include/library/utf8.hpp

   1 #ifndef _library__utf8__hpp__included__
   2 #define _library__utf8__hpp__included__
   3
   4 #include <cstdint>
   5 #include <cstdlib>
   6 #include <string>
   7
   8 /**
   9  * Initial state for UTF-8 parser.
  10  */
  11 extern const uint16_t utf8_initial_state;
  12 /**
  13  * Parse a byte.
  14  *
  15  * Parameter ch: The character to parse. -1 for end of string.
  16  * Parameter state: The state. Mutated.
  17  * Returns: The codepoint, or -1 if no codepoint emitted.
  18  *
  19  * Note: When called with EOF, max 1 codepoint can be emitted.
  20  */
  21 int32_t utf8_parse_byte(int ch, uint16_t& state) throw();
  22 /**
  23  * Return length of string in UTF-8 codepoints.
  24  *
  25  * Parameter str: The string.
  26  * Returns: The length in codepoints.
  27  */
  28 size_t utf8_strlen(const std::string& str) throw();
  29
  30 /**
  31  * Iterator copy from UTF-8 to UTF-32
  32  */
  33 template<typename srcitr, typename dstitr>
  34 inline void copy_from_utf8(srcitr begin, srcitr end, dstitr target)
  35 {
  36         uint16_t state = utf8_initial_state;
  37         for(srcitr i = begin; i != end; i++) {
  38                 int32_t x = utf8_parse_byte(*i, state);
  39                 if(x >= 0) {
  40                         *target = x;
  41                         ++target;
  42                 }
  43         }
  44         int32_t x = utf8_parse_byte(-1, state);
  45         if(x >= 0) {
  46                 *target = x;
  47                 ++target;
  48         }
  49 }
  50
  51 #endif