tidy-html5/src/utf8.h

47 lines
1.4 KiB
C
Raw Normal View History

2011-11-17 02:44:16 +00:00
#ifndef __UTF8_H__
#define __UTF8_H__
/* utf8.h -- convert characters to/from UTF-8
(c) 1998-2006 (W3C) MIT, ERCIM, Keio University
See tidy.h for the copyright notice.
*/
#include "tidyplatform.h"
#include "tidybuffio.h"
2011-11-17 02:44:16 +00:00
/* UTF-8 encoding/decoding support
** Does not convert character "codepoints", i.e. to/from 10646.
*/
TY_PRIVATE int TY_(DecodeUTF8BytesToChar)( uint* c, uint firstByte, ctmbstr successorBytes,
2011-11-17 02:44:16 +00:00
TidyInputSource* inp, int* count );
TY_PRIVATE int TY_(EncodeCharToUTF8Bytes)( uint c, tmbstr encodebuf,
2011-11-17 02:44:16 +00:00
TidyOutputSink* outp, int* count );
TY_PRIVATE uint TY_(GetUTF8)( ctmbstr str, uint *ch );
TY_PRIVATE tmbstr TY_(PutUTF8)( tmbstr buf, uint c );
2011-11-17 02:44:16 +00:00
#define UNICODE_BOM_BE 0xFEFF /* big-endian (default) UNICODE BOM */
#define UNICODE_BOM UNICODE_BOM_BE
#define UNICODE_BOM_LE 0xFFFE /* little-endian UNICODE BOM */
#define UNICODE_BOM_UTF8 0xEFBBBF /* UTF-8 UNICODE BOM */
TY_PRIVATE Bool TY_(IsValidUTF16FromUCS4)( tchar ucs4 );
TY_PRIVATE Bool TY_(IsHighSurrogate)( tchar ch );
TY_PRIVATE Bool TY_(IsLowSurrogate)( tchar ch );
2011-11-17 02:44:16 +00:00
TY_PRIVATE Bool TY_(IsCombinedChar)( tchar ch );
TY_PRIVATE Bool TY_(IsValidCombinedChar)( tchar ch );
2011-11-17 02:44:16 +00:00
TY_PRIVATE tchar TY_(CombineSurrogatePair)( tchar high, tchar low );
TY_PRIVATE Bool TY_(SplitSurrogatePair)( tchar utf16, tchar* high, tchar* low );
2011-11-17 02:44:16 +00:00
#endif /* __UTF8_H__ */