public void Insert()
{
ClearDb();
using (SqlConnection conn = new SqlConnection(ConnectString))
{
ISpider spider = new DefaultSpider("test", new Site());
MsSqlEntityPipeline insertPipeline = new MsSqlEntityPipeline(ConnectString);
insertPipeline.InitiEntity(EntitySpider.ParseEntityMetaData(typeof(Product).GetTypeInfo()));
insertPipeline.InitPipeline(spider);
// Common data
JObject data1 = new JObject {
{ "sku", "110" }, { "category", "3C" }, { "url", "http://jd.com/110" }, { "cdate", "2016-08-13" }
};
JObject data2 = new JObject {
{ "sku", "111" }, { "category", "3C" }, { "url", "http://jd.com/111" }, { "cdate", "2016-08-13" }
};
// Value is null
JObject data3 = new JObject {
{ "sku", "112" }, { "category", null }, { "url", "http://jd.com/111" }, { "cdate", "2016-08-13" }
};
insertPipeline.Process(new List <JObject> {
data1, data2, data3
});
var list = conn.Query <Product>($"use test;select * from sku_{DateTime.Now.ToString("yyyy_MM_dd")}").ToList();
Assert.AreEqual(3, list.Count);
Assert.AreEqual("110", list[0].Sku);
Assert.AreEqual("111", list[1].Sku);
Assert.AreEqual(null, list[2].Category);
}
ClearDb();
}
public void PushAndPollBreadthFirst()
{
QueueDuplicateRemovedScheduler scheduler = new QueueDuplicateRemovedScheduler();
scheduler.DepthFirst = false;
ISpider spider = new DefaultSpider("test", new Site());
scheduler.Init(spider);
scheduler.Push(new Request("http://www.a.com", 1, null));
scheduler.Push(new Request("http://www.a.com", 1, null));
scheduler.Push(new Request("http://www.a.com", 1, null));
scheduler.Push(new Request("http://www.b.com", 1, null));
var request = scheduler.Poll();
Assert.Equal(request.Url.ToString(), "http://www.a.com/");
long left = scheduler.GetLeftRequestsCount();
long total = scheduler.GetTotalRequestsCount();
Assert.Equal(left, 1);
Assert.Equal(total, 2);
}
public void Load()
{
QueueDuplicateRemovedScheduler scheduler = new QueueDuplicateRemovedScheduler();
ISpider spider = new DefaultSpider("test");
scheduler.Push(new Request("http://www.a.com/", null));
scheduler.Push(new Request("http://www.b.com/", null));
scheduler.Push(new Request("http://www.c.com/", null));
scheduler.Push(new Request("http://www.d.com/", null));
Extension.Scheduler.RedisScheduler redisScheduler = GetRedisScheduler(spider.Identity);
redisScheduler.Dispose();
redisScheduler.Reload(scheduler.All);
Assert.Equal("http://www.d.com/", redisScheduler.Poll().Url.ToString());
Assert.Equal("http://www.c.com/", redisScheduler.Poll().Url.ToString());
Assert.Equal("http://www.b.com/", redisScheduler.Poll().Url.ToString());
Assert.Equal("http://www.a.com/", redisScheduler.Poll().Url.ToString());
redisScheduler.Dispose();
}
public void Clear()
{
Extension.Scheduler.RedisScheduler scheduler = GetRedisScheduler();
ISpider spider = new DefaultSpider();
scheduler.Init(spider);
scheduler.Dispose();
Request request1 = new Request("http://www.ibm.com/1", null);
Request request2 = new Request("http://www.ibm.com/2", null);
Request request3 = new Request("http://www.ibm.com/3", null);
Request request4 = new Request("http://www.ibm.com/4", null);
scheduler.Push(request1);
scheduler.Push(request2);
scheduler.Push(request3);
scheduler.Push(request4);
Request result = scheduler.Poll();
Assert.AreEqual("http://www.ibm.com/4", result.Url.ToString());
scheduler.Dispose();
}
public void PushAndPollBreadthFirst()
{
QueueDuplicateRemovedScheduler scheduler = new QueueDuplicateRemovedScheduler();
scheduler.DepthFirst = false;
ISpider spider = new DefaultSpider("test", new Site());
scheduler.Init(spider);
scheduler.Push(new Request("http://www.a.com", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.a.com", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.a.com", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.b.com", null)
{
Site = spider.Site
});
var request = scheduler.Poll();
Assert.Equal("http://www.a.com/", request.Url.ToString());
long left = scheduler.LeftRequestsCount;
long total = scheduler.TotalRequestsCount;
Assert.Equal(1, left);
