protected override dynamic FormateValue(dynamic value)
{
var name = Path.GetFileName(value);
if (name != null)
{
Task <byte[]> task = Client.GetByteArrayAsync(value);
task.ContinueWith(t =>
{
if (t.Exception != null)
{
throw t.Exception;
}
var fileData = t.Result;
string file = Path.Combine(SpiderConsts.GlobalDirectory, "images", name);
if (!File.Exists(file))
{
var stream = BasePipeline.PrepareFile(file).OpenWrite();
foreach (var b in fileData)
{
stream.WriteByte(b);
}
stream.Flush();
stream.Dispose();
}
});
}
return(name);
}
public override string Formate(string value)
{
try
{
var name = Path.GetFileName(value);
if (name != null)
{
var fileData = Client.GetByteArrayAsync(value).Result;
string file = Path.Combine(SpiderEnviroment.GlobalDirectory, "images", name);
if (File.Exists(file))
{
return(value);
}
var stream = BasePipeline.PrepareFile(file).OpenWrite();
foreach (var b in fileData)
{
stream.WriteByte(b);
}
stream.Flush();
stream.Dispose();
}
return(value);
}
catch (Exception e)
{
Logger.SaveLog(LogInfo.Create($"Download file: {value} failed.", Logger.Name, null, LogLevel.Error, e));
throw;
}
}
public void InitComponent()
{
if (_init)
{
return;
}
if (Pipelines == null || Pipelines.Count == 0)
{
throw new SpiderException("Pipelines should not be null.");
}
Scheduler.Init(this);
#if !NET_CORE
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
#else
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppContext.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
#endif
Console.CancelKeyPress += ConsoleCancelKeyPress;
foreach (var pipeline in Pipelines)
{
pipeline.InitPipeline(this);
}
if (Site.StartRequests != null && Site.StartRequests.Count > 0)
{
this.Log($"[步骤 1] 添加链接到调度中心, 数量: {Site.StartRequests.Count}.", LogLevel.Info);
//Logger.SaveLog(LogInfo.Create(, Logger.Name, this, LogLevel.Info));
if ((Scheduler is QueueDuplicateRemovedScheduler) || (Scheduler is PriorityScheduler))
{
Parallel.ForEach(Site.StartRequests, new ParallelOptions()
{
MaxDegreeOfParallelism = 4
}, request =>
{
Scheduler.Push(request);
});
}
else
{
Scheduler.Load(new HashSet <Request>(Site.StartRequests));
ClearStartRequests();
}
}
else
{
this.Log("[步骤 1] 添加链接到调度中心, 数量: 0.", LogLevel.Info);
}
_waitCountLimit = EmptySleepTime / WaitInterval;
if (Site.MinSleepTime > Site.MaxSleepTime)
{
Site.MaxSleepTime = Site.MinSleepTime;
}
_init = true;
}
protected override dynamic FormateValue(dynamic value)
{
var name = Path.GetFileName(value);
if (name != null)
{
Task <byte[]> task = Client.GetByteArrayAsync(value);
task.ContinueWith(t =>
{
if (t.Exception != null)
{
Logger.SaveLog(LogInfo.Create($"下载文件: {value} 失败.", Logger.Name, null, LogLevel.Warn, t.Exception));
return;
}
var fileData = t.Result;
string file = Path.Combine(SpiderEnviroment.GlobalDirectory, "images", name);
if (!File.Exists(file))
{
var stream = BasePipeline.PrepareFile(file).OpenWrite();
foreach (var b in fileData)
{
stream.WriteByte(b);
}
stream.Flush();
stream.Dispose();
}
});
}
return(name);
}
public override void InitPipeline(ISpider spider)
{
base.InitPipeline(spider);
#if !NET_CORE
DataFolder = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, spider.Identity, "mysql");
#else
DataFolder = Path.Combine(AppContext.BaseDirectory, spider.Identity, "mysql");
#endif
Writer = BasePipeline.PrepareFile(Path.Combine(DataFolder, $"{Schema.Database}.{Schema.TableName}.data")).AppendText();
Writer.AutoFlush = true;
}
/// <summary>
/// Create a spider with pageProcessor.
