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     wxUniChar(long int c) { m_value = c; }
  69
  70     wxUniChar(const wxUniCharRef& c);
  71
  72     // Returns Unicode code point value of the character
  73     value_type GetValue() const { return m_value; }
  74
  75 #if wxUSE_UNICODE_UTF8
  76     // buffer for single UTF-8 character
  77     struct Utf8CharBuffer
  78     {
  79         char data[5];
  80         operator const char*() const { return data; }
  81     };
  82
  83     // returns the character encoded as UTF-8
  84     // (NB: implemented in stringops.cpp)
  85     Utf8CharBuffer AsUTF8() const;
  86 #endif // wxUSE_UNICODE_UTF8
  87
  88     // Returns true if the character is an ASCII character:
  89     bool IsAscii() const { return m_value < 0x80; }
  90
  91     // Conversions to char and wchar_t types: all of those are needed to be
  92     // able to pass wxUniChars to verious standard narrow and wide character
  93     // functions
  94     operator char() const { return To8bit(m_value); }
  95     operator wchar_t() const { return m_value; }
  96     operator int() const { return m_value; }
  97     operator long int() const { return m_value; }
  98 #ifndef wxWCHAR_T_IS_UINT
  99     operator unsigned int() const { return m_value; }
 100 #endif
 101
 102     // More conversions needed for other standard functions: uchar is for VC++
 103     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
 104     // and some wide character functions take wint_t which happens to be the
 105     // same as wchar_t for Windows compilers but not for g++ (except for the
 106     // special Apple version)
 107     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
 108 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 109     operator wint_t() const { return m_value; }
 110 #endif
 111
 112     // We need this operator for the "*p" part of expressions like "for (
 113     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
 114     // compilation would fail without it because the conversion to bool would
 115     // be ambiguous (there are all these int types conversions...). (And adding
 116     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 117     operator bool() const { return m_value != 0; }
 118     bool operator!() const { return !((bool)*this); }
 119 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 120     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 121     // We need this for VC++ < 8 or DigitalMars and expressions like
 122     // "str[0] && *p":
 123     bool operator&&(bool v) const { return (bool)*this && v; }
 124 #endif
 125
 126     // Assignment operators:
 127     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 128     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 129     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 130     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 131     wxUniChar& operator=(int c) { m_value = c; return *this; }
 132     wxUniChar& operator=(long int c) { m_value = c; return *this; }
 133 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 134     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 135 #endif
 136
 137     // Comparison operators:
 138
 139     // define the given comparison operator for all the types
 140 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 141     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 142     bool operator op(char c) const { return m_value op From8bit(c); }         \
 143     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 144     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 145     bool operator op(int c) const { return m_value op (value_type)c; }        \
 146     bool operator op(long int c) const { return m_value op (value_type)c; }   \
 147     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 148
 149     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 150
 151 #undef wxDEFINE_UNICHAR_OPERATOR
 152
 153     // this is needed for expressions like 'Z'-c
 154     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 155     int operator-(char c) const { return m_value - From8bit(c); }
 156     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 157     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 158 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 159     int operator-(wint_t c) const { return m_value - (value_type)c; }
 160 #endif
 161
 162 private:
 163     static value_type From8bit(char c);
 164     static char To8bit(value_type c);
 165
 166 private:
 167     value_type m_value;
 168 };
 169
 170
 171 // Writeable reference to a character in wxString.
 172 //
 173 // This class can be used in the same way wxChar is used, except that changing
 174 // its value updates the underlying string object.
 175 class WXDLLIMPEXP_BASE wxUniCharRef
 176 {
 177 private:
 178     typedef wxStringImpl::iterator iterator;
 179
 180     // create the reference
 181 #if wxUSE_UNICODE_UTF8
 182     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 183 #else
 184     wxUniCharRef(iterator pos) : m_pos(pos) {}
 185 #endif
 186
 187 public:
 188     // NB: we have to make this public, because we don't have wxString
 189     //     declaration available here and so can't declare wxString::iterator
 190     //     as friend; so at least don't use a ctor but a static function
 191     //     that must be used explicitly (this is more than using 'explicit'
 192     //     keyword on ctor!):
 193 #if wxUSE_UNICODE_UTF8
 194     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 195         { return wxUniCharRef(node, pos); }
 196 #else
 197     static wxUniCharRef CreateForString(iterator pos)
 198         { return wxUniCharRef(pos); }
 199 #endif
 200
 201     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 202
 203 #if wxUSE_UNICODE_UTF8
 204     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 205 #endif // wxUSE_UNICODE_UTF8
 206
 207     bool IsAscii() const { return UniChar().IsAscii(); }
 208
 209     // Assignment operators:
 210 #if wxUSE_UNICODE_UTF8
 211     wxUniCharRef& operator=(const wxUniChar& c);
 212 #else
 213     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 214 #endif
 215
 216     wxUniCharRef& operator=(const wxUniCharRef& c)
 217         { return *this = c.UniChar(); }
 218
 219     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 220     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 221     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 222     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 223     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 224 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 225     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 226 #endif
 227
 228     // Conversions to the same types as wxUniChar is convertible too:
 229     operator char() const { return UniChar(); }
 230     operator int() const { return UniChar(); }
 231     operator long int() const { return UniChar(); }
 232     operator unsigned char() const { return UniChar(); }
 233     operator wchar_t() const { return UniChar(); }
 234 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 235     operator wint_t() const { return UniChar(); }
 236 #endif
 237 #ifndef wxWCHAR_T_IS_UINT
 238     operator unsigned int() const { return UniChar(); }
 239 #endif
 240
 241     // see wxUniChar::operator bool etc. for explanation
 242     operator bool() const { return (bool)UniChar(); }
 243     bool operator!() const { return !UniChar(); }
 244 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 245     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 246     bool operator&&(bool v) const { return UniChar() && v; }
 247 #endif
 248
 249     // Comparison operators:
 250 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 251     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 252     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 253     bool operator op(char c) const { return UniChar() op c; }                 \
 254     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 255     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 256     bool operator op(int c) const { return UniChar() op c; }                  \
 257     bool operator op(long int c) const { return UniChar() op c; }             \
 258     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 259
 260     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 261
 262 #undef wxDEFINE_UNICHARREF_OPERATOR
 263
 264     // for expressions like c-'A':
 265     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 266     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 267     int operator-(char c) const { return UniChar() - c; }
 268     int operator-(unsigned char c) const { return UniChar() - c; }
 269     int operator-(wchar_t c) const { return UniChar() - c; }
 270 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 271     int operator-(wint_t c) const { return UniChar() - c; }
 272 #endif
 273
 274 private:
 275 #if wxUSE_UNICODE_UTF8
 276     wxUniChar UniChar() const;
 277 #else
 278     wxUniChar UniChar() const { return *m_pos; }
 279 #endif
 280
 281     friend class WXDLLIMPEXP_BASE wxUniChar;
 282
 283 private:
 284     // reference to the string and pointer to the character in string
 285 #if wxUSE_UNICODE_UTF8
 286     wxStringIteratorNode& m_node;
 287 #endif
 288     iterator m_pos;
 289 };
 290
 291 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 292 {
 293     m_value = c.UniChar().m_value;
 294 }
 295
 296 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 297 // implemented in terms of member comparison functions
 298
 299 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 300
 301 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 302 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 303
 304 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 305 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 306
 307 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 308 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 309 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 310 #endif
 311
 312 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 313
 314 #undef wxCMP_REVERSE
 315
 316 // for expressions like c-'A':
 317 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 318 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 319 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 320 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 321 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 322 #endif
 323
 324 #endif /* _WX_UNICHAR_H_ */