/// <summary>
/// 初始化
/// </summary>
private void Initialize()
{
//实例化布隆筛选器
BloomVisitedQueue = new BloomVisitedQueue();
//todo 获取config
Config();
InitSeedsUrl();
InitThreads();
}
/// <summary>
/// 初始化
/// </summary>
private void Initialize()
{
//实例化布隆筛选器
BloomVisitedQueue = new BloomVisitedQueue();
//todo 获取config
Config();
InitSeedsUrl();
InitThreads();
}
/// <summary>
/// 线程执行方法
/// </summary>
/// <param name="threadIndex"></param>
protected void CrawlProcess(object threadIndex)
{
var currentThreadIndex = (int)threadIndex;
//Redis工作队列
ITodoQueue <UrlInfo> workQueue = new RedisWorkQueue();
while (true)
{
var info = workQueue.GetNext();
// 根据队列中取出数据是否为null & 空闲线程的数量，判断线程是睡眠还是退出
if (info == null)
{
threadStatus[currentThreadIndex] = true;
if (!threadStatus.Any(t => t == false))
{
Console.WriteLine(currentThreadIndex + "处理完啦");
break;
}
Thread.Sleep(2000);
continue;
}
threadStatus[currentThreadIndex] = false;
HttpWebRequest request = null;
HttpWebResponse response = null;
try
{
if (!BloomVisitedQueue.Contains(info.Key))
{
//todo 检查是否在数据库且有更新
request = WebRequest.Create(info.UrlString) as HttpWebRequest;
ConfigRequest(request);
Console.WriteLine("分析网页");
Console.WriteLine("把超链接加入todo");
BloomVisitedQueue.Add(info.Key);
Console.WriteLine("把本连接加入visited");
}
else
{
Console.WriteLine(info.Key + "...完成过了");
}
}
catch (Exception)
{
//日志
//把弹出的info加入回work队列
workQueue.Add(info);
}
//try
//{
// //// 1~5 秒随机间隔的自动限速
// //if (this.Settings.AutoSpeedLimit)
// //{
// // int span = this.random.Next(1000, 5000);
// // Thread.Sleep(span);
// //}
// // 创建并配置Web请求
// request = WebRequest.Create(urlInfo.UrlString) as HttpWebRequest;
// this.ConfigRequest(request);
// if (request != null)
// {
// response = request.GetResponse() as HttpWebResponse;
// }
// if (response != null)
// {
// this.PersistenceCookie(response);
// Stream stream = null;
// // 如果页面压缩，则解压数据流
// if (response.ContentEncoding == "gzip")
// {
// var responseStream = response.GetResponseStream();
// if (responseStream != null)
// {
// stream = new GZipStream(responseStream, CompressionMode.Decompress);
// }
// }
// else
// {
// stream = response.GetResponseStream();
// }
// using (stream)
// {
// string html = this.ParseContent(stream, response.CharacterSet);
// this.ParseLinks(urlInfo, html);
// if (this.DataReceivedEvent != null)
// {
// this.DataReceivedEvent(
// new DataReceivedEventArgs
// {
// Url = urlInfo.UrlString,
// Depth = urlInfo.Depth,
// Html = html
// });
// }
// if (stream != null)
// {
// stream.Close();
// }
// }
// }
//}
//catch (Exception exception)
//{
// if (this.CrawlErrorEvent != null)
// {
// if (urlInfo != null)
// {
// this.CrawlErrorEvent(
// new CrawlErrorEventArgs {Url = urlInfo.UrlString, Exception = exception});
// }
// }
//}
//finally
//{
// if (request != null)
// {
// request.Abort();
// }
// if (response != null)
// {
// response.Close();
// }
//}
}
}
