public static void Run(string loaction)
{
Downloader.GetInstance().Start();
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddThroughMessageQueue();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "网易云音乐"; // 设置任务名称
spider.Speed = 10; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 5; // 设置采集深度
spider.AddDataFlow(new MusicListDataParser());
//spider.AddRequests("https://music.163.com/#/playlist?id=2964757969"); // 设置起始链接
spider.AddRequests(loaction);
spider.RunAsync(); // 启动
}
public static Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddKafka();
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 1; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
spider.AddDataFlow(new DataParser <EntitySpider.CnblogsEntry>())
.AddDataFlow(spider.GetDefaultStorage());
spider.AddRequests(
new Request("https://news.cnblogs.com/n/page/1/", new Dictionary <string, string> {
{ "网站", "博客园" }
}),
new Request("https://news.cnblogs.com/n/page/2/", new Dictionary <string, string> {
{ "网站", "博客园" }
}));
return(spider.RunAsync()); // 启动
}
public static Task Run2()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
}).Register <EntitySpider>();
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 1; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
var options = provider.GetRequiredService <SpiderOptions>();
spider.AddDataFlow(new CnblogsDataParser()).AddDataFlow(new MongoEntityStorage(options.StorageConnectionString));
spider.AddRequests("http://www.cnblogs.com/"); // 设置起始链接
return(spider.RunAsync()); // 启动
}
public static void Run1()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 1; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
spider.AddDataFlow(new DataParser
{
SelectableFactory = context => context.GetSelectable(ContentType.Html),
Required = DataParserHelper.CheckIfRequiredByRegex("cnblogs\\.com"),
GetFollowRequests = DataParserHelper.QueryFollowRequestsByXPath(".")
}).AddDataFlow(new ConsoleStorage()); // 控制台打印采集结果
spider.AddRequests("http://www.cnblogs.com/"); // 设置起始链接
spider.RunAsync(); // 启动
}
public static void Run()
{
ImageDownloader.GetInstance().Start();
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddThroughMessageQueue();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
}).Register <EntitySpider>();
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "宅男女神图片采集"; // 设置任务名称
spider.Speed = 2; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 5; // 设置采集深度
//spider.AddDataFlow(new NvshensTagIndexDataParser());
spider.AddDataFlow(new NvshensFirstPageTagDataParser());
spider.AddDataFlow(new NvshensPageTagDataParser());
spider.AddDataFlow(new NvshensFirstPageDetailDataParser());
spider.AddDataFlow(new NvshensPageDetailDataParser());
//spider.AddRequests("https://www.nvshens.com/gallery/"); // 设置起始链接
spider.AddRequests("https://www.nvshens.com/gallery/luoli/"); // 设置起始链接
spider.RunAsync(); // 启动
}
public static Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalMessageQueue();
services.AddLocalDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddLocalDownloadCenter();
services.AddSpiderStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.NewGuidId(); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 1; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
spider.DownloaderSettings.Type = DownloaderType.HttpClient; // 使用普通下载器, 无关 Cookie, 干净的 HttpClient
spider.AddDataFlow(new CnblogsDataParser()).AddDataFlow(new ConsoleStorage());
spider.AddRequests(new Request("http://www.cnblogs.com/", new Dictionary <string, string>
{
{ "key1", "value1" }
})); // 设置起始链接
return(spider.RunAsync()); // 启动
}
public static void StartWithHost(string[] args)
{
var configure = new LoggerConfiguration()
#if DEBUG
.MinimumLevel.Verbose()
#else
.MinimumLevel.Information()
#endif
.MinimumLevel.Override("Microsoft", LogEventLevel.Warning)
.Enrich.FromLogContext()
.WriteTo.Console().WriteTo
.RollingFile("dotnet-spider.log");
Log.Logger = configure.CreateLogger();
var hostBuilder = new SpiderHostBuilder()
