src/common/string.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        string.cpp
   3 // Purpose:     wxString class
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifdef __GNUG__
  13   #pragma implementation "string.h"
  14 #endif
  15
  16 /*
  17  * About ref counting:
  18  *  1) all empty strings use g_strEmpty, nRefs = -1 (set in Init())
  19  *  2) AllocBuffer() sets nRefs to 1, Lock() increments it by one
  20  *  3) Unlock() decrements nRefs and frees memory if it goes to 0
  21  */
  22
  23 // ===========================================================================
  24 // headers, declarations, constants
  25 // ===========================================================================
  26
  27 // For compilers that support precompilation, includes "wx.h".
  28 #include "wx/wxprec.h"
  29
  30 #ifdef __BORLANDC__
  31   #pragma hdrstop
  32 #endif
  33
  34 #ifndef WX_PRECOMP
  35   #include "wx/defs.h"
  36   #include "wx/string.h"
  37   #include "wx/intl.h"
  38   #include "wx/thread.h"
  39 #endif
  40
  41 #include <ctype.h>
  42 #include <string.h>
  43 #include <stdlib.h>
  44
  45 #ifdef __SALFORDC__
  46   #include <clib.h>
  47 #endif
  48
  49 #if wxUSE_WCSRTOMBS
  50   #include <wchar.h>    // for wcsrtombs(), see comments where it's used
  51 #endif // GNU
  52
  53 #ifdef  WXSTRING_IS_WXOBJECT
  54   IMPLEMENT_DYNAMIC_CLASS(wxString, wxObject)
  55 #endif  //WXSTRING_IS_WXOBJECT
  56
  57 #if wxUSE_UNICODE
  58 #undef wxUSE_EXPERIMENTAL_PRINTF
  59 #define wxUSE_EXPERIMENTAL_PRINTF 1
  60 #endif
  61
  62 // allocating extra space for each string consumes more memory but speeds up
  63 // the concatenation operations (nLen is the current string's length)
  64 // NB: EXTRA_ALLOC must be >= 0!
  65 #define EXTRA_ALLOC       (19 - nLen % 16)
  66
  67 #if defined(__VISAGECPP__) && __IBMCPP__ >= 400
  68 // must define this in .cpp for VA or else you get multiply defined symbols everywhere
  69
  70 // maximum possible length for a string means "take all string" everywhere
  71 //  (as sizeof(StringData) is unknown here, we substract 100)
  72 const unsigned int wxSTRING_MAXLEN = UINT_MAX - 100;
  73
  74 #endif
  75
  76 // ---------------------------------------------------------------------------
  77 // static class variables definition
  78 // ---------------------------------------------------------------------------
  79
  80 #ifdef  wxSTD_STRING_COMPATIBILITY
  81   const size_t wxString::npos = wxSTRING_MAXLEN;
  82 #endif // wxSTD_STRING_COMPATIBILITY
  83
  84 // ----------------------------------------------------------------------------
  85 // static data
  86 // ----------------------------------------------------------------------------
  87
  88 // for an empty string, GetStringData() will return this address: this
  89 // structure has the same layout as wxStringData and it's data() method will
  90 // return the empty string (dummy pointer)
  91 static const struct
  92 {
  93   wxStringData data;
  94   wxChar dummy;
  95 } g_strEmpty = { {-1, 0, 0}, wxT('\0') };
  96
  97 // empty C style string: points to 'string data' byte of g_strEmpty
  98 extern const wxChar WXDLLEXPORT *wxEmptyString = &g_strEmpty.dummy;
  99
 100 // ----------------------------------------------------------------------------
 101 // conditional compilation
 102 // ----------------------------------------------------------------------------
 103
 104 #if !defined(__WXSW__) && wxUSE_UNICODE
 105   #ifdef wxUSE_EXPERIMENTAL_PRINTF
 106     #undef wxUSE_EXPERIMENTAL_PRINTF
 107   #endif
 108   #define wxUSE_EXPERIMENTAL_PRINTF 1
 109 #endif
 110
 111 // we want to find out if the current platform supports vsnprintf()-like
 112 // function: for Unix this is done with configure, for Windows we test the
 113 // compiler explicitly.
 114 //
 115 // FIXME currently, this is only for ANSI (!Unicode) strings, so we call this
 116 //       function wxVsnprintfA (A for ANSI), should also find one for Unicode
 117 //       strings in Unicode build
 118 #ifdef __WXMSW__
 119     #if (defined(__VISUALC__) || defined(wxUSE_NORLANDER_HEADERS)) && !defined(__MINGW32__)
 120         #define wxVsnprintfA     _vsnprintf
 121     #endif
 122 #else   // !Windows
 123     #ifdef HAVE_VSNPRINTF
 124         #define wxVsnprintfA       vsnprintf
 125     #endif
 126 #endif  // Windows/!Windows
 127
 128 #ifndef wxVsnprintfA
 129     // in this case we'll use vsprintf() (which is ANSI and thus should be
 130     // always available), but it's unsafe because it doesn't check for buffer
 131     // size - so give a warning
 132     #define wxVsnprintfA(buf, len, format, arg) vsprintf(buf, format, arg)
 133
 134     #if defined(__VISUALC__)
 135         #pragma message("Using sprintf() because no snprintf()-like function defined")
 136     #elif defined(__GNUG__) && !defined(__UNIX__)
 137         #warning "Using sprintf() because no snprintf()-like function defined"
 138     #elif defined(__MWERKS__)
 139         #warning "Using sprintf() because no snprintf()-like function defined"
 140     #endif //compiler
 141 #endif // no vsnprintf
 142
 143 #ifdef _AIX
 144   // AIX has vsnprintf, but there's no prototype in the system headers.
 145   extern "C" int vsnprintf(char* str, size_t n, const char* format, va_list ap);
 146 #endif
 147
 148 // ----------------------------------------------------------------------------
 149 // global functions
 150 // ----------------------------------------------------------------------------
 151
 152 #if defined(wxSTD_STRING_COMPATIBILITY) && wxUSE_STD_IOSTREAM
 153
 154 // MS Visual C++ version 5.0 provides the new STL headers as well as the old
 155 // iostream ones.
 156 //
 157 // ATTN: you can _not_ use both of these in the same program!
 158
 159 istream& operator>>(istream& is, wxString& WXUNUSED(str))
 160 {
 161 #if 0
 162   int w = is.width(0);
 163   if ( is.ipfx(0) ) {
 164     streambuf *sb = is.rdbuf();
 165     str.erase();
 166     while ( true ) {
 167       int ch = sb->sbumpc ();
 168       if ( ch == EOF ) {
 169         is.setstate(ios::eofbit);
 170         break;
 171       }
 172       else if ( isspace(ch) ) {
 173         sb->sungetc();
 174         break;
 175       }
 176
 177       str += ch;
 178       if ( --w == 1 )
 179         break;
 180     }
 181   }
 182
 183   is.isfx();
 184   if ( str.length() == 0 )
 185     is.setstate(ios::failbit);
 186 #endif
 187   return is;
 188 }
 189
 190 ostream& operator<<(ostream& os, const wxString& str)
 191 {
 192   os << str.c_str();
 193   return os;
 194 }
 195
 196 #endif  //std::string compatibility
 197
 198 extern int WXDLLEXPORT wxVsnprintf(wxChar *buf, size_t len,
 199                                    const wxChar *format, va_list argptr)
 200 {
 201 #if wxUSE_UNICODE
 202     // FIXME should use wvsnprintf() or whatever if it's available
 203     wxString s;
 204     int iLen = s.PrintfV(format, argptr);
 205     if ( iLen != -1 )
 206     {
 207         wxStrncpy(buf, s.c_str(), iLen);
 208     }
 209
 210     return iLen;
 211 #else // ANSI
 212     // vsnprintf() will not terminate the string with '\0' if there is not
 213     // enough place, but we want the string to always be NUL terminated
 214     int rc = wxVsnprintfA(buf, len - 1, format, argptr);
 215     if ( rc == -1 )
 216     {
 217         buf[len] = 0;
 218     }
 219
 220     return rc;
 221 #endif // Unicode/ANSI
 222 }
 223
 224 extern int WXDLLEXPORT wxSnprintf(wxChar *buf, size_t len,
 225                                   const wxChar *format, ...)
