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