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 #if wxUSE_THREADS
  39   #include <wx/thread.h>
  40 #endif
  41 #endif
  42
  43 #include <ctype.h>
  44 #include <string.h>
  45 #include <stdlib.h>
  46
  47 #ifdef __SALFORDC__
  48   #include <clib.h>
  49 #endif
  50
  51 #if wxUSE_WCSRTOMBS
  52   #include <wchar.h>    // for wcsrtombs(), see comments where it's used
  53 #endif // GNU
  54
  55 #ifdef  WXSTRING_IS_WXOBJECT
  56   IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
  57 #endif  //WXSTRING_IS_WXOBJECT
  58
  59 // allocating extra space for each string consumes more memory but speeds up
  60 // the concatenation operations (nLen is the current string's length)
  61 // NB: EXTRA_ALLOC must be >= 0!
  62 #define EXTRA_ALLOC       (19 - nLen % 16)
  63
  64 // ---------------------------------------------------------------------------
  65 // static class variables definition
  66 // ---------------------------------------------------------------------------
  67
  68 #ifdef  wxSTD_STRING_COMPATIBILITY
  69   const size_t wxString::npos = wxSTRING_MAXLEN;
  70 #endif // wxSTD_STRING_COMPATIBILITY
  71
  72 // ----------------------------------------------------------------------------
  73 // static data
  74 // ----------------------------------------------------------------------------
  75
  76 // for an empty string, GetStringData() will return this address: this
  77 // structure has the same layout as wxStringData and it's data() method will
  78 // return the empty string (dummy pointer)
  79 static const struct
  80 {
  81   wxStringData data;
  82   wxChar dummy;
  83 } g_strEmpty = { {-1, 0, 0}, _T('\0') };
  84
  85 // empty C style string: points to 'string data' byte of g_strEmpty
  86 extern const wxChar WXDLLEXPORT *g_szNul = &g_strEmpty.dummy;
  87
  88 // ----------------------------------------------------------------------------
  89 // conditional compilation
  90 // ----------------------------------------------------------------------------
  91
  92 // we want to find out if the current platform supports vsnprintf()-like
  93 // function: for Unix this is done with configure, for Windows we test the
  94 // compiler explicitly.
  95 #ifdef __WXMSW__
  96     #ifdef __VISUALC__
  97         #define wxVsnprintf     _vsnprintf
  98     #endif
  99 #else   // !Windows
 100     #ifdef HAVE_VSNPRINTF
 101         #define wxVsnprintf       vsnprintf
 102     #endif
 103 #endif  // Windows/!Windows
 104
 105 #ifndef wxVsnprintf
 106     // in this case we'll use vsprintf() (which is ANSI and thus should be
 107     // always available), but it's unsafe because it doesn't check for buffer
 108     // size - so give a warning
 109     #define wxVsnprintf(buffer,len,format,argptr) vsprintf(buffer,format, argptr)
 110
 111     #if defined(__VISUALC__)
 112         #pragma message("Using sprintf() because no snprintf()-like function defined")
 113     #elif defined(__GNUG__) && !defined(__UNIX__)
 114         #warning "Using sprintf() because no snprintf()-like function defined"
 115     #elif defined(__MWERKS__)
 116             #warning "Using sprintf() because no snprintf()-like function defined"
 117     #endif //compiler
 118 #endif // no vsnprintf
 119
 120 #ifdef _AIX
 121   // AIX has vsnprintf, but there's no prototype in the system headers.
 122   extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
 123 #endif
 124
 125 // ----------------------------------------------------------------------------
 126 // global functions
 127 // ----------------------------------------------------------------------------
 128
 129 #ifdef  wxSTD_STRING_COMPATIBILITY
 130
 131 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
 132 // iostream ones.
 133 //
 134 // ATTN: you can _not_ use both of these in the same program!
 135
 136 istream& operator>>(istream& is, wxString& WXUNUSED(str))
 137 {
 138 #if 0
 139   int w = is.width(0);
 140   if ( is.ipfx(0) ) {
 141     streambuf *sb = is.rdbuf();
 142     str.erase();
 143     while ( true ) {
 144       int ch = sb->sbumpc ();
 145       if ( ch == EOF ) {
 146         is.setstate(ios::eofbit);
 147         break;
 148       }
 149       else if ( isspace(ch) ) {
 150         sb->sungetc();
 151         break;
 152       }
 153
 154       str += ch;
 155       if ( --w == 1 )
 156         break;
 157     }
 158   }
 159
 160   is.isfx();
 161   if ( str.length() == 0 )
 162     is.setstate(ios::failbit);
 163 #endif
 164   return is;
 165 }
 166
 167 #endif  //std::string compatibility
 168
 169 // ----------------------------------------------------------------------------
 170 // private classes
 171 // ----------------------------------------------------------------------------
 172
 173 // this small class is used to gather statistics for performance tuning
 174 //#define WXSTRING_STATISTICS
 175 #ifdef  WXSTRING_STATISTICS
 176   class Averager
 177   {
 178   public:
 179     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 180    ~Averager()
 181    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 182
 183     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 184
 185   private:
 186     size_t m_nCount, m_nTotal;
 187     const char *m_sz;
 188   } g_averageLength("allocation size"),
 189     g_averageSummandLength("summand length"),
 190     g_averageConcatHit("hit probability in concat"),
 191     g_averageInitialLength("initial string length");
 192
 193   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 194 #else
 195   #define STATISTICS_ADD(av, val)
 196 #endif // WXSTRING_STATISTICS
 197
 198 // ===========================================================================
 199 // wxString class core
 200 // ===========================================================================
 201
 202 // ---------------------------------------------------------------------------
 203 // construction
 204 // ---------------------------------------------------------------------------
 205
 206 // constructs string of <nLength> copies of character <ch>
 207 wxString::wxString(wxChar ch, size_t nLength)
 208 {
 209   Init();
 210
 211   if ( nLength > 0 ) {
 212     AllocBuffer(nLength);
 213
 214 #if wxUSE_UNICODE
 215     // memset only works on char
 216     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 217 #else
 218     memset(m_pchData, ch, nLength);
 219 #endif
 220   }
 221 }
 222
 223 // takes nLength elements of psz starting at nPos
 224 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 225 {
 226   Init();
 227
 228   wxASSERT( nPos <= wxStrlen(psz) );
 229
 230   if ( nLength == wxSTRING_MAXLEN )
 231     nLength = wxStrlen(psz + nPos);
 232
 233   STATISTICS_ADD(InitialLength, nLength);
 234
 235   if ( nLength > 0 ) {
 236     // trailing '\0' is written in AllocBuffer()
 237     AllocBuffer(nLength);
 238     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 239   }
 240 }
 241
 242 #ifdef  wxSTD_STRING_COMPATIBILITY
 243
 244 // poor man's iterators are "void *" pointers
 245 wxString::wxString(const void *pStart, const void *pEnd)
 246 {
 247   InitWith((const wxChar *)pStart, 0,
 248            (const wxChar *)pEnd - (const wxChar *)pStart);
 249 }
 250
 251 #endif  //std::string compatibility
 252
 253 #if wxUSE_UNICODE
 254
 255 // from multibyte string
 256 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 257 {
 258   // first get necessary size
 259
 260   size_t nLen = conv.MB2WC((wchar_t *) NULL, psz, 0);
 261
 262   // nLength is number of *Unicode* characters here!
 263   if (nLen > nLength)
 264     nLen = nLength;
 265
 266   // empty?
 267   if ( nLen != 0 ) {
 268     AllocBuffer(nLen);
 269     conv.MB2WC(m_pchData, psz, nLen);
 270   }
 271   else {
 272     Init();
 273   }
 274 }
 275
 276 #else
 277
 278 // from wide string
 279 wxString::wxString(const wchar_t *pwz)
 280 {
 281   // first get necessary size
 282
 283   size_t nLen = wxWC2MB((char *) NULL, pwz, 0);
 284
 285   // empty?
