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 // Copyright:   (c) 2007 REA Elektronik GmbH
   7 // Licence:     wxWindows licence
   8 ///////////////////////////////////////////////////////////////////////////////
   9
  10 #ifndef _WX_UNICHAR_H_
  11 #define _WX_UNICHAR_H_
  12
  13 #include "wx/defs.h"
  14 #include "wx/chartype.h"
  15 #include "wx/stringimpl.h"
  16
  17 class WXDLLIMPEXP_FWD_BASE wxUniCharRef;
  18 class WXDLLIMPEXP_FWD_BASE wxString;
  19
  20 // This class represents single Unicode character. It can be converted to
  21 // and from char or wchar_t and implements commonly used character operations.
  22 class WXDLLIMPEXP_BASE wxUniChar
  23 {
  24 public:
  25     // NB: this is not wchar_t on purpose, it needs to represent the entire
  26     //     Unicode code points range and wchar_t may be too small for that
  27     //     (e.g. on Win32 where wchar_t* is encoded in UTF-16)
  28     typedef wxUint32 value_type;
  29
  30     wxUniChar() : m_value(0) {}
  31
  32     // Create the character from 8bit character value encoded in the current
  33     // locale's charset.
  34     wxUniChar(char c) { m_value = From8bit(c); }
  35     wxUniChar(unsigned char c) { m_value = From8bit((char)c); }
  36
  37 #define wxUNICHAR_DEFINE_CTOR(type) \
  38     wxUniChar(type c) { m_value = (value_type)c; }
  39     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_CTOR)
  40 #undef wxUNICHAR_DEFINE_CTOR
  41
  42     wxUniChar(const wxUniCharRef& c);
  43
  44     // Returns Unicode code point value of the character
  45     value_type GetValue() const { return m_value; }
  46
  47 #if wxUSE_UNICODE_UTF8
  48     // buffer for single UTF-8 character
  49     struct Utf8CharBuffer
  50     {
  51         char data[5];
  52         operator const char*() const { return data; }
  53     };
  54
  55     // returns the character encoded as UTF-8
  56     // (NB: implemented in stringops.cpp)
  57     Utf8CharBuffer AsUTF8() const;
  58 #endif // wxUSE_UNICODE_UTF8
  59
  60     // Returns true if the character is an ASCII character:
  61     bool IsAscii() const { return m_value < 0x80; }
  62
  63     // Returns true if the character is representable as a single byte in the
  64     // current locale encoding and return this byte in output argument c (which
  65     // must be non-NULL)
  66     bool GetAsChar(char *c) const
  67     {
  68 #if wxUSE_UNICODE
  69         if ( !IsAscii() )
  70         {
  71 #if !wxUSE_UTF8_LOCALE_ONLY
  72             if ( GetAsHi8bit(m_value, c) )
  73                 return true;
  74 #endif // !wxUSE_UTF8_LOCALE_ONLY
  75
  76             return false;
  77         }
  78 #endif // wxUSE_UNICODE
  79
  80         *c = wx_truncate_cast(char, m_value);
  81         return true;
  82     }
  83
  84     // Conversions to char and wchar_t types: all of those are needed to be
  85     // able to pass wxUniChars to verious standard narrow and wide character
  86     // functions
  87     operator char() const { return To8bit(m_value); }
  88     operator unsigned char() const { return (unsigned char)To8bit(m_value); }
  89
  90 #define wxUNICHAR_DEFINE_OPERATOR_PAREN(type) \
  91     operator type() const { return (type)m_value; }
  92     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_PAREN)
  93 #undef wxUNICHAR_DEFINE_OPERATOR_PAREN
  94
  95     // We need this operator for the "*p" part of expressions like "for (
  96     // const_iterator p = begin() + nStart; *p; ++p )". In this case,
  97     // compilation would fail without it because the conversion to bool would
  98     // be ambiguous (there are all these int types conversions...). (And adding
  99     // operator unspecified_bool_type() would only makes the ambiguity worse.)
