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 class WXDLLIMPEXP_FWD_BASE wxUniCharRef;
  19 class WXDLLIMPEXP_FWD_BASE wxStringIteratorNode;
  20
  21 // This class represents single Unicode character. It can be converted to
  22 // and from char or wchar_t and implements commonly used character operations.
  23 class WXDLLIMPEXP_BASE wxUniChar
  24 {
  25 public:
  26     // NB: this is not wchar_t on purpose, it needs to represent the entire
  27     //     Unicode code points range and wchar_t may be too small for that
  28     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  29     typedef wxUint32 value_type;
  30
  31     wxUniChar() : m_value(0) {}
  32
  33     // Create the character from 8bit character value encoded in the current
  34     // locale's charset.
  35     wxUniChar(char c) { m_value = From8bit(c); }
  36     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  37
  38     // Create the character from a wchar_t character value.
  39 #if wxWCHAR_T_IS_REAL_TYPE
  40     wxUniChar(wchar_t c) { m_value = c; }
  41 #endif
  42
  43     wxUniChar(int c) { m_value = c; }
  44     wxUniChar(unsigned int c) { m_value = c; }
  45     wxUniChar(long int c) { m_value = c; }
  46     wxUniChar(unsigned long int c) { m_value = c; }
  47     wxUniChar(short int c) { m_value = c; }
  48     wxUniChar(unsigned short int c) { m_value = c; }
  49
  50     wxUniChar(const wxUniCharRef& c);
  51
  52     // Returns Unicode code point value of the character
  53     value_type GetValue() const { return m_value; }
  54
  55 #if wxUSE_UNICODE_UTF8
  56     // buffer for single UTF-8 character
  57     struct Utf8CharBuffer
  58     {
  59         char data[5];
  60         operator const char*() const { return data; }
  61     };
  62
  63     // returns the character encoded as UTF-8
  64     // (NB: implemented in stringops.cpp)
  65     Utf8CharBuffer AsUTF8() const;
  66 #endif // wxUSE_UNICODE_UTF8
  67
  68     // Returns true if the character is an ASCII character:
  69     bool IsAscii() const { return m_value < 0x80; }
  70
  71     // Conversions to char and wchar_t types: all of those are needed to be
  72     // able to pass wxUniChars to verious standard narrow and wide character
  73     // functions
  74     operator char() const { return To8bit(m_value); }
  75     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  76 #if wxWCHAR_T_IS_REAL_TYPE
  77     operator wchar_t() const { return (wchar_t)m_value; }
  78 #endif
  79     operator int() const { return (int)m_value; }
  80     operator unsigned int() const { return (unsigned int)m_value; }
  81     operator long int() const { return (long int)m_value; }
  82     operator unsigned long int() const { return (unsigned long)m_value; }
  83     operator short int() const { return (short int)m_value; }
  84     operator unsigned short int() const { return (unsigned short int)m_value; }
  85
  86     // We need this operator for the "*p" part of expressions like "for (
  87     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  88     // compilation would fail without it because the conversion to bool would
  89     // be ambiguous (there are all these int types conversions...). (And adding
  90     // operator unspecified_bool_type() would only makes the ambiguity worse.)
  91     operator bool() const { return m_value != 0; }
  92     bool operator!() const { return !((bool)*this); }
  93
  94     // And this one is needed by some (not all, but not using ifdefs makes the
  95     // code easier) compilers to parse "str[0] && *p" successfully
  96     bool operator&&(bool v) const { return (bool)*this && v; }
  97
  98     // Assignment operators:
  99     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 100     wxUniChar& operator=(const wxUniCharRef& c);
 101     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 102     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 103 #if wxWCHAR_T_IS_REAL_TYPE
 104     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 105 #endif
 106     wxUniChar& operator=(int c) { m_value = c; return *this; }
 107     wxUniChar& operator=(unsigned int c) { m_value = c; return *this; }
 108     wxUniChar& operator=(long int c) { m_value = c; return *this; }
 109     wxUniChar& operator=(unsigned long int c) { m_value = c; return *this; }
 110     wxUniChar& operator=(short int c) { m_value = c; return *this; }
 111     wxUniChar& operator=(unsigned short int c) { m_value = c; return *this; }
 112
 113     // Comparison operators:
 114
 115     // define the given comparison operator for all the types
 116 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 117     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 118     bool operator op(char c) const { return m_value op From8bit(c); }         \
 119     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 120     wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return m_value op (value_type)c; } )    \
 121     bool operator op(int c) const { return m_value op (value_type)c; }        \
 122     bool operator op(unsigned int c) const { return m_value op (value_type)c; }        \
 123     bool operator op(short int c) const { return m_value op (value_type)c; }  \
 124     bool operator op(unsigned short int c) const { return m_value op (value_type)c; }  \
 125     bool operator op(long int c) const { return m_value op (value_type)c; }   \
 126     bool operator op(unsigned long int c) const { return m_value op (value_type)c; }
 127
 128     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 129
 130 #undef wxDEFINE_UNICHAR_OPERATOR
 131
 132     // this is needed for expressions like 'Z'-c
 133     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 134     int operator-(char c) const { return m_value - From8bit(c); }
 135     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 136     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 137
 138
 139 private:
 140     // notice that we implement these functions inline for 7-bit ASCII
 141     // characters purely for performance reasons
 142     static value_type From8bit(char c)
 143     {
 144         if ( (unsigned char)c < 0x80 )
 145             return c;
 146
 147         return FromHi8bit(c);
 148     }
 149
 150     static char To8bit(value_type c)
 151     {
 152         if ( c < 0x80 )
 153             return c;
 154
 155         return ToHi8bit(c);
 156     }
 157
 158     // helpers of the functions above called to deal with non-ASCII chars
 159     static value_type FromHi8bit(char c);
 160     static char ToHi8bit(value_type c);
 161
 162 private:
 163     value_type m_value;
 164 };
 165
 166
 167 // Writeable reference to a character in wxString.