 286   if ( nLen != 0 ) {
 287     AllocBuffer(nLen);
 288     wxWC2MB(m_pchData, pwz, nLen);
 289   }
 290   else {
 291     Init();
 292   }
 293 }
 294
 295 #endif
 296
 297 // ---------------------------------------------------------------------------
 298 // memory allocation
 299 // ---------------------------------------------------------------------------
 300
 301 // allocates memory needed to store a C string of length nLen
 302 void wxString::AllocBuffer(size_t nLen)
 303 {
 304   wxASSERT( nLen >  0         );    //
 305   wxASSERT( nLen <= INT_MAX-1 );    // max size (enough room for 1 extra)
 306
 307   STATISTICS_ADD(Length, nLen);
 308
 309   // allocate memory:
 310   // 1) one extra character for '\0' termination
 311   // 2) sizeof(wxStringData) for housekeeping info
 312   wxStringData* pData = (wxStringData*)
 313     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 314   pData->nRefs        = 1;
 315   pData->nDataLength  = nLen;
 316   pData->nAllocLength = nLen + EXTRA_ALLOC;
 317   m_pchData           = pData->data();  // data starts after wxStringData
 318   m_pchData[nLen]     = _T('\0');
 319 }
 320
 321 // must be called before changing this string
 322 void wxString::CopyBeforeWrite()
 323 {
 324   wxStringData* pData = GetStringData();
 325
 326   if ( pData->IsShared() ) {
 327     pData->Unlock();                // memory not freed because shared
 328     size_t nLen = pData->nDataLength;
 329     AllocBuffer(nLen);
 330     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 331   }
 332
 333   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 334 }
 335
 336 // must be called before replacing contents of this string
 337 void wxString::AllocBeforeWrite(size_t nLen)
 338 {
 339   wxASSERT( nLen != 0 );  // doesn't make any sense
 340
 341   // must not share string and must have enough space
 342   wxStringData* pData = GetStringData();
 343   if ( pData->IsShared() || (nLen > pData->nAllocLength) ) {
 344     // can't work with old buffer, get new one
 345     pData->Unlock();
 346     AllocBuffer(nLen);
 347   }
 348   else {
 349     // update the string length
 350     pData->nDataLength = nLen;
 351   }
 352
 353   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 354 }
 355
 356 // allocate enough memory for nLen characters
 357 void wxString::Alloc(size_t nLen)
 358 {
 359   wxStringData *pData = GetStringData();
 360   if ( pData->nAllocLength <= nLen ) {
 361     if ( pData->IsEmpty() ) {
 362       nLen += EXTRA_ALLOC;
 363
 364       wxStringData* pData = (wxStringData*)
 365         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 366       pData->nRefs = 1;
 367       pData->nDataLength = 0;
 368       pData->nAllocLength = nLen;
 369       m_pchData = pData->data();  // data starts after wxStringData
 370       m_pchData[0u] = _T('\0');
 371     }
 372     else if ( pData->IsShared() ) {
 373       pData->Unlock();                // memory not freed because shared
 374       size_t nOldLen = pData->nDataLength;
 375       AllocBuffer(nLen);
 376       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 377     }
 378     else {
 379       nLen += EXTRA_ALLOC;
 380
 381       wxStringData *p = (wxStringData *)
 382         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 383
 384       if ( p == NULL ) {
 385         // @@@ what to do on memory error?
 386         return;
 387       }
 388
 389       // it's not important if the pointer changed or not (the check for this
 390       // is not faster than assigning to m_pchData in all cases)
 391       p->nAllocLength = nLen;
 392       m_pchData = p->data();
 393     }
 394   }
 395   //else: we've already got enough
 396 }
 397
 398 // shrink to minimal size (releasing extra memory)
 399 void wxString::Shrink()
 400 {
 401   wxStringData *pData = GetStringData();
 402
 403   // this variable is unused in release build, so avoid the compiler warning by
 404   // just not declaring it
 405 #ifdef __WXDEBUG__
 406   void *p =
 407 #endif
 408   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 409
 410   wxASSERT( p != NULL );  // can't free memory?
 411   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 412 }
 413
 414 // get the pointer to writable buffer of (at least) nLen bytes
 415 wxChar *wxString::GetWriteBuf(size_t nLen)
 416 {
 417   AllocBeforeWrite(nLen);
 418
 419   wxASSERT( GetStringData()->nRefs == 1 );
 420   GetStringData()->Validate(FALSE);
 421
 422   return m_pchData;
 423 }
 424
 425 // put string back in a reasonable state after GetWriteBuf
 426 void wxString::UngetWriteBuf()
 427 {
 428   GetStringData()->nDataLength = wxStrlen(m_pchData);
 429   GetStringData()->Validate(TRUE);
 430 }
 431
 432 // ---------------------------------------------------------------------------
 433 // data access
 434 // ---------------------------------------------------------------------------
 435
 436 // all functions are inline in string.h
 437
 438 // ---------------------------------------------------------------------------
 439 // assignment operators
 440 // ---------------------------------------------------------------------------
 441
 442 // helper function: does real copy
 443 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 444 {
 445   if ( nSrcLen == 0 ) {
 446     Reinit();
 447   }
 448   else {
 449     AllocBeforeWrite(nSrcLen);
 450     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 451     GetStringData()->nDataLength = nSrcLen;
 452     m_pchData[nSrcLen] = _T('\0');
 453   }
 454 }
 455
 456 // assigns one string to another
 457 wxString& wxString::operator=(const wxString& stringSrc)
 458 {
 459   wxASSERT( stringSrc.GetStringData()->IsValid() );
 460
 461   // don't copy string over itself
 462   if ( m_pchData != stringSrc.m_pchData ) {
 463     if ( stringSrc.GetStringData()->IsEmpty() ) {
 464       Reinit();
 465     }
 466     else {
 467       // adjust references
 468       GetStringData()->Unlock();
 469       m_pchData = stringSrc.m_pchData;
 470       GetStringData()->Lock();
 471     }
 472   }
 473
 474   return *this;
 475 }
 476
 477 // assigns a single character
 478 wxString& wxString::operator=(wxChar ch)
 479 {
 480   AssignCopy(1, &ch);
 481   return *this;
 482 }
 483
 484 // assigns C string
 485 wxString& wxString::operator=(const wxChar *psz)
 486 {
 487   AssignCopy(wxStrlen(psz), psz);
 488   return *this;
 489 }
 490
 491 #if !wxUSE_UNICODE
 492
 493 // same as 'signed char' variant
 494 wxString& wxString::operator=(const unsigned char* psz)
 495 {
 496   *this = (const char *)psz;
 497   return *this;
 498 }
 499
 500 wxString& wxString::operator=(const wchar_t *pwz)
 501 {
 502   wxString str(pwz);
 503   *this = str;
 504   return *this;
 505 }
 506
 507 #endif
 508
 509 // ---------------------------------------------------------------------------
 510 // string concatenation
 511 // ---------------------------------------------------------------------------
 512
 513 // add something to this string
 514 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 515 {
 516   STATISTICS_ADD(SummandLength, nSrcLen);
 517
 518   // concatenating an empty string is a NOP
