gdbm.c

C编写的用来实现search engine的推荐功能

/****************************************************************************************
* Copyright (c) 2006
* All rights reserved.
*       
* $Id
****************************************************************************************/

#include <gdbm.h>
#include <stdio.h>
#include <fcntl.h>
#include <string.h>

//定义大数组
#define MAXKEYLEN	255
#define MAXVALUELEN	3096

int query_gdbm(GDBM_FILE db,char *key);
int create_gdbm(GDBM_FILE db,char *srcFile);

int main(int argc,char **argv)
{
	
	GDBM_FILE db = NULL;
	char *dbName=NULL;
	char *srcFile =NULL;
	char *use;
	
	int beginSecond;
	int endSecond;
	int totalSecond;
	
	if (argc != 4)
	{
		printf("usage %s -c|-s dbname srcfile\n",argv[0]);
		return 1;
	}
	
	use = argv[1];
	dbName =argv[2];
	
	//db = gdbm_open(dbName, 1024, GDBM_READER |GDBM_WRITER |GDBM_WRCREAT, 0660 ,0);
	
	if (strcmp(use,"-c")==0)
	{
		/*
		*创建数据库，dbName是数据库文件名，1024是从硬盘到内存的块的大小，GDBM_WRITER是写的权限，GDBM_WRCREAT是创建的权限
		*0660是mode,0表示用缺省的函数
		*/
		db = gdbm_open(dbName, 2048, GDBM_WRITER |GDBM_WRCREAT, 0644 ,0);
		
		if (db ==NULL)
		{
			printf("create %s error\n",dbName);
			return 0;
		}
		
		//记录开始的时间，单位为秒
		beginSecond = time((time_t *)NULL);
		srcFile = argv[3];
		
		//调用函数
		create_gdbm(db,srcFile);
		
		//关闭数据库文件
		gdbm_close(db);
		
		//记录结束时间
		endSecond = time((time_t *)NULL);
		
		//得到创建数据库文件所发费的总时间
		totalSecond = endSecond - beginSecond;
		printf("spend total time is :%d seconds\n", totalSecond);
		return 0;
	}
	
	else if (strcmp(use,"-s")==0)
	{
		/*
		*打开数据库，dbName是数据库文件名，1024是从硬盘到内存的块的大小，GDBM_READER是读的权限
		*0660是mode,0表示用缺省的函数
		*/
		db = gdbm_open(dbName, 2048, GDBM_READER , 0400 ,0);
		
		if (db ==NULL)
		{
			printf("open %s error\n",dbName);
			return 1;
		}
		

		//调用查询函数
		query_gdbm(db,argv[3]);
		
		//关闭数据库文件
		gdbm_close(db);
		return 0;
	}
	else
	{
		gdbm_close(db);
		printf("use error\n");
		return 1;
	}
}

//创建gdbm数据库
int create_gdbm(GDBM_FILE db,char *srcFile)
{
	
	datum name;
	datum value;
	datum mytest;
	FILE *fp;
	int ret;
	
	char word[MAXKEYLEN];
	char url[MAXVALUELEN];
	char line[30960];
	char *ptr = NULL;
	char *pptr = NULL;
	fp = fopen(srcFile,"r");
	
	if (fp ==NULL)
	{
		printf(" error open %s\n",srcFile);
		return 1;
	}
	


	while (fgets(line, 3096, fp) != NULL)
	{
		line[3095] = 0;
		memset(word,0,MAXKEYLEN);
		memset(url,0,MAXVALUELEN);
		ptr = strchr(line, '\t');
		if(ptr == NULL) continue;
		*ptr = '\0';
		strncpy(word, line, 255);
		word[254] = '\0';
		pptr = ptr + 1;
		if(pptr == NULL) continue;
		strncpy(url, pptr, 3096);
		url[3095] = 0;
		ptr = strchr(url, '\n');
		if(ptr != NULL) *ptr = '\0';
		if(url[0] == '\0' || strlen(url) <=1) continue;
		//fscanf(fp,"%s\t%s\n", word,url);
		
		//printf("%s\t%s\n", word,url);
		//word[strlen(word)]='\0';
		//url[strlen(url)]='\0';
		
		name.dptr = word;
		name.dsize = strlen(word);
		value.dptr = url;
		value.dsize = strlen(url);

		
		//把结构体name,value的内容插入到数据库文件中去
		ret = gdbm_store(db, name, value, GDBM_INSERT);
		
	}
	
	strcpy(word,"PLASTICSPACK");
	word[strlen(word)]='\0';
	memset(&name,0,sizeof(name));
	name.dptr=word;
	name.dsize=sizeof(word);
	printf("name.dprt=%s\tlen dptr=%d\tname.dsize=%d\n",name.dptr,strlen(name.dptr),name.dsize);
	
	mytest = gdbm_fetch(db, name);

	if (mytest.dptr)
		printf("name %s\tvalue %s\n",name.dptr,mytest.dptr);
	
	
	memset(word,0,MAXKEYLEN);
	strcpy(word,"IMPORTER");
	word[strlen(word)]='\0';
	memset(&name,0,sizeof(name));
	name.dptr=word;
	name.dsize=sizeof(word);
	printf("name.dprt=%s\tlen dptr=%d\tname.dsize=%d\n",name.dptr,strlen(name.dptr),name.dsize);
	
	//根据nam在数据库中查找value的值
	mytest = gdbm_fetch(db, name);

	if (mytest.dptr)
		printf("name %s\tvalue %s\n",name.dptr,mytest.dptr);
	
	fclose(fp);
	return 0;
}

//查询一条记录
int query_gdbm(GDBM_FILE db,char *key)
{	
	datum name;
	datum value;
	char word[MAXKEYLEN];
	

	name.dptr=key;
	name.dsize=strlen(key);

	//根据nam在数据库中查找value的值
	value = gdbm_fetch(db, name);
	
	if (value.dptr !=NULL)
	{
		printf("name %s\tvalue %s\n",name.dptr,value.dptr);
		free(value.dptr);
	}
	
	else
	{
		printf("found not \n");
	}
	
	
	return 0;
}