n++);
2325         buf[c] = (n>0xff) ? '?' : n;
2326       }
2327     } else {
2328       // latin-1 (direct)
2329       for (size_t c=0; c<n; c++)
2330         buf[c] = (psz[c]>0xff) ? '?' : psz[c];
2331     }
2332     return n;
2333   }
2334   return wcslen(psz);
2335 }
2336
2337 #endif//wxUSE_WCHAR_T
2338
2339 #if wxUSE_WCHAR_T
2340 const wxWCharBuffer wxMBConv::cMB2WC(const char *psz) const
2341     {
2342       if (psz) {
2343         size_t nLen = MB2WC((wchar_t *) NULL, psz, 0);
2344         wxWCharBuffer buf(nLen);
2345         MB2WC(WCSTRINGCAST buf, psz, nLen);
2346         return buf;
2347       } else return wxWCharBuffer((wchar_t *) NULL);
2348     }
2349
2350 const wxCharBuffer wxMBConv::cWC2MB(const wchar_t *psz) const
2351     {
2352       if (psz) {
2353         size_t nLen = WC2MB((char *) NULL, psz, 0);
2354         wxCharBuffer buf(nLen);
2355         WC2MB(MBSTRINGCAST buf, psz, nLen);
2356         return buf;
2357       } else return wxCharBuffer((char *) NULL);
2358     }
2359
2360 #endif//wxUSE_WCHAR_T
2361