 519   if ( nSrcLen > 0 ) {
 520     wxStringData *pData = GetStringData();
 521     size_t nLen = pData->nDataLength;
 522     size_t nNewLen = nLen + nSrcLen;
 523
 524     // alloc new buffer if current is too small
 525     if ( pData->IsShared() ) {
 526       STATISTICS_ADD(ConcatHit, 0);
 527
 528       // we have to allocate another buffer
 529       wxStringData* pOldData = GetStringData();
 530       AllocBuffer(nNewLen);
 531       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 532       pOldData->Unlock();
 533     }
 534     else if ( nNewLen > pData->nAllocLength ) {
 535       STATISTICS_ADD(ConcatHit, 0);
 536
 537       // we have to grow the buffer
 538       Alloc(nNewLen);
 539     }
 540     else {
 541       STATISTICS_ADD(ConcatHit, 1);
 542
 543       // the buffer is already big enough
 544     }
 545
 546     // should be enough space
 547     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 548
 549     // fast concatenation - all is done in our buffer
 550     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 551
 552     m_pchData[nNewLen] = _T('\0');          // put terminating '\0'
 553     GetStringData()->nDataLength = nNewLen; // and fix the length
 554   }
 555   //else: the string to append was empty
 556 }
 557
 558 /*
 559  * concatenation functions come in 5 flavours:
 560  *  string + string
 561  *  char   + string      and      string + char
 562  *  C str  + string      and      string + C str
 563  */
 564
 565 wxString operator+(const wxString& string1, const wxString& string2)
 566 {
 567   wxASSERT( string1.GetStringData()->IsValid() );
 568   wxASSERT( string2.GetStringData()->IsValid() );
 569
 570   wxString s = string1;
 571   s += string2;
 572
 573   return s;
 574 }
 575
 576 wxString operator+(const wxString& string, wxChar ch)
 577 {
 578   wxASSERT( string.GetStringData()->IsValid() );
 579
 580   wxString s = string;
 581   s += ch;
 582
 583   return s;
 584 }
 585
 586 wxString operator+(wxChar ch, const wxString& string)
 587 {
 588   wxASSERT( string.GetStringData()->IsValid() );
 589
 590   wxString s = ch;
 591   s += string;
 592
 593   return s;
 594 }
 595
 596 wxString operator+(const wxString& string, const wxChar *psz)
 597 {
 598   wxASSERT( string.GetStringData()->IsValid() );
 599
 600   wxString s;
 601   s.Alloc(wxStrlen(psz) + string.Len());
 602   s = string;
 603   s += psz;
 604
 605   return s;
 606 }
 607
 608 wxString operator+(const wxChar *psz, const wxString& string)
 609 {
 610   wxASSERT( string.GetStringData()->IsValid() );
 611
 612   wxString s;
 613   s.Alloc(wxStrlen(psz) + string.Len());
 614   s = psz;
 615   s += string;
 616
 617   return s;
 618 }
 619
 620 // ===========================================================================
 621 // other common string functions
 622 // ===========================================================================
 623
 624 // ---------------------------------------------------------------------------
 625 // simple sub-string extraction
 626 // ---------------------------------------------------------------------------
 627
 628 // helper function: clone the data attached to this string
 629 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 630 {
 631   if ( nCopyLen == 0 ) {
 632     dest.Init();
 633   }
 634   else {
 635     dest.AllocBuffer(nCopyLen);
 636     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 637   }
 638 }
 639
 640 // extract string of length nCount starting at nFirst
 641 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 642 {
 643   wxStringData *pData = GetStringData();
 644   size_t nLen = pData->nDataLength;
 645
 646   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 647   if ( nCount == wxSTRING_MAXLEN )
 648   {
 649     nCount = nLen - nFirst;
 650   }
 651
 652   // out-of-bounds requests return sensible things
 653   if ( nFirst + nCount > nLen )
 654   {
 655     nCount = nLen - nFirst;
 656   }
 657
 658   if ( nFirst > nLen )
 659   {
 660     // AllocCopy() will return empty string
 661     nCount = 0;
 662   }
 663
 664   wxString dest;
 665   AllocCopy(dest, nCount, nFirst);
 666
 667   return dest;
 668 }
 669
 670 // extract nCount last (rightmost) characters
 671 wxString wxString::Right(size_t nCount) const
 672 {
 673   if ( nCount > (size_t)GetStringData()->nDataLength )
 674     nCount = GetStringData()->nDataLength;
 675
 676   wxString dest;
 677   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 678   return dest;
 679 }
 680
 681 // get all characters after the last occurence of ch
 682 // (returns the whole string if ch not found)
 683 wxString wxString::AfterLast(wxChar ch) const
 684 {
 685   wxString str;
 686   int iPos = Find(ch, TRUE);
 687   if ( iPos == wxNOT_FOUND )
 688     str = *this;
 689   else
 690     str = c_str() + iPos + 1;
 691
 692   return str;
 693 }
 694
 695 // extract nCount first (leftmost) characters
 696 wxString wxString::Left(size_t nCount) const
 697 {
 698   if ( nCount > (size_t)GetStringData()->nDataLength )
 699     nCount = GetStringData()->nDataLength;
 700
 701   wxString dest;
 702   AllocCopy(dest, nCount, 0);
 703   return dest;
 704 }
 705
 706 // get all characters before the first occurence of ch
 707 // (returns the whole string if ch not found)
 708 wxString wxString::BeforeFirst(wxChar ch) const
 709 {
 710   wxString str;
 711   for ( const wxChar *pc = m_pchData; *pc != _T('\0') && *pc != ch; pc++ )
 712     str += *pc;
 713
 714   return str;
 715 }
 716
 717 /// get all characters before the last occurence of ch
 718 /// (returns empty string if ch not found)
 719 wxString wxString::BeforeLast(wxChar ch) const
 720 {
 721   wxString str;
 722   int iPos = Find(ch, TRUE);
 723   if ( iPos != wxNOT_FOUND && iPos != 0 )
 724     str = wxString(c_str(), iPos);
 725
 726   return str;
 727 }
 728
 729 /// get all characters after the first occurence of ch
 730 /// (returns empty string if ch not found)
 731 wxString wxString::AfterFirst(wxChar ch) const
 732 {
 733   wxString str;
 734   int iPos = Find(ch);
 735   if ( iPos != wxNOT_FOUND )
 736     str = c_str() + iPos + 1;
 737
 738   return str;
 739 }
 740
 741 // replace first (or all) occurences of some substring with another one
 742 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 743 {
 744   size_t uiCount = 0;   // count of replacements made
 745
 746   size_t uiOldLen = wxStrlen(szOld);
 747
 748   wxString strTemp;
 749   const wxChar *pCurrent = m_pchData;
 750   const wxChar *pSubstr;
 751   while ( *pCurrent != _T('\0') ) {
 752     pSubstr = wxStrstr(pCurrent, szOld);
 753     if ( pSubstr == NULL ) {
 754       // strTemp is unused if no replacements were made, so avoid the copy
 755       if ( uiCount == 0 )
 756         return 0;
 757
 758       strTemp += pCurrent;    // copy the rest
 759       break;                  // exit the loop
 760     }
 761     else {
 762       // take chars before match
 763       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 764       strTemp += szNew;
 765       pCurrent = pSubstr + uiOldLen;  // restart after match
 766
 767       uiCount++;
 768
 769       // stop now?
