-
C#教程之C#中使用Socket获取网页源代码的代码
C#使用Socket获取网页源代码的代码,需要的朋友可以参考下。WebToolkit类:
复制代码 代码如下:
using System;
using System.Net.Sockets;
using System.Text;
namespace ConsoleApplication1
{
class WebToolkit
{
/// <summary>
/// Url结构
/// </summary>
struct UrlInfo
{
public string Host;
public int Port;
public string File;
public string Body;
}
/// <summary>
/// 解析URL
/// </summary>
/// <param name="url"></param>
/// <returns></returns>
private static UrlInfo ParseURL(string url)
{
UrlInfo urlInfo = new UrlInfo();
string[] strTemp = null;
urlInfo.Host = "";
urlInfo.Port = 80;
urlInfo.File = "/";
urlInfo.Body = "";
int intIndex = url.ToLower().IndexOf("http://");
if (intIndex != -1)
{
url = url.Substring(7);
intIndex = url.IndexOf("/");
if (intIndex == -1)
{
urlInfo.Host = url;
}
else
{
urlInfo.Host = url.Substring(0, intIndex);
url = url.Substring(intIndex);
intIndex = urlInfo.Host.IndexOf(":");
if (intIndex != -1)
{
strTemp = urlInfo.Host.Split(':');
urlInfo.Host = strTemp[0];
int.TryParse(strTemp[1], out urlInfo.Port);
}
intIndex = url.IndexOf("?");
if (intIndex == -1)
{
urlInfo.File = url;
}
else
{
strTemp = url.Split('?');
urlInfo.File = strTemp[0];
urlInfo.Body = strTemp[1];
}
}
}
return urlInfo;
}
/// <summary>
/// 发出请求并获取响应
/// </summary>
/// <param name="host"></param>
/// <param name="port"></param>
/// <param name="body"></param>
/// <param name="encode"></param>
/// <returns></returns>
private static string GetResponse(string host, int port, string body, Encoding encode)
{
string strResult = string.Empty;
byte[] bteSend = Encoding.ASCII.GetBytes(body);
byte[] bteReceive = new byte[1024];
int intLen = 0;
using (Socket socket = new Socket(AddressFamily.InterNetwork, SocketType.Stream, ProtocolType.Tcp))
{
try
{
socket.Connect(host, port);
if (socket.Connected)
{
socket.Send(bteSend, bteSend.Length, 0);
while ((intLen = socket.Receive(bteReceive, bteReceive.Length, 0)) > 0)
{
strResult += encode.GetString(bteReceive, 0, intLen);
}
}
socket.Close();
}
catch { }
}
return strResult;
}
/// <summary>
/// GET请求
/// </summary>
/// <param name="url"></param>
/// <param name="encode"></param>
/// <returns></returns>
public static string Get(string url, Encoding encode)
{
UrlInfo urlInfo = ParseURL(url);
string strRequest = string.Format("GET {0}?{1} HTTP/1.1\r\nHost:{2}:{3}\r\nConnection:Close\r\n\r\n", urlInfo.File, urlInfo.Body, urlInfo.Host, urlInfo.Port.ToString());
return GetResponse(urlInfo.Host, urlInfo.Port, strRequest, encode);
}
/// <summary>
/// POST请求
/// </summary>
/// <param name="url"></param>
/// <param name="encode"></param>
/// <returns></returns>
public static string Post(string url, Encoding encode)
{
UrlInfo urlInfo = ParseURL(url);
string strRequest = string.Format("POST {0} HTTP/1.1\r\nHost:{1}:{2}\r\nContent-Length:{3}\r\nContent-Type:application/x-www-form-urlencoded\r\nConnection:Close\r\n\r\n{4}", urlInfo.File, urlInfo.Host, urlInfo.Port.ToString(), urlInfo.Body.Length, urlInfo.Body);
return GetResponse(urlInfo.Host, urlInfo.Port, strRequest, encode);
}
}
}
调用示例:
复制代码 代码如下:
using System;
using System.Text;
namespace ConsoleApplication1
{
//调用示例
class Program
{
public static void Main(string[] args)
{
Console.WriteLine(WebToolkit.Get("http://www.jb51.net/t.asp?keyword=vbscript", Encoding.Default));
Console.ReadKey();
}
}
}
最新更新
nodejs爬虫
Python正则表达式完全指南
爬取豆瓣Top250图书数据
shp 地图文件批量添加字段
爬虫小试牛刀(爬取学校通知公告)
【python基础】函数-初识函数
【python基础】函数-返回值
HTTP请求:requests模块基础使用必知必会
Python初学者友好丨详解参数传递类型
如何有效管理爬虫流量?
2个场景实例讲解GaussDB(DWS)基表统计信息估
常用的 SQL Server 关键字及其含义
动手分析SQL Server中的事务中使用的锁
openGauss内核分析:SQL by pass & 经典执行
一招教你如何高效批量导入与更新数据
天天写SQL,这些神奇的特性你知道吗?
openGauss内核分析:执行计划生成
[IM002]Navicat ODBC驱动器管理器 未发现数据
初入Sql Server 之 存储过程的简单使用
SQL Server -- 解决存储过程传入参数作为s
JavaScript判断两个数组相等的四类方法
js如何操作video标签
React实战--利用甘特图和看板,强化Paas平
【记录】正则替换的偏方
前端下载 Blob 类型整理
抽象语法树AST必知必会
关于JS定时器的整理
JS中使用Promise.all控制所有的异步请求都完
js中字符串的方法
import-local执行流程与node模块路径解析流程