欢迎您访问程序员文章站本站旨在为大家提供分享程序员计算机编程知识!
您现在的位置是: 首页

利用socket获取网页内容

程序员文章站 2024-02-21 22:29:28
...
#include        <stdio.h>
#include        <winsock.h>
#include        <string.h>
#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 failed\n");
		return;
	}

	/*
	*        分离url中的主机地址和相对路径
	*/
	strcpy(myurl, url);
	for (pHost = myurl; *pHost != '/' && *pHost != '\0'; ++pHost);
	if ( (int)(pHost - myurl) == strlen(myurl) )
		strcpy(GET, "/");
	else
		strcpy(GET, pHost);
	*pHost = '\0';
	strcpy(host, myurl);
	printf("%s\n%s\n", 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/1.1\r\n");
	strcat(header, "HOST: ");
	strcat(header, host);
	strcat(header, "\r\nConnection: Close\r\n\r\n");

	/*
	*        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, '\0', BUFSIZ);
	}

	closesocket(sockfd);

	WSACleanup();
}

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

转载于:https://www.cnblogs.com/lgh1992314/archive/2013/02/03/5835205.html