 770       if ( !bReplaceAll ) {
 771         strTemp += pCurrent;    // copy the rest
 772         break;                  // exit the loop
 773       }
 774     }
 775   }
 776
 777   // only done if there were replacements, otherwise would have returned above
 778   *this = strTemp;
 779
 780   return uiCount;
 781 }
 782
 783 bool wxString::IsAscii() const
 784 {
 785   const wxChar *s = (const wxChar*) *this;
 786   while(*s){
 787     if(!isascii(*s)) return(FALSE);
 788     s++;
 789   }
 790   return(TRUE);
 791 }
 792
 793 bool wxString::IsWord() const
 794 {
 795   const wxChar *s = (const wxChar*) *this;
 796   while(*s){
 797     if(!wxIsalpha(*s)) return(FALSE);
 798     s++;
 799   }
 800   return(TRUE);
 801 }
 802
 803 bool wxString::IsNumber() const
 804 {
 805   const wxChar *s = (const wxChar*) *this;
 806   while(*s){
 807     if(!wxIsdigit(*s)) return(FALSE);
 808     s++;
 809   }
 810   return(TRUE);
 811 }
 812
 813 wxString wxString::Strip(stripType w) const
 814 {
 815     wxString s = *this;
 816     if ( w & leading ) s.Trim(FALSE);
 817     if ( w & trailing ) s.Trim(TRUE);
 818     return s;
 819 }
 820
 821 // ---------------------------------------------------------------------------
 822 // case conversion
 823 // ---------------------------------------------------------------------------
 824
 825 wxString& wxString::MakeUpper()
 826 {
 827   CopyBeforeWrite();
 828
 829   for ( wxChar *p = m_pchData; *p; p++ )
 830     *p = (wxChar)wxToupper(*p);
 831
 832   return *this;
 833 }
 834
 835 wxString& wxString::MakeLower()
 836 {
 837   CopyBeforeWrite();
 838
 839   for ( wxChar *p = m_pchData; *p; p++ )
 840     *p = (wxChar)wxTolower(*p);
 841
 842   return *this;
 843 }
 844
 845 // ---------------------------------------------------------------------------
 846 // trimming and padding
 847 // ---------------------------------------------------------------------------
 848
 849 // trims spaces (in the sense of isspace) from left or right side
 850 wxString& wxString::Trim(bool bFromRight)
 851 {
 852   // first check if we're going to modify the string at all
 853   if ( !IsEmpty() &&
 854        (
 855         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 856         (!bFromRight && wxIsspace(GetChar(0u)))
 857        )
 858      )
 859   {
 860     // ok, there is at least one space to trim
 861     CopyBeforeWrite();
 862
 863     if ( bFromRight )
 864     {
 865       // find last non-space character
 866       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
 867       while ( wxIsspace(*psz) && (psz >= m_pchData) )
 868         psz--;
 869
 870       // truncate at trailing space start
 871       *++psz = _T('\0');
 872       GetStringData()->nDataLength = psz - m_pchData;
 873     }
 874     else
 875     {
 876       // find first non-space character
 877       const wxChar *psz = m_pchData;
 878       while ( wxIsspace(*psz) )
 879         psz++;
 880
 881       // fix up data and length
 882       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
 883       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
 884       GetStringData()->nDataLength = nDataLength;
 885     }
 886   }
 887
 888   return *this;
 889 }
 890
 891 // adds nCount characters chPad to the string from either side
 892 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
 893 {
 894   wxString s(chPad, nCount);
 895
 896   if ( bFromRight )
 897     *this += s;
 898   else
 899   {
 900     s += *this;
 901     *this = s;
 902   }
 903
 904   return *this;
 905 }
 906
 907 // truncate the string
 908 wxString& wxString::Truncate(size_t uiLen)
 909 {
 910   if ( uiLen < Len() ) {
 911     CopyBeforeWrite();
 912
 913     *(m_pchData + uiLen) = _T('\0');
 914     GetStringData()->nDataLength = uiLen;
 915   }
 916   //else: nothing to do, string is already short enough
 917
 918   return *this;
 919 }
 920
 921 // ---------------------------------------------------------------------------
 922 // finding (return wxNOT_FOUND if not found and index otherwise)
 923 // ---------------------------------------------------------------------------
 924
 925 // find a character
 926 int wxString::Find(wxChar ch, bool bFromEnd) const
 927 {
 928   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
 929
 930   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 931 }
 932
 933 // find a sub-string (like strstr)
 934 int wxString::Find(const wxChar *pszSub) const
 935 {
 936   const wxChar *psz = wxStrstr(m_pchData, pszSub);
 937
 938   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
 939 }
 940
 941 // ---------------------------------------------------------------------------
 942 // stream-like operators
 943 // ---------------------------------------------------------------------------
 944 wxString& wxString::operator<<(int i)
 945 {
 946     wxString res;
 947     res.Printf(_T("%d"), i);
 948
 949     return (*this) << res;
 950 }
 951
 952 wxString& wxString::operator<<(float f)
 953 {
 954     wxString res;
 955     res.Printf(_T("%f"), f);
 956
 957     return (*this) << res;
 958 }
 959
 960 wxString& wxString::operator<<(double d)
 961 {
 962     wxString res;
 963     res.Printf(_T("%g"), d);
 964
 965     return (*this) << res;
 966 }
 967
 968 // ---------------------------------------------------------------------------
 969 // formatted output
 970 // ---------------------------------------------------------------------------
 971 int wxString::Printf(const wxChar *pszFormat, ...)
 972 {
 973   va_list argptr;
 974   va_start(argptr, pszFormat);
 975
 976   int iLen = PrintfV(pszFormat, argptr);
 977
 978   va_end(argptr);
 979
 980   return iLen;
 981 }
 982
 983 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
 984 {
 985   // static buffer to avoid dynamic memory allocation each time
 986   static char s_szScratch[1024];
 987 #if wxUSE_THREADS
 988   // protect the static buffer
 989   static wxCriticalSection critsect;
 990   wxCriticalSectionLocker lock(critsect);
 991 #endif
 992
 993 #if 1 // the new implementation
 994
 995   Reinit();
 996   for (size_t n = 0; pszFormat[n]; n++)
 997     if (pszFormat[n] == _T('%')) {
 998       static char s_szFlags[256] = "%";
 999       size_t flagofs = 1;
1000       bool adj_left = FALSE, in_prec = FALSE,
1001            prec_dot = FALSE, done = FALSE;
1002       int ilen = 0;
1003       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1004       do {
1005 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1006         switch (pszFormat[++n]) {
1007         case _T('\0'):
1008           done = TRUE;
1009           break;
1010         case _T('%'):
1011           *this += _T('%');
1012           done = TRUE;
1013           break;
1014         case _T('#'):
1015         case _T('0'):
1016         case _T(' '):
1017         case _T('+'):
1018         case _T('\''):
1019           CHECK_PREC
1020           s_szFlags[flagofs++] = pszFormat[n];
1021           break;
1022         case _T('-'):
1023           CHECK_PREC
1024           adj_left = TRUE;
1025           s_szFlags[flagofs++] = pszFormat[n];
1026           break;
1027         case _T('.'):
1028           CHECK_PREC
1029           in_prec = TRUE;
1030           prec_dot = FALSE;
1031           max_width = 0;
1032           // dot will be auto-added to s_szFlags if non-negative number follows
1033           break;
1034         case _T('h'):
1035           ilen = -1;
1036           CHECK_PREC
1037           s_szFlags[flagofs++] = pszFormat[n];
1038           break;
1039         case _T('l'):
1040           ilen = 1;
1041           CHECK_PREC
1042           s_szFlags[flagofs++] = pszFormat[n];
1043           break;
1044         case _T('q'):
1045         case _T('L'):
1046           ilen = 2;
1047           CHECK_PREC
1048           s_szFlags[flagofs++] = pszFormat[n];
1049           break;
1050         case _T('Z'):
1051           ilen = 3;
1052           CHECK_PREC
1053           s_szFlags[flagofs++] = pszFormat[n];
1054           break;
1055         case _T('*'):
1056           {
1057             int len = va_arg(argptr, int);
1058             if (in_prec) {
1059               if (len<0) break;
1060               CHECK_PREC
1061               max_width = len;
1062             } else {
1063               if (len<0) {
1064                 adj_left = !adj_left;
1065                 s_szFlags[flagofs++] = '-';
1066                 len = -len;
1067               }
1068               min_width = len;
1069             }
1070             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1071           }
1072           break;
1073         case _T('1'): case _T('2'): case _T('3'):
1074         case _T('4'): case _T('5'): case _T('6'):
1075         case _T('7'): case _T('8'): case _T('9'):
1076           {
1077             int len = 0;
1078             CHECK_PREC
1079             while ((pszFormat[n]>=_T('0')) && (pszFormat[n]<=_T('9'))) {
1080               s_szFlags[flagofs++] = pszFormat[n];
1081               len = len*10 + (pszFormat[n] - _T('0'));
1082               n++;
1083             }
1084             if (in_prec) max_width = len;
1085             else min_width = len;
1086             n--; // the main loop pre-increments n again
1087           }
1088           break;
1089         case _T('d'):
1090         case _T('i'):
1091         case _T('o'):
1092         case _T('u'):
1093         case _T('x'):
1094         case _T('X'):
1095           CHECK_PREC
1096           s_szFlags[flagofs++] = pszFormat[n];
1097           s_szFlags[flagofs] = '\0';
1098           if (ilen == 0 ) {
1099             int val = va_arg(argptr, int);
1100             ::sprintf(s_szScratch, s_szFlags, val);
1101           }
1102           else if (ilen == -1) {
1103             short int val = va_arg(argptr, short int);
1104             ::sprintf(s_szScratch, s_szFlags, val);
1105           }
1106           else if (ilen == 1) {
1107             long int val = va_arg(argptr, long int);
1108             ::sprintf(s_szScratch, s_szFlags, val);
1109           }
1110           else if (ilen == 2) {
1111 #if SIZEOF_LONG_LONG
1112             long long int val = va_arg(argptr, long long int);
1113             ::sprintf(s_szScratch, s_szFlags, val);
1114 #else
1115             long int val = va_arg(argptr, long int);
1116             ::sprintf(s_szScratch, s_szFlags, val);
1117 #endif
1118           }
1119           else if (ilen == 3) {
1120             size_t val = va_arg(argptr, size_t);
1121             ::sprintf(s_szScratch, s_szFlags, val);
1122           }
1123           *this += wxString(s_szScratch);
1124           done = TRUE;
1125           break;
1126         case _T('e'):
1127         case _T('E'):
1128         case _T('f'):
1129         case _T('g'):
1130         case _T('G'):
1131           CHECK_PREC
1132           s_szFlags[flagofs++] = pszFormat[n];
1133           s_szFlags[flagofs] = '\0';
1134           if (ilen == 2) {
1135             long double val = va_arg(argptr, long double);
1136             ::sprintf(s_szScratch, s_szFlags, val);
1137           } else {
1138             double val = va_arg(argptr, double);
1139             ::sprintf(s_szScratch, s_szFlags, val);
1140           }
1141           *this += wxString(s_szScratch);
1142           done = TRUE;
1143           break;
1144         case _T('p'):
1145           {
1146             void *val = va_arg(argptr, void *);
1147             CHECK_PREC
1148             s_szFlags[flagofs++] = pszFormat[n];
1149             s_szFlags[flagofs] = '\0';
1150             ::sprintf(s_szScratch, s_szFlags, val);
1151             *this += wxString(s_szScratch);
1152             done = TRUE;
1153           }
1154           break;
1155         case _T('c'):
1156           {
1157             wxChar val = va_arg(argptr, int);
1158             // we don't need to honor padding here, do we?
