utf8.h File Reference

Go to the source code of this file.

Macros
#define	UNICODE_BOM   UNICODE_BOM_BE
#define	UNICODE_BOM_BE   0xFEFF /* big-endian (default) UNICODE BOM */
#define	UNICODE_BOM_LE   0xFFFE /* little-endian UNICODE BOM */
#define	UNICODE_BOM_UTF8   0xEFBBBF /* UTF-8 UNICODE BOM */

Functions
TY_PRIVATE tchar	TY_❪CombineSurrogatePair❫ (tchar high, tchar low)
TY_PRIVATE int	TY_❪DecodeUTF8BytesToChar❫ (uint *c, uint firstByte, ctmbstr successorBytes, TidyInputSource *inp, int *count)
TY_PRIVATE int	TY_❪EncodeCharToUTF8Bytes❫ (uint c, tmbstr encodebuf, TidyOutputSink *outp, int *count)
TY_PRIVATE uint	TY_❪GetUTF8❫ (ctmbstr str, uint *ch)
TY_PRIVATE Bool	TY_❪IsCombinedChar❫ (tchar ch)
TY_PRIVATE Bool	TY_❪IsHighSurrogate❫ (tchar ch)
TY_PRIVATE Bool	TY_❪IsLowSurrogate❫ (tchar ch)
TY_PRIVATE Bool	TY_❪IsValidCombinedChar❫ (tchar ch)
TY_PRIVATE Bool	TY_❪IsValidUTF16FromUCS4❫ (tchar ucs4)
TY_PRIVATE tmbstr	TY_❪PutUTF8❫ (tmbstr buf, uint c)
TY_PRIVATE Bool	TY_❪SplitSurrogatePair❫ (tchar utf16, tchar *high, tchar *low)

Macro Definition Documentation

UNICODE_BOM

#define UNICODE_BOM   UNICODE_BOM_BE

UNICODE_BOM_BE

#define UNICODE_BOM_BE   0xFEFF /* big-endian (default) UNICODE BOM */

UNICODE_BOM_LE

#define UNICODE_BOM_LE   0xFFFE /* little-endian UNICODE BOM */

UNICODE_BOM_UTF8

#define UNICODE_BOM_UTF8   0xEFBBBF /* UTF-8 UNICODE BOM */

Function Documentation

TY_❪CombineSurrogatePair❫()

TY_PRIVATE tchar TY_❪CombineSurrogatePair❫	(	tchar	high,
		tchar	low
	)

TY_❪DecodeUTF8BytesToChar❫()

TY_PRIVATE int TY_❪DecodeUTF8BytesToChar❫	(	uint *	c,
		uint	firstByte,
		ctmbstr	successorBytes,
		TidyInputSource *	inp,
		int *	count
	)

TY_❪EncodeCharToUTF8Bytes❫()

TY_PRIVATE int TY_❪EncodeCharToUTF8Bytes❫	(	uint	c,
		tmbstr	encodebuf,
		TidyOutputSink *	outp,
		int *	count
	)

TY_❪GetUTF8❫()

TY_PRIVATE uint TY_❪GetUTF8❫	(	ctmbstr	str,
		uint *	ch
	)

TY_❪IsCombinedChar❫()

TY_PRIVATE Bool TY_❪IsCombinedChar❫

(

tchar

ch

)

TY_❪IsHighSurrogate❫()

TY_PRIVATE Bool TY_❪IsHighSurrogate❫

(

tchar

ch

)

TY_❪IsLowSurrogate❫()

TY_PRIVATE Bool TY_❪IsLowSurrogate❫

(

tchar

ch

)

TY_❪IsValidCombinedChar❫()

TY_PRIVATE Bool TY_❪IsValidCombinedChar❫

(

tchar

ch

)

TY_❪IsValidUTF16FromUCS4❫()

TY_PRIVATE Bool TY_❪IsValidUTF16FromUCS4❫

(

tchar

ucs4

)

TY_❪PutUTF8❫()

TY_PRIVATE tmbstr TY_❪PutUTF8❫	(	tmbstr	buf,
		uint	c
	)

TY_❪SplitSurrogatePair❫()

TY_PRIVATE Bool TY_❪SplitSurrogatePair❫	(	tchar	utf16,
		tchar *	high,
		tchar *	low
	)