Assert.Equal(2, total);
}
public void InsertAndIgnoreDuplicate()
{
#if NETSTANDARD
if (!RuntimeInformation.IsOSPlatform(OSPlatform.Windows))
{
return;
}
#endif
ClearDb();
ISpider spider = new DefaultSpider("test", new Site());
CassandraEntityPipeline insertPipeline = new CassandraEntityPipeline(connectString);
var metadata = new EntityDefine <ProductInsert>();
insertPipeline.AddEntity(metadata);
insertPipeline.Init();
// Common data
var data1 = new ProductInsert {
Sku = "110", Category = "3C", Url = "http://jd.com/110", CDate = new DateTime(2016, 8, 13)
};
var data2 = new ProductInsert {
Sku = "111", Category = "3C", Url = "http://jd.com/111", CDate = new DateTime(2016, 8, 13)
};
var data3 = new ProductInsert {
Sku = "112", Category = null, Url = "http://jd.com/111", CDate = new DateTime(2016, 8, 13)
};
// Value is null
insertPipeline.Process(metadata.Name, new List <dynamic> {
data1, data2, data3
}, spider);
var cluster = CassandraUtil.CreateCluster(connectString);
var session = cluster.Connect();
session.ChangeKeyspace("test");
var rows = session.Execute($"SELECT * FROM test.sku_cassandra_{DateTime.Now.ToString("yyyy_MM_dd")}").GetRows().ToList();
var results = new List <ProductInsert>();
foreach (var row in rows)
{
results.Add(new ProductInsert
{
Sku = row.GetValue <string>("sku"),
Category = row.GetValue <string>("category"),
Id = row.GetValue <Guid>("id")
});
}
insertPipeline.DefaultPipelineModel = PipelineMode.InsertAndIgnoreDuplicate;
var sku = results.First().Sku;
var data4 = new ProductInsert {
Id = results.First().Id, Sku = "113", Category = "asdfasf", Url = "http://jd.com/111", CDate = new DateTime(2016, 8, 13)
};
insertPipeline.Process(metadata.Name, new List <dynamic> {
data4
}, spider);
rows = session.Execute($"SELECT * FROM test.sku_cassandra_{DateTime.Now.ToString("yyyy_MM_dd")}").GetRows().ToList();
results = new List <ProductInsert>();
foreach (var row in rows)
{
results.Add(new ProductInsert
{
Sku = row.GetValue <string>("sku"),
Category = row.GetValue <string>("category")
});
}
Assert.Equal(3, results.Count);
Assert.DoesNotContain(results, r => r.Sku == sku);
Assert.Contains(results, r => r.Sku == "113");
Assert.Contains(results, r => r.Category == "asdfasf");
ClearDb();
}
public virtual void DataTypes()
{
using (MySqlConnection conn = new MySqlConnection("Database='mysql';Data Source=localhost;User ID=root;Password=;Port=3306;SslMode=None;"))
{
try
{
conn.Execute("use test; drop table table15;");
}
catch
{
}
var spider = new DefaultSpider();
EntityProcessor <Entity15> processor = new EntityProcessor <Entity15>();
var pipeline = new MySqlEntityPipeline("Database='mysql';Data Source=localhost;User ID=root;Password=;Port=3306;SslMode=None;");
var resultItems = new ResultItems();
resultItems.Request = new Request();
resultItems.AddOrUpdateResultItem(processor.Model.Identity,
new Tuple <IModel, IList <dynamic> >(processor.Model, new[] {
new Dictionary <string, dynamic>
{
{ "int", "1" },
{ "bool", "1" },
{ "bigint", "11" },
{ "string", "aaa" },
{ "time", "2018-06-12" },
{ "float", "1" },
{ "double", "1" },
{ "string1", "abc" },
{ "string2", "abcdd" },
{ "decimal", "1" }
}
}));
pipeline.Process(new ResultItems[] { resultItems }, spider.Logger, spider);
var columns = conn.Query <ColumnInfo>("SELECT COLUMN_NAME as `Name`, COLUMN_TYPE as `Type` FROM information_schema.columns WHERE table_name='table15' AND table_schema = 'test';").ToList();;
Assert.Equal(12, columns.Count);
Assert.Equal("int".ToLower(), columns[0].Name);
Assert.Equal("bool".ToLower(), columns[1].Name);
Assert.Equal("bigint".ToLower(), columns[2].Name);
Assert.Equal("string".ToLower(), columns[3].Name);
Assert.Equal("time".ToLower(), columns[4].Name);
Assert.Equal("float".ToLower(), columns[5].Name);
Assert.Equal("double".ToLower(), columns[6].Name);
Assert.Equal("string1".ToLower(), columns[7].Name);
Assert.Equal("string2".ToLower(), columns[8].Name);
Assert.Equal("decimal".ToLower(), columns[9].Name);
Assert.Equal("creation_time".ToLower(), columns[10].Name);
