Linux下C语言用socket获取网页源码

#include
#include
#include
#pragma comment(lib, “ws2_32.lib”)

void geturl(char *url)
{
WSADATA WSAData={0};
SOCKET sockfd;
struct sockaddr_in addr;
struct hostent *pURL;
char myurl[BUFSIZ];
char *pHost = 0, *pGET = 0;
char host[BUFSIZ], GET[BUFSIZ];
char header[BUFSIZ] = “”;
static char text[BUFSIZ];
int i;

/*
* windows下使用socket必须用WSAStartup初始化，否则不能调用
*/
if(WSAStartup(MAKEWORD(2,2), &WSAData))
{
printf(“WSA failedn”);
return;
}

/*
* 分离url中的主机地址和相对路径
*/
strcpy(myurl, url);
for (pHost = myurl; *pHost != ‘/’ && *pHost != ‘’; ++pHost);
if ( (int)(pHost – myurl) == strlen(myurl) )
strcpy(GET, “/”);
else
strcpy(GET, pHost);
*pHost = ‘’;
strcpy(host, myurl);
printf(“%sn%sn”, host, GET);

/*
* 设定socket参数,并未真正初始化
*/
sockfd = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP);
pURL = gethostbyname(host);
addr.sin_family = AF_INET;
addr.sin_addr.s_addr = *((unsigned long*)pURL->h_addr);
addr.sin_port = htons(80);

/*
* 组织发送到web服务器的信息
* 为何要发送下面的信息请参考HTTP协议的约定
*/
strcat(header, “GET “);
strcat(header, GET);
strcat(header, ” HTTP/1.1rn”);
strcat(header, “HOST: “);
strcat(header, host);
strcat(header, “rnConnection: Closernrn”);

/*
* 连接到服务器，发送请求header，并接受反馈（即网页源代码）
*/
connect(sockfd,(SOCKADDR *)&addr,sizeof(addr));

send(sockfd, header, strlen(header), 0);

while ( recv(sockfd, text, BUFSIZ, 0) > 0)
{
printf(“%s”, text);
strnset(text, ‘’, BUFSIZ);
}

closesocket(sockfd);

WSACleanup();
}

int main()
{
char url[256];
printf(“http://”);
scanf(“%s”, url);
geturl(url);
return 0;
}