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 wxString;
  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) { if (&c != this) 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 wxUSE_UNICODE
 145         if ( (unsigned char)c < 0x80 )
 146             return c;
 147
 148         return FromHi8bit(c);
 149 #else
 150         return c;
 151 #endif
 152     }
 153
 154     static char To8bit(value_type c)
 155     {
 156 #if wxUSE_UNICODE
 157         if ( c < 0x80 )
 158             return c;
 159
 160         return ToHi8bit(c);
 161 #else
 162         return c;
 163 #endif
 164     }
 165
 166     // helpers of the functions above called to deal with non-ASCII chars
 167     static value_type FromHi8bit(char c);
 168     static char ToHi8bit(value_type c);
 169
 170 private:
 171     value_type m_value;
 172 };
 173
 174
 175 // Writeable reference to a character in wxString.
 176 //
 177 // This class can be used in the same way wxChar is used, except that changing
 178 // its value updates the underlying string object.
 179 class WXDLLIMPEXP_BASE wxUniCharRef
 180 {
 181 private:
 182     typedef wxStringImpl::iterator iterator;
 183
 184     // create the reference
 185 #if wxUSE_UNICODE_UTF8
 186     wxUniCharRef(wxString& str, iterator pos) : m_str(str), m_pos(pos) {}
 187 #else
 188     wxUniCharRef(iterator pos) : m_pos(pos) {}
 189 #endif
 190
 191 public:
 192     // NB: we have to make this public, because we don't have wxString
 193     //     declaration available here and so can't declare wxString::iterator
 194     //     as friend; so at least don't use a ctor but a static function
 195     //     that must be used explicitly (this is more than using 'explicit'
 196     //     keyword on ctor!):
 197 #if wxUSE_UNICODE_UTF8
 198     static wxUniCharRef CreateForString(wxString& str, iterator pos)
 199         { return wxUniCharRef(str, pos); }
 200 #else
 201     static wxUniCharRef CreateForString(iterator pos)
 202         { return wxUniCharRef(pos); }
 203 #endif
 204
 205     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 206
 207 #if wxUSE_UNICODE_UTF8
 208     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 209 #endif // wxUSE_UNICODE_UTF8
 210
 211     bool IsAscii() const { return UniChar().IsAscii(); }
 212
 213     // Assignment operators:
 214 #if wxUSE_UNICODE_UTF8
 215     wxUniCharRef& operator=(const wxUniChar& c);
 216 #else
 217     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 218 #endif
 219
 220     wxUniCharRef& operator=(const wxUniCharRef& c)
 221         { if (&c != this) *this = c.UniChar(); return *this; }
 222
 223     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 224     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 225 #if wxWCHAR_T_IS_REAL_TYPE
 226     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 227 #endif
 228     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 229     wxUniCharRef& operator=(unsigned int c) { return *this = wxUniChar(c); }
 230     wxUniCharRef& operator=(short int c) { return *this = wxUniChar(c); }
 231     wxUniCharRef& operator=(unsigned short int c) { return *this = wxUniChar(c); }
 232     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 233     wxUniCharRef& operator=(unsigned long int c) { return *this = wxUniChar(c); }
 234
 235     // Conversions to the same types as wxUniChar is convertible too:
 236     operator char() const { return UniChar(); }
 237     operator unsigned char() const { return UniChar(); }
 238 #if wxWCHAR_T_IS_REAL_TYPE
 239     operator wchar_t() const { return UniChar(); }
 240 #endif
 241     operator int() const { return UniChar(); }
 242     operator unsigned int() const { return UniChar(); }
 243     operator short int() const { return UniChar(); }
 244     operator unsigned short int() const { return UniChar(); }
 245     operator long int() const { return UniChar(); }
 246     operator unsigned long int() const { return UniChar(); }
 247
 248     // see wxUniChar::operator bool etc. for explanation
 249     operator bool() const { return (bool)UniChar(); }
 250     bool operator!() const { return !UniChar(); }
 251     bool operator&&(bool v) const { return UniChar() && v; }
 252
 253     // Comparison operators:
 254 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 255     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 256     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 257     bool operator op(char c) const { return UniChar() op c; }                 \
 258     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 259     wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return UniChar() op c; } ) \
 260     bool operator op(int c) const { return UniChar() op c; }                  \
 261     bool operator op(unsigned int c) const { return UniChar() op c; }         \
 262     bool operator op(short int c) const { return UniChar() op c; }             \
 263     bool operator op(unsigned short int c) const { return UniChar() op c; }    \
 264     bool operator op(long int c) const { return UniChar() op c; }             \
 265     bool operator op(unsigned long int c) const { return UniChar() op c; }
 266
 267     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 268
 269 #undef wxDEFINE_UNICHARREF_OPERATOR
 270
 271     // for expressions like c-'A':
 272     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 273     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 274     int operator-(char c) const { return UniChar() - c; }
 275     int operator-(unsigned char c) const { return UniChar() - c; }
 276     int operator-(wchar_t c) const { return UniChar() - c; }
 277
 278 private:
 279 #if wxUSE_UNICODE_UTF8
 280     wxUniChar UniChar() const;
 281 #else
 282     wxUniChar UniChar() const { return *m_pos; }
 283 #endif
 284
 285     friend class WXDLLIMPEXP_FWD_BASE wxUniChar;
 286
 287 private:
 288     // reference to the string and pointer to the character in string
 289 #if wxUSE_UNICODE_UTF8
 290     wxString& m_str;
 291 #endif
 292     iterator m_pos;
 293 };
 294
 295 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 296 {
 297     m_value = c.UniChar().m_value;
 298 }
 299
 300 inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c)
 301 {
 302     m_value = c.UniChar().m_value;
 303     return *this;
 304 }
 305
 306 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 307 // implemented in terms of member comparison functions
 308
 309 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 310
 311 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 312 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 313
 314 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 315 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 316
 317 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 318
 319 #undef wxCMP_REVERSE
 320
 321 // for expressions like c-'A':
 322 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 323 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 324 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 325
 326 #endif /* _WX_UNICHAR_H_ */