Charset.h File Reference

#include <dcl/Config.h>
#include <bits/types/mbstate_t.h>
#include <dcl/Object.h>
#include <dcl/Exception.h>
#include <dcl/String.h>

Go to the source code of this file.

Classes
class	CharsetConvertException
class	CharsetEncoder
class	CharsetDecoder
class	UTF8Encoder
class	UTF8Decoder
class	UTF16Encoder
class	UTF16Decoder
class	UTF32Encoder
class	UTF32Decoder
class	AsciiEncoder
class	AsciiDecoder
class	Latin1Encoder
class	Latin1Decoder
class	LocaleEncoder
class	LocaleDecoder

Macros
#define	__DCL_CHARSET_H__   20071008
#define	IS_UTF8(bom)
#define	IS_UTF16BE(bom)
#define	IS_UTF16LE(bom)
#define	IS_UTF16(bom)
#define	IS_UTF32BE(bom)
#define	IS_UTF32LE(bom)
#define	IS_UTF32(bom)

Typedefs
typedef uint16_t	utf16_t
typedef uint32_t	utf32_t

Enumerations
enum	Charset {   CS_LOCALE = (int)0 , CS_ASCII , CS_LATIN1 , CS_UTF8 ,   CS_UTF16 , CS_UTF32 }
enum	UnicodeByteOrder { CS_DEFAULT_ENDIAN = __BYTE_ORDER__ , CS_LITTLE_ENDIAN = __ORDER_LITTLE_ENDIAN__ , CS_BIG_ENDIAN = __ORDER_BIG_ENDIAN__ }
enum	{   CS_NOERROR = 0 , CS_ILLEGAL_UCS , CS_SOURCE_FEW , CS_ILLEGAL_SEQUENCE ,   CS_ILLEGAL_UCS2 }

Variables
__DCL_BEGIN_NAMESPACE typedef uint32_t	ucs4_t

Macro Definition Documentation

__DCL_CHARSET_H__

#define __DCL_CHARSET_H__   20071008

Definition at line 2 of file Charset.h.

IS_UTF16

#define IS_UTF16

(

bom

)

Value:

(IS_UTF16BE(bom) || IS_UTF16LE(bom))

Definition at line 41 of file Charset.h.

IS_UTF16BE

#define IS_UTF16BE

(

bom

)

Value:

((bom[0] == '\xfe') && (bom[1] == '\xff'))

Definition at line 39 of file Charset.h.

IS_UTF16LE

#define IS_UTF16LE

(

bom

)

Value:

((bom[0] == '\xff') && (bom[1] == '\xfe'))

Definition at line 40 of file Charset.h.

IS_UTF32

#define IS_UTF32

(

bom

)

Value:

(IS_UTF32BE(bom) || IS_UTF32LE(bom))

Definition at line 46 of file Charset.h.

IS_UTF32BE

#define IS_UTF32BE

(

bom

)

Value:

                            ((bom[0] == '\x00') && (bom[1] == '\x00') \
                            && (bom[2] == '\xfe') && (bom[3] == '\xff'))

Definition at line 42 of file Charset.h.

#define IS_UTF32BE(bom) ((bom[0] == '\x00') && (bom[1] == '\x00') \
                            && (bom[2] == '\xfe') && (bom[3] == '\xff'))

IS_UTF32LE

#define IS_UTF32LE

(

bom

)

Value:

                            ((bom[0] == '\xff') && (bom[1] == '\xfe') \
                            && (bom[2] == '\x00') && (bom[3] == '\x00'))

Definition at line 44 of file Charset.h.

#define IS_UTF32LE(bom) ((bom[0] == '\xff') && (bom[1] == '\xfe') \
                            && (bom[2] == '\x00') && (bom[3] == '\x00'))

IS_UTF8

#define IS_UTF8

(

bom

)

Value:

                            ((bom[0] == '\xef') && (bom[1] == '\xbb') \
                            && (bom[2] == '\xbf'))

Definition at line 37 of file Charset.h.

#define IS_UTF8(bom)    ((bom[0] == '\xef') && (bom[1] == '\xbb') \
                            && (bom[2] == '\xbf'))

Typedef Documentation

utf16_t

typedef uint16_t utf16_t

Definition at line 34 of file Charset.h.

utf32_t

typedef uint32_t utf32_t

Definition at line 35 of file Charset.h.

Enumeration Type Documentation

anonymous enum

anonymous enum

Enumerator
CS_NOERROR
CS_ILLEGAL_UCS
CS_SOURCE_FEW
CS_ILLEGAL_SEQUENCE
CS_ILLEGAL_UCS2

Definition at line 65 of file Charset.h.

     {
    CS_NOERROR = 0,
    // encode()
    CS_ILLEGAL_UCS,               // can't UCS4 ==> MB
    // decode()
    CS_SOURCE_FEW,             // source bytes few
    CS_ILLEGAL_SEQUENCE,     // illegal bytes sequence
    CS_ILLEGAL_UCS2              // can't UCS4 ==> UCS2 , on sizeof(wchar_t) == 2
};

Charset

enum Charset

Enumerator
CS_LOCALE
CS_ASCII
CS_LATIN1
CS_UTF8
CS_UTF16
CS_UTF32

Definition at line 48 of file Charset.h.

{
    CS_LOCALE = (int)0,
    CS_ASCII,                   // 7bit, US_ASCII
    CS_LATIN1,                 // 8bit, ISO-8859-1
    CS_UTF8,                 // http://www.faqs.org/rfcs/rfc3629.html
    CS_UTF16,
    CS_UTF32
};

UnicodeByteOrder

enum UnicodeByteOrder

Enumerator
CS_DEFAULT_ENDIAN
CS_LITTLE_ENDIAN
CS_BIG_ENDIAN

Definition at line 58 of file Charset.h.

                      {
    // UTF16, UTF32
    CS_DEFAULT_ENDIAN = __BYTE_ORDER__,    // platform dependent default
    CS_LITTLE_ENDIAN = __ORDER_LITTLE_ENDIAN__,
    CS_BIG_ENDIAN = __ORDER_BIG_ENDIAN__
};

Variable Documentation

ucs4_t

__DCL_BEGIN_NAMESPACE typedef uint32_t ucs4_t

Definition at line 33 of file Charset.h.