include/wx/stringimpl.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        wx/stringimpl.h
   3 // Purpose:     wxStringImpl class, implementation of wxString
   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 licence
  10 ///////////////////////////////////////////////////////////////////////////////
  11
  12 /*
  13     This header implements std::string-like string class, wxStringImpl, that is
  14     used by wxString to store the data. Alternatively, if wxUSE_STL=1,
  15     wxStringImpl is just a typedef to std:: string class.
  16 */
  17
  18 #ifndef _WX_WXSTRINGIMPL_H__
  19 #define _WX_WXSTRINGIMPL_H__
  20
  21 // ----------------------------------------------------------------------------
  22 // headers
  23 // ----------------------------------------------------------------------------
  24
  25 #include "wx/defs.h"        // everybody should include this
  26 #include "wx/chartype.h"    // for wxChar
  27 #include "wx/wxcrtbase.h"   // for wxStrlen() etc.
  28
  29 #include <stdlib.h>
  30
  31 // ---------------------------------------------------------------------------
  32 // macros
  33 // ---------------------------------------------------------------------------
  34
  35 // implementation only
  36 #define   wxASSERT_VALID_INDEX(i) \
  37     wxASSERT_MSG( (size_t)(i) <= length(), _T("invalid index in wxString") )
  38
  39
  40 // ----------------------------------------------------------------------------
  41 // global data
  42 // ----------------------------------------------------------------------------
  43
  44 // global pointer to empty string
  45 extern WXDLLIMPEXP_DATA_BASE(const wxChar*) wxEmptyString;
  46 #if wxUSE_UNICODE_UTF8
  47 // FIXME-UTF8: we should have only one wxEmptyString
  48 extern WXDLLIMPEXP_DATA_BASE(const wxStringCharType*) wxEmptyStringImpl;
  49 #endif
  50
  51
  52 // ----------------------------------------------------------------------------
  53 // deal with STL/non-STL/non-STL-but-wxUSE_STD_STRING
  54 // ----------------------------------------------------------------------------
  55
  56 #define wxUSE_STL_BASED_WXSTRING  wxUSE_STL
  57
  58 // in both cases we need to define wxStdString
  59 #if wxUSE_STL_BASED_WXSTRING || wxUSE_STD_STRING
  60
  61 #include "wx/beforestd.h"
  62 #include <string>
  63 #include "wx/afterstd.h"
  64
  65 #ifdef HAVE_STD_WSTRING
  66     typedef std::wstring wxStdWideString;
  67 #else
  68     typedef std::basic_string<wchar_t> wxStdWideString;
  69 #endif
  70
  71 #if wxUSE_UNICODE_WCHAR
  72     typedef wxStdWideString wxStdString;
  73 #else
  74     typedef std::string wxStdString;
  75 #endif
  76
  77 #endif // wxUSE_STL_BASED_WXSTRING || wxUSE_STD_STRING
  78
  79
  80 #if wxUSE_STL_BASED_WXSTRING
  81
  82     // we always want ctor from std::string when using std::string internally
  83     #undef wxUSE_STD_STRING
  84     #define wxUSE_STD_STRING 1
  85
  86     #if (defined(__GNUG__) && (__GNUG__ < 3)) || \
  87         (defined(_MSC_VER) && (_MSC_VER <= 1200))
  88         #define wxSTRING_BASE_HASNT_CLEAR
  89     #endif
  90
  91     typedef wxStdString wxStringImpl;
  92 #else // if !wxUSE_STL_BASED_WXSTRING
  93
  94 // in non-STL mode, compare() is implemented in wxString and not wxStringImpl
  95 #undef HAVE_STD_STRING_COMPARE
  96
  97 // ---------------------------------------------------------------------------
  98 // string data prepended with some housekeeping info (used by wxString class),
  99 // is never used directly (but had to be put here to allow inlining)
 100 // ---------------------------------------------------------------------------
 101
 102 struct WXDLLIMPEXP_BASE wxStringData
 103 {
 104   int     nRefs;        // reference count
 105   size_t  nDataLength,  // actual string length
 106           nAllocLength; // allocated memory size
 107
 108   // mimics declaration 'wxStringCharType data[nAllocLength]'
 109   wxStringCharType* data() const { return (wxStringCharType*)(this + 1); }
 110
 111   // empty string has a special ref count so it's never deleted
 112   bool  IsEmpty()   const { return (nRefs == -1); }
 113   bool  IsShared()  const { return (nRefs > 1);   }
 114
 115   // lock/unlock
 116   void  Lock()   { if ( !IsEmpty() ) nRefs++;                    }
 117
 118   // VC++ will refuse to inline Unlock but profiling shows that it is wrong
