private void MapOne(Dictionary <string, object> obj)
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "id"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "user.id"));
// [retweeted_tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "retweeted_status.id"));
// [retweeted_user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "retweeted_status.user.id"));
// [created_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "created_at"));
// [retweeted_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "retweeted_status.created_at"));
BulkWriter.EndLine();
}
public async Task CanWriteSyncWithExistingConnectionAndTransaction()
{
string tableName = DropCreate(nameof(BulkWriterTestsMyTestClass));
using (var connection = new SqlConnection(_connectionString))
{
await connection.OpenAsync();
using (var transaction = connection.BeginTransaction())
{
var writer = new BulkWriter <BulkWriterTestsMyTestClass>(connection, transaction);
var items = Enumerable.Range(1, 1000)
.Select(i => new BulkWriterTestsMyTestClass {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(connection, $"SELECT COUNT(1) FROM {tableName}", transaction);
Assert.Equal(1000, count);
transaction.Rollback();
count = (int)await TestHelpers.ExecuteScalar(connection, $"SELECT COUNT(1) FROM {tableName}");
Assert.Equal(0, count);
}
}
}
public async Task Should_Handle_Column_VarBinary_Large()
{
string tableName = nameof(MyTestClassForVarBinary);
TestHelpers.ExecuteNonQuery(_connectionString, $"DROP TABLE IF EXISTS [dbo].[{tableName}]");
TestHelpers.ExecuteNonQuery(_connectionString,
"CREATE TABLE [dbo].[" + tableName + "](" +
"[Id] [int] IDENTITY(1,1) NOT NULL," +
"[Data] [varbinary](MAX) NULL," +
"CONSTRAINT [PK_" + tableName + "] PRIMARY KEY CLUSTERED ([Id] ASC)" +
")");
var writer = new BulkWriter <MyTestClassForVarBinary>(_connectionString);
var items = new[] { new MyTestClassForVarBinary {
Id = 1, Data = new byte[1024 * 1024 * 1]
} };
new Random().NextBytes(items.First().Data);
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName}");
var data = (byte[])await TestHelpers.ExecuteScalar(_connectionString, $"SELECT TOP 1 Data FROM {tableName}");
Assert.Equal(items.First().Data, data);
Assert.Equal(1, count);
}
public async Task PartiallyCompletesWhenAStepThrows()
{
var tableName = TestHelpers.DropCreate(nameof(PipelineTestsMyTestClass));
using (var writer = new BulkWriter <PipelineTestsMyTestClass>(_connectionString))
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var pipeline = EtlPipeline
.StartWith(items)
.TransformInPlace(i =>
{
if (i.Id > 500)
{
throw new Exception("Transform exception");
}
i.Id -= 1;
i.Name = $"Alice {i.Id}";
})
.WriteTo(writer);
var pipelineTask = pipeline.ExecuteAsync();
await Assert.ThrowsAsync <Exception>(() => pipelineTask);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName}");
Assert.Equal(500, count);
}
}
public async Task CanWriteSyncWithOptions()
{
var tableName = TestHelpers.DropCreate(nameof(BulkWriterTestsMyTestClass));
var tableNameWithKey = TestHelpers.DropCreate(nameof(BulkWriterTestsMyTestClassWithKey));
var writer = new BulkWriter <BulkWriterTestsMyTestClass>(_connectionString);
var writerWithOptions = new BulkWriter <BulkWriterTestsMyTestClassWithKey>(_connectionString, SqlBulkCopyOptions.KeepIdentity);
var items = Enumerable.Range(11, 20).Select(i => new BulkWriterTestsMyTestClass {
Id = i, Name = "Bob"
});
var itemsWithKey = Enumerable.Range(11, 20).Select(i => new BulkWriterTestsMyTestClassWithKey {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
writerWithOptions.WriteToDatabase(itemsWithKey);
var minId = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT MIN(Id) FROM {tableName}");
var minIdWithKey = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT MIN(Id) FROM {tableNameWithKey}");
Assert.Equal(1, minId);
Assert.Equal(11, minIdWithKey);
}
public async Task CanSetBulkCopyParameters()
{
const int timeout = 10;
