include/wx/unichar.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        wx/unichar.h
   3 // Purpose:     wxUniChar and wxUniCharRef classes
   4 // Author:      Vaclav Slavik
   5 // Created:     2007-03-19
   6 // RCS-ID:      $Id$
   7 // Copyright:   (c) 2007 REA Elektronik GmbH
   8 // Licence:     wxWindows licence
   9 ///////////////////////////////////////////////////////////////////////////////
  10
  11 #ifndef _WX_UNICHAR_H_
  12 #define _WX_UNICHAR_H_
  13
  14 #include "wx/defs.h"
  15 #include "wx/chartype.h"
  16 #include "wx/stringimpl.h"
  17
  18 // wint_t is just a typedef for wchar_t for many old compilers but for modern
  19 // ones it's a separate type and we must provide a conversion to it to allow
  20 // passing wxUniChar[Ref] to functions taking wint_t such as iswalnum() &c
  21 #if (defined(__GNUC__) && !defined(__DARWIN__) && !defined(__OS2__)) || \
  22     (defined(__VISUALC__) && defined(_NATIVE_WCHAR_T_DEFINED))
  23     #define wxWINT_T_IS_SEPARATE_TYPE
  24 #endif
  25
  26 // helper macro for doing something dependent on whether wint_t is or isn't a
  27 // typedef inside another macro
  28 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  29     #define wxIF_WINT_T_TYPE(x) x
  30 #else // !wxWINT_T_IS_SEPARATE_TYPE
  31     #define wxIF_WINT_T_TYPE(x)
  32 #endif // wxWINT_T_IS_SEPARATE_TYPE/!wxWINT_T_IS_SEPARATE_TYPE
  33
  34 // wchar_t seems to be defined as unsigned short by all Windows compilers but
  35 // unsigned int everywhere else
  36 #ifndef __WIN32__
  37     #define wxWCHAR_T_IS_UINT
  38 #endif
  39
  40 class WXDLLIMPEXP_BASE wxUniCharRef;
  41
  42 // This class represents single Unicode character. It can be converted to
  43 // and from char or wchar_t and implements commonly used character operations.
  44 class WXDLLIMPEXP_BASE wxUniChar
  45 {
  46 public:
  47     // NB: this is not wchar_t on purpose, it needs to represent the entire
  48     //     Unicode code points range and wchar_t may be too small for that
  49     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  50     typedef wxUint32 value_type;
  51
  52     wxUniChar() : m_value(0) {}
  53
  54     // Create the character from 8bit character value encoded in the current
  55     // locale's charset.
  56     wxUniChar(char c) { m_value = From8bit(c); }
  57     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  58
  59     // Create the character from a wchar_t character value.
  60     wxUniChar(wchar_t c) { m_value = c; }
  61 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  62     wxUniChar(wint_t c) { m_value = c; }
  63 #endif
  64
  65     wxUniChar(int c) { m_value = c; }
  66
  67     wxUniChar(const wxUniCharRef& c);
  68
  69     // Returns Unicode code point value of the character
  70     value_type GetValue() const { return m_value; }
  71
  72     // Conversions to char and wchar_t types: all of those are needed to be
  73     // able to pass wxUniChars to verious standard narrow and wide character
  74     // functions
  75     operator char() const { return To8bit(m_value); }
  76     operator wchar_t() const { return m_value; }
  77     operator int() const { return m_value; }
  78 #ifndef wxWCHAR_T_IS_UINT
  79     operator unsigned int() const { return m_value; }
  80 #endif
  81
  82     // More conversions needed for other standard functions: uchar is for VC++
  83     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
  84     // and some wide character functions take wint_t which happens to be the
  85     // same as wchar_t for Windows compilers but not for g++ (except for the
  86     // special Apple version)
  87     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  88 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  89     operator wint_t() const { return m_value; }
  90 #endif
  91
  92     // We need this operator for the "*p" part of expressions like "for (
  93     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  94     // compilation would fail without it because the conversion to bool would
  95     // be ambiguous (there are all these int types conversions...). (And adding
  96     // operator unspecified_bool_type() would only makes the ambiguity worse.)