/// </summary>
/// <param name="identity"></param>
/// <param name="pageProcessor"></param>
/// <param name="scheduler"></param>
protected Spider(Site site, string identity, string userid, string taskGroup, IPageProcessor pageProcessor, IScheduler scheduler)
{
if (string.IsNullOrWhiteSpace(identity))
{
Identity = string.IsNullOrEmpty(Site.Domain) ? Guid.NewGuid().ToString() : Site.Domain;
}
else
{
//if (!IdentifyRegex.IsMatch(identity))
//{
// throw new SpiderExceptoin("任务ID不能有空字符.");
//}
Identity = identity;
}
UserId = string.IsNullOrEmpty(userid) ? "DotnetSpider" : userid;
TaskGroup = string.IsNullOrEmpty(taskGroup) ? "DotnetSpider" : taskGroup;
Logger = new Logger(Identity, UserId, TaskGroup);
_waitCount = 0;
if (pageProcessor == null)
{
throw new SpiderException("PageProcessor should not be null.");
}
PageProcessor = pageProcessor;
Site = site;
if (Site == null)
{
throw new SpiderException("Site should not be null.");
}
PageProcessor.Site = site;
StartRequests = Site.StartRequests;
Scheduler = scheduler ?? new QueueDuplicateRemovedScheduler();
Scheduler.Init(this);
#if !NET_CORE
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
#else
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppContext.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
#endif
}
public void InitComponent()
{
if (_init)
{
return;
}
if (Pipelines == null || Pipelines.Count == 0)
{
throw new SpiderException("Pipelines should not be null.");
}
_monitors = IocManager.GetServices <IMonitor>().ToList();
if (_monitors.Count == 0)
{
_monitors = new List <IMonitor> {
new NLogMonitor()
};
}
Scheduler.Init(this);
_monitorTask = Task.Factory.StartNew(() =>
{
var monitor = GetMonitor();
while (!_scheduler.IsExited)
{
ReportStatus();
Thread.Sleep(2000);
}
monitor.IsExited = true;
ReportStatus();
});
#if !NET_CORE
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
#else
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppContext.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
#endif
Console.CancelKeyPress += ConsoleCancelKeyPress;
foreach (var pipeline in Pipelines)
{
pipeline.InitPipeline(this);
}
if (Site.StartRequests != null && Site.StartRequests.Count > 0)
{
this.Log($"准备步骤: 添加链接到调度中心, 数量 {Site.StartRequests.Count}.", LogLevel.Info);
//Logger.SaveLog(LogInfo.Create(, Logger.Name, this, LogLevel.Info));
if ((Scheduler is QueueDuplicateRemovedScheduler) || (Scheduler is PriorityScheduler))
{
Parallel.ForEach(Site.StartRequests, new ParallelOptions()
{
MaxDegreeOfParallelism = 4
}, request =>
{
Scheduler.Push(request);
});
}
else
{
Scheduler.Import(new HashSet <Request>(Site.StartRequests));
ClearStartRequests();
}
}
else
{
this.Log("准备步骤: 添加链接到调度中心, 数量 0.", LogLevel.Info);
}
_waitCountLimit = EmptySleepTime / WaitInterval;
if (Site.MinSleepTime > Site.MaxSleepTime)
{
Site.MaxSleepTime = Site.MinSleepTime;
}
_init = true;
}
protected virtual void InitComponent(params string[] arguments)
{
if (_init)
{
return;
}
this.Log("构建内部模块、准备爬虫数据...", LogLevel.Info);
if (Pipelines == null || Pipelines.Count == 0)
{
throw new SpiderException("Pipelines should not be null.");
}
PreInitComponent(arguments);
_monitor = IocManager.Resolve <IMonitor>() ?? new NLogMonitor();
if (CookieInjector != null)
{
CookieInjector.Inject(this, false);
}
Scheduler.Init(this);
_monitorTask = Task.Factory.StartNew(() =>
{
while (!Monitorable.IsExited)
{
ReportStatus();
Thread.Sleep(2000);
}
ReportStatus();
});
#if !NET_CORE
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
#else
_errorRequestFile = BasePipeline.PrepareFile(Path.Combine(AppContext.BaseDirectory, "ErrorRequests", Identity, "errors.txt"));
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
#endif
Console.CancelKeyPress += ConsoleCancelKeyPress;
foreach (var pipeline in Pipelines)
{
pipeline.InitPipeline(this);
}
if (Site.StartRequests != null && Site.StartRequests.Count > 0)
{
this.Log($"准备步骤: 添加链接到调度中心, 数量 {Site.StartRequests.Count}.", LogLevel.Info);
//Logger.SaveLog(LogInfo.Create(, Logger.Name, this, LogLevel.Info));
if ((Scheduler is QueueDuplicateRemovedScheduler) || (Scheduler is PriorityScheduler))
{
Parallel.ForEach(Site.StartRequests, new ParallelOptions()
{
MaxDegreeOfParallelism = 4
}, request =>
{
Scheduler.Push(request);
});
}
else
{
Scheduler.Import(new HashSet <Request>(Site.StartRequests));
ClearStartRequests();
}
}
else
{
this.Log("准备步骤: 添加链接到调度中心, 数量 0.", LogLevel.Info);
}
_waitCountLimit = EmptySleepTime / WaitInterval;
AfterInitComponent(arguments);
_init = true;
}