 100     operator bool() const { return m_value != 0; }
 101     bool operator!() const { return !((bool)*this); }
 102
 103     // And this one is needed by some (not all, but not using ifdefs makes the
 104     // code easier) compilers to parse "str[0] && *p" successfully
 105     bool operator&&(bool v) const { return (bool)*this && v; }
 106
 107     // Assignment operators:
 108     wxUniChar& operator=(const wxUniChar& c) { if (&c != this) m_value = c.m_value; return *this; }
 109     wxUniChar& operator=(const wxUniCharRef& c);
 110     wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; }
 111     wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; }
 112
 113 #define wxUNICHAR_DEFINE_OPERATOR_EQUAL(type) \
 114     wxUniChar& operator=(type c) { m_value = (value_type)c; return *this; }
 115     wxDO_FOR_INT_TYPES(wxUNICHAR_DEFINE_OPERATOR_EQUAL)
 116 #undef wxUNICHAR_DEFINE_OPERATOR_EQUAL
 117
 118     // Comparison operators:
 119 #define wxDEFINE_UNICHAR_CMP_WITH_INT(T, op) \
 120     bool operator op(T c) const { return m_value op (value_type)c; }
 121
 122     // define the given comparison operator for all the types
 123 #define wxDEFINE_UNICHAR_OPERATOR(op)                                         \
 124     bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\
 125     bool operator op(char c) const { return m_value op From8bit(c); }         \
 126     bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \
 127     wxDO_FOR_INT_TYPES_1(wxDEFINE_UNICHAR_CMP_WITH_INT, op)
 128
 129     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR)
 130
 131 #undef wxDEFINE_UNICHAR_OPERATOR
 132 #undef wxDEFINE_UNCHAR_CMP_WITH_INT
 133
 134     // this is needed for expressions like 'Z'-c
 135     int operator-(const wxUniChar& c) const { return m_value - c.m_value; }
 136     int operator-(char c) const { return m_value - From8bit(c); }
 137     int operator-(unsigned char c) const { return m_value - From8bit((char)c); }
 138     int operator-(wchar_t c) const { return m_value - (value_type)c; }
 139
 140
 141 private:
 142     // notice that we implement these functions inline for 7-bit ASCII
 143     // characters purely for performance reasons
 144     static value_type From8bit(char c)
 145     {
 146 #if wxUSE_UNICODE
 147         if ( (unsigned char)c < 0x80 )
 148             return c;
 149
 150         return FromHi8bit(c);
 151 #else
 152         return c;
 153 #endif
 154     }
 155
 156     static char To8bit(value_type c)
 157     {
 158 #if wxUSE_UNICODE
 159         if ( c < 0x80 )
 160             return wx_truncate_cast(char, c);
 161
 162         return ToHi8bit(c);
 163 #else
 164         return c;
 165 #endif
 166     }
 167
 168     // helpers of the functions above called to deal with non-ASCII chars
 169     static value_type FromHi8bit(char c);
 170     static char ToHi8bit(value_type v);
 171     static bool GetAsHi8bit(value_type v, char *c);
 172
 173 private:
 174     value_type m_value;
 175 };
 176
 177
 178 // Writeable reference to a character in wxString.
 179 //
 180 // This class can be used in the same way wxChar is used, except that changing
 181 // its value updates the underlying string object.
