code.h

此源码功能是捕获网页中的链接并进行分析

//Judge to which kind of char set a character belongs

#ifndef __ENCODE_H__
#define __ENCODE_H__

#ifdef __cplusplus
extern "C" {
#endif

//======================================================================
//from jin:gbk_big5.h
//======================================================================
#define GB2312_CHN_CHAR_NUM	6768

#define GB2312_SYM_HIMIN	0xA1	/* GB symbol : minmum value of first byte */
#define GB2312_SYM_HIMAX	0xF7	/* GB symbol : maxmum value of first byte */
#define GB2312_SYM_LOMIN	0xA1	/* GB symbol : minmum value of second byte */
#define GB2312_SYM_LOMAX	0xFE	/* GB symbol : maxmum value of second byte */

#define GB2312_CHAR_HIMIN	0xB0	/* GB char   : min - 1st byte */
#define GB2312_CHAR_HIMAX	0xF7	/* GB char   : max - 1st byte */
#define GB2312_CHAR_LOMIN	0xA1	/* GB char   : min - 2nd byte */
#define GB2312_CHAR_LOMAX	0xFE	/* GB char   : max - 2nd byte */

#define GB2312_MARK_HIMIN	0xA1	/* GB mark   : min - 1st byte */
#define GB2312_MARK_HIMAX	0xAF	/* GB mark   : max - 1st byte */
#define GB2312_MARK_LOMIN	0xA1	/* GB mark   : min - 2nd byte */
#define GB2312_MARK_LOMAX	0xFE	/* GB mark   : max - 2nd byte */


//========================================================================
#define GBK_CHN_CHAR_NUM	((0xfe-0x81+1)*(0xfe-0x40+1))

#define GBK_SYM_HIMIN	0x81	/* GBK symbol : minmum value of first byte */
#define GBK_SYM_HIMAX	0xFE	/* GBK symbol : maxmum value of first byte */
#define GBK_SYM_LOMIN	0x40	/* GBK symbol : minmum value of second byte */
#define GBK_SYM_LOMAX	0xFE	/* GBK symbol : maxmum value of second byte */

#define GBK_CHAR_HIMIN	0xB0	/* GB 2312 char   : min - 1st byte */
#define GBK_CHAR_HIMAX	0xF7	/* GB 2312 char   : max - 1st byte */
#define GBK_CHAR_LOMIN	0xA1	/* GB 2312 char   : min - 2nd byte */
#define GBK_CHAR_LOMAX	0xFE	/* GB 2312 char   : max - 2nd byte */

#define GBK_CHAR1_HIMIN	0x81	/* GBK 1 char   : min - 1st byte */
#define GBK_CHAR1_HIMAX	0xA0	/* GBK 1 char   : max - 1st byte */
#define GBK_CHAR1_LOMIN	0x40	/* GBK 1 char   : min - 2nd byte */
#define GBK_CHAR1_LOMAX	0xFE	/* GBK 1 char   : max - 2nd byte */

#define GBK_CHAR2_HIMIN	0xAA	/* GBK 2 char   : min - 1st byte */
#define GBK_CHAR2_HIMAX	0xFE	/* GBK 2 char   : max - 1st byte */
#define GBK_CHAR2_LOMIN	0x40	/* GBK 2 char   : min - 2nd byte */
#define GBK_CHAR2_LOMAX	0xA0	/* GBK 2 char   : max - 2nd byte */


#define GBK_MARK_HIMIN	0xA1	/* GB 2312 mark   : min - 1st byte */
#define GBK_MARK_HIMAX	0xA9	/* GB 2312 mark   : max - 1st byte */
#define GBK_MARK_LOMIN	0xA1	/* GB 2312 mark   : min - 2nd byte */
#define GBK_MARK_LOMAX	0xFE	/* GB 2312 mark   : max - 2nd byte */


#define GBK_MARK1_HIMIN	0xA8	/* GBK mark   : min - 1st byte */
#define GBK_MARK1_HIMAX	0xA9	/* GBK mark   : max - 1st byte */
#define GBK_MARK1_LOMIN	0x40	/* GBK mark   : min - 2nd byte */
#define GBK_MARK1_LOMAX	0xFE	/* GBK mark   : max - 2nd byte */


