include/a/utf.h

   1 /*!
   2  @file utf.h
   3  @brief Unicode Transformation Format
   4   | Unicode | UTF-8 |
   5   |---------|-------|
   6   | U+0000000 ~ U+0000007F | 0XXXXXXX |
   7   | U+0000080 ~ U+000007FF | 110XXXXX 10XXXXXX |
   8   | U+0000800 ~ U+0000FFFF | 1110XXXX 10XXXXXX 10XXXXXX |
   9   | U+0010000 ~ U+001FFFFF | 11110XXX 10XXXXXX 10XXXXXX 10XXXXXX |
  10   | U+0200000 ~ U+03FFFFFF | 111110XX 10XXXXXX 10XXXXXX 10XXXXXX 10XXXXXX |
  11   | U+4000000 ~ U+7FFFFFFF | 1111110X 10XXXXXX 10XXXXXX 10XXXXXX 10XXXXXX 10XXXXXX |
  12 */
  13
  14 #ifndef LIBA_UTF_H
  15 #define LIBA_UTF_H
  16
  17 #include "a.h"
  18
  19 /*!
  20  @ingroup A
  21  @addtogroup A_UTF Unicode Transformation Format
  22  @{
  23 */
  24
  25 #if defined(__cplusplus)
  26 extern "C" {
  27 #endif /* __cplusplus */
  28
  29 /*!
  30  @brief encode one unicode code point into UTF-8
  31  @param[in] str a buffer to store a UTF-8 character
  32  @param[in] val the unicode value to be encoded
  33  @return offset to the next character
  34 */
  35 A_EXTERN unsigned int a_utf_encode(void *str, a_u32 val);
  36
  37 /*!
  38  @brief decode one unicode code point from UTF-8
  39  @param[in] str string terminated with a null character
  40  @param[in,out] val a pointer to a variable that stores unicode
  41  @return offset to the next character
  42   @retval 0 null character or invalid
  43 */
  44 A_EXTERN unsigned int a_utf_decode(void const *str, a_u32 *val);
  45
  46 /*!
  47  @brief length of a UTF-8 string terminated with a null character
  48  @param[in] str string terminated with a null character
  49  @return length of the UTF-8 string
  50 */
  51 A_EXTERN a_size a_utf_length(void const *str);
  52 #define a_utf_len(str) a_utf_length(str)
  53
  54 #if defined(__cplusplus)
  55 } /* extern "C" */
  56 #endif /* __cplusplus */
  57
  58 /*! @} A_UTF */
  59
  60 #endif /* a/utf.h */