utf8.h

   1 #ifndef UTF8_H
   2 #define UTF8_H
   3
   4 enum gwm_encoding {
   5     ENCODING_COMPOUND, /* X Consortium Compound Text (ISO 2022-like) */
   6     ENCODING_LATIN_1 /* ISO 8859-1 */
   7 };
   8
   9 extern char *to_utf8( enum gwm_encoding encoding, const char *in, size_t len );
  10
  11 /* Return a pointer to the first illegal byte in STR, or NULL if the
  12    entire string is legal. */
  13 extern PURE unsigned char *utf8_illegal( const unsigned char *str );
  14
  15 /* Return the number of characters in STR. */
  16 extern PURE int utf8_length( const unsigned char *str );
  17
  18 /* Duplicate a UTF-8 string.  The output is allocated with malloc() and
  19    is guaranteed to be a legal UTF-8 sequence, even if the input was not. */
  20 extern MALLOC unsigned char *utf8_dup_valid_len( const unsigned char *str,
  21                                                  int num_bytes );
  22 extern MALLOC unsigned char *utf8_dup_valid( const unsigned char *str );
  23
  24 /* Return the character pointed to by P (which is advanced). */
  25 extern uint32_t utf8_next( const unsigned char **p );
  26
  27 #if DEBUG
  28 extern void cleanup_utf8( void );
  29 #endif
  30
  31 #endif