.ConfigureAppConfiguration(x =>
{
if (File.Exists("appsettings.json"))
{
x.AddJsonFile("appsettings.json");
}
x.AddCommandLine(args);
//x.AddEnvironmentVariables();
})
.ConfigureLogging(x => { x.AddSerilog(); })
.ConfigureServices((services) =>
{
services.AddLocalEventBus();
services.AddSingleton <IScheduler>(new MyScheduler());
services.AddLocalDownloadCenter();
services.AddDownloaderAgent((x) =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter((x) =>
{
x.UseMemory();
});
});
hostBuilder.Register <IndexSpider>();
hostBuilder.Register <InfoSpider>();
var host = hostBuilder.Build();
host.Start();
var spider1 = host.Create <IndexSpider>();
Task task = spider1.RunAsync();
task.ContinueWith((t) =>
{
var spider2 = host.Create <InfoSpider>();
spider2.RunAsync(args);
});
}
protected override void ConfigureService(IConfiguration configuration, SpiderHostBuilder builder)
{
builder.ConfigureLogging(b =>
{
#if DEBUG
b.SetMinimumLevel(LogLevel.Debug);
#else
b.SetMinimumLevel(LogLevel.Information);
#endif
b.AddSerilog();
});
var config = configuration["DOTNET_SPIDER_CONFIG"];
builder.ConfigureAppConfiguration(x =>
{
if (!string.IsNullOrWhiteSpace(config) && File.Exists(config))
{
// 添加 JSON 配置文件
x.AddJsonFile(config);
}
else
{
if (File.Exists("appsettings.json"))
{
x.AddJsonFile("appsettings.json");
}
}
x.AddCommandLine(Environment.GetCommandLineArgs(), Framework.SwitchMappings);
x.AddEnvironmentVariables();
});
var distributed = configuration["DOTNET_SPIDER_DISTRIBUTED"] == "false";
builder.ConfigureServices(services =>
{
if (distributed)
{
services.AddLocalMessageQueue();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x =>
{
// 添加内存统计服务
x.UseMemory();
});
}
else
{
services.AddKafka();
}
});
}
protected override void ConfigureService(IConfiguration configuration, SpiderHostBuilder builder)
{
builder.ConfigureLogging(b =>
{
#if DEBUG
b.SetMinimumLevel(LogLevel.Debug);
#else
b.SetMinimumLevel(LogLevel.Information);
#endif
b.AddSerilog();
});
var config = configuration["config"];
builder.ConfigureAppConfiguration(b =>
{
if (!string.IsNullOrWhiteSpace(config) && File.Exists(config))
{
// 添加 JSON 配置文件
b.AddJsonFile(config);
}
else
{
b.AddJsonFile("appsettings.json");
}
b.AddCommandLine(Environment.GetCommandLineArgs(), Framework.SwitchMappings);
b.AddEnvironmentVariables();
});
var local = configuration["local"] == "true";
builder.ConfigureServices(services =>
{
if (local)
{
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x =>
{
// 添加内存统计服务
x.UseMemory();
});
}
else
{
services.AddKafkaEventBus();
}
});
}
public void RetryWhenResultIsEmpty()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.NewGuidId();
spider.Name = "RetryWhenResultIsEmpty";
spider.EmptySleepTime = 15;
spider.RetryWhenResultIsEmpty = true;
spider.Scheduler = new QueueDistinctBfsScheduler();
spider.AddRequests(new Request("http://www.RetryWhenResultIsEmpty.com")
{
DownloaderType = DownloaderType.Empty,
RetryTimes = 5
});
spider.RunAsync().Wait();
var statisticsStore = provider.GetRequiredService <IStatisticsStore>();
var s = statisticsStore.GetSpiderStatisticsAsync(spider.Id).Result;
var dss = statisticsStore.GetDownloadStatisticsListAsync(1, 10).Result;
while (dss.Count == 0)
{
Thread.Sleep(1000);
}
var ds = dss[0];
Assert.Equal(1, s.Total);
Assert.Equal(1, s.Failed);
Assert.Equal(0, s.Success);
Assert.Equal(0, ds.Failed);
Assert.Equal(6, ds.Success);
}
static void Main(string[] args)
{
ImageDownloader.GetInstance().Start();
// var configure = new LoggerConfiguration()
//#if DEBUG
// .MinimumLevel.Verbose()
//#else
// .MinimumLevel.Information()
//#endif
// .MinimumLevel.Override("Microsoft", LogEventLevel.Warning)
// .Enrich.FromLogContext()
// .WriteTo.Console().WriteTo
// .RollingFile("dotnet-spider.log");
// Log.Logger = configure.CreateLogger();
//Startup.Execute<SsqSpider>(args);
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
//services.AddKafkaEventBus();
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.Id = Guid.NewGuid().ToString("N"); // 设置任务标识
