private void ExecuteClearCommand()
{
lock (_object)
{
UrlsList.Clear();
}
}
private void ExecuteStopCommand()
{
_tokenSource.Cancel();
foreach (var item in UrlsList.Where(item => item.Status == UrlStatus.Waiting))
{
item.Status = UrlStatus.None;
}
IsEnable = true;
}
public override void Cleanup()
{
_tokenSource.Cancel();
UrlsList.Clear();
lock (_object)
{
_numberList.Clear();
}
base.Cleanup();
}
private void ExecuteClearCompleteCommand()
{
lock (_object)
{
for (var i = UrlsList.Count - 1; i >= 0; i--)
{
if (UrlsList[i].Status == UrlStatus.Completed)
{
UrlsList.RemoveAt(i);
}
}
}
}
/// <summary>
/// 开始下载提取
/// </summary>
/// <param name="item"></param>
private void _begin(UrlItem item)
{
var request = new Request();
var html = request.Get(item.Url);
#region 控制深度
if (item.Depth < Depth || Depth <= 0)
{
var htmlHelper = new Html(html);
var links = htmlHelper.GetLinks(item.Url);
foreach (var newItem in from link in links let newItem = new UrlItem(link, item.Depth + 1, UrlStatus.Waiting) where !UrlsList.Contains <UrlItem>(newItem, new UrlItemComparer()) && (Kind != 1 || htmlHelper.Comparer(item.Url, link)) select newItem)
{
lock (_object)
{
Application.Current.Dispatcher.Invoke(() =>
{
UrlsList.Add(newItem);
});
}
}
}
if (!string.IsNullOrEmpty(_pattern) && Regex.IsMatch(html, _pattern))
{
return;
}
var helper = new NumberHelper();
var lists = helper.GetNumberWithHtml(html);
#endregion
lock (_object)
{
_numberList.AddRange(lists);
_numberList = _numberList.Distinct().ToList();
Application.Current.Dispatcher.Invoke(() =>
{
Message = $"目前已提取{_numberList.Count}个号码！";
});
}
}
private void ExecuteStartCommand()
{
if (UrlsList.Count == 0)
{
UrlsList.Add(new UrlItem(UrlHelper.GetUrl(BaseUrl)));
}
Message = "自动爬虫启动！";
IsEnable = false;
_init();
if (!string.IsNullOrWhiteSpace(Keywords))
{
_pattern = Keywords.Replace(" ", "|");
}
#region 创造主线程，去分配多个下载线程
_tokenSource = new CancellationTokenSource();
var token = _tokenSource.Token;
Task.Factory.StartNew(() =>
{
var index = 0;
while (!token.IsCancellationRequested)
{
#region 创建执行下载的线程数组
var tasksLength = Math.Min(Count, UrlsList.Count - index);
var tasks = new Task[tasksLength];
for (var i = 0; i < tasksLength; i++)
{
var index1 = index;
tasks[i] = new Task(() =>
{
_begin(UrlsList[index1]);
});
index++;
}
#endregion
#region 监视线程数组完成
var continuation = Task.Factory.ContinueWhenAll(tasks, (task) =>
{ }, token);
foreach (var task in tasks)
{
task.Start();
}
while (!continuation.IsCompleted)
{
Thread.Sleep(1000);
}
#endregion
_changedStatus(index - tasksLength, tasksLength, UrlStatus.Completed);
if (index >= UrlsList.Count)
{
_tokenSource.Cancel();
break;
}
}
Application.Current.Dispatcher.Invoke(() =>
{
IsEnable = true;
});
}, token);
#endregion
#region 计时器 做自动导出操作
var token1 = _tokenSource.Token;
Task.Factory.StartNew(() =>
{
var time = Space;
while (!token1.IsCancellationRequested)
{
Thread.Sleep(10000);
time--;
if (time > 0 && !token1.IsCancellationRequested)
{
continue;
}
var file = ExportHelper.ExportRandomName(_numberList, "爬虫导出-");
lock (_object)
{
_numberList.Clear();
Message = "执行自动保存操作，保存到：" + file;
}
time = Space;
}
}, token1);
#endregion
}
