UTF8.hpp

   1 #ifndef UTF8_DOT_HPP
   2 #define UTF8_DOT_HPP
   3
   4 // clang-format off
   5
   6 struct UTF8_tail : range<'\x80', '\xBF'> {};
   7
   8 struct UTF8_1 : range<'\x00', '\x7F'> {};
   9
  10 struct UTF8_2 : seq<range<'\xC2', '\xDF'>, UTF8_tail> {};
  11
  12 struct UTF8_3 : sor<seq<one<'\xE0'>, range<'\xA0', '\xBF'>, UTF8_tail>,
  13                     seq<range<'\xE1', '\xEC'>, rep<2, UTF8_tail>>,
  14                     seq<one<'\xED'>, range<'\x80', '\x9F'>, UTF8_tail>,
  15                     seq<range<'\xEE', '\xEF'>, rep<2, UTF8_tail>>> {};
  16
  17 struct UTF8_4 : sor<seq<one<'\xF0'>, range<'\x90', '\xBF'>, rep<2, UTF8_tail>>,
  18                     seq<range<'\xF1', '\xF3'>, rep<3, UTF8_tail>>,
  19                     seq<one<'\xF4'>, range<'\x80', '\x8F'>, rep<2, UTF8_tail>>> {};
  20
  21 struct UTF8_non_ascii : sor<UTF8_2, UTF8_3, UTF8_4> {};
  22
  23 struct VUCHAR : sor<VCHAR, UTF8_non_ascii> {};
  24
  25 // clang-format on
  26
  27 #endif // UTF8_DOT_HPP