 168 //
 169 // This class can be used in the same way wxChar is used, except that changing
 170 // its value updates the underlying string object.
 171 class WXDLLIMPEXP_BASE wxUniCharRef
 172 {
 173 private:
 174     typedef wxStringImpl::iterator iterator;
 175
 176     // create the reference
 177 #if wxUSE_UNICODE_UTF8
 178     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 179 #else
 180     wxUniCharRef(iterator pos) : m_pos(pos) {}
 181 #endif
 182
 183 public:
 184     // NB: we have to make this public, because we don't have wxString
 185     //     declaration available here and so can't declare wxString::iterator
 186     //     as friend; so at least don't use a ctor but a static function
 187     //     that must be used explicitly (this is more than using 'explicit'
 188     //     keyword on ctor!):
 189 #if wxUSE_UNICODE_UTF8
 190     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 191         { return wxUniCharRef(node, pos); }
 192 #else
 193     static wxUniCharRef CreateForString(iterator pos)
 194         { return wxUniCharRef(pos); }
 195 #endif
 196
 197     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 198
 199 #if wxUSE_UNICODE_UTF8
 200     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 201 #endif // wxUSE_UNICODE_UTF8
 202
 203     bool IsAscii() const { return UniChar().IsAscii(); }
 204
 205     // Assignment operators:
 206 #if wxUSE_UNICODE_UTF8
 207     wxUniCharRef& operator=(const wxUniChar& c);
 208 #else
 209     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 210 #endif
 211
 212     wxUniCharRef& operator=(const wxUniCharRef& c)
 213         { return *this = c.UniChar(); }
 214
 215     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 216     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 217 #if wxWCHAR_T_IS_REAL_TYPE
 218     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 219 #endif
 220     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 221     wxUniCharRef& operator=(unsigned int c) { return *this = wxUniChar(c); }
 222     wxUniCharRef& operator=(short int c) { return *this = wxUniChar(c); }
 223     wxUniCharRef& operator=(unsigned short int c) { return *this = wxUniChar(c); }
 224     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 225     wxUniCharRef& operator=(unsigned long int c) { return *this = wxUniChar(c); }
 226
 227     // Conversions to the same types as wxUniChar is convertible too:
 228     operator char() const { return UniChar(); }
 229     operator unsigned char() const { return UniChar(); }
 230 #if wxWCHAR_T_IS_REAL_TYPE
 231     operator wchar_t() const { return UniChar(); }
 232 #endif
 233     operator int() const { return UniChar(); }
 234     operator unsigned int() const { return UniChar(); }
 235     operator short int() const { return UniChar(); }
 236     operator unsigned short int() const { return UniChar(); }
 237     operator long int() const { return UniChar(); }
 238     operator unsigned long int() const { return UniChar(); }
 239
 240     // see wxUniChar::operator bool etc. for explanation
 241     operator bool() const { return (bool)UniChar(); }
 242     bool operator!() const { return !UniChar(); }
 243     bool operator&&(bool v) const { return UniChar() && v; }
 244
 245     // Comparison operators:
 246 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 247     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 248     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 249     bool operator op(char c) const { return UniChar() op c; }                 \
 250     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 251     wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return UniChar() op c; } ) \
 252     bool operator op(int c) const { return UniChar() op c; }                  \
 253     bool operator op(unsigned int c) const { return UniChar() op c; }         \
 254     bool operator op(short int c) const { return UniChar() op c; }             \
 255     bool operator op(unsigned short int c) const { return UniChar() op c; }    \
 256     bool operator op(long int c) const { return UniChar() op c; }             \
 257     bool operator op(unsigned long int c) const { return UniChar() op c; }
 258
 259     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 260
 261 #undef wxDEFINE_UNICHARREF_OPERATOR
 262
 263     // for expressions like c-'A':
 264     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 265     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 266     int operator-(char c) const { return UniChar() - c; }
 267     int operator-(unsigned char c) const { return UniChar() - c; }
 268     int operator-(wchar_t c) const { return UniChar() - c; }
 269
 270 private:
 271 #if wxUSE_UNICODE_UTF8
 272     wxUniChar UniChar() const;
 273 #else
 274     wxUniChar UniChar() const { return *m_pos; }
 275 #endif
 276
 277     friend class WXDLLIMPEXP_FWD_BASE wxUniChar;
 278
 279 private:
 280     // reference to the string and pointer to the character in string
 281 #if wxUSE_UNICODE_UTF8
 282     wxStringIteratorNode& m_node;
 283 #endif
 284     iterator m_pos;
 285 };
 286
 287 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 288 {
 289     m_value = c.UniChar().m_value;
 290 }
 291
 292 inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c)
 293 {
 294     m_value = c.UniChar().m_value;
 295     return *this;
 296 }
 297
 298 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 299 // implemented in terms of member comparison functions
 300
 301 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 302
 303 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 304 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 305
 306 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 307 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 308
 309 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 310
 311 #undef wxCMP_REVERSE
 312
 313 // for expressions like c-'A':
 314 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 315 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 316 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 317
 318 #endif /* _WX_UNICHAR_H_ */