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 class WXDLLIMPEXP_BASE wxStringIteratorNode;
  43
  44 // This class represents single Unicode character. It can be converted to
  45 // and from char or wchar_t and implements commonly used character operations.
  46 class WXDLLIMPEXP_BASE wxUniChar
  47 {
  48 public:
  49     // NB: this is not wchar_t on purpose, it needs to represent the entire
  50     //     Unicode code points range and wchar_t may be too small for that
  51     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  52     typedef wxUint32 value_type;
  53
  54     wxUniChar() : m_value(0) {}
  55
  56     // Create the character from 8bit character value encoded in the current
  57     // locale's charset.
  58     wxUniChar(char c) { m_value = From8bit(c); }
  59     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  60
  61     // Create the character from a wchar_t character value.
  62     wxUniChar(wchar_t c) { m_value = c; }
  63 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  64     wxUniChar(wint_t c) { m_value = c; }
  65 #endif
  66
  67     wxUniChar(int c) { m_value = c; }
  68
  69     wxUniChar(const wxUniCharRef& c);
  70
  71     // Returns Unicode code point value of the character
  72     value_type GetValue() const { return m_value; }
  73
  74     // Returns true if the character is an ASCII character:
  75     bool IsAscii() const { return m_value < 0x80; }
  76
  77     // Conversions to char and wchar_t types: all of those are needed to be
  78     // able to pass wxUniChars to verious standard narrow and wide character
  79     // functions
  80     operator char() const { return To8bit(m_value); }
  81     operator wchar_t() const { return m_value; }
  82     operator int() const { return m_value; }
  83 #ifndef wxWCHAR_T_IS_UINT
  84     operator unsigned int() const { return m_value; }
  85 #endif
  86
  87     // More conversions needed for other standard functions: uchar is for VC++
  88     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
  89     // and some wide character functions take wint_t which happens to be the
  90     // same as wchar_t for Windows compilers but not for g++ (except for the
  91     // special Apple version)
  92     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  93 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  94     operator wint_t() const { return m_value; }
  95 #endif
  96
  97     // We need this operator for the "*p" part of expressions like "for (
  98     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  99     // compilation would fail without it because the conversion to bool would
 100     // be ambiguous (there are all these int types conversions...). (And adding
 101     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 102     operator bool() const { return m_value != 0; }
 103     bool operator!() const { return !((bool)*this); }
 104 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 105     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 106     // We need this for VC++ < 8 or DigitalMars and expressions like
 107     // "str[0] && *p":
 108     bool operator&&(bool v) const { return (bool)*this && v; }
 109 #endif
 110
 111     // Assignment operators:
 112     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 113     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 114     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 115     wxUniChar& operator=(int c) { m_value = c; return *this; }
 116 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 117     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 118 #endif
 119
 120     // Comparison operators:
 121
 122     // define the given comparison operator for all the types
 123 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 124     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 125     bool operator op(char c) const { return m_value op From8bit(c); }         \
 126     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 127     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 128
 129     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 130
 131 #undef wxDEFINE_UNICHAR_OPERATOR
 132
 133     // this is needed for expressions like 'Z'-c
 134     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 135     int operator-(char c) const { return m_value - From8bit(c); }
 136     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 137 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 138     int operator-(wint_t c) const { return m_value - (value_type)c; }
 139 #endif
 140
 141 private:
 142     static value_type From8bit(char c);
 143     static char To8bit(value_type c);
 144
 145 private:
 146     value_type m_value;
 147 };
 148
 149
 150 // Writeable reference to a character in wxString.
 151 //
 152 // This class can be used in the same way wxChar is used, except that changing
 153 // its value updates the underlying string object.
 154 class WXDLLIMPEXP_BASE wxUniCharRef
 155 {
 156 private:
 157     typedef wxStringImpl::iterator iterator;
 158
 159     // create the reference
 160 #if wxUSE_UNICODE_UTF8
 161     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 162 #else
 163     wxUniCharRef(iterator pos) : m_pos(pos) {}
 164 #endif
 165
 166 public:
 167     // NB: we have to make this public, because we don't have wxString
 168     //     declaration available here and so can't declare wxString::iterator
 169     //     as friend; so at least don't use a ctor but a static function
 170     //     that must be used explicitly (this is more than using 'explicit'
 171     //     keyword on ctor!):
 172 #if wxUSE_UNICODE_UTF8
 173     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 174         { return wxUniCharRef(node, pos); }
 175 #else
 176     static wxUniCharRef CreateForString(iterator pos)
 177         { return wxUniCharRef(pos); }
 178 #endif
 179
 180     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 181     bool IsAscii() const { return UniChar().IsAscii(); }
 182
 183     // Assignment operators:
 184 #if wxUSE_UNICODE_UTF8
 185     wxUniCharRef& operator=(const wxUniChar& c);
 186 #else
 187     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 188 #endif
 189
 190     wxUniCharRef& operator=(const wxUniCharRef& c)
 191         { return *this = c.UniChar(); }
 192
 193     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 194     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 195     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 196 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 197     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 198 #endif
 199
 200     // Conversions to the same types as wxUniChar is convertible too:
 201     operator char() const { return UniChar(); }
 202     operator int() const { return UniChar(); }
 203     operator unsigned char() const { return UniChar(); }
 204     operator wchar_t() const { return UniChar(); }
 205 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 206     operator wint_t() const { return UniChar(); }
 207 #endif
 208 #ifndef wxWCHAR_T_IS_UINT
 209     operator unsigned int() const { return UniChar(); }
 210 #endif
 211
 212     // see wxUniChar::operator bool etc. for explanation
 213     operator bool() const { return (bool)UniChar(); }
 214     bool operator!() const { return !UniChar(); }
 215 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 216     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 217     bool operator&&(bool v) const { return UniChar() && v; }
 218 #endif
 219
 220     // Comparison operators:
 221 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 222     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 223     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 224     bool operator op(char c) const { return UniChar() op c; }                 \
 225     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 226     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 227
 228     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 229
 230 #undef wxDEFINE_UNICHARREF_OPERATOR
 231
 232     // for expressions like c-'A':
 233     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 234     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 235     int operator-(char c) const { return UniChar() - c; }
 236     int operator-(wchar_t c) const { return UniChar() - c; }
 237 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 238     int operator-(wint_t c) const { return UniChar() - c; }
 239 #endif
 240
 241 private:
 242 #if wxUSE_UNICODE_UTF8
 243     wxUniChar UniChar() const;
 244 #else
 245     wxUniChar UniChar() const { return *m_pos; }
 246 #endif
 247
 248     friend class WXDLLIMPEXP_BASE wxUniChar;
 249
 250 private:
 251     // reference to the string and pointer to the character in string
 252 #if wxUSE_UNICODE_UTF8
 253     wxStringIteratorNode& m_node;
 254 #endif
 255     iterator m_pos;
 256 };
 257
 258 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 259 {
 260     m_value = c.UniChar().m_value;
 261 }
 262
 263 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 264 // implemented in terms of member comparison functions
 265
 266 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 267
 268 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 269 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 270
 271 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 272 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 273
 274 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 275 //wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 276 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 277 #endif
 278
 279 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 280
 281 #undef wxCMP_REVERSE
 282
 283 // for expressions like c-'A':
 284 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 285 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 286 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 287 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 288 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 289 #endif
 290
 291 #endif /* _WX_UNICHAR_H_ */