1159             *this += val;
1160             done = TRUE;
1161           }
1162           break;
1163         case _T('s'):
1164           if (ilen == -1) {
1165             // wx extension: we'll let %hs mean non-Unicode strings
1166             char *val = va_arg(argptr, char *);
1167 #if wxUSE_UNICODE
1168             // ASCII->Unicode constructor handles max_width right
1169             wxString s(val, wxConv_libc, max_width);
1170 #else
1171             size_t len = wxSTRING_MAXLEN;
1172             if (val) {
1173               for (len = 0; val[len] && (len<max_width); len++);
1174             } else val = _T("(null)");
1175             wxString s(val, len);
1176 #endif
1177             if (s.Len() < min_width)
1178               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1179             *this += s;
1180           } else {
1181             wxChar *val = va_arg(argptr, wxChar *);
1182             size_t len = wxSTRING_MAXLEN;
1183             if (val) {
1184               for (len = 0; val[len] && (len<max_width); len++);
1185             } else val = _T("(null)");
1186             wxString s(val, len);
1187             if (s.Len() < min_width)
1188               s.Pad(min_width - s.Len(), _T(' '), adj_left);
1189             *this += s;
1190             done = TRUE;
1191           }
1192           break;
1193         case _T('n'):
1194           if (ilen == 0) {
1195             int *val = va_arg(argptr, int *);
1196             *val = Len();
1197           }
1198           else if (ilen == -1) {
1199             short int *val = va_arg(argptr, short int *);
1200             *val = Len();
1201           }
1202           else if (ilen >= 1) {
1203             long int *val = va_arg(argptr, long int *);
1204             *val = Len();
1205           }
1206           done = TRUE;
1207           break;
1208         default:
1209           if (wxIsalpha(pszFormat[n]))
1210             // probably some flag not taken care of here yet
1211             s_szFlags[flagofs++] = pszFormat[n];
1212           else {
1213             // bad format
1214             *this += _T('%'); // just to pass the glibc tst-printf.c
1215             n--;
1216             done = TRUE;
1217           }
1218           break;
1219         }
1220 #undef CHECK_PREC
1221       } while (!done);
1222     } else *this += pszFormat[n];
1223
1224 #else
1225   // NB: wxVsnprintf() may return either less than the buffer size or -1 if there
1226   //     is not enough place depending on implementation
1227   int iLen = wxVsnprintf(s_szScratch, WXSIZEOF(s_szScratch), pszFormat, argptr);
1228   char *buffer;
1229   if ( iLen < (int)WXSIZEOF(s_szScratch) ) {
1230     buffer = s_szScratch;
1231   }
1232   else {
1233       int size = WXSIZEOF(s_szScratch) * 2;
1234       buffer = (char *)malloc(size);
1235       while ( buffer != NULL ) {
1236           iLen = wxVsnprintf(buffer, WXSIZEOF(s_szScratch), pszFormat, argptr);
1237           if ( iLen < size ) {
1238               // ok, there was enough space
1239               break;
1240           }
1241
1242           // still not enough, double it again
1243           buffer = (char *)realloc(buffer, size *= 2);
1244       }
1245
1246       if ( !buffer ) {
1247           // out of memory
1248           return -1;
1249       }
1250   }
1251
1252   wxString s(buffer);
1253   *this = s;
1254
1255   if ( buffer != s_szScratch )
1256       free(buffer);
1257 #endif
1258
1259   return Len();
1260 }
1261
1262 // ----------------------------------------------------------------------------
1263 // misc other operations
1264 // ----------------------------------------------------------------------------
1265 bool wxString::Matches(const wxChar *pszMask) const
1266 {
1267   // check char by char
1268   const wxChar *pszTxt;
1269   for ( pszTxt = c_str(); *pszMask != _T('\0'); pszMask++, pszTxt++ ) {
1270     switch ( *pszMask ) {
1271       case _T('?'):
1272         if ( *pszTxt == _T('\0') )
1273           return FALSE;
1274
1275         pszTxt++;
1276         pszMask++;
1277         break;
1278
1279       case _T('*'):
1280         {
1281           // ignore special chars immediately following this one
1282           while ( *pszMask == _T('*') || *pszMask == _T('?') )
1283             pszMask++;
1284
1285           // if there is nothing more, match
1286           if ( *pszMask == _T('\0') )
1287             return TRUE;
1288
1289           // are there any other metacharacters in the mask?
1290           size_t uiLenMask;
1291           const wxChar *pEndMask = wxStrpbrk(pszMask, _T("*?"));
1292
1293           if ( pEndMask != NULL ) {
1294             // we have to match the string between two metachars
1295             uiLenMask = pEndMask - pszMask;
1296           }
1297           else {
1298             // we have to match the remainder of the string
1299             uiLenMask = wxStrlen(pszMask);
1300           }
1301
1302           wxString strToMatch(pszMask, uiLenMask);
1303           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1304           if ( pMatch == NULL )
1305             return FALSE;
1306
1307           // -1 to compensate "++" in the loop
1308           pszTxt = pMatch + uiLenMask - 1;
1309           pszMask += uiLenMask - 1;
1310         }
1311         break;
1312
1313       default:
1314         if ( *pszMask != *pszTxt )
1315           return FALSE;
1316         break;
1317     }
1318   }
1319
1320   // match only if nothing left
1321   return *pszTxt == _T('\0');
1322 }
1323
1324 // Count the number of chars
1325 int wxString::Freq(wxChar ch) const
1326 {
1327     int count = 0;
1328     int len = Len();
1329     for (int i = 0; i < len; i++)
1330     {
1331         if (GetChar(i) == ch)
1332             count ++;
1333     }
1334     return count;
1335 }
1336
1337 // convert to upper case, return the copy of the string
1338 wxString wxString::Upper() const
1339 { wxString s(*this); return s.MakeUpper(); }
1340
1341 // convert to lower case, return the copy of the string
1342 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1343
1344 int wxString::sprintf(const wxChar *pszFormat, ...)