 119 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 120   __forceinline
 121 #endif
 122   // VC++ free must take place in same DLL as allocation when using non dll
 123   // run-time library (e.g. Multithreaded instead of Multithreaded DLL)
 124 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 125   void  Unlock() { if ( !IsEmpty() && --nRefs == 0) Free();  }
 126   // we must not inline deallocation since allocation is not inlined
 127   void  Free();
 128 #else
 129   void  Unlock() { if ( !IsEmpty() && --nRefs == 0) free(this);  }
 130 #endif
 131
 132   // if we had taken control over string memory (GetWriteBuf), it's
 133   // intentionally put in invalid state
 134   void  Validate(bool b)  { nRefs = (b ? 1 : 0); }
 135   bool  IsValid() const   { return (nRefs != 0); }
 136 };
 137
 138 class WXDLLIMPEXP_BASE wxStringImpl
 139 {
 140 public:
 141   // an 'invalid' value for string index, moved to this place due to a CW bug
 142   static const size_t npos;
 143
 144 protected:
 145   // points to data preceded by wxStringData structure with ref count info
 146   wxStringCharType *m_pchData;
 147
 148   // accessor to string data
 149   wxStringData* GetStringData() const { return (wxStringData*)m_pchData - 1; }
 150
 151   // string (re)initialization functions
 152     // initializes the string to the empty value (must be called only from
 153     // ctors, use Reinit() otherwise)
 154 #if wxUSE_UNICODE_UTF8
 155   void Init() { m_pchData = (wxStringCharType *)wxEmptyStringImpl; } // FIXME-UTF8
 156 #else
 157   void Init() { m_pchData = (wxStringCharType *)wxEmptyString; }
 158 #endif
 159     // initializes the string with (a part of) C-string
 160   void InitWith(const wxStringCharType *psz, size_t nPos = 0, size_t nLen = npos);
 161     // as Init, but also frees old data
 162   void Reinit() { GetStringData()->Unlock(); Init(); }
 163
 164   // memory allocation
 165     // allocates memory for string of length nLen
 166   bool AllocBuffer(size_t nLen);
 167     // effectively copies data to string
 168   bool AssignCopy(size_t, const wxStringCharType *);
 169
 170   // append a (sub)string
 171   bool ConcatSelf(size_t nLen, const wxStringCharType *src, size_t nMaxLen);
 172   bool ConcatSelf(size_t nLen, const wxStringCharType *src)
 173     { return ConcatSelf(nLen, src, nLen); }
 174
 175   // functions called before writing to the string: they copy it if there
 176   // are other references to our data (should be the only owner when writing)
 177   bool CopyBeforeWrite();
 178   bool AllocBeforeWrite(size_t);
 179
 180     // compatibility with wxString
 181   bool Alloc(size_t nLen);
 182
 183 public:
 184   // standard types
 185   typedef wxStringCharType value_type;
 186   typedef wxStringCharType char_type;
 187   typedef size_t size_type;
 188   typedef value_type& reference;
 189   typedef const value_type& const_reference;
 190   typedef value_type* pointer;
 191   typedef const value_type* const_pointer;
 192
 193   // macro to define the bulk of iterator and const_iterator classes
 194   #define WX_DEFINE_STRINGIMPL_ITERATOR(iterator_name, ref_type, ptr_type)    \
 195     public:                                                                   \
 196         typedef wxStringCharType value_type;                                  \
 197         typedef ref_type reference;                                           \
 198         typedef ptr_type pointer;                                             \
 199         typedef int difference_type;                                          \
 200                                                                               \
 201         iterator_name() : m_ptr(NULL) { }                                     \
 202         iterator_name(pointer ptr) : m_ptr(ptr) { }                           \
 203                                                                               \
 204         reference operator*() const { return *m_ptr; }                        \
 205                                                                               \
 206         iterator_name& operator++() { m_ptr++; return *this; }                \
 207         iterator_name operator++(int)                                         \
 208         {                                                                     \