 226 {
 227     va_list argptr;
 228     va_start(argptr, format);
 229
 230     int iLen = wxVsnprintf(buf, len, format, argptr);
 231
 232     va_end(argptr);
 233
 234     return iLen;
 235 }
 236
 237 // ----------------------------------------------------------------------------
 238 // private classes
 239 // ----------------------------------------------------------------------------
 240
 241 // this small class is used to gather statistics for performance tuning
 242 //#define WXSTRING_STATISTICS
 243 #ifdef  WXSTRING_STATISTICS
 244   class Averager
 245   {
 246   public:
 247     Averager(const char *sz) { m_sz = sz; m_nTotal = m_nCount = 0; }
 248    ~Averager()
 249    { printf("wxString: average %s = %f\n", m_sz, ((float)m_nTotal)/m_nCount); }
 250
 251     void Add(size_t n) { m_nTotal += n; m_nCount++; }
 252
 253   private:
 254     size_t m_nCount, m_nTotal;
 255     const char *m_sz;
 256   } g_averageLength("allocation size"),
 257     g_averageSummandLength("summand length"),
 258     g_averageConcatHit("hit probability in concat"),
 259     g_averageInitialLength("initial string length");
 260
 261   #define STATISTICS_ADD(av, val) g_average##av.Add(val)
 262 #else
 263   #define STATISTICS_ADD(av, val)
 264 #endif // WXSTRING_STATISTICS
 265
 266 // ===========================================================================
 267 // wxString class core
 268 // ===========================================================================
 269
 270 // ---------------------------------------------------------------------------
 271 // construction
 272 // ---------------------------------------------------------------------------
 273
 274 // constructs string of <nLength> copies of character <ch>
 275 wxString::wxString(wxChar ch, size_t nLength)
 276 {
 277   Init();
 278
 279   if ( nLength > 0 ) {
 280     AllocBuffer(nLength);
 281
 282 #if wxUSE_UNICODE
 283     // memset only works on char
 284     for (size_t n=0; n<nLength; n++) m_pchData[n] = ch;
 285 #else
 286     memset(m_pchData, ch, nLength);
 287 #endif
 288   }
 289 }
 290
 291 // takes nLength elements of psz starting at nPos
 292 void wxString::InitWith(const wxChar *psz, size_t nPos, size_t nLength)
 293 {
 294   Init();
 295
 296   wxASSERT( nPos <= wxStrlen(psz) );
 297
 298   if ( nLength == wxSTRING_MAXLEN )
 299     nLength = wxStrlen(psz + nPos);
 300
 301   STATISTICS_ADD(InitialLength, nLength);
 302
 303   if ( nLength > 0 ) {
 304     // trailing '\0' is written in AllocBuffer()
 305     AllocBuffer(nLength);
 306     memcpy(m_pchData, psz + nPos, nLength*sizeof(wxChar));
 307   }
 308 }
 309
 310 #ifdef  wxSTD_STRING_COMPATIBILITY
 311
 312 // poor man's iterators are "void *" pointers
 313 wxString::wxString(const void *pStart, const void *pEnd)
 314 {
 315   InitWith((const wxChar *)pStart, 0,
 316            (const wxChar *)pEnd - (const wxChar *)pStart);
 317 }
 318
 319 #endif  //std::string compatibility
 320
 321 #if wxUSE_UNICODE
 322
 323 // from multibyte string
 324 wxString::wxString(const char *psz, wxMBConv& conv, size_t nLength)
 325 {
 326   // first get necessary size
 327   size_t nLen = psz ? conv.MB2WC((wchar_t *) NULL, psz, 0) : 0;
 328
 329   // nLength is number of *Unicode* characters here!
 330   if ((nLen != (size_t)-1) && (nLen > nLength))
 331     nLen = nLength;
 332
 333   // empty?
 334   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 335     AllocBuffer(nLen);
 336     conv.MB2WC(m_pchData, psz, nLen);
 337   }
 338   else {
 339     Init();
 340   }
 341 }
 342
 343 #else // ANSI
 344
 345 #if wxUSE_WCHAR_T
 346 // from wide string
 347 wxString::wxString(const wchar_t *pwz)
 348 {
 349   // first get necessary size
 350   size_t nLen = pwz ? wxWC2MB((char *) NULL, pwz, 0) : 0;
 351
 352   // empty?
 353   if ( (nLen != 0) && (nLen != (size_t)-1) ) {
 354     AllocBuffer(nLen);
 355     wxWC2MB(m_pchData, pwz, nLen);
 356   }
 357   else {
 358     Init();
 359   }
 360 }
 361 #endif // wxUSE_WCHAR_T
 362
 363 #endif // Unicode/ANSI
 364
 365 // ---------------------------------------------------------------------------
 366 // memory allocation
 367 // ---------------------------------------------------------------------------
 368
 369 // allocates memory needed to store a C string of length nLen
 370 void wxString::AllocBuffer(size_t nLen)
 371 {
 372   // allocating 0 sized buffer doesn't make sense, all empty strings should
 373   // reuse g_strEmpty
 374   wxASSERT( nLen >  0 );
 375
 376   // make sure that we don't overflow
 377   wxASSERT( nLen < (INT_MAX / sizeof(wxChar)) -
 378                    (sizeof(wxStringData) + EXTRA_ALLOC + 1) );
 379
 380   STATISTICS_ADD(Length, nLen);
 381
 382   // allocate memory:
 383   // 1) one extra character for '\0' termination
 384   // 2) sizeof(wxStringData) for housekeeping info
 385   wxStringData* pData = (wxStringData*)
 386     malloc(sizeof(wxStringData) + (nLen + EXTRA_ALLOC + 1)*sizeof(wxChar));
 387   pData->nRefs        = 1;
 388   pData->nDataLength  = nLen;
 389   pData->nAllocLength = nLen + EXTRA_ALLOC;
 390   m_pchData           = pData->data();  // data starts after wxStringData
 391   m_pchData[nLen]     = wxT('\0');
 392 }
 393
 394 // must be called before changing this string
 395 void wxString::CopyBeforeWrite()
 396 {
 397   wxStringData* pData = GetStringData();
 398
 399   if ( pData->IsShared() ) {
 400     pData->Unlock();                // memory not freed because shared
 401     size_t nLen = pData->nDataLength;
 402     AllocBuffer(nLen);
 403     memcpy(m_pchData, pData->data(), nLen*sizeof(wxChar));
 404   }
 405
 406   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 407 }
 408
 409 // must be called before replacing contents of this string
 410 void wxString::AllocBeforeWrite(size_t nLen)
 411 {
 412   wxASSERT( nLen != 0 );  // doesn't make any sense
 413
 414   // must not share string and must have enough space
 415   wxStringData* pData = GetStringData();
 416   if ( pData->IsShared() || pData->IsEmpty() ) {
 417     // can't work with old buffer, get new one
 418     pData->Unlock();
 419     AllocBuffer(nLen);
 420   }
 421   else {
 422     if ( nLen > pData->nAllocLength ) {
 423       // realloc the buffer instead of calling malloc() again, this is more
 424       // efficient
 425       STATISTICS_ADD(Length, nLen);
 426
 427       nLen += EXTRA_ALLOC;
 428
 429       wxStringData *pDataOld = pData;
 430       pData = (wxStringData*)
 431           realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 432       if ( !pData ) {
 433         // out of memory
 434         free(pDataOld);
 435
 436         // FIXME we're going to crash...
 437         return;
 438       }
 439
 440       pData->nAllocLength = nLen;
 441       m_pchData = pData->data();
 442     }
 443
 444     // now we have enough space, just update the string length
 445     pData->nDataLength = nLen;
 446   }
 447
 448   wxASSERT( !GetStringData()->IsShared() );  // we must be the only owner
 449 }
 450
 451 // allocate enough memory for nLen characters
 452 void wxString::Alloc(size_t nLen)
 453 {
 454   wxStringData *pData = GetStringData();
 455   if ( pData->nAllocLength <= nLen ) {
 456     if ( pData->IsEmpty() ) {
 457       nLen += EXTRA_ALLOC;
 458
 459       wxStringData* pData = (wxStringData*)
 460         malloc(sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 461       pData->nRefs = 1;
 462       pData->nDataLength = 0;
 463       pData->nAllocLength = nLen;
 464       m_pchData = pData->data();  // data starts after wxStringData
 465       m_pchData[0u] = wxT('\0');
 466     }
 467     else if ( pData->IsShared() ) {
 468       pData->Unlock();                // memory not freed because shared
 469       size_t nOldLen = pData->nDataLength;
 470       AllocBuffer(nLen);
 471       memcpy(m_pchData, pData->data(), nOldLen*sizeof(wxChar));
 472     }
 473     else {
 474       nLen += EXTRA_ALLOC;
 475
 476       wxStringData *pDataOld = pData;
 477       wxStringData *p = (wxStringData *)
 478         realloc(pData, sizeof(wxStringData) + (nLen + 1)*sizeof(wxChar));
 479
 480       if ( p == NULL ) {
 481         // don't leak memory
 482         free(pDataOld);
 483
 484         // FIXME what to do on memory error?
