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                                                                               \
 188         iterator_name(pointer ptr) : m_ptr(ptr) { }                           \
 189                                                                               \
 190         reference operator*() const { return *m_ptr; }                        \
 191                                                                               \
 192         iterator_name& operator++() { m_ptr++; return *this; }                \
 193         iterator_name operator++(int)                                         \
 194         {                                                                     \
 195             const iterator_name tmp(*this);                                   \
 196             m_ptr++;                                                          \
 197             return tmp;                                                       \
 198         }                                                                     \
 199                                                                               \
 200         iterator_name& operator--() { m_ptr--; return *this; }                \
 201         iterator_name operator--(int)                                         \
 202         {                                                                     \
 203             const iterator_name tmp(*this);                                   \
 204             m_ptr--;                                                          \
 205             return tmp;                                                       \
 206         }                                                                     \
 207                                                                               \
 208         iterator_name operator+(int n) const                                  \
 209             { return iterator_name(m_ptr + n); }                              \
 210         iterator_name operator+(size_t n) const                               \
 211             { return iterator_name(m_ptr + n); }                              \
 212         iterator_name operator-(int n) const                                  \
 213             { return iterator_name(m_ptr - n); }                              \
 214         iterator_name operator-(size_t n) const                               \
 215             { return iterator_name(m_ptr - n); }                              \
 216         iterator_name& operator+=(int n)                                      \
 217             { m_ptr += n; return *this; }                                     \
 218         iterator_name& operator+=(size_t n)                                   \
 219             { m_ptr += n; return *this; }                                     \
 220         iterator_name& operator-=(int n)                                      \
 221             { m_ptr -= n; return *this; }                                     \
 222         iterator_name& operator-=(size_t n)                                   \
 223             { m_ptr -= n; return *this; }                                     \
 224                                                                               \
 225         size_t operator-(const iterator_name& i) const                        \
 226             { return m_ptr - i.m_ptr; }                                       \
 227                                                                               \
 228         bool operator==(const iterator_name& i) const                         \
 229           { return m_ptr == i.m_ptr; }                                        \
 230         bool operator!=(const iterator_name& i) const                         \
 231           { return m_ptr != i.m_ptr; }                                        \
 232                                                                               \
 233         bool operator<(const iterator_name& i) const                          \
 234           { return m_ptr < i.m_ptr; }                                         \
 235         bool operator>(const iterator_name& i) const                          \
 236           { return m_ptr > i.m_ptr; }                                         \
 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     private:                                                                  \
 243         /* for wxStringImpl use only */                                       \
 244         operator pointer() const { return m_ptr; }                            \
 245                                                                               \
