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 class WXDLLIMPEXP_FWD_BASE wxUniCharRef;
  19 class WXDLLIMPEXP_FWD_BASE wxString;
  20
  21 // This class represents single Unicode character. It can be converted to
  22 // and from char or wchar_t and implements commonly used character operations.
  23 class WXDLLIMPEXP_BASE wxUniChar
  24 {
  25 public:
  26     // NB: this is not wchar_t on purpose, it needs to represent the entire
  27     //     Unicode code points range and wchar_t may be too small for that
  28     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  29     typedef wxUint32 value_type;
  30
  31     wxUniChar() : m_value(0) {}
  32
  33     // Create the character from 8bit character value encoded in the current
  34     // locale's charset.
  35     wxUniChar(char c) { m_value = From8bit(c); }
  36     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  37
  38 #define wxUNICHAR_DEFINE_CTOR(type) \
  39     wxUniChar(type c) { m_value = (value_type)c; }
  40     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_CTOR)
  41 #undef wxUNICHAR_DEFINE_CTOR
  42
  43     wxUniChar(const wxUniCharRef& c);
  44
  45     // Returns Unicode code point value of the character
  46     value_type GetValue() const { return m_value; }
  47
  48 #if wxUSE_UNICODE_UTF8
  49     // buffer for single UTF-8 character
  50     struct Utf8CharBuffer
  51     {
  52         char data[5];
  53         operator const char*() const { return data; }
  54     };
  55
  56     // returns the character encoded as UTF-8
  57     // (NB: implemented in stringops.cpp)
  58     Utf8CharBuffer AsUTF8() const;
  59 #endif // wxUSE_UNICODE_UTF8
  60
  61     // Returns true if the character is an ASCII character:
  62     bool IsAscii() const { return m_value < 0x80; }
  63
  64     // Returns true if the character is representable as a single byte in the
  65     // current locale encoding and return this byte in output argument c (which
  66     // must be non-NULL)
  67     bool GetAsChar(char *c) const
  68     {
  69 #if wxUSE_UNICODE
  70         if ( !IsAscii() )
  71         {
  72 #if !wxUSE_UTF8_LOCALE_ONLY
  73             if ( GetAsHi8bit(m_value, c) )
  74                 return true;
  75 #endif // !wxUSE_UTF8_LOCALE_ONLY
  76
  77             return false;
  78         }
  79 #endif // wxUSE_UNICODE
  80
  81         *c = wx_truncate_cast(char, m_value);
  82         return true;
  83     }
  84
  85     // Conversions to char and wchar_t types: all of those are needed to be
  86     // able to pass wxUniChars to verious standard narrow and wide character
  87     // functions
  88     operator char() const { return To8bit(m_value); }
  89     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  90
  91 #define wxUNICHAR_DEFINE_OPERATOR_PAREN(type) \
  92     operator type() const { return (type)m_value; }
  93     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_PAREN)
  94 #undef wxUNICHAR_DEFINE_OPERATOR_PAREN
  95
  96     // We need this operator for the "*p" part of expressions like "for (
  97     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  98     // compilation would fail without it because the conversion to bool would
  99     // be ambiguous (there are all these int types conversions...). (And adding
 100     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 101     operator bool() const { return m_value != 0; }
 102     bool operator!() const { return !((bool)*this); }
 103
 104     // And this one is needed by some (not all, but not using ifdefs makes the
 105     // code easier) compilers to parse "str[0] && *p" successfully
 106     bool operator&&(bool v) const { return (bool)*this && v; }
 107
 108     // Assignment operators:
 109     wxUniChar& operator=(const wxUniChar& c) { if (&c != this) m_value = c.m_value; return *this; }
 110     wxUniChar& operator=(const wxUniCharRef& c);
 111     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 112     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 113
 114 #define wxUNICHAR_DEFINE_OPERATOR_EQUAL(type) \
 115     wxUniChar& operator=(type c) { m_value = (value_type)c; return *this; }
 116     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_EQUAL)
 117 #undef wxUNICHAR_DEFINE_OPERATOR_EQUAL
 118
 119     // Comparison operators:
 120 #define wxDEFINE_UNICHAR_CMP_WITH_INT(T, op) \
 121     bool operator op(T c) const { return m_value op (value_type)c; }
 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     wxDO_FOR_INT_TYPES_1(wxDEFINE_UNICHAR_CMP_WITH_INT, op)
 129
 130     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 131
 132 #undef wxDEFINE_UNICHAR_OPERATOR
 133 #undef wxDEFINE_UNCHAR_CMP_WITH_INT
 134
 135     // this is needed for expressions like 'Z'-c
 136     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 137     int operator-(char c) const { return m_value - From8bit(c); }
 138     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 139     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 140
 141
 142 private:
 143     // notice that we implement these functions inline for 7-bit ASCII
 144     // characters purely for performance reasons
 145     static value_type From8bit(char c)
 146     {
 147 #if wxUSE_UNICODE
 148         if ( (unsigned char)c < 0x80 )
 149             return c;
 150
 151         return FromHi8bit(c);
 152 #else
 153         return c;
 154 #endif
 155     }
 156
 157     static char To8bit(value_type c)
 158     {
 159 #if wxUSE_UNICODE
 160         if ( c < 0x80 )
 161             return wx_truncate_cast(char, c);
 162
 163         return ToHi8bit(c);
 164 #else
 165         return c;
 166 #endif
 167     }
 168
 169     // helpers of the functions above called to deal with non-ASCII chars
 170     static value_type FromHi8bit(char c);
 171     static char ToHi8bit(value_type v);
 172     static bool GetAsHi8bit(value_type v, char *c);
 173
 174 private:
 175     value_type m_value;
 176 };
 177
 178
 179 // Writeable reference to a character in wxString.