 209             const iterator_name tmp(*this);                                   \
 210             m_ptr++;                                                          \
 211             return tmp;                                                       \
 212         }                                                                     \
 213                                                                               \
 214         iterator_name& operator--() { m_ptr--; return *this; }                \
 215         iterator_name operator--(int)                                         \
 216         {                                                                     \
 217             const iterator_name tmp(*this);                                   \
 218             m_ptr--;                                                          \
 219             return tmp;                                                       \
 220         }                                                                     \
 221                                                                               \
 222         iterator_name operator+(ptrdiff_t n) const                            \
 223             { return iterator_name(m_ptr + n); }                              \
 224         iterator_name operator-(ptrdiff_t n) const                            \
 225             { return iterator_name(m_ptr - n); }                              \
 226         iterator_name& operator+=(ptrdiff_t n)                                \
 227             { m_ptr += n; return *this; }                                     \
 228         iterator_name& operator-=(ptrdiff_t n)                                \
 229             { m_ptr -= n; return *this; }                                     \
 230                                                                               \
 231         difference_type operator-(const iterator_name& i) const               \
 232             { return m_ptr - i.m_ptr; }                                       \
 233                                                                               \
 234         bool operator==(const iterator_name& i) const                         \
 235           { return m_ptr == i.m_ptr; }                                        \
 236         bool operator!=(const iterator_name& i) const                         \
 237           { return m_ptr != i.m_ptr; }                                        \
 238                                                                               \
 239         bool operator<(const iterator_name& i) const                          \
 240           { return m_ptr < i.m_ptr; }                                         \
 241         bool operator>(const iterator_name& i) const                          \
 242           { return m_ptr > i.m_ptr; }                                         \
 243         bool operator<=(const iterator_name& i) const                         \
 244           { return m_ptr <= i.m_ptr; }                                        \
 245         bool operator>=(const iterator_name& i) const                         \
 246           { return m_ptr >= i.m_ptr; }                                        \
 247                                                                               \
 248     private:                                                                  \
 249         /* for wxStringImpl use only */                                       \
 250         pointer GetPtr() const { return m_ptr; }                              \
 251                                                                               \
 252         friend class wxStringImpl;                                            \
 253                                                                               \
 254         pointer m_ptr
 255
 256   // we need to declare const_iterator in wxStringImpl scope, the friend
 257   // declaration inside iterator class itself is not enough, or at least not
 258   // for g++ 3.4 (g++ 4 is ok)
 259   class WXDLLIMPEXP_FWD_BASE const_iterator;
 260
 261   class WXDLLIMPEXP_BASE iterator
 262   {
 263     WX_DEFINE_STRINGIMPL_ITERATOR(iterator,
 264                                   wxStringCharType&,
 265                                   wxStringCharType*);
 266
 267     friend class const_iterator;
