/// <summary>
/// 处理网页
/// </summary>
/// <param name="html"></param>
protected override void Process(string html)
{
try
{
Regex re = new Regex(@"href=(?<web_url>[\s\S]*?)>|href=""(?<web_url>[\s\S]*?)""|href='(?<web_url>[\s\S]*?)'");
MatchCollection mc = re.Matches(html);
foreach (Match m in mc)
{
string url = m.Groups["web_url"].ToString();
url = this.RemoveQuotation(url);
if (url.IndexOf("http://") != -1)
{
UrlQueue.GetInstance().Enqueue(url);
}
}
string title = string.Empty;
re = new Regex(@"<title[\s\S]*?>(?<title>[\s\S]*?)</title>");
Match temp = re.Match(html.ToLower());
title = temp.Groups["title"].ToString();
if (!string.IsNullOrEmpty(title))
{
Console.WriteLine(string.Format("网页标题：{0}", title));
Console.WriteLine(string.Format("网页URL：{0}", this.Url));
}
}
catch
{
}
}
static void Main(string[] args)
{
try
{
Console.Title = System.Configuration.ConfigurationManager.AppSettings["Title"].ToString();
Console.WriteLine("Process is running！");
string url = System.Configuration.ConfigurationManager.AppSettings["URL"].ToString();
UrlQueue.GetInstance().Enqueue(url);
ThreadManager thread = new ThreadManager();
thread.Start();
}
catch (Exception ex)
{
}
Console.ReadLine();
}