Assert.Equal("creation_date".ToLower(), columns[11].Name);
Assert.Equal("int(11)", columns[0].Type);
Assert.Equal("tinyint(1)", columns[1].Type);
Assert.Equal("bigint(20)", columns[2].Type);
Assert.Equal("varchar(255)", columns[3].Type);
Assert.Equal("timestamp", columns[4].Type);
Assert.Equal("float", columns[5].Type);
Assert.Equal("double", columns[6].Type);
Assert.Equal("varchar(100)", columns[7].Type);
Assert.Equal("longtext", columns[8].Type);
Assert.Equal("decimal(18,2)", columns[9].Type);
Assert.Equal("timestamp", columns[10].Type);
Assert.Equal("date", columns[11].Type);
try
{
conn.Execute("use test; drop table table15;");
}
catch
{
}
}
}
public void Status()
{
Extension.Scheduler.RedisScheduler scheduler = GetRedisScheduler();
ISpider spider = new DefaultSpider("test", new Site());
scheduler.Init(spider);
scheduler.Dispose();
scheduler.Push(new Request("http://www.a.com/", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.b.com/", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.c.com/", null)
{
Site = spider.Site
});
scheduler.Push(new Request("http://www.d.com/", null)
{
Site = spider.Site
});
Assert.Equal(0, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.LeftRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.IncreaseErrorCount();
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(0, scheduler.SuccessRequestsCount);
scheduler.IncreaseSuccessCount();
Assert.Equal(1, scheduler.SuccessRequestsCount);
scheduler.Poll();
Assert.Equal(3, scheduler.LeftRequestsCount);
Assert.Equal(1, scheduler.SuccessRequestsCount);
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.Poll();
Assert.Equal(2, scheduler.LeftRequestsCount);
Assert.Equal(1, scheduler.SuccessRequestsCount);
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.Poll();
Assert.Equal(1, scheduler.LeftRequestsCount);
Assert.Equal(1, scheduler.SuccessRequestsCount);
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.Poll();
Assert.Equal(0, scheduler.LeftRequestsCount);
Assert.Equal(1, scheduler.SuccessRequestsCount);
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.Poll();
scheduler.Poll();
Assert.Equal(0, scheduler.LeftRequestsCount);
Assert.Equal(1, scheduler.SuccessRequestsCount);
Assert.Equal(1, scheduler.ErrorRequestsCount);
Assert.Equal(4, scheduler.TotalRequestsCount);
scheduler.Dispose();
}
public void SubContentHandler()
{
var spider = new DefaultSpider("test", new Site());
TestDownloader downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c"
});
var request1 = new Request("http://a.com/", null);
Page page = downloader1.Download(request1, spider);
Assert.AreEqual("aabbc", page.Content);
downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c",
EndOffset = 1
});
page = downloader1.Download(request1, spider);
Assert.AreEqual("aabb", page.Content);
downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 1
});
page = downloader1.Download(request1, spider);
Assert.AreEqual("abbc", page.Content);
downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 1,
EndOffset = 1
});
page = downloader1.Download(request1, spider);
Assert.AreEqual("abb", page.Content);
downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 10
});
var downloader2 = downloader1;
try
{
page = downloader2.Download(request1, spider);
throw new System.Exception("test failed.");
}
catch (SpiderException exception)
{
Assert.AreEqual("Sub content failed. Please check your settings.", exception.Message);
}
downloader1 = new TestDownloader();
downloader1.AddAfterDownloadCompleteHandler(new SubContentHandler
{
Start = "a",
End = "c",
EndOffset = 20
});
try
{
page = downloader1.Download(request1, spider);
throw new System.Exception("test failed.");
}
catch (SpiderException exception)
{
Assert.AreEqual("Sub content failed. Please check your settings.", exception.Message);
}
}
public void SubContentHandler()
{
var spider = new DefaultSpider("test", new Site());
TestDownloader downloader1 = new TestDownloader()
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c"
}