spider.Name = "优美图片采集"; // 设置任务名称
spider.Speed = 2; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 5; // 设置采集深度
spider.AddDataFlow(new YouMeiSpider());
spider.AddDataFlow(new YouMeiDetailSpider());
//spider.AddDataFlow(new NvshensPageTagDataParser());
//spider.AddDataFlow(new NvshensFirstPageDetailDataParser());
//spider.AddDataFlow(new NvshensPageDetailDataParser());
//spider.AddRequests("https://www.nvshens.com/gallery/"); // 设置起始链接
spider.AddRequests("http://www.umei.cc/p/gaoqing/cn/1.htm"); // 设置起始链接
spider.RunAsync(); // 启动
// await DistributedSpider.Run();
Console.Read();
}
public static async Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddKafkaEventBus();
services.AddDownloadCenter(x => x.UseLocalDownloaderAgentStore());
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var bus = provider.GetRequiredService <IEventBus>();
bus.Subscribe("test-topic", evt => { Console.WriteLine("i am consumer 1"); });
bus.Subscribe("test-topic", evt => { Console.WriteLine("i am consumer 2"); });
for (int i = 0; i < 100; ++i)
{
await bus.PublishAsync("test-topic", new Event());
}
Console.Read();
var spider = provider.Create <Spider>();
spider.NewGuidId(); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 10; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
spider.AddDataFlow(new CnblogsDataParser()).AddDataFlow(new ConsoleStorage());
spider.AddRequests(new Request("http://www.cnblogs.com/", new Dictionary <string, string>
{
{ "key1", "value1" }
})); // 设置起始链接
await spider.RunAsync(); // 启动
}
protected SpiderHostBuilder GetLocalSpiderHostBuilder()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddThroughMessageQueue();
services.AddLocalDownloadCenter();
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
return(builder);
}
public SpiderStart()
{
_hostBuilder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => {
x.AddJsonFile("appsettings.json");
})
.ConfigureServices(services =>
{
services.AddLocalMessageQueue();
services.AddDownloadCenter(x => x.UseLocalDownloaderAgentStore());
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
_provider = _hostBuilder.Build();
}
/// <summary>
/// 运行爬虫
/// </summary>
/// <returns></returns>
public static Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloadAgent(x =>
{
x.UseFileLocker();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseRedis());
}).Register <PCNewsSpider>();
var provider = builder.Build();
var spider = provider.Create <PCNewsSpider>();
return(spider.RunAsync());
}
public static Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalMessageQueue();
services.AddLocalDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddLocalDownloadCenter();
services.AddSpiderStatisticsCenter(x => x.UseMemory());
}).Register<EntitySpider>();
var provider = builder.Build();
var spider = provider.Create<EntitySpider>();
return spider.RunAsync();
}
/// <summary>
/// 运行爬虫
/// </summary>
/// <returns></returns>
public static Task Run()
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
//这里也使用本地事件注入，目前不需要分布式消息队列Kafka
services.AddLocalEventBus();
services.AddLocalDownloadCenter();
services.AddDownloadAgent(x =>
{
x.UseFileLocker();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
}).Register <WeiXinSoGouSpider>();
var provider = builder.Build();
var spider = provider.Create <WeiXinSoGouSpider>();
return(spider.RunAsync());
}
public static async Task Run()
{
try
{
var builder = new SpiderHostBuilder()
.ConfigureLogging(x => x.AddSerilog())
.ConfigureAppConfiguration(x => x.AddJsonFile("appsettings.json"))
.ConfigureServices(services =>
{
services.AddLocalEventBus();
services.AddDownloadCenter(x => x.UseLocalDownloaderAgentStore());
services.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
services.AddStatisticsCenter(x => x.UseMemory());
});
var provider = builder.Build();
var spider = provider.Create <Spider>();
spider.NewGuidId(); // 设置任务标识
spider.Name = "博客园全站采集"; // 设置任务名称
spider.Speed = 10; // 设置采集速度, 表示每秒下载多少个请求, 大于 1 时越大速度越快, 小于 1 时越小越慢, 不能为0.
