geturlterms_test.c

C编写的用来实现search engine的推荐功能

/**
 * testing program to extract terms from urls
 * zhiyong, zhang, March, 1st, 2006
 **/
#include <stdio.h>
#include "getUrlTermFreqs.h"

int main(int argc, char** argv)
{
	int i,iRet;
	PAGE_TERMS terms1, terms2;
	if(argc < 4)
	{
		fprintf(stdout,"Usage:%s <url1> <url2> <gdbm>\n", argv[0]);
		return 0;
	}
	bzero(&terms1, sizeof(terms1));
	bzero(&terms2, sizeof(terms2));
	iRet = getPageTerms(argv[1], &terms1, argv[3]);
	if(iRet < 0)
	{
		fprintf(stderr, "err getPageTerms \n");
		return -1;
	}
	fprintf(stdout, "url1 term number = %d\n", terms1.termNum);
	for(i=0;i<terms1.termNum;i++)
	{
		fprintf(stdout, "%s(%d) \t", 
			terms1.terms[i].term, terms1.terms[i].freq);
	}
	fprintf(stdout, "\n"); 
	
	iRet = getPageTerms(argv[2], &terms2, argv[3]);
	if(iRet < 0)
	{
		fprintf(stderr, "err getPageTerms \n");
		return -1;
	}
	fprintf(stdout, "url2 term2 number = %d\n", terms2.termNum);
	for(i=0;i<terms2.termNum;i++)
	{
		fprintf(stdout, "%s(%d) \t", 
			terms2.terms[i].term, terms2.terms[i].freq);
	}
	fprintf(stdout, "\n");

	iRet = mergePageTerms(&terms1, &terms2);
	if(iRet < 0)
	{
		fprintf(stderr, "err getTermsFromURLS \n");
		return -1;
	}
	fprintf(stdout, "merge result term number = %d\n", terms1.termNum);
	for(i=0;i<terms1.termNum;i++)
	{
		fprintf(stdout, "%s(%d) \t", 
			terms1.terms[i].term, terms1.terms[i].freq);
	}
	fprintf(stdout, "\n");
	return 0;
	
}