1
我目前正在C#中编写一个基本的webcrawler。它是多线程的,但是使用当前的结构,它始终在无限循环中创建线程。无尽循环继续创建新主题
履带:我应该在此代码更改
public static void Start(Uri url)
{
if (!RobotsParser.IsDomainParsed(LinkParser.GetDomainUrl(url)))
{
RobotsParser.Parse(url);
}
if (!CrawlQueue.CrawledList.Contains(url) && RobotsParser.IsCrawlingAllowed(url.ToString()))
{
CrawlQueue.CrawledList.Add(url);
CrawlQueue.QueueList.Remove(url);
//Crawling logic happens here...
CrawlQueue.Add(LinkParser.Find(doc, responseUri));
}
}
}
LinkParser:
public static HashSet<Uri> Find(HtmlDocument doc, string url)
{
//returns list of found urls
return list;
}
队列(这是问题的所在)
internal static class CrawlQueue
{
public static HashSet<Uri> QueueList = new HashSet<Uri>();
public static HashSet<Uri> CrawledList = new HashSet<Uri>();
private static void Start()
{
Parallel.ForEach(QueueList.ToList(), new ParallelOptions { MaxDegreeOfParallelism = 7 }, url =>
{
try { CrawlEngine.Start(url); }
catch (Exception e) { Debug.WriteLine(url + " "+ e.ToString()); }
});
}
public static void Add(HashSet<Uri> list)
{
int counter = 0;
foreach (var site in list.Where(site => !QueueList.Contains(site)))
{
QueueList.Add(site);
counter++;
}
if (counter >= 1)
{
Logging.Log(counter + " items added to crawling queue");
Start();
}
}
}
对于什么样的我应该在设计清单时进行设计吗?
您可能不需要超过单个线程,因为绝大多数工作都是IO绑定。 – Servy
为什么首先使用多线程? – Maarten