}
};
var request1 = new Request("http://a.com/", 0, null);
Page page = downloader1.Download(request1, spider);
Assert.Equal("aabbc", page.Content);
downloader1 = new TestDownloader
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c",
EndOffset = 1
}
}
};
page = downloader1.Download(request1, spider);
Assert.Equal("aabb", page.Content);
downloader1 = new TestDownloader
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 1
}
}
};
page = downloader1.Download(request1, spider);
Assert.Equal("abbc", page.Content);
downloader1 = new TestDownloader
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 1,
EndOffset = 1
}
}
};
page = downloader1.Download(request1, spider);
Assert.Equal("abb", page.Content);
downloader1 = new TestDownloader
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c",
StartOffset = 10
}
}
};
var downloader2 = downloader1;
var exception = Assert.Throws <SpiderException>(() =>
{
page = downloader2.Download(request1, spider);
});
Assert.Equal("Sub content failed. Please check your settings.", exception.Message);
downloader1 = new TestDownloader
{
DownloadCompleteHandlers = new IDownloadCompleteHandler[]
{
new SubContentHandler
{
Start = "a",
End = "c",
EndOffset = 20
}
}
};
exception = Assert.Throws <SpiderException>(() =>
{
page = downloader1.Download(request1, spider);
});
Assert.Equal("Sub content failed. Please check your settings.", exception.Message);
}
public override void DataTypes()
{
if (!Env.IsWindows)
{
return;
}
using (var conn = new SqlConnection("Server=.\\SQLEXPRESS;Database=master;Trusted_Connection=True;MultipleActiveResultSets=true"))
{
try
{
conn.Execute("create database test;");
}
catch
{
}
try
{
conn.Execute("USE [test]; drop table [test].dbo.[table15]");
}
catch
{
}
var spider = new DefaultSpider();
EntityProcessor <Entity15> processor = new EntityProcessor <Entity15>();
var pipeline = new SqlServerEntityPipeline("Server=.\\SQLEXPRESS;Database=master;Trusted_Connection=True;MultipleActiveResultSets=true");
var resultItems = new ResultItems();
resultItems.Request = new Request();
resultItems.AddOrUpdateResultItem(processor.Model.Identity, new Tuple <IModel, IEnumerable <dynamic> >(processor.Model, new dynamic[] {
new Dictionary <string, dynamic>
{
{ "int", "1" },
{ "bool", "1" },
{ "bigint", "11" },
{ "string", "aaa" },
{ "time", "2018-06-12" },
{ "float", "1" },
{ "double", "1" },
{ "string1", "abc" },
{ "string2", "abcdd" },
{ "decimal", "1" }
}
}));
pipeline.Process(new ResultItems[] { resultItems }, spider);
var columns = conn.Query <ColumnInfo>("USE [test];select b.name Name,c.name+'(' + cast(c.length as varchar)+')' [Type] from sysobjects a,syscolumns b,systypes c where a.id=b.id and a.name='table15' and a.xtype='U'and b.xtype=c.xtype").ToList();
Assert.Equal(15, columns.Count);
Assert.Equal("creation_date".ToLower(), columns[0].Name);
Assert.Equal("int".ToLower(), columns[1].Name);
Assert.Equal("time".ToLower(), columns[2].Name);
Assert.Equal("creation_time".ToLower(), columns[3].Name);
Assert.Equal("float".ToLower(), columns[4].Name);
Assert.Equal("double".ToLower(), columns[5].Name);
Assert.Equal("bool".ToLower(), columns[6].Name);
Assert.Equal("decimal".ToLower(), columns[7].Name);
Assert.Equal("bigint".ToLower(), columns[8].Name);
Assert.Equal("string".ToLower(), columns[9].Name);
Assert.Equal("string1".ToLower(), columns[10].Name);
Assert.Equal("string2".ToLower(), columns[11].Name);
Assert.Equal("date(3)", columns[0].Type);
Assert.Equal("int(4)", columns[1].Type);
Assert.Equal("datetime(8)", columns[2].Type);
Assert.Equal("datetime(8)", columns[3].Type);
Assert.Equal("float(8)", columns[4].Type);
Assert.Equal("float(8)", columns[5].Type);
Assert.Equal("bit(1)", columns[6].Type);
Assert.Equal("decimal(17)", columns[7].Type);
Assert.Equal("bigint(8)", columns[8].Type);
Assert.Equal("nvarchar(8000)", columns[9].Type);
Assert.Equal("nvarchar(8000)", columns[10].Type);
Assert.Equal("nvarchar(8000)", columns[11].Type);
conn.Execute("USE [test]; drop table [test].dbo.[table15]");
}
}