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__) && \
  22         !defined(__DARWIN__) && !defined(__OS2__) && !defined(__DOS__)) || \
  23     (defined(__VISUALC__) && defined(_NATIVE_WCHAR_T_DEFINED))
  24     #define wxWINT_T_IS_SEPARATE_TYPE
  25 #endif
  26
  27 // helper macro for doing something dependent on whether wint_t is or isn't a
  28 // typedef inside another macro
  29 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  30     #define wxIF_WINT_T_TYPE(x) x
  31 #else // !wxWINT_T_IS_SEPARATE_TYPE
  32     #define wxIF_WINT_T_TYPE(x)
  33 #endif // wxWINT_T_IS_SEPARATE_TYPE/!wxWINT_T_IS_SEPARATE_TYPE
  34
  35 // wchar_t seems to be defined as unsigned short by all Windows compilers but
  36 // unsigned int everywhere else
  37 #ifndef __WIN32__
  38     #define wxWCHAR_T_IS_UINT
  39 #endif
  40
  41 class WXDLLIMPEXP_BASE wxUniCharRef;
  42
  43 // This class represents single Unicode character. It can be converted to
  44 // and from char or wchar_t and implements commonly used character operations.
  45 class WXDLLIMPEXP_BASE wxUniChar
  46 {
  47 public:
  48     // NB: this is not wchar_t on purpose, it needs to represent the entire
  49     //     Unicode code points range and wchar_t may be too small for that
  50     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  51     typedef wxUint32 value_type;
  52
  53     wxUniChar() : m_value(0) {}
  54
  55     // Create the character from 8bit character value encoded in the current
  56     // locale's charset.
  57     wxUniChar(char c) { m_value = From8bit(c); }
  58     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  59
  60     // Create the character from a wchar_t character value.
  61     wxUniChar(wchar_t c) { m_value = c; }
  62 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  63     wxUniChar(wint_t c) { m_value = c; }
  64 #endif
  65
  66     wxUniChar(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     // Returns true if the character is an ASCII character:
  74     bool IsAscii() const { return m_value < 0x80; }
  75
  76     // Conversions to char and wchar_t types: all of those are needed to be
  77     // able to pass wxUniChars to verious standard narrow and wide character
  78     // functions
  79     operator char() const { return To8bit(m_value); }
  80     operator wchar_t() const { return m_value; }
  81     operator int() const { return m_value; }
  82 #ifndef wxWCHAR_T_IS_UINT
  83     operator unsigned int() const { return m_value; }
  84 #endif
  85
  86     // More conversions needed for other standard functions: uchar is for VC++
  87     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
  88     // and some wide character functions take wint_t which happens to be the
  89     // same as wchar_t for Windows compilers but not for g++ (except for the
  90     // special Apple version)
  91     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  92 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  93     operator wint_t() const { return m_value; }
  94 #endif
  95
  96     // We need this operator for the "*p" part of expressions like "for (
  97     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  98     // compilation would fail without it because the conversion to bool would
  99     // be ambiguous (there are all these int types conversions...). (And adding
 100     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 101     operator bool() const { return m_value != 0; }
 102     bool operator!() const { return !((bool)*this); }
 103 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 104     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 105     // We need this for VC++ < 8 or DigitalMars and expressions like
 106     // "str[0] && *p":
 107     bool operator&&(bool v) const { return (bool)*this && v; }
 108 #endif
 109
 110     // Assignment operators:
 111     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 112     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 113     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 114     wxUniChar& operator=(int c) { m_value = c; return *this; }
 115 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 116     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 117 #endif
 118
 119     // Comparison operators:
 120
 121     // define the given comparison operator for all the types
 122 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 123     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 124     bool operator op(char c) const { return m_value op From8bit(c); }         \
 125     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 126     wxIF_WINT_T_TYPE( bool operator op(wint_t 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-(wchar_t c) const { return m_value - (value_type)c; }
 136 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 137     int operator-(wint_t c) const { return m_value - (value_type)c; }
 138 #endif
 139
 140 private:
 141     static value_type From8bit(char c);
 142     static char To8bit(value_type c);
 143
 144 private:
 145     value_type m_value;
 146 };
 147
 148
 149 // Writeable reference to a character in wxString.