 182 class WXDLLIMPEXP_BASE wxUniCharRef
 183 {
 184 private:
 185     typedef wxStringImpl::iterator iterator;
 186
 187     // create the reference
 188 #if wxUSE_UNICODE_UTF8
 189     wxUniCharRef(wxString& str, iterator pos) : m_str(str), m_pos(pos) {}
 190 #else
 191     wxUniCharRef(iterator pos) : m_pos(pos) {}
 192 #endif
 193
 194 public:
 195     // NB: we have to make this public, because we don't have wxString
 196     //     declaration available here and so can't declare wxString::iterator
 197     //     as friend; so at least don't use a ctor but a static function
 198     //     that must be used explicitly (this is more than using 'explicit'
 199     //     keyword on ctor!):
 200 #if wxUSE_UNICODE_UTF8
 201     static wxUniCharRef CreateForString(wxString& str, iterator pos)
 202         { return wxUniCharRef(str, pos); }
 203 #else
 204     static wxUniCharRef CreateForString(iterator pos)
 205         { return wxUniCharRef(pos); }
 206 #endif
 207
 208     wxUniChar::value_type GetValue() const { return UniChar().GetValue(); }
 209
 210 #if wxUSE_UNICODE_UTF8
 211     wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); }
 212 #endif // wxUSE_UNICODE_UTF8
 213
 214     bool IsAscii() const { return UniChar().IsAscii(); }
 215     bool GetAsChar(char *c) const { return UniChar().GetAsChar(c); }
 216
 217     // Assignment operators:
 218 #if wxUSE_UNICODE_UTF8
 219     wxUniCharRef& operator=(const wxUniChar& c);
 220 #else
 221     wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; }
 222 #endif
 223
 224     wxUniCharRef& operator=(const wxUniCharRef& c)
 225         { if (&c != this) *this = c.UniChar(); return *this; }
 226
 227 #define wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL(type) \
 228     wxUniCharRef& operator=(type c) { return *this = wxUniChar(c); }
 229     wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL)
 230 #undef wxUNICHAR_REF_DEFINE_OPERATOR_EQUAL
 231
 232     // Conversions to the same types as wxUniChar is convertible too:
 233 #define wxUNICHAR_REF_DEFINE_OPERATOR_PAREN(type) \
 234     operator type() const { return UniChar(); }
 235     wxDO_FOR_CHAR_INT_TYPES(wxUNICHAR_REF_DEFINE_OPERATOR_PAREN)
 236 #undef wxUNICHAR_REF_DEFINE_OPERATOR_PAREN
 237
 238     // see wxUniChar::operator bool etc. for explanation
 239     operator bool() const { return (bool)UniChar(); }
 240     bool operator!() const { return !UniChar(); }
 241     bool operator&&(bool v) const { return UniChar() && v; }
 242
 243 #define wxDEFINE_UNICHARREF_CMP_WITH_INT(T, op) \
 244     bool operator op(T c) const { return UniChar() op c; }
 245
 246     // Comparison operators:
 247 #define wxDEFINE_UNICHARREF_OPERATOR(op)                                      \
 248     bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\
 249     bool operator op(const wxUniChar& c) const { return UniChar() op c; }     \
 250     wxDO_FOR_CHAR_INT_TYPES_1(wxDEFINE_UNICHARREF_CMP_WITH_INT, op)
 251
 252     wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR)
 253
 254 #undef wxDEFINE_UNICHARREF_OPERATOR
 255 #undef wxDEFINE_UNICHARREF_CMP_WITH_INT
 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
 264 private:
 265 #if wxUSE_UNICODE_UTF8
 266     wxUniChar UniChar() const;
 267 #else
 268     wxUniChar UniChar() const { return *m_pos; }
 269 #endif
 270
 271     friend class WXDLLIMPEXP_FWD_BASE wxUniChar;
 272
 273 private:
 274     // reference to the string and pointer to the character in string
 275 #if wxUSE_UNICODE_UTF8
 276     wxString& m_str;
 277 #endif
 278     iterator m_pos;
 279 };
 280
 281 inline wxUniChar::wxUniChar(const wxUniCharRef& c)
 282 {
 283     m_value = c.UniChar().m_value;
 284 }
 285
 286 inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c)
 287 {
 288     m_value = c.UniChar().m_value;
 289     return *this;
 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 wxDEFINE_COMPARISONS_BY_REV(char, const wxUniChar&)
 296 wxDEFINE_COMPARISONS_BY_REV(char, const wxUniCharRef&)
 297
 298 wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniChar&)
 299 wxDEFINE_COMPARISONS_BY_REV(wchar_t, const wxUniCharRef&)
 300
 301 wxDEFINE_COMPARISONS_BY_REV(const wxUniChar&, const wxUniCharRef&)
 302
 303 // for expressions like c-'A':
 304 inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 305 inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 306 inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); }
 307
 308 #endif /* _WX_UNICHAR_H_ */