C#使用Socket获取网页源代码的实例代码
本文导语: 首先,我们创建一个WebToolkit类: 代码示例: using System; using System.Net.Sockets; using System.Text; namespace ConsoleApplication1 { class WebToolkit { /// /// Url结构 /// struct UrlInfo { public string Host; public int Port; public string File; ...
首先,我们创建一个WebToolkit类:
using System;
using System.Net.Sockets;
using System.Text;
namespace ConsoleApplication1
{
class WebToolkit
{
///
/// Url结构
///
struct UrlInfo
{
public string Host;
public int Port;
public string File;
public string Body;
}
///
/// 解析URL
///
///
///
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;
}
///
/// 发出请求并获取响应
///
///
///
///
///
///
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;
}
///
/// GET请求
///
///
///
///
public static string Get(string url, Encoding encode)
{
UrlInfo urlInfo = ParseURL(url);
string strRequest = string.Format("GET {0}?{1} HTTP/1.1rnHost:{2}:{3}rnConnection:Closernrn", urlInfo.File, urlInfo.Body, urlInfo.Host, urlInfo.Port.ToString());
return GetResponse(urlInfo.Host, urlInfo.Port, strRequest, encode);
}
///
/// POST请求
///
///
///
///
public static string Post(string url, Encoding encode)
{
UrlInfo urlInfo = ParseURL(url);
string strRequest = string.Format("POST {0} HTTP/1.1rnHost:{1}:{2}rnContent-Length:{3}rnContent-Type:application/x-www-form-urlencodedrnConnection:Closernrn{4}", urlInfo.File, urlInfo.Host, urlInfo.Port.ToString(), urlInfo.Body.Length, urlInfo.Body);
return GetResponse(urlInfo.Host, urlInfo.Port, strRequest, encode);
}
}
}
在以下的代码中,使用上面的类,获取网页的源代码。
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./get.asp?keyword=vbscript", Encoding.Default));
Console.ReadKey();
}
}
}