这个是获取源码的代码,至于有什么作用和整理数据在下一篇博客
#define _CRT_SECURE_NO_WARNINGS
#include <iostream>
#include<string>//包含string类
#include<windows.h>
#include<winsock.h>
// Need to link with Ws2_32.lib
#pragma comment(lib, "ws2_32.lib")
using namespace std;
//http://www.imetou.com/nvsheng/257564.html
int main() {
//用户输入一个网址
string str;
cout << "请输入一个网址" << endl;
cin >> str;
cout << "你输入的网址是" <<str<< endl;
//网址解析获得域名和路径
//三段 第一段是http://
//第二段是域名
//第三段是地址
char zhuji[256] = {0};//给每一个字符都赋值为0
char dizhi[256] = {0};
const char* p = strstr(str.c_str(), "http://");
if (NULL == p)
{
cout << "没找到" << endl;
}
else
{
cout << "找到了" << endl;
cout << p << endl;
}
p += 7;//将指针往后移动七位,将指针指向第一个w
sscanf(p, "%[^/]%s", zhuji, dizhi);//通过/来将前后分开(这个是c中的用法)
cout << zhuji << endl;
cout << dizhi << endl;
//1 获取协议版本号
WSADATA wsaData;
WSAStartup(MAKEWORD(2,2), & wsaData);
if (LOBYTE(wsaData.wVersion) != 2 || HIBYTE(wsaData.wVersion) != 2)
{
cout << "请求版本号失败" << endl;
return -1;
}
cout<<"请求版本号成功!"<<endl;
//2 创建socket
SOCKET g_socket = socket(AF_INET, SOCK_STREAM, 0);
if (SOCKET_ERROR == g_socket) {
cout << "创建socket失败!" << endl;
return-1;
}
cout << "创建socket成功!" << endl;
//3 创建协议地址族
SOCKADDR_IN addr = { 0 };
addr.sin_family = AF_INET;//必须和socket函数第一个参数一致
//4 绑定
int r=bind(g_socket,(sockaddr*)&addr, sizeof addr);
if (-1 == r)
{
cout<<"绑定失败!"<<endl;
return -1;
}
printf("绑定成功!\n");
//5 拿到主机ip地址
struct hostent* p1 = gethostbyname(zhuji);//192.168.0.44利用主机获取ip地址
if (NULL == p1)
{
cout << "获取主机地址失败!" << endl;
return -1;
}
cout<<"获取主机地址成功!"<<endl;
memcpy(&addr.sin_addr, p1->h_addr, 4);//把主机地址放入协议地址
addr.sin_port = htons(80);//设置主机端口号 了浏览器
//6 连接服务器
r = connect(g_socket, (sockaddr*)&addr, sizeof addr);
if (-1 == r)
{
cout << "连接服务器失败!"<<endl;
return -1;
}
cout << "连接服务器成功!" << endl;
//7 通信:发送获取源代码请求
//请求信息
string regInfo = "GET " + (string)dizhi + " HTTP/1.1\r\nHost:"+(string)zhuji +"\r\nconnection:close\r\n\rn";
//发送请求信息到服务器
r=send(g_socket,regInfo.c_str(),regInfo.size(),NULL);
if (r > 0)
{
printf("发送请求信息成功!\n");
}
else
{
printf("发送请求信息失败,失败原因:%d\n", WSAGetLastError());
}
//网络连接到域名
//申请获得网页源代码
string allHtml;//村源代码的字符串
char buff[1024];//临时存放
while (1)
{
r = recv(g_socket, buff, 1023, NULL);
if (r > 0)
{
buff[r] = 0;//添加结束符号
allHtml += string(buff);//强转
}
else
{
break;
}
}
cout << "源码如下:" << endl;
cout << allHtml << endl;
cout << "==========================" << endl;
//网页源代码中解析
//网址
//图片链接地址
//循环2-5步
//同时并发的下载图片
return 0;
}