 485         return;
 486       }
 487
 488       // it's not important if the pointer changed or not (the check for this
 489       // is not faster than assigning to m_pchData in all cases)
 490       p->nAllocLength = nLen;
 491       m_pchData = p->data();
 492     }
 493   }
 494   //else: we've already got enough
 495 }
 496
 497 // shrink to minimal size (releasing extra memory)
 498 void wxString::Shrink()
 499 {
 500   wxStringData *pData = GetStringData();
 501
 502   // this variable is unused in release build, so avoid the compiler warning
 503   // by just not declaring it
 504 #ifdef __WXDEBUG__
 505   void *p =
 506 #endif
 507   realloc(pData, sizeof(wxStringData) + (pData->nDataLength + 1)*sizeof(wxChar));
 508
 509   // we rely on a reasonable realloc() implementation here - so far I haven't
 510   // seen any which wouldn't behave like this
 511
 512   wxASSERT( p != NULL );  // can't free memory?
 513   wxASSERT( p == pData ); // we're decrementing the size - block shouldn't move!
 514 }
 515
 516 // get the pointer to writable buffer of (at least) nLen bytes
 517 wxChar *wxString::GetWriteBuf(size_t nLen)
 518 {
 519   AllocBeforeWrite(nLen);
 520
 521   wxASSERT( GetStringData()->nRefs == 1 );
 522   GetStringData()->Validate(FALSE);
 523
 524   return m_pchData;
 525 }
 526
 527 // put string back in a reasonable state after GetWriteBuf
 528 void wxString::UngetWriteBuf()
 529 {
 530   GetStringData()->nDataLength = wxStrlen(m_pchData);
 531   GetStringData()->Validate(TRUE);
 532 }
 533
 534 // ---------------------------------------------------------------------------
 535 // data access
 536 // ---------------------------------------------------------------------------
 537
 538 // all functions are inline in string.h
 539
 540 // ---------------------------------------------------------------------------
 541 // assignment operators
 542 // ---------------------------------------------------------------------------
 543
 544 // helper function: does real copy
 545 void wxString::AssignCopy(size_t nSrcLen, const wxChar *pszSrcData)
 546 {
 547   if ( nSrcLen == 0 ) {
 548     Reinit();
 549   }
 550   else {
 551     AllocBeforeWrite(nSrcLen);
 552     memcpy(m_pchData, pszSrcData, nSrcLen*sizeof(wxChar));
 553     GetStringData()->nDataLength = nSrcLen;
 554     m_pchData[nSrcLen] = wxT('\0');
 555   }
 556 }
 557
 558 // assigns one string to another
 559 wxString& wxString::operator=(const wxString& stringSrc)
 560 {
 561   wxASSERT( stringSrc.GetStringData()->IsValid() );
 562
 563   // don't copy string over itself
 564   if ( m_pchData != stringSrc.m_pchData ) {
 565     if ( stringSrc.GetStringData()->IsEmpty() ) {
 566       Reinit();
 567     }
 568     else {
 569       // adjust references
 570       GetStringData()->Unlock();
 571       m_pchData = stringSrc.m_pchData;
 572       GetStringData()->Lock();
 573     }
 574   }
 575
 576   return *this;
 577 }
 578
 579 // assigns a single character
 580 wxString& wxString::operator=(wxChar ch)
 581 {
 582   AssignCopy(1, &ch);
 583   return *this;
 584 }
 585
 586 // assigns C string
 587 wxString& wxString::operator=(const wxChar *psz)
 588 {
 589   AssignCopy(wxStrlen(psz), psz);
 590   return *this;
 591 }
 592
 593 #if !wxUSE_UNICODE
 594
 595 // same as 'signed char' variant
 596 wxString& wxString::operator=(const unsigned char* psz)
 597 {
 598   *this = (const char *)psz;
 599   return *this;
 600 }
 601
 602 #if wxUSE_WCHAR_T
 603 wxString& wxString::operator=(const wchar_t *pwz)
 604 {
 605   wxString str(pwz);
 606   *this = str;
 607   return *this;
 608 }
 609 #endif
 610
 611 #endif
 612
 613 // ---------------------------------------------------------------------------
 614 // string concatenation
 615 // ---------------------------------------------------------------------------
 616
 617 // add something to this string
 618 void wxString::ConcatSelf(int nSrcLen, const wxChar *pszSrcData)
 619 {
 620   STATISTICS_ADD(SummandLength, nSrcLen);
 621
 622   // concatenating an empty string is a NOP
 623   if ( nSrcLen > 0 ) {
 624     wxStringData *pData = GetStringData();
 625     size_t nLen = pData->nDataLength;
 626     size_t nNewLen = nLen + nSrcLen;
 627
 628     // alloc new buffer if current is too small
 629     if ( pData->IsShared() ) {
 630       STATISTICS_ADD(ConcatHit, 0);
 631
 632       // we have to allocate another buffer
 633       wxStringData* pOldData = GetStringData();
 634       AllocBuffer(nNewLen);
 635       memcpy(m_pchData, pOldData->data(), nLen*sizeof(wxChar));
 636       pOldData->Unlock();
 637     }
 638     else if ( nNewLen > pData->nAllocLength ) {
 639       STATISTICS_ADD(ConcatHit, 0);
 640
 641       // we have to grow the buffer
 642       Alloc(nNewLen);
 643     }
 644     else {
 645       STATISTICS_ADD(ConcatHit, 1);
 646
 647       // the buffer is already big enough
 648     }
 649
 650     // should be enough space
 651     wxASSERT( nNewLen <= GetStringData()->nAllocLength );
 652
 653     // fast concatenation - all is done in our buffer
 654     memcpy(m_pchData + nLen, pszSrcData, nSrcLen*sizeof(wxChar));
 655
 656     m_pchData[nNewLen] = wxT('\0');          // put terminating '\0'
 657     GetStringData()->nDataLength = nNewLen; // and fix the length
 658   }
 659   //else: the string to append was empty
 660 }
 661
 662 /*
 663  * concatenation functions come in 5 flavours:
 664  *  string + string
 665  *  char   + string      and      string + char
 666  *  C str  + string      and      string + C str
 667  */
 668
 669 wxString operator+(const wxString& string1, const wxString& string2)
 670 {
 671   wxASSERT( string1.GetStringData()->IsValid() );
 672   wxASSERT( string2.GetStringData()->IsValid() );
 673
 674   wxString s = string1;
 675   s += string2;
 676
 677   return s;
 678 }
 679
 680 wxString operator+(const wxString& string, wxChar ch)
 681 {
 682   wxASSERT( string.GetStringData()->IsValid() );
 683
 684   wxString s = string;
 685   s += ch;
 686
 687   return s;
 688 }
 689
 690 wxString operator+(wxChar ch, const wxString& string)
 691 {
 692   wxASSERT( string.GetStringData()->IsValid() );
 693
 694   wxString s = ch;
 695   s += string;
 696
 697   return s;
 698 }
 699
 700 wxString operator+(const wxString& string, const wxChar *psz)
 701 {
 702   wxASSERT( string.GetStringData()->IsValid() );
 703
 704   wxString s;
 705   s.Alloc(wxStrlen(psz) + string.Len());
 706   s = string;
 707   s += psz;
 708
 709   return s;
 710 }
 711
 712 wxString operator+(const wxChar *psz, const wxString& string)
 713 {
 714   wxASSERT( string.GetStringData()->IsValid() );
 715
 716   wxString s;
 717   s.Alloc(wxStrlen(psz) + string.Len());
 718   s = psz;
 719   s += string;
 720
 721   return s;
 722 }
 723
 724 // ===========================================================================
 725 // other common string functions
 726 // ===========================================================================
 727
 728 // ---------------------------------------------------------------------------
 729 // simple sub-string extraction
 730 // ---------------------------------------------------------------------------
 731
 732 // helper function: clone the data attached to this string
 733 void wxString::AllocCopy(wxString& dest, int nCopyLen, int nCopyIndex) const
 734 {
 735   if ( nCopyLen == 0 ) {
 736     dest.Init();
 737   }
 738   else {
 739     dest.AllocBuffer(nCopyLen);
 740     memcpy(dest.m_pchData, m_pchData + nCopyIndex, nCopyLen*sizeof(wxChar));
 741   }
 742 }
 743
 744 // extract string of length nCount starting at nFirst
 745 wxString wxString::Mid(size_t nFirst, size_t nCount) const
 746 {
 747   wxStringData *pData = GetStringData();
 748   size_t nLen = pData->nDataLength;
 749
 750   // default value of nCount is wxSTRING_MAXLEN and means "till the end"
 751   if ( nCount == wxSTRING_MAXLEN )
 752   {
 753     nCount = nLen - nFirst;
 754   }
 755
 756   // out-of-bounds requests return sensible things
 757   if ( nFirst + nCount > nLen )
 758   {
 759     nCount = nLen - nFirst;
 760   }
 761
 762   if ( nFirst > nLen )
 763   {
 764     // AllocCopy() will return empty string
 765     nCount = 0;
 766   }
 767
 768   wxString dest;
 769   AllocCopy(dest, nCount, nFirst);
 770
 771   return dest;
 772 }
 773
 774 // extract nCount last (rightmost) characters
 775 wxString wxString::Right(size_t nCount) const
 776 {
 777   if ( nCount > (size_t)GetStringData()->nDataLength )
 778     nCount = GetStringData()->nDataLength;
 779
 780   wxString dest;
 781   AllocCopy(dest, nCount, GetStringData()->nDataLength - nCount);
 782   return dest;
 783 }
 784
 785 // get all characters after the last occurence of ch
 786 // (returns the whole string if ch not found)
 787 wxString wxString::AfterLast(wxChar ch) const
 788 {
 789   wxString str;
 790   int iPos = Find(ch, TRUE);