const int batchSize = 1000;
var writer = new BulkWriter <BulkWriterInitializationTestsMyTestClass>(_connectionString)
{
BulkCopyTimeout = timeout,
BatchSize = batchSize,
BulkCopySetup = bcp =>
{
Assert.Equal(timeout, bcp.BulkCopyTimeout);
Assert.Equal(batchSize, bcp.BatchSize);
}
};
var items = Enumerable.Range(1, 10)
.Select(i => new BulkWriterInitializationTestsMyTestClass {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {_tableName}");
Assert.Equal(10, count);
}
public async Task CanSetBulkCopyParameters_Respects_Table_Annotation()
{
string tableName = DropCreate("TestClass2");
const int timeout = 10;
const int batchSize = 1000;
bool setupCallbackInvoked = false;
var writer = new BulkWriter <BulkWriterInitializationTestsMyTestClassAnnotation>(_connectionString)
{
BulkCopyTimeout = timeout,
BatchSize = batchSize,
BulkCopySetup = bcp =>
{
setupCallbackInvoked = true;
Assert.Equal("TestClass2", bcp.DestinationTableName);
Assert.Equal(timeout, bcp.BulkCopyTimeout);
Assert.Equal(batchSize, bcp.BatchSize);
}
};
var items = Enumerable.Range(1, 10)
.Select(i => new BulkWriterInitializationTestsMyTestClassAnnotation {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName }");
Assert.Equal(10, count);
Assert.True(setupCallbackInvoked);
}
private void MapOne(Dictionary <string, object> obj)
{
ArrayList um = (ArrayList)JsonUtil.GetValue(obj, "entities.hashtags");
for (int i = 0; i < um.Count; i++)
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "id"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "user.id"));
// [tag] [nvarchar](50) NOT NULL
//string tag = (string)((Dictionary<string, object>)um[i])["text"];
//WriteTinyInt(tag.Substring(0, Math.Min(tag.Length, 50)));
BulkWriter.WriteVarChar(JsonUtil.GetString(um[i], "text"), 50);
// [created_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "created_at"));
BulkWriter.EndLine();
}
}
public override void RunAsync(CancellationToken cancellationToken)
{
var enumerable = _inCollection.GetConsumingEnumerable(cancellationToken);
using (var bulkWriter = new BulkWriter <T>(_context.ConnectionString))
{
bulkWriter.WriteToDatabase(enumerable);
}
}
public async Task BulkWriterAsyncEnumerable()
{
await using var sqlConnection = DbHelpers.OpenSqlConnection();
using var bulkWriter = new BulkWriter <DomainEntity>(sqlConnection)
{
BulkCopyTimeout = 0,
BatchSize = 10000
};
var items = GetTestRecords().ToAsyncEnumerable();
await bulkWriter.WriteToDatabaseAsync(items);
}
private void MapOne(Dictionary <string, object> obj)
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "scrub_geo.user_id"));
// [up_to_status_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "scrub_geo.up_to_status_id"));
BulkWriter.EndLine();
}
private void MapOne(Dictionary <string, object> obj)
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "delete.status.id"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "delete.status.user_id"));
BulkWriter.EndLine();
}
public async Task CanWriteSync()
{
var writer = new BulkWriter <BulkWriterTestsMyTestClass>(_connectionString);
var items = Enumerable.Range(1, 1000).Select(i => new BulkWriterTestsMyTestClass {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {_tableName}");
Assert.Equal(1000, count);
}
private void WriteUrl(Dictionary <string, object> obj, ArrayList um)
{
if (um != null)
{
for (int i = 0; i < um.Count; i++)
{
string url = (string)((Dictionary <string, object>)um[i])["url"];
if (url.StartsWith("http://t.co/"))
{
url = url.Substring(12);
}
else if (url.StartsWith("http://t.co/"))
{
url = url.Substring(13);
}
else
{
continue;
}
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "id"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "user.id"));
// [url_id] [char](8) NOT NULL
BulkWriter.WriteChar(url, 8);
// [created_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "created_at"));
// [expanded_url] [varchar](8000)
if (((Dictionary <string, object>)um[i]).ContainsKey("expanded_url"))