 246         friend class WXDLLIMPEXP_BASE wxStringImpl;                           \
 247                                                                               \
 248         pointer m_ptr
 249
 250   // we need to declare const_iterator in wxStringImpl scope, the friend
 251   // declaration inside iterator class itself is not enough, or at least not
 252   // for g++ 3.4 (g++ 4 is ok)
 253   class const_iterator;
 254
 255   class iterator
 256   {
 257     WX_DEFINE_STRINGIMPL_ITERATOR(iterator,
 258                                   wxStringCharType&,
 259                                   wxStringCharType*);
 260
 261     friend class const_iterator;
 262   };
 263
 264   class const_iterator
 265   {
 266   public:
 267       const_iterator(iterator i) : m_ptr(i.m_ptr) { }
 268
 269       WX_DEFINE_STRINGIMPL_ITERATOR(const_iterator,
 270                                     const wxStringCharType&,
 271                                     const wxStringCharType*);
 272   };
 273
 274
 275   // constructors and destructor
 276     // ctor for an empty string
 277   wxStringImpl() { Init(); }
 278     // copy ctor
 279   wxStringImpl(const wxStringImpl& stringSrc)
 280   {
 281     wxASSERT_MSG( stringSrc.GetStringData()->IsValid(),
 282                   _T("did you forget to call UngetWriteBuf()?") );
 283
 284     if ( stringSrc.empty() ) {
 285       // nothing to do for an empty string
 286       Init();
 287     }
 288     else {
 289       m_pchData = stringSrc.m_pchData;            // share same data
 290       GetStringData()->Lock();                    // => one more copy
 291     }
 292   }
 293     // string containing nRepeat copies of ch
 294   wxStringImpl(size_type nRepeat, wxStringCharType ch);
 295     // ctor takes first nLength characters from C string
 296     // (default value of npos means take all the string)
 297   wxStringImpl(const wxStringCharType *psz)
 298       { InitWith(psz, 0, npos); }
 299   wxStringImpl(const wxStringCharType *psz, size_t nLength)
 300       { InitWith(psz, 0, nLength); }
 301     // take nLen chars starting at nPos
 302   wxStringImpl(const wxStringImpl& str, size_t nPos, size_t nLen)
 303   {
 304     wxASSERT_MSG( str.GetStringData()->IsValid(),
 305                   _T("did you forget to call UngetWriteBuf()?") );
 306     Init();
 307     size_t strLen = str.length() - nPos; nLen = strLen < nLen ? strLen : nLen;
 308     InitWith(str.c_str(), nPos, nLen);
 309   }
 310     // take everything between start and end
 311   wxStringImpl(const_iterator start, const_iterator end);
 312
 313     // dtor is not virtual, this class must not be inherited from!
 314   ~wxStringImpl()
 315   {
 316 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 317       //RN - according to the above VC++ does indeed inline this,
 318       //even though it spits out two warnings
 319       #pragma warning (disable:4714)
 320 #endif
 321
 322       GetStringData()->Unlock();
 323   }
 324
 325 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 326     //re-enable inlining warning
 327     #pragma warning (default:4714)
 328 #endif
 329   // overloaded assignment
 330     // from another wxString
 331   wxStringImpl& operator=(const wxStringImpl& stringSrc);
 332     // from a character
 333   wxStringImpl& operator=(wxStringCharType ch);
 334     // from a C string
 335   wxStringImpl& operator=(const wxStringCharType *psz);
 336
 337     // return the length of the string
 338   size_type length() const { return GetStringData()->nDataLength; }
 339     // return the length of the string
 340   size_type size() const { return length(); }
 341     // return the maximum size of the string
 342   size_type max_size() const { return npos; }
 343     // resize the string, filling the space with c if c != 0
 344   void resize(size_t nSize, wxStringCharType ch = '\0');
 345     // delete the contents of the string
 346   void clear() { erase(0, npos); }
 347     // returns true if the string is empty
 348   bool empty() const { return length() == 0; }
 349     // inform string about planned change in size
 350   void reserve(size_t sz) { Alloc(sz); }
 351   size_type capacity() const { return GetStringData()->nAllocLength; }
 352
 353   // lib.string.access
 354     // return the character at position n
 355   value_type at(size_type n) const
 356     { wxASSERT_VALID_INDEX( n ); return m_pchData[n]; }
 357     // returns the writable character at position n
 358   reference at(size_type n)
 359   {
 360     wxASSERT_VALID_INDEX( n );
 361     CopyBeforeWrite();
 362     return m_pchData[n];
 363   } // FIXME-UTF8: not useful for us...?