 150 //
 151 // This class can be used in the same way wxChar is used, except that changing
 152 // its value updates the underlying string object.
 153 class WXDLLIMPEXP_BASE wxUniCharRef
 154 {
 155 private:
 156     typedef wxStringImpl::iterator iterator;
 157
 158     // create the reference
 159 #if wxUSE_UNICODE_UTF8
 160     wxUniCharRef(wxStringImpl& str, iterator pos) : m_str(str), m_pos(pos) {}
 161 #else
 162     wxUniCharRef(iterator pos) : m_pos(pos) {}
 163 #endif
 164
 165 public:
 166     // NB: we have to make this public, because we don't have wxString
 167     //     declaration available here and so can't declare wxString::iterator
 168     //     as friend; so at least don't use a ctor but a static function
 169     //     that must be used explicitly (this is more than using 'explicit'
 170     //     keyword on ctor!):
 171 #if wxUSE_UNICODE_UTF8
 172     static wxUniCharRef CreateForString(wxStringImpl& str, iterator pos)
 173         { return wxUniCharRef(str, pos); }
 174 #else
 175     static wxUniCharRef CreateForString(iterator pos)
 176         { return wxUniCharRef(pos); }
 177 #endif
 178
 179     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 180     bool IsAscii() const { return UniChar().IsAscii(); }
 181
 182     // Assignment operators:
 183 #if wxUSE_UNICODE_UTF8
 184     wxUniCharRef& operator=(const wxUniChar& c);
 185 #else
 186     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 187 #endif
 188
 189     wxUniCharRef& operator=(const wxUniCharRef& c)
 190         { return *this = c.UniChar(); }
 191
 192     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 193     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 194     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 195 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 196     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 197 #endif
 198
 199     // Conversions to the same types as wxUniChar is convertible too:
 200     operator char() const { return UniChar(); }
 201     operator int() const { return UniChar(); }
 202     operator unsigned char() const { return UniChar(); }
 203     operator wchar_t() const { return UniChar(); }
 204 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 205     operator wint_t() const { return UniChar(); }
 206 #endif
 207 #ifndef wxWCHAR_T_IS_UINT
 208     operator unsigned int() const { return UniChar(); }
 209 #endif
 210
 211     // see wxUniChar::operator bool etc. for explanation
 212     operator bool() const { return (bool)UniChar(); }
 213     bool operator!() const { return !UniChar(); }
 214 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 215     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 216     bool operator&&(bool v) const { return UniChar() && v; }
 217 #endif
 218
 219     // Comparison operators:
 220 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 221     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 222     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 223     bool operator op(char c) const { return UniChar() op c; }                 \
 224     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 225     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 226
 227     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 228
 229 #undef wxDEFINE_UNICHARREF_OPERATOR
 230
 231     // for expressions like c-'A':
 232     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 233     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 234     int operator-(char c) const { return UniChar() - c; }
 235     int operator-(wchar_t c) const { return UniChar() - c; }
 236 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 237     int operator-(wint_t c) const { return UniChar() - c; }
 238 #endif
 239
 240 private:
 241 #if wxUSE_UNICODE_UTF8
 242     wxUniChar UniChar() const;
 243 #else
 244     wxUniChar UniChar() const { return *m_pos; }
 245 #endif
 246
 247     friend class WXDLLIMPEXP_BASE wxUniChar;
 248
 249 private:
 250     // reference to the string and pointer to the character in string
 251 #if wxUSE_UNICODE_UTF8
 252     wxStringImpl& m_str;
 253 #endif
 254     iterator m_pos;
 255 };
 256
 257 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 258 {
 259     m_value = c.UniChar().m_value;
 260 }
 261
 262 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 263 // implemented in terms of member comparison functions
 264
 265 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 266
 267 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 268 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 269
 270 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 271 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 272
 273 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 274 //wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 275 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 276 #endif
 277
 278 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 279
 280 #undef wxCMP_REVERSE
 281
 282 // for expressions like c-'A':
 283 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 284 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 285 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 286 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 287 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 288 #endif
 289
 290 #endif /* _WX_UNICHAR_H_ */