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 #ifndef wxWCHAR_T_IS_SEPARATE_TYPE
  19     // older versions of VC++ have wchar_t as typedef by default; this is
  20     // configurable, so we have to check which behaviour is enabled
  21     #if defined(__VISUALC__) && !defined(_NATIVE_WCHAR_T_DEFINED)
  22         #define wxWCHAR_T_IS_SEPARATE_TYPE 0
  23     #else
  24         #define wxWCHAR_T_IS_SEPARATE_TYPE 1
  25     #endif
  26 #endif
  27
  28 // helper macro for doing something dependent on whether wchar_t is or isn't a
  29 // typedef inside another macro
  30 #if wxWCHAR_T_IS_SEPARATE_TYPE
  31     #define wxIF_WCHAR_T_TYPE(x) x
  32 #else // !wxWCHAR_T_IS_SEPARATE_TYPE
  33     #define wxIF_WCHAR_T_TYPE(x)
  34 #endif // wxWCHAR_T_IS_SEPARATE_TYPE/!wxWCHAR_T_IS_SEPARATE_TYPE
  35
  36 class WXDLLIMPEXP_BASE wxUniCharRef;
  37 class WXDLLIMPEXP_BASE wxStringIteratorNode;
  38
  39 // This class represents single Unicode character. It can be converted to
  40 // and from char or wchar_t and implements commonly used character operations.
  41 class WXDLLIMPEXP_BASE wxUniChar
  42 {
  43 public:
  44     // NB: this is not wchar_t on purpose, it needs to represent the entire
  45     //     Unicode code points range and wchar_t may be too small for that
  46     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  47     typedef wxUint32 value_type;
  48
  49     wxUniChar() : m_value(0) {}
  50
  51     // Create the character from 8bit character value encoded in the current
  52     // locale's charset.
  53     wxUniChar(char c) { m_value = From8bit(c); }
  54     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  55
  56     // Create the character from a wchar_t character value.
  57 #if wxWCHAR_T_IS_SEPARATE_TYPE
  58     wxUniChar(wchar_t c) { m_value = c; }
  59 #endif
  60
  61     wxUniChar(int c) { m_value = c; }
  62     wxUniChar(unsigned int c) { m_value = c; }
  63     wxUniChar(long int c) { m_value = c; }
  64     wxUniChar(unsigned long int c) { m_value = c; }
  65     wxUniChar(short int c) { m_value = c; }
  66     wxUniChar(unsigned short int c) { m_value = c; }
  67
  68     wxUniChar(const wxUniCharRef& c);
  69
  70     // Returns Unicode code point value of the character
  71     value_type GetValue() const { return m_value; }
  72
  73 #if wxUSE_UNICODE_UTF8
  74     // buffer for single UTF-8 character
  75     struct Utf8CharBuffer
  76     {
  77         char data[5];
  78         operator const char*() const { return data; }
  79     };
  80
  81     // returns the character encoded as UTF-8
  82     // (NB: implemented in stringops.cpp)
  83     Utf8CharBuffer AsUTF8() const;
  84 #endif // wxUSE_UNICODE_UTF8
  85
  86     // Returns true if the character is an ASCII character:
  87     bool IsAscii() const { return m_value < 0x80; }
  88
  89     // Conversions to char and wchar_t types: all of those are needed to be
  90     // able to pass wxUniChars to verious standard narrow and wide character
  91     // functions
  92     operator char() const { return To8bit(m_value); }
  93     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  94 #if wxWCHAR_T_IS_SEPARATE_TYPE
  95     operator wchar_t() const { return (wchar_t)m_value; }
  96 #endif
  97     operator int() const { return (int)m_value; }
  98     operator unsigned int() const { return (unsigned int)m_value; }
  99     operator long int() const { return (long int)m_value; }
 100     operator unsigned long int() const { return (unsigned long)m_value; }
 101     operator short int() const { return (short int)m_value; }
 102     operator unsigned short int() const { return (unsigned short int)m_value; }
 103
 104     // We need this operator for the "*p" part of expressions like "for (
 105     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
 106     // compilation would fail without it because the conversion to bool would
 107     // be ambiguous (there are all these int types conversions...). (And adding
 108     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 109     operator bool() const { return m_value != 0; }
 110     bool operator!() const { return !((bool)*this); }
 111
 112     // And this one is needed by some (not all, but not using ifdefs makes the
 113     // code easier) compilers to parse "str[0] && *p" successfully
 114     bool operator&&(bool v) const { return (bool)*this && v; }
 115
 116     // Assignment operators:
 117     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 118     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 119     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 120 #if wxWCHAR_T_IS_SEPARATE_TYPE
 121     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 122 #endif
 123     wxUniChar& operator=(int c) { m_value = c; return *this; }
 124     wxUniChar& operator=(unsigned int c) { m_value = c; return *this; }
 125     wxUniChar& operator=(long int c) { m_value = c; return *this; }
 126     wxUniChar& operator=(unsigned long int c) { m_value = c; return *this; }
 127     wxUniChar& operator=(short int c) { m_value = c; return *this; }
 128     wxUniChar& operator=(unsigned short int c) { m_value = c; return *this; }
 129
 130     // Comparison operators:
 131
 132     // define the given comparison operator for all the types
 133 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 134     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 135     bool operator op(char c) const { return m_value op From8bit(c); }         \
 136     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 137     wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return m_value op (value_type)c; } )    \
 138     bool operator op(int c) const { return m_value op (value_type)c; }        \
 139     bool operator op(unsigned int c) const { return m_value op (value_type)c; }        \
 140     bool operator op(short int c) const { return m_value op (value_type)c; }  \
 141     bool operator op(unsigned short int c) const { return m_value op (value_type)c; }  \
 142     bool operator op(long int c) const { return m_value op (value_type)c; }   \
 143     bool operator op(unsigned long int c) const { return m_value op (value_type)c; }
 144
 145     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 146
 147 #undef wxDEFINE_UNICHAR_OPERATOR
 148
 149     // this is needed for expressions like 'Z'-c
 150     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 151     int operator-(char c) const { return m_value - From8bit(c); }
 152     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 153     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 154
 155
 156 private:
 157     static value_type From8bit(char c);
 158     static char To8bit(value_type c);
 159
 160 private:
 161     value_type m_value;
 162 };
 163
 164
 165 // Writeable reference to a character in wxString.