{
BulkWriter.WriteVarChar((string)((Dictionary <string, object>)um[i])["expanded_url"], 8000);
}
else
{
BulkWriter.WriteVarChar(null, 8000);
}
BulkWriter.EndLine();
}
}
}
public async Task Should_Handle_Both_Ordinal_And_ColumnName_For_Destination_Mapping()
{
string tableName = DropCreate(nameof(OrdinalAndColumnNameExampleType));
var writer = new BulkWriter <OrdinalAndColumnNameExampleType>(_connectionString);
var items = new[] { new OrdinalAndColumnNameExampleType {
Id = 1, Name2 = "Bob"
} };
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName}");
Assert.Equal(1, count);
}
public async Task RaisesExceptionsForAllStepsThatThrow()
{
var tableName = TestHelpers.DropCreate(nameof(PipelineTestsMyTestClass));
using (var writer = new BulkWriter <PipelineTestsMyTestClass>(_connectionString))
{
var items = Enumerable
.Range(1, 1000)
.Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
})
.ToAsyncEnumerable();
var pipeline = EtlPipeline
.StartWith(items)
.Project(i =>
{
//pump a few values through to ensure the next pipeline step actually
//gets run
if (i.Id >= 400)
{
throw new Exception("Projection exception 1");
}
return(i);
})
.Project(i =>
{
if (i.Id >= 200)
{
throw new Exception("Projection exception 2");
}
return(i);
})
.WriteTo(writer);
var pipelineTask = pipeline.ExecuteAsync();
await Assert.ThrowsAsync <Exception>(() => pipelineTask);
Assert.Equal(2, pipelineTask.Exception.InnerExceptions.Count);
Assert.Equal(1, pipelineTask.Exception.InnerExceptions.Count(e => e.Message == "Projection exception 1"));
Assert.Equal(1, pipelineTask.Exception.InnerExceptions.Count(e => e.Message == "Projection exception 2"));
}
}
private void WriteProjectDependencies(IEnumerable <Dependency> dependencies, IDictionary <string, int> projectMap)
{
var projectDependencyIntCounter = new IntCounter(-1);
var projectDependencies = dependencies
.Select(x => new ProjectDependency
{
Id = projectDependencyIntCounter.Next(),
ProjectFromId = projectMap[x.ProjectName],
ProjectToId = projectMap[x.Id],
Version = x.Version,
TargetFramework = x.Framework,
Type = x.Type.ToString()
});
using (var bulkWriter = new BulkWriter <ProjectDependency>(_connectionString))
{
bulkWriter.WriteToDatabase(projectDependencies);
}
}
public async Task CanWriteSyncWithExistingConnectionAndTransactionAndOptions()
{
var tableName = TestHelpers.DropCreate(nameof(BulkWriterTestsMyTestClass));
var tableNameWithKey = TestHelpers.DropCreate(nameof(BulkWriterTestsMyTestClassWithKey));
using (var connection = new SqlConnection(_connectionString))
{
await connection.OpenAsync();
using (var transaction = connection.BeginTransaction())
{
var writer = new BulkWriter <BulkWriterTestsMyTestClass>(connection, transaction);
var writerWithOptions = new BulkWriter <BulkWriterTestsMyTestClassWithKey>(connection, SqlBulkCopyOptions.KeepIdentity, transaction);
var items = Enumerable.Range(11, 20).Select(i => new BulkWriterTestsMyTestClass {
Id = i, Name = "Bob"
});
var itemsWithKey = Enumerable.Range(11, 20).Select(i => new BulkWriterTestsMyTestClassWithKey {
Id = i, Name = "Bob"
});
writer.WriteToDatabase(items);
writerWithOptions.WriteToDatabase(itemsWithKey);
var minId = (int?)await TestHelpers.ExecuteScalar(connection, $"SELECT MIN(Id) FROM {tableName}", transaction);
var minIdWithKey = (int?)await TestHelpers.ExecuteScalar(connection, $"SELECT MIN(Id) FROM {tableNameWithKey}", transaction);
Assert.Equal(1, minId);
Assert.Equal(11, minIdWithKey);
transaction.Rollback();
var emptyMinId = await TestHelpers.ExecuteScalar(connection, $"SELECT MIN(Id) FROM {tableName}");
var emptyMinIdWithKey = await TestHelpers.ExecuteScalar(connection, $"SELECT MIN(Id) FROM {tableNameWithKey}");
Assert.Equal(emptyMinId, System.DBNull.Value);
Assert.Equal(emptyMinIdWithKey, System.DBNull.Value);
}
}
}
public async Task ThrowsWhenAStepThrows()
{
var tableName = TestHelpers.DropCreate(nameof(PipelineTestsMyTestClass));