  97     operator bool() const { return m_value != 0; }
  98     bool operator!() const { return !((bool)*this); }
  99 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 100     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 101     // We need this for VC++ < 8 or DigitalMars and expressions like
 102     // "str[0] && *p":
 103     bool operator&&(bool v) const { return (bool)*this && v; }
 104 #endif
 105
 106     // Assignment operators:
 107     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 108     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 109     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 110     wxUniChar& operator=(int c) { m_value = c; return *this; }
 111 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 112     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 113 #endif
 114
 115     // Comparison operators:
 116
 117     // define the given comparison operator for all the types
 118 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 119     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 120     bool operator op(char c) const { return m_value op From8bit(c); }         \
 121     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 122     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 123
 124     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 125
 126 #undef wxDEFINE_UNICHAR_OPERATOR
 127
 128     // this is needed for expressions like 'Z'-c
 129     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 130     int operator-(char c) const { return m_value - From8bit(c); }
 131     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 132 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 133     int operator-(wint_t c) const { return m_value - (value_type)c; }
 134 #endif
 135
 136 private:
 137     static value_type From8bit(char c);
 138     static char To8bit(value_type c);
 139
 140 private:
 141     value_type m_value;
 142 };
 143
 144
 145 // Writeable reference to a character in wxString.
 146 //
 147 // This class can be used in the same way wxChar is used, except that changing
 148 // its value updates the underlying string object.
 149 class WXDLLIMPEXP_BASE wxUniCharRef
 150 {
 151 private:
 152     typedef wxStringImpl::iterator iterator;
 153
 154     // create the reference
 155     wxUniCharRef(iterator pos) : m_pos(pos) {}
 156
 157 public:
 158     // NB: we have to make this public, because we don't have wxString
 159     //     declaration available here and so can't declare wxString::iterator
 160     //     as friend; so at least don't use a ctor but a static function
 161     //     that must be used explicitly (this is more than using 'explicit'
 162     //     keyword on ctor!):
 163     static wxUniCharRef CreateForString(iterator pos)
 164         { return wxUniCharRef(pos); }
 165
 166     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 167
 168     // Assignment operators:
 169     wxUniCharRef& operator=(const wxUniCharRef& c)
 170     {
 171         *m_pos = *c.m_pos;
 172         return *this;
 173     };
 174
 175     wxUniCharRef& operator=(const wxUniChar& c)
 176     {
 177         *m_pos = c;
 178         return *this;
 179     };
 180
 181     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 182     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 183     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 184 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 185     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 186 #endif
 187
 188     // Conversions to the same types as wxUniChar is convertible too:
 189     operator char() const { return UniChar(); }
 190     operator int() const { return UniChar(); }
 191     operator unsigned char() const { return UniChar(); }
 192     operator wchar_t() const { return UniChar(); }
 193 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 194     operator wint_t() const { return UniChar(); }
 195 #endif
 196 #ifndef wxWCHAR_T_IS_UINT
 197     operator unsigned int() const { return UniChar(); }
 198 #endif
 199
 200     // see wxUniChar::operator bool etc. for explanation
 201     operator bool() const { return (bool)UniChar(); }
 202     bool operator!() const { return !UniChar(); }
 203 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 204     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 205     bool operator&&(bool v) const { return UniChar() && v; }
 206 #endif
 207
 208     // Comparison operators:
 209 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 210     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 211     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 212     bool operator op(char c) const { return UniChar() op c; }                 \
 213     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 214     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 215
 216     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 217
 218 #undef wxDEFINE_UNICHARREF_OPERATOR
 219
 220     // for expressions like c-'A':
 221     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 222     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 223     int operator-(char c) const { return UniChar() - c; }
 224     int operator-(wchar_t c) const { return UniChar() - c; }
 225 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 226     int operator-(wint_t c) const { return UniChar() - c; }
 227 #endif
 228
 229 private:
 230     wxUniChar UniChar() const { return *m_pos; }
 231     friend class WXDLLIMPEXP_BASE wxUniChar;
 232
 233 private:
 234     // pointer to the character in string
 235     iterator m_pos;
 236 };
 237
 238 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 239 {
 240     m_value = c.UniChar().m_value;
 241 }
 242
 243 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 244 // implemented in terms of member comparison functions
 245
 246 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 247
 248 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 249 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 250
 251 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 252 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 253
 254 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 255 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 256 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 257 #endif
 258
 259 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 260
 261 #undef wxCMP_REVERSE
 262
 263 // for expressions like c-'A':
 264 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 265 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 266 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 267 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 268 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 269 #endif
 270
 271 #endif /* _WX_UNICHAR_H_ */