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+(int n) const                                  \
 223             { return iterator_name(m_ptr + n); }                              \
 224         iterator_name operator+(size_t n) const                               \
 225             { return iterator_name(m_ptr + n); }                              \
 226         iterator_name operator-(int n) const                                  \
 227             { return iterator_name(m_ptr - n); }                              \
 228         iterator_name operator-(size_t n) const                               \
 229             { return iterator_name(m_ptr - n); }                              \
 230         iterator_name& operator+=(int n)                                      \
 231             { m_ptr += n; return *this; }                                     \
 232         iterator_name& operator+=(size_t n)                                   \
 233             { m_ptr += n; return *this; }                                     \
 234         iterator_name& operator-=(int n)                                      \
 235             { m_ptr -= n; return *this; }                                     \
 236         iterator_name& operator-=(size_t n)                                   \
 237             { m_ptr -= n; return *this; }                                     \
 238                                                                               \
 239         difference_type 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                                                                               \
 247         bool operator<(const iterator_name& i) const                          \
 248           { return m_ptr < i.m_ptr; }                                         \
 249         bool operator>(const iterator_name& i) const                          \
 250           { return m_ptr > i.m_ptr; }                                         \
 251         bool operator<=(const iterator_name& i) const                         \
 252           { return m_ptr <= i.m_ptr; }                                        \
 253         bool operator>=(const iterator_name& i) const                         \
 254           { return m_ptr >= i.m_ptr; }                                        \
 255                                                                               \
 256     private:                                                                  \
 257         /* for wxStringImpl use only */                                       \
 258         operator pointer() const { return m_ptr; }                            \
 259                                                                               \
 260         friend class wxStringImpl;                                            \
 261                                                                               \
 262         pointer m_ptr
 263
 264   // we need to declare const_iterator in wxStringImpl scope, the friend
 265   // declaration inside iterator class itself is not enough, or at least not
 266   // for g++ 3.4 (g++ 4 is ok)
 267   class WXDLLIMPEXP_FWD_BASE const_iterator;
 268
 269   class WXDLLIMPEXP_BASE iterator
 270   {
 271     WX_DEFINE_STRINGIMPL_ITERATOR(iterator,
 272                                   wxStringCharType&,
 273                                   wxStringCharType*);
 274
 275     friend class const_iterator;
 276   };
 277
 278   class WXDLLIMPEXP_BASE const_iterator
 279   {
 280   public:
 281       const_iterator(iterator i) : m_ptr(i.m_ptr) { }
 282
 283       WX_DEFINE_STRINGIMPL_ITERATOR(const_iterator,
 284                                     const wxStringCharType&,
 285                                     const wxStringCharType*);
 286   };
 287
 288   #undef WX_DEFINE_STRINGIMPL_ITERATOR
 289
 290
 291   // constructors and destructor
 292     // ctor for an empty string
 293   wxStringImpl() { Init(); }
 294     // copy ctor
 295   wxStringImpl(const wxStringImpl& stringSrc)
 296   {
 297     wxASSERT_MSG( stringSrc.GetStringData()->IsValid(),
 298                   _T("did you forget to call UngetWriteBuf()?") );
 299
 300     if ( stringSrc.empty() ) {
 301       // nothing to do for an empty string
 302       Init();
 303     }
 304     else {
 305       m_pchData = stringSrc.m_pchData;            // share same data
 306       GetStringData()->Lock();                    // => one more copy
 307     }
 308   }
 309     // string containing nRepeat copies of ch
 310   wxStringImpl(size_type nRepeat, wxStringCharType ch);
 311     // ctor takes first nLength characters from C string
 312     // (default value of npos means take all the string)
 313   wxStringImpl(const wxStringCharType *psz)
 314       { InitWith(psz, 0, npos); }
 315   wxStringImpl(const wxStringCharType *psz, size_t nLength)
 316       { InitWith(psz, 0, nLength); }
 317     // take nLen chars starting at nPos
 318   wxStringImpl(const wxStringImpl& str, size_t nPos, size_t nLen)
 319   {
 320     wxASSERT_MSG( str.GetStringData()->IsValid(),
 321                   _T("did you forget to call UngetWriteBuf()?") );
 322     Init();
 323     size_t strLen = str.length() - nPos; nLen = strLen < nLen ? strLen : nLen;
 324     InitWith(str.c_str(), nPos, nLen);
 325   }
 326     // take everything between start and end
 327   wxStringImpl(const_iterator start, const_iterator end);
 328
 329
 330     // ctor from and conversion to std::string
 331 #if wxUSE_STD_STRING
 332   wxStringImpl(const wxStdString& impl)
 333       { InitWith(impl.c_str(), 0, impl.length()); }
 334
 335   operator wxStdString() const
 336       { return wxStdString(c_str(), length()); }
 337 #endif
 338
 339
 340     // dtor is not virtual, this class must not be inherited from!