using (var writer = new BulkWriter <PipelineTestsMyTestClass>(_connectionString))
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var pipeline = EtlPipeline
.StartWith(items)
.Project <PipelineTestsMyTestClass>(i => throw new Exception("Projection exception"))
.WriteTo(writer);
var pipelineTask = pipeline.ExecuteAsync();
var exception = await Assert.ThrowsAsync <Exception>(() => pipelineTask);
Assert.Equal("Projection exception", exception.Message);
}
}
public async Task WritesToBulkWriter()
{
var tableName = TestHelpers.DropCreate(nameof(PipelineTestsMyTestClass));
using (var writer = new BulkWriter <PipelineTestsMyTestClass>(_connectionString))
{
var items = Enumerable.Range(1, 1000).Select(i => new PipelineTestsMyTestClass {
Id = i, Name = "Bob"
});
var pipeline = EtlPipeline
.StartWith(items)
.WriteTo(writer);
await pipeline.ExecuteAsync();
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName}");
Assert.Equal(1000, count);
}
}
private IDictionary <string, int> WriteProjects(IEnumerable <Dependency> dependencies)
{
var projectIntCounter = new IntCounter(-1);
var projects = dependencies
.SelectMany(x => new List <string> {
x.ProjectName, x.Id
})
.Distinct()
.Select(x => new Project
{
Id = projectIntCounter.Next(),
Name = x
})
.ToList();
using (var bulkWriter = new BulkWriter <Project>(_connectionString))
{
bulkWriter.WriteToDatabase(projects);
}
return(projects.ToDictionary(x => x.Name, x => x.Id));
}
private static async Task Main(string[] args)
{
SetupDb();
var timer = new Stopwatch();
using (var bulkWriter = new BulkWriter <MyDomainEntity>(@"Data Source=.\sqlexpress;Database=BulkWriter.Demo;Trusted_Connection=True;Connection Timeout=300")
{
BulkCopyTimeout = 0,
BatchSize = 10000
})
{
var items = GetDomainEntities();
timer.Start();
await bulkWriter.WriteToDatabaseAsync(items);
timer.Stop();
}
Console.WriteLine(timer.ElapsedMilliseconds);
Console.ReadKey();
}
public async Task Should_Handle_Column_Nvarchar_With_Length_Max()
{
string tableName = nameof(MyTestClassForNvarCharMax);
TestHelpers.ExecuteNonQuery(_connectionString, $"DROP TABLE IF EXISTS [dbo].[{tableName}]");
TestHelpers.ExecuteNonQuery(_connectionString,
"CREATE TABLE [dbo].[" + tableName + "](" +
"[Id] [int] IDENTITY(1,1) NOT NULL," +
"[Name] [nvarchar](MAX) NULL," +
"CONSTRAINT [PK_" + tableName + "] PRIMARY KEY CLUSTERED ([Id] ASC)" +
")");
var writer = new BulkWriter <MyTestClassForNvarCharMax>(_connectionString);
var items = new[] { new MyTestClassForNvarCharMax {
Id = 1, Name = "Bob"
} };
writer.WriteToDatabase(items);
var count = (int)await TestHelpers.ExecuteScalar(_connectionString, $"SELECT COUNT(1) FROM {tableName}");
Assert.Equal(1, count);
}
private void MapOne(Dictionary <string, object> obj)
{
ArrayList um = (ArrayList)JsonUtil.GetValue(obj, "entities.user_mentions");
for (int i = 0; i < um.Count; i++)
{
if (!String.IsNullOrWhiteSpace((string)((Dictionary <string, object>)um[i])["id"]))
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "id"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "user.id"));
// [mentioned_user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(um[i], "id"));
BulkWriter.EndLine();
}
}
}
private void WriteBlocksToDisk(ConcurrentIOPoolManager readPool, ConcurrentIOPoolManager writePool,
Algorithm.Collection.ICollectionOnDisk parent, IDictionary <long, Sop.DataBlock> blocks)
{
// todo: do this Async way, when time permits. :)
if (BinaryWriter == null)
{
if (BufferStream == null)
{
BufferStream = new MemoryStream();
}
BinaryWriter = new BinaryWriter(BufferStream, parent.File.Server.Encoding);
}
else
{
BinaryWriter.Seek(0, SeekOrigin.Begin);
}
const int sizeOfNumerics = Sop.DataBlock.OverheadSize;
var writeBuffer = _writeBuffer;
int bufferIndex = 0, currentTargetBufferIndex = 0;
long runningAddress = -1, startBlockAddress = -1;
var bulkWriter = new BulkWriter();
var dataChunks = new List <BulkWriter.DataChunk>(4);
#region resize data file before appending to it...
