/*=============================================================================
* drt/sys/kchar.h
*	DRT Unicode character classes.
*
*	These classes provide a portable representation of Unicode characters.
*
* Notes
*	The 'DrtChar' type is used to represent Unicode characters rather than
*	the 'wchar_t' because the latter varies across implementations and can
*	be as small as 8 bits or as large as 64 bits.
*	Using this class type provides a better interface with better control
*	and more functionality.
*
* History
*	1.00, 1998-08-19, David R Tribble.
*	First cut, converted from C++ to C.
*
* Copyright ©1999, by David R. Tribble, all rights reserved.
* See "drt/sys/copyr.txt" for more information.
*----------------------------------------------------------------------------*/

#ifndef drt_sys_kchar_h
#define drt_sys_kchar_h		100

#ifdef __cplusplus
extern "C"
{
#endif


/* Identification */

#ifndef NO_H_IDENT
static const char	drt_sys_kchar_h_id[] =
    "@(#)drt/sys/kchar.h 1.00";
#endif


/* Special includes */

#ifndef drt_sys_kdefs_h
 #include "kdefs.h"
#endif


/* Local includes */

#ifndef drt_sys_kdebug_h
 #include "kdebug.h"
#endif


/* Local wrappers */

#include "klib1.h"


/*-----------------------------------------------------------------------------
* Type drt_char_t
*	Unicode character code datatype (16 bit).
*
* History
*	100, 1998-08-19, David R Tribble.
*	First cut.
*----------------------------------------------------------------------------*/

#define DRT_CHAR_VS	100		/* Type version			*/

typedef drt_uint16_t	drt_char_t;	/* Unicode character type	*/


/* Public constants */

extern const int	drt_char_vs;	/* Library version		*/

extern const unsigned char	drt_atoe_01047[0x100];
					/* ASCII to EBCDIC table	*/
extern const unsigned char	drt_etoa_01047[0x100];
					/* EBCDIC to ASCII table	*/

#define drt_char_code_VS	100	/* Enum version			*/

enum /*anonymous*/
{
    C_NUL =	0x0000,			/* Null				*/
    C_HT =	0x0009,			/* Tab				*/
    C_LF =	0x000A,			/* Linefeed			*/
    C_VT =	0x000B,			/* Vertical tab			*/
    C_FF =	0x000C,			/* Formfeed			*/
    C_CR =	0x000D,			/* Carriage return		*/
    C_SP =	0x0020,			/* Space			*/
    C_DEL =	0x007F,			/* Delete			*/
    C_NBSP =	0x00A0,			/* Nonbreaking space		*/
    C_EOLN =	0x2028,			/* End-of-line			*/
    C_SURR_H =	0xD800,			/* Surrogate, high half		*/
    C_SURR_L =	0xDC00,			/* Surrogate, low half		*/
    C_EOF =	0xFFFF,			/* End-of-file			*/
    C_NAC =	0xFFFF,			/* Not-a-char			*/
    C_BOM =	0xFEFF,			/* Byte-order-mark		*/
    C_BOM_R =	0xFFFE,			/* Byte-order-mark, swapped	*/
    C_MAX =	0xFFFF			/* Highest char code		*/
};


/* Public variables */

extern struct DrtTraceGrp	drt_char_grp;
					/* Debugging trace group	*/


/* Public functions */

extern bool		drt_char_is_alnum(drt_char_t c);
					/* Is alphanumeric		*/
extern bool		drt_char_is_alpha(drt_char_t c);
					/* Is alphabetic		*/
extern bool		drt_char_is_control(drt_char_t c);
					/* Is control			*/
extern bool		drt_char_is_digit(drt_char_t c);
					/* Is a digit			*/
extern bool		drt_char_is_eoln(drt_char_t c);
					/* Is an end-of-line char	*/
extern bool		drt_char_is_hex(drt_char_t c);
					/* Is a hexdecimal digit	*/
extern bool		drt_char_is_lower(drt_char_t c);
					/* Is lowercase alphabetic	*/
extern bool		drt_char_is_null(drt_char_t c);
					/* Is null char			*/
extern bool		drt_char_is_octal(drt_char_t c);
					/* Is an octal digit		*/
extern bool		drt_char_is_print(drt_char_t c);
					/* Is printable			*/
extern bool		drt_char_is_print_8(drt_char_t c);
					/* Is printable as 8-bit ASCII	*/
extern bool		drt_char_is_punct(drt_char_t c);
					/* Is punctuation		*/
extern bool		drt_char_is_space(drt_char_t c);
					/* Is space			*/
extern bool		drt_char_is_surr(drt_char_t c);
					/* Is a surrogate half-char	*/
extern bool		drt_char_is_upper(drt_char_t c);
					/* Is uppercase alphabetic	*/
extern bool		drt_char_is_white(drt_char_t c);
					/* Is whitespace		*/

extern bool		drt_char_is_ASCII_7(drt_char_t c);
					/* Is 7-bit ASCII/ISO-646	*/
extern bool		drt_char_is_ISO_8859_1(drt_char_t c);
					/* Is 8-bit ISO-8859-1/Latin-1	*/

extern int		drt_char_to_UTF_7(char *s, drt_char_t c);
					/* Convert to UTF-7 seq		*/
extern int		drt_char_to_UTF_8(char *s, drt_char_t c);
					/* Convert to UTF-8 seq		*/
extern int		drt_char_to_ISO_8859_1(char *s, drt_char_t c);
					/* Convert to ISO-8859-1 seq	*/

extern int		drt_char_from_UTF_7(drt_char_t *c, const char *s);
					/* Convert from UTF-7 seq	*/
extern int		drt_char_from_UTF_8(drt_char_t *c, const char *s);
					/* Convert from UTF-8 seq	*/
extern int		drt_char_from_ISO_8859_1(drt_char_t *c, const char *s);
					/* Convert from ISO-8859-1 seq	*/

extern int		drt_char_copy(drt_char_t d[], const drt_char_t s[],
			    int len);	/* Copy buffer contents		*/


/* Wrapper end */

#include "klib2.h"


#ifdef __cplusplus
}
#endif

#endif /* drt_sys_kchar_h */

/* End drt/sys/kchar.h */