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=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 115     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 116     wxUniChar& operator=(int c) { m_value = c; return *this; }
 117 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 118     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 119 #endif
 120
 121     // Comparison operators:
 122
 123     // define the given comparison operator for all the types
 124 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 125     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 126     bool operator op(char c) const { return m_value op From8bit(c); }         \
 127     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 128     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 129     bool operator op(int c) const { return m_value op (value_type)c; }        \
 130     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 131
 132     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 133
 134 #undef wxDEFINE_UNICHAR_OPERATOR
 135
 136     // this is needed for expressions like 'Z'-c
 137     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 138     int operator-(char c) const { return m_value - From8bit(c); }
 139     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 140     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 141 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 142     int operator-(wint_t c) const { return m_value - (value_type)c; }
 143 #endif
 144
 145 private:
 146     static value_type From8bit(char c);
 147     static char To8bit(value_type c);
 148
 149 private:
 150     value_type m_value;
 151 };
 152
 153
 154 // Writeable reference to a character in wxString.
 155 //
 156 // This class can be used in the same way wxChar is used, except that changing
 157 // its value updates the underlying string object.
 158 class WXDLLIMPEXP_BASE wxUniCharRef
 159 {
 160 private:
 161     typedef wxStringImpl::iterator iterator;
 162
 163     // create the reference
 164 #if wxUSE_UNICODE_UTF8
 165     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 166 #else
 167     wxUniCharRef(iterator pos) : m_pos(pos) {}
 168 #endif
 169
 170 public:
 171     // NB: we have to make this public, because we don't have wxString
 172     //     declaration available here and so can't declare wxString::iterator
 173     //     as friend; so at least don't use a ctor but a static function
 174     //     that must be used explicitly (this is more than using 'explicit'
 175     //     keyword on ctor!):
 176 #if wxUSE_UNICODE_UTF8
 177     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 178         { return wxUniCharRef(node, pos); }
 179 #else
 180     static wxUniCharRef CreateForString(iterator pos)
 181         { return wxUniCharRef(pos); }
 182 #endif
 183
 184     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 185     bool IsAscii() const { return UniChar().IsAscii(); }
 186
 187     // Assignment operators:
 188 #if wxUSE_UNICODE_UTF8
 189     wxUniCharRef& operator=(const wxUniChar& c);
 190 #else
 191     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 192 #endif
 193
 194     wxUniCharRef& operator=(const wxUniCharRef& c)
 195         { return *this = c.UniChar(); }
 196
 197     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 198     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 199     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 200     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 201 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 202     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 203 #endif
 204
 205     // Conversions to the same types as wxUniChar is convertible too:
 206     operator char() const { return UniChar(); }
 207     operator int() const { return UniChar(); }
 208     operator unsigned char() const { return UniChar(); }
 209     operator wchar_t() const { return UniChar(); }
 210 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 211     operator wint_t() const { return UniChar(); }
 212 #endif
 213 #ifndef wxWCHAR_T_IS_UINT
 214     operator unsigned int() const { return UniChar(); }
 215 #endif
 216
 217     // see wxUniChar::operator bool etc. for explanation
 218     operator bool() const { return (bool)UniChar(); }
 219     bool operator!() const { return !UniChar(); }
 220 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 221     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 222     bool operator&&(bool v) const { return UniChar() && v; }
 223 #endif
 224
 225     // Comparison operators:
 226 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 227     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 228     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 229     bool operator op(char c) const { return UniChar() op c; }                 \
 230     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 231     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 232     bool operator op(int c) const { return UniChar() op c; }                  \
 233     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 234
 235     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 236
 237 #undef wxDEFINE_UNICHARREF_OPERATOR
 238
 239     // for expressions like c-'A':
 240     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 241     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 242     int operator-(char c) const { return UniChar() - c; }
 243     int operator-(unsigned char c) const { return UniChar() - c; }
 244     int operator-(wchar_t c) const { return UniChar() - c; }
 245 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 246     int operator-(wint_t c) const { return UniChar() - c; }
 247 #endif
 248
 249 private:
 250 #if wxUSE_UNICODE_UTF8
 251     wxUniChar UniChar() const;
 252 #else
 253     wxUniChar UniChar() const { return *m_pos; }
 254 #endif
 255
 256     friend class WXDLLIMPEXP_BASE wxUniChar;
 257
 258 private:
 259     // reference to the string and pointer to the character in string
 260 #if wxUSE_UNICODE_UTF8
 261     wxStringIteratorNode& m_node;
 262 #endif
 263     iterator m_pos;
 264 };
 265
 266 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 267 {
 268     m_value = c.UniChar().m_value;
 269 }
 270
 271 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 272 // implemented in terms of member comparison functions
 273
 274 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 275
 276 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 277 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 278
 279 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 280 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 281
 282 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 283 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 284 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 285 #endif
 286
 287 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 288
 289 #undef wxCMP_REVERSE
 290
 291 // for expressions like c-'A':
 292 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 293 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 294 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 295 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 296 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 297 #endif
 298
 299 #endif /* _WX_UNICHAR_H_ */