tse.h

小型搜索引擎,用C/C++编写,属于全文搜索引擎

#ifndef _TSE_H_030719_
#define _TSE_H_030719_

#include <string>

using namespace std;

const string DATA_FILE_NAME("WebData.db");
const string INDEX_FILE_NAME("WebData.idx");

const string DATA_TIANWANG_FILE("Tianwang.raw");

const string VISITED_FILE(".visited.url");
const string UNREACH_HOST("unreachHost.list");
const string VISITED_LINK_FILE(".visitedlink.url");
const string HISTORY_LINK_FILE(".historylink.url");
const string IP_BLOCKED_URL(".ipblocked.url");
const string MD5_FILE(".md5.visitedurl");

const string IP_BLOCK_FILE("ipblock");

const int MAX_UNREACHABLE_HOST_NUM = 100;

/* So we can say strcmp(a, b) == EQ rather than strcmp(a, b) == 0 or
   the really awful !strcmp(a, b). */
const unsigned int NUM_WORKERS		= 50;
const unsigned int NUM_WORKERS_ON_A_SITE = 4;

/*============== Include file ==================*/

#ifndef         _IOSTREAM_
#define         _IOSTREAM_
#include        <iostream>
#endif

#ifndef         _FSTREAM_
#define         _FSTREAM_
#include        <fstream>
#endif

#ifndef         _STDIO_H_
#define         _STDIO_H_
#include        <stdio.h>
#endif

#ifndef         _STDLIB_H_
#define         _STDLIB_H_
#include        <stdlib.h>
#endif

#ifndef         _CSTDLIB_
#define         _CSTDLIB_
#include        <cstdlib>
#endif

#ifndef         _STRING_
#define         _STRING_
#include        <string>
#endif

#ifndef         _CSTRING_
#define         _CSTRING_
#include        <cstring>
#endif

#ifndef         _DIRENT_H_
#define         _DIRENT_H_
#include        <dirent.h>
#endif

#ifndef         _UNISTD_H_
#define         _UNISTD_H_
#include        <unistd.h>
#endif

#ifndef         _S_DIR_H_
#define         _S_DIR_H_
#include        <sys/dir.h>
#endif

#ifndef         _S_TYPES_H_
#define         _S_TYPES_H_
#include        <sys/types.h>
#endif

#ifndef         _S_STAT_H_
#define         _S_STAT_H_
#include        <sys/stat.h>
#endif

#ifndef         _FTW_H_
#define         _FTW_H_
#include        <ftw.h>
#endif

#ifndef         _LIST_
#define         _LIST_
#include        <list>
#endif

#ifndef         _MAP_
#define         _MAP_
#include        <map>
#endif


#ifndef         _ERROR_H_
#define         _ERROR_H_
#include        <error.h>
#endif

#ifndef		_STREAMBUF_
#define		_STREAMBUF_
#include 	<streambuf>
#endif

#ifndef		_OPS_MD5_H_
#define		_OPS_MD5_H_	
#include	 <openssl/md5.h>
#endif

#ifndef		_IOMANIP_
#define		_IOMANIP_	
#include	 <iomanip>
#endif

#ifndef		_TIME_H_
#define		_TIME_H_	
#include	 <time.h>
#endif

#ifndef		_CTIME_
#define		_CTIME_
#include 	<ctime>
#endif

#ifndef		_ALGORITHM_
#define		_ALGORITHM_
#include 	<algorithm>
#endif

#ifndef		_CCTYPE_
#define		_CCTYPE_
#include 	<cctype>
#endif


#ifndef		_VECTOR_
#define		_VECTOR_
#include 	<vector>
#endif

#ifndef		_ITERATOR_
#define		_ITERATOR_
#include	<iterator>
#endif

#ifndef		_LIST_
#define		_LIST_
#include 	<list>
#endif

#ifndef		_DEQUE_
#define		_DEQUE_
#include 	<deque>
#endif

#ifndef		_MAP_
#define		_MAP_
#include 	<map>
#endif

#ifndef		_SET_
#define		_SET_
#include 	<set>
#endif

#ifndef		_CASSERT_
#define		_CASSERT_
#include	<cassert>
#endif

#ifndef		_SIGNAL_H_
#define		_SIGNAL_H_
#include	<signal.h>
#endif

#ifndef		_SOCKET_H_	
#define		_SOCKET_H_
#include	<sys/socket.h>
#endif

#ifndef		_IN_H_
#define		_IN_H_
#include	<netinet/in.h>
#endif

#ifndef		_INET_H_
#define		_INET_H_
#include	<arpa/inet.h>
#endif
/*==========================================================================*/

/*
#ifdef		_DMALLOC_
#include	<dmalloc.h>
#endif		 
*/

/*==========================================================================*/

#endif /* _TSE_H_030719_ */