 166 //
 167 // This class can be used in the same way wxChar is used, except that changing
 168 // its value updates the underlying string object.
 169 class WXDLLIMPEXP_BASE wxUniCharRef
 170 {
 171 private:
 172     typedef wxStringImpl::iterator iterator;
 173
 174     // create the reference
 175 #if wxUSE_UNICODE_UTF8
 176     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 177 #else
 178     wxUniCharRef(iterator pos) : m_pos(pos) {}
 179 #endif
 180
 181 public:
 182     // NB: we have to make this public, because we don't have wxString
 183     //     declaration available here and so can't declare wxString::iterator
 184     //     as friend; so at least don't use a ctor but a static function
 185     //     that must be used explicitly (this is more than using 'explicit'
 186     //     keyword on ctor!):
 187 #if wxUSE_UNICODE_UTF8
 188     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 189         { return wxUniCharRef(node, pos); }
 190 #else
 191     static wxUniCharRef CreateForString(iterator pos)
 192         { return wxUniCharRef(pos); }
 193 #endif
 194
 195     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 196
 197 #if wxUSE_UNICODE_UTF8
 198     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 199 #endif // wxUSE_UNICODE_UTF8
 200
 201     bool IsAscii() const { return UniChar().IsAscii(); }
 202
 203     // Assignment operators:
 204 #if wxUSE_UNICODE_UTF8
 205     wxUniCharRef& operator=(const wxUniChar& c);
 206 #else
 207     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 208 #endif
 209
 210     wxUniCharRef& operator=(const wxUniCharRef& c)
 211         { return *this = c.UniChar(); }
 212
 213     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 214     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 215 #if wxWCHAR_T_IS_SEPARATE_TYPE
 216     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 217 #endif
 218     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 219     wxUniCharRef& operator=(unsigned int c) { return *this = wxUniChar(c); }
 220     wxUniCharRef& operator=(short int c) { return *this = wxUniChar(c); }
 221     wxUniCharRef& operator=(unsigned short int c) { return *this = wxUniChar(c); }
 222     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 223     wxUniCharRef& operator=(unsigned long int c) { return *this = wxUniChar(c); }
 224
 225     // Conversions to the same types as wxUniChar is convertible too:
 226     operator char() const { return UniChar(); }
 227     operator unsigned char() const { return UniChar(); }
 228 #if wxWCHAR_T_IS_SEPARATE_TYPE
 229     operator wchar_t() const { return UniChar(); }
 230 #endif
 231     operator int() const { return UniChar(); }
 232     operator unsigned int() const { return UniChar(); }
 233     operator short int() const { return UniChar(); }
 234     operator unsigned short int() const { return UniChar(); }
 235     operator long int() const { return UniChar(); }
 236     operator unsigned long int() const { return UniChar(); }
 237
 238     // see wxUniChar::operator bool etc. for explanation
 239     operator bool() const { return (bool)UniChar(); }
 240     bool operator!() const { return !UniChar(); }
 241     bool operator&&(bool v) const { return UniChar() && v; }
 242
 243     // Comparison operators:
 244 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 245     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 246     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 247     bool operator op(char c) const { return UniChar() op c; }                 \
 248     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 249     wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return UniChar() op c; } ) \
 250     bool operator op(int c) const { return UniChar() op c; }                  \
 251     bool operator op(unsigned int c) const { return UniChar() op c; }         \
 252     bool operator op(short int c) const { return UniChar() op c; }             \
 253     bool operator op(unsigned short int c) const { return UniChar() op c; }    \
 254     bool operator op(long int c) const { return UniChar() op c; }             \
 255     bool operator op(unsigned long int c) const { return UniChar() op c; }
 256
 257     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 258
 259 #undef wxDEFINE_UNICHARREF_OPERATOR
 260
 261     // for expressions like c-'A':
 262     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 263     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 264     int operator-(char c) const { return UniChar() - c; }
 265     int operator-(unsigned char c) const { return UniChar() - c; }
 266     int operator-(wchar_t c) const { return UniChar() - c; }
 267
 268 private:
 269 #if wxUSE_UNICODE_UTF8
 270     wxUniChar UniChar() const;
 271 #else
 272     wxUniChar UniChar() const { return *m_pos; }
 273 #endif
 274
 275     friend class WXDLLIMPEXP_BASE wxUniChar;
 276
 277 private:
 278     // reference to the string and pointer to the character in string
 279 #if wxUSE_UNICODE_UTF8
 280     wxStringIteratorNode& m_node;
 281 #endif
 282     iterator m_pos;
 283 };
 284
 285 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 286 {
 287     m_value = c.UniChar().m_value;
 288 }
 289
 290 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 291 // implemented in terms of member comparison functions
 292
 293 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 294
 295 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 296 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 297
 298 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 299 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 300
 301 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 302
 303 #undef wxCMP_REVERSE
 304
 305 // for expressions like c-'A':
 306 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 307 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 308 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 309
 310 #endif /* _WX_UNICHAR_H_ */