 791   if ( iPos == wxNOT_FOUND )
 792     str = *this;
 793   else
 794     str = c_str() + iPos + 1;
 795
 796   return str;
 797 }
 798
 799 // extract nCount first (leftmost) characters
 800 wxString wxString::Left(size_t nCount) const
 801 {
 802   if ( nCount > (size_t)GetStringData()->nDataLength )
 803     nCount = GetStringData()->nDataLength;
 804
 805   wxString dest;
 806   AllocCopy(dest, nCount, 0);
 807   return dest;
 808 }
 809
 810 // get all characters before the first occurence of ch
 811 // (returns the whole string if ch not found)
 812 wxString wxString::BeforeFirst(wxChar ch) const
 813 {
 814   wxString str;
 815   for ( const wxChar *pc = m_pchData; *pc != wxT('\0') && *pc != ch; pc++ )
 816     str += *pc;
 817
 818   return str;
 819 }
 820
 821 /// get all characters before the last occurence of ch
 822 /// (returns empty string if ch not found)
 823 wxString wxString::BeforeLast(wxChar ch) const
 824 {
 825   wxString str;
 826   int iPos = Find(ch, TRUE);
 827   if ( iPos != wxNOT_FOUND && iPos != 0 )
 828     str = wxString(c_str(), iPos);
 829
 830   return str;
 831 }
 832
 833 /// get all characters after the first occurence of ch
 834 /// (returns empty string if ch not found)
 835 wxString wxString::AfterFirst(wxChar ch) const
 836 {
 837   wxString str;
 838   int iPos = Find(ch);
 839   if ( iPos != wxNOT_FOUND )
 840     str = c_str() + iPos + 1;
 841
 842   return str;
 843 }
 844
 845 // replace first (or all) occurences of some substring with another one
 846 size_t wxString::Replace(const wxChar *szOld, const wxChar *szNew, bool bReplaceAll)
 847 {
 848   size_t uiCount = 0;   // count of replacements made
 849
 850   size_t uiOldLen = wxStrlen(szOld);
 851
 852   wxString strTemp;
 853   const wxChar *pCurrent = m_pchData;
 854   const wxChar *pSubstr;
 855   while ( *pCurrent != wxT('\0') ) {
 856     pSubstr = wxStrstr(pCurrent, szOld);
 857     if ( pSubstr == NULL ) {
 858       // strTemp is unused if no replacements were made, so avoid the copy
 859       if ( uiCount == 0 )
 860         return 0;
 861
 862       strTemp += pCurrent;    // copy the rest
 863       break;                  // exit the loop
 864     }
 865     else {
 866       // take chars before match
 867       strTemp.ConcatSelf(pSubstr - pCurrent, pCurrent);
 868       strTemp += szNew;
 869       pCurrent = pSubstr + uiOldLen;  // restart after match
 870
 871       uiCount++;
 872
 873       // stop now?
 874       if ( !bReplaceAll ) {
 875         strTemp += pCurrent;    // copy the rest
 876         break;                  // exit the loop
 877       }
 878     }
 879   }
 880
 881   // only done if there were replacements, otherwise would have returned above
 882   *this = strTemp;
 883
 884   return uiCount;
 885 }
 886
 887 bool wxString::IsAscii() const
 888 {
 889   const wxChar *s = (const wxChar*) *this;
 890   while(*s){
 891     if(!isascii(*s)) return(FALSE);
 892     s++;
 893   }
 894   return(TRUE);
 895 }
 896
 897 bool wxString::IsWord() const
 898 {
 899   const wxChar *s = (const wxChar*) *this;
 900   while(*s){
 901     if(!wxIsalpha(*s)) return(FALSE);
 902     s++;
 903   }
 904   return(TRUE);
 905 }
 906
 907 bool wxString::IsNumber() const
 908 {
 909   const wxChar *s = (const wxChar*) *this;
 910   while(*s){
 911     if(!wxIsdigit(*s)) return(FALSE);
 912     s++;
 913   }
 914   return(TRUE);
 915 }
 916
 917 wxString wxString::Strip(stripType w) const
 918 {
 919     wxString s = *this;
 920     if ( w & leading ) s.Trim(FALSE);
 921     if ( w & trailing ) s.Trim(TRUE);
 922     return s;
 923 }
 924
 925 // ---------------------------------------------------------------------------
 926 // case conversion
 927 // ---------------------------------------------------------------------------
 928
 929 wxString& wxString::MakeUpper()
 930 {
 931   CopyBeforeWrite();
 932
 933   for ( wxChar *p = m_pchData; *p; p++ )
 934     *p = (wxChar)wxToupper(*p);
 935
 936   return *this;
 937 }
 938
 939 wxString& wxString::MakeLower()
 940 {
 941   CopyBeforeWrite();
 942
 943   for ( wxChar *p = m_pchData; *p; p++ )
 944     *p = (wxChar)wxTolower(*p);
 945
 946   return *this;
 947 }
 948
 949 // ---------------------------------------------------------------------------
 950 // trimming and padding
 951 // ---------------------------------------------------------------------------
 952
 953 // trims spaces (in the sense of isspace) from left or right side
 954 wxString& wxString::Trim(bool bFromRight)
 955 {
 956   // first check if we're going to modify the string at all
 957   if ( !IsEmpty() &&
 958        (
 959         (bFromRight && wxIsspace(GetChar(Len() - 1))) ||
 960         (!bFromRight && wxIsspace(GetChar(0u)))
 961        )
 962      )
 963   {
 964     // ok, there is at least one space to trim
 965     CopyBeforeWrite();
 966
 967     if ( bFromRight )
 968     {
 969       // find last non-space character
 970       wxChar *psz = m_pchData + GetStringData()->nDataLength - 1;
 971       while ( wxIsspace(*psz) && (psz >= m_pchData) )
 972         psz--;
 973
 974       // truncate at trailing space start
 975       *++psz = wxT('\0');
 976       GetStringData()->nDataLength = psz - m_pchData;
 977     }
 978     else
 979     {
 980       // find first non-space character
 981       const wxChar *psz = m_pchData;
 982       while ( wxIsspace(*psz) )
 983         psz++;
 984
 985       // fix up data and length
 986       int nDataLength = GetStringData()->nDataLength - (psz - (const wxChar*) m_pchData);
 987       memmove(m_pchData, psz, (nDataLength + 1)*sizeof(wxChar));
 988       GetStringData()->nDataLength = nDataLength;
 989     }
 990   }
 991
 992   return *this;
 993 }
 994
 995 // adds nCount characters chPad to the string from either side
 996 wxString& wxString::Pad(size_t nCount, wxChar chPad, bool bFromRight)
 997 {
 998   wxString s(chPad, nCount);
 999
1000   if ( bFromRight )
1001     *this += s;
1002   else
1003   {
1004     s += *this;
1005     *this = s;
1006   }
1007
1008   return *this;
1009 }
1010
1011 // truncate the string
1012 wxString& wxString::Truncate(size_t uiLen)
1013 {
1014   if ( uiLen < Len() ) {
1015     CopyBeforeWrite();
1016
1017     *(m_pchData + uiLen) = wxT('\0');
1018     GetStringData()->nDataLength = uiLen;
1019   }
1020   //else: nothing to do, string is already short enough
1021
1022   return *this;
1023 }
1024
1025 // ---------------------------------------------------------------------------
1026 // finding (return wxNOT_FOUND if not found and index otherwise)
1027 // ---------------------------------------------------------------------------
1028
1029 // find a character
1030 int wxString::Find(wxChar ch, bool bFromEnd) const
1031 {
1032   const wxChar *psz = bFromEnd ? wxStrrchr(m_pchData, ch) : wxStrchr(m_pchData, ch);
1033
1034   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1035 }
1036
1037 // find a sub-string (like strstr)
1038 int wxString::Find(const wxChar *pszSub) const
1039 {
1040   const wxChar *psz = wxStrstr(m_pchData, pszSub);
1041
1042   return (psz == NULL) ? wxNOT_FOUND : psz - (const wxChar*) m_pchData;
1043 }
1044
1045 // ----------------------------------------------------------------------------
1046 // conversion to numbers
1047 // ----------------------------------------------------------------------------
1048
1049 bool wxString::ToLong(long *val) const
1050 {
1051     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToLong") );
1052
1053     const wxChar *start = c_str();
1054     wxChar *end;
1055     *val = wxStrtol(start, &end, 10);
1056
1057     // return TRUE only if scan was stopped by the terminating NUL and if the
1058     // string was not empty to start with
1059     return !*end && (end != start);
1060 }
1061
1062 bool wxString::ToULong(unsigned long *val) const
1063 {
1064     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToULong") );
1065
1066     const wxChar *start = c_str();
1067     wxChar *end;
1068     *val = wxStrtoul(start, &end, 10);
1069
1070     // return TRUE only if scan was stopped by the terminating NUL and if the
1071     // string was not empty to start with
1072     return !*end && (end != start);
1073 }
1074
1075 bool wxString::ToDouble(double *val) const
1076 {
1077     wxCHECK_MSG( val, FALSE, _T("NULL pointer in wxString::ToDouble") );
1078
1079     const wxChar *start = c_str();
1080     wxChar *end;
1081     *val = wxStrtod(start, &end);
1082
1083     // return TRUE only if scan was stopped by the terminating NUL and if the
1084     // string was not empty to start with
1085     return !*end && (end != start);
1086 }
1087
1088 // ---------------------------------------------------------------------------
1089 // formatted output
1090 // ---------------------------------------------------------------------------
1091
1092 /* static */
1093 wxString wxString::Format(const wxChar *pszFormat, ...)
