bittoken.h

一个很好的html网页分析工具

#ifndef __NETBIT_BITTOKEN_H__
#define __NETBIT_BITTOKEN_H__

//--------------------------------------------
//                 BIT HTML LIB
//
//file:  BitTokent.h
//usage: for token
//model: BitTokenContext
//---------------------------------------------

#include "BitHtmlDtd.h"

typedef struct BitTokenAttr
{
int type;
char *value;	
}BitTokenAttr,*BitPTokenAttr;

typedef struct AttrList
{
 BitTokenAttr *attr;
 struct AttrList *priou;
 struct AttrList *next;
}
BitTokenAttrList,*BitPTokenAttrList;

typedef struct BitToken
{
int type;
char *pData;
int end;
BitTokenAttrList *attrList;
BitTokenAttrList *attrList_tail;
}BitToken,*BitPToken;


typedef struct TokenList
{
 BitToken *token;
 struct TokenList *priou;
 struct TokenList *next;
}
BitTokenList,*BitPTokenList;


typedef struct BitTokenContext
{
char * strBuffer;
int    bufferLength;
int    curPosition;

char * global_strBuffer;
int    global_bufferLength;
int    global_curPosition;

BitTokenList *tokenList;
BitTokenList *tokenList_tail;

BitPTagList	pTagList;

}BitTokenContext,*BitPTokenContext;


//--------------- function declare -----------------

BitTokenContext * Bit_NewContext();
BitToken *Bit_NewToken(int type);
BitTokenAttr *Bit_NewTokenAttribute(int type);

int Bit_Tokenize(BitTokenContext *global_cx);
void Bit_BeginToken(BitTokenContext *global_cx);
int Bit_DestroyToken(BitTokenContext *global_cx);

void Token_EraseSpace(BitTokenContext *global_cx);
char *Token_ReadUntil(BitTokenContext *global_cx,char *sUntil);
void Token_ThrowTo(BitTokenContext *global_cx,char *sUntil);
char *Token_GetAttribute(BitTokenContext *global_cx);
void Token_ConvertRtToSpace(BitTokenContext *global_cx,char * aString);

void Token_ConvertIfNeed(char * aString);

int Bit_GetTagNumber(BitTokenContext *global_cx,char * name_element);
int Bit_GetAtrNumber(BitTokenContext *global_cx,int Type,char * name_element);

int Token_ConsumTag(BitTokenContext *global_cx);
int Token_Consum_PlainText(BitTokenContext *global_cx);

void Bit_ShowTokenResult(BitTokenContext *global_cx);
void Bit_SaveTokenResult(BitTokenContext *global_cx,char * filename);

#endif