 364
 365   // lib.string.modifiers
 366     // append elements str[pos], ..., str[pos+n]
 367   wxStringImpl& append(const wxStringImpl& str, size_t pos, size_t n)
 368   {
 369     wxASSERT(pos <= str.length());
 370     ConcatSelf(n, str.c_str() + pos, str.length() - pos);
 371     return *this;
 372   }
 373     // append a string
 374   wxStringImpl& append(const wxStringImpl& str)
 375     { ConcatSelf(str.length(), str.c_str()); return *this; }
 376     // append first n (or all if n == npos) characters of sz
 377   wxStringImpl& append(const wxStringCharType *sz)
 378     { ConcatSelf(wxStrlen(sz), sz); return *this; }
 379   wxStringImpl& append(const wxStringCharType *sz, size_t n)
 380     { ConcatSelf(n, sz); return *this; }
 381     // append n copies of ch
 382   wxStringImpl& append(size_t n, wxStringCharType ch);
 383     // append from first to last
 384   wxStringImpl& append(const_iterator first, const_iterator last)
 385     { ConcatSelf(last - first, first); return *this; }
 386
 387     // same as `this_string = str'
 388   wxStringImpl& assign(const wxStringImpl& str)
 389     { return *this = str; }
 390     // same as ` = str[pos..pos + n]
 391   wxStringImpl& assign(const wxStringImpl& str, size_t pos, size_t n)
 392     { clear(); return append(str, pos, n); }
 393     // same as `= first n (or all if n == npos) characters of sz'
 394   wxStringImpl& assign(const wxStringCharType *sz)
 395     { clear(); return append(sz, wxStrlen(sz)); }
 396   wxStringImpl& assign(const wxStringCharType *sz, size_t n)
 397     { clear(); return append(sz, n); }
 398     // same as `= n copies of ch'
 399   wxStringImpl& assign(size_t n, wxStringCharType ch)
 400     { clear(); return append(n, ch); }
 401     // assign from first to last
 402   wxStringImpl& assign(const_iterator first, const_iterator last)
 403     { clear(); return append(first, last); }
 404
 405     // first valid index position
 406   const_iterator begin() const { return m_pchData; }
 407   iterator begin();
 408     // position one after the last valid one
 409   const_iterator end() const { return m_pchData + length(); }
 410   iterator end();
 411
 412     // insert another string
 413   wxStringImpl& insert(size_t nPos, const wxStringImpl& str)
 414   {
 415     wxASSERT( str.GetStringData()->IsValid() );
 416     return insert(nPos, str.c_str(), str.length());
 417   }
 418     // insert n chars of str starting at nStart (in str)
 419   wxStringImpl& insert(size_t nPos, const wxStringImpl& str, size_t nStart, size_t n)
 420   {
 421     wxASSERT( str.GetStringData()->IsValid() );
 422     wxASSERT( nStart < str.length() );
 423     size_t strLen = str.length() - nStart;
 424     n = strLen < n ? strLen : n;
 425     return insert(nPos, str.c_str() + nStart, n);
 426   }
 427     // insert first n (or all if n == npos) characters of sz
 428   wxStringImpl& insert(size_t nPos, const wxStringCharType *sz, size_t n = npos);
 429     // insert n copies of ch
 430   wxStringImpl& insert(size_t nPos, size_t n, wxStringCharType ch)// FIXME-UTF8: tricky
 431     { return insert(nPos, wxStringImpl(n, ch)); }
 432   iterator insert(iterator it, wxStringCharType ch) // FIXME-UTF8: tricky
 433     { size_t idx = it - begin(); insert(idx, 1, ch); return begin() + idx; }
 434   void insert(iterator it, const_iterator first, const_iterator last)
 435     { insert(it - begin(), first, last - first); }
 436   void insert(iterator it, size_type n, wxStringCharType ch)
 437     { insert(it - begin(), n, ch); }
 438
 439     // delete characters from nStart to nStart + nLen
 440   wxStringImpl& erase(size_type pos = 0, size_type n = npos);
 441   iterator erase(iterator first, iterator last)
 442   {
 443     size_t idx = first - begin();
 444     erase(idx, last - first);
 445     return begin() + idx;
 446   }
 447   iterator erase(iterator first);
 448
 449   // explicit conversion to C string (use this with printf()!)
