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