#define BIG5_CHN_CHAR_NUM	13468

#define BIG5_SYM_HIMIN	0xA0	/* GBK symbol : minmum value of first byte */
#define BIG5_SYM_HIMAX	0xFE	/* GBK symbol : maxmum value of first byte */
#define BIG5_SYM1_LOMIN	0x40	/* GBK symbol : minmum value of second byte */
#define BIG5_SYM1_LOMAX	0x7E	/* GBK symbol : maxmum value of second byte */
#define BIG5_SYM2_LOMIN	0xA1	/* GBK symbol : minmum value of second byte */
#define BIG5_SYM2_LOMAX	0xFE	/* GBK symbol : maxmum value of second byte */

#define BIG5_CHAR1_HIMIN	0xA4	/* BIG5 char   : min - 1st byte */
#define BIG5_CHAR1_HIMAX	0xC6	/* BIG5 char   : max - 1st byte */
#define BIG5_CHAR1_LOMIN	0x40	/* BIG5 char   : min - 2nd byte */
#define BIG5_CHAR1_LOMAX	0x7E	/* BIG5 char   : max - 2nd byte */

#define BIG5_CHAR2_HIMIN	0xC9	/* BIG5 ext char   : min - 1st byte */
#define BIG5_CHAR2_HIMAX	0xF9	/* BIG5 ext char   : max - 1st byte */
#define BIG5_CHAR2_LOMIN	0x40	/* BIG5 ext char   : min - 2nd byte */
#define BIG5_CHAR2_LOMAX	0xFE	/* BIG5 ext char   : max - 2nd byte */

#define BIG5_MARK_HIMIN	0xA1	/* BIG5 mark   : min - 1st byte */
#define BIG5_MARK_HIMAX	0xA3	/* BIG5 mark   : max - 1st byte */
#define BIG5_MARK_LOMIN	0x40	/* BIG5 mark   : min - 2nd byte */
#define BIG5_MARK_LOMAX	0xFE	/* BIG5 mark   : max - 2nd byte */

/* define the range of gb */
#define IN_RANGE(ch, min, max) ( (((unsigned char)(ch))>=min) && (((unsigned char)(ch))<=max) )

#define IS_GBK(cst)		( IN_RANGE((cst)[0], GBK_SYM_HIMIN, GBK_SYM_HIMAX) && IN_RANGE((cst)[1], GBK_SYM_LOMIN, GBK_SYM_LOMAX) )

#define IS_BIG5(cst)	(	\
						(IN_RANGE((cst)[0], BIG5_SYM_HIMIN, BIG5_SYM_HIMAX) \
						&& IN_RANGE((cst)[1], BIG5_SYM1_LOMIN, BIG5_SYM1_LOMAX))	\
						||	\
						(IN_RANGE((cst)[0], BIG5_SYM_HIMIN, BIG5_SYM_HIMAX)	\
						&& IN_RANGE((cst)[1], BIG5_SYM2_LOMIN, BIG5_SYM2_LOMAX))	\
						)

#define IS_GB2312_CHAR(cst)	( IN_RANGE((cst)[0], GB2312_CHAR_HIMIN, GB2312_CHAR_HIMAX) && IN_RANGE((cst)[1], GB2312_CHAR_LOMIN, GB2312_CHAR_LOMAX) )

#define IS_GBK_CHAR1(cst)	( IN_RANGE((cst)[0], GBK_CHAR1_HIMIN, GBK_CHAR1_HIMAX) && IN_RANGE((cst)[1], GBK_CHAR1_LOMIN, GBK_CHAR1_LOMAX) )

#define IS_BIG5_CHAR1(cst)	( IN_RANGE((cst)[0], BIG5_CHAR1_HIMIN, BIG5_CHAR1_HIMAX) && IN_RANGE((cst)[1], BIG5_CHAR1_LOMIN, BIG5_CHAR1_LOMAX) )




#ifdef __cplusplus
}
#endif

bool is_InASCII( unsigned char* word)
{               
	    if(*word < 128) return true;
		    else  return false;
}           
 
#endif