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 #if wxUSE_UNICODE_UTF8
  75     // buffer for single UTF-8 character
  76     struct Utf8CharBuffer
  77     {
  78         char data[5];
  79         operator const char*() const { return data; }
  80     };
  81
  82     // returns the character encoded as UTF-8
  83     // (NB: implemented in stringops.cpp)
  84     Utf8CharBuffer AsUTF8() const;
  85 #endif // wxUSE_UNICODE_UTF8
  86
  87     // Returns true if the character is an ASCII character:
  88     bool IsAscii() const { return m_value < 0x80; }
  89
  90     // Conversions to char and wchar_t types: all of those are needed to be
  91     // able to pass wxUniChars to verious standard narrow and wide character
  92     // functions
  93     operator char() const { return To8bit(m_value); }
  94     operator wchar_t() const { return m_value; }
  95     operator int() const { return m_value; }
  96 #ifndef wxWCHAR_T_IS_UINT
  97     operator unsigned int() const { return m_value; }
  98 #endif
  99
 100     // More conversions needed for other standard functions: uchar is for VC++
 101     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
 102     // and some wide character functions take wint_t which happens to be the
 103     // same as wchar_t for Windows compilers but not for g++ (except for the
 104     // special Apple version)
 105     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
 106 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 107     operator wint_t() const { return m_value; }
 108 #endif
 109
 110     // We need this operator for the "*p" part of expressions like "for (
 111     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
 112     // compilation would fail without it because the conversion to bool would
 113     // be ambiguous (there are all these int types conversions...). (And adding
 114     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 115     operator bool() const { return m_value != 0; }
 116     bool operator!() const { return !((bool)*this); }
 117 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 118     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 119     // We need this for VC++ < 8 or DigitalMars and expressions like
 120     // "str[0] && *p":
 121     bool operator&&(bool v) const { return (bool)*this && v; }
 122 #endif
 123
 124     // Assignment operators:
 125     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 126     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 127     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 128     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 129     wxUniChar& operator=(int c) { m_value = c; return *this; }
 130 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 131     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 132 #endif
 133
 134     // Comparison operators:
 135
 136     // define the given comparison operator for all the types
 137 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 138     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 139     bool operator op(char c) const { return m_value op From8bit(c); }         \
 140     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 141     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 142     bool operator op(int c) const { return m_value op (value_type)c; }        \
 143     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 144
 145     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 146
 147 #undef wxDEFINE_UNICHAR_OPERATOR
 148
 149     // this is needed for expressions like 'Z'-c
 150     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 151     int operator-(char c) const { return m_value - From8bit(c); }
 152     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 153     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 154 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 155     int operator-(wint_t c) const { return m_value - (value_type)c; }
 156 #endif
 157
 158 private:
 159     static value_type From8bit(char c);
 160     static char To8bit(value_type c);
 161
 162 private:
 163     value_type m_value;
 164 };
 165
 166
 167 // Writeable reference to a character in wxString.
 168 //
 169 // This class can be used in the same way wxChar is used, except that changing
 170 // its value updates the underlying string object.
 171 class WXDLLIMPEXP_BASE wxUniCharRef
 172 {
 173 private:
 174     typedef wxStringImpl::iterator iterator;
 175
 176     // create the reference
 177 #if wxUSE_UNICODE_UTF8
 178     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 179 #else
 180     wxUniCharRef(iterator pos) : m_pos(pos) {}
 181 #endif
 182
 183 public:
 184     // NB: we have to make this public, because we don't have wxString
 185     //     declaration available here and so can't declare wxString::iterator
 186     //     as friend; so at least don't use a ctor but a static function
 187     //     that must be used explicitly (this is more than using 'explicit'
 188     //     keyword on ctor!):
 189 #if wxUSE_UNICODE_UTF8
 190     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 191         { return wxUniCharRef(node, pos); }
 192 #else
 193     static wxUniCharRef CreateForString(iterator pos)
 194         { return wxUniCharRef(pos); }
 195 #endif
 196
 197     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 198
 199 #if wxUSE_UNICODE_UTF8
 200     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 201 #endif // wxUSE_UNICODE_UTF8
 202
 203     bool IsAscii() const { return UniChar().IsAscii(); }
 204
 205     // Assignment operators:
 206 #if wxUSE_UNICODE_UTF8
 207     wxUniCharRef& operator=(const wxUniChar& c);
 208 #else
 209     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 210 #endif
 211
 212     wxUniCharRef& operator=(const wxUniCharRef& c)
 213         { return *this = c.UniChar(); }
 214
 215     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 216     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 217     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 218     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 219 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 220     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 221 #endif
 222
 223     // Conversions to the same types as wxUniChar is convertible too:
 224     operator char() const { return UniChar(); }
 225     operator int() const { return UniChar(); }
 226     operator unsigned char() const { return UniChar(); }
 227     operator wchar_t() const { return UniChar(); }
 228 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 229     operator wint_t() const { return UniChar(); }
 230 #endif
 231 #ifndef wxWCHAR_T_IS_UINT
 232     operator unsigned int() const { return UniChar(); }
 233 #endif
 234
 235     // see wxUniChar::operator bool etc. for explanation
 236     operator bool() const { return (bool)UniChar(); }
 237     bool operator!() const { return !UniChar(); }
 238 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 239     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 240     bool operator&&(bool v) const { return UniChar() && v; }
 241 #endif
 242
 243     // Comparison operators:
 244 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 245     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 246     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 247     bool operator op(char c) const { return UniChar() op c; }                 \
 248     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 249     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 250     bool operator op(int c) const { return UniChar() op c; }                  \
 251     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 252
 253     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 254
 255 #undef wxDEFINE_UNICHARREF_OPERATOR
 256
 257     // for expressions like c-'A':
 258     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 259     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 260     int operator-(char c) const { return UniChar() - c; }
 261     int operator-(unsigned char c) const { return UniChar() - c; }
 262     int operator-(wchar_t c) const { return UniChar() - c; }
 263 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 264     int operator-(wint_t c) const { return UniChar() - c; }
 265 #endif
 266
 267 private:
 268 #if wxUSE_UNICODE_UTF8
 269     wxUniChar UniChar() const;
 270 #else
 271     wxUniChar UniChar() const { return *m_pos; }
 272 #endif
 273
 274     friend class WXDLLIMPEXP_BASE wxUniChar;
 275
 276 private:
 277     // reference to the string and pointer to the character in string
 278 #if wxUSE_UNICODE_UTF8
 279     wxStringIteratorNode& m_node;
 280 #endif
 281     iterator m_pos;
 282 };
 283
 284 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 285 {
 286     m_value = c.UniChar().m_value;
 287 }
 288
 289 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 290 // implemented in terms of member comparison functions
 291
 292 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 293
 294 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 295 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 296
 297 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 298 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 299
 300 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 301 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 302 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 303 #endif
 304
 305 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 306
 307 #undef wxCMP_REVERSE
 308
 309 // for expressions like c-'A':
 310 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 311 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 312 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 313 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 314 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 315 #endif
 316
 317 #endif /* _WX_UNICHAR_H_ */