spider.Depth = 3; // 设置采集深度
spider.AddDataFlow(new CnblogsDataParser()).AddDataFlow(new ConsoleStorage());
spider.AddRequests(new Request("http://www.cnblogs.com/", new Dictionary <string, string>
{
{ "key1", "value1" }
})); // 设置起始链接
await spider.RunAsync(); // 启动
}
catch (Exception ex)
{
Console.WriteLine(ex.ToString());
}
}
/// <summary>
/// 运行
/// </summary>
/// <param name="args">运行参数</param>
public static void Execute(params string[] args)
{
ConfigureSerialLog();
Framework.SetEncoding();
Framework.SetMultiThread();
var configurationBuilder = new ConfigurationBuilder();
configurationBuilder.SetBasePath(AppDomain.CurrentDomain.BaseDirectory);
configurationBuilder.AddEnvironmentVariables();
configurationBuilder.AddCommandLine(Environment.GetCommandLineArgs(), Framework.SwitchMappings);
var configuration = configurationBuilder.Build();
string spiderTypeName = configuration["type"];
if (string.IsNullOrWhiteSpace(spiderTypeName))
{
Log.Logger.Error("未指定需要执行的爬虫类型");
return;
}
var name = configuration["name"];
var id = configuration["id"] ?? Guid.NewGuid().ToString("N");
var config = configuration["config"];
var arguments = configuration["args"]?.Split(' ');
var local = configuration["local"] == "true";
PrintEnvironment(args);
var spiderTypes = DetectSpiders();
if (spiderTypes == null || spiderTypes.Count == 0)
{
return;
}
var spiderType = spiderTypes.FirstOrDefault(x => x.Name.ToLower() == spiderTypeName.ToLower());
if (spiderType == null)
{
Log.Logger.Error($"未找到爬虫: {spiderTypeName}", 0, ConsoleColor.DarkYellow);
return;
}
var builder = new SpiderHostBuilder();
builder.ConfigureLogging(b =>
{
#if DEBUG
b.SetMinimumLevel(LogLevel.Debug);
#else
b.SetMinimumLevel(LogLevel.Information);
#endif
b.AddSerilog();
});
builder.ConfigureAppConfiguration(b =>
{
// 添加 JSON 配置文件
b.AddJsonFile(config);
b.AddCommandLine(args);
b.AddEnvironmentVariables();
});
if (local)
{
builder.ConfigureServices(b =>
{
b.AddLocalEventBus();
b.AddLocalDownloadCenter();
b.AddDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
b.AddStatisticsCenter(x =>
{
// 添加内存统计服务
x.UseMemory();
});
});
}
else
{
builder.ConfigureServices(b => { b.AddKafkaEventBus(); });
}
builder.Register(spiderType);
var provider = builder.Build();
var instance = provider.Create(spiderType);
if (instance != null)
{
instance.Name = name;
instance.Id = id;
instance.RunAsync(arguments).ConfigureAwait(false).GetAwaiter();
}
else
{
Log.Logger.Error("创建爬虫对象失败", 0, ConsoleColor.DarkYellow);
}
}
/// <summary>
/// 运行
/// </summary>
/// <param name="args">运行参数</param>
public static void Run(params string[] args)
{
Framework.SetEncoding();
var configurationBuilder = Framework.CreateConfigurationBuilder(null, args);
var configuration = configurationBuilder.Build();
string spider = configuration["spider"];
if (string.IsNullOrWhiteSpace(spider))
{
throw new SpiderException("未指定需要执行的爬虫");
}
var name = configuration["name"];
var id = configuration["id"] ?? Guid.NewGuid().ToString("N");
var config = configuration["config"];
var arguments = configuration["args"]?.Split(' ');