 341   ~wxStringImpl()
 342   {
 343 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 344       //RN - according to the above VC++ does indeed inline this,
 345       //even though it spits out two warnings
 346       #pragma warning (disable:4714)
 347 #endif
 348
 349       GetStringData()->Unlock();
 350   }
 351
 352 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 353     //re-enable inlining warning
 354     #pragma warning (default:4714)
 355 #endif
 356   // overloaded assignment
 357     // from another wxString
 358   wxStringImpl& operator=(const wxStringImpl& stringSrc);
 359     // from a character
 360   wxStringImpl& operator=(wxStringCharType ch);
 361     // from a C string
 362   wxStringImpl& operator=(const wxStringCharType *psz);
 363
 364     // return the length of the string
 365   size_type length() const { return GetStringData()->nDataLength; }
 366     // return the length of the string
 367   size_type size() const { return length(); }
 368     // return the maximum size of the string
 369   size_type max_size() const { return npos; }
 370     // resize the string, filling the space with c if c != 0
 371   void resize(size_t nSize, wxStringCharType ch = '\0');
 372     // delete the contents of the string
 373   void clear() { erase(0, npos); }
 374     // returns true if the string is empty
 375   bool empty() const { return length() == 0; }
 376     // inform string about planned change in size
 377   void reserve(size_t sz) { Alloc(sz); }
 378   size_type capacity() const { return GetStringData()->nAllocLength; }
 379
 380   // lib.string.access
 381     // return the character at position n
 382   value_type at(size_type n) const
 383     { wxASSERT_VALID_INDEX( n ); return m_pchData[n]; }
 384     // returns the writable character at position n
 385   reference at(size_type n)
 386   {
 387     wxASSERT_VALID_INDEX( n );
 388     CopyBeforeWrite();
 389     return m_pchData[n];
 390   } // FIXME-UTF8: not useful for us...?
 391
 392   // lib.string.modifiers
 393     // append elements str[pos], ..., str[pos+n]
 394   wxStringImpl& append(const wxStringImpl& str, size_t pos, size_t n)
 395   {
 396     wxASSERT(pos <= str.length());
 397     ConcatSelf(n, str.c_str() + pos, str.length() - pos);
 398     return *this;
 399   }
 400     // append a string
 401   wxStringImpl& append(const wxStringImpl& str)
 402     { ConcatSelf(str.length(), str.c_str()); return *this; }
 403     // append first n (or all if n == npos) characters of sz
 404   wxStringImpl& append(const wxStringCharType *sz)
 405     { ConcatSelf(wxStrlen(sz), sz); return *this; }
 406   wxStringImpl& append(const wxStringCharType *sz, size_t n)
 407     { ConcatSelf(n, sz); return *this; }
 408     // append n copies of ch
 409   wxStringImpl& append(size_t n, wxStringCharType ch);
 410     // append from first to last
 411   wxStringImpl& append(const_iterator first, const_iterator last)
 412     { ConcatSelf(last - first, first); return *this; }
 413
 414     // same as `this_string = str'
 415   wxStringImpl& assign(const wxStringImpl& str)
 416     { return *this = str; }
 417     // same as ` = str[pos..pos + n]
 418   wxStringImpl& assign(const wxStringImpl& str, size_t pos, size_t n)
 419     { return replace(0, npos, str, pos, n); }
 420     // same as `= first n (or all if n == npos) characters of sz'
 421   wxStringImpl& assign(const wxStringCharType *sz)
 422     { return replace(0, npos, sz, wxStrlen(sz)); }
 423   wxStringImpl& assign(const wxStringCharType *sz, size_t n)
 424     { return replace(0, npos, sz, n); }
 425     // same as `= n copies of ch'
 426   wxStringImpl& assign(size_t n, wxStringCharType ch)
 427     { return replace(0, npos, n, ch); }
 428     // assign from first to last
 429   wxStringImpl& assign(const_iterator first, const_iterator last)
 430     { return replace(begin(), end(), first, last); }
 431
 432     // first valid index position
 433   const_iterator begin() const { return m_pchData; }
 434   iterator begin();
 435     // position one after the last valid one
 436   const_iterator end() const { return m_pchData + length(); }
 437   iterator end();
 438
 439     // insert another string
 440   wxStringImpl& insert(size_t nPos, const wxStringImpl& str)
 441   {
 442     wxASSERT( str.GetStringData()->IsValid() );
 443     return insert(nPos, str.c_str(), str.length());
 444   }
 445     // insert n chars of str starting at nStart (in str)
 446   wxStringImpl& insert(size_t nPos, const wxStringImpl& str, size_t nStart, size_t n)
 447   {
 448     wxASSERT( str.GetStringData()->IsValid() );
 449     wxASSERT( nStart < str.length() );
 450     size_t strLen = str.length() - nStart;
 451     n = strLen < n ? strLen : n;
 452     return insert(nPos, str.c_str() + nStart, n);
 453   }
 454     // insert first n (or all if n == npos) characters of sz
 455   wxStringImpl& insert(size_t nPos, const wxStringCharType *sz, size_t n = npos);
 456     // insert n copies of ch
 457   wxStringImpl& insert(size_t nPos, size_t n, wxStringCharType ch)
 458     { return insert(nPos, wxStringImpl(n, ch)); }
 459   iterator insert(iterator it, wxStringCharType ch)
 460     { size_t idx = it - begin(); insert(idx, 1, ch); return begin() + idx; }
 461   void insert(iterator it, const_iterator first, const_iterator last)
 462     { insert(it - begin(), first, last - first); }
 463   void insert(iterator it, size_type n, wxStringCharType ch)
 464     { insert(it - begin(), n, ch); }
 465
 466     // delete characters from nStart to nStart + nLen
 467   wxStringImpl& erase(size_type pos = 0, size_type n = npos);
 468   iterator erase(iterator first, iterator last)
 469   {
 470     size_t idx = first - begin();
 471     erase(idx, last - first);
 472     return begin() + idx;
 473   }
 474   iterator erase(iterator first);
 475
 476   // explicit conversion to C string (use this with printf()!)
