public void Run()
{
// Config encoding, header, cookie, proxy etc... 定义采集的 Site 对象, 设置 Header、Cookie、代理等
//var site = new Site { EncodingName = "UTF-8" };
var site = new Site();
// Add start/feed urls. 添加初始采集链接
var context = new StockLearningEntities();
var stocks = context.Stocks.ToList();
foreach (var stock in stocks)
{
string range = "sh";
if (stock.StockId.StartsWith("0") || stock.StockId.StartsWith("3"))
{
range = "sz";
}
site.AddStartUrl($"http://f9.eastmoney.com/{range}{stock.StockId}.html");
}
DotnetSpider.Core.Spider spider = DotnetSpider.Core.Spider.Create(site,
// use memoery queue scheduler. 使用内存调度
new QueueDuplicateRemovedScheduler(),
// use custmize processor for Processor
new StockJJRPageProcessor())
// use custmize pipeline for Pipeline
.AddPipeline(new StockJJRPipeline());
spider.Downloader = new HttpClientDownloader();
spider.ThreadNum = 1;
spider.EmptySleepTime = 3000;
// Start crawler 启动爬虫
spider.Run();
}
public static void Run()
{
DotnetSpider.Core.Spider spider = DotnetSpider.Core.Spider.Create(
// use memoery queue scheduler. 使用内存调度
new QueueDuplicateRemovedScheduler(),
// use custmize processor for youku 为优酷自定义的 Processor
new YoukuPageProcessor())
// use custmize pipeline for youku 为优酷自定义的 Pipeline
.AddPipeline(new YoukuPipeline());
// Start crawler 启动爬虫
spider.EncodingName = "UTF-8";
for (int i = 1; i < 5; ++i)
{
// Add start/feed urls. 添加初始采集链接
spider.AddRequests($"http://list.youku.com/category/show/c_96_s_1_d_1_p_{i}.html");
}
spider.Run();
}
private void CustmizeProcessorAndPipeline()
{
var site = new Site {
EncodingName = "UTF-8", RemoveOutboundLinks = true
};
string url = this.tb_url.Text;
site.AddStartUrl(url);
DotnetSpider.Core.Spider spider = DotnetSpider.Core.Spider.Create(site, new QueueDuplicateRemovedScheduler(), new TouTiaoPageProcessor());
spider.AddPipeline(new TotiaoPipeline());
spider.Downloader = new HttpClientDownloader();
spider.ThreadNum = 1;
spider.EmptySleepTime = 3000;
// Start crawler 启动爬虫
spider.Run();
}
public void Run()
{
// Config encoding, header, cookie, proxy etc... 定义采集的 Site 对象, 设置 Header、Cookie、代理等
//var site = new Site { EncodingName = "UTF-8" };
var site = new Site();
// Add start/feed urls. 添加初始采集链接
site.AddStartUrl("http://quote.eastmoney.com/stocklist.html");
DotnetSpider.Core.Spider spider = DotnetSpider.Core.Spider.Create(site,
// use memoery queue scheduler. 使用内存调度
new QueueDuplicateRemovedScheduler(),
// use custmize processor for Processor
new StockListPageProcessor())
// use custmize pipeline for Pipeline
.AddPipeline(new StockListPipeline());
spider.Downloader = new HttpClientDownloader();
spider.ThreadNum = 1;
spider.EmptySleepTime = 3000;
// Start crawler 启动爬虫
spider.Run();
}
public static void Run(params string[] args)
{
Console.WriteLine("");
Spider.PrintInfo();
Console.WriteLine("");
Console.ForegroundColor = ConsoleColor.Cyan;
var commands = string.Join(" ", args);
Console.WriteLine("Args: " + commands);
Console.WriteLine("");
Console.ForegroundColor = ConsoleColor.White;
#if NET_CORE
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
#endif
Dictionary <string, string> arguments = new Dictionary <string, string>();
foreach (var arg in args)
{
var results = arg.Split(':');
if (results.Length == 2)
{
var key = results[0].Trim();
if (arguments.ContainsKey(key))
{
arguments[key] = results[1].Trim();
}
else
{
arguments.Add(key, results[1].Trim());
}
}
else if (results.Length == 1)
{
var key = results[0].Trim();
if (!arguments.ContainsKey(key))
{
arguments.Add(key, string.Empty);
}
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine("Please use command like: -s:[spider type name] -i:[identity] -a:[arg1,arg2...] -tid:[taskId]");