1094 {
1095     va_list argptr;
1096     va_start(argptr, pszFormat);
1097
1098     wxString s;
1099     s.PrintfV(pszFormat, argptr);
1100
1101     va_end(argptr);
1102
1103     return s;
1104 }
1105
1106 /* static */
1107 wxString wxString::FormatV(const wxChar *pszFormat, va_list argptr)
1108 {
1109     wxString s;
1110     s.Printf(pszFormat, argptr);
1111     return s;
1112 }
1113
1114 int wxString::Printf(const wxChar *pszFormat, ...)
1115 {
1116   va_list argptr;
1117   va_start(argptr, pszFormat);
1118
1119   int iLen = PrintfV(pszFormat, argptr);
1120
1121   va_end(argptr);
1122
1123   return iLen;
1124 }
1125
1126 int wxString::PrintfV(const wxChar* pszFormat, va_list argptr)
1127 {
1128 #if wxUSE_EXPERIMENTAL_PRINTF
1129   // the new implementation
1130
1131   // buffer to avoid dynamic memory allocation each time for small strings
1132   char szScratch[1024];
1133
1134   Reinit();
1135   for (size_t n = 0; pszFormat[n]; n++)
1136     if (pszFormat[n] == wxT('%')) {
1137       static char s_szFlags[256] = "%";
1138       size_t flagofs = 1;
1139       bool adj_left = FALSE, in_prec = FALSE,
1140            prec_dot = FALSE, done = FALSE;
1141       int ilen = 0;
1142       size_t min_width = 0, max_width = wxSTRING_MAXLEN;
1143       do {
1144 #define CHECK_PREC if (in_prec && !prec_dot) { s_szFlags[flagofs++] = '.'; prec_dot = TRUE; }
1145         switch (pszFormat[++n]) {
1146         case wxT('\0'):
1147           done = TRUE;
1148           break;
1149         case wxT('%'):
1150           *this += wxT('%');
1151           done = TRUE;
1152           break;
1153         case wxT('#'):
1154         case wxT('0'):
1155         case wxT(' '):
1156         case wxT('+'):
1157         case wxT('\''):
1158           CHECK_PREC
1159           s_szFlags[flagofs++] = pszFormat[n];
1160           break;
1161         case wxT('-'):
1162           CHECK_PREC
1163           adj_left = TRUE;
1164           s_szFlags[flagofs++] = pszFormat[n];
1165           break;
1166         case wxT('.'):
1167           CHECK_PREC
1168           in_prec = TRUE;
1169           prec_dot = FALSE;
1170           max_width = 0;
1171           // dot will be auto-added to s_szFlags if non-negative number follows
1172           break;
1173         case wxT('h'):
1174           ilen = -1;
1175           CHECK_PREC
1176           s_szFlags[flagofs++] = pszFormat[n];
1177           break;
1178         case wxT('l'):
1179           ilen = 1;
1180           CHECK_PREC
1181           s_szFlags[flagofs++] = pszFormat[n];
1182           break;
1183         case wxT('q'):
1184         case wxT('L'):
1185           ilen = 2;
1186           CHECK_PREC
1187           s_szFlags[flagofs++] = pszFormat[n];
1188           break;
1189         case wxT('Z'):
1190           ilen = 3;
1191           CHECK_PREC
1192           s_szFlags[flagofs++] = pszFormat[n];
1193           break;
1194         case wxT('*'):
1195           {
1196             int len = va_arg(argptr, int);
1197             if (in_prec) {
1198               if (len<0) break;
1199               CHECK_PREC
1200               max_width = len;
1201             } else {
1202               if (len<0) {
1203                 adj_left = !adj_left;
1204                 s_szFlags[flagofs++] = '-';
1205                 len = -len;
1206               }
1207               min_width = len;
1208             }
1209             flagofs += ::sprintf(s_szFlags+flagofs,"%d",len);
1210           }
1211           break;
1212         case wxT('1'): case wxT('2'): case wxT('3'):
1213         case wxT('4'): case wxT('5'): case wxT('6'):
1214         case wxT('7'): case wxT('8'): case wxT('9'):
1215           {
1216             int len = 0;
1217             CHECK_PREC
1218             while ((pszFormat[n]>=wxT('0')) && (pszFormat[n]<=wxT('9'))) {
1219               s_szFlags[flagofs++] = pszFormat[n];
1220               len = len*10 + (pszFormat[n] - wxT('0'));
1221               n++;
1222             }
1223             if (in_prec) max_width = len;
1224             else min_width = len;
1225             n--; // the main loop pre-increments n again
1226           }
1227           break;
1228         case wxT('d'):
1229         case wxT('i'):
1230         case wxT('o'):
1231         case wxT('u'):
1232         case wxT('x'):
1233         case wxT('X'):
1234           CHECK_PREC
1235           s_szFlags[flagofs++] = pszFormat[n];
1236           s_szFlags[flagofs] = '\0';
1237           if (ilen == 0 ) {
1238             int val = va_arg(argptr, int);
1239             ::sprintf(szScratch, s_szFlags, val);
1240           }
1241           else if (ilen == -1) {
1242             short int val = va_arg(argptr, short int);
1243             ::sprintf(szScratch, s_szFlags, val);
1244           }
1245           else if (ilen == 1) {
1246             long int val = va_arg(argptr, long int);
1247             ::sprintf(szScratch, s_szFlags, val);
1248           }
1249           else if (ilen == 2) {
1250 #if SIZEOF_LONG_LONG
1251             long long int val = va_arg(argptr, long long int);
1252             ::sprintf(szScratch, s_szFlags, val);
1253 #else
1254             long int val = va_arg(argptr, long int);
1255             ::sprintf(szScratch, s_szFlags, val);
1256 #endif
1257           }
1258           else if (ilen == 3) {
1259             size_t val = va_arg(argptr, size_t);
1260             ::sprintf(szScratch, s_szFlags, val);
1261           }
1262           *this += wxString(szScratch);
1263           done = TRUE;
1264           break;
1265         case wxT('e'):
1266         case wxT('E'):
1267         case wxT('f'):
1268         case wxT('g'):
1269         case wxT('G'):
1270           CHECK_PREC
1271           s_szFlags[flagofs++] = pszFormat[n];
1272           s_szFlags[flagofs] = '\0';
1273           if (ilen == 2) {
1274             long double val = va_arg(argptr, long double);
1275             ::sprintf(szScratch, s_szFlags, val);
1276           } else {
1277             double val = va_arg(argptr, double);
1278             ::sprintf(szScratch, s_szFlags, val);
1279           }
1280           *this += wxString(szScratch);
1281           done = TRUE;
1282           break;
1283         case wxT('p'):
1284           {
1285             void *val = va_arg(argptr, void *);
1286             CHECK_PREC
1287             s_szFlags[flagofs++] = pszFormat[n];
1288             s_szFlags[flagofs] = '\0';
1289             ::sprintf(szScratch, s_szFlags, val);
1290             *this += wxString(szScratch);
1291             done = TRUE;
1292           }
1293           break;
1294         case wxT('c'):
1295           {
1296             wxChar val = va_arg(argptr, int);
1297             // we don't need to honor padding here, do we?