1345   {
1346     va_list argptr;
1347     va_start(argptr, pszFormat);
1348     int iLen = PrintfV(pszFormat, argptr);
1349     va_end(argptr);
1350     return iLen;
1351   }
1352
1353 // ---------------------------------------------------------------------------
1354 // standard C++ library string functions
1355 // ---------------------------------------------------------------------------
1356 #ifdef  wxSTD_STRING_COMPATIBILITY
1357
1358 wxString& wxString::insert(size_t nPos, const wxString& str)
1359 {
1360   wxASSERT( str.GetStringData()->IsValid() );
1361   wxASSERT( nPos <= Len() );
1362
1363   if ( !str.IsEmpty() ) {
1364     wxString strTmp;
1365     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1366     wxStrncpy(pc, c_str(), nPos);
1367     wxStrcpy(pc + nPos, str);
1368     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1369     strTmp.UngetWriteBuf();
1370     *this = strTmp;
1371   }
1372
1373   return *this;
1374 }
1375
1376 size_t wxString::find(const wxString& str, size_t nStart) const
1377 {
1378   wxASSERT( str.GetStringData()->IsValid() );
1379   wxASSERT( nStart <= Len() );
1380
1381   const wxChar *p = wxStrstr(c_str() + nStart, str);
1382
1383   return p == NULL ? npos : p - c_str();
1384 }
1385
1386 // VC++ 1.5 can't cope with the default argument in the header.
1387 #if !defined(__VISUALC__) || defined(__WIN32__)
1388 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1389 {
1390   return find(wxString(sz, n == npos ? 0 : n), nStart);
1391 }
1392 #endif // VC++ 1.5
1393
1394 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1395 #if !defined(__BORLANDC__)
1396 size_t wxString::find(wxChar ch, size_t nStart) const
1397 {
1398   wxASSERT( nStart <= Len() );
1399
1400   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1401
1402   return p == NULL ? npos : p - c_str();
1403 }
1404 #endif
1405
1406 size_t wxString::rfind(const wxString& str, size_t nStart) const
1407 {
1408   wxASSERT( str.GetStringData()->IsValid() );
1409   wxASSERT( nStart <= Len() );
1410
1411   // # could be quicker than that
1412   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1413   while ( p >= c_str() + str.Len() ) {
1414     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1415       return p - str.Len() - c_str();
1416     p--;
1417   }
1418
1419   return npos;
1420 }
1421
1422 // VC++ 1.5 can't cope with the default argument in the header.
1423 #if !defined(__VISUALC__) || defined(__WIN32__)
1424 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1425 {
1426   return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1427 }
1428
1429 size_t wxString::rfind(wxChar ch, size_t nStart) const
1430 {
1431   wxASSERT( nStart <= Len() );
1432
1433   const wxChar *p = wxStrrchr(c_str() + nStart, ch);
1434
1435   return p == NULL ? npos : p - c_str();
1436 }
1437 #endif // VC++ 1.5
1438
1439 wxString wxString::substr(size_t nStart, size_t nLen) const
1440 {
1441   // npos means 'take all'
1442   if ( nLen == npos )
1443     nLen = 0;
1444
1445   wxASSERT( nStart + nLen <= Len() );
1446
1447   return wxString(c_str() + nStart, nLen == npos ? 0 : nLen);
1448 }
1449
1450 wxString& wxString::erase(size_t nStart, size_t nLen)
1451 {
1452   wxString strTmp(c_str(), nStart);
1453   if ( nLen != npos ) {
1454     wxASSERT( nStart + nLen <= Len() );
1455
1456     strTmp.append(c_str() + nStart + nLen);
1457   }
1458
1459   *this = strTmp;
1460   return *this;
1461 }
1462
1463 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1464 {
1465   wxASSERT( nStart + nLen <= wxStrlen(sz) );
1466
1467   wxString strTmp;
1468   if ( nStart != 0 )
1469     strTmp.append(c_str(), nStart);
1470   strTmp += sz;
1471   strTmp.append(c_str() + nStart + nLen);
1472
1473   *this = strTmp;
1474   return *this;
1475 }
1476
1477 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1478 {
1479   return replace(nStart, nLen, wxString(ch, nCount));
1480 }
1481
1482 wxString& wxString::replace(size_t nStart, size_t nLen,
1483                             const wxString& str, size_t nStart2, size_t nLen2)
1484 {
1485   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1486 }
1487
1488 wxString& wxString::replace(size_t nStart, size_t nLen,
1489                         const wxChar* sz, size_t nCount)
1490 {
1491   return replace(nStart, nLen, wxString(sz, nCount));
1492 }
1493
1494 #endif  //std::string compatibility
1495
1496 // ============================================================================
1497 // ArrayString
1498 // ============================================================================
1499
1500 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1501 #define   ARRAY_MAXSIZE_INCREMENT       4096
1502 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1503   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1504 #endif
1505
1506 #define   STRING(p)   ((wxString *)(&(p)))
1507
1508 // ctor
1509 wxArrayString::wxArrayString()
1510 {
1511   m_nSize  =
1512   m_nCount = 0;
1513   m_pItems = (wxChar **) NULL;
1514 }
1515
1516 // copy ctor
1517 wxArrayString::wxArrayString(const wxArrayString& src)
1518 {
1519   m_nSize  =
1520   m_nCount = 0;
1521   m_pItems = (wxChar **) NULL;
1522
1523   *this = src;
1524 }
1525
1526 // assignment operator
1527 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1528 {
1529   if ( m_nSize > 0 )
1530     Clear();
1531
1532   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1533     Alloc(src.m_nCount);
1534
1535   // we can't just copy the pointers here because otherwise we would share
1536   // the strings with another array
1537   for ( size_t n = 0; n < src.m_nCount; n++ )
1538     Add(src[n]);
1539
1540   if ( m_nCount != 0 )
1541     memcpy(m_pItems, src.m_pItems, m_nCount*sizeof(wxChar *));
1542
1543   return *this;
1544 }
1545
1546 // grow the array
1547 void wxArrayString::Grow()
1548 {
1549   // only do it if no more place
1550   if( m_nCount == m_nSize ) {
1551     if( m_nSize == 0 ) {
1552       // was empty, alloc some memory
1553       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1554       m_pItems = new wxChar *[m_nSize];
1555     }
1556     else {
1557       // otherwise when it's called for the first time, nIncrement would be 0
1558       // and the array would never be expanded
1559       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1560
1561       // add 50% but not too much
1562       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1563                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1564       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1565         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1566       m_nSize += nIncrement;
1567       wxChar **pNew = new wxChar *[m_nSize];
1568
1569       // copy data to new location
1570       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1571
1572       // delete old memory (but do not release the strings!)
1573       wxDELETEA(m_pItems);
1574
1575       m_pItems = pNew;
1576     }
1577   }
1578 }
1579
1580 void wxArrayString::Free()
1581 {
1582   for ( size_t n = 0; n < m_nCount; n++ ) {
1583     STRING(m_pItems[n])->GetStringData()->Unlock();
1584   }
1585 }
1586
1587 // deletes all the strings from the list
1588 void wxArrayString::Empty()
1589 {
1590   Free();
1591
1592   m_nCount = 0;
1593 }
1594
1595 // as Empty, but also frees memory
1596 void wxArrayString::Clear()
1597 {
1598   Free();
1599
1600   m_nSize  =
1601   m_nCount = 0;
1602
1603   wxDELETEA(m_pItems);
1604 }
1605
1606 // dtor
1607 wxArrayString::~wxArrayString()
1608 {
1609   Free();
1610
1611   wxDELETEA(m_pItems);
1612 }
1613
1614 // pre-allocates memory (frees the previous data!)
1615 void wxArrayString::Alloc(size_t nSize)
1616 {
1617   wxASSERT( nSize > 0 );
1618
1619   // only if old buffer was not big enough
1620   if ( nSize > m_nSize ) {
1621     Free();
1622     wxDELETEA(m_pItems);
1623     m_pItems = new wxChar *[nSize];
1624     m_nSize  = nSize;
1625   }
1626
1627   m_nCount = 0;
1628 }
1629
1630 // minimizes the memory usage by freeing unused memory
1631 void wxArrayString::Shrink()
1632 {
1633   // only do it if we have some memory to free
1634   if( m_nCount < m_nSize ) {
1635     // allocates exactly as much memory as we need
1636     wxChar **pNew = new wxChar *[m_nCount];
1637
1638     // copy data to new location
1639     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1640     delete [] m_pItems;
1641     m_pItems = pNew;
1642   }
1643 }
1644
1645 // searches the array for an item (forward or backwards)
1646 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1647 {
1648   if ( bFromEnd ) {
1649     if ( m_nCount > 0 ) {
1650       size_t ui = m_nCount;
1651       do {
1652         if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1653           return ui;
1654       }
1655       while ( ui != 0 );
1656     }
1657   }
1658   else {
1659     for( size_t ui = 0; ui < m_nCount; ui++ ) {
1660       if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1661         return ui;
1662     }
1663   }
1664
1665   return wxNOT_FOUND;
1666 }
1667
1668 // add item at the end
1669 void wxArrayString::Add(const wxString& str)
1670 {
1671   wxASSERT( str.GetStringData()->IsValid() );
1672
1673   Grow();
1674
1675   // the string data must not be deleted!