 180 //
 181 // This class can be used in the same way wxChar is used, except that changing
 182 // its value updates the underlying string object.
 183 class WXDLLIMPEXP_BASE wxUniCharRef
 184 {
 185 private:
 186     typedef wxStringImpl::iterator iterator;
 187
 188     // create the reference
 189 #if wxUSE_UNICODE_UTF8
 190     wxUniCharRef(wxString& str, iterator pos) : m_str(str), m_pos(pos) {}
 191 #else
 192     wxUniCharRef(iterator pos) : m_pos(pos) {}
 193 #endif
 194
 195 public:
 196     // NB: we have to make this public, because we don't have wxString
 197     //     declaration available here and so can't declare wxString::iterator
 198     //     as friend; so at least don't use a ctor but a static function
 199     //     that must be used explicitly (this is more than using 'explicit'
 200     //     keyword on ctor!):
 201 #if wxUSE_UNICODE_UTF8
 202     static wxUniCharRef CreateForString(wxString& str, iterator pos)
 203         { return wxUniCharRef(str, pos); }
 204 #else
 205     static wxUniCharRef CreateForString(iterator pos)
 206         { return wxUniCharRef(pos); }
 207 #endif
 208
 209     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 210
 211 #if wxUSE_UNICODE_UTF8
 212     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 213 #endif // wxUSE_UNICODE_UTF8
 214
 215     bool IsAscii() const { return UniChar().IsAscii(); }
 216     bool GetAsChar(char *c) const { return UniChar().GetAsChar(c); }
 217
 218     // Assignment operators:
 219 #if wxUSE_UNICODE_UTF8
 220     wxUniCharRef& operator=(const wxUniChar& c);
 221 #else
 222     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 223 #endif
 224
 225     wxUniCharRef& operator=(const wxUniCharRef& c)
 226         { if (&c != this) *this = c.UniChar(); return *this; }
 227
 228 #define wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL(type) \
 229     wxUniCharRef& operator=(type c) { return *this = wxUniChar(c); }
 230     wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL)
 231 #undef wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL
 232
 233     // Conversions to the same types as wxUniChar is convertible too:
 234 #define wxUNICHAR_REF_DEFINE_OPERATOR_PAREN(type) \
 235     operator type() const { return UniChar(); }
 236     wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_PAREN)
 237 #undef wxUNICHAR_REF_DEFINE_OPERATOR_PAREN
 238
 239     // see wxUniChar::operator bool etc. for explanation
 240     operator bool() const { return (bool)UniChar(); }
 241     bool operator!() const { return !UniChar(); }
 242     bool operator&&(bool v) const { return UniChar() && v; }
 243
 244 #define wxDEFINE_UNICHARREF_CMP_WITH_INT(T, op) \
 245     bool operator op(T c) const { return UniChar() op c; }
 246
 247     // Comparison operators:
 248 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 249     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 250     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 251     wxDO_FOR_CHAR_INT_TYPES_1(wxDEFINE_UNICHARREF_CMP_WITH_INT, op)
 252
 253     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 254
 255 #undef wxDEFINE_UNICHARREF_OPERATOR
 256 #undef wxDEFINE_UNICHARREF_CMP_WITH_INT
 257
 258     // for expressions like c-'A':
 259     int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); }
 260     int operator-(const wxUniChar& c) const { return UniChar() - c; }
 261     int operator-(char c) const { return UniChar() - c; }
 262     int operator-(unsigned char c) const { return UniChar() - c; }
 263     int operator-(wchar_t c) const { return UniChar() - c; }
 264
 265 private:
 266 #if wxUSE_UNICODE_UTF8
 267     wxUniChar UniChar() const;
 268 #else
 269     wxUniChar UniChar() const { return *m_pos; }
 270 #endif
 271
 272     friend class WXDLLIMPEXP_FWD_BASE wxUniChar;
 273
 274 private:
 275     // reference to the string and pointer to the character in string
 276 #if wxUSE_UNICODE_UTF8
 277     wxString& m_str;
 278 #endif
 279     iterator m_pos;
 280 };
 281
 282 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 283 {
 284     m_value = c.UniChar().m_value;
 285 }
 286
 287 inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c)
 288 {
 289     m_value = c.UniChar().m_value;
 290     return *this;
 291 }
 292
 293 // Comparison operators for the case when wxUniChar(Ref) is the second operand
 294 // implemented in terms of member comparison functions
 295
 296 wxDEFINE_COMPARISONS_BY_REV(char, const wxUniChar&)
 297 wxDEFINE_COMPARISONS_BY_REV(char, const wxUniCharRef&)
 298
 299 wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniChar&)
 300 wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniCharRef&)
 301
 302 wxDEFINE_COMPARISONS_BY_REV(const wxUniChar&, const wxUniCharRef&)
 303
 304 // for expressions like c-'A':
 305 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 306 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 307 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 308
 309 #endif /* _WX_UNICHAR_H_ */