 268   };
 269
 270   class WXDLLIMPEXP_BASE const_iterator
 271   {
 272   public:
 273       const_iterator(iterator i) : m_ptr(i.m_ptr) { }
 274
 275       WX_DEFINE_STRINGIMPL_ITERATOR(const_iterator,
 276                                     const wxStringCharType&,
 277                                     const wxStringCharType*);
 278   };
 279
 280   #undef WX_DEFINE_STRINGIMPL_ITERATOR
 281
 282
 283   // constructors and destructor
 284     // ctor for an empty string
 285   wxStringImpl() { Init(); }
 286     // copy ctor
 287   wxStringImpl(const wxStringImpl& stringSrc)
 288   {
 289     wxASSERT_MSG( stringSrc.GetStringData()->IsValid(),
 290                   _T("did you forget to call UngetWriteBuf()?") );
 291
 292     if ( stringSrc.empty() ) {
 293       // nothing to do for an empty string
 294       Init();
 295     }
 296     else {
 297       m_pchData = stringSrc.m_pchData;            // share same data
 298       GetStringData()->Lock();                    // => one more copy
 299     }
 300   }
 301     // string containing nRepeat copies of ch
 302   wxStringImpl(size_type nRepeat, wxStringCharType ch);
 303     // ctor takes first nLength characters from C string
 304     // (default value of npos means take all the string)
 305   wxStringImpl(const wxStringCharType *psz)
 306       { InitWith(psz, 0, npos); }
 307   wxStringImpl(const wxStringCharType *psz, size_t nLength)
 308       { InitWith(psz, 0, nLength); }
 309     // take nLen chars starting at nPos
 310   wxStringImpl(const wxStringImpl& str, size_t nPos, size_t nLen)
 311   {
 312     wxASSERT_MSG( str.GetStringData()->IsValid(),
 313                   _T("did you forget to call UngetWriteBuf()?") );
 314     Init();
 315     size_t strLen = str.length() - nPos; nLen = strLen < nLen ? strLen : nLen;
 316     InitWith(str.c_str(), nPos, nLen);
 317   }
 318     // take everything between start and end
 319   wxStringImpl(const_iterator start, const_iterator end);
 320
 321
 322     // ctor from and conversion to std::string
 323 #if wxUSE_STD_STRING
 324   wxStringImpl(const wxStdString& impl)
 325       { InitWith(impl.c_str(), 0, impl.length()); }
 326
 327   operator wxStdString() const
 328       { return wxStdString(c_str(), length()); }
 329 #endif
 330
 331
 332     // dtor is not virtual, this class must not be inherited from!
 333   ~wxStringImpl()
 334   {
 335 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 336       //RN - according to the above VC++ does indeed inline this,
 337       //even though it spits out two warnings
 338       #pragma warning (disable:4714)
 339 #endif
 340
 341       GetStringData()->Unlock();
 342   }
 343
 344 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 345     //re-enable inlining warning
 346     #pragma warning (default:4714)
 347 #endif
 348   // overloaded assignment
 349     // from another wxString
 350   wxStringImpl& operator=(const wxStringImpl& stringSrc);
 351     // from a character
 352   wxStringImpl& operator=(wxStringCharType ch);
 353     // from a C string
 354   wxStringImpl& operator=(const wxStringCharType *psz);
 355
 356     // return the length of the string
 357   size_type length() const { return GetStringData()->nDataLength; }
 358     // return the length of the string
 359   size_type size() const { return length(); }
 360     // return the maximum size of the string
 361   size_type max_size() const { return npos; }
 362     // resize the string, filling the space with c if c != 0
 363   void resize(size_t nSize, wxStringCharType ch = '\0');
 364     // delete the contents of the string
 365   void clear() { erase(0, npos); }
 366     // returns true if the string is empty
 367   bool empty() const { return length() == 0; }
 368     // inform string about planned change in size
 369   void reserve(size_t sz) { Alloc(sz); }
 370   size_type capacity() const { return GetStringData()->nAllocLength; }
 371
 372   // lib.string.access
 373     // return the character at position n
 374   value_type at(size_type n) const
 375     { wxASSERT_VALID_INDEX( n ); return m_pchData[n]; }
 376     // returns the writable character at position n
 377   reference at(size_type n)
 378   {
 379     wxASSERT_VALID_INDEX( n );
 380     CopyBeforeWrite();
 381     return m_pchData[n];
 382   } // FIXME-UTF8: not useful for us...?
