C语言获取网页源代码
11月23日更新：
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <winsock2.h>
#pragma comment(lib, "ws2_32.lib")
int geturl(char *url)
{
/****************解析URL，解析资源名、主机名********************/
const char *parseptr1;
const char *parseptr2;
char host[BUFSIZ];
char resource[BUFSIZ];
int len, i;
parseptr2 = url;
parseptr1 = strstr(parseptr2, "http://");
if(NULL == parseptr1) //解析用户输入URL形式如"www.baidu.com" 情况
{
parseptr1 = strchr(parseptr2, '/');
len = parseptr1 - parseptr2;
if(NULL == parseptr1) //解析主机、资源名
{
strcpy(host, parseptr2);
strcpy(resource, "/");
}
else //解析用户输入URL形式如"www.baidu.com/xxx"情况
{
for(i = 0; i < len; i++)
{
host[i] = parseptr2[i];//解析主机名
}
host[len] = '\0';
for(i = 0; i < len; i++)
{
parseptr2++;
}
strcpy(resource, parseptr2);//解析资源名
}
printf("host: %s\nresouce: %s\n", host, resource);
}
else
{
parseptr1 = strchr(parseptr2, ':'); //解析用户输入URL形式如"http://www.baidu.com"情况
len = parseptr1 - parseptr2;
printf("protocol: "); //解析协议
for(i = 0; i < len; i++)
{
printf("%c", parseptr2[i]);
}
for(i = 0; i < 3; i++)
{
parseptr1++;
}
parseptr2 = parseptr1;
parseptr1 = strchr(parseptr2, '/');
len = parseptr1 - parseptr2;
if(NULL == parseptr1)
{
strcpy(host, parseptr2); //解析主机名
strcpy(resource, "/"); //解析资源名
}
else //解析用户输入URL形式如"http://www.baidu.com/xxx.."情况
{
for(i = 0; i < len; i++)
{
host[i] = parseptr2[i];
}
host[len] = '\0';
parseptr2 = parseptr1;
strcpy(resource, parseptr2);
}
printf("\nhost: %s\nresource: %s\n", host, resource);
}
/*****************创建socket************/
//初始化套接字
WSADATA wsaData;
WSAStartup(MAKEWORD(2,2), &wsaData);
//创建套接字
SOCKET sock = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
if(sock == INVALID_SOCKET)
{
printf("Failed socket().\n");
WSACleanup();
return 0;
}
//设置socket参数
struct sockaddr_in sockAddr;
memset(&sockAddr, 0, sizeof(sockAddr));
sockAddr.sin_family = AF_INET;
sockAddr.sin_port = htons(80);
//获取主机名和地址信息
struct hostent *hp = gethostbyname(host);
if(hp == NULL)
{
printf("Can not find host address.\n");
return 0;
}
sockAddr.sin_addr.s_addr = *((unsigned long *)hp->h_addr);
//连接到服务器
if(connect(sock, (SOCKADDR *)&sockAddr, sizeof(sockAddr)) == -1)
{
printf("Failed connect().\n");
WSACleanup();
return 0;
}
/****************与服务器通信，收发数据***************/
//准备发送数据
char request[BUFSIZ] = "";
//request = "GET " + resource + " HTTP/1.1\r\nHost:" + host + "\r\nConnection:Close\r\n\r\n";
strcat(request, "GET ");
strcat(request, resource);
strcat(request, " HTTP/1.1\r\nHost:");
strcat(request, host);
strcat(request, "\r\nConnection:Close\r\n\r\n");
//发送数据
if(SOCKET_ERROR==send(sock, request, sizeof(request), 0))
{
printf("Send error.\n");
closesocket(sock);
return 0;
}
//接收数据
static char pageBuf[BUFSIZ];
printf("Read: ");
int ret = 1;
while(ret > 0)
{
ret = recv(sock, pageBuf, BUFSIZ, 0);
printf("%s", pageBuf);
strnset(pageBuf, '\0', BUFSIZ);
}
closesocket(sock);
WSACleanup();
return 0;
}
int main(int argc, char *argv[])
{
char url[256];
puts("Enter URL:");
scanf("%s", url);
geturl(url);
return 0;
}
11月20日更新下面源代码：
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <winsock2.h>
#pragma comment(lib, "ws2_32.lib")
int geturl(char *url)
{
/***********解析URL，解析出主机名，资源名**********/
char host[BUFSIZ];
char resource[BUFSIZ];
char myurl[BUFSIZ];
char *pHost = 0;
if(strlen(url) > 2000)
{
return 0;
}
//解析出主机和资源名
strcpy(myurl, url);
for(pHost = myurl; *pHost != '/' && *pHost != '\0'; ++pHost);
if((int)(pHost - myurl) == strlen(myurl))
{
strcpy(resource, "/");
}
else
{
strcpy(resource, pHost);
}
*pHost = '\0';
strcpy(host, myurl);
printf("%s\n%s\n", host, resource);
/*****************创建socket************/
//初始化套接字
WSADATA wsaData;
WSAStartup(MAKEWORD(2,2), &wsaData);
//创建套接字
SOCKET sock = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
if(sock == INVALID_SOCKET)
{
printf("Failed socket().\n");
WSACleanup();
return 0;
}
//设置socket参数
struct sockaddr_in sockAddr;
memset(&sockAddr, 0, sizeof(sockAddr));
sockAddr.sin_family = AF_INET;
sockAddr.sin_port = htons(80);
//获取主机名和地址信息
struct hostent *hp = gethostbyname(host);
if(hp == NULL)
{
printf("Can not find host address.\n");
return 0;
}
sockAddr.sin_addr.s_addr = *((unsigned long *)hp->h_addr);
//连接到服务器
if(connect(sock, (SOCKADDR *)&sockAddr, sizeof(sockAddr)) == -1)
{
printf("Failed connect().\n");
WSACleanup();
return 0;
}
/****************与服务器通信，收发数据***************/
//准备发送数据
char request[BUFSIZ] = "";
//request = "GET " + resource + " HTTP/1.1\r\nHost:" + host + "\r\nConnection:Close\r\n\r\n";
strcat(request, "GET ");
strcat(request, resource);
strcat(request, " HTTP/1.1\r\nHost:");
strcat(request, host);
strcat(request, "\r\nConnection:Close\r\n\r\n");
//发送数据
if(SOCKET_ERROR==send(sock, request, sizeof(request), 0))
{
printf("Send error.\n");
closesocket(sock);
return 0;
}
//接收数据
static char pageBuf[BUFSIZ];
printf("Read: ");
int ret = 1;
while(ret > 0)
{
ret = recv(sock, pageBuf, BUFSIZ, 0);
printf("%s", pageBuf);
strnset(pageBuf, '\0', BUFSIZ);
}
closesocket(sock);
WSACleanup();
return 0;
}
int main(void)
{
char url[256];
printf("http://");
scanf("%s", url);
geturl(url);
return 0;
}