1298             *this += val;
1299             done = TRUE;
1300           }
1301           break;
1302         case wxT('s'):
1303           if (ilen == -1) {
1304             // wx extension: we'll let %hs mean non-Unicode strings
1305             char *val = va_arg(argptr, char *);
1306 #if wxUSE_UNICODE
1307             // ASCII->Unicode constructor handles max_width right
1308             wxString s(val, wxConvLibc, max_width);
1309 #else
1310             size_t len = wxSTRING_MAXLEN;
1311             if (val) {
1312               for (len = 0; val[len] && (len<max_width); len++);
1313             } else val = wxT("(null)");
1314             wxString s(val, len);
1315 #endif
1316             if (s.Len() < min_width)
1317               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1318             *this += s;
1319           } else {
1320             wxChar *val = va_arg(argptr, wxChar *);
1321             size_t len = wxSTRING_MAXLEN;
1322             if (val) {
1323               for (len = 0; val[len] && (len<max_width); len++);
1324             } else val = wxT("(null)");
1325             wxString s(val, len);
1326             if (s.Len() < min_width)
1327               s.Pad(min_width - s.Len(), wxT(' '), adj_left);
1328             *this += s;
1329           }
1330           done = TRUE;
1331           break;
1332         case wxT('n'):
1333           if (ilen == 0) {
1334             int *val = va_arg(argptr, int *);
1335             *val = Len();
1336           }
1337           else if (ilen == -1) {
1338             short int *val = va_arg(argptr, short int *);
1339             *val = Len();
1340           }
1341           else if (ilen >= 1) {
1342             long int *val = va_arg(argptr, long int *);
1343             *val = Len();
1344           }
1345           done = TRUE;
1346           break;
1347         default:
1348           if (wxIsalpha(pszFormat[n]))
1349             // probably some flag not taken care of here yet
1350             s_szFlags[flagofs++] = pszFormat[n];
1351           else {
1352             // bad format
1353             *this += wxT('%'); // just to pass the glibc tst-printf.c
1354             n--;
1355             done = TRUE;
1356           }
1357           break;
1358         }
1359 #undef CHECK_PREC
1360       } while (!done);
1361     } else *this += pszFormat[n];
1362
1363 #else
1364   // buffer to avoid dynamic memory allocation each time for small strings
1365   char szScratch[1024];
1366
1367   // NB: wxVsnprintf() may return either less than the buffer size or -1 if
1368   //     there is not enough place depending on implementation
1369   int iLen = wxVsnprintfA(szScratch, WXSIZEOF(szScratch), pszFormat, argptr);
1370   if ( iLen != -1 ) {
1371     // the whole string is in szScratch
1372     *this = szScratch;
1373   }
1374   else {
1375       bool outOfMemory = FALSE;
1376       int size = 2*WXSIZEOF(szScratch);
1377       while ( !outOfMemory ) {
1378           char *buf = GetWriteBuf(size);
1379           if ( buf )
1380             iLen = wxVsnprintfA(buf, size, pszFormat, argptr);
1381           else
1382             outOfMemory = TRUE;
1383
1384           UngetWriteBuf();
1385
1386           if ( iLen != -1 ) {
1387               // ok, there was enough space
1388               break;
1389           }
1390
1391           // still not enough, double it again
1392           size *= 2;
1393       }
1394
1395       if ( outOfMemory ) {
1396           // out of memory
1397           return -1;
1398       }
1399   }
1400 #endif // wxUSE_EXPERIMENTAL_PRINTF/!wxUSE_EXPERIMENTAL_PRINTF
1401
1402   return Len();
1403 }
1404
1405 // ----------------------------------------------------------------------------
1406 // misc other operations
1407 // ----------------------------------------------------------------------------
1408
1409 // returns TRUE if the string matches the pattern which may contain '*' and
1410 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1411 // of them)
1412 bool wxString::Matches(const wxChar *pszMask) const
1413 {
1414   // check char by char
1415   const wxChar *pszTxt;
1416   for ( pszTxt = c_str(); *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
1417     switch ( *pszMask ) {
1418       case wxT('?'):
1419         if ( *pszTxt == wxT('\0') )
1420           return FALSE;
1421
1422         // pszText and pszMask will be incremented in the loop statement
1423
1424         break;
1425
1426       case wxT('*'):
1427         {
1428           // ignore special chars immediately following this one
1429           while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
1430             pszMask++;
1431
1432           // if there is nothing more, match
1433           if ( *pszMask == wxT('\0') )
1434             return TRUE;
1435
1436           // are there any other metacharacters in the mask?
1437           size_t uiLenMask;
1438           const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
1439
1440           if ( pEndMask != NULL ) {
1441             // we have to match the string between two metachars
1442             uiLenMask = pEndMask - pszMask;
1443           }
1444           else {
1445             // we have to match the remainder of the string
1446             uiLenMask = wxStrlen(pszMask);
1447           }
1448
1449           wxString strToMatch(pszMask, uiLenMask);
1450           const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
1451           if ( pMatch == NULL )
1452             return FALSE;
1453
1454           // -1 to compensate "++" in the loop
1455           pszTxt = pMatch + uiLenMask - 1;
1456           pszMask += uiLenMask - 1;
1457         }
1458         break;
1459
1460       default:
1461         if ( *pszMask != *pszTxt )
1462           return FALSE;
1463         break;
1464     }
1465   }
1466
1467   // match only if nothing left
1468   return *pszTxt == wxT('\0');
1469 }
1470
1471 // Count the number of chars
1472 int wxString::Freq(wxChar ch) const
1473 {
1474     int count = 0;
1475     int len = Len();
1476     for (int i = 0; i < len; i++)
1477     {
1478         if (GetChar(i) == ch)
1479             count ++;
1480     }
1481     return count;
1482 }
1483
1484 // convert to upper case, return the copy of the string
1485 wxString wxString::Upper() const
1486 { wxString s(*this); return s.MakeUpper(); }
1487
1488 // convert to lower case, return the copy of the string
1489 wxString wxString::Lower() const { wxString s(*this); return s.MakeLower(); }
1490
1491 int wxString::sprintf(const wxChar *pszFormat, ...)
1492   {
1493     va_list argptr;
1494     va_start(argptr, pszFormat);
1495     int iLen = PrintfV(pszFormat, argptr);
1496     va_end(argptr);
1497     return iLen;
1498   }
1499
1500 // ---------------------------------------------------------------------------
1501 // standard C++ library string functions
1502 // ---------------------------------------------------------------------------
1503 #ifdef  wxSTD_STRING_COMPATIBILITY
1504
1505 wxString& wxString::insert(size_t nPos, const wxString& str)
1506 {
1507   wxASSERT( str.GetStringData()->IsValid() );
1508   wxASSERT( nPos <= Len() );
1509
1510   if ( !str.IsEmpty() ) {
1511     wxString strTmp;
1512     wxChar *pc = strTmp.GetWriteBuf(Len() + str.Len());
1513     wxStrncpy(pc, c_str(), nPos);
1514     wxStrcpy(pc + nPos, str);
1515     wxStrcpy(pc + nPos + str.Len(), c_str() + nPos);
1516     strTmp.UngetWriteBuf();
1517     *this = strTmp;
1518   }
1519
1520   return *this;
1521 }
1522
1523 size_t wxString::find(const wxString& str, size_t nStart) const
1524 {
1525   wxASSERT( str.GetStringData()->IsValid() );
1526   wxASSERT( nStart <= Len() );
1527
1528   const wxChar *p = wxStrstr(c_str() + nStart, str);
1529
1530   return p == NULL ? npos : p - c_str();
1531 }
1532
1533 // VC++ 1.5 can't cope with the default argument in the header.
1534 #if !defined(__VISUALC__) || defined(__WIN32__)
1535 size_t wxString::find(const wxChar* sz, size_t nStart, size_t n) const
1536 {
1537   return find(wxString(sz, n), nStart);
1538 }
1539 #endif // VC++ 1.5
1540
1541 // Gives a duplicate symbol (presumably a case-insensitivity problem)
1542 #if !defined(__BORLANDC__)
1543 size_t wxString::find(wxChar ch, size_t nStart) const
1544 {
1545   wxASSERT( nStart <= Len() );
1546
1547   const wxChar *p = wxStrchr(c_str() + nStart, ch);
1548
1549   return p == NULL ? npos : p - c_str();
1550 }
1551 #endif
1552
1553 size_t wxString::rfind(const wxString& str, size_t nStart) const
1554 {
1555   wxASSERT( str.GetStringData()->IsValid() );
1556   wxASSERT( nStart <= Len() );
1557
1558   // TODO could be made much quicker than that
1559   const wxChar *p = c_str() + (nStart == npos ? Len() : nStart);
1560   while ( p >= c_str() + str.Len() ) {
1561     if ( wxStrncmp(p - str.Len(), str, str.Len()) == 0 )
1562       return p - str.Len() - c_str();
1563     p--;
1564   }
1565
1566   return npos;
1567 }
1568
1569 // VC++ 1.5 can't cope with the default argument in the header.