 383
 384   // lib.string.modifiers
 385     // append elements str[pos], ..., str[pos+n]
 386   wxStringImpl& append(const wxStringImpl& str, size_t pos, size_t n)
 387   {
 388     wxASSERT(pos <= str.length());
 389     ConcatSelf(n, str.c_str() + pos, str.length() - pos);
 390     return *this;
 391   }
 392     // append a string
 393   wxStringImpl& append(const wxStringImpl& str)
 394     { ConcatSelf(str.length(), str.c_str()); return *this; }
 395     // append first n (or all if n == npos) characters of sz
 396   wxStringImpl& append(const wxStringCharType *sz)
 397     { ConcatSelf(wxStrlen(sz), sz); return *this; }
 398   wxStringImpl& append(const wxStringCharType *sz, size_t n)
 399     { ConcatSelf(n, sz); return *this; }
 400     // append n copies of ch
 401   wxStringImpl& append(size_t n, wxStringCharType ch);
 402     // append from first to last
 403   wxStringImpl& append(const_iterator first, const_iterator last)
 404     { ConcatSelf(last - first, first.GetPtr()); return *this; }
 405
 406     // same as `this_string = str'
 407   wxStringImpl& assign(const wxStringImpl& str)
 408     { return *this = str; }
 409     // same as ` = str[pos..pos + n]
 410   wxStringImpl& assign(const wxStringImpl& str, size_t pos, size_t n)
 411     { return replace(0, npos, str, pos, n); }
 412     // same as `= first n (or all if n == npos) characters of sz'
 413   wxStringImpl& assign(const wxStringCharType *sz)
 414     { return replace(0, npos, sz, wxStrlen(sz)); }
 415   wxStringImpl& assign(const wxStringCharType *sz, size_t n)
 416     { return replace(0, npos, sz, n); }
 417     // same as `= n copies of ch'
 418   wxStringImpl& assign(size_t n, wxStringCharType ch)
 419     { return replace(0, npos, n, ch); }
 420     // assign from first to last
 421   wxStringImpl& assign(const_iterator first, const_iterator last)
 422     { return replace(begin(), end(), first, last); }
 423
 424     // first valid index position
 425   const_iterator begin() const { return m_pchData; }
 426   iterator begin();
 427     // position one after the last valid one
 428   const_iterator end() const { return m_pchData + length(); }
 429   iterator end();
 430
 431     // insert another string
 432   wxStringImpl& insert(size_t nPos, const wxStringImpl& str)
 433   {
 434     wxASSERT( str.GetStringData()->IsValid() );
 435     return insert(nPos, str.c_str(), str.length());
 436   }
 437     // insert n chars of str starting at nStart (in str)
 438   wxStringImpl& insert(size_t nPos, const wxStringImpl& str, size_t nStart, size_t n)
 439   {
 440     wxASSERT( str.GetStringData()->IsValid() );
 441     wxASSERT( nStart < str.length() );
 442     size_t strLen = str.length() - nStart;
 443     n = strLen < n ? strLen : n;
 444     return insert(nPos, str.c_str() + nStart, n);
 445   }
 446     // insert first n (or all if n == npos) characters of sz
 447   wxStringImpl& insert(size_t nPos, const wxStringCharType *sz, size_t n = npos);
 448     // insert n copies of ch
 449   wxStringImpl& insert(size_t nPos, size_t n, wxStringCharType ch)
 450     { return insert(nPos, wxStringImpl(n, ch)); }
 451   iterator insert(iterator it, wxStringCharType ch)
 452     { size_t idx = it - begin(); insert(idx, 1, ch); return begin() + idx; }
 453   void insert(iterator it, const_iterator first, const_iterator last)
 454     { insert(it - begin(), first.GetPtr(), last - first); }
 455   void insert(iterator it, size_type n, wxStringCharType ch)
 456     { insert(it - begin(), n, ch); }
 457
 458     // delete characters from nStart to nStart + nLen
 459   wxStringImpl& erase(size_type pos = 0, size_type n = npos);
 460   iterator erase(iterator first, iterator last)
 461   {
 462     size_t idx = first - begin();
 463     erase(idx, last - first);
 464     return begin() + idx;
 465   }
 466   iterator erase(iterator first);
 467
 468   // explicit conversion to C string (use this with printf()!)