if (readPool == null || writePool == null)
{
long currentAddress = -1;
if (((Collections.Generic.ISortedDictionary <long, Sop.DataBlock>)blocks).MoveLast())
{
currentAddress = ((Collections.Generic.ISortedDictionary <long, Sop.DataBlock>)blocks).CurrentKey;
}
if (currentAddress > -1)
{
// thread safe increase File Size to accomodate data to be appended...
var FileSize = currentAddress + (int)parent.DataBlockSize;
if (parent.FileStream.Length < FileSize)
{
if (parent.Transaction != null)
{
lock (parent.Transaction)
{
if (parent.FileStream.Length < FileSize)
{
parent.FileStream.SetLength(FileSize);
}
}
}
else
{
parent.FileStream.SetLength(FileSize);
}
}
}
}
#endregion
Sop.DataBlock[] blocksCopy = new Sop.DataBlock[blocks.Count];
blocks.Values.CopyTo(blocksCopy, 0);
foreach (Sop.DataBlock block in blocksCopy)
{
SetIsDirty(block, false);
if (block.DataAddress >= 0)
{
#region Process special states, e.g. - buffer is full, current block Address is fragmented from previous block's
if (startBlockAddress == -1)
{
startBlockAddress = runningAddress = block.DataAddress;
}
else
{
bool bufferIsFull = bufferIndex + sizeOfNumerics + block.Data.Length +
currentTargetBufferIndex > writeBuffer.Length - block.Length;
if (block.DataAddress != runningAddress || bufferIsFull)
{
dataChunks.Add(new BulkWriter.DataChunk
{
TargetDataAddress = startBlockAddress,
Index = currentTargetBufferIndex, // Index in the buffer of 1st byte of this segment
Size = bufferIndex // size of the segment
});
if (bufferIsFull)
{
//** write to disk
if (readPool != null && writePool != null)
{
bulkWriter.Backup(readPool, writePool, parent, writeBuffer, dataChunks);
if (writePool.AsyncThreadException != null)
{
throw writePool.AsyncThreadException;
}
else if (readPool.AsyncThreadException != null)
{
throw readPool.AsyncThreadException;
}
}
else if (writePool != null)
{
bulkWriter.Write(writePool, parent, writeBuffer, dataChunks);
if (writePool.AsyncThreadException != null)
{
throw writePool.AsyncThreadException;
}
}
else
{
throw new SopException("WriteBlocksToDisk has a bug!");
}
// create new buffer for succeeding chunks...
dataChunks = new List <BulkWriter.DataChunk>(4);
writeBuffer = new byte[writeBuffer.Length];
currentTargetBufferIndex = 0;
}
else
{
currentTargetBufferIndex += bufferIndex;
}
bufferIndex = 0;
runningAddress = startBlockAddress = block.DataAddress;
}
}
#endregion
}
else
{
throw new InvalidOperationException("Invalid (-) Block.DataAddress detected.");
}
//**** write Block Header and Data to disk
BinaryWriter.Seek(0, SeekOrigin.Begin);
// Byte 0 to 7: Next Item Address (64 bit long int) = 0 (no next item)
BinaryWriter.Write(block.NextItemAddress);
// Byte 8 to 11: Size Occupied
BinaryWriter.Write(block.SizeOccupied);
// Byte 12 to 19: Low-level next datablock address
BinaryWriter.Write(block.InternalNextBlockAddress);
// Byte 20: count of member blocks, max is 65535.
ushort memberCount = 0;
if (block.IsHead)
{
int cm = block.CountMembers(true);
memberCount = cm > Sop.DataBlock.MaxChainMemberCount ? Sop.DataBlock.MaxChainMemberCount : (ushort)cm;
}
BinaryWriter.Write(memberCount);
byte[] b2 = BufferStream.GetBuffer();
Array.Copy(b2, 0, writeBuffer, currentTargetBufferIndex + bufferIndex, sizeOfNumerics);
bufferIndex += sizeOfNumerics;
//** Byte 20 to 20 + Data Length: USER DATA
int cs = block.Data.Length;
//if (currentTargetBufferIndex + cs + bufferIndex > writeBuffer.Length - block.Length)
// cs = writeBuffer.Length - (currentTargetBufferIndex + bufferIndex);
Array.Copy(block.Data, 0, writeBuffer, currentTargetBufferIndex + bufferIndex, cs);
bufferIndex += block.Data.Length;
runningAddress += block.Length;
}
// write the last chunk set to disk...