 477   const wxStringCharType* c_str() const { return m_pchData; }
 478   const wxStringCharType* data() const { return m_pchData; }
 479
 480     // replaces the substring of length nLen starting at nStart
 481   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringCharType* sz)
 482     { return replace(nStart, nLen, sz, npos); }
 483     // replaces the substring of length nLen starting at nStart
 484   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringImpl& str)
 485     { return replace(nStart, nLen, str.c_str(), str.length()); }
 486     // replaces the substring with nCount copies of ch
 487   wxStringImpl& replace(size_t nStart, size_t nLen,
 488                         size_t nCount, wxStringCharType ch)
 489     { return replace(nStart, nLen, wxStringImpl(nCount, ch)); }
 490     // replaces a substring with another substring
 491   wxStringImpl& replace(size_t nStart, size_t nLen,
 492                         const wxStringImpl& str, size_t nStart2, size_t nLen2)
 493     { return replace(nStart, nLen, str.substr(nStart2, nLen2)); }
 494     // replaces the substring with first nCount chars of sz
 495   wxStringImpl& replace(size_t nStart, size_t nLen,
 496                         const wxStringCharType* sz, size_t nCount);
 497
 498   wxStringImpl& replace(iterator first, iterator last, const_pointer s)
 499     { return replace(first - begin(), last - first, s); }
 500   wxStringImpl& replace(iterator first, iterator last, const_pointer s,
 501                         size_type n)
 502     { return replace(first - begin(), last - first, s, n); }
 503   wxStringImpl& replace(iterator first, iterator last, const wxStringImpl& s)
 504     { return replace(first - begin(), last - first, s); }
 505   wxStringImpl& replace(iterator first, iterator last, size_type n, wxStringCharType c)
 506     { return replace(first - begin(), last - first, n, c); }
 507   wxStringImpl& replace(iterator first, iterator last,
 508                         const_iterator first1, const_iterator last1)
 509     { return replace(first - begin(), last - first, first1, last1 - first1); }
 510
 511     // swap two strings
 512   void swap(wxStringImpl& str);
 513
 514     // All find() functions take the nStart argument which specifies the
 515     // position to start the search on, the default value is 0. All functions
 516     // return npos if there were no match.
 517
 518     // find a substring
 519   size_t find(const wxStringImpl& str, size_t nStart = 0) const;
 520
 521     // find first n characters of sz
 522   size_t find(const wxStringCharType* sz, size_t nStart = 0, size_t n = npos) const;
 523
 524     // find the first occurrence of character ch after nStart
 525   size_t find(wxStringCharType ch, size_t nStart = 0) const;
 526
 527     // rfind() family is exactly like find() but works right to left
 528
 529     // as find, but from the end
 530   size_t rfind(const wxStringImpl& str, size_t nStart = npos) const;
 531
 532     // as find, but from the end
 533   size_t rfind(const wxStringCharType* sz, size_t nStart = npos,
 534                size_t n = npos) const;
 535     // as find, but from the end
 536   size_t rfind(wxStringCharType ch, size_t nStart = npos) const;
 537
 538   size_type copy(wxStringCharType* s, size_type n, size_type pos = 0);
 539
 540   // substring extraction
 541   wxStringImpl substr(size_t nStart = 0, size_t nLen = npos) const;
 542
 543       // string += string
 544   wxStringImpl& operator+=(const wxStringImpl& s) { return append(s); }
 545       // string += C string
 546   wxStringImpl& operator+=(const wxStringCharType *psz) { return append(psz); }
 547       // string += char
 548   wxStringImpl& operator+=(wxStringCharType ch) { return append(1, ch); }
 549
 550   // helpers for wxStringBuffer and wxStringBufferLength
 551   wxStringCharType *DoGetWriteBuf(size_t nLen);
 552   void DoUngetWriteBuf();
 553   void DoUngetWriteBuf(size_t nLen);
 554
 555   friend class WXDLLIMPEXP_FWD_BASE wxString;
 556 };
 557
 558 #endif // !wxUSE_STL_BASED_WXSTRING
 559
 560 // don't pollute the library user's name space
 561 #undef wxASSERT_VALID_INDEX
 562
 563 #endif  // _WX_WXSTRINGIMPL_H__