1676   str.GetStringData()->Lock();
1677   m_pItems[m_nCount++] = (wxChar *)str.c_str();
1678 }
1679
1680 // add item at the given position
1681 void wxArrayString::Insert(const wxString& str, size_t nIndex)
1682 {
1683   wxASSERT( str.GetStringData()->IsValid() );
1684
1685   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Insert") );
1686
1687   Grow();
1688
1689   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
1690           (m_nCount - nIndex)*sizeof(wxChar *));
1691
1692   str.GetStringData()->Lock();
1693   m_pItems[nIndex] = (wxChar *)str.c_str();
1694
1695   m_nCount++;
1696 }
1697
1698 // removes item from array (by index)
1699 void wxArrayString::Remove(size_t nIndex)
1700 {
1701   wxCHECK_RET( nIndex <= m_nCount, _("bad index in wxArrayString::Remove") );
1702
1703   // release our lock
1704   Item(nIndex).GetStringData()->Unlock();
1705
1706   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
1707           (m_nCount - nIndex - 1)*sizeof(wxChar *));
1708   m_nCount--;
1709 }
1710
1711 // removes item from array (by value)
1712 void wxArrayString::Remove(const wxChar *sz)
1713 {
1714   int iIndex = Index(sz);
1715
1716   wxCHECK_RET( iIndex != wxNOT_FOUND,
1717                _("removing inexistent element in wxArrayString::Remove") );
1718
1719   Remove(iIndex);
1720 }
1721
1722 // ----------------------------------------------------------------------------
1723 // sorting
1724 // ----------------------------------------------------------------------------
1725
1726 // we can only sort one array at a time with the quick-sort based
1727 // implementation
1728 #if wxUSE_THREADS
1729   // need a critical section to protect access to gs_compareFunction and
1730   // gs_sortAscending variables
1731   static wxCriticalSection *gs_critsectStringSort = NULL;
1732
1733   // call this before the value of the global sort vars is changed/after
1734   // you're finished with them
1735   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
1736                            gs_critsectStringSort = new wxCriticalSection;     \
1737                            gs_critsectStringSort->Enter()
1738   #define END_SORT()       gs_critsectStringSort->Leave();                    \
1739                            delete gs_critsectStringSort;                      \
1740                            gs_critsectStringSort = NULL
1741 #else // !threads
1742   #define START_SORT()
1743   #define END_SORT()
1744 #endif // wxUSE_THREADS
1745
1746 // function to use for string comparaison
1747 static wxArrayString::CompareFunction gs_compareFunction = NULL;
1748
1749 // if we don't use the compare function, this flag tells us if we sort the
1750 // array in ascending or descending order
1751 static bool gs_sortAscending = TRUE;
1752
1753 // function which is called by quick sort
1754 static int wxStringCompareFunction(const void *first, const void *second)
1755 {
1756   wxString *strFirst = (wxString *)first;
1757   wxString *strSecond = (wxString *)second;
1758
1759   if ( gs_compareFunction ) {
1760     return gs_compareFunction(*strFirst, *strSecond);
1761   }
1762   else {
1763     // maybe we should use wxStrcoll
1764     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
1765
1766     return gs_sortAscending ? result : -result;
1767   }
1768 }
1769
1770 // sort array elements using passed comparaison function
1771 void wxArrayString::Sort(CompareFunction compareFunction)
1772 {
1773   START_SORT();
1774
1775   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1776   gs_compareFunction = compareFunction;
1777
1778   DoSort();
1779
1780   END_SORT();
1781 }
1782
1783 void wxArrayString::Sort(bool reverseOrder)
1784 {
1785   START_SORT();
1786
1787   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
1788   gs_sortAscending = !reverseOrder;
1789
1790   DoSort();
1791
1792   END_SORT();
1793 }
1794
1795 void wxArrayString::DoSort()
1796 {
1797   // just sort the pointers using qsort() - of course it only works because
1798   // wxString() *is* a pointer to its data
1799   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
1800 }
1801
1802 // ============================================================================
1803 // MBConv
1804 // ============================================================================
1805
1806 WXDLLEXPORT_DATA(wxMBConv *) wxConv_current = &wxConv_libc;
1807
1808 // ----------------------------------------------------------------------------
1809 // standard libc conversion
1810 // ----------------------------------------------------------------------------
1811
1812 WXDLLEXPORT_DATA(wxMBConv) wxConv_libc;
1813
1814 size_t wxMBConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1815 {
1816   return wxMB2WC(buf, psz, n);
1817 }
1818
1819 size_t wxMBConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1820 {
1821   return wxWC2MB(buf, psz, n);
1822 }
1823
1824 // ----------------------------------------------------------------------------
1825 // standard file conversion
1826 // ----------------------------------------------------------------------------
1827
1828 WXDLLEXPORT_DATA(wxMBConv_file) wxConv_file;
1829
1830 // just use the libc conversion for now
1831 size_t wxMBConv_file::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1832 {
1833   return wxMB2WC(buf, psz, n);
1834 }
1835
1836 size_t wxMBConv_file::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1837 {
1838   return wxWC2MB(buf, psz, n);
1839 }
1840
1841 // ----------------------------------------------------------------------------
1842 // standard gdk conversion
1843 // ----------------------------------------------------------------------------
1844
1845 #if defined(__WXGTK__) && (GTK_MINOR_VERSION > 0)
1846 WXDLLEXPORT_DATA(wxMBConv_gdk) wxConv_gdk;
1847
1848 #include <gdk/gdk.h>
1849
1850 size_t wxMBConv_gdk::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1851 {
1852   if (buf) {
1853     return gdk_mbstowcs((GdkWChar *)buf, psz, n);
1854   } else {
1855     GdkWChar *nbuf = new GdkWChar[n=strlen(psz)];
1856     size_t len = gdk_mbstowcs(nbuf, psz, n);
1857     delete [] nbuf;
1858     return len;
1859   }
1860 }
1861
1862 size_t wxMBConv_gdk::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1863 {
1864   char *mbstr = gdk_wcstombs((GdkWChar *)psz);
1865   size_t len = mbstr ? strlen(mbstr) : 0;
1866   if (buf) {
1867     if (len > n) len = n;
1868     memcpy(buf, psz, len);
1869     if (len < n) buf[len] = 0;
1870   }
1871   return len;
1872 }
1873 #endif // GTK > 1.0
1874
1875 // ----------------------------------------------------------------------------
1876 // UTF-7
1877 // ----------------------------------------------------------------------------
1878
1879 WXDLLEXPORT_DATA(wxMBConv_UTF7) wxConv_UTF7;
1880
1881 // TODO: write actual implementations of UTF-7 here
1882 size_t wxMBConv_UTF7::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1883 {
1884   return 0;
1885 }
1886
1887 size_t wxMBConv_UTF7::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1888 {
1889   return 0;
1890 }
1891
1892 // ----------------------------------------------------------------------------
1893 // UTF-8
1894 // ----------------------------------------------------------------------------
1895
1896 WXDLLEXPORT_DATA(wxMBConv_UTF8) wxConv_UTF8;
1897