Console.ForegroundColor = ConsoleColor.White;
return;
}
}
string spiderName;
if (arguments.Count == 0 || !arguments.ContainsKey("-s") || !arguments.ContainsKey("-tid"))
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine("-s or -tid are necessary.");
Console.ForegroundColor = ConsoleColor.White;
return;
}
else
{
spiderName = arguments["-s"];
}
#if NET_CORE
var deps = DependencyContext.Default;
#endif
var spiders = new Dictionary <string, object>();
#if NET_CORE
foreach (var library in deps.CompileLibraries.Where(l => l.Name.ToLower().EndsWith("dotnetspider.sample") || l.Name.ToLower().EndsWith("spiders") || l.Name.ToLower().EndsWith("crawlers")))
{
var asm = Assembly.Load(new AssemblyName(library.Name));
var types = asm.GetTypes();
#else
foreach (var file in DetectDlls())
{
var asm = Assembly.LoadFrom(file);
var types = asm.GetTypes();
#endif
Console.WriteLine($"Fetch assembly: {asm.FullName}.");
foreach (var type in types)
{
bool hasNonParametersConstructor = type.GetConstructors().Any(c => c.IsPublic && c.GetParameters().Length == 0);
if (hasNonParametersConstructor)
{
var interfaces = type.GetInterfaces();
var isNamed = interfaces.Any(t => t.FullName == "DotnetSpider.Core.INamed");
var isIdentity = interfaces.Any(t => t.FullName == "DotnetSpider.Core.IIdentity");
var isRunnable = interfaces.Any(t => t.FullName == "DotnetSpider.Core.IRunable");
if (isNamed && isRunnable && isIdentity)
{
var property = type.GetProperties().First(p => p.Name == "Name");
object runner = Activator.CreateInstance(type);
var name = (string)property.GetValue(runner);
if (!spiders.ContainsKey(name))
{
spiders.Add(name, runner);
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine();
Console.WriteLine($"Spider {name} are duplicate.");
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.White;
return;
}
}
}
}
}
if (spiders.Count == 0)
{
Console.ForegroundColor = ConsoleColor.DarkYellow;
Console.WriteLine();
Console.WriteLine("Did not detect any spider.");
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.White;
return;
}
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.Cyan;
Console.WriteLine($"Detected {spiders.Keys.Count} crawlers.");
Console.ForegroundColor = ConsoleColor.White;
Console.WriteLine();
Console.WriteLine("=================================================================");
Console.WriteLine();
if (!spiders.ContainsKey(spiderName))
{
Console.WriteLine($"There is no spider named: {spiderName}.");
return;
}
var spider = spiders[spiderName];
string identity = "";
if (arguments.ContainsKey("-i"))
{
var property = spider.GetType().GetProperties().First(p => p.Name == "Identity");
identity = arguments["-i"].ToLower();
if (arguments["-i"].ToLower() == "guid")
{
property.SetValue(spider, Guid.NewGuid().ToString("N"));
}
else
{
if (!string.IsNullOrEmpty(identity))
{
property.SetValue(spider, arguments["-i"]);
}
}
}
if (arguments.ContainsKey("-tid"))
{
var property = spider.GetType().GetProperties().First(p => p.Name == "TaskId");
if (arguments["-tid"].ToLower() == "guid")
{
property.SetValue(spider, Guid.NewGuid().ToString("N"));
}
else
{
if (!string.IsNullOrEmpty(identity))
{
property.SetValue(spider, arguments["-tid"]);
}
}
}
var spiderType = spider.GetType();
var method = spiderType.GetMethod("Run");
if (!arguments.ContainsKey("-a"))
{
method.Invoke(spider, new object[] { new string[] { } });
}
else
{
var parameters = arguments["-a"].Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries);
if (parameters.Contains("report"))
{
var emptySleepTime = spiderType.GetProperty("EmptySleepTime");
if (emptySleepTime != null)
{
emptySleepTime.SetValue(spider, 1000);
}
}
method.Invoke(spider, new object[] { parameters });
}
}