/// <summary>
/// Constructs a DbDataJoiner instance
/// </summary>
/// <param name="joinOn">Property path from the root object down to the lookup key</param>
/// <param name="dimTableTarget">Table information of the remote dimension table</param>
/// <param name="option">Option to use for this dataflow</param>
/// <param name="batchSize">The batch size for a batched remote look up</param>
/// <param name="cacheSize">The local cache item count (part of the remote table)</param>
public DbDataJoiner(Expression <Func <TIn, TLookupKey> > joinOn, TargetTable dimTableTarget, DataflowOptions option, int batchSize = 8 * 1024, int cacheSize = 1024 * 1024)
: base(option)
{
m_dimTableTarget = dimTableTarget;
m_batchSize = batchSize;
m_batcher = new BatchBlock <TIn>(batchSize, option.ToGroupingBlockOption()).ToDataflow(option);
m_batcher.Name = "Batcher";
m_lookupNode = new TransformManyDataflow <JoinBatch <TIn>, TIn>(this.JoinBatch, option);
m_lookupNode.Name = "LookupNode";
m_typeAccessor = TypeAccessorManager <TIn> .GetAccessorForTable(dimTableTarget);
m_keyComparer = typeof(TLookupKey) == typeof(byte[])
? (IEqualityComparer <TLookupKey>)((object)new ByteArrayEqualityComparer())
: EqualityComparer <TLookupKey> .Default;
m_rowCache = new RowCache <TLookupKey>(cacheSize, m_keyComparer);
m_logger = Utils.GetNamespaceLogger();
m_joinOnMapping = m_typeAccessor.DbColumnMappings.First(m => m.Host.PropertyInfo == this.ExtractPropertyInfo(joinOn));
var transformer =
new TransformBlock <TIn[], JoinBatch <TIn> >(
array => new JoinBatch <TIn>(array, CacheLookupStrategy.RemoteLookup), option.ToExecutionBlockOption()).ToDataflow(option);
transformer.Name = "ArrayToJoinBatchConverter";
m_batcher.LinkTo(transformer);
transformer.LinkTo(m_lookupNode);
RegisterChild(m_batcher);
RegisterChild(transformer);
RegisterChild(m_lookupNode);
m_dimInserter = new DimTableInserter(this, dimTableTarget, joinOn, option)
{
Name = "DimInserter"
};
var hb = new HeartbeatNode <JoinBatch <TIn> >(option);
m_dimInserter.RegisterDependency(m_lookupNode);
m_dimInserter.LinkTo(hb);
hb.LinkTo(m_lookupNode);
RegisterChild(m_dimInserter);
RegisterChild(hb);
RegisterChildRing(transformer.CompletionTask, m_lookupNode, m_dimInserter, hb);
}
/// <summary>
/// Override this method to customize row cache initialization logic
/// </summary>
protected virtual void InitializeCache(RowCache <TLookupKey> cache)
{
string selectPartial = string.Format(
"select TOP {3} [{0}], [{1}] from {2} order by [{0}] desc",
Utils.GetAutoIncrementColumn(m_typeAccessor.SchemaTable.Columns).ColumnName,
m_joinOnMapping.DestColumnName,
m_dimTableTarget.TableName,
cache.SizeLimit);
m_logger.DebugFormat("[{0}] Start to initialize cache using sql: {1}", FullName, selectPartial);
using (var connection = new SqlConnection(m_dimTableTarget.ConnectionString))
{
connection.Open();
SqlCommand command = new SqlCommand(selectPartial, connection);
var autoKeyColumn = Utils.GetAutoIncrementColumn(m_typeAccessor.SchemaTable.Columns);
bool is64BitAutoKey = autoKeyColumn.DataType == typeof(long);
SqlDataReader reader = command.ExecuteReader();
while (reader.Read())
{
long dimKey = is64BitAutoKey ? reader.GetInt64(0) : reader.GetInt32(0); //$inserted.AutoKey
TLookupKey joinColumnValue = (TLookupKey)reader[1]; //$inserted.JoinOnColumn
//add to the subcache no matter it is an "UPDATE" or "INSERT"
cache.TryAdd(
joinColumnValue,
new PartialDimRow <TLookupKey> {
AutoIncrementKey = dimKey, JoinOn = joinColumnValue
});
}
}
m_logger.DebugFormat("[{0}] Global cache initialized with {1} items", FullName, cache.Count);
}