if (startBlockAddress != -1)
{
//** write to disk
dataChunks.Add(new BulkWriter.DataChunk
{
TargetDataAddress = startBlockAddress,
Index = currentTargetBufferIndex,
Size = bufferIndex
});
}
if (dataChunks.Count > 0)
{
if (readPool != null && writePool != null)
{
bulkWriter.Backup(readPool, writePool, parent, writeBuffer, dataChunks);
}
else if (writePool != null)
{
bulkWriter.Write(writePool, parent, writeBuffer, dataChunks);
}
else
{
throw new SopException("WriteBlocksToDisk has a bug!");
}
}
}
/// <summary>
/// Write a group of Blocks into Disk. NOTE: it will be more optimal if Blocks
/// are sorted by its Data Address so this function can write contiguous blocks
/// in one async write.
/// </summary>
public int WriteBlocksToDisk(Algorithm.Collection.ICollectionOnDisk parent,
IDictionary <long, Sop.DataBlock> blocks, bool clear)
{
if (!parent.IsOpen)
{
return(0);
}
var blockSize = (int)parent.DataBlockSize;
int r = blocks.Count;
if (BinaryWriter == null)
{
if (BufferStream == null)
{
BufferStream = new MemoryStream();
}
BinaryWriter = new BinaryWriter(BufferStream, parent.File.Server.Encoding);
}
else
{
BinaryWriter.Seek(0, SeekOrigin.Begin);
}
const int sizeOfNumerics = Sop.DataBlock.OverheadSize;
int chunkSize = (int)DataBlockSize.FiveTwentyFourTwoEightyEight * 4;
if (chunkSize > blocks.Count * blockSize)
{
chunkSize = blocks.Count * blockSize;
}
if (_writeBuffer == null || _writeBuffer.Length < chunkSize)
{
_writeBuffer = new byte[chunkSize];
}
int bufferIndex = 0, startIndex = 0, currentTargetBufferIndex = 0;
long runningAddress = -1, startBlockAddress = -1;
var bulkWriter = new BulkWriter();
var dataChunks = new List <BulkWriter.DataChunk>(4);
foreach (Sop.DataBlock block in blocks.Values)
{
SetIsDirty(block, false);
if (block.DataAddress >= 0)
{
#region Process special states, e.g. - buffer is full
if (startBlockAddress == -1)
{
startBlockAddress = runningAddress = block.DataAddress;
}
else
{
bool bufferIsFull = (bufferIndex - startIndex) + sizeOfNumerics + block.Data.Length +
currentTargetBufferIndex > _writeBuffer.Length - block.Length;
if (block.DataAddress != runningAddress || bufferIsFull)
{
dataChunks.Add(new BulkWriter.DataChunk
{
TargetDataAddress =
startBlockAddress == -1
? block.DataAddress
: startBlockAddress,
Index = currentTargetBufferIndex + startIndex,
Size = bufferIndex - startIndex
});
if (bufferIsFull)
{
//** write to disk
bulkWriter.Write(parent, _writeBuffer, dataChunks);
//** reset buffer
dataChunks.Clear();
currentTargetBufferIndex = 0;
}
else
{
currentTargetBufferIndex += (bufferIndex - startIndex);
}
startIndex = bufferIndex = 0;
runningAddress = startBlockAddress = block.DataAddress;
}
}
#endregion
}
else
{
throw new InvalidOperationException("Invalid (-) Block.DataAddress detected.");
}
//**** write Block Header and Data to disk
BinaryWriter.Seek(0, SeekOrigin.Begin);
// Byte 0 to 7: Next Item Address (64 bit long int) = 0 (no next item)
BinaryWriter.Write(block.NextItemAddress);
// Byte 8 to 11: Size Occupied
BinaryWriter.Write(block.SizeOccupied);
// Byte 12 to 19: Low-level next datablock address
BinaryWriter.Write(block.InternalNextBlockAddress);
// Byte 20: count of member blocks, max is 255.
