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
  17 class WXDLLIMPEXP_BASE wxUniCharRef;
  18
  19 // This class represents single Unicode character. It can be converted to
  20 // and from char or wchar_t and implements commonly used character operations.
  21 class WXDLLIMPEXP_BASE wxUniChar
  22 {
  23 public:
  24     // NB: this is not wchar_t on purpose, it needs to represent the entire
  25     //     Unicode code points range and wchar_t may be too small for that
  26     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  27     typedef wxUint32 value_type;
  28
  29     wxUniChar() : m_value(0) {}
  30
  31     // Create the character from 8bit character value encoded in the current
  32     // locale's charset.
  33     wxUniChar(char c) { m_value = From8bit(c); }
  34     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  35
  36     // Create the character from a wchar_t character value.
  37     wxUniChar(wchar_t c) { m_value = c; }
  38
  39     wxUniChar(int c) { m_value = c; }
  40
  41     wxUniChar(const wxUniCharRef& c);
  42
  43     // Returns Unicode code point value of the character
  44     value_type GetValue() const { return m_value; }
  45
  46     // Conversions to char and wchar_t types: all of those are needed to be
  47     // able to pass wxUniChars to verious standard narrow and wide character
  48     // functions
  49     operator char() const { return To8bit(m_value); }
  50     operator wchar_t() const { return m_value; }
  51     operator int() const { return m_value; }
  52
  53     // More conversions needed for other standard functions: uchar is for VC++
  54     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
  55     // and some wide character functions take wint_t which happens to be the
  56     // same as wchar_t for Windows compilers but not for g++ (except for the
  57     // special Apple version)
  58     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  59 #if defined(__GNUC__) && !defined(__DARWIN__)
  60     operator wint_t() const { return m_value; }
  61 #endif
  62
  63     // We need this operator for the "*p" part of expressions like "for (
  64     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  65     // compilation would fail without it because the conversion to bool would
  66     // be ambiguous (there are all these int types conversions...). (And adding
  67     // operator unspecified_bool_type() would only makes the ambiguity worse.)
  68     operator bool() const { return m_value != 0; }
  69     bool operator!() const { return !((bool)*this); }
  70 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
  71     defined(__DIGITALMARS__) || defined(__BORLANDC__)
  72     // We need this for VC++ < 8 or DigitalMars and expressions like
  73     // "str[0] && *p":
  74     bool operator&&(bool v) const { return (bool)*this && v; }
  75 #endif
  76
  77     // Assignment operators:
  78     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
  79     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
  80     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
  81
  82     // Comparison operators:
  83
  84     // define the given comparison operator for all the types
  85 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
  86     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
  87     bool operator op(char c) const { return m_value op From8bit(c); }         \
  88     bool operator op(wchar_t c) const { return m_value op (value_type)c; }
  89
  90     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
  91
  92 #undef wxDEFINE_UNICHAR_OPERATOR
  93
  94     // this is needed for expressions like 'Z'-c
  95     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
  96     int operator-(char c) const { return m_value - From8bit(c); }
  97     int operator-(wchar_t c) const { return m_value - (value_type)c; }
  98
  99 private:
 100     static value_type From8bit(char c);
 101     static char To8bit(value_type c);
 102
 103 private:
 104     value_type m_value;
 105 };
 106
 107
 108 // Writeable reference to a character in wxString.
 109 //
 110 // This class can be used in the same way wxChar is used, except that changing
 111 // its value updates the underlying string object.
 112 class WXDLLIMPEXP_BASE wxUniCharRef
 113 {
 114 private:
 115     // create the reference
 116     // FIXME-UTF8: the interface will need changes for UTF-8 build
 117     wxUniCharRef(wxChar *pos) : m_pos(pos) {}
 118
 119 public:
 120     // NB: we have to make this public, because we don't have wxString
 121     //     declaration available here and so can't declare wxString::iterator
 122     //     as friend; so at least don't use a ctor but a static function
 123     //     that must be used explicitly (this is more than using 'explicit'
 124     //     keyword on ctor!):
 125     //
 126     // FIXME-UTF8: the interface will need changes for UTF-8 build
 127     static wxUniCharRef CreateForString(wxChar *pos)
 128         { return wxUniCharRef(pos); }
 129
 130     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 131
 132     // Assignment operators:
 133     wxUniCharRef& operator=(const wxUniCharRef& c)
 134     {
 135         *m_pos = *c.m_pos;
 136         return *this;
 137     };
 138
 139     wxUniCharRef& operator=(const wxUniChar& c)
 140     {
 141         *m_pos = c;
 142         return *this;
 143     };
 144
 145     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 146     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 147
 148     // Conversions to the same types as wxUniChar is convertible too:
 149     operator char() const { return UniChar(); }
 150     operator wchar_t() const { return UniChar(); }
 151     operator int() const { return UniChar(); }
 152     operator unsigned char() const { return UniChar(); }
 153 #if defined(__GNUC__) && !defined(__DARWIN__)
 154     operator wint_t() const { return UniChar(); }
 155 #endif
 156
 157     // see wxUniChar::operator bool etc. for explanation
 158     operator bool() const { return (bool)UniChar(); }
 159     bool operator!() const { return !UniChar(); }
 160 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 161     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 162     bool operator&&(bool v) const { return UniChar() && v; }
 163 #endif
 164
 165     // Comparison operators:
 166 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 167     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 168     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 169     bool operator op(char c) const { return UniChar() op c; }                 \
 170     bool operator op(wchar_t c) const { return UniChar() op c; }
 171
 172     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 173
 174 #undef wxDEFINE_UNICHARREF_OPERATOR
 175
 176     // for expressions like c-'A':
 177     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 178     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 179     int operator-(char c) const { return UniChar() - c; }
 180     int operator-(wchar_t c) const { return UniChar() - c; }
 181
 182 private:
 183     wxUniChar UniChar() const { return *m_pos; }
 184     friend class WXDLLIMPEXP_BASE wxUniChar;
 185
 186 private:
 187     // pointer to the character in string
 188     wxChar *m_pos;
 189 };
 190
 191 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 192 {
 193     m_value = c.UniChar().m_value;
 194 }
 195
 196 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 197 // implemented in terms of member comparison functions
 198
 199 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 200
 201 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 202 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 203
 204 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 205 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 206
 207 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 208
 209 #undef wxCMP_REVERSE
 210
 211 // for expressions like c-'A':
 212 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 213 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 214 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 215
 216 #endif /* _WX_UNICHAR_H_ */