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     wxCHECK_SUNCC_VERSION(5, 9)
  25     #define wxWINT_T_IS_SEPARATE_TYPE
  26 #endif
  27
  28 // helper macro for doing something dependent on whether wint_t is or isn't a
  29 // typedef inside another macro
  30 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  31     #define wxIF_WINT_T_TYPE(x) x
  32 #else // !wxWINT_T_IS_SEPARATE_TYPE
  33     #define wxIF_WINT_T_TYPE(x)
  34 #endif // wxWINT_T_IS_SEPARATE_TYPE/!wxWINT_T_IS_SEPARATE_TYPE
  35
  36 // wchar_t seems to be defined as unsigned short by all Windows compilers but
  37 // unsigned int everywhere else
  38 #ifndef __WIN32__
  39     #define wxWCHAR_T_IS_UINT
  40 #endif
  41
  42 class WXDLLIMPEXP_BASE wxUniCharRef;
  43 class WXDLLIMPEXP_BASE wxStringIteratorNode;
  44
  45 // This class represents single Unicode character. It can be converted to
  46 // and from char or wchar_t and implements commonly used character operations.
  47 class WXDLLIMPEXP_BASE wxUniChar
  48 {
  49 public:
  50     // NB: this is not wchar_t on purpose, it needs to represent the entire
  51     //     Unicode code points range and wchar_t may be too small for that
  52     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  53     typedef wxUint32 value_type;
  54
  55     wxUniChar() : m_value(0) {}
  56
  57     // Create the character from 8bit character value encoded in the current
  58     // locale's charset.
  59     wxUniChar(char c) { m_value = From8bit(c); }
  60     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  61
  62     // Create the character from a wchar_t character value.
  63     wxUniChar(wchar_t c) { m_value = c; }
  64 #ifdef wxWINT_T_IS_SEPARATE_TYPE
  65     wxUniChar(wint_t c) { m_value = c; }
  66 #endif
  67
  68     wxUniChar(int c) { m_value = c; }
  69     wxUniChar(long int c) { m_value = c; }
  70
  71     wxUniChar(const wxUniCharRef& c);
  72
  73     // Returns Unicode code point value of the character
  74     value_type GetValue() const { return m_value; }
  75
  76 #if wxUSE_UNICODE_UTF8
  77     // buffer for single UTF-8 character
  78     struct Utf8CharBuffer
  79     {
  80         char data[5];
  81         operator const char*() const { return data; }
  82     };
  83
  84     // returns the character encoded as UTF-8
  85     // (NB: implemented in stringops.cpp)
  86     Utf8CharBuffer AsUTF8() const;
  87 #endif // wxUSE_UNICODE_UTF8
  88
  89     // Returns true if the character is an ASCII character:
  90     bool IsAscii() const { return m_value < 0x80; }
  91
  92     // Conversions to char and wchar_t types: all of those are needed to be
  93     // able to pass wxUniChars to verious standard narrow and wide character
  94     // functions
  95     operator char() const { return To8bit(m_value); }
  96     operator wchar_t() const { return m_value; }
  97     operator int() const { return m_value; }
  98     operator long int() const { return m_value; }
  99 #ifndef wxWCHAR_T_IS_UINT
 100     operator unsigned int() const { return m_value; }
 101 #endif
 102
 103     // More conversions needed for other standard functions: uchar is for VC++
 104     // _mbxxx() ones (to which toxxx/isxxx() are mapped when _MBCS is defined)
 105     // and some wide character functions take wint_t which happens to be the
 106     // same as wchar_t for Windows compilers but not for g++ (except for the
 107     // special Apple version)
 108     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
 109 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 110     operator wint_t() const { return m_value; }
 111 #endif
 112
 113     // We need this operator for the "*p" part of expressions like "for (
 114     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
 115     // compilation would fail without it because the conversion to bool would
 116     // be ambiguous (there are all these int types conversions...). (And adding
 117     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 118     operator bool() const { return m_value != 0; }
 119     bool operator!() const { return !((bool)*this); }
 120 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 121     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 122     // We need this for VC++ < 8 or DigitalMars and expressions like
 123     // "str[0] && *p":
 124     bool operator&&(bool v) const { return (bool)*this && v; }
 125 #endif
 126
 127     // Assignment operators:
 128     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 129     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 130     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 131     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 132     wxUniChar& operator=(int c) { m_value = c; return *this; }
 133     wxUniChar& operator=(long int c) { m_value = c; return *this; }
 134 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 135     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 136 #endif
 137
 138     // Comparison operators:
 139
 140     // define the given comparison operator for all the types
 141 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 142     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 143     bool operator op(char c) const { return m_value op From8bit(c); }         \
 144     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 145     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 146     bool operator op(int c) const { return m_value op (value_type)c; }        \
 147     bool operator op(long int c) const { return m_value op (value_type)c; }   \
 148     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 149
 150     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 151
 152 #undef wxDEFINE_UNICHAR_OPERATOR
 153
 154     // this is needed for expressions like 'Z'-c
 155     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 156     int operator-(char c) const { return m_value - From8bit(c); }
 157     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 158     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 159 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 160     int operator-(wint_t c) const { return m_value - (value_type)c; }
 161 #endif
 162
 163 private:
 164     static value_type From8bit(char c);
 165     static char To8bit(value_type c);
 166
 167 private:
 168     value_type m_value;
 169 };
 170
 171
 172 // Writeable reference to a character in wxString.