1898 // TODO: write actual implementations of UTF-8 here
1899 size_t wxMBConv_UTF8::MB2WC(wchar_t *buf, const char *psz, size_t n) const
1900 {
1901   return wxMB2WC(buf, psz, n);
1902 }
1903
1904 size_t wxMBConv_UTF8::WC2MB(char *buf, const wchar_t *psz, size_t n) const
1905 {
1906   return wxWC2MB(buf, psz, n);
1907 }
1908
1909 // ----------------------------------------------------------------------------
1910 // specified character set
1911 // ----------------------------------------------------------------------------
1912
1913 class wxCharacterSet
1914 {
1915 public:
1916   wxArrayString names;
1917   wchar_t *data;
1918 };
1919
1920 #ifndef WX_PRECOMP
1921   #include "wx/dynarray.h"
1922   #include "wx/filefn.h"
1923   #include "wx/textfile.h"
1924   #include "wx/tokenzr.h"
1925   #include "wx/utils.h"
1926 #endif
1927
1928 WX_DECLARE_OBJARRAY(wxCharacterSet, wxCSArray);
1929 #include "wx/arrimpl.cpp"
1930 WX_DEFINE_OBJARRAY(wxCSArray);
1931
1932 static wxCSArray wxCharsets;
1933
1934 static void wxLoadCharacterSets(void)
1935 {
1936   static bool already_loaded = FALSE;
1937
1938   if (already_loaded) return;
1939
1940 #if defined(__UNIX__)
1941   // search through files in /usr/share/i18n/charmaps
1942   wxString fname;
1943   for (fname = ::wxFindFirstFile(_T("/usr/share/i18n/charmaps/*"));
1944        !fname.IsEmpty();
1945        fname = ::wxFindNextFile()) {
1946     wxTextFile cmap(fname);
1947     if (cmap.Open()) {
1948       wxCharacterSet *cset = new wxCharacterSet;
1949       wxString comchar,escchar;
1950       bool in_charset = FALSE;
1951
1952       // wxFprintf(stderr,_T("Loaded: %s\n"),fname.c_str());
1953
1954       wxString line;
1955       for (line = cmap.GetFirstLine();
1956            !cmap.Eof();
1957            line = cmap.GetNextLine()) {
1958         // wxFprintf(stderr,_T("line contents: %s\n"),line.c_str());
1959         wxStringTokenizer token(line);
1960         wxString cmd = token.GetNextToken();
1961         if (cmd == comchar) {
1962           if (token.GetNextToken() == _T("alias")) {
1963             wxStringTokenizer names(token.GetNextToken(),_T("/"));
1964             wxString name;
1965             while (!(name = names.GetNextToken()).IsEmpty())
1966               cset->names.Add(name);
1967           }
1968         }
1969         else if (cmd == _T("<code_set_name>"))
1970           cset->names.Add(token.GetNextToken());
1971         else if (cmd == _T("<comment_char>"))
1972           comchar = token.GetNextToken();
1973         else if (cmd == _T("<escape_char>"))
1974           escchar = token.GetNextToken();
1975         else if (cmd == _T("<mb_cur_min>")) {
1976           delete cset;
1977           cset = (wxCharacterSet *) NULL;
1978           break; // we don't support multibyte charsets ourselves (yet)
1979         }
1980         else if (cmd == _T("CHARMAP")) {
1981           cset->data = (wchar_t *)calloc(256, sizeof(wchar_t));
1982           in_charset = TRUE;
1983         }
1984         else if (cmd == _T("END")) {
1985           if (token.GetNextToken() == _T("CHARMAP"))
1986             in_charset = FALSE;
1987         }
1988         else if (in_charset) {
1989           // format: <NUL> /x00 <U0000> NULL (NUL)
1990           //         <A>   /x41 <U0041> LATIN CAPITAL LETTER A
1991           wxString hex = token.GetNextToken();
1992           // skip whitespace (why doesn't wxStringTokenizer do this?)
1993           while (wxIsEmpty(hex) && token.HasMoreTokens()) hex = token.GetNextToken();
1994           wxString uni = token.GetNextToken();
1995           // skip whitespace again
1996           while (wxIsEmpty(uni) && token.HasMoreTokens()) uni = token.GetNextToken();
1997
1998           if ((hex.GetChar(0) == escchar) && (hex.GetChar(1) == _T('x')) &&
1999               (uni.Left(2) == _T("<U"))) {
2000             hex.MakeUpper(); uni.MakeUpper();
2001             int pos = ::wxHexToDec(hex.Mid(2,2));
2002             if (pos>=0) {
2003               unsigned long uni1 = ::wxHexToDec(uni.Mid(2,2));
2004               unsigned long uni2 = ::wxHexToDec(uni.Mid(4,2));
2005               cset->data[pos] = (uni1 << 16) | uni2;
2006               // wxFprintf(stderr,_T("char %02x mapped to %04x (%c)\n"),pos,cset->data[pos],cset->data[pos]);
2007             }
2008           }
2009         }
2010       }
2011       if (cset) {
2012         cset->names.Shrink();
2013         wxCharsets.Add(cset);
2014       }
2015     }
2016   }
2017 #endif
2018   wxCharsets.Shrink();
2019   already_loaded = TRUE;
2020 }
2021
2022 static wxCharacterSet *wxFindCharacterSet(const wxChar *charset)
2023 {
2024   wxLoadCharacterSets();
2025   for (size_t n=0; n<wxCharsets.GetCount(); n++)
2026     if (wxCharsets[n].names.Index(charset) != wxNOT_FOUND)
2027       return &(wxCharsets[n]);
2028   return (wxCharacterSet *)NULL;
2029 }
2030
2031 WXDLLEXPORT_DATA(wxCSConv) wxConv_local((const wxChar *)NULL);
2032
2033 wxCSConv::wxCSConv(const wxChar *charset)
2034 {
2035   if (!charset) {
2036 #ifdef __UNIX__
2037     wxChar *lang = wxGetenv(_T("LANG"));
2038     wxChar *dot = lang ? wxStrchr(lang, _T('.')) : (wxChar *)NULL;
2039     if (dot) charset = dot+1;
2040 #endif
2041   }
2042   m_cset = (wxCharacterSet *) NULL;
2043   m_deferred = FALSE;
2044   if (charset) {
2045 #ifdef __UNIX__
2046     // first, convert the character set name to standard form
2047     wxString codeset;
2048     if (wxString(charset,3).CmpNoCase(_T("ISO")) == 0) {
2049       // make sure it's represented in the standard form: ISO_8859-1
2050       codeset = _T("ISO_");
2051       charset += 3;
2052       if ((*charset == _T('-')) || (*charset == _T('_'))) charset++;
2053       if (wxStrlen(charset)>4) {
2054         if (wxString(charset,4) == _T("8859")) {
2055           codeset << _T("8859-");
2056           if (*charset == _T('-')) charset++;
2057         }
2058       }
2059     }
2060     codeset << charset;
2061     codeset.MakeUpper();
2062     m_name = wxStrdup(codeset.c_str());
2063     m_deferred = TRUE;
2064 #endif
2065   }
2066 }
2067
2068 wxCSConv::~wxCSConv()
2069 {
2070   free(m_name);
2071 }
2072
2073 void wxCSConv::LoadNow()
2074 {
2075 //  wxPrintf(_T("Conversion request\n"));
2076   if (m_deferred) {
2077     m_cset = wxFindCharacterSet(m_name);
2078     m_deferred = FALSE;
2079   }
2080 }
2081
2082 size_t wxCSConv::MB2WC(wchar_t *buf, const char *psz, size_t n) const
2083 {
2084   ((wxCSConv *)this)->LoadNow(); // discard constness
2085   if (buf) {
2086     if (m_cset) {
2087       for (size_t c=0; c<=n; c++)
2088         buf[c] = m_cset->data[psz[c]];
2089     } else {
2090       // latin-1 (direct)
2091       for (size_t c=0; c<=n; c++)
2092         buf[c] = psz[c];
2093     }
2094   }
2095   return n;
2096 }
2097
2098 size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
2099 {
2100   ((wxCSConv *)this)->LoadNow(); // discard constness
2101   if (buf) {
2102     if (m_cset) {
2103       for (size_t c=0; c<=n; c++) {
2104         size_t n;
2105         for (n=0; (n<256) && (m_cset->data[n] != psz[c]); n++);
2106         buf[c] = (n>0xff) ? '?' : n;
2107       }
2108     } else {
2109       // latin-1 (direct)
2110       for (size_t c=0; c<=n; c++)
2111         buf[c] = (psz[c]>0xff) ? '?' : psz[c];
2112     }
2113   }
2114   return n;
2115 }
2116