var distribute = configuration["distribute"] == "true";
PrintEnvironment(args);
var spiderTypes = DetectSpiders();
if (spiderTypes == null || spiderTypes.Count == 0)
{
return;
}
var spiderType = spiderTypes.FirstOrDefault(x => x.Name.ToLower() == spider.ToLower());
if (spiderType == null)
{
ConsoleHelper.WriteLine($"未找到爬虫: {spider}", 0, ConsoleColor.DarkYellow);
return;
}
var builder = new SpiderHostBuilder();
builder.ConfigureLogging(b =>
{
#if DEBUG
b.SetMinimumLevel(LogLevel.Debug);
#else
b.SetMinimumLevel(LogLevel.Information);
#endif
b.AddSerilog();
});
builder.ConfigureAppConfiguration(b =>
{
// 添加 JSON 配置文件
b.AddJsonFile(config);
b.AddCommandLine(args);
});
if (!distribute)
{
builder.ConfigureServices(b =>
{
b.AddLocalMessageQueue();
b.AddLocalDownloaderAgent(x =>
{
x.UseFileLocker();
x.UseDefaultAdslRedialer();
x.UseDefaultInternetDetector();
});
b.AddLocalDownloadCenter();
b.AddSpiderStatisticsCenter(x =>
{
// 添加内存统计服务
x.UseMemory();
});
});
}
builder.Register(spiderType);
var provider = builder.Build();
var instance = provider.Create(spiderType);
if (instance != null)
{
instance.Name = name;
instance.Id = id;
instance.RunAsync(arguments);
}
else
{
ConsoleHelper.WriteLine("创建爬虫对象失败", 0, ConsoleColor.DarkYellow);
}
}
static void Main(string[] args)
{
try
{
var builder = new SpiderHostBuilder();
var configurationBuilder = Framework.CreateConfigurationBuilder(null, args);
var configuration = configurationBuilder.Build();
var @class = configuration["dotnetspider.spider.class"];
var spiderId = configuration["dotnetspider.spider.id"];
@class = "DotnetSpider.Spiders.CnblogsSpider";
spiderId = "xxxxxxxx";
var folder = Directory.Exists("/logs/") ? "/logs/" : "";
var logPath = string.IsNullOrWhiteSpace(spiderId)
? $"{folder}{DateTime.Now:yyyy-MM-dd HH:mm:ss}.log"
: $"{folder}{spiderId}.log";
var loggerConfiguration = new LoggerConfiguration()
.MinimumLevel.Information()
.MinimumLevel.Override("Microsoft", LogEventLevel.Warning)
.Enrich.FromLogContext()
.WriteTo.Console().WriteTo
.RollingFile(logPath);
Log.Logger = loggerConfiguration.CreateLogger();
var spiderName = configuration["dotnetspider.spider.name"];
if (string.IsNullOrWhiteSpace(@class) ||
string.IsNullOrWhiteSpace(spiderId) ||
string.IsNullOrWhiteSpace(spiderName)
)
{
Log.Logger.Error($"执行爬虫的参数不正确: class {@class}, id {spiderId}, name {spiderName}");
return;
}
var type = Type.GetType(@class);
if (type == null)
{
Log.Logger.Error($"未找到爬虫类型: {@class}");
return;
}
Log.Logger.Information($"获取爬虫类型 {type.FullName} 成功");
builder.ConfigureAppConfiguration(x =>
{
x.AddCommandLine(args);
});
builder.ConfigureLogging(x =>
{
x.AddSerilog();
});
builder.ConfigureServices(services =>
{
services.AddKafkaMessageQueue();
});
builder.Register(type);
var provider = builder.Build();
var spider = provider.Create(type);
Log.Logger.Information($"创建爬虫实例成功");
spider.Id = spiderId;
spider.Name = spiderName;
Log.Logger.Information($"尝试启动爬虫实例");
spider.Run();
Log.Logger.Information($"爬虫实例退出");
}
catch (Exception e)
{
Log.Logger.Error($"执行失败: {e}");
}
}