 173 //
 174 // This class can be used in the same way wxChar is used, except that changing
 175 // its value updates the underlying string object.
 176 class WXDLLIMPEXP_BASE wxUniCharRef
 177 {
 178 private:
 179     typedef wxStringImpl::iterator iterator;
 180
 181     // create the reference
 182 #if wxUSE_UNICODE_UTF8
 183     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 184 #else
 185     wxUniCharRef(iterator pos) : m_pos(pos) {}
 186 #endif
 187
 188 public:
 189     // NB: we have to make this public, because we don't have wxString
 190     //     declaration available here and so can't declare wxString::iterator
 191     //     as friend; so at least don't use a ctor but a static function
 192     //     that must be used explicitly (this is more than using 'explicit'
 193     //     keyword on ctor!):
 194 #if wxUSE_UNICODE_UTF8
 195     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 196         { return wxUniCharRef(node, pos); }
 197 #else
 198     static wxUniCharRef CreateForString(iterator pos)
 199         { return wxUniCharRef(pos); }
 200 #endif
 201
 202     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 203
 204 #if wxUSE_UNICODE_UTF8
 205     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 206 #endif // wxUSE_UNICODE_UTF8
 207
 208     bool IsAscii() const { return UniChar().IsAscii(); }
 209
 210     // Assignment operators:
 211 #if wxUSE_UNICODE_UTF8
 212     wxUniCharRef& operator=(const wxUniChar& c);
 213 #else
 214     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 215 #endif
 216
 217     wxUniCharRef& operator=(const wxUniCharRef& c)
 218         { return *this = c.UniChar(); }
 219
 220     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 221     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 222     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 223     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 224     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 225 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 226     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 227 #endif
 228
 229     // Conversions to the same types as wxUniChar is convertible too:
 230     operator char() const { return UniChar(); }
 231     operator int() const { return UniChar(); }
 232     operator long int() const { return UniChar(); }
 233     operator unsigned char() const { return UniChar(); }
 234     operator wchar_t() const { return UniChar(); }
 235 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 236     operator wint_t() const { return UniChar(); }
 237 #endif
 238 #ifndef wxWCHAR_T_IS_UINT
 239     operator unsigned int() const { return UniChar(); }
 240 #endif
 241
 242     // see wxUniChar::operator bool etc. for explanation
 243     operator bool() const { return (bool)UniChar(); }
 244     bool operator!() const { return !UniChar(); }
 245 #if (defined(__VISUALC__) && __VISUALC__ < 1400) || \
 246     defined(__DIGITALMARS__) || defined(__BORLANDC__)
 247     bool operator&&(bool v) const { return UniChar() && v; }
 248 #endif
 249
 250     // Comparison operators:
 251 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 252     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 253     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 254     bool operator op(char c) const { return UniChar() op c; }                 \
 255     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 256     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 257     bool operator op(int c) const { return UniChar() op c; }                  \
 258     bool operator op(long int c) const { return UniChar() op c; }             \
 259     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 260
 261     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 262
 263 #undef wxDEFINE_UNICHARREF_OPERATOR
 264
 265     // for expressions like c-'A':
 266     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 267     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 268     int operator-(char c) const { return UniChar() - c; }
 269     int operator-(unsigned char c) const { return UniChar() - c; }
 270     int operator-(wchar_t c) const { return UniChar() - c; }
 271 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 272     int operator-(wint_t c) const { return UniChar() - c; }
 273 #endif
 274
 275 private:
 276 #if wxUSE_UNICODE_UTF8
 277     wxUniChar UniChar() const;
 278 #else
 279     wxUniChar UniChar() const { return *m_pos; }
 280 #endif
 281
 282     friend class WXDLLIMPEXP_BASE wxUniChar;
 283
 284 private:
 285     // reference to the string and pointer to the character in string
 286 #if wxUSE_UNICODE_UTF8
 287     wxStringIteratorNode& m_node;
 288 #endif
 289     iterator m_pos;
 290 };
 291
 292 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 293 {
 294     m_value = c.UniChar().m_value;
 295 }
 296
 297 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 298 // implemented in terms of member comparison functions
 299
 300 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 301
 302 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 303 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 304
 305 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 306 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 307
 308 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 309 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 310 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 311 #endif
 312
 313 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 314
 315 #undef wxCMP_REVERSE
 316
 317 // for expressions like c-'A':
 318 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 319 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 320 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 321 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 322 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 323 #endif
 324
 325 #endif /* _WX_UNICHAR_H_ */