linux利用socket编程来下载网页的html代码

http://hi.baidu.com/zzniu1984/blog/item/f9c45a029dd96a7f3912bbd0.html

linux利用socket编程来下载网页的html代码

2009-05-20 14:53

编译该程序: gcc -g wclient.c -o wclient
运行该程序: ./wclient www.baidu.com 80
抓取百度首页的html代码

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdarg.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netdb.h>

//建立一个http tcp 连接的辅助函数
int htconnect(char *domain, int port)
{
int white_sock;
struct hostent *site;
struct sockaddr_in me;

//获得服务器的名称
site = gethostbyname(domain);
if(NULL == site)
{
    return (-2);
}
//建立套接字
white_sock = socket(AF_INET, SOCK_STREAM, 0);
if(white_sock < 0)
{
    return (-1);
}
//初始化为0
memset(&me, 0, sizeof(struct sockaddr_in));
memcpy(&me.sin_addr, site->h_addr_list[0], site->h_length);
me.sin_family = AF_INET;
me.sin_port = htons(port);

//建立连接
return ( (connect(white_sock, (struct sockaddr*)&me, sizeof(struct sockaddr))<0)?1:white_sock ); 
}

void main(int argc, char** argv)
{
int black_sock;
char bugs_bunny[3];
if(argc < 2)
{
    printf("usage:\nwClient host\n");
    return;
}
black_sock = htconnect(argv[1], 80);

if (black_sock<0)
{
       printf ( "Socket Connect Error!\n") ;
       return ;
}

char *msg = "GET / HTTP/1.1\r\n\r\n";
if( send(black_sock, msg, strlen(msg), 0)<0 )
{
   perror("error in send msg\n");
   exit(1);
}

char buf[1000];
while(1)
{
   while((recv(black_sock,buf,1000,MSG_WAITALL))>0)
   {
   printf("%s",buf);
   }
}
close(black_sock);
}