1570 #if !defined(__VISUALC__) || defined(__WIN32__)
1571 size_t wxString::rfind(const wxChar* sz, size_t nStart, size_t n) const
1572 {
1573     return rfind(wxString(sz, n == npos ? 0 : n), nStart);
1574 }
1575
1576 size_t wxString::rfind(wxChar ch, size_t nStart) const
1577 {
1578     if ( nStart == npos )
1579     {
1580         nStart = Len();
1581     }
1582     else
1583     {
1584         wxASSERT( nStart <= Len() );
1585     }
1586
1587     const wxChar *p = wxStrrchr(c_str(), ch);
1588
1589     if ( p == NULL )
1590         return npos;
1591
1592     size_t result = p - c_str();
1593     return ( result > nStart ) ? npos : result;
1594 }
1595 #endif // VC++ 1.5
1596
1597 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
1598 {
1599     const wxChar *start = c_str() + nStart;
1600     const wxChar *firstOf = wxStrpbrk(start, sz);
1601     if ( firstOf )
1602         return firstOf - start;
1603     else
1604         return npos;
1605 }
1606
1607 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
1608 {
1609     if ( nStart == npos )
1610     {
1611         nStart = Len();
1612     }
1613     else
1614     {
1615         wxASSERT( nStart <= Len() );
1616     }
1617
1618     for ( const wxChar *p = c_str() + length() - 1; p >= c_str(); p-- )
1619     {
1620         if ( wxStrchr(sz, *p) )
1621             return p - c_str();
1622     }
1623
1624     return npos;
1625 }
1626
1627 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
1628 {
1629     if ( nStart == npos )
1630     {
1631         nStart = Len();
1632     }
1633     else
1634     {
1635         wxASSERT( nStart <= Len() );
1636     }
1637
1638     size_t nAccept = wxStrspn(c_str() + nStart, sz);
1639     if ( nAccept >= length() - nStart )
1640         return npos;
1641     else
1642         return nAccept;
1643 }
1644
1645 size_t wxString::find_first_not_of(wxChar ch, size_t nStart) const
1646 {
1647     wxASSERT( nStart <= Len() );
1648
1649     for ( const wxChar *p = c_str() + nStart; *p; p++ )
1650     {
1651         if ( *p != ch )
1652             return p - c_str();
1653     }
1654
1655     return npos;
1656 }
1657
1658 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
1659 {
1660     if ( nStart == npos )
1661     {
1662         nStart = Len();
1663     }
1664     else
1665     {
1666         wxASSERT( nStart <= Len() );
1667     }
1668
1669     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1670     {
1671         if ( !wxStrchr(sz, *p) )
1672             return p - c_str();
1673     }
1674
1675     return npos;
1676 }
1677
1678 size_t wxString::find_last_not_of(wxChar ch, size_t nStart) const
1679 {
1680     if ( nStart == npos )
1681     {
1682         nStart = Len();
1683     }
1684     else
1685     {
1686         wxASSERT( nStart <= Len() );
1687     }
1688
1689     for ( const wxChar *p = c_str() + nStart - 1; p >= c_str(); p-- )
1690     {
1691         if ( *p != ch )
1692             return p - c_str();
1693     }
1694
1695     return npos;
1696 }
1697
1698 wxString& wxString::erase(size_t nStart, size_t nLen)
1699 {
1700   wxString strTmp(c_str(), nStart);
1701   if ( nLen != npos ) {
1702     wxASSERT( nStart + nLen <= Len() );
1703
1704     strTmp.append(c_str() + nStart + nLen);
1705   }
1706
1707   *this = strTmp;
1708   return *this;
1709 }
1710
1711 wxString& wxString::replace(size_t nStart, size_t nLen, const wxChar *sz)
1712 {
1713   wxASSERT( nStart + nLen <= wxStrlen(sz) );
1714
1715   wxString strTmp;
1716   if ( nStart != 0 )
1717     strTmp.append(c_str(), nStart);
1718   strTmp += sz;
1719   strTmp.append(c_str() + nStart + nLen);
1720
1721   *this = strTmp;
1722   return *this;
1723 }
1724
1725 wxString& wxString::replace(size_t nStart, size_t nLen, size_t nCount, wxChar ch)
1726 {
1727   return replace(nStart, nLen, wxString(ch, nCount));
1728 }
1729
1730 wxString& wxString::replace(size_t nStart, size_t nLen,
1731                             const wxString& str, size_t nStart2, size_t nLen2)
1732 {
1733   return replace(nStart, nLen, str.substr(nStart2, nLen2));
1734 }
1735
1736 wxString& wxString::replace(size_t nStart, size_t nLen,
1737                         const wxChar* sz, size_t nCount)
1738 {
1739   return replace(nStart, nLen, wxString(sz, nCount));
1740 }
1741
1742 #endif  //std::string compatibility
1743
1744 // ============================================================================
1745 // ArrayString
1746 // ============================================================================
1747
1748 // size increment = max(50% of current size, ARRAY_MAXSIZE_INCREMENT)
1749 #define   ARRAY_MAXSIZE_INCREMENT       4096
1750 #ifndef   ARRAY_DEFAULT_INITIAL_SIZE    // also defined in dynarray.h
1751   #define   ARRAY_DEFAULT_INITIAL_SIZE    (16)
1752 #endif
1753
1754 #define   STRING(p)   ((wxString *)(&(p)))
1755
1756 // ctor
1757 wxArrayString::wxArrayString(bool autoSort)
1758 {
1759   m_nSize  =
1760   m_nCount = 0;
1761   m_pItems = (wxChar **) NULL;
1762   m_autoSort = autoSort;
1763 }
1764
1765 // copy ctor
1766 wxArrayString::wxArrayString(const wxArrayString& src)
1767 {
1768   m_nSize  =
1769   m_nCount = 0;
1770   m_pItems = (wxChar **) NULL;
1771   m_autoSort = src.m_autoSort;
1772
1773   *this = src;
1774 }
1775
1776 // assignment operator
1777 wxArrayString& wxArrayString::operator=(const wxArrayString& src)
1778 {
1779   if ( m_nSize > 0 )
1780     Clear();
1781
1782   Copy(src);
1783
1784   return *this;
1785 }
1786
1787 void wxArrayString::Copy(const wxArrayString& src)
1788 {
1789   if ( src.m_nCount > ARRAY_DEFAULT_INITIAL_SIZE )
1790     Alloc(src.m_nCount);
1791
1792   for ( size_t n = 0; n < src.m_nCount; n++ )
1793     Add(src[n]);
1794 }
1795
1796 // grow the array
1797 void wxArrayString::Grow()
1798 {
1799   // only do it if no more place
1800   if( m_nCount == m_nSize ) {
1801     if( m_nSize == 0 ) {
1802       // was empty, alloc some memory
1803       m_nSize = ARRAY_DEFAULT_INITIAL_SIZE;
1804       m_pItems = new wxChar *[m_nSize];
1805     }
1806     else {
1807       // otherwise when it's called for the first time, nIncrement would be 0
1808       // and the array would never be expanded
1809 #if defined(__VISAGECPP__) && defined(__WXDEBUG__)
1810       int array_size = ARRAY_DEFAULT_INITIAL_SIZE;
1811       wxASSERT( array_size != 0 );
1812 #else
1813       wxASSERT( ARRAY_DEFAULT_INITIAL_SIZE != 0 );
1814 #endif
1815
1816       // add 50% but not too much
1817       size_t nIncrement = m_nSize < ARRAY_DEFAULT_INITIAL_SIZE
1818                           ? ARRAY_DEFAULT_INITIAL_SIZE : m_nSize >> 1;
1819       if ( nIncrement > ARRAY_MAXSIZE_INCREMENT )
1820         nIncrement = ARRAY_MAXSIZE_INCREMENT;
1821       m_nSize += nIncrement;
1822       wxChar **pNew = new wxChar *[m_nSize];
1823
1824       // copy data to new location
1825       memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1826
1827       // delete old memory (but do not release the strings!)