byte memberCount = 0;
if (block.IsHead)
{
int cm = block.CountMembers(true);
memberCount = cm > byte.MaxValue ? byte.MaxValue : (byte)cm;
}
BinaryWriter.Write(memberCount);
byte[] b2 = BufferStream.GetBuffer();
Array.Copy(b2, 0, _writeBuffer, currentTargetBufferIndex + bufferIndex, sizeOfNumerics);
bufferIndex += sizeOfNumerics;
//** Byte 20 to 20 + Data Length: USER DATA
int cs = block.Data.Length;
if (currentTargetBufferIndex + cs + bufferIndex > _writeBuffer.Length - block.Length)
{
cs = _writeBuffer.Length - (currentTargetBufferIndex + bufferIndex);
}
Array.Copy(block.Data, 0, _writeBuffer, currentTargetBufferIndex + bufferIndex, cs);
bufferIndex += block.Data.Length;
runningAddress += block.Length;
}
if (startBlockAddress != -1)
{
//** write to disk
dataChunks.Add(new BulkWriter.DataChunk
{
TargetDataAddress = startBlockAddress,
Index = currentTargetBufferIndex + startIndex,
Size = bufferIndex - startIndex
});
}
if (dataChunks.Count > 0)
{
bulkWriter.Write(parent, _writeBuffer, dataChunks);
}
return(r);
}
private void MapOne(Dictionary <string, object> obj)
{
var user = (Dictionary <string, object>)obj["user"];
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(user, "id"));
// [created_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(user, "created_at"));
// [tweeted_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "created_at"));
// [screen_name] [nvarchar](50) NOT NULL
BulkWriter.WriteVarChar(JsonUtil.GetString(user, "screen_name") ?? "", 50);
// [description] [nvarchar](160) NOT NULL
BulkWriter.WriteVarChar(Util.UnescapeText(JsonUtil.GetString(user, "description")) ?? "", 160);
// [favourites_count] [int] NOT NULL
BulkWriter.WriteInt(JsonUtil.GetNullableInt32(user, "favourites_count") ?? 0);
// [followers_count] [int] NOT NULL
BulkWriter.WriteInt(JsonUtil.GetNullableInt32(user, "followers_count") ?? 0);
// [friends_count] [int] NOT NULL
BulkWriter.WriteInt(JsonUtil.GetNullableInt32(user, "friends_count") ?? 0);
// [statuses_count] [int] NOT NULL
BulkWriter.WriteInt(JsonUtil.GetNullableInt32(user, "statuses_count") ?? 0);
// [geo_enabled] [bit] NOT NULL
BulkWriter.WriteBit(JsonUtil.GetNullableBoolean(user, "geo_enabled") ?? false);
// [lang] [char](5) NOT NULL,
BulkWriter.WriteChar(JsonUtil.GetString(user, "lang") ?? "", 5);
// [location] [nvarchar](100) NULL
BulkWriter.WriteVarChar(Util.UnescapeText(JsonUtil.GetString(user, "location")), 100);
// [name] [nvarchar](30) NOT NULL
BulkWriter.WriteVarChar(Util.UnescapeText(JsonUtil.GetString(user, "name")) ?? "", 30);
// [profile_background_color] [char](6) NOT NULL
BulkWriter.WriteChar(JsonUtil.GetString(user, "profile_background_color") ?? "000000", 6);
// [profile_text_color] [char](6) NOT NULL
BulkWriter.WriteChar(JsonUtil.GetString(user, "profile_text_color") ?? "000000", 6);
// [protected] [bit] NOT NULL
BulkWriter.WriteBit(JsonUtil.GetNullableBoolean(user, "protected") ?? false);
// [show_all_inline_media] [bit] NOT NULL
BulkWriter.WriteBit(JsonUtil.GetNullableBoolean(user, "show_all_inline_media") ?? false);
// [utc_offset] [int] NULL
BulkWriter.WriteNullableInt(JsonUtil.GetNullableInt32(user, "utc_offset"));
// [verified] [bit] NOT NULL
BulkWriter.WriteBit(JsonUtil.GetNullableBoolean(user, "verified") ?? false);
