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