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
 121     // And this one is needed by some (not all, but not using ifdefs makes the
 122     // code easier) compilers to parse "str[0] && *p" successfully
 123     bool operator&&(bool v) const { return (bool)*this && v; }
 124
 125     // Assignment operators:
 126     wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; }
 127     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 128     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 129     wxUniChar& operator=(wchar_t c) { m_value = c; return *this; }
 130     wxUniChar& operator=(int c) { m_value = c; return *this; }
 131     wxUniChar& operator=(long int c) { m_value = c; return *this; }
 132 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 133     wxUniChar& operator=(wint_t c) { m_value = c; return *this; }
 134 #endif
 135
 136     // Comparison operators:
 137
 138     // define the given comparison operator for all the types
 139 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 140     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 141     bool operator op(char c) const { return m_value op From8bit(c); }         \
 142     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 143     bool operator op(wchar_t c) const { return m_value op (value_type)c; }    \
 144     bool operator op(int c) const { return m_value op (value_type)c; }        \
 145     bool operator op(long int c) const { return m_value op (value_type)c; }   \
 146     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return m_value op (value_type)c; } )
 147
 148     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 149
 150 #undef wxDEFINE_UNICHAR_OPERATOR
 151
 152     // this is needed for expressions like 'Z'-c
 153     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 154     int operator-(char c) const { return m_value - From8bit(c); }
 155     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 156     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 157 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 158     int operator-(wint_t c) const { return m_value - (value_type)c; }
 159 #endif
 160
 161 private:
 162     static value_type From8bit(char c);
 163     static char To8bit(value_type c);
 164
 165 private:
 166     value_type m_value;
 167 };
 168
 169
 170 // Writeable reference to a character in wxString.
 171 //
 172 // This class can be used in the same way wxChar is used, except that changing
 173 // its value updates the underlying string object.
 174 class WXDLLIMPEXP_BASE wxUniCharRef
 175 {
 176 private:
 177     typedef wxStringImpl::iterator iterator;
 178
 179     // create the reference
 180 #if wxUSE_UNICODE_UTF8
 181     wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {}
 182 #else
 183     wxUniCharRef(iterator pos) : m_pos(pos) {}
 184 #endif
 185
 186 public:
 187     // NB: we have to make this public, because we don't have wxString
 188     //     declaration available here and so can't declare wxString::iterator
 189     //     as friend; so at least don't use a ctor but a static function
 190     //     that must be used explicitly (this is more than using 'explicit'
 191     //     keyword on ctor!):
 192 #if wxUSE_UNICODE_UTF8
 193     static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos)
 194         { return wxUniCharRef(node, pos); }
 195 #else
 196     static wxUniCharRef CreateForString(iterator pos)
 197         { return wxUniCharRef(pos); }
 198 #endif
 199
 200     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 201
 202 #if wxUSE_UNICODE_UTF8
 203     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 204 #endif // wxUSE_UNICODE_UTF8
 205
 206     bool IsAscii() const { return UniChar().IsAscii(); }
 207
 208     // Assignment operators:
 209 #if wxUSE_UNICODE_UTF8
 210     wxUniCharRef& operator=(const wxUniChar& c);
 211 #else
 212     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 213 #endif
 214
 215     wxUniCharRef& operator=(const wxUniCharRef& c)
 216         { return *this = c.UniChar(); }
 217
 218     wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); }
 219     wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); }
 220     wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); }
 221     wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); }
 222     wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); }
 223 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 224     wxUniCharRef& operator=(wint_t c) { return *this = wxUniChar(c); }
 225 #endif
 226
 227     // Conversions to the same types as wxUniChar is convertible too:
 228     operator char() const { return UniChar(); }
 229     operator int() const { return UniChar(); }
 230     operator long int() const { return UniChar(); }
 231     operator unsigned char() const { return UniChar(); }
 232     operator wchar_t() const { return UniChar(); }
 233 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 234     operator wint_t() const { return UniChar(); }
 235 #endif
 236 #ifndef wxWCHAR_T_IS_UINT
 237     operator unsigned int() const { return UniChar(); }
 238 #endif
 239
 240     // see wxUniChar::operator bool etc. for explanation
 241     operator bool() const { return (bool)UniChar(); }
 242     bool operator!() const { return !UniChar(); }
 243     bool operator&&(bool v) const { return UniChar() && v; }
 244
 245     // Comparison operators:
 246 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 247     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 248     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 249     bool operator op(char c) const { return UniChar() op c; }                 \
 250     bool operator op(unsigned char c) const { return UniChar() op c; }        \
 251     bool operator op(wchar_t c) const { return UniChar() op c; }              \
 252     bool operator op(int c) const { return UniChar() op c; }                  \
 253     bool operator op(long int c) const { return UniChar() op c; }             \
 254     wxIF_WINT_T_TYPE( bool operator op(wint_t c) const { return UniChar() op c; } )
 255
 256     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 257
 258 #undef wxDEFINE_UNICHARREF_OPERATOR
 259
 260     // for expressions like c-'A':
 261     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 262     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 263     int operator-(char c) const { return UniChar() - c; }
 264     int operator-(unsigned char c) const { return UniChar() - c; }
 265     int operator-(wchar_t c) const { return UniChar() - c; }
 266 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 267     int operator-(wint_t c) const { return UniChar() - c; }
 268 #endif
 269
 270 private:
 271 #if wxUSE_UNICODE_UTF8
 272     wxUniChar UniChar() const;
 273 #else
 274     wxUniChar UniChar() const { return *m_pos; }
 275 #endif
 276
 277     friend class WXDLLIMPEXP_BASE wxUniChar;
 278
 279 private:
 280     // reference to the string and pointer to the character in string
 281 #if wxUSE_UNICODE_UTF8
 282     wxStringIteratorNode& m_node;
 283 #endif
 284     iterator m_pos;
 285 };
 286
 287 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 288 {
 289     m_value = c.UniChar().m_value;
 290 }
 291
 292 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 293 // implemented in terms of member comparison functions
 294
 295 #define wxCMP_REVERSE(c1, c2, op) c2 op c1
 296
 297 wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE)
 298 wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE)
 299
 300 wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE)
 301 wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE)
 302
 303 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 304 wxDEFINE_COMPARISONS(wint_t, const wxUniChar&, wxCMP_REVERSE)
 305 wxDEFINE_COMPARISONS(wint_t, const wxUniCharRef&, wxCMP_REVERSE)
 306 #endif
 307
 308 wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE)
 309
 310 #undef wxCMP_REVERSE
 311
 312 // for expressions like c-'A':
 313 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 314 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 315 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 316 #ifdef wxWINT_T_IS_SEPARATE_TYPE
 317 inline int operator-(wint_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 318 #endif
 319
 320 #endif /* _WX_UNICHAR_H_ */