BulkWriter.EndLine();
}
private void MapOne(Dictionary <string, object> obj)
{
// [run_id] [smallint] NOT NULL
BulkWriter.WriteSmallInt(RunID);
// [tweet_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "id"));
// [created_at] [datetime] NOT NULL
BulkWriter.WriteDateTime(JsonUtil.GetDateTime(obj, "created_at"));
// [utc_offset] [int] NULL
BulkWriter.WriteNullableInt(JsonUtil.GetNullableInt32(obj, "user.utc_offset"));
// [user_id] [bigint] NOT NULL
BulkWriter.WriteBigInt(JsonUtil.GetInt64(obj, "user.id"));
// [place_id] [char](16) NULL
BulkWriter.WriteNullableChar(JsonUtil.GetString(obj, "place.id"), 16);
// [lon] [float] NULL
// [lat] [float] NULL
// [cx] [float] NOT NULL
// [cy] [float] NOT NULL
// [cz] [float] NOT NULL
// [htm_id] [bigint] NOT NULL
if (obj.ContainsKey("coordinates") && obj["coordinates"] != null)
{
string[] coords = (string[])((ArrayList)JsonUtil.GetValue(obj, "coordinates.coordinates")).ToArray(typeof(string));
var lon = double.Parse(coords[0], System.Globalization.CultureInfo.InvariantCulture);
var lat = double.Parse(coords[1], System.Globalization.CultureInfo.InvariantCulture);
var c = new Cartesian(lon, lat);
BulkWriter.WriteNullableFloat(lon);
BulkWriter.WriteNullableFloat(lat);
BulkWriter.WriteFloat(c.X);
BulkWriter.WriteFloat(c.Y);
BulkWriter.WriteFloat(c.Z);
BulkWriter.WriteBigInt(Trixel.CartesianToHid20(c));
}
else
{
BulkWriter.WriteNullableFloat(null);
BulkWriter.WriteNullableFloat(null);
BulkWriter.WriteFloat(0);
BulkWriter.WriteFloat(0);
BulkWriter.WriteFloat(0);
BulkWriter.WriteBigInt(0);
}
// [in_reply_to_tweet_id] [bigint] NULL
BulkWriter.WriteNullableBigInt(JsonUtil.GetNullableInt64(obj, "in_reply_to_status_id"));
// [in_reply_to_user_id] [bigint] NULL
BulkWriter.WriteNullableBigInt(JsonUtil.GetNullableInt64(obj, "in_reply_to_user_id"));
// [possibly_sensitive] [bit] NULL
// [possibly_sensitive_editable] [bit] NULL
if (obj.ContainsKey("possibly_sensitive") && obj.ContainsKey("possibly_sensitive_editable"))
{
BulkWriter.WriteNullableBit(JsonUtil.GetNullableBoolean(obj, "possibly_sensitive"));
BulkWriter.WriteNullableBit(JsonUtil.GetNullableBoolean(obj, "possibly_sensitive_editable"));
}
else
{
BulkWriter.WriteNullableBit(null);
BulkWriter.WriteNullableBit(null);
}
// [retweet_count] [int] NOT NULL
BulkWriter.WriteInt(JsonUtil.GetNullableInt32(obj, "retweet_count") ?? 0);
// [text] [nvarchar](150) NOT NULL
var text = Util.UnescapeText(System.Web.HttpUtility.HtmlDecode(JsonUtil.GetString(obj, "text")));
BulkWriter.WriteVarChar(text, 150);
// [truncated] [bit] NOT NULL
BulkWriter.WriteBit(JsonUtil.GetNullableBoolean(obj, "truncated") ?? false);
// [lang] [char](5) NOT NULL
BulkWriter.WriteChar(JsonUtil.GetString(obj, "user.lang") ?? "??", 5);
// [lang_word_count] [tinyint] NOT NULL
// [lang_guess1] [char](2) NOT NULL
// [lang_guess2] [char](2) NOT NULL
int words;
string lang1, lang2;
if (LanguageUtil.DetectLanguage(text, out words, out lang1, out lang2))
{
BulkWriter.WriteTinyInt((sbyte)words); // [lang_word_count]
BulkWriter.WriteChar(lang1, 2); // [lang_guess1]
BulkWriter.WriteChar(lang2, 2); // [lang_guess2]
}
else
{
BulkWriter.WriteTinyInt(0);
BulkWriter.WriteChar("??", 2);
BulkWriter.WriteChar("??", 2);
}
BulkWriter.EndLine();
}