 450   const wxStringCharType* c_str() const { return m_pchData; }
 451   const wxStringCharType* data() const { return m_pchData; }
 452
 453     // replaces the substring of length nLen starting at nStart
 454   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringCharType* sz);
 455     // replaces the substring of length nLen starting at nStart
 456   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringImpl& str)
 457     { return replace(nStart, nLen, str.c_str()); }
 458     // replaces the substring with nCount copies of ch
 459   wxStringImpl& replace(size_t nStart, size_t nLen, size_t nCount, wxStringCharType ch);
 460     // replaces a substring with another substring
 461   wxStringImpl& replace(size_t nStart, size_t nLen,
 462                         const wxStringImpl& str, size_t nStart2, size_t nLen2);
 463     // replaces the substring with first nCount chars of sz
 464   wxStringImpl& replace(size_t nStart, size_t nLen,
 465                         const wxStringCharType* sz, size_t nCount);
 466   wxStringImpl& replace(iterator first, iterator last, const_pointer s)
 467     { return replace(first - begin(), last - first, s); }
 468   wxStringImpl& replace(iterator first, iterator last, const_pointer s,
 469                         size_type n)
 470     { return replace(first - begin(), last - first, s, n); }
 471   wxStringImpl& replace(iterator first, iterator last, const wxStringImpl& s)
 472     { return replace(first - begin(), last - first, s); }
 473   wxStringImpl& replace(iterator first, iterator last, size_type n, wxStringCharType c)
 474     { return replace(first - begin(), last - first, n, c); }
 475   wxStringImpl& replace(iterator first, iterator last,
 476                         const_iterator first1, const_iterator last1)
 477     { return replace(first - begin(), last - first, first1, last1 - first1); }
 478
 479     // swap two strings
 480   void swap(wxStringImpl& str);
 481
 482     // All find() functions take the nStart argument which specifies the
 483     // position to start the search on, the default value is 0. All functions
 484     // return npos if there were no match.
 485
 486     // find a substring
 487   size_t find(const wxStringImpl& str, size_t nStart = 0) const;
 488
 489     // find first n characters of sz
 490   size_t find(const wxStringCharType* sz, size_t nStart = 0, size_t n = npos) const;
 491
 492     // find the first occurrence of character ch after nStart
 493   size_t find(wxStringCharType ch, size_t nStart = 0) const;
 494
 495     // rfind() family is exactly like find() but works right to left
 496
 497     // as find, but from the end
 498   size_t rfind(const wxStringImpl& str, size_t nStart = npos) const;
 499
 500     // as find, but from the end
 501   size_t rfind(const wxStringCharType* sz, size_t nStart = npos,
 502                size_t n = npos) const;
 503     // as find, but from the end
 504   size_t rfind(wxStringCharType ch, size_t nStart = npos) const;
 505
 506   size_type copy(wxStringCharType* s, size_type n, size_type pos = 0);
 507
 508   // substring extraction
 509   wxStringImpl substr(size_t nStart = 0, size_t nLen = npos) const;
 510
 511       // string += string
 512   wxStringImpl& operator+=(const wxStringImpl& s) { return append(s); }
 513       // string += C string
 514   wxStringImpl& operator+=(const wxStringCharType *psz) { return append(psz); }
 515       // string += char
 516   wxStringImpl& operator+=(wxStringCharType ch) { return append(1, ch); }
 517
 518 #if !wxUSE_UNICODE_UTF8
 519   // helpers for wxStringBuffer and wxStringBufferLength
 520   wxStringCharType *DoGetWriteBuf(size_t nLen);
 521   void DoUngetWriteBuf();
 522   void DoUngetWriteBuf(size_t nLen);
 523 #endif
 524
 525   friend class WXDLLIMPEXP_BASE wxString;
 526 };
 527
 528 #endif // !wxUSE_STL_BASED_WXSTRING
 529
 530 // don't pollute the library user's name space
 531 #undef wxASSERT_VALID_INDEX
 532
 533 #endif  // _WX_WXSTRINGIMPL_H__