1828       wxDELETEA(m_pItems);
1829
1830       m_pItems = pNew;
1831     }
1832   }
1833 }
1834
1835 void wxArrayString::Free()
1836 {
1837   for ( size_t n = 0; n < m_nCount; n++ ) {
1838     STRING(m_pItems[n])->GetStringData()->Unlock();
1839   }
1840 }
1841
1842 // deletes all the strings from the list
1843 void wxArrayString::Empty()
1844 {
1845   Free();
1846
1847   m_nCount = 0;
1848 }
1849
1850 // as Empty, but also frees memory
1851 void wxArrayString::Clear()
1852 {
1853   Free();
1854
1855   m_nSize  =
1856   m_nCount = 0;
1857
1858   wxDELETEA(m_pItems);
1859 }
1860
1861 // dtor
1862 wxArrayString::~wxArrayString()
1863 {
1864   Free();
1865
1866   wxDELETEA(m_pItems);
1867 }
1868
1869 // pre-allocates memory (frees the previous data!)
1870 void wxArrayString::Alloc(size_t nSize)
1871 {
1872   wxASSERT( nSize > 0 );
1873
1874   // only if old buffer was not big enough
1875   if ( nSize > m_nSize ) {
1876     Free();
1877     wxDELETEA(m_pItems);
1878     m_pItems = new wxChar *[nSize];
1879     m_nSize  = nSize;
1880   }
1881
1882   m_nCount = 0;
1883 }
1884
1885 // minimizes the memory usage by freeing unused memory
1886 void wxArrayString::Shrink()
1887 {
1888   // only do it if we have some memory to free
1889   if( m_nCount < m_nSize ) {
1890     // allocates exactly as much memory as we need
1891     wxChar **pNew = new wxChar *[m_nCount];
1892
1893     // copy data to new location
1894     memcpy(pNew, m_pItems, m_nCount*sizeof(wxChar *));
1895     delete [] m_pItems;
1896     m_pItems = pNew;
1897   }
1898 }
1899
1900 // searches the array for an item (forward or backwards)
1901 int wxArrayString::Index(const wxChar *sz, bool bCase, bool bFromEnd) const
1902 {
1903   if ( m_autoSort ) {
1904     // use binary search in the sorted array
1905     wxASSERT_MSG( bCase && !bFromEnd,
1906                   wxT("search parameters ignored for auto sorted array") );
1907
1908     size_t i,
1909            lo = 0,
1910            hi = m_nCount;
1911     int res;
1912     while ( lo < hi ) {
1913       i = (lo + hi)/2;
1914
1915       res = wxStrcmp(sz, m_pItems[i]);
1916       if ( res < 0 )
1917         hi = i;
1918       else if ( res > 0 )
1919         lo = i + 1;
1920       else
1921         return i;
1922     }
1923
1924     return wxNOT_FOUND;
1925   }
1926   else {
1927     // use linear search in unsorted array
1928     if ( bFromEnd ) {
1929       if ( m_nCount > 0 ) {
1930         size_t ui = m_nCount;
1931         do {
1932           if ( STRING(m_pItems[--ui])->IsSameAs(sz, bCase) )
1933             return ui;
1934         }
1935         while ( ui != 0 );
1936       }
1937     }
1938     else {
1939       for( size_t ui = 0; ui < m_nCount; ui++ ) {
1940         if( STRING(m_pItems[ui])->IsSameAs(sz, bCase) )
1941           return ui;
1942       }
1943     }
1944   }
1945
1946   return wxNOT_FOUND;
1947 }
1948
1949 // add item at the end
1950 size_t wxArrayString::Add(const wxString& str)
1951 {
1952   if ( m_autoSort ) {
1953     // insert the string at the correct position to keep the array sorted
1954     size_t i,
1955            lo = 0,
1956            hi = m_nCount;
1957     int res;
1958     while ( lo < hi ) {
1959       i = (lo + hi)/2;
1960
1961       res = wxStrcmp(str, m_pItems[i]);
1962       if ( res < 0 )
1963         hi = i;
1964       else if ( res > 0 )
1965         lo = i + 1;
1966       else {
1967         lo = hi = i;
1968         break;
1969       }
1970     }
1971
1972     wxASSERT_MSG( lo == hi, wxT("binary search broken") );
1973
1974     Insert(str, lo);
1975
1976     return (size_t)lo;
1977   }
1978   else {
1979     wxASSERT( str.GetStringData()->IsValid() );
1980
1981     Grow();
1982
1983     // the string data must not be deleted!
1984     str.GetStringData()->Lock();
1985
1986     // just append
1987     m_pItems[m_nCount] = (wxChar *)str.c_str(); // const_cast
1988
1989     return m_nCount++;
1990   }
1991 }
1992
1993 // add item at the given position
1994 void wxArrayString::Insert(const wxString& str, size_t nIndex)
1995 {
1996   wxASSERT( str.GetStringData()->IsValid() );
1997
1998   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Insert") );
1999
2000   Grow();
2001
2002   memmove(&m_pItems[nIndex + 1], &m_pItems[nIndex],
2003           (m_nCount - nIndex)*sizeof(wxChar *));
2004
2005   str.GetStringData()->Lock();
2006   m_pItems[nIndex] = (wxChar *)str.c_str();
2007
2008   m_nCount++;
2009 }
2010
2011 // removes item from array (by index)
2012 void wxArrayString::Remove(size_t nIndex)
2013 {
2014   wxCHECK_RET( nIndex <= m_nCount, wxT("bad index in wxArrayString::Remove") );
2015
2016   // release our lock
2017   Item(nIndex).GetStringData()->Unlock();
2018
2019   memmove(&m_pItems[nIndex], &m_pItems[nIndex + 1],
2020           (m_nCount - nIndex - 1)*sizeof(wxChar *));
2021   m_nCount--;
2022 }
2023
2024 // removes item from array (by value)
2025 void wxArrayString::Remove(const wxChar *sz)
2026 {
2027   int iIndex = Index(sz);
2028
2029   wxCHECK_RET( iIndex != wxNOT_FOUND,
2030                wxT("removing inexistent element in wxArrayString::Remove") );
2031
2032   Remove(iIndex);
2033 }
2034
2035 // ----------------------------------------------------------------------------
2036 // sorting
2037 // ----------------------------------------------------------------------------
2038
2039 // we can only sort one array at a time with the quick-sort based
2040 // implementation
2041 #if wxUSE_THREADS
2042   // need a critical section to protect access to gs_compareFunction and
2043   // gs_sortAscending variables
2044   static wxCriticalSection *gs_critsectStringSort = NULL;
2045
2046   // call this before the value of the global sort vars is changed/after
2047   // you're finished with them
2048   #define START_SORT()     wxASSERT( !gs_critsectStringSort );                \
2049                            gs_critsectStringSort = new wxCriticalSection;     \
2050                            gs_critsectStringSort->Enter()
2051   #define END_SORT()       gs_critsectStringSort->Leave();                    \
2052                            delete gs_critsectStringSort;                      \
2053                            gs_critsectStringSort = NULL
2054 #else // !threads
2055   #define START_SORT()
2056   #define END_SORT()
2057 #endif // wxUSE_THREADS
2058
2059 // function to use for string comparaison
2060 static wxArrayString::CompareFunction gs_compareFunction = NULL;
2061
2062 // if we don't use the compare function, this flag tells us if we sort the
2063 // array in ascending or descending order
2064 static bool gs_sortAscending = TRUE;
2065
2066 // function which is called by quick sort
2067 static int LINKAGEMODE wxStringCompareFunction(const void *first, const void *second)
2068 {
2069   wxString *strFirst = (wxString *)first;
2070   wxString *strSecond = (wxString *)second;
2071
2072   if ( gs_compareFunction ) {
2073     return gs_compareFunction(*strFirst, *strSecond);
2074   }
2075   else {
2076     // maybe we should use wxStrcoll
2077     int result = wxStrcmp(strFirst->c_str(), strSecond->c_str());
2078
2079     return gs_sortAscending ? result : -result;
2080   }
2081 }
2082
2083 // sort array elements using passed comparaison function
2084 void wxArrayString::Sort(CompareFunction compareFunction)
2085 {
2086   START_SORT();
2087
2088   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2089   gs_compareFunction = compareFunction;
2090
2091   DoSort();
2092
2093   END_SORT();
2094 }
2095
2096 void wxArrayString::Sort(bool reverseOrder)
2097 {
2098   START_SORT();
2099
2100   wxASSERT( !gs_compareFunction );  // must have been reset to NULL
2101   gs_sortAscending = !reverseOrder;
2102
2103   DoSort();
2104
2105   END_SORT();
2106 }
2107
2108 void wxArrayString::DoSort()
2109 {
2110   wxCHECK_RET( !m_autoSort, wxT("can't use this method with sorted arrays") );
2111
2112   // just sort the pointers using qsort() - of course it only works because
2113   // wxString() *is* a pointer to its data
2114   qsort(m_pItems, m_nCount, sizeof(wxChar *), wxStringCompareFunction);
2115 }
2116