/// <summary>
/// Maps each group of the current DataFrame using a UDF and
/// returns the result as a DataFrame.
///
/// The user-defined function should take an Apache Arrow RecordBatch
/// and return another Apache Arrow RecordBatch. For each group, all
/// columns are passed together as a RecordBatch to the user-function and
/// the returned RecordBatch are combined as a DataFrame.
///
/// The returned <see cref="RecordBatch"/> can be of arbitrary length and its
/// schema must match <paramref name="returnType"/>.
/// </summary>
/// <param name="returnType">
/// The <see cref="StructType"/> that represents the shape of the return data set.
/// </param>
/// <param name="func">A grouped map user-defined function.</param>
/// <returns>New DataFrame object with the UDF applied.</returns>
public DataFrame Apply(StructType returnType, Func <RecordBatch, RecordBatch> func)
{
ArrowGroupedMapWorkerFunction.ExecuteDelegate wrapper =
new ArrowGroupedMapUdfWrapper(func).Execute;
UserDefinedFunction udf = UserDefinedFunction.Create(
Reference.Jvm,
func.Method.ToString(),
CommandSerDe.Serialize(
wrapper,
CommandSerDe.SerializedMode.Row,
CommandSerDe.SerializedMode.Row),
UdfUtils.PythonEvalType.SQL_GROUPED_MAP_PANDAS_UDF,
returnType.Json);
IReadOnlyList <string> columnNames = _dataFrame.Columns();
var columns = new Column[columnNames.Count];
for (int i = 0; i < columnNames.Count; ++i)
{
columns[i] = _dataFrame[columnNames[i]];
}
Column udfColumn = udf.Apply(columns);
return(new DataFrame((JvmObjectReference)Reference.Invoke(
"flatMapGroupsInPandas",
udfColumn.Expr())));
}
