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 // wchar_t seems to be defined as unsigned short by all Windows compilers but
  35 // unsigned int everywhere else
  36 #ifndef __WIN32__
  37     #define wxWCHAR_T_IS_UINT
  38 #endif
  39
  40 class WXDLLIMPEXP_BASE wxUniCharRef;
  41
  42 // This class represents single Unicode character. It can be converted to
  43 // and from char or wchar_t and implements commonly used character operations.
  44 class WXDLLIMPEXP_BASE wxUniChar
  45 {
  46 public:
  47     // NB: this is not wchar_t on purpose, it needs to represent the entire
  48     //     Unicode code points range and wchar_t may be too small for that
  49     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  50     typedef wxUint32 value_type;
  51
  52     wxUniChar() : m_value(0) {}
  53
  54     // Create the character from 8bit character value encoded in the current
  55     // locale's charset.
  56     wxUniChar(char c) { m_value = From8bit(c); }
  57     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  58
  59     // Create the character from a wchar_t character value.
  60     wxUniChar(wchar_t c) { m_value = c; }
  61 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  62     wxUniChar(wint_t c) { m_value = c; }
  63 #endif
  64
  65     wxUniChar(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     // Returns true if the character is an ASCII character:
  73     bool IsAscii() const { return m_value < 0x80; }
  74
  75     // Conversions to char and wchar_t types: all of those are needed to be
  76     // able to pass wxUniChars to verious standard narrow and wide character
  77     // functions
  78     operator char() const { return To8bit(m_value); }
  79     operator wchar_t() const { return m_value; }
  80     operator int() const { return m_value; }
  81 #ifndef wxWCHAR_T_IS_UINT
  82     operator unsigned int() const { return m_value; }
  83 #endif
  84
  85     // More conversions needed for other standard functions: uchar is for VC++
  86     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
  87     // and some wide character functions take wint_t which happens to be the
  88     // same as wchar_t for Windows compilers but not for g++ (except for the
  89     // special Apple version)
  90     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  91 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  92     operator wint_t() const { return m_value; }
  93 #endif
  94
  95     // We need this operator for the "*p" part of expressions like "for (
  96     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  97     // compilation would fail without it because the conversion to bool would
  98     // be ambiguous (there are all these int types conversions...). (And adding
  99     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 100     operator bool() const { return m_value != 0; }
 101     bool operator!() const { return !((bool)*this); }
 102 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 103     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 104     // We need this for VC++ < 8 or DigitalMars and expressions like
 105     // "str[0] && *p":
 106     bool operator&&(bool v) const { return (bool)*this && v; }
 107 #endif
 108
 109     // Assignment operators:
 110     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 111     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 112     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 113     wxUniChar& operator=(int c) { m_value = c; return *this; }
 114 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 115     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 116 #endif
 117
 118     // Comparison operators:
 119
 120     // define the given comparison operator for all the types
 121 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 122     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 123     bool operator op(char c) const { return m_value op From8bit(c); }         \
 124     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 125     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 126
 127     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 128
 129 #undef wxDEFINE_UNICHAR_OPERATOR
 130
 131     // this is needed for expressions like 'Z'-c
 132     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 133     int operator-(char c) const { return m_value - From8bit(c); }
 134     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 135 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 136     int operator-(wint_t c) const { return m_value - (value_type)c; }
 137 #endif
 138
 139 private:
 140     static value_type From8bit(char c);
 141     static char To8bit(value_type c);
 142
 143 private:
 144     value_type m_value;
 145 };
 146
 147
 148 // Writeable reference to a character in wxString.
 149 //
 150 // This class can be used in the same way wxChar is used, except that changing
 151 // its value updates the underlying string object.