 469   const wxStringCharType* c_str() const { return m_pchData; }
 470   const wxStringCharType* data() const { return m_pchData; }
 471
 472     // replaces the substring of length nLen starting at nStart
 473   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringCharType* sz)
 474     { return replace(nStart, nLen, sz, npos); }
 475     // replaces the substring of length nLen starting at nStart
 476   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringImpl& str)
 477     { return replace(nStart, nLen, str.c_str(), str.length()); }
 478     // replaces the substring with nCount copies of ch
 479   wxStringImpl& replace(size_t nStart, size_t nLen,
 480                         size_t nCount, wxStringCharType ch)
 481     { return replace(nStart, nLen, wxStringImpl(nCount, ch)); }
 482     // replaces a substring with another substring
 483   wxStringImpl& replace(size_t nStart, size_t nLen,
 484                         const wxStringImpl& str, size_t nStart2, size_t nLen2)
 485     { return replace(nStart, nLen, str.substr(nStart2, nLen2)); }
 486     // replaces the substring with first nCount chars of sz
 487   wxStringImpl& replace(size_t nStart, size_t nLen,
 488                         const wxStringCharType* sz, size_t nCount);
 489
 490   wxStringImpl& replace(iterator first, iterator last, const_pointer s)
 491     { return replace(first - begin(), last - first, s); }
 492   wxStringImpl& replace(iterator first, iterator last, const_pointer s,
 493                         size_type n)
 494     { return replace(first - begin(), last - first, s, n); }
 495   wxStringImpl& replace(iterator first, iterator last, const wxStringImpl& s)
 496     { return replace(first - begin(), last - first, s); }
 497   wxStringImpl& replace(iterator first, iterator last, size_type n, wxStringCharType c)
 498     { return replace(first - begin(), last - first, n, c); }
 499   wxStringImpl& replace(iterator first, iterator last,
 500                         const_iterator first1, const_iterator last1)
 501     { return replace(first - begin(), last - first, first1.GetPtr(), last1 - first1); }
 502
 503     // swap two strings
 504   void swap(wxStringImpl& str);
 505
 506     // All find() functions take the nStart argument which specifies the
 507     // position to start the search on, the default value is 0. All functions
 508     // return npos if there were no match.
 509
 510     // find a substring
 511   size_t find(const wxStringImpl& str, size_t nStart = 0) const;
 512
 513     // find first n characters of sz
 514   size_t find(const wxStringCharType* sz, size_t nStart = 0, size_t n = npos) const;
 515
 516     // find the first occurrence of character ch after nStart
 517   size_t find(wxStringCharType ch, size_t nStart = 0) const;
 518
 519     // rfind() family is exactly like find() but works right to left
 520
 521     // as find, but from the end
 522   size_t rfind(const wxStringImpl& str, size_t nStart = npos) const;
 523
 524     // as find, but from the end
 525   size_t rfind(const wxStringCharType* sz, size_t nStart = npos,
 526                size_t n = npos) const;
 527     // as find, but from the end
 528   size_t rfind(wxStringCharType ch, size_t nStart = npos) const;
 529
 530   size_type copy(wxStringCharType* s, size_type n, size_type pos = 0);
 531
 532   // substring extraction
 533   wxStringImpl substr(size_t nStart = 0, size_t nLen = npos) const;
 534
 535       // string += string
 536   wxStringImpl& operator+=(const wxStringImpl& s) { return append(s); }
 537       // string += C string
 538   wxStringImpl& operator+=(const wxStringCharType *psz) { return append(psz); }
 539       // string += char
 540   wxStringImpl& operator+=(wxStringCharType ch) { return append(1, ch); }
 541
 542   // helpers for wxStringBuffer and wxStringBufferLength
 543   wxStringCharType *DoGetWriteBuf(size_t nLen);
 544   void DoUngetWriteBuf();
 545   void DoUngetWriteBuf(size_t nLen);
 546
 547   friend class WXDLLIMPEXP_FWD_BASE wxString;
 548 };
 549
 550 #endif // !wxUSE_STL_BASED_WXSTRING
 551
 552 // don't pollute the library user's name space
 553 #undef wxASSERT_VALID_INDEX
 554
 555 #endif  // _WX_WXSTRINGIMPL_H__