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