 152 class WXDLLIMPEXP_BASE wxUniCharRef
 153 {
 154 private:
 155     typedef wxStringImpl::iterator iterator;
 156
 157     // create the reference
 158 #if wxUSE_UNICODE_UTF8
 159     wxUniCharRef(wxStringImpl& str, iterator pos) : m_str(str), m_pos(pos) {}
 160 #else
 161     wxUniCharRef(iterator pos) : m_pos(pos) {}
 162 #endif
 163
 164 public:
 165     // NB: we have to make this public, because we don't have wxString
 166     //     declaration available here and so can't declare wxString::iterator
 167     //     as friend; so at least don't use a ctor but a static function
 168     //     that must be used explicitly (this is more than using 'explicit'
 169     //     keyword on ctor!):
 170 #if wxUSE_UNICODE_UTF8
 171     static wxUniCharRef CreateForString(wxStringImpl& str, iterator pos)
 172         { return wxUniCharRef(str, pos); }
 173 #else
 174     static wxUniCharRef CreateForString(iterator pos)
 175         { return wxUniCharRef(pos); }
 176 #endif
 177
 178     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 179     bool IsAscii() const { return UniChar().IsAscii(); }
 180
 181     // Assignment operators:
 182 #if wxUSE_UNICODE_UTF8
 183     wxUniCharRef& operator=(const wxUniChar& c);
 184 #else
 185     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 186 #endif
 187
 188     wxUniCharRef& operator=(const wxUniCharRef& c)
 189         { return *this = c.UniChar(); }
 190
 191     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 192     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 193     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 194 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 195     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 196 #endif
 197
 198     // Conversions to the same types as wxUniChar is convertible too:
 199     operator char() const { return UniChar(); }
 200     operator int() const { return UniChar(); }
 201     operator unsigned char() const { return UniChar(); }
 202     operator wchar_t() const { return UniChar(); }
 203 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 204     operator wint_t() const { return UniChar(); }
 205 #endif
 206 #ifndef wxWCHAR_T_IS_UINT
 207     operator unsigned int() const { return UniChar(); }
 208 #endif
 209
 210     // see wxUniChar::operator bool etc. for explanation
 211     operator bool() const { return (bool)UniChar(); }
 212     bool operator!() const { return !UniChar(); }
 213 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 214     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 215     bool operator&&(bool v) const { return UniChar() && v; }
 216 #endif
 217
 218     // Comparison operators:
 219 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 220     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 221     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 222     bool operator op(char c) const { return UniChar() op c; }                 \
 223     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 224     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 225
 226     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 227
 228 #undef wxDEFINE_UNICHARREF_OPERATOR
 229
 230     // for expressions like c-'A':
 231     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 232     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 233     int operator-(char c) const { return UniChar() - c; }
 234     int operator-(wchar_t c) const { return UniChar() - c; }
 235 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 236     int operator-(wint_t c) const { return UniChar() - c; }
 237 #endif
 238
 239 private:
 240     wxUniChar UniChar() const
 241     {
 242 #if wxUSE_UNICODE_UTF8
 243         return DecodeChar(m_pos);
 244 #else
 245         return *m_pos;
 246 #endif
 247     }
 248
 249 #if wxUSE_UNICODE_UTF8
 250     // FIXME-UTF8: move this to a separate 'string operations' class
 251     static wxUniChar DecodeChar(wxStringImpl::const_iterator i);
 252     friend class WXDLLIMPEXP_BASE wxString;
 253 #endif
 254
 255     friend class WXDLLIMPEXP_BASE wxUniChar;
 256
 257 private:
 258     // reference to the string and pointer to the character in string
 259 #if wxUSE_UNICODE_UTF8
 260     wxStringImpl& m_str;
 261 #endif
 262     iterator m_pos;
 263 };
 264
 265 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 266 {
 267     m_value = c.UniChar().m_value;
 268 }
 269
 270 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 271 // implemented in terms of member comparison functions
 272
 273 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 274
 275 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 276 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 277
 278 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 279 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 280
 281 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 282 //wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 283 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 284 #endif
 285
 286 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 287
 288 #undef wxCMP_REVERSE
 289
 290 // for expressions like c-'A':
 291 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 292 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 293 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 294 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 295 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 296 #endif
 297
 298 #endif /* _WX_UNICHAR_H_ */