private static Delegate DeserializeUdfs <T>(
UdfWrapperData data,
ref int nodeIndex,
ref int udfIndex)
{
UdfWrapperNode node = data.UdfWrapperNodes[nodeIndex++];
Type nodeType = Type.GetType(node.TypeName);
if (node.HasUdf)
{
var udfs = new object[node.NumChildren];
for (int i = 0; i < node.NumChildren; ++i)
{
udfs[i] = UdfSerDe.Deserialize(data.Udfs[udfIndex++]);
}
return(CreateUdfWrapperDelegate <T>(nodeType, udfs));
}
var udfWrappers = new object[node.NumChildren];
for (int i = 0; i < node.NumChildren; ++i)
{
udfWrappers[i] = DeserializeUdfs <T>(data, ref nodeIndex, ref udfIndex);
}
return(CreateUdfWrapperDelegate <T>(nodeType, udfWrappers));
}
internal static object DeserializeArrowOrDataFrameUdf(
Stream stream,
out SerializedMode serializerMode,
out SerializedMode deserializerMode,
out string runMode)
{
UdfWrapperData udfWrapperData = GetUdfWrapperDataFromStream(
stream,
out serializerMode,
out deserializerMode,
out runMode);
int nodeIndex = 0;
int udfIndex = 0;
UdfWrapperNode node = udfWrapperData.UdfWrapperNodes[nodeIndex];
Type nodeType = Type.GetType(node.TypeName);
Delegate udf = null;
if (nodeType == typeof(DataFrameGroupedMapUdfWrapper))
{
udf = (DataFrameGroupedMapWorkerFunction.ExecuteDelegate)DeserializeUdfs <DataFrameGroupedMapWorkerFunction.ExecuteDelegate>(
udfWrapperData,
ref nodeIndex,
ref udfIndex);
}
else if (nodeType == typeof(DataFrameWorkerFunction) || nodeType.IsSubclassOf(typeof(DataFrameUdfWrapper)))
{
udf = (DataFrameWorkerFunction.ExecuteDelegate)DeserializeUdfs <DataFrameWorkerFunction.ExecuteDelegate>(
udfWrapperData,
ref nodeIndex,
ref udfIndex);
}
else if (nodeType == typeof(ArrowGroupedMapUdfWrapper))
{
udf = (ArrowGroupedMapWorkerFunction.ExecuteDelegate)DeserializeUdfs <ArrowGroupedMapWorkerFunction.ExecuteDelegate>(
udfWrapperData,
ref nodeIndex,
ref udfIndex);
}
else
{
udf = (ArrowWorkerFunction.ExecuteDelegate)
DeserializeUdfs <ArrowWorkerFunction.ExecuteDelegate>(
udfWrapperData,
ref nodeIndex,
ref udfIndex);
}
// Check all the data is consumed.
Debug.Assert(nodeIndex == udfWrapperData.UdfWrapperNodes.Length);
Debug.Assert(udfIndex == udfWrapperData.Udfs.Length);
return(udf);
}
private static void SerializeUdfs(
Delegate func,
UdfWrapperNode parent,
List <UdfWrapperNode> udfWrapperNodes,
List <UdfSerDe.UdfData> udfs)
{
UdfSerDe.UdfData udfData = UdfSerDe.Serialize(func);
if ((udfData.MethodName != UdfWrapperMethodName) ||
!Attribute.IsDefined(func.Target.GetType(), typeof(UdfWrapperAttribute)))
{
// Found the actual UDF.
if (parent != null)
{
parent.HasUdf = true;
Debug.Assert(parent.NumChildren == 1);
}
udfs.Add(udfData);
return;
}
UdfSerDe.FieldData[] fields = udfData.TargetData.Fields;
if ((fields.Length == 0) || (fields.Length > 2))
{
throw new Exception(
$"Invalid number of children ({fields.Length}) for {udfData.TypeData.Name}");
}
var curNode = new UdfWrapperNode
{
TypeName = udfData.TypeData.Name,
NumChildren = fields.Length,
HasUdf = false
};
udfWrapperNodes.Add(curNode);
foreach (UdfSerDe.FieldData field in fields)
{
SerializeUdfs((Delegate)field.Value, curNode, udfWrapperNodes, udfs);
}
}
