src/common/string.cpp

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