session.c

C编写的用来实现search engine的推荐功能

/**
 * session.c, program to get sessions
 * zhiyong zhang
 * louisville May 23th
 **/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <gdbm.h>

#include "recdef.h"

static char session_gdbm[256] = "/home/zhiyong/software/apache2/cgi-bin/session.gdbm";
static int compareWeight(const void* wt1, const void* wt2);
static int compareUrl(const void* url1, const void* url2);

void myStrToLower(unsigned char* str)
{
	for(;*str;str++)
	{
		if(*str>='A' && *str<='Z')
			*str = *str - ('A'-'a');
	}		
}

//urls must be preallocated to at least MAX_LINKS_PER_SESSION
int getSessionUrlsFromGdbm(char* sessionID, SESSION_LINKS* links)
{
	int i,iRet;
	char sessions[2048];
	char temp[2048];
	char* ptr = NULL;
	char* pptr = NULL;
	GDBM_FILE db = NULL;
	datum name;
	datum value;
	db = gdbm_open(session_gdbm, 2048, GDBM_READER, 0644 ,0);
	if (db ==NULL)
	{
		fprintf(stderr, "open %s for read error\n",session_gdbm);
		return -1;
	}
	
	name.dptr = sessionID;
	name.dsize = strlen(sessionID);
	iRet = gdbm_exists(db, name);
	if(iRet != 0) 
	{
		value = gdbm_fetch(db, name);
		if(value.dptr != NULL)
		{
			strcpy(sessions, value.dptr);
			free(value.dptr);
			value.dptr = NULL;
			sessions[strlen(sessions)] = '\0';
		}
		else
		{
			fprintf(stderr, "error fetching gdbm lib file");
			gdbm_close(db);
			return -1;
		}
	}
	else
	{
		gdbm_close(db);
		return 0;
	}
	
	gdbm_close(db);
	strcpy(temp, sessions);
	pptr = temp;
	i = 0;
	while((ptr = strchr(pptr, ';')) != NULL)
	{
		*ptr = '\0';
		strcpy(links->urls[i].url, pptr);
		if(strlen(links->urls[i].url) > 1 && links->urls[i].url[strlen(links->urls[i].url)-1] == '/')
		{
			links->urls[i].url[strlen(links->urls[i].url)-1] = '\0';
		}
		i++;
		links->num++;
		pptr = ptr+1;
		if((i>=MAX_LINKS_PER_SESSION) || (*pptr == '\0') || (pptr == NULL)) break;
	}
	return i;
}

static int compareUrl(const void* url1, const void* url2)
{
	REC_URL* rec_url1 = (REC_URL*)url1;
	REC_URL* rec_url2 = (REC_URL*)url2;
	if (strcmp(rec_url1->url, rec_url2->url) < 0)
	{
		return -1;
	}
	else if (strcmp(rec_url1->url, rec_url2->url) > 0)
	{
		return 1;
	}
	else
	{
		return 0;
	}
	
}

static int compareWeight(const void* wt1, const void* wt2)
{
	REC_URL* weight1 = (REC_URL*)wt1;
	REC_URL* weight2 = (REC_URL*)wt2;
	if (weight1->wt > weight2->wt)
	{
		return -1;
	}
	else if (weight1->wt < weight2->wt)
	{
		return 1;
	}
	else
	{
		return 0;
	}
	
}
/**
 * merge recommendation url links and output result in weight order
 **/
int mergeRecLinks(REC_LINKS *rec1, REC_LINKS *rec2)
{
	int i=0,j=0;
	int k=0;
	int i2,j2;
	REC_LINKS rec_merge;
	if(rec1->num == 0)
	{
		*rec1 = *rec2;
		return rec1->num;
	}
	bzero(&rec_merge, sizeof(REC_LINKS));
	qsort(rec1->urls, rec1->num, sizeof(REC_URL), compareUrl);
	qsort(rec2->urls, rec2->num, sizeof(REC_URL), compareUrl);
	while(i<rec1->num && j<rec2->num)
	{
			if(strcmp(rec2->urls[j].url, rec1->urls[i].url) < 0)
			{
				//only occurs in urls2
				strcpy(rec_merge.urls[k].url, rec2->urls[j].url);
				rec_merge.urls[k++].wt = rec2->urls[j].wt;
				rec_merge.num++;
				j++;
			}
			else if(strcmp(rec2->urls[j].url, rec1->urls[i].url) > 0)
			{
				//only occurs in urls1
				strcpy(rec_merge.urls[k].url, rec1->urls[i].url);
				rec_merge.urls[k++].wt = rec1->urls[i].wt;
				rec_merge.num++;
				i++;
			}
			else
			{
				//same url occurs in both urls1 and urls2
				strcpy(rec_merge.urls[k].url, rec1->urls[i].url);
				rec_merge.urls[k++].wt = rec1->urls[i].wt + rec2->urls[j].wt;
				rec_merge.num++;
				j++;
				i++;
			}	
		
		if(rec_merge.num >= MAX_RECS_PER_VISIT) break;
	}
	if(rec_merge.num < MAX_RECS_PER_VISIT)
	{
		if(j>=rec2->num && i<rec1->num)
		{
			for(i2=0;i2<rec1->num-i;i2++)
			{
				strcpy(rec_merge.urls[k].url, rec1->urls[i+i2].url);
				rec_merge.urls[k++].wt = rec1->urls[i+i2].wt;
				rec_merge.num++;
			}
						
		}
		else if(i>=rec1->num && j<rec2->num)
		{
			for(j2=0;j2<rec2->num-j;j2++)
			{
				strcpy(rec_merge.urls[k].url, rec2->urls[j+j2].url);
				rec_merge.urls[k++].wt = rec2->urls[j+j2].wt;
				rec_merge.num++;
			}
		}
	}
	qsort(rec_merge.urls, rec_merge.num, sizeof(REC_URL), compareWeight);
	